In This Issue:

VOLUME LXXiv, NUMBER 4, WinteR 2012 Minding Brains • Mending Retinas • Pumping Blood 2 www.caltech.edu Stay connected to campus, wherever you are.

3 From the President

4 Random Walk Scintillating Neutrinos • Sporulating Membranes • Selenocentric Spacecraft • And More!

14 Different Minds by Lori Oliwenstein The word autism covers what is really a spectrum of disorders, but they all appear to result from nonstandard connections between various parts of the brain.

20 Connecting the Dots by Katie Neith Implantable electronics can reconnect parts of the nervous system that have been damaged by disease or disability.

26 Programming Molecular Apps by Marcus Y. Woo The future of programming may lie with DNA—but the vision behind tomorrow’s apps extends beyond biology.

34 Naturally Inspired by Kimm Fesenmaier The Wright Brothers designed flying machines by watching birds. Caltech bioengineers, inspired by zebrafish embryos, are designing “living” pumps.

38 Some Assembly Required by Kathy Svitil Artificial tissues made from lifelike proteins could give “growing your own” in a medical context an entirely new meaning.

42 Alumni Impact by Katharine Gammon Caltech alumni began leaving their mark in translational medicine nearly a century before there was such a field.

44 Endnotes We asked alumni . . . VOLUME LXXiv, NUMBER 4, WinteR 2012 www.caltech.edu Stay connected no matter where you are

WATCH, LISTEN, AND LEARN headlines JOIN THE COMMUNITY

Featured Videos News and Information today.caltech.edu itunes.caltech.edu youtube.com/user/Caltech @caltech on Twitter

1 “The Caltech basketball team is the least Senior Caltech intimidating team ever . . . until they graduate.” Nobelist Reflects on Prize-Winning Work California Institute of Technology on Facebook 2011 Convocation Ceremony 2 LIKE Meet the Class of 2015!

Smart Petri Dish 3 Caltech Welcomes New Faculty Members Caltech/Sci-Arc Solar Decathlon series Visit E&S online at www.eands.caltech.edu

Read more on Caltech Today Volume LXXIV, Number 4, winter 2012

From the President Editor Douglas L. Smith Contributing Editors You might not think of Caltech when you think of advances in clinical medicine. Lori Oliwenstein, Kathy Svitil But perhaps you should. Contributing Writers No, we’re not building a hospital, or opening a medical school. But we Kimm Fesenmaier, Binti Harvey, Katie Neith, Doreese are establishing enduring partnerships with top facilities in Southern California Norman, Lori Oliwenstein, Ann Wendland, Marcus Y. Woo and beyond. Copy Editors Allie Akmal, Michael Farquhar Though these partnerships—and our ever-sharpening focus on developing Art Director solutions to medical, biomedical, and biotechnological problems—we are taking Jenny K. Somerville vital research from laboratory benches to patients’ bedsides. Staff Artist But we can’t do it alone. Which is why Joel Burdick is working with neurobi- Lance Hayashida ologists from UCLA and physicians from the University of Louisville to test his Design and Layout “The Caltech basketball team is the least paralysis-reversing electrode array; it’s why Yu-Chong Tai is working with oph- Etch Creative Business Manager intimidating team ever . . . until they graduate.” thalmologists from USC to test the retinal implants he’s devised. (Both projects are described in “Connecting the Dots,” page 20.) It’s why David Tirrell and Julia Rosanne Lombardi Kornfield are collaborating with scientists at UC San Francisco on implantable The California Institute of Technology materials meant to heal damaged corneas. (“Some Assembly Required,” page 38.) Peter B. Dervan—Vice President for Development and Institute Relations It’s why the Broad Foundation gave Caltech and UCLA $5 million to start a Kristen Brown—Assistant Vice President for Marketing Joint Center for Translational Medicine. And it’s why, this summer, an anonymous and Communications donor gave $3 million each to Caltech and City of Hope to strengthen scientific Binti Harvey—Director of Institute Marketing collaborations between these institutions. The Caltech Alumni Association We know that to accelerate the speed at which game-changing discoveries Charles Halloran—President move from the laboratory to patient care, you need extraordinary people doing Read E&S on the go at http://EandS.caltech.edu extraordinary work. And you need other extraordinary people to fund that research. Contact E&S at [email protected] Take, for instance, Ben and Donna Rosen. Ben Letters selected for publication may be edited for length, (BS ’54) is a Caltech trustee, known for his leader- content, and clarity. ship as a venture capitalist and in the computer industry—and now for his leadership at Caltech. Engineering & Science (ISSN 0013-7812) is published quarterly at the California Institute of Technology, 1200 Our three-year-old Donna and Benjamin Rosen East California Boulevard, Pasadena, CA 91125. Annual Bioengineering Center has already become a hub subscription $20.00 domestic, $30.00 foreign air mail; for collaborations. single copies $5.00. Send subscriptions to Engineering & Science, Caltech Mail Code 5-32, Pasadena, CA 91125. There’s also the Anna L. Rosen Professorship— Third-class postage paid at Pasadena, CA. All rights Ben and Donna Rosen named for Ben’s mother—which is held by Scott reserved. Reproduction of material contained herein for- Fraser, whose bioimaging work is described in “Pro- bidden without authorization. © 2011, California Institute of Technology. Published by Caltech and the Alumni gramming Molecular Apps” on page 26, and “Naturally Inspired” on page 34. Association. Telephone: 626-395-8371. Among the more than 50 beneficiaries of the Rosen Graduate Fellowship Fund at Caltech are Hesham Azizgolshani and Derek Rinderknecht, Printed by Sinclair Printing, Los Angeles, California who have helped develop tiny implantable valveless pumps (described in “Naturally Inspired”). PICTURE CREDITS Caltech’s mission is “to expand human knowledge and benefit society through Cover, inside front cover, 1, 14, 21, 27, 34, 39, 44 — Bill E&S Youngblood; 4–5 — Roy Kaltschmidt, Lawrence Berkeley research integrated with education.” In this issue of , you’ll see just how National Laboratory; 6–7 — Everett Kane; 8–9 — NASA/ seriously we take that charge. JPL; 9 — NASA/KSC; 11, 20–21, 22, 30, 34, 36, 40, 42, 42–43, 44 — Lance Hayashida; 11 — NASA/Goddard/ Yours in discovery, ASU; 15, 16, 19 — Mary Roy; 17 — Daniel Kennedy and Ralph Adolphs; 23 — Mandheerei (Monty) Nandra; 25 — University of Louisville; 28 — Paul Rothemund; 31 — Harry Choi; 35 — Derek Rinderknecht; 37 — Adrianus Aria, Jijie Zhou; 38, 40 — Sidney Harris; 39 — Bob Paz; 42 — Briana Ticehurst, Caltech Archives, Rockefeller University; 43 — Caltech Archives, NIH/NLM/UCLA, the Herzenberg research group; 44 — iStock.com

Special thanks to the Polytechnic School for the use of Mandy the mannequin, and to the Caltech Store for the loan of the fashion wear. Random Walk

This image is a glimpse into the depths of a neutrino detector that’s part of the Daya Bay Neutrino Experiment, buried under the moun- tains of southern China near Hong Kong. The detector consists of two inner nested transpar- ent acrylic cylinders. Each of the experiment’s eight 100-ton detectors is filled with a clear, liquid scintillator that flashes when an antineu- trino—a neutrino’s antimatter counterpart—zips through and interacts. The alien-looking globules along the inner walls are photomultiplier tubes that amplify and record the signals. The experi- ment, for which Caltech designed and built the 24 calibration devices, started taking data last August, probing the nature of the neutrino, that elusive particle that flies through the cosmos at nearly the speed of light. Physicists hope the small particles will help reveal answers to big questions—for example, why the universe has more matter than antimatter.

4 ENGINEERING & SCIENCE winter 2012 Random WalkThings that caught our eye . . .

winter 2012 ENGINEERING & SCIENCE 5 random walk

6 ENGINEERING & SCIENCE winter 2012 Flip It and Reverse It

A team of Caltech researchers has Using a found what it’s calling a “missing link.” powerful technique No, not that missing called electron cryoto- membrane, link—a link you mography (ECT), Professor of even though it origi- probably didn’t Biology Grant Jensen and members of nated as part of the mother cell’s know was his lab captured the first high-resolution inner membrane. This, they say, missing in the images of A. longum undergoing sporu- means that sporulation could have first place. lation. The results were surprising: “We been the mechanism by which the The group is saw that a piece of the inner membrane bacterial outer membrane arose. referring to a actually becomes the new cell’s outer The whole study started after bacterium, called membrane,” says Jensen, who is also Jensen spoke with Professor of Acetonema longum, a Howard Hughes Medical Institute Environmental Microbiology Jared which seems to be investigator. Leadbetter about the capabilities the link between In the case of a double-membraned of ECT, which allows researchers bacterial species bacterium such as A. longum, sporula- to image biological specimens in a with two tion begins with the inner membrane near-native state rather than requiring membranes pinching together asymmetrically, them to be dehydrated, embedded and those with creating a mother cell and a smaller in plastic, sectioned, and stained. just one. daughter cell, both covered by the outer Leadbetter, who has long been A. longum membrane. Next, the mother cell engulfs interested in the process of sporula- belongs to a the daughter, giving the daughter a tion, wondered if the technique might little-known second layer of what was originally inner be used to image sporulating cells. family of bac- membrane. At this point, the daughter is Alas, the model organism for studying teria that have a spore, surrounded by two membranes sporulation, Bacilus subtilis, is too two membranes within the mother cell. Having done her thick to be imaged using ECT. But A. and respond maternal duty, the mother cell dies away, longum saved the day—in addition to extreme leaving the spore protected by those to its tendency to sporulate, it’s also environmental two membranes and a protein coat. relatively skinny. stresses by When conditions improve and the Elitza Tocheva, a postdoc in forming protec- spore germinates, part of its protective Jensen’s lab, is the lead author of the tive spores, a protein coat cracks open and the new paper describing the work, which process known cell outgrows its former shell. Unlike a appeared in the September 2 issue as sporulation. single-membraned bacterium, which of Cell. The other authors include Outside of this would at this point also shed its outer postdoc Eric Matson, grad student small family, membrane, the double-membraned Dylan Morris, and Farshid Moussavi the only bacteria bacterium retains it. of Stanford. —KF known to sporulate Jensen’s group found that the new are those with a outer membrane has the structure and single membrane. all of the functions of a typical outer

An artist’s representation of the entire sporulation process as a single drawing.

winter 2012 ENGINEERING & SCIENCE 7 Moonward Bound GRAIL, NASA’s Gravity Recovery and Interior Labora- tory, will answer longstanding questions about the moon and give scientists a better understanding of how Earth and other rocky planets in the solar system formed. The twin spacecraft lifted off on a single rocket from Cape Canaveral at 9:08 a.m. Florida time on September 10, and they will fly in tandem orbits around the moon to measure its gravity field. GRAIL-A is sched- uled to reach the moon on New Year’s Eve 2011, while GRAIL-B will arrive New Year’s Day 2012. The science- collection phase for GRAIL is expected to last 82 days. This artist’s conception shows the two spacecraft using radio links to each other to measure the distance between themselves while continuously relaying the information back to Earth—a feat they can accomplish even when the moon is between one of them and us. The inset photo was shot on August 23, as technicians were testing the clamshell fairing that protects the copper-foil-covered spacecraft during launch. JPL manages the GRAIL mission, which is part of NASA’s Discovery Program. Lockheed Martin Space Systems in Denver built the spacecraft. — DN

