If I Only Had a …

B ODYBUILDING:THE B IONIC H UMAN I NTRODUCTION If I Only Had a …

could stay young and chipper and I’d lock it with a zipper, if I only had a heart,” “ sang the Tin Man to his companions in the 1939 movie The Wizard of Oz. There is hope yet for the Tin Man and for the many thousands of patients with diseased hearts, damaged livers, or injured tissues. Spurred on by a severe shortage of donor organs for transplantation, bioengineers, cell biologists, and clinicians are combining their tal- ents to design off-the-shelf replacement parts for the human body. This dream is not Inew, as our historical timeline on the next two pages reveals. A daring goal that has captured the public imagination is the implantation of a totally artificial heart into patients with end-stage heart failure (p. 1000), a remarkable achievement despite recent setbacks. Of simpler design, ven- CONTENTS tricular assist devices, which support the left ventricle until the heart repairs itself, may benefit the majority of patients with heart disease (p. 998). Similar- 998 Mechanical Circulatory Support ly, a bioartificial liver device, composed of liver cells nurtured in a bioreactor, —a Long and Winding Road P. M. McCarthy and W. A. Smith is undergoing clinical testing as a bridge to support the failing liver until it can regenerate (p. 1005). Meanwhile, despite numerous regulatory hurdles 1000 A Space Age Vision Advances in (p. 1003), researchers are developing artificial blood substitutes (p. 1002) for the Clinic use instead of donor blood during lengthy surgeries or at disaster sites. 1002 Artificial Blood From tendons (p. 1011) to bladders, bioengineers are manufacturing J. E. Squires “ready-to-wear” designer tissues in the laboratory (p. 1009). They are coaxing 1003 Not Blood Simple cells to assemble into three-dimensional structures on biodegradable scaffolds 1005 A Bioartificial Liver—State of that can be implanted in patients at sites of tissue injury. Challenges to be ad- the Art dressed before engineered tissues enter the clinic include identifying sources of A. J. Strain and J. M. Neuberger suitable cells, developing new scaffolding biomaterials (p. 1014), and scaling 1008 Wanted: Pig Transplants That up production, not to mention discovering ways to preserve the tissue product Work (p. 1015) and to prevent immune rejection. Animal models of spinal cord in- 1009 Tissue Engineering—Current jury are being used to test different bioengineering and molecular strategies for

Challenges and Expanding on February 3, 2016 their ability to induce the injured spinal cord to regenerate (p. 1029). Opportunities Sophisticated microelectronics for signal processing are bringing the L. G. Griffith and G. Naughton dream of merging man and machine closer to reality. A brain-computer in- 1011 Tending Tender Tendons terface that enables patients to control artificial limbs through brain neural 1014 Third-Generation Biomedical

CREDITS: (TOP TO BOTTOM) UNDERWOOD & UNDERWOOD/CORBIS; UNDERWOOD CREDITS: BOTTOM) DENISE LEAR RUTH TO (TOP pathways is under development (p. 1018). The success of cochlear implants, Materials which enable some profoundly deaf patients to hear (p. 1025), has ac- L. L. Hench and J. M. Polak celerated engineering of the next generation of electrode implants 1015 New Prospects for Putting

for brainstem auditory centers. Researchers facing the enor- Organs on Ice Downloaded from mous challenge of trying to restore vision to the blind are de- 1018 The Bionic Man: Restoring Mobility veloping electrode arrays that activate the retina, optic W. Craelius nerve, or even the visual cortex itself (pp. 1022 and 1026). 1020 Part Man, Part Computer: Designing bionic devices that enhance human ca- Researcher Tests the Limits pabilities (cybernetics) is raising a host of ethical questions (p. 1020), as is the transplant of xenogeneic tis- 1022 Will Retinal Implants Restore Vision? sues from other species into human patients (p. 1008). E. Zrenner Even older than the science of replacing body parts is the art of staying young, and antidotes to aging are be- 1025 Sending Sound to the Brain J. P. Rauschecker and R.V. Shannon coming big business as millions of baby boomers hit their 50s (p. 1032). 1026 The Confusing Mix of Hype and The next 38 pages reveal how far we have come in Hope our dream of engineering designer parts for the human 1029 Repairing the Injured Spinal Cord body to replace those lost through injury or disease. M. E. Schwab However, this special section also highlights the many 1032 The Quest to Reverse Time’s Toll hurdles yet to be overcome before these fascinating Cracking the Secrets of Aging technologies enter routine clinical use. See also Science's STKE, page 923. –MARC LAVINE,LESLIE ROBERTS,ORLA SMITH Seeking to serve all of Science’s read- ership, we recommend Boing-Boing the Bionic Cat to those of our very young readers who are contemplat- ing a future career in bioengineering.

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 995 Prosthetics 1504 1888 & devices ❙ ❙ Blood, organs, 1628 1667 1822 1881 Historical Highlights ❙ ❙ & tissue ❙ ❙ ❙ ❙ ❙ 1666 1682 1883 in Bionics and Related Medicine 1597 1807 1842 1858 ❙ ❙ ❙ ❙ Bionics has had a rich and fascinating history. In the development of arm and leg Medicine ❙ ❙ 1847 1863 prostheses, progress has come as a slow and gradual process. In other fields, such as organ transplantation, there have been decades between the first attempts at a new 1504 Iron prosthetic hand with flexible procedure and the routine achievement of success. This punctuated progress has finger joints. frequently been contingent on accumulating new insights from other disciplines, such 1597 Reconstruction of the nose by as the role of the immune system in tissue rejection, or on serendipitous discovery, as tissue grafting. was the case for several anti-rejection drugs. Rather than attempting to be compre- hensive, we have portrayed in this timeline, highlights from different areas of medical 1628 Early theory on the circulatory research. Similarly the images, although not always connected to specific dates or system developed. events, are intended to portray both interesting artifacts from the past, and today's 1666 Blood transfusion between two high-tech advances, which provocatively hint at developments yet to come. dogs. 1667 Blood transfusion between sheep and human. 1682 Repair of human skull with dog ABIOMED, INC. skull bone. 1807 Development of an endoscope for minimally invasive surgery. 1822 First successful fresh skin autograft. 1842–1847 Anesthetic properties of a Subretinal chip in a cat eye number of compounds first discovered and demonstrated.

DR. H. SACHS, REGENSBURG UNIVERSITY EYE HOSPITAL 1847 Introduction of silver amalgam for dental fillings. 1858 Publication of the first edition of Gray's Anatomy. AbioCor self-contained, replacement heart 1863 Introduction of antiseptic surgical techniques. JON VAN ALLEN,JON VAN UIHC MEDICAL MUSEUM 1881 First temporary skin graft.

Glass eye prostheses c. 1900 1883 Development of Ringer’s solution for keeping tissues alive outside BIOVASCULAR, INC. LIFECELL CORPORATION the body. 1888 First reports on use of contact lenses to correct vision. MEDTRONIC INC.MEDTRONIC 1901 Identification of different blood MEDTRONIC, INC. Ocu-Guard orbital implant groups. wrap AlloDerm® processed 1905 & 1906 First reports of corneal human dermis transplants. Mechanical Medtronic Hall aortic valve 1905 Discovery of technique for growing Bioprosthetic Mosaic tissue cells in vitro. aortic valve COCHLEAR LTD. ARROW INTERNATIONAL ARROW

NATIONAL MUSEUM OF HEALTH AND MEDICINE MUSEUM OF HEALTH NATIONAL Nucleus 24 Contour cochlear electrode array

Debakey left ventricular assist device, c. 1966

Emerson respirator (iron lung), c. 1950 AND MEDICINE MUSEUM OF HEALTH NATIONAL IK HEART INC.IK HEART RV TTOBOCK ORTHOPEDIC INDUSTRY, ORTHOPEDIC TTOBOCK INC. O JA

LionHeart left ventricular assist device

NATIONAL MUSEUM OF HEALTH AND MEDICINE Jarvik-7 heart, 1986 Responsive prosthetic Prosthetic arm, c. 1920’s SensorHand 1905 1928 1939 1943 1951 1957 1969 1972 1979 1980 1982 1987 1993 2000 2001 ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ 1953 1958 1901 1908 1954 1963 1966 1967 1973 1981 1983 1984 1986 1990 1995 1997 ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ 1905 1956 1998 1914 1949 1953 1958 1969 1978 1982 1990 ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ ❙ 1911 1905 Successful direct blood transfusion 1966 First successful pancreas transplant. 1990 First living donor lung transplant. between humans. 1967 First successful heart transplant. 1993 FDA approval of left ventricular 1905 Early attempt at an artificial hip Patient survived 18 days. assist device as a bridge to heart replacement. 1969 First biocompatible ceramic that transplantation. 1908 Early attempt at knee replacement could bond to collagen and bone 1995 Jeff Getty receives a baboon bone surgery (using a cadaver for the developed. marrow transplant. replacement part). 1969 Total artificial heart implanted in a 1997 Transplant of pig neurons in patients 1911 Paraffin injection to treat vocal human as a temporary measure. with Parkinson's disease. fold paralysis. 1972 Testing of modern design 1998 Human hand transplant. 1914 Citrate identified as a blood anti- steel/polymer hip joint. 1998 Total larynx transplant. coagulant, allowing for blood 1973 Successful unrelated bone marrow storage. 2000 Implantation of a prototype artificial transplant. pancreas. 1928 Iron lung developed for treatment 1978 The immunosuppressant of polio victims. 2001 Implantation of the AbioCor, a cyclosporine is introduced. permanent self contained total 1939 Hard (plastic) contact lenses 1979 First auditory brainstem implant. heart replacement. introduced. 1980 First successful single-channel 1943 Kidney dialysis machine developed. cochlear implant in a child. 1949 Role of immune system in tissue 1981 A peripheral nerve bridge is rejection identified. implanted into the injured spinal A. ATALA, HARVARD 1951 First artificial heart valve implanted. cord of an adult rat. 1953 Development of the heart-lung 1982 Genetically engineered insulin be- machine. comes commercially available (the 1953 Demonstration of acquired immune first genetically engineered drug).

tolerance to foreign grafts. 1982 Implantation of the Jarvik-7, a AND MEDICINE MUSEUM OF HEALTH NATIONAL permanent total artificial heart. 1954 Kidney transplant between identical An engineered human twins. 1983 First successful single lung bladder 1956 First successful bone marrow transplant. transplant. 1984 Baby Fae receives heart from 1957 First cochlear implant developed. baboon and survives 20 days. 1958 Early attempts at developing an 1986 First successful double lung Artificial leg, implantable pacemaker. transplant. c. 1937 1958 Identification of the importance of 1987 First clinical use of a bioartificial the histocompatibility system for liver device. tissue matching. 1990 FK506 immunosuppressant

1963 First liver transplant. becomes available. AND MEDICINE MUSEUM OF HEALTH NATIONAL

Biomet hip implant, late 20th century BOCK ORTHOPEDIC INDUSTRY,BOCK ORTHOPEDIC INC. TO OT

Porcelain denture, c. 1850 NATIONAL MUSEUM OF HEALTH AND MEDICINE MUSEUM OF HEALTH NATIONAL NATIONAL MUSEUM OF HEALTH AND MEDICINE MUSEUM OF HEALTH NATIONAL Kolff-Brigham dialysis unit (artificial kidney), c. 1950 C-Leg System, leg

NICK ROMANENKO, SERVICES PHOTO RUTGERS prosthesis Tendon-controlled artificial hand

Main sources: National Museum of Health and Medicine, Armed Forces Institute of Pathology, Washington, D.C. M. Facklam, H. Facklam, Spare Parts for People (Harcourt Brace Jovanovich New York, NY, 1987). PEDIC INDUSTRY, INC. AL MUSEUM OF HEALTH AND MEDICINE AL MUSEUM OF HEALTH M OF HEALTH AND MEDICINE M OF HEALTH U W. B. Murphy, Spare Parts: From Peg Legs to Gene Splices (Twenty-First Century THO R