8 ENGINEERING & SCIENCE winter 2012 winter 2012 ENGINEERING & SCIENCE 9 10 ENGINEERING & SCIENCE winter 2012 Jackie Barton Wins National Medal of Science Chemistry professor and division chair single mismatch in the DNA’s nucleic- electron to the other, the wire is unbroken Jacqueline K. (Jackie) Barton has acid sequence can prevent the transfer and the DNA is OK. Mutations in these good reason to celebrate. She is one from happening. “The DNA’s base pairs proteins have been linked to predisposi- of a handful of recipients of this year’s are something like a stack of pennies,” tions to colon and breast cancer. National Medal of Science, the highest Barton says. “If you interrupt the stack Barton joined the Caltech faculty as a honor bestowed by the United States in some way—if you distort even one professor of chemistry in 1989. She was government on scientists. The award penny—it interrupts the conductivity of named the Arthur and Marian Hanisch recognizes her discovery of a new the stack.” Memorial Professor in 1997 and became property of the DNA helix—long-range Nature might use this conductiv- the chair of the Division of Chemistry and electron transfer. ity to locate and repair lesions in the Chemical Engineering in 2009. Barton is Over more than 20 years, Barton has DNA, Barton says. There are metal- also the recipient of the 1985 National pieced together an understanding of containing proteins that are believed to Science Foundation Waterman Award, how double-helical DNA can behave be involved in the repair process; in her the 1988 American Chemical Society like a wire, allowing the transfer of elec- model, two of these proteins “test the Award in Pure Chemistry, and a 1991 trons across long molecular distances. wire” by attaching themselves to the MacArthur Foundation Fellowship. Her experiments have revealed that a DNA at widely separated points. Then, —KF if one protein successfully sends an

The Moon and Footprints

This high-resolution image of the Apollo 17 landing site, taken by the Lunar Reconnaissance Orbiter last September, has a pixel size of 27 by 56 centime- ters—almost enough to distinguish the individual footprints left by astronauts Eugene Cernan and Harrison Schmitt (BS ’57), who is the only geologist ever to walk on another world. The Apollo Lunar Surface Experiments Package (ALSEP), the descent stage of the lunar module Challenger, and the lunar rover (LRV) are all clearly visible, and the foot trails made by the astronauts are easily distinguishable from the dual tracks left by their rover.

winter 2012 ENGINEERING & SCIENCE 11 the art of science Caltech’s third annual Art of Science competition was held last June. Here are some of our favorite images from the show.

12 ENGINEERING & SCIENCE winter 2012 FIRST-PRIZE WINNER (left) ARTIST: Floris Van Breugel (grad student) TITLE: Fluorescent Treasures Scheelite (blue/white), calcite (red), uranyl ions (green), fluorite (pink), and “desert varnish” (yellow/orange) fluoresce under ultraviolet light in this 20-minute-long time-lapse exposure of tailings at a tungsten mine near Darwin, California.

SECOND-PRIZE WINNER (right) ARTIST: Young Shik Shin (MS ’06, PhD ’11) TITLE: Single-Cell Barcode Chip This image shows 960 sets of protein data from individual cells in a microfluidic device. Each red bar represents a specific the art of science cancer-related protein; the green bars are reference proteins. Such analysis can offer information missed by a conventional cell screening in bulk.

THIRD-PRIZE WINNER (left) ARTISTS: Andrew Leenheer (grad student) and Nick Strandwitz (postdoc) TITLE: Crystalrise The topography, as measured by confocal reflectance microscopy, of gallium phosphide crystals grown on a silicon microwire array to be used for photoelectro- chemical solar fuels. The rainbow color scale covers 15 microns height.

winter 2012 ENGINEERING & SCIENCE 13 Different Minds

If there’s one thing Ralph Or not. The fact that there are few An infectious theory Adolphs wants you to under- absolutes is part of why the set of dis- Much of that insight comes from Pat- stand about autism, it’s this: “It’s orders—autism, Asperger’s, pervasive terson, who pioneered the study of the wrong to call many of the people developmental disorder not otherwise connections between the brain and the on the autism spectrum impaired,” says specified (PDD-NOS)—that fall under immune system in autism, schizophrenia, the Caltech neuroscientist. “They’re autism’s umbrella are referred to as and depression a decade ago. simply different.” a spectrum. As with the spectrum of The main focus of that connection? These differences are in no way insig- visible light—where red morphs into Some kind of viral infection during preg- nificant—they are, after all, why so much orange, which morphs into yellow— nancy, Patterson explains. Or, rather, the effort and passion is being put into un- it is difficult to draw sharp lines immune response that infection inevitably derstanding autism’s most troublesome between the various diagnoses in engenders. traits—but neither are they as inevitably the autism spectrum. To bolster his argument, Patterson devastating as has often been depicted. And, like the autism spectrum itself, points to a recent study by Hjordis O. They are simply differences; intriguing, the spectrum of autism research at Atladottir of Aarhus University, Denmark, fleeting glimpses into minds that work in Caltech also runs a gamut. Adolphs, and colleagues—“an extraordinary look at ways most of us don’t quite understand, for example, studies brain differences over 10,000 autism cases” in the Danish and yet which may ultimately give each between adults on the high-functioning Medical Register, which is a comprehen- and every one of us a little more insight portion of the autism spectrum and the sive database of every Dane’s medical into our own minds, our own selves. general public—the so-called neuro- encounter from cradle to grave—that What makes autism so fascinating, typical population. Neurobiologist John showed a strong epidemiological link Adolphs notes, is what also makes it so Allman’s work on a particular set of neu- between autism in a child and a first- difficult to study, to get a good grasp on: rons is providing insight into a possible trimester viral infection in the mother. the diversity of the population itself. If basis for some of autism’s social quirks. “And we’ve found that if you give a mouse you’ve met one person with autism, the And then there’s neurobiologist Paul the flu during pregnancy, or activate the saying goes, you’ve met one person Patterson, who literally has just finished mother’s immune system directly without with autism. writing the book on the connection a pathogen, the offspring will show a Still, there are characteristic tenden- between the immune system and neuropathology characteristic of autism, cies and traits, gifts and gaps. According autism, and who is collaborating with as well as the three cardinal symptoms to psychiatry’s diagnostic bible, familiarly biologist Sarkis Mazmanian to explore of autism.” known as the DSM-IV, autism involves the connections between gastrointes- In mouse terms, that means exhibit- an “impairment” in social interaction and tinal symptoms and the brain in autism ing repetitive behavior such as excessive communication, as well as “restricted re- spectrum disorders. self-grooming and compulsively and petitive . . . patterns of behavior, interests, They and half a dozen or so of their furiously burying marbles placed on its and activities.” What this means to the colleagues have come together in a bedding. Freaking out, rather than making rest of us is some combination of these: sort of informal Autism Working Group, friends, when faced with an unfamiliar Lousy eye contact. Social awkwardness. which meets regularly to talk about their mouse. And communicating much less Hand flapping.T oe walking. Late-talking individual findings and to brainstorm as a young pup with its mother, or as an babies. Kids who can’t seem to get the together. “People don’t think of Caltech adult with other mice. (Mice “talk” to one hang of playing cops and robbers with when they think about autism research,” another at ultrasonic frequencies, requir- friends, yet can talk for an hour about says Patterson. “But we have a lot go- ing a special microphone and recording Egyptian gods, elevators, or elephants. ing on here; a lot of insight to offer.” equipment to make their conversations Adults who speak in flat tones, who audible to human ears.) come off as rigid, uncomfortable, brilliant.

14 ENGINEERING & SCIENCE winter 2012 Different MinBy Lori Oliwenstein ds

Caltech researchers are gaining insight into what autism is, what autism gives, and where autism lives.

winter 2012 ENGINEERING & SCIENCE 15 This is an impor- dysfunction there—then the breakdown tant clue connecting product gets into the bloodstream. And the mother’s im- research has indeed found that this mune reaction to the happens more often in kids with autism fetus’s brain develop- than in typical kids.” ment. It’s known that There’s similarly strong evidence, he the interleukins are says, implicating our normally helpful inflammation-boosters, and cooperative intestinal bacteria—as and inflammation is it turns out, the various types of bacteria all well and good found in the digestive systems of typical when you need to and autism spectrum kids tend to differ. wake up the immune But while the bowel-bacterium-brain system and get it to connection is clear, it’s still not known pour white cells into just how widespread it is. “Estimates an infected area. But of the percentage of people with autism it has its unintended who are affected by these gastro- side effects as well; intestinal differences vary from 20 to Patterson observed the same neuro- the inflammatory response often takes 70 percent,” Patterson says. “Clearly, pathology and behavior in mice who had a scattershot approach, meaning that the science is lagging; there’s still a lot never had an infection, but were instead healthy tissues can also get caught up of work to be done.” given synthetic molecules that mimic in the onslaught. Which is why the science-and- the presence of a virus. In other words, For instance, grad student Elaine advocacy group Autism Speaks recently Patterson found that simply mounting Hsiao is finding that IL-6 and other brought Patterson and Mazmanian to- an immune response may be enough cytokines not only work directly on the gether with immunologist Paul Ashwood to steer a fetus down a path that could fetal brain, but can also alter the func- at UC Davis and pediatrician Alessio eventually lead to an autism-like condi- tion of the placenta, which would have Fasano from the University of Maryland tion after birth. indirect effects on fetal development. to explore the gut-brain-immune con- That “may” is key. There is definitely Why does an immune response nections in mice and in kids with autism a connection between infections during sometimes result in a developmental spectrum disorders. pregnancy—especially the early months disability—and sometimes not? Pat- “You have Sarkis’s and Alessio’s of pregnancy—and increased risk of terson thinks it could be the timing of expertise in the human gastrointestinal autism (and, as Patterson has also the infection, or the genetic makeup of system, you have Paul’s expertise in shown, schizophrenia) in the offspring, the mother, or her fetus that modulates children with autism, and then you have he says, but that is not the same as the outcome. our mouse model,” says Patterson. “That saying it’s a direct, 100 percent, if-you- The questions Patterson is asking makes for a pretty exciting consortium. It get-sick-while-you’re-pregnant-your- about inflammation and the brain might should allow us to look at autism in new baby-will-have-autism correlation. apply to other parts of the body as and unprecedented ways.” How does it all work? Patterson thinks well. So he and biologist Sarkis cytokines—proteins produced by the Mazmanian are looking at the gastro- Looking and connecting body in response to an infection—play intestinal problems somewhat com- Ralph Adolphs, too, is looking at autism a role. In fact, grad student Steve Smith monly found in kids with autism, and in new ways—and looking at looking in (PhD ’08) found that giving pregnant how those might be related to the autism as well. mice one cytokine in particular, called development of autism. Adolphs, who runs the Autism and interleukin-6 (IL-6), leads to offspring “The evidence that there are gastro- Asperger Syndrome Research Program who exhibit autistic and schizophrenic intestinal issues in children with autism at Caltech, spends most of his time features. “It was really unexpected that is pretty good,” Patterson notes. “You working with high-functioning adults a single injection of a single cytokine test it by feeding a child a molecule with autism. “We’re exploring the social would exert such a powerful effect,” that gets broken down in the gut; differences people with autism have,” Patterson notes. if the gut is leaky—if there is some says Adolphs. “We’re looking at every-