ATION Books, Brookfield, CT, 2001). MUSE N International Encyclopedia of Science and Technology (Oxford University Press,

OBOCK O Silver dentures with ivory Oxford, UK, 1999). T NATIONAL NATIONAL Howmedica Sphero- OT teeth, c.1800 S. M. Zeitels, M.D., Harvard Medical School and Massachusetts Eye and Ear Infirmary. centric knee implant Mechanical knee joint Authors of our pieces. c. 1970’s B ODYBUILDING:THE B IONIC H UMAN VIEWPOINT Mechanical Circulatory Support—a Long and Winding Road Patrick M. McCarthy1,2* and William A. Smith2

The highly public reintroduction of the total artificial heart last year has match between right and left blood flows. prompted renewed interest in mechanical circulatory support systems for Existing TAHs are large and do not fit many the treatment of end-stage heart disease. male candidates, most females, or any children. Should the TAH become infected, even Heart transplants are thought of as miracles heart transplant patients. The AbioCor TAH intravenous antibiotics may not be effective. because of their impact on length and qual- that has made news headlines is a titanium Despite these problems, the TAH is a desir- ity of life. Unfortunately, like miracles, and polymer construct that uses an electrohy- able device for certain patients with severe they are all too rare. Each year in the draulic actuator system to shuttle blood alter- heart muscle damage, or for those where the United States, about 2000 patients with nately between the right and left pumps (3). continued presence of the native heart is a end-stage heart failure receive heart trans- Currently under development are electrome- liability (4). However, because of the inher- plants. Yet 400,000 or more individuals chanical TAH devices that directly convert ent limitations in current TAH technology, develop heart failure annually, of whom electrical energy into mechanical action. A there is wide agreement that only about 10% 30,000 to 100,000 potentially could benefit recent improvement has been the use of a of the patients with end-stage heart disease from a transplant. This has led to a system transcutaneous energy transmission system (approximately 5000 to 10,000 U.S. candi- of rationing where committees have to de- (TETS) that transmits power through intact dates per year) should be treated with this termine whether a candidate is eligible for a skin, eliminating the risk of infection entering radical approach (5). transplant. Many patients have to be reject- the body along the percutaneous power line Most clinicians believe that 90% of ended because of advanced age, other medical that was used with earlier implanted blood stage heart disease patients can be managed conditions that limit life expectancy or pumps. A rechargeable implanted battery with a ventricular assist device (VAD) that quality of life, poor medical compliance, or provides brief periods of freedom from the supports the left ventricle, the locus of the bad habits such as smoking or alcohol external battery and electronics associated most critical heart damage (6, 7). With a abuse. Physicians strive to give the limited with the TETS, allowing the patient to show- VAD, the right ventricle benefits from reduc- number of hearts to those who will best er or, theoretically, even to swim. tion of left atrial and pulmonary arterial pres- take care of this gift. No one wants to be in A TAH presents many design challenges. sures and the heart can start to repair itself. on February 3, 2016 the uncomfortable position of rationing ac- In all probability the patient will die very While “resting” during VAD support, dam- cess to heart transplants, so physicians have rapidly should the machine malfunction. aged cardiac muscle starts to repair itself and worked for decades to develop complex Therefore, it must perform flawlessly. A ma- to contract with more vigor. The benefit of a machines that, like cardiac pacemakers or jor potential problem are blood clots that VAD versus a TAH is that the natural heart heart valves, can save desperately ill pa- form around the four artificial plastic valves remains in place—if the VAD is stopped or tients. The ultimate goal is to develop a and in response to the synthetic material with removed, the patient’s heart can still support machine that is as good as or even better which the blood is constantly in contact. Pul- the circulation. Clinical experience has dem- than the donor heart, avoiding the need for monary problems may develop from a mis- onstrated that once a damaged heart is “rest- Downloaded from a transplant altogether. As a result of the highly public reintro- Fig. 1. The implantable Heart- duction of the total artificial heart (TAH) in Mate VAD, a pulsatile-flow de- July 2001, the interest of the medical com- vice that supports the left ven- munity and the public has been drawn again tricle of the damaged heart. to mechanical circulatory support systems for Blood flows from the left atrium the treatment of end-stage heart disease (see through the left ventricle and page 1000) (1). A TAH implant is a dramatic the pump inflow valve into the VAD, which then pumps the event that removes the patient’s heart and blood into the aorta. [Image results in complete dependence on the most courtesy of Cleveland Clinic sophisticated technology ever implanted in Foundation] humans. A TAH must “beat” approximately 35 to 40 million times a year, providing a cardiac output of 5 to 6 liters/min of blood (2). Current design goals are 90% reliability after 5 years of operation, which is compara- ble to the 70% 5-year survival for current

1Department of Thoracic and Cardiovascular Surgery, George M. and Linda H. Kaufman Center for Heart Failure, 2Department of Biomedical Engineering, Cleveland Clinic Foundation, Cleveland, OH 44195, USA. *To whom correspondence should be addressed. E- mail: [email protected]

998 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

Fig. 2. Size comparison of the Heart- Mate pulsatile VAD (right) with the newer and much smaller continuous- ﬂow (nonpulsatile) Jarvik 2000 VAD (left). Continuous-ﬂow VADs are currently in clinical trials. [Photo courtesy of Cleveland Clinic Foundation]

ed” through VAD support, the heart can re- Prototypes of the continuous-flow VAD pair itself and its function improves. In a few include the Jarvik 2000 (Fig. 2), Micro-Med cases, the VAD was removed and the patient DeBakey pump, and HeartMate II pumps, all no longer needed a transplant. The goal is for now in clinical trials. Each of these devices wider application of VAD support so that has an axial flow impeller supported on Fig. 3. Barney Clark was the first patient to heart failure patients are supported for weeks blood-lubricated pivot points. So far, the de- receive the Jarvik-7 total artificial heart. Clark, implanted with the Jarvik-7 in 1982 and shown or months until cardiac damage has been vices have been implanted with a percutane- here pumping an exercycle under the watchful repaired sufficiently such that a transplant is ous wire that connects the device to an exter- eye of his physical therapist, lived for 112 days not necessary. nal controller and battery. Nothing prevents after the operation. [Photo courtesy of Brad Design of the VAD began as early as 1963 the use of a TETS system in future versions, Nelson/University of Utah Health Sciences but did not progress rapidly until the 1990s. although this would add hardware to the im- Center] Since then, thousands of patients have been plant. The next generation of VAD blood successfully supported, sometimes for years pumps (still being tested in the laboratory) myoblasts have been injected into the dam- (8). The most commonly used VAD is the use three-dimensional magnetic suspension aged heart, improving local cardiac function HeartMate, a rotating torque motor and cam of the pump rotor (12, 13). These pumps and abrogating the problem of rejection asso- system that displaces 75 cc of blood from the address residual concerns regarding friction, ciated with a heart transplant (16, 17). Gene left ventricle into the aorta (Fig. 1). The wear, and the crevice-like geometry of the therapy can be combined with cell therapy control system can increase or decrease the pivot supports (which trap blood cells), and such that the muscle cells to be transplanted pump rate to match demand. The surfaces are predicted to remain reliable for 10 years are engineered to carry genes encoding pro- that make contact with blood are textured so or more. This technology, the prototype of teins that will promote muscle cell survival that an adherent layer of blood cells settles on which is hopefully to be launched this year, and growth. These clinical efforts toward car- the surface, enabling blood to flow smoothly. brings us much closer to the day when small diac repair are not science fiction; they have Blood is directed through the pump by inflow reliable blood pumps with few complications already begun and show encouraging early and outflow valves made of pig tissue (9). In become as common as pacemakers and heart results. the Cleveland Clinic experience, the inci- valves. dence of emboli (blood clots) or stroke is The future most likely includes two sce- References 1. E. Marshall, Science 295, 1000 (2002). very low, about 2%, using only aspirin as an narios: small, magnetically suspended blood 2. G. M. Pantalos et al., Ann. Thorac. Surg. 66, 1852 anticoagulant. pumps that are completely implanted and (1998). A new type of VAD—in which rotary dy- powered through transcutaneous energy 3. R. T. Kung et al., ASAIO J. 41, M245 (1995). 4. P. M. McCarthy et al., J. Thorac. Cardiovasc. Surg. namic pumps produce a continuous nonpulsa- sources; and the “synergistic” combination of 108, 420 (1994). tile flow of blood—is currently in clinical trials. these pumps with new biological therapies. 5. V. Willman, Expert Panel Review of the NHLBI Total In some situations, patients with this new VAD Such therapies include gene therapy, trans- Artificial Heart (TAH) Program June 1998 to No- vember 1999 (www.nhlbi.nih.gov/resources/docs/ have a normal cardiac output of 5 to 6 liters/ plants of muscle stem cells or cardiac myo- tah-rpt.htm). min, but no palpable pulse or measurable blood cytes, and treatment with drugs such as the 6. Cleveland Clinic Heart Center press release, 23 pressure! There was concern that this would beta-agonist Clenbuterol. Patients on VAD April 2001 (www.clevelandclinic.org/heartcenter/ pub/news/archive/2001/heartpump_4_23.asp). cause physiologic abnormalities, but experience support who are treated for several months 7. P. M. McCarthy et al., J. Thorac. Cardiovasc. Surg. with animal models and early clinical results with Clenbuterol show physiologic hypertro- 115, 904 (1998). suggest that the difficulties may have been phy (thickening) of their heart muscle, which 8. E. A. Rose et al., N. Engl. J. Med. 345, 1435 (2001). overestimated (10, 11). These continuous-flow contracts more vigorously (14). A patient 9. J. E. Slater et al., Ann. Thorac. Surg. 62, 1321 (1996). 10. G. M. Wieselthaler et al., Ann. Thorac. Surg. 71, S139 devices are much smaller and simpler than the could receive a small blood pump to “rest” (2001). pulsatile pumps (Fig. 2) and thus fit the major- the heart, meanwhile being treated with bio- 11. O. H. Frazier et al., Circulation, in press. ity of adults and children. They may also be less logical therapies to promote healing of the 12. Y. Ochiai et al., Ann. Thorac. Surg. 72, 747 (2001). 13. S. Saito et al., Eur. J. Cardiothorac. Surg. 19, 678 expensive. Most current designs use highly pol- cardiac muscle. Once the heart is repaired, (2001). ished titanium surfaces that appear to be com- the pump would be removed. Already an 14. M. H. Yacoub, E. J. Birks, P. Tansley, M. Y. Henien, C. T. patible with blood, at least for the short resi- oxygen-starved heart has been induced to Bowles, J. Congest. Heart Fail. Circ. Support 2,27 (2001). dence times during which blood is in contact grow new blood vessels through injection of 15. A. T. Askari, M. S. Penn, Semin. Thorac. Cardiovasc. with the pump. As a “partial” assist VAD that the gene encoding vascular endothelial Surg., in press. elevates the total cardiac output of a weak heart growth factor directly into the damaged car- 16. Cleveland Clinic Heart Center press release, 14 June 2001 (www.clevelandclinic.org/heartcenter/pub/news/ closer to the normal range, these devices show diac muscle (15). Recently, new heart muscle archive/2001/skeletalcelltransplant_6_14.asp). tremendous potential. cells grown from a patient’s own skeletal 17. P. Menasche et al., Lancet 357, 279 (2001).