16 ENGINEERING & SCIENCE winter 2012 thing from eye-tracking, to reasoning, to about what you notice around you, and not only to your emotional awareness complex theory-of-mind stuff, which is how you respond to it, rather than a lack and empathy, but also to your internal about understanding what others feel of caring. “What’s interesting is figuring awareness—of the state of your or think. We hope one day to be able to out what people with autism pay atten- digestive system, the sensation of tell a story about the roles of particular tion to,” says Adolphs. “It’s not that pain, and the like. This fits well with brain areas, and how the white-matter their amygdala doesn’t work, it’s that the known tendency of people with connections between them look abnor- it works differently.” autism to be hypersensitive to sound, mal, and how that causes the effects Neurobiologists John Allman and light, temperature and other stimuli. we see.” Atiya Hakeem (BS ’93) have also been The missed or unusual connec- For example, people with autism look focusing on a component of the brain tions both Adolphs and Allman are at faces differently—focusing not on the that may work differently in people looking at may be a critical part of the other person’s eyes, the way a neuro- with autism. Called the von Economo story, Adolphs adds. Indeed, he says, typical person would, but on the center neurons, these cells are thought to play there’s a sort of “globally abnormal of the face or on a point beside the a key role in your ability to quickly and connectivity in the brains of people face. Several research groups, including intuitively assess a situation—say, for with autism.” Adolphs’s lab, have discovered that this example, when you meet someone for Which is why Adolphs and Caltech is linked to the brain’s amygdala, which the first time, and come to an immedi- psychology researcher Lynn Paul is known to play roles in face recogni- ate conclusion about how to respond have been studying a group of people tion and in the processing of emotions— to their greeting. Those sorts of snap born without a corpus callosum—the both our own and others’. judgments, Allman says, are much bundle of nerve fibers that connects The amygdala is also involved in more difficult for people with autism, the brain’s two hemispheres. Intrigu- social interactions, so Adolphs’s team who tend to have trouble realizing that ingly, a full third of such people also is looking at how being aware of other what you say when you first meet a new meet the diagnostic criteria for having people affects behavior. In one set of client is very different from the way you an autism spectrum disorder. experiments, volunteers were asked should respond upon first meeting your The team recently found that to make charitable donations (with brother’s new girlfriend or your long-lost despite the missing 200 million or their own money!) in the presence or Aunt Sadie. so connections between neurons absence of an observer. As expected, And that, he posits, may well be due that the corpus callosum normally neurotypical people gave bigger sums to abnormalities in the connections provides, these folks’ left and right when other people were watching. By these neurons make or where they wind hemispheres still manage to carry on contrast, the people with high-function- up in the brain. The latter notion has communications with one another. ing autism gave the same amount either since begun to gain traction else- “This finding really amazed us,” says way. “For most of us, when other people where—a recent study led by Micaela Adolphs, “and it shows that you can are watching, we act differently,” he Santos of the Mount Sinai School of generate remarkably normal func- says. “But we’ve found that people with Medicine in New York has found that tional networks that communicate autism lack that behavioral change. They children with autism have an oversupply between different parts of the brain don’t seem to think about what other of von Economo neurons in the fronto- even when the wiring is all scrambled. people think of them.” insular cortex relative to another type There must be some fundamental But at its root, this seems to be more of neuron. This brain structure is linked principle about how the brain orga-

These eye-tracking studies show how we “read” other people’s faces. Left: Neurotypical people look at the eyes. Middle: People with autism tend to focus on or around the bridge of the nose. Right: Superposing the two, with red being the neurotypical folks and blue being people with autism.

winter 2012 ENGINEERING & SCIENCE 17 nizes itself functionally during develop- ity,” says Adolphs. ment, such that it can do so even when “People with autism the corpus callosum is entirely missing.” have strengths and Although the corpus callosum study they have weakness- looked at a specialized population, most es, just as we all do.” of Adolphs’s other findings have come Mapping these about thanks to the cooperation of some peaks and valleys 40 adult volunteers with diagnoses of will not only help high-functioning autism, Asperger’s, or academics like PDD-NOS. Adolphs and his Why study high-functioning adults? Caltech colleagues Adolphs admits that at least some of learn the ins and outs the impetus is purely practical. It’s much of autism, it will help easier to get adults to sit through long society as a whole hours of testing and MRI scanning than understand how it is to get any child—with or without best to help people autism—to do the same. But the payoff is with autism exploit worth it. Those hours they’ve spent lying their strengths and work around their Paul Patterson is the Anne P. and motionless in the MRI mean that Adolphs weaknesses. Benjamin R. Biaggini Professor of knows these people inside and out. And lest anyone think these lessons Biological Sciences at Caltech. His And there are other draws as well. apply only to those with a diagnosis that research on autism and schizophrenia is For one thing, the overwhelming majority falls clearly onto the autism spectrum, funded by Autism Speaks, the of autism research is done on children Adolphs has a few words of caution. International Rett Syndrome Foundation, and is about children. “You hear a great “Autism is just one extreme of this broad the National Institute of Mental Health deal about children with autism, but not spectrum of individual differences that (NIMH), the U.S.-Israel Binational Science so much about adults,” Adolphs says. stretch across the general population,” Foundation, the Simons Foundation, the “In fact, although autism starts to be he says. For instance, in a 2008 paper McGrath Foundation, the Weston Havens noticed in childhood, it’s not a child- published in Current Biology, Adolphs Foundation, the Della Martin Founda- tion, the Department of Defense, and the hood disorder. Instead, it’s a disorder of and colleagues showed that the par- Caltech Innovation Initiative. His book, development, and those children who ents of children with autism—who were Infectious Behavior: Brain-Immune Con- have autism become adults with autism, not themselves autistic—frequently nections in Autism, Schizophrenia, and and eventually elderly adults with autism. respond to faces in a manner similar Depression, was published in September We know extremely little about how to children with autism, spending more by the MIT Press. autism interacts with the aging process, time looking at other people’s mouths for instance.” than at their eyes. Ralph Adolphs is the Bren Professor Indeed, he notes, while some of “It turns out that what we’re learning of Psychology and Neuroscience and autism’s symptoms are most appar- about autism,” Adolphs notes, “may be professor of biology, as well as direc- ent in childhood—delayed speech, for relevant to almost anyone who is a little tor of the Caltech Brain Imaging Center. instance—others don’t make themselves bit different, who does things differently.” His research is supported by grants from known until later in life and are a direct That means you over there in the cor- Autism Speaks, NIMH, and the Simons result of the atypical development that ner, who hates large parties and small Foundation. came before. “Autism affects develop- talk. And you, the person who plugs ment,” Adolphs says, “and in turn, devel- your ears at a concert, because you feel John Allman is the Frank P. Hixon Pro- opment affects autism. And so we want physically assaulted by the noise. And fessor of Neurobiology. His research is to look at the interaction of autism with you, spending hours upon hours fixing funded by the James S. McDonnell Foun- the entire developmental lifespan.” up old computers, or old cars, or orga- dation, NIMH, and the Simons Foundation. They also want to look at the way au- nizing your model-train collection. He and neuropsychiatrist Peter Williamson tism interacts to create not only deficits, Not diseased. Not impaired. Just are the coauthors of The Human Illnesses: but benefits. “One of the things we’re different. Just like you and me. Neuropsychiatric Disorders and the Nature getting with such detailed assessments of the Human Brain, published in January is a profile of peaks and valleys of abil- 2011, by the Oxford University Press.

winter 2012 ENGINEERING & SCIENCE 19 A multidisciplinary approach to building implantable neural devices could help blind people see, paralyzed people stand, and even endow robotic limbs with a sense of touch. By Katie Neith Like the fiber-optic electrical pulses and/or restore the ideas,” says Tai of the beginning of network beneath a right electrical pulses,” explains engi- his partnership with Mark Humayun. teeming metropo- neer Yu-Chong Tai, who specializes “Since then, we’ve been working on lis, a labyrinth of in making micro- and nano-devices. retinal implants nonstop.” neurons sends signals up, down, “We can help the nerve to operate That was eight years ago. Now, and across your body. This marvel- properly, possibly even better than their second-generation implant, the ously complex system helps us to before. And we do this by making Argus II, has been approved for com- see stunning sunsets, hear brilliant electrodes.” mercial use, receiving patent number symphonies, and achieve athletic Just as a current moving through 8,000,000—a milestone for the feats—things that most people take a coil of wire can generate a current patent office that netted the device for granted until there is a break in in a second coil, an electrode can a mention on National Public Radio’s the matrix. Accidents, injuries, and induce a current within a neuron. If the Morning Edition. degenerative disease can easily electrode is properly positioned on Designed for people with degen- render the eyes sightless, the ears the far side of the broken pathway, it erative eye diseases such as macular deaf, and limbs paralyzed by discon- can jolt the neuron back into activ- degeneration, the implant is em- necting neural pathways. At Caltech, ity, helping a paralyzed person’s legs placed inside the eye, up against the researchers are working on innovative decide, for example, that it’s time to retina. The implant’s electrodes are ways to restore nerve functions via stand. Explains Tai, “The ‘start’ signal thus in direct contact with the gan- implantable devices. from the brain has stopped, but we glion cells, which lie on the retina’s provide the signal with the implant.” inner surface and collectively form Hitting a Nerve the optic nerve. The user wears a Information flows through nerve cells, The Body Electric pair of glasses equipped with a video or neurons, in the form of electrical Tai entered the field after a chance camera plugged in to a tiny computer impulses that travel from one end of meeting with an ophthalmologist that fits in the user’s pocket. The the cell to the other. In some cases, from the Doheny Eye Institute at the processed images are sent wirelessly this is quite a journey—for example, University of Southern California, who to a receiver chip implanted behind the sciatic nerve is a single set of cells had an idea for a retinal implant but the conjunctiva, which is the mucous that sends signals from the spinal needed an engineer to help with the membrane surrounding the eye. The cord all the way to your feet. technology. signals then travel by a thin intraocu- “If neurons can propagate electrical “I’m a small-device person, so when lar cable to the ganglion cells in the pulses a long distance in the body, we started to brainstorm, you can retina. From there, they take the nor- then we can interfere with those imagine that we came up with many mal route to the visual cortex at the

As Yu-Chong Tai says, 16 pixels (opposite page) is not much resolution for seeing the world. But as the resolution increases from 64 pixels (far left) to 256 pixels (center) to 1,024 pixels (left), a familiar face emerges.

winter 2012 ENGINEERING & SCIENCE 21 One in trol of his bladder and bowels even every 50 when the device is off—a “surprise” Americans outcome, Burdick says, which if suffers from replicated in other patients would be some form a huge quality-of-life improvement. of paralysis. The implant uses an off-the-shelf However, electrode array approved by the FDA says Burdick, for treating back pain. Meanwhile, “an amaz- Tai has been developing a more ing fact is that advanced device from scratch us- after spinal-cord ing microelectrodes instead of the injury, there is standard-sized ones, hoping that circuitry in the lower putting more power into fewer neu- spinal cord that can rons will give better results. But due control standing and to tough FDA regulations that clas- stepping and that doesn’t sify all long-term electronic implants degenerate. So we have the at the highest level of risk, his device potential to exploit this circuitry, still has a number of years before it even years after the injury, by stimu- will be ready for human trials. lating the nerve pathways that are still intact.” Burdick, who is also a Think and Do back of the brain. Despite the lengthy mechanical engineer, has been help- A short walk away from Tai’s lab, explanation, this process happens so ing to develop a device for doing just neuroscientist Richard Andersen quickly that users are able detect light, that as part of a research team that (not to be confused with actor Rich- identify objects, and even perceive mo- includes neurobiologists from ard Anderson, of Six Million Dollar tion in real time. UCLA and neurosurgeons from the Man and Bionic Woman fame) “When I started working on retinal im- University of Louisville, where Sum- leads a research group focusing plants, we were working on a 16-elec- mers received his implant. The elec- on “the idea that paralyzed people trode device,” says Tai. “You can imagine trode array is implanted in the small or people with amputations could a 16-pixel camera—that’s not much of the patient’s back, in the same control a wheelchair or a robotic resolution for seeing the world. The lat- area where pregnant women get limb via a neural implant,” he says. est implant has 60 electrodes, and we epidural anesthesia during childbirth. “It’s kind of sci-fi sounding, but it are working on 256-electrode devices. “By basically beaming energy actually works.” In less than a year, we’ll be working on in there, we try to raise the level of Back in the late ’80s, Andersen, thousand-electrode devices.” excitability of the existing neurons,” an early pioneer of cortical-function These prostheses have been implant- explains Burdick. “This does at least studies, discovered that a part of the ed in less than 20 people so far, but two things, and probably a lot more: brain called the posterior parietal they’ve been very successful. “We think it replaces the descending input cortex (PPC) that was known to be the direction we are going is absolutely from the brain and tells the circuitry involved in spatial awareness also correct,” says Tai. “It’s not just a science to ‘go,’ and it excites certain parts of fed the intention of movement to story. We all think we will produce a the spinal cord that help process the the motor cortex. In other words, the device that will benefit mankind.” sensory information needed to tell PPC helps us plan how to pick up a Retinal implants aren’t the only such muscles how to do their job.” pen, or type a certain word on a key- hardware to have gone from the bench After two years of using the array, board. His lab, in collaboration with a to the bedside with the help of Caltech Summers can now stand, balance, company called Microprobe (whose faculty. Tai and bioengineer Joel Burdick and step when the device is turned main line of work is building testing have lent their talents to a neuropros- on. (According to the terms of the devices to listen in on electronic thesis in the spine of Rob Summers, a FDA’s testing protocol, the device components), has designed a micro- now-25-year-old college baseball star can only be used during designated electrode implant that eavesdrops on who was paralyzed by a hit-and-run exercise periods.) However, Sum- the PPC and decodes the subject’s driver in 2006. mers has also gained improved con- intent by using algorithms gleaned