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 999 B ODYBUILDING:THE B IONIC H UMAN NEWS A Space Age Vision Advances in the Clinic

The total mechanical heart is making headlines, but many say a less first R&D program, with a budget that rose radical technology—the heart assist device—is leading the revolution to $10 million a year by 1970. Contracts, mainly with aerospace firms, yielded sever- After barely surviving for nearly 2 decades, ef- the National Heart, Lung, and Blood Insti- al test devices, including the Jarvik-7. De- forts to build a “total artificial heart” have tute (NHLBI) who oversees this area. But veloped at the University of Utah under the gained a new lease on life. Surgeons in that success has rekindled a debate about the direction of Willem Kolff, it was the first Louisville, Kentucky, removed the failing heart best way to meet the needs of the estimated artificial heart implanted in a human with of a 59-year-old man last July and replaced 100,000 people in the United States who, FDA’s approval. Kolff recruited Utah heart it with a machine. Indeed, they implanted a like Tools, face heart failure each year yet surgeon William DeVries to lead the clini- whole network of machines: a 1-kilogram have little prospect of obtaining one of the cal work and in 1982 won FDA’s consent to pump, a computer, a battery, a power converter, 2200 donor hearts available for transplant. treat five patients. DeVries and others re- and cables to link them all together. AbioCor’s comparative success has led ported results in 1988. An accompanying It was a dramatic procedure, the first to- supporters to argue that it could eventually editorial in the Journal of the American tal heart implant in the United States since Medical Association noted “many compli- the mid-1980s. The patient, Robert Tools, cations” that “seem to be related specifical- had been facing certain death and was too ly to the design of the Jarvik-7,” such as sick for a donor heart. He agreed to be the blood clots and infections along the skin- first to try a new mechanical replacement penetrating power lines. heart made by Abiomed Corp. of Danvers, Three months later, NHLBI director Claude Massachusetts. And it was a success, at least Lenfant announced that he was ending the to- in some respects: Tools survived the implan- tal artificial heart program because NHLBI tation at Louisville’s Jewish Hospital and lat- couldn’t afford it, and many clinicians believed er gained enough strength to venture out- that VAD systems then under development doors a couple of times. But he suffered a were closer to practical use. But two powerful stroke—caused possibly by a clot that senators—Edward Kennedy (D–MA) and formed in the artificial heart. On 30 Novem- Orrin Hatch (R–UT)—got NHLBI to reverse ber he died after 151 days on the machine. that decision and renew funding.

Tools’s experience was certainly better than Over the years, NHLBI has spent about on February 3, 2016 that of his famous predecessor—Barney $100 million on the total artificial heart, Clark, the Seattle dentist who in 1982 was the Watson estimates. By the late 1990s, how- first to receive the celebrated “Jarvik-7” heart, Heart of the heart. The AbioCor pump is con- ever, only two competitors remained: named for its designer Robert Jarvik. Clark trolled by a small computer and power supply Abiomed and a group led by William Pierce, lived 112 days tethered to an external power system, all implanted. Walter Pae, and Gerson Rosenberg of Penn- unit the size of a washing machine, never able sylvania State University’s Hershey Medical to leave the hospital room. In news clips he ap- be an alternative to transplantation. But Center. Penn State lost the race to the clinic,

peared miserable and uncomfortable. Even critics dismiss the totally implantable heart Rosenberg says, when its industrial backer, Downloaded from though another patient with a Jarvik-7 lived as “obsolete,” a hugely expensive technol- 3M Corp., pulled out. In 2000 the university for 620 days, the artificial heart became linked ogy that may never deliver an acceptable sold the “Penn State Heart” to Abiomed, in the public mind with Barney Clark’s ordeal. quality of life. Many experts, including which plans to develop it. Hoping to erase those memories, Abiomed Watson, believe that, at least in the near took a different tack. It created a more user- term, desperate patients are more likely to Survivor friendly device, called AbioCor, in which the be helped by machines known as ventricu- What makes AbioCor exceptional, says pumping components are fully implanted, lar assist devices (VADs). These mechani- company vice president Edward Berger, is powered by radio-frequency energy transmit- cal pumps, in use since the early 1990s, that the system fits within the body. Unlike ted through the skin. Doing away with con- boost rather than replace the ailing heart by predecessors, it doesn’t require a line into the nector lines may lower the risk of infection increasing blood flow through a single body for pressure relief or power supply. It and frees the patient to walk around for sever- chamber, or ventricle, and may even help relies on a quiet centrifugal electric pump al hours with batteries. Company president the heart get stronger. that shuttles hydraulic fluid alternately be- and CEO David Lederman also sought to pro- All these devices are now just stopgaps. tween two chambers that press on blood sacs tect patients’privacy and limit harsh publicity: The U.S. Food and Drug Administration corresponding to the left and right sides of Hospitals agreed not to give daily reports on (FDA) has approved them only as a “bridge the heart. In addition to reducing infection medical details. And he lowered expectations, to transplant,” not as a permanent solution. risks, according to Abiomed, this should en- telling reporters that they should consider it a But supporters of AbioCor and VADs are able the patient “to remain mobile and con- success if the first patients lived 60 days. At both hoping to gain FDA’s blessing for far tinue a productive lifestyle.” this writing, four of six had passed that mark more extensive use of these ultimate bion- The AbioCor has not been priced yet, and three were still living. Time magazine ic devices. Berger says, although many predict that it will dubbed AbioCor “the invention of the year.” cost about $75,000. Berger notes: “Our goal is The AbioCor trials have gone “remark- Aerospace baby to eventually sell it for less than half that ably well” considering how sick the patients The U.S. effort to build artificial hearts be- amount,” assuming manufacturing can be im-

are, says John Watson, the senior official at gan in 1964 when NHLBI established the proved. The implantation procedure might CREDIT: ABIOMED INC.

1000 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN cost $150,000, observers say. The heart would maker,” to suggest something as routine as aby says he and colleagues at the Texas Heart be offered to patients with moribund left and a pacemaker. Jarvik argues that the risk of Institute in Houston and the Cleveland Clinic right heart chambers who could not be saved infection is greatly reduced because his de- in Ohio have all seen this phenomenon. This by a VAD. No one is sure how many people vice is one of several new models powered is “my biggest interest” now, Westaby says. “I might be candidates; Abiomed estimates by a miniature rotary pump that generates intend to spend a lot of time on it.” He imag- “many thousands” per year in the United no pulse and requires no artificial valves. ines that it may be possible someday to rou- States, whereas an execu- The entire mechanism tinely use small pumps, drugs, and cell thera- tive at a VAD manufac- sits in the ventricle im- py to restore sick hearts. turing firm doubts that mersed in blood, an en- Many other VAD designers are also de- the number would reach vironment hostile to veloping miniature rotary pumps. The 3000. VADs, by contrast, bacteria. biggest U.S. company in the field, Thoratec are thought to have a po- One surgeon pioneer- Corp. in Pleasanton, California, is working tential market of 30,000 ing the use of Jarvik on several generations of VADs simultane- to 100,000 per year. 2000, Stephen Westaby ously. The newest, called HeartMate II and Before Abiomed can of John Radcliffe Hospi- HeartMate III, will include small rotary sell this product, however, tal at Oxford University, pump devices. Terumo Corp. of Tokyo was it must demonstrate to U.K., seconds Jarvik’s among the first to use a levitating magnet FDA that its device will views. “The total artifi- system to eliminate bearings and friction that extend and improve a pa- cial heart puts us back can damage blood cells, and similar designs tient’s life. Jarvik thinks 30 years,” Westaby says. are being developed by Thoratec and the this will be hard. The Instead, he praises the Berlin Heart company in Germany. AbioCor system “repre- “thumb-sized” Jarvik Keith Grossman, CEO of Thoratec, claims sents old technology,” 2000 because it is easy that the HeartMate II VAD pump is designed Jarvik argues. It uses a to sew inside the pa- to run for 5 to 8 years and HeartMate III, displacement diaphragm tient’s left ventricle— for up to 10 years. These goals are twice the pump that requires artifi- and it can be removed design life of Thoratec’s current standby, the cial heart valves, he adds, just as simply if the nat- HeartMate XVE. Grossman expects that a making for a complex ural heart improves. Al- growing number of patients who cannot get system that’s likely to ready, Westaby says he donor hearts will use VADs as a permanent cause clots and infections. is using the Jarvik 2000 solution. Indeed, Thoratec has asked FDA Jarvik predicts that fiscal experimentally as a per- for permission to begin marketing VADs as managers will balk at to- manent prosthesis: “I’ve a therapy for heart disease, and FDA has tal replacement hearts: Pioneers. Robert Jarvik (left) and Bar- had three in the commu- already scheduled an advisory panel re- Because they are intended ney Clark, the patient who received the nity for 7 to 19 months view for 4 March. for the sickest heart pa- first Jarvik-7 heart, in 1982. now, and they’re all do- Thoratec and others plan to cite the tients, Jarvik predicts that ing terrifically well.” REMATCH trial,* reported last November, implanting them will often be traumatic, lead- The 19-month patient, Peter Houghton, as proof that mechanical hearts are robust ing to “million-dollar patients.” who according to Westaby was “within a enough now to be used as standard therapy. Abiomed’s Berger rejects this skepticism, couple of weeks of death” before surgery, This trial of 129 patients showed that those noting in an e-mail that “reliability testing in recently flew across the Atlantic Ocean on a with VADs were more likely to survive the laboratory leads us to think … we may al- commercial plane to see relatives and brief over a 2-year period, and to enjoy a good ready have an AbioCor that will sustain a pa- NHLBI on his health. quality of life, than those on standard drug tient for more than a year.” It is too early to tell Researchers are surprised to discover that therapy. Eric Rose, lead author of the whether blood clots will be a problem for the the use of VADs enables some patients’ hearts REMATCH study at Columbia University first-generation AbioCor, he says, although to regain strength by resting the muscle. West- in New York City, predicts that an FDA results so far suggest that “the issue is ad- policy change could come as dressable through anticoagulation manage- early as this summer. ment” similar to that already given to patients The major task after the with artificial heart valves. Recently Abiomed FDA review, Rose says, will be announced that it is removing an outer “cage” to argue that Medicare should to lower the risk of clots in future trials. Even- reimburse medical centers for tually, 5 to 10 years from now, Abiomed hopes VADs even if patients are not its device will run for 5 years and be an alter- waiting for a transplant. Rose is native to heart transplantation. already consulting with the De- partment of Health and Human Reversing field Services about costs. If Medi- As Abiomed seeks to prove that its device care gives a green light, the ar- can replace a failing heart, others—includ- tificial heart–makers will enter ing Jarvik—are pursuing a different strate- a brave new world. gy. “Removing the natural heart is an obso- –ELIOT MARSHALL lete approach,” Jarvik says today. His own company in New York City is developing * E.A. Rose et al.,“Long-Term Use of small, simple pumps that go inside the a Left Ventricular Assist Device for End-Stage Heart Failure,” New Eng- heart and potentially salvage it. He calls his Next generation. Robert Tools (left), with doctors; Tools lived 151 land Journal of Medicine 345,

CREDITS: (TOP TO BOTTOM) BETTMANN/CORBIS;CREDITS: BOTTOM) HOSPITAL TO AFP PHOTO/JEWISH (TOP new device the Jarvik 2000, or the “flow- days on the AbioCor device but suffered a stroke and fatal bleeding. 1435–1443 (2001).