22 ENGINEERING & SCIENCE winter 2012 from the lab’s research. With such an being grasped is hard or soft, whether it own parts out of position. Andersen implant, a severely paralyzed person is slipping through the hand or is firmly is also exploring the idea of making the could, in theory, move their wheelchair grasped, and how to manipulate the implants wireless. With no connections forward just by thinking about it. “The object with dexterity. Clinical trials are penetrating the skin, there’s no path initial application would be for typing planned in collaboration with Huntington for infection. or possibly operating an iPad or other Hospital, just three miles west “The challenge there is that you have tablet-type computer,” says Andersen. of campus. to broadcast a very large amount of “With all the tablet applications that information and you don’t want to heat exist today, one might be able to do G ray Matters the tissue up, so you need special elec- quite a lot without being able to move.” While neural devices have been shown tronics of very low power but very high His lab is also involved in a sprawl- to work well, their longevity remains a bandwidth,” he says. ing collaboration, directed by the challenge. It is particularly hard to keep Figuring out a general method for Defense Advanced Research Projects brain implants working for more than a tailoring the therapy to the patient is Agency and the Applied Physics Lab- year or so. also a high priority. “We have a large oratory at Johns Hopkins, that is at- “The brain is a lot like the ocean—it’s number of electrodes, and we can tempting to build a robotic arm—from very corrosive,” says Andersen, pointing change the voltage and frequency on shoulder to fingertips—that has all the out that finding and/or developing better every one,” Burdick says of the spinal- freedom of movement of the human materials will be part of improving the cord implant. For example, the array variety. His lab is working on two technology. placed in Summers has 16 electrodes, brain implants: one controls grasping, Tai and Burdick agree. “We are each of which stimulates a handful and the other, the movements of the constantly looking for materials that our of neurons. But the spinal cord is so hand as a whole. In addition, they are bodies like, or at least react neutrally to,” densely packed that only a few of the developing a feedback loop to the says Tai. neurons in contact with the array will brain so that activating sensors on the Beyond biocompatibility, the compo- belong to the correct circuit, and even robot’s fingertips will lead to electrical nents need to be flexible, so that they those few will generally require different stimulation of the somatosensory cor- can bend and stretch with the body. levels of stimulation. “We have an enor- tex, the part of the brain responsible In addition, the implants need to be mous number of parameters to work for the feeling of touch. This feedback engineered to fit within tiny spaces: with to figure out the best combination will give the hand more finesse by behind the eye, perhaps, or between of stimuli for each person.” telling the brain whether the object vertebrae—without moving the body’s All of these projects are collaborative efforts, both on a Caltech scale and a national scale. The research requires expertise from many fields—from the engineers and scientists who build the devices to the medical doctors who oversee the human trials and help trans- late the technology into general use. “Neural implants are complicated and involve so many different backgrounds— you cannot be a lone wolf,” says Tai.

Tai’s prototype spinal-cord implant has success- fully helped paralyzed rats to stand. Researchers hope to have a human-scale device ready for trials in a few years.

winter 2012 ENGINEERING & SCIENCE 23 Keeping an Electrical A Bionic Future? “There are interesting techni- The researchers look forward to seeing cal challenges as an engineer, and Eye on the Future their work move out of the laboratory really gratifying opportunities to see and into people’s lives. positive results with human patients,” Just upstairs from Yu-Chong Tai, Azita Emami Andersen envisions his communica- says Burdick. Adds Andersen, “The has a toyshop of her own where she, too, is tion prosthesis initially helping people medical component of these projects toiling away on the next generation of eye with severe paralysis or with conditions is incredible. You can actually see implants. A part of the Caltech-USC collabo- such as Lou Gehrig’s disease, but his something work, and show why it is ration that built the 60-pixel Argus II, her lab long-term goal is to enable amputees so important.” has recently developed a stimulator chip for to operate prosthetic limbs with the “A lot of my engineering research a future self-contained retinal prosthesis. instinctive ease of their natural ones. He ends in a paper, and I never know if “The main focus is to put the chip, and the also imagines endowing stroke patients or when it will become useful,” says electrodes and coils that Tai is making, into with the ability to rewire damaged parts Tai. “But with neural implants, I clearly an integrated package,” says Emami. “The of the brain in order to recover lost feel that my research will not be in ophthalmologist can put the whole device in functions. To that end, his lab is examin- vain. The experience is wonderful in the eye, and once it’s healed there is much ing whether targeted stimulation of the the sense that the project is never- less risk for infection.” healthy neural circuitry that remains can ending. We can always keep The chip would supplant the external, accelerate the relearning needed for improving, to make better devices that pocket-sized computer that currently process- brain repair. help more and more people. For es the video-camera images and sends the While Burdick and Tai are building those who can’t walk or see, we have digital information to the implanted electrodes. the next generation of smaller, more to be the marathon runners.” “For a person to read, or even have func- powerful spinal-cord implants, Burdick tional sight for everyday tasks, you need many, is anticipating wider clinical trials of the many points of stimulation in the retina—over first-generation one. Yu-Chong Tai is a professor of elec- 1,000 at the least,” she explains. “This requires “In five years, hopefully we will be in trical engineering and mechanical extremely low power consumption if you want clinical trials for one or two generations engineering. His work is funded to put everything inside the eye.” of newer technology for human use,” he by the National Science Found- This is where Emami and her graduate says. However, he is quick to point out ation and the National Institutes students come in. As masters of microelec- that this is not a cure for paralysis. “We of Health (NIH). tronics, they have devised a tiny, tiny nanochip call it a therapy, intended to improve life capable of delivering more efficient stimulation quality, but not to fully restore locomo- Joel Burdick is a professor of from far less power than the current model. tion at this point,” explains Burdick. “In mechanical engineering and The chip, which gets its juice from an the long run, a biologic approach will bioengineering. His work is funded inductance coil, uses less than one-tenth as be the right solution, whether it’s stem by the NIH, the Telemedicine and Advanced Technology Research much power as the Argus II, even though it cells, genetic engineering, or tissue Center, and the Christopher and powers nearly 20 times as many electrodes, implants. But even then, it’s naïve to Dana Reeve Foundation. says Emami. “The power consumption has think that someone who’s just experi- been reduced to levels that we think will work enced a major injury will go in to the Richard Andersen is the James G. inside the eye.” operating room, have some of their Boswell Professor of Neuroscience. The team hopes to test the entire intraocular spinal cord snipped out, receive an His work is funded by the NIH, prosthesis within the next two years. —KN injection of some stem cells, and get the Boswell Foundation, the Gordon up off the table and walk away. So and Betty Moore Foundation, and the we think our strategy will be useful in Azita Emami-Neyestanak is an assistant Swartz Foundation; the robotic limb rehabilitation, even when biological professor of electrical engineering. project is funded by the Defense Ad- solutions are in place.” Her work is supported by the National vanced Research Projects Agency. For all three men, helping people Science Foundation. overcome disabilities is a big part of what pushes them forward.

24 ENGINEERING & SCIENCE winter 2012 Robots in Disguise

Joel Burdick likes to call himself an “old robot guy,” can lower the cost so people can continue therapy even though he’s not old and his extensive background over a long period of time, it’s a win for everybody. Rob in robotics is still being put to good use. In fact, when is continuing to improve after 18 months of rehab, so he’s not busy working on computational models for clearly there is a benefit to extensive TP .” neural implants, he and a team of undergrads from his Burdick built his first stand frame out of plastic robotics lab are building physical-therapy equipment pipes in his garage this spring. The next generation will for Rob Summers, the first recipient of the spinal-cord be built out of aluminum and may include an ellipti- implant Burdick helped design. cal trainer, so that Summers can practice his walking “I’m building Rob a series of increasingly sophisti- motions. Burdick also plans to go a few steps beyond, cated devices that he can use in his apartment,” says building a frame with sensors on it. One set of sensors Burdick. “It’s a prototype phase, so that if our therapy would measure how much force Summers puts on continues to be successful in more patients, we’ll have the frame, while a set under his feet would determine them available.” how much weight is being supported by his body. In Burdick’s team is developing a home version of the addition, a set of motion sensors would track his whole “stand frame,” which supports a patient in an upright body in real time to document his training sessions position as the person learns to stand and walk again. and monitor his progress. All this information could “This device is often used in spinal-cord injuries,” says then be emailed to his doctors and therapists after Burdick. “But a home version needs to be more auto- each session. mated, to ensure safety without the help of a techni- “If you can have quantitative information from the cian. It has different straps and adjustments to stabilize robotic devices about how the PT is working, it’s useful Rob so that he can manage all the parameters of his for the therapists, doctors, and insurance companies, training by himself. but also for the scientists so we can test hypotheses,” “Physical therapy is expensive, and insurance says Burdick. “Plus I’m an old robot guy, so it keeps companies only pay so much,” Burdick explains. “If you me excited.” —KN

Rob Summers works out in a standard physical- therapy-type stand frame at the University of Louisville’s Human Locomotion Research Center at the Frazier Rehab Institute as a camera crew from ESPN shoots a story about his progress. Summers is talking to Louisville professors Claudia Angeli and Susan Harkema (in red), who are at the computer consoles, while Caltech’s Joel Burdick (to Harkema’s right) looks on.

winter 2012 ENGINEERING & SCIENCE 25

Programming M olecular Apps By Marcus Y. Woo By learning how to program molecules to do everything from assembling a nanorobot to fighting cancer, we may be embarking on the next technological revolution.

Computers haven’t That next step? Molecular program- From Binary to BASIC taken over—yet. There ming. Instead of telling electrons how Fortunately, a powerful computing are no Hals or homi- to call Mom, brew an espresso, or solve language already exists in the form of cidal cyborgs on Har- a complex equation, these researchers DNA molecules, which encode all the leys, but computers hope to tell molecules how to diagnose information any organism needs to de- still pervade every facet of our daily lives. diabetes, assemble into a nanobot, or at- velop, grow, and reproduce, whether it’s As adding machines and vacuum tubes tack a cancer cell. Such molecules might a bacterium, an elephant, or a towering gave way to electronic transistors in the not only seek out the cancerous cell, but redwood. This information is spelled out 1940s and 1950s, and as room-sized based on their evaluation of the cell type, by sequences of four chemical “bases” computers gave way to laptops several its environment, and the cancer’s state (commonly called A, T, C, and G) that act decades later, computer technology has of progression, they would release the as the letters of the DNA alphabet. The been evolving faster than anyone ever appropriate drug at the proper dose letters follow strict rules—A is paired with anticipated. From cell phones to coffee and time. T, and C with G—so the sequence of let- makers, nearly every electronic device is “It’s hard to tell where things are going ters in a strand determines the sequence now equipped with a silicon chip. to go,” says engineer Richard Murray of letters in the strand that will bind to it. These chips not only allow people (BS ’85), who along with Rothemund is Couple this intrinsic logic to the ability to to program their latest smart phone, part of Caltech’s Molecular Programming write any sequence of letters you want they can embed autonomous decision- Project, or MPP. “But I suspect we’ll into a DNA strand—now a routine part of making—some might even say intel- use molecular programming the way we bioengineering—and you’re on your way ligence—into what’s otherwise a dumb now think of electronics.” Like computer to writing a molecular program. machine. “When you attach a computer programming, molecular programming DNA and its cousin, RNA, are easily to something, it becomes much more is an engineering endeavor, he adds. If made in labs and are integral to biology. powerful,” says computer scientist Paul the field advances as rapidly as Mur- If you want to learn how proteins func- Rothemund (BS ’94). But while the ray, Rothemund, and their colleagues tion, or inject molecules into the body to computer revolution continues, Rothe- hope it will, in a few decades molecular combat cancer, then it only makes sense mund is among a group of scientists and programming could be as ubiquitous as to consider DNA. engineers who think we’re on the verge the electronic kind, changing not just The potential for molecular pro- of yet another revolution. how we live, but how we understand the gramming, however, goes far beyond world and life itself.

Left: City of Life, by Ann Erpino (http://www.annerpino.com/). Copyright 2006. Reprinted with permission.

winter 2012 ENGINEERING & SCIENCE 27 To the left: This map of North and South America, made using DNA origami, is only about 100 nanometers wide. The image was taken using an atomic-force microscope, which works by dragging a needle across the object, measuring the bumps of every atom. To the right: Woo and Rothemund have developed a jigsaw-puzzle-like way to assemble DNA tiles. Here, four tiles bearing the raised letters A, B, C, and D bind only if the shapes of their edges match. Top: a diagram of the scheme. Bottom: an atomic-force microscope image of the actual tiles.

we’re trying to think in terms of informa- lent of machine language—developing tion science,” Murray explains. “What the first low-level assembly languages are the different levels of abstraction and the compilers that read them—that medicine. we can use to describe the system?” will enable the next generation of mo- Consider a plastics For example, a specific combination of lecular software developers to write the factory where the ingredients DNA strands that behaves in a certain equivalent of BASIC and FORTRAN. themselves control the manufacturing way can be thought of as an indepen- The MPP “is a quintessential engi- process through built-in feedback loops, dent component—a black box with a neering activity, in that we’re trying to rather than relying on humans (or even well-defined function. Engineers can understand how you design things in computers) to mix the right amounts of mix and match these boxes to cre- a systematic way,” says Murray, whose chemicals A and B to get the maximum ate larger, more complex components expertise is in feedback and control production of C. Programmable mol- that will eventually lead to a level of systems. Since feedback is also crucial ecules could assemble themselves into abstraction on a par with the high-level to biology—it’s how your body regulates entirely new kinds of composite materi- programming languages we now have your blood-sugar levels, for instance— als, or even complicated structures. in computer science. “If you sit down he’s applying control-system principles Maybe one day our cell phones (or what- and use Microsoft Word, you don’t to molecular programming. In fact, Eric ever communication devices we’ll have have to think about what the individual Klavins of the University of Washing- in the future) will be grown, molecule transistors are doing,” Murray says. “You ton, another member of the MPP and by molecule. Perhaps more fundamen- think on the level of writing a macro to a former postdoc of Murray’s, has built tally, molecular programming is already do something you want to do.” networks of genes that behave in a providing tools for studying biology like The key is to be able to program mol- homeostatic manner. In other words, never before. ecules without having to wade knee- just as the amount of glucose in your The MPP researchers envision a deep into the intricacies of molecular blood remains within certain limits, world where molecular programming is biology. Writing out the strings of bases regardless of whether you’re lifting accessible to all regardless of technical needed for individual DNA strands to weights or watching TV, these networks training. Just as anyone can teach them- perform specific tasks is the equivalent adjust the production rate of a given selves how to write software or design a of hand-coding applets in binary. The substance depending on how much web page, someone in the future might MPP is creating the molecular equiva- of it is being consumed. just as easily write a molecular program. “In a hundred years, someone who has no clue about biology might make a huge contribution to medicine by writing something on the level of an iPhone app,” says information scientist Shuki Bruck, another member of the MPP team. “The difference between the MPP and what other people are doing with nano- technology and biotechnology is that

Each sheet of DNA origami “paper” is a closed loop of DNA called a plasmid. Any two-dimen- sional shape you like can be made by folding the loop back and forth on itself.

28 ENGINEERING & SCIENCE winter 2012 Woo and Rothemund, Nature Chemistry, vol. 3, pp. 620–627. Published online July 10, 2011. Copyright © 2011, Nature Publishing Group.

From Origami to Nanobots so-called stacking interaction—in fact, ged edges: all of the logs had sticky ends, Paul Rothemund is perhaps best known this interaction appears to be the main but the logs themselves were of varying for his tiny smiley faces. In 2006, he force holding the ladder together. lengths. And behold, properly matched pioneered DNA origami—a method for If you put a hinge between two edges clicked into place like the pieces of folding strands of DNA into any two- rungs and fold the ladder in half, the a jigsaw puzzle. dimensional shape, including smiley rungs next to the hinge will be “hang- This particular set of jigsaw tiles had 16 faces just 100 nanometers across and 2 ing in the breeze, without a partner possible edge shapes; 16-bit binary tiles nanometers thick. (He has also made a to stick to on one side,” says Rothe- would have thousands of easily distinguish- DNA map of North and South America, mund. This blunt end, as it’s called, able edge patterns. But merely having 16 spelled out the letters “DNA” in DNA, will readily stick to any other blunt possible edges to play with would enable and formed DNA snowflakes.) Each end. “You can visualize the blunt end people to make much more complex shape consists of a long strand of DNA as the flat, sawed-off end of a log,” devices than can be made through origami, folded at intervals so that it runs back Rothemund continues, noting that the including simple logic circuits rather than and forth in a series of parallel line seg- origami blocks—or, more accurately, the single transistors that have been made ments that fill in the shape.T he lines are two-dimensional tiles—look like tiny so far. held in place by “staples,” about 200 of log-cabin walls. By including logs that In addition to enabling larger structures, them—short strands of DNA that bind ended in floppy, nonadhesive loops of these techniques could ease us past a to the long strand as it folds, guiding DNA as well as sticky blunt ends, the sticking point on the way to a full-on, sci- the adjoining segments into their proper researchers created a binary system: fi, self-assembling nanobot: the problem positions. Researchers around the world “0” for non-sticky and “1” for sticky. of moving parts. “A human-scale analogy now use DNA origami to build everything Woo and Rothemund demonstrated would be to take all the parts for a car, from tiny boxes to nanoscale transistors. that these tiles bound preferentially paint them with glue, throw them in a bag, Just as the wingspan of a folded to other tiles whose edges encoded shake it, and have a working car pop out,” crane is limited by the size of the sheet the same binary sequence, mean- Rothemund says. But DNA base-pair of paper, the size of a single piece of ing that such tiles could be strung binding makes for a very powerful molecu- DNA origami is limited by the length of together in any order one might care lar adhesive—your car’s motor wouldn’t the strand being folded. Strands longer to program. “A lot of people, including run and the wheels wouldn’t turn. Even the than the ones Rothemund is using have us, have made DNA shapes with blunt wipers would be stuck to the windshield. proven hard to come by, so over the last ends that cause the shapes to stick “We think that we will be able to design two years, Rothemund and grad student together into random clumps—piles stacking bonds in which the parts of a Sungwook Woo have worked out new of junk,” Rothemund says. “But if you nanomachine will be able to self-assemble techniques for assembling origami build- are careful, you can harness the power and then slide freely past each other,” he ing blocks into larger structures. The of blunt-end stacking interactions— continues. “The parts won’t look exactly standard high-school biology view of a converting something that used to be like interlocking log walls, but they will DNA molecule is as a twisted ladder, considered a ‘bug’ into a ‘feature.’” work on the same basic principle.” with the rungs consisting of pairs of The binary-sequence tiles were bases that cling to each other. How- rectangular, with straight edges. But From Nanobots to Cell Phones ever, each rung also sticks strongly to Woo and Rothemund also created a In order to make these parts self-assemble, the rungs above and below it through a set of shape-recognizing tiles with jag- however, each one has to know exactly

winter 2012 ENGINEERING & SCIENCE 29 a* b*

c* b* a* b b* c* b* a* bb* + = + = cc* b* b bb* I

a a* a H1 c b* b H2

Spring-loaded hairpins Initiator Chain reaction begins Spring-loaded hairpins a* a

Chain reaction propagates

H1 H1 H1 • • • where it’s going and how it fits into the internal structure extends all the way are building DNA circuits, replacing H2 H2 H2 grand design. And this is a problem: down to the arrangement of the indi- handfuls of transistors with test tubes say you had several thousand unique vidual proteinAmplification molecules polymer in its cells. In full of molecules. In a digital logic gate, shapes at your disposal—that’s still not other words, a chunk of whale has a electrons either flow or they don’t.I n a enough to orchestrate the spontaneous lot more going on than an equal-sized DNA-based logic gate, the DNA strands coalescence of a piece of human-scale chunk of submarine. either bind or they don’t. Early this year, technology. Another advantage of molecular Winfree and postdoc Lulu Qian created, “If we wanted to self-assemble a cell self-assembly is the potential for cheap from scratch, the largest and most com- phone, origami won’t do it,” says Erik manufacturing. It costs billions of dol- plex DNA circuit ever made. It used a set Winfree (PhD ’98), a computer scientist lars to build a factory that makes sili- of standardized components—a crucial and director of the MPP. “But algorith- con chips. “Biology has always been requirement for developing higher-level mic processes could.” For the last 15 the opposite of that,” Winfree says. “If molecular programming languages, years, first as a grad student and now you have a few seeds, they’ll just grow. as well as for scaling up the circuits as a faculty member, Winfree has been Mold will grow in your refrigerator even themselves. working on ways to embed algorithms, though you didn’t want it to.” DNA This circuit consists of 74 kinds of the abstract ideas at the heart of self-assembly mimics how biological DNA molecules, and can calculate the computer programs, into the DNA itself. organisms grow, cell by cell, protein square root of any integer up to 15— “You design a set of molecules that fit by protein, and as a result, it’s much or, for the technically inclined, any four- together according to a certain logic, cheaper than conventional manufac- bit number. (The circuit does round and by controlling which molecule fits turing. Using molecular programming the result down to the nearest whole at which location, you can program the to “grow” a cell phone would be more number, however, as dealing with a growth of a whole structure,” he explains. akin to kitchen chemistry than to an decimal point is a bit beyond its capacity Then all you need to do is pour the expensive, factory-based process, he at the moment.) molecules into a beaker, stir briskly, and says. (He also adds that we’re a long The original dream for DNA comput- voilà! The algorithm executes itself. If you way from growing anything as big and ing was to solve big, complex problems. are clever enough writing the rules, you complicated as a cell phone.) “Our “That hasn’t panned out,” Winfree says. can create very complex structures with lab,” Winfree says, “is very simple. DNA computing just isn’t efficient just a few different molecular bricks. Most of our procedures involve order- enough; calculating that square root But why work molecule by molecule? ing a few DNA strands, mixing them took 10 hours. “Nevertheless, a tiny “To build things that are more struc- together, and letting the molecules do bit of computing goes a long way in tured than current technology allows,” the hard work.” the molecular world,” remarks Rothe- Winfree says. “If you can build things out mund. “Molecular computers, no matter of bricks that are a few nanometers in From Cell Phones to the Brain how simple, can be used to control size, you can do a lot more in the same “In a really fundamental sense, algo- other molecular phenomena.” No silicon space.” Compare a whale and a sub- rithmic self-assembly is a form of com- computer has this power, but it’s DNA’s marine, he explains. Both are about the putation,” says Rothemund. “In fact, it natural role. “The whole process of same size, and both propel themselves is far more powerful than circuits.” But embryonic development is controlled underwater. A submarine is made up of these hardworking molecules can also by a molecular computer performing relatively large pieces—steel panels and be made to “compute” in the tradition- logical operations—‘If X then Y, but only pipes, screws and bolts. But a whale’s al fashion: Winfree and his colleagues if A hasn’t happened.’ And even though