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1001 B ODYBUILDING:THE B IONIC H UMAN VIEWPOINT Artiﬁcial Blood Jerry E. Squires

Blood substitutes are under development for transfusion in place of are two types of blood substitute currently donor blood during emergencies and lengthy surgeries. The ﬁrst gen- under development: those derived from he- eration of blood substitutes is currently in clinical trials. moglobin, called hemoglobin-based oxygen carriers (HBOCs), and those that use perfluorocarbon emulsions (3). The idea of using a blood substitute instead of reasons for developing blood substitutes. Hu- human blood for transfusing patients is not man red blood cells have exacting storage new. In the 17th century, Sir Christopher requirements designed to prolong clinical ef- Hemoglobin-Based Oxygen Carriers Wren suggested that ale, wine, and even opi- fectiveness and to reduce the risk of bacterial The hemoglobin in adult erythrocytes is a um could be used as substitutes for human contamination. This substantially limits the tetramer of two ␣ and two ␤ polypeptide blood. It was not until 1901 that the modern availability of blood at disaster sites or on the chains, each of which is bound to an iron- age of blood transfusion began with the dis- battlefield. Blood substitutes with less strin- containing heme group. Each heme group covery of human blood group antigens by gent storage requirements would be valuable binds one oxygen molecule. This oxygen- Karl Landsteiner (1, 2). He categorized hu- in these situations. In addition, blood substi- heme bond results in a conformational man blood into A, B, and C (later renamed O) tutes are more amenable to sterilization to (shape) change in the hemoglobin molecule, groups, and a year later the AB group was remove infectious pathogens, and they do not which in turn progressively increases the af- added to the list. These studies were critical require cross-matching because they do not finity of hemoglobin for additional oxygen to understanding why blood transfusions had harbor blood group antigens. Donor blood molecules. Thus, a small change in oxygen failed thus far. Ottenberg in 1913 was the first shortages make blood substitutes attractive partial pressure results in a large change in to apply blood group serology to transfusion for the short-term replacement of erythro- the amount of oxygen bound or released by practice. Even with Ottenberg’s description cytes during surgery. In the 1980s, the real- hemoglobin (Fig. 1). A number of additional of blood compatibility testing, blood transfu- ization that the human immunodeficiency vi- factors, such as temperature and pH, can alter sions were severely limited because of a lack rus (HIV ) could be transmitted through blood the oxygen-hemoglobin dissociation curve of suitable anticoagulants and storage meth- transfusion provided renewed impetus for the (Fig. 1). Similarly, 2,3-diphosphoglycerate ods. The First and Second World Wars pre- development of “disease-free” blood substi- (2,3-DPG), a product of the red blood cell cipitated progress on these two fronts, at last tutes. Before the development of a specific glycolytic pathway and present in normal allowing blood transfusion to become a stan- test for HIV, the risk of transfusion-associat- human erythrocytes, also directly affects the on February 3, 2016 dard part of medical treatment. ed acquired immunodeficiency syndrome binding of oxygen to hemoglobin. As the One of the principal tasks of blood is to was about 38 per 100,000 transfused patients. concentration of 2,3-DPG rises, the oxygen- transport oxygen throughout the body and Researchers interested in developing hemoglobin dissociation curve “shifts to the then to release the oxygen to tissues, picking blood substitutes (dubbed artificial blood) right,” thereby allowing the release of oxygen up carbon dioxide in its stead (1, 2). All of have predominantly concentrated on mimick- to tissues at higher than normal oxygen par- this is accomplished by hemoglobin, the ox- ing the oxygen-carrying capacity of hemoglo- tial pressures (Fig. 1).

ygen-carrying protein contained within red bin. However, in addition to being able to A cell-free solution of hemoglobin can be Downloaded from blood cells (erythrocytes). The other cellular transport oxygen, an ideal blood substitute used as a blood substitute because hemoglo- constituents of blood are white cells, which would also (i) require no cross-matching or bin maintains its ability to transport oxygen are important for the immune response, and compatibility testing, (ii) be suitable for long- outside of the red blood cell. One advantage cellular fragments called platelets that are term storage (preferably at room tempera- of HBOCs is that compatibility testing is not crucial for blood clotting and wound healing. ture), (iii) be able to survive in the circulation required. In contrast, transfusion of donor Although transfusion of donor blood is a for several weeks (the intravascular “dwell” blood requires careful cross-matching to routine and safe procedure, there are several time) before being cleared by the kidney, (iv) avoid the consequences of a hemolytic trans- be free of side effects, (v) be free of patho- fusion reaction in the recipient, the cause of a American Red Cross Biomedical Services, Arlington, gens, and (vi) not only transport but also small but significant transfusion-associated VA 22307, USA. E-mail: [email protected] effectively deliver oxygen to tissues. There morbidity and mortality. Another advantage of cell-free hemoglobin is that it can be ster- Fig. 1. Oxygen-hemoglobin dis- ↑ pH,↓ temperature,↓ 2,3-DPG ilized by ultrafiltration and low heat to inac- sociation curves. A small change pH 7.4/37ºC ↓ pH,↑ temperature,↑ 2,3-DPG tivate infectious agents, a strategy that is not in oxygen partial pressure results possible with red blood cells. in a large change in the amount Early attempts to develop blood substi- of oxygen released or bound by native hemoglobin. Under condi- tutes in the 1970s concentrated on cell-free tions of decreased pH, increased solutions of human hemoglobin (4). But temperature, or an increase in problems with these HBOCs included un- the concentration of 2,3-DPG, acceptably short survival times in the cir- the oxygen-hemoglobin dissoci- culation, an abnormally high oxygen affin- ation curve is shifted to the right, Oxygen content (ml/dl) ity, and clinical side effects such as mal- thereby enabling the release of oxygen to tissues at higher than Partial pressure of oxygen aise, abdominal pain, hemoglobinuria, and normal oxygen partial pressures. pO2 (mm Hg) renal toxicity. If HBOCs are to become An increase in pH or a decrease clinically effective blood substitutes, two in temperature or 2,3-DPG concentration shifts the oxygen-hemoglobin dissociation curve to the left, resulting in an increased afﬁnity of hemoglobin for oxygen. CONTINUED ON PAGE 1004

1002 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

CONTINUED FROM PAGE 1002 or polyoxyethylene, retarding the rate at which derived from humans or animals. Of it is cleared from the circulation. In this in- course, the problem with this approach is principal challenges must be overcome: stance, the intravascular dwell time can be ex- the cost of scaling up operations sufficient- their short intravascular dwell times, and tended to 48 hours (5). New HBOCs under ly to yield the large volumes of hemoglobin their reduced ability to oxygenate tissues development include polyhemoglobin linked to required for routine transfusions. relative to normal hemoglobin. antioxidant enzymes such as superoxide dis- Several other problems must be addressed The first challenge is to prevent the rapid mutase and catalase in an effort to minimize before HBOCs can become broadly useful breakdown and elimination by the kidney of ischemia-reperfusion injury (6). In addition, transfusion products. Erythrocytes exert little if hemoglobin molecules that are no longer within some groups are trying to package human he- any colloidal osmotic pressure, whereas hemo- the protective environment of the red blood cell. moglobin in artificial red blood cells made from globin (like other plasma proteins) has a colloi- Intracellular hemoglobin has the same life-span biodegradable polymer nanocapsules or lipid dal osmotic effect. Cell-free hemoglobin thus as the erythrocyte, about 120 days, but in solu- vesicles (6). acts as a plasma expander, because the colloidal tion the hemoglobin tetramer readily dissociates There are at least three modified hemo- osmotic pressure it exerts alters the intravascu- into monomers and dimers that are quickly globin products currently in advanced clinical lar volume by more than the volume transfused. eliminated by the kidneys. The dissociation of trials (see the News story by Stokstad, page Polyhemoglobin exerts much less colloidal os- oxygen from intracellular hemoglobin is mod- 1003). Northfield Laboratories has developed motic pressure and can be prepared as an iso-

ified by 2,3-DPG, resulting in a normal P50 Polyheme, a polymerized human hemoglobin oncotic solution (6). Some hemoglobin solu- value (the oxygen partial pressure at which product. Biopure has prepared polymerized tions also have vasopressor effects, that is, they hemoglobin is 50% saturated) of about 27 hemoglobin from bovine red blood cells (He- increase blood pressure and decrease cardiac mmHg for human blood. In hemoglobin solu- mopure), which has received approval for use output. Several studies have indicated that this tions, the lack of 2,3-DPG causes a “left shift” as a blood substitute in the Republic of South vasopressor effect is partly attributable to the in the oxygen dissociation curve, that is, an Africa and is now awaiting FDA review of a ability of cell-free hemoglobin to scavenge ni- increase in the affinity of hemoglobin for oxy- phase III clinical trial in the United States. tric oxide, a cellular chemical messenger that gen. Although this increased affinity does not Finally, Hemosol has developed Hemolink, a stimulates blood vessel relaxation. Although limit the ability of HBOCs to transport oxygen, partially polymerized human hemoglobin this vasopressor effect is undesirable for most it does seriously limit their ability to “unload” product that is also under FDA review. applications, it may turn out to be clinically or deliver oxygen to tissues. There are several problems with HBOCs advantageous for treating a small subpopulation These two challenges have been addressed that must be overcome. The first is the source of of patients with septic shock who suffer from an by chemically modifying cell-free hemoglobin. the hemoglobin used in the blood substitute. An uncontrollable decrease in blood pressure (7). In the first class of modified HBOCs, specific obvious candidate is human hemoglobin de- chemical cross-links are established between rived from outdated donor blood (red blood Perﬂuorocarbon-Based Products hemoglobin polypeptide chains to prevent the cells that have exceeded the approved storage For religious reasons, certain groups of peo- dissociation of the hemoglobin tetramer, thus period), but the limited supply of human blood ple are unable to accept transfusions of either retarding renal elimination. Treatment with 3,5- has compounded difficulties in developing a donor blood or human and animal proteins dibromosalicyl fumarate results in a strong co- human-derived HBOC. A more readily avail- such as hemoglobin. For these patients, per- valent bond that maintains the integrity of the able (and cheaper) source of hemoglobin is fluorocarbons—molecules that are structural- hemoglobin tetramer (5, 6). This results in a bovine blood. Bovine hemoglobin does not ly similar to hydrocarbons except that hydro-

prolonged intravascular dwell time of up to 12 have 2,3-DPG, resulting in a P50 of about 30 gen atoms are replaced with fluorine atoms— hours; by contrast, untreated cell-free hemoglo- mmHg both inside and outside the red blood may be their only option if a blood transfu- bin is eliminated by the kidney in less than 6 cell (similar to that of the intracellular native sion is required. Perfluorocarbon liquids have hours. Alternatively, hemoglobin can be treated human hemoglobin). In addition, once stripped an excellent capacity for carrying oxygen and with bifunctional cross-linking agents, such as of all other proteins, bovine hemoglobin is not carbon dioxide without actually binding to o-raffinose or glutaraldehyde, that target specif- recognized by the human immune system as these gases (3). Through partial liquid venti- ic amino groups and polymerize the hemoglo- foreign. However, one potential obstacle to the lation of the lungs, perfluorocarbons have bin molecule (5, 6). These reagents produce acceptance of a bovine-derived HBOC is the been successfully used to provide oxygen to polyhemoglobin, composed of four or five he- fear that it could harbor the prion pathogen that premature infants with severe respiratory dis- moglobin molecules, which varies in molecular causes bovine spongiform encephalopathy. Ge- tress syndrome. size and configuration and has intravascular netically engineering bacteria to produce a Perfluorocarbons provide an entirely dif- dwell times of up to 24 hours. Finally, hemo- recombinant source of human hemoglobin ferent approach to oxygen transport. In cell- globin can be conjugated to a variety of larger would eliminate the concerns about disease free hemoglobin solutions, oxygen is bound molecules such as dextran, polyethylene glycol, transmission associated with hemoglobin to hemoglobin in the same way as it is bound to the native molecule. In contrast, oxygen readily dissolves in the chemically inert per- Cell-free Fig. 2. Oxygen dissociation curves for fluorocarbon liquid and can be easily extract- hemoglobin cell-free hemoglobin and for a perfluo- ed by oxygen-deprived tissues (Fig. 2). Per