30 ENGINEERING & SCIENCE winter 2012 A single strand of RNA can fold onto itself to form a hairpin molecule. A fluorescent molecule, labeled H1, is attached to such a hairpin. When an RNA initiator comes along, it binds to the hairpin and pops it open. The resulting molecule attaches to yet another hairpin, which carries another fluorescent molecule (H2). The chain reaction continues, producing a long molecule with many glowing attachments to form a really bright marker.

it doesn’t solve anything we’d recognize From the Brain to the molecular instruments—tools he hopes as a computationally difficult problem, Organism will revolutionize biological research. its computation serves to make a very While his colleagues were concerned “There are profound questions hanging complicated object.” with computational power, bioengi- over the heads of biologists about how Still, if doing sixth-grade math isn’t neer Niles Pierce was all about motive development works, about how dis- enough for you, Qian, Winfree, and Shuki power. In 2004, he and grad student eases work,” he says. “Biologists will be Bruck have created the first-ever DNA Robert Dirks (PhD ’05) came up with a able to make faster progress with more circuit that has brainlike behavior. That’s way to make DNA “fuel” in the form of powerful experimental techniques.” right—using the same methods em- spring-loaded hairpins of single-strand- Answering many of these questions ployed to design the square-root calcula- ed DNA that could pop open on cue. requires finding out when and where tor, they have made a neural network that Four years later, Pierce and postdoc genes are switched on in different cells. plays a mind-reading game. To play, you Peng Yin demonstrated that a prop- For the last 40 years, biologists have first think of a scientist.T hen you answer erly scripted set of cues can nudge been using a method called in situ one or more of four previously defined molecular machines into performing a hybridization, or ISH for short, to pin- yes-or-no questions—for example, “Was surprising variety of feats. One of their point the locations of messenger RNA the scientist British?”—by dropping the creations was a molecular “tree” that (mRNA) molecules that serve as prox- DNA strands corresponding to those grew dendritically from seed to leaf; an- ies for their activated genes. (RNA, the answers into the test tube. Now you other was a DNA “walker” that strolled chemical first cousin to DNA, encodes shake the test tube vigorously, and if the along a DNA track. “In theory, DNA information in an almost identical set of facts you provided match any of the four motors could provide us a different way bases, and mRNAs deliver the DNA’s scientists programmed into the network’s to build things,” says Pierce. “Rather protein-making instructions to the cell’s memory, the test tube lights up with a than just letting things stick together, if protein factories, the ribosomes.) After color-coded fluorescent signal.T he net- pieces can be moved around actively, flooding a thinly sliced tissue sample work also tells you if it doesn’t know the by DNA walkers, it may be possible or a Petri dish full of cells with an RNA answer. It even tells you why it doesn’t to build complex objects much more probe designed to bind to the mRNA of know: whether the scientist you picked is quickly and efficiently.” interest, the unbound probe molecules not in its memory, or whether you didn’t Meanwhile, Pierce has begun apply- are rinsed away. A fluorescent “tag” give it enough clues for it to narrow its ing what he had learned from engineer- on the probe then lets you image the choice to just one person. ing molecular machinery to developing mRNA targets under a microscope. Taking an incomplete pattern and figuring out what it might mean has been one of the hallmarks of a living brain. Maybe it’s not just silicon chips that we have to worry about becoming self-aware and taking over the world. We might have to watch out for those molecules, too.

This image shows the fluorescent amplification technique being used in a zebrafish embryo, illuminating five different kinds of mRNA with five different colors.

winter 2012 ENGINEERING & SCIENCE 31 To the right: A “smart drug” might one day consist of a molecular robot that recognizes malignant cells by their surface proteins. After docking with the marker protein, the robot could crawl along the cell’s membrane, slicing it open and destroying the cell.

But when you try ISH on a vertebrate method is catching on elsewhere—the that life is driven by the programming embryo, which are the ones of most Pierce lab is providing probes and power of DNA. “The universe just hap- interest to scientists studying human hairpins, as well as technical support, pens to be that way,” he says. “Biology development and diseases, you need to biologists around the world. has exploited that inherent essence to boost the fluorescence in order to While the researchers are continu- of nature to do what biology wants to see it. Biologists add an enzyme that ing to enhance this technique—for do: to reproduce, to evolve, to build deposits extra tag molecules near the example, figuring out how to zoom in really complex animals, and to build probe, amplifying the fluorescence. To and achieve molecule-scale resolution brains.” The ambitions behind molecular see two mRNAs at once, you repeat in order to map mRNA locations quan- programming go beyond a descriptive the process with the same enzyme and titatively—they’re also pondering other understanding of biology, as research- a different dye. types of molecular instruments. One ers strive for deeper insight into what This is a big problem, Pierce says. possibility, Pierce says, is to design life is at its most fundamental level. “It’s cumbersome, and an impediment tools that would turn gene B on (or “Technological developments have to studying genetic circuits, where off) depending on whether gene A is historically led to new concepts, and biologists want to look at many genes already on or off. “This would provide the languages needed to express interacting with each other.” With the unprecedented tools for studying them,” says Winfree. “And these new standard approach—first dyeing one genetic circuits at specific times and languages change how we look at the mRNA red, say, and then another one locations within developing embryos,” world and reason about it.” green—it takes about five days to do he says. Being able to program con- Just as miners and engineers tinker- three colors. ditional gene activation (or silencing) ing with pumps led to the develop- But if the fluorescent amplifiers were would also have medical potential, ment of the steam engine, which in programmable, they could operate he adds. turn led to the discovery of the laws of independently, seeking out their various “These are all dreams right now,” thermodynamics and eventually to the mRNAs at the same time. Pierce real- Pierce says. But he hopes that in the development of statistical mechanics— ized that the spring-loaded hairpins that next 10 to 15 years molecular instru- which is now used to analyze ev- powered the walker could run an ampli- ments will become indispensible for erything from galactic evolution to fier as well. With each type of hairpin research. “The possibilities,” he says, stock-market rallies—the languages carrying a different dye molecule, the “are essentially endless.” that might spring from molecular pro- only limit to the number of colors would gramming could give us the conceptual be the number of dyes that can be dis- From the Organism Back tools needed to think about complex tinguished through the microscope. to Basics biological systems in a whole new way. After five years of hard work, Pierce Although the MPP is informed by the The Industrial Revolution and the and his coworkers, including grad stu- computer revolution, it’s rooted in as- Computer Age came about by accident dent Harry Choi (PhD ’10) and biology pirations to understand how life works. and happenstance, but Rothemund professor Scott Fraser, successfully “Understanding biological systems is hopes that the Molecular Era will occur tested fluorescentI SH, or FISH—on, the most important challenge for the by design. “The fact that we have appropriately enough, zebrafish embry- next 100 years,” says Bruck, whose started to recognize the features of os—by targeting five species of mRNA background is electrical engineering, such revolutions gives society tolerance simultaneously. The method passed but who now focuses his research on for us to play around and see if we can with, well, flying colors. T“ hese amplifi- computing with biological circuits. “If build another one,” he says. “This kind ers, which started as a proof-of-principle you compare our world to anything of forethought is, I think, a hallmark of exercise in molecular programming, are else in the universe, based on what we our age. Research grants and start-up now a research tool at Caltech,” Pierce know so far, we have life here and not companies have regularized, ritualized, says; while Fraser continues with the anywhere else. That’s the most pre- and mechanized innovation. The MPP, zebrafish studies, biologists Marianne cious thing we have here, and we still for example, is supported by a very Bronner, Dianne Newman, and Eric don’t understand it.” forward-looking program run by the Davidson are using the technique to For a computer scientist like Win- National Science Foundation called study genetic circuits in birds, bacteria, free, the MPP is about the idea that Expeditions in Computing. It’s really and sea urchins respectively. And the information is the essence of nature, amazing to me that we can come up

32 ENGINEERING & SCIENCE winter 2012 Process of Elimination, by Ann Erpino (http://www.annerpino.com/). Copyright 2007. Reprinted with permission. with a futuristic, almost science-fiction the Moore Professor of Computation supported by the National Institutes of vision, organize around it, and have and Neural Systems and Electrical Health, the International Human Fron- society buy into it.” Engineering. Erik Winfree (PhD ’98) tier Science Program, the Ralph M. is a professor of computer science, Parsons Foundation, the Charles Lee computation and neural systems, Powell Foundation, the Semiconduc- Paul Rothemund (BS ’94) is a senior and bioengineering. Niles Pierce is a tor Research Corporation, Microsoft research associate in bioengineering, professor of applied and computational Corporation, Nanorex Corporation, the computer science, and computational mathematics and bioengineering and Benjamin M. Rosen Family Founda- and neural systems. Richard Murray (BS executive officer for bioengineering. tion, the Caltech Center for Biological ’85) is the Everhart Professor of Control Circuit Design, the Beckman Institute and Dynamical Systems and Bioen- Besides the National Science Founda- at Caltech, and the Gates Grubstake gineering. Jehoshua “Shuki” Bruck is tion’s Expeditions in Computing Pro- Foundation at Caltech. gram, various parts of this work were

winter 2012 ENGINEERING & SCIENCE 33 Naturally Inspired By Kimm Fesenmaier

In the Charyk Lab of Bioinspired Design, even an embry- onic zebrafish heart can be an engineering muse. Here, Caltech engineers continue the time-honored tradition of teasing apart nature’s tricks and borrowing the best bits.

When Mory Gharib (PhD was more to its tiny pulsing heart than The striped, pinkie-sized zebrafish ’83) studies zebrafish, he meets the eye. has become a mainstay for develop- sees straight to the heart of Since then they’ve picked apart the mental biologists—it was one of the the matter. Literally. Since properties of the fish’s embryonic heart, first animals to have its entire genome zebrafish embryos are and are now applying what they’ve sequenced; it reproduces in large almost transparent, it’s learned to attack problems as diverse numbers and has a short life cycle; and quite possible to peer right through as ringing in the ears and overheated since its see-through embryo devel- them. But when Gharib and his group electronics. They’ve even developed the ops outside the mother, the changes started studying Danio rerio about a first pump built entirely from biological it undergoes as it matures are easy to decade ago, they had a feeling there building blocks. follow. Gharib was introduced to this

34 ENGINEERING & SCIENCE winter 2012 An unlikely muse: The embryonic zebrafish heart is little more than a tiny tube of cells, but it gets the job done. This three-dimensional reconstruction shows fluorescently labeled heart-muscle cells and traces their movements over the course of two cardiac cycles. The Y-shaped scale bar indicates 20 microns in each direction.

at both ends to less flex- duty, without any degradation of ible tubes—even tubes performance. made of the same mate- The impedance pump’s elegant rial will suffice, as long simplicity is seductive. Clearly, it can as the walls are thick operate on a very small scale—nature enough to make them only needs a handful of cells to make substantially stiffer. The one. “With something like peristalsis, A. Forouhar, et. al., Science, vol. 312, pp. 751–753. key is to make the transi- you need to coordinate the motion © 2006, American Association for the Advancement of Science. tion abrupt, because of many different cells all contracting little marvel by biologist Scott Fraser, the point where the wall suddenly in succession. But with an imped- director of the Rosen Bioengineering becomes rigid acts as a reflector, ance pump, all you need is a simple Center, whose confocal microscopy bouncing the pressure wave back band that’s continually beating at one lab Gahrib uses. “We now have a very into the flexible section of the tube. frequency,” says Derek Rinderknecht fruitful collaboration between the two By compressing the flexible section (PhD ’08), senior research scientist groups,” Gharib says. in just the right (off-center) location in aerospace and a member of the Gharib is an aeronautical engineer, at just the right rhythm, the reflected Gharib lab. “This allows you to have not a developmental biologist, so he’s and newly generated waves interfere a very simple way of moving fluid interested in species like the zebrafish constructively to create a pressure around at a very early stage in the for other reasons. He likes to steal their gradient that drives a flow through fish’s development.” tricks, one-up them by enhancing what the system. The pumps are also astonish- Mother Nature has accomplished, and Though the pump is simple, its fluid ingly efficient. T“ aking advantage of find new applications where the tricks dynamics are complex. Thus the resonance allows you to have a much could come in handy. researchers in Gharib’s lab built higher output for any given input,” “We can actually be more clever scaled-up pumps in order to tinker Rinderknecht says. “That means less than nature,” Gharib says. “We can get with their parameters and to discover energy is required.” inspired by nature and use engineering the ins and outs of their function. The Armed with a solid understanding to come up with better functions. first mechanical pumps were centime- of the physics at play, the researchers Just look at 747s—they fly from LAX ters in diameter—small, but still a hun- were ready to put their new tricks to to La Guardia much more efficiently dred times larger than the zebrafish’s work. “We knew we had a ‘platform than any bird could.” That’s why Gharib tiny heart tube. Since then, they’ve technology’ that could hold value for runs the Charyk Laboratory for Bioin- been scaled back down. Some models many different areas of medicine,” spired Design. were simple tubes, made with a two- Gharib says. “These pumps could be In the case of the zebrafish, the centimeter length of latex tubing and useful pretty much anywhere you need group homed in on a very early stage a couple of glass capillaries. Others localized, controlled delivery.” of the species’ development, when its were more sophisticated, constructed The inner ear, for example. A tiny, heart is little more than a short tube of from micromachined metals or silicone implantable pump could deliver a cells. By studying the patterns of flow and built to lie flat on a tabletop. created in and by the tube, Gharib Since impedance pumps don’t have and his team determined that the ac- valves, there are very few moving parts tion wasn’t driven by peristalsis, as to break down. Some of the group’s had widely been assumed. Peristalsis creations have managed to go through uses a traveling wave of contraction to more than 200 million cycles, which move the fluid along—think squeezing is to say two weeks of continuous a toothpaste tube from the bottom up to push the contents onto your brush. Instead, they found a much simpler mechanism, first described in 1954 by German cardiologist Gerhart Liebau A prototype impedance pump for treating and which the Gharib lab has dubbed “the impedance pump.” chronic ringing in the ears nestles in All you need for an impedance pump Rinderknecht’s gloved palm. is a flexible, fluid-filled tube connected

winter 2012 ENGINEERING & SCIENCE 35 Right: In these frames from a video, water droplets do skateboarding tricks on a carbon- nanotube-coated surface. Left: Grad student Hesham Azizgolshani watches the workings of a pump he made from biological building blocks.

setting.” The prototype version has just been completed, and now awaits testing. The Gharib group has also found nonmedical applications. In the hot field of electronic cooling, for example, many researchers are look- ing for low-power ways to dissipate or redistribute the heat generated by the processors, circuits, and other components of our ever-shrinking electronics. The lab is now develop- ing a circuit-board overlay, thin as a credit card, that would circulate a fluid over the board’s electronic hot spots, carrying heat away to cooler areas. “For electronic cooling, the un- steady flow that these pumps create could be a good thing,” Rinderknecht steady, long-term trickle of medicine no cure. Part of the problem lies in says. “It can be used to move heat to help treat a variety of hearing the fact that tinnitus can spring from more uniformly into the fluid. It’s disorders, and the Gharib team has several causes. Some cases appear almost like sloshing the contents of a partnered with pharmaceutical giant to originate in the inner ear; others in cup in order to mix them up.” Sanofi to develop just such a device. the nerves and brain regions associ- While these ventures may be ef- The pilot version is designed to ated with it. forts to one-up Mother Nature, anoth- alleviate ringing in the ears. Chronic Rinderknecht is building an imped- er project could be described as an tinnitus, to use the medical term, ance pump that could be implanted attempt to tune up Mother Nature—to affects at least 20 million Americans, in the mastoid bone, close to which- fix something biological that isn’t according to the National Institutes of ever tissues are affected. “The tinni- working properly. Toward that end, Health. It’s also becoming a com- tus project has been a good chance grad student Hesham Azizgolshani mon disability in the military, where to go from a pump that works on a has actually built a “living” impedance loud explosions in close proximity are bench top to an implantable, proof- pump using heart cells and intestinal unfortunately a part of life. Drugs, de- of-concept device,” Rinderknecht tissue from rats. He strips the intes- vices to mask the “sound” of tinnitus, says. “That’s where we’re heading tinal cells from the tissue, leaving a and neural stimulation have all been right now. We’re starting to lay the tubular scaffold that he repopulates tried as remedies, with varying levels groundwork for eventually, hopefully, with heart-muscle cells. of success, but there is currently moving into some sort of clinical

Right: An electron-microscope image of a carbon-nanotube array. Each tube is about 40-millionths of a meter long. Far right: Manipulating the nanotubes’ degree of hydrophobicity can create gradients across the array that will drive a fluid sample placed in the cen- ter into different wells for analysis. This chip is called “the Zen garden.”

Image by Michael Bronikowski (BS ’86), reprinted from Carbon 44: 2822-2832 (2006) © 2006, Elsevier Ltd.

36 ENGINEERING & SCIENCE winter 2012 “We wanted to show that it’s pos- tor for two weeks. But eventually, SUPER-HYDRO-PHOBIC- sible to make a pump fully out of bio- durable, long-lasting pumps made logical materials,” Azizgolshani says. from a patient’s own cells could be LOTUS-ACTING- He plays a piece of video captured implanted into human veins. This NANO-TUBULES by a high-speed camera mounted on would assist diabetics, for example, a stereomicroscope. Sure enough, who often suffer from poor circula- Zebrafish aren’t Gharib’s only muses. There’s also the pump is pulsing, driven by the tion, and who would benefit greatly the lotus leaf, which is exceptionally good at repel- compression of a ring of cells around from “booster pumps” that would ling water. Many of Gharib’s students are working the tube. But this isn’t enough, he help move blood back to the heart with carbon nanotubes, minuscule rolled-up graph- says. “I can’t just claim that I have from the peripheries. ite lattices that look, at the atomic level, like hollow made a pump. I have to verify and “Instead of trying to get the body bundles of molecular chicken wire. The superlatives characterize its function.” to accept our man-made materials, fly when scientists talk about these tubes—it’s not He starts a second video; this one which can easily be rejected, uncommon to hear them described as the strongest features silver-coated glass particles the goal is to use the biological material ever made, or the most promising. But the suspended in the fluid within the materials that are available to us,” descriptor grad student Adrianus Aria (MS ’08) has tube. The particles slosh back and Azizgolshani says. “The possibili- focused on is “most hydrophobic.” forth, so it’s hard to tell if they are ac- ties are limitless.” Some arrays of carbon nanotubes are more tually flowing. But when Azizgolshani water-repellant than others, and Aria has found out plays the video in fast forward, a net why. It’s all in how you grow them—if their surfaces left-to-right flow is apparent. Morteza “Mory” Gharib is get slightly oxidized, the oxygen atoms attract water “We started from a biological sys- the Liepmann Professor of molecules. So Aria developed a recipe for cooking tem, we learned from it, and after we Aeronautics and professor of bio- the arrays in a vacuum chamber to get rid of any felt we were confident and comfort- inspired engineering. He joined the stray oxygen atoms. The result? Superhydrophobic able in our understanding, we faculty at Caltech in 1992, and he nanotubes. decided it was time to try to make our currently serves as one of the Insti- Like the flower itself, the lotus effect lasts only a own biological pump,” Azizgolshani tute’s vice provosts. The Caltech- few days. But if Aria can figure out a way to make says. “I think one of the beauties Sanofi nanopump project is funded the effect more or less permanent, the potential of this project is that we are closing by a Sponsored Research Agree- applications are enormous. A ship with a nanotube- that circle.” ment under the Caltech-Sanofi coated hull, for example, would knife through the Currently, this pump is only a Alliance Framework. The nanotube water with almost no drag at all, consuming far less proof of concept. In his best effort work is supported by the Charyk fuel and perhaps carrying much more cargo. so far, Azizgolshani has managed Foundation and the Fletcher Jones Aria can also make the nanotubes less hydro- to keep a pump going in a bioreac- Foundation. phobic, by treating them with ultraviolet light in the presence of air. “That means I can control the wet- ting properties,” he says. That could come in very handy if, for example, you want to package a water- soluble drug in a waterproof container in order to ensure delivery to the target organ. A video of water droplets bouncing off Aria’s nanotube arrays or rolling down their hydrophobic surfaces—during one segment, it’s hard not to envi- sion skateboard champs in a half-pipe—recently showed up on YouTube. Within about two months, it had garnered 400,000 hits. Gharib smiles as he plays it. “Sometimes we just like to have fun, too.”

winter 2012 ENGINEERING & SCIENCE 37 By Kathy Svitil

We’re a long way from transplantable organs grown to order, but Caltech chemists are developing some of the tools we will need to make that happen.

38 ENGINEERING & SCIENCE winter 2012 At any given tissues, and he offers no claims that it distributed— moment, more ever will: “It may be that in 100 years, which can be David Tirrell than 100,000 something we’re involved in now may as simple as stir- people in the lead to artificial tissue,” he cautions. ring the mixture—you add another protein United States What he does do is far more basic: to the gel to make it harden like Jell-O in a are awaiting organ transplants. Although He invents proteins made with amino refrigerator. thousands of transplants are performed acids not found in nature that function Tirrell, working with bioengineering grad each year, thousands of other people in ways that normal proteins do not. In student Alborz Mahdavi and and with H. die because there just aren’t enough these artificial proteins, as in natural Teresa Ku of the City of Hope, recently donated organs. One solution? Making ones, the sequence of amino acids spiked one of his ECMs with pancreatic organs, from scratch. Imagine a big within the molecule determines how it stem cells. Nestled within the matrix, the assembly line, churning out kidney after contorts itself into a three-dimensional cells flourished—and, Ku says, differenti- kidney after kidney. shape, and that shape in turn determines ated into endocrine cells, as hoped. “We Science fiction? Of course. But the protein’s function. But the sequence are planning to transplant the cell colonies through the development of synthetic of a protein also determines how it into mice in the near future to see whether tissuelike biomaterials, “artificial” pro- behaves en masse, when surrounded these cells will secrete insulin and correct teins with programmable properties, and by countless numbers of its fellows. diabetes in a mouse model,” she says. methods for the pinpoint placement of And that behavior matters if you’re The usual way to build a “simple” artifi- cells, Caltech chemists are inching us trying to build an artificial tissue, which cial organ starts with a prefab scaffolding closer to the elusive goal of made-to- needs to include not just the cells but constructed of a porous, biocompatible order organs. the scaffold on which they hang. That material such as a water-swollen gel—or In truth, artificial organs have been framework—constructed mainly out “hydrogel”—made from polyethylene in use for decades. Back in 1982, for of protein molecules—is called the glycol. (PEG, as it is known in the outside example, the first Jarvik-7 replaced the extracellular matrix, or ECM. Tirrell world, is a common ingredient in skin ailing heart of a Seattle dentist, who builds artificial ECMs to order, control- cream, shampoo, and toothpaste.) The gel lived for 112 days (albeit hooked up to ling their properties by monkeying is molded into the rough shape of the or- machines); artificial ears, a.k.a. cochlear with the genetic blueprints of their gan and seeded with cells. The challenge, implants, are now commonplace. But constituent proteins. of course, is steering all of the various cell the ersatz organs available today are far When you design these genes, he types to their proper locations—including from perfect substitutes—quite under- says, “you have to think not only about the cells that form blood vessels. Without standably, because human tissues are the protein itself but about the behavior blood flow, after all, any tissue will die. exceedingly complex, with a daunting of the material that results—is it stiff, is A method developed by grad students variety of strategically placed cells and it loose, what other properties does it Udi and Ophir Vermesh (both PhD ’11) a complicated infrastructure of nerves have?” For example, say you want to and their colleagues in the laboratory of and blood vessels. This architecture has, build an artificial tissue composed of chemist Jim Heath may meet this chal- so far, proven impossible to duplicate; liver cells, or from the insulin-producing lenge. By pressing a silicone template indeed, the bladder—essentially a bal- beta cells of the pancreas. An ECM— containing microfluidic channels against loon of soft, stretchy tissue—remains the artificial or otherwise—is an elastic, but a microscope slide, a gridwork of anchor only living lab-grown replacement organ solid, gel. When the ECM grows up with points is created on the glass. Each yet developed. the organ, getting the cells inside it is anchor can be tailored to adhere to one A crucial first step in building artificial not a problem because they are already specific cell type, allowing individual cells organs that are more lifelike is creating there, but otherwise “it’s hard to get to be placed in prescribed locations just a lifelike artificial tissues. At Caltech, such cells into an elastic solid,” Tirrell says. “If few millionths of a meter apart. “We then materials are the purview of chemist you design your gel so that it’s initially a encase the patterned cells in a hydrogel David Tirrell. Admittedly, Tirrell’s work is liquid, you can distribute the cells within matrix and stack these hydrogel sheets not focused explicitly on making such it.” Then, once the cells are properly to form 3-D tissue constructs,” explains

winter 2012 ENGINEERING & SCIENCE 39 Julia Kornfield

humans the alpha and beta cells are just sort of scattered willy-nilly, surrounded by blood vessels. And although random clumps of cells aren’t prohibitively dif- ficult to make, making blood vessels is another issue. To do so, you’d have to snake tunnels, constructed out of the appropriate cells, through your matrix, which means accurately creating and lin- ing up holes in the stacked 2-D sheets. “We’re still trying to figure that part out,” Vermesh admits. The experiment was merely intended to demonstrate the ability to control where specific cells went, so “we didn’t place much emphasis on incorporat- ing ECM proteins into our hydrogels,” Ophir, who is now in his fourth year of neat rows of insulin-secreting beta cells Vermesh says. “But the types of proteins medical school at UCLA as part of a alternating with equally neat rows of found in an ECM play an important role joint MD/PhD program. non-insulin-secreting alpha cells. These in keeping cells alive in real tissues. The team tested the process by build- cells were actually much better orga- So while these cells survived for a ing very rudimentary islets of Langer- nized than absolutely necessary: in rats time, I would expect they would need hans—as the insulin-producing parts of and mice, the beta cells form clusters those ECM proteins in their immediate the pancreas are known—by setting out encircled by the alpha cells, whereas in surroundings, especially in an implant- able device.” ECM proteins are also key to a different implantable material that Tirrell, chemist Robert Grubbs, chemical engineer Julia Kornfield (BS ’83, MS ’85), and collaborators at UC San Francisco have developed for healing damaged corneas. The cornea is the eye’s half-milli- meter-thick outer “skin.” It’s a complex sandwich whose filling contains three collagen-rich layers, where long fibrils of collagen are placed so precisely that the resulting structure is as transparent as a perfect piece of glass. If these fibrils’ orientation gets disrupted by a scratch or infection, the glass can turn cloudy, resulting in corneal blindness—some 2,000,000 cases worldwide. (Cataracts, the most common cause of blindness, are caused by cloudiness of the lens—the light- focusing crystal located deeper within

40 ENGINEERING & SCIENCE winter 2012 the eye.) “We want to understand how The idea is not far-fetched, she argues. David Tirrell is the Ross McCollum– we can guide wound healing so that it Because it has to be transparent, “the William H. Corcoran Professor and does not lead to scarring and blindness,” precision with which the cornea has professor of chemistry and chemical Kornfield says. to be built is extreme,” she says. “If we engineering. Research on artificial ECM Corneal scarring results from the understood that, and could keep that proteins has been funded by the National body’s overly enthusiastic effort to slap process from going wrong, it would be Institutes of Health and by Caltech’s a Band-Aid on a wound. Your system’s easier to promote healing without scar- Jacobs Institute for Molecular Engineering first responders at the scene of any injury ring in other tissues, including the skin.” for Medicine. are the constantly circulating platelets Surgical incisions could someday be in the bloodstream. They set up a triage closed without a trace—but also wounds Jim Heath is the Elizabeth W. Gilloon center by forming a makeshift ECM, caused by accidents, say, or even burns. Professor and professor of chemistry. The upon which the other incoming cells on the emergency team alight so they can fix things. Soon, fresh collagen fibers are being plastered down, mending the damaged area. “The healing response is Although random clumps of cells designed to close a wound as quickly as possible,” Kornfield says. I“ t doesn’t have aren’t prohibitively difficult to to be pretty, just get it closed.” And, indeed, the result is not pretty, make, making blood vessels is because the patch’s fibers lack the regi- mented organization of their neighbors. another issue. To do so, you’d have Instead, they are strewn haphazardly, like a microscopic crazy quilt. And instead of to snake tunnels, constructed transmitting light unscathed, the patch scatters it. The cornea takes on a milky out of the appropriate cells, hue, and vision is obscured. “What we need to do,” Kornfield says, through your matrix, which means “is tell the system how to come in and clean up the mess without making a accurately creating and lining up mess.” The way to do that, she and her colleagues have found, is with an implant holes in the stacked 2-D sheets. somewhat akin to a contact lens, but made of Tirrell’s specially crafted gel. In addition to the ECM proteins that form the gel itself, the implant contains This type of carefully orchestrated cell-patterning project was funded by the signaling molecules that slow the triage repair would likely take longer than the National Cancer Institute, the Ivy Founda- process and buy valuable time for the natural healing process, but, Kornfield tion, and the Grand Duchy of Luxembourg. body to lay down collagen fibers the right says, the results “would be perfect”—and way. “We need to get in there within a that, she adds, is worth the wait. Nobel Laureate Robert Grubbs is the few hours to prevent the wrong stuff from Victor and Elizabeth Atkins Professor of laying down,” she continues. Chemistry. “We’re very enthusiastic about the cornea work,” Kornfield says, “but my Julia Kornfield is a professor of chemi- hope is that we can create a gel that cal engineering. Her research on corneal could prevent any kind of scar,” anywhere wound healing is also sponsored by the in the body. Jacobs Institute for Molecular Engineering for Medicine.

winter 2012 ENGINEERING & SCIENCE 41 Alumni Impact Caltech Alumni Shine in Tr anslational Medicine Whether it’s understanding transplant rejection or building a better prosthesis, Caltech alu mni have made an impact in translational medicine. Here are some of the highlights: By Katharine Gammon

1929 Albert Tyler 2010 PhD 1929, Biology Heather Agnew PhD 2010, Chemistry Albert Tyler earned the first doctorate ever to be granted in the Division of Biology. He stayed on as a Heather Agnew, now a principal research investigator at Integrated member of the faculty, where his studies of embryonic Diagnostics, designs binding molecules that detect specific disease- differentiation in sea urchins helped transform classi- related proteins and allow doctors to discover them earlier. At Caltech, cal embryology into modern developmental science. He she was awarded the Lemelson-MIT Caltech Student Prize for mak- helped introduce biochemical methods into the field, ing heat-stable antigen-detecting compounds that could one day developing techniques to detect cellular antigens that replace tests based on Tyler’s methods and make cheap, reliable underlie many of the diagnostic methods used today. diagnostic kits available to the Third World. He died in 1968.

1983 Morteza Gharib PhD 1983, Aeronautics Caltech professor Mory Gharib looks at everything from space to cells and finds a way to improve the human condition. His work, which includes studying the dynamics of heart valves and creating valveless pumps, sits on the cusp of biology and engineering. Gharib finds inspiration in the natural world, designing sustainable biomedical devices and harvesting sustainable energy within the body to run them. 1983 Julia Kornfield BS 1983, Chemistry; MS 1985, Chemical Engineering An undergraduate research fellowship on nerve cells got Julie Kornfield hooked on biotech.N ow a Caltech professor, she stud- 1974 ies how polymers can be exploited—in and out of the human David Ho body. One of her projects focuses on giving cataract patients BS 1974, Biology better vision through laser-adjustable lens implants; another is David Ho changed gears as he entered his graduate stud- helping to build artificial tissues with real cells. Kornfield was the ies and was drawn toward medicine. Ho’s AIDS research first alumna to gain tenure at theI nstitute. caused him to be named Man of the Year for 1996 by TIME magazine. He pioneered treating HIV-infected pa- tients with protease inhibitors, and championed the use of combination anti-retroviral therapy early in the dis- ease’s course. Ho is currently a professor at Rockefeller University in New York.

42 ENGINEERING & SCIENCE winter 2012 Caltech Alumni Shine in Tr anslational Medicine Whether it’s understanding transplant rejection or building a better prosthesis, Caltech alu mni have made an impact in translational medicine. Here are some of the highlights:

1941 William Corcoran BS 1941, Chemistry; MS 1942, PhD 1948, Chemical Engineering Bill Corcoran had interests everywhere—he played all the intra- mural sports, wrote for the California Tech, and even worked on the firing mechanism for the atomic bomb during the Manhattan Project. After World War II, he returned to his studies and earned one of the first doctorates given in chemical engineering at Caltech. As 1956 a Caltech professor, he studied the fluid mechanics of heart valves William Hildemann by shooting laser beams through them to accurately measure the PhD 1956, Biology flow. He also helped develop better disposable hospital equipment. After serving in World War II, Bill Hildemann came to Caltech Corcoran died in 1982. to study with legendary immunologist Ray Owens, who was working on the problem of tissue rejection. Hildemann dis- covered the role of antibodies in this process, laying the groundwork for successful organ transplants between unre- lated donors and recipients. He died in 1983.

1956 Leonard A. Herzenberg PhD 1956, Biochemistry In 1970, Len Herzenberg developed the fluorescence-activated cell sort- er—a device that revolutionized immunology and cancer biology and is the basis for the purification of adult stem cells. Herzenberg continues 1960 to work on FACS development with his wife, Leonore (Lee)—whom he Leroy Hood met while she was a research assistant for Albert Tyler. BS 1960, Biology; PhD 1968, Biochemistry The future got a boost from Lee Hood. While a Caltech professor, he and his colleagues pioneered four instruments—the DNA sequencer and synthesizer, and the protein synthesizer and sequencer—which comprise the technological foundation for contemporary molecular biology. Hood now heads the Institute for Systems Biology in Seattle, where he is pioneering a transition from reactive medicine to an ap- proach that’s proactive—as well as predictive, preventive, personal- ized, and participatory.

winter 2012 ENGINEERING & SCIENCE 43 We asked Caltech alums to tell us their favorite classroom endnotes or laboratory memory. Here’s what they had to say:

Linus Pauling’s Chem 1 lecture where he held a chunk of sodium Meeting over what he led us to believe was a bucket of water. Students in the Einstein. front row were especially fearful.

It was a quantum mechanics class and the previous professor had left all the blackboards full. Dr. Leighton [BS ’41, MS ’44, PhD ’47] walked in and spent a minute or two absorbing this mass of chalk, which was physical chemistry (quantum mechanics for chemists). He then announced that we were going to make a “few” changes of notation, and proceeded to give the hour lecture using the other guy’s boards, finishing, on time, at the last equation. I discovered that it was just as unclear in P-chem notation as physics. I flunked the course.

I was one of six students enrolled in Richard Feynman’s Quantum Electrodynamics course. The final exam was simultaneously frightful and delightful: a solo oral examination standing at Feynman’s office blackboard. He convinced me that I understood far more than I realized. I was neck and neck with Les Ingber [BS ’62] regarding breakage in Chem I lab. However, when he dropped a balance, I gave up. Friday afternoon seminar in which von Kármán and Hsue-Shen Tsien [PhD ’39] would argue about rocketry theory.

To read more favorite memories, visit http://EandS.caltech.edu/endnotes/memories

44 ENGINEERING & SCIENCE winter 2012 Friday afternoon seminar in which von Kármán and Hsue-Shen Tsien [PhD ’39] would argue about rocketry theory.

To read more favorite memories, visit http://EandS.caltech.edu/endnotes/memories California Institute of Technology Pasadena, California 91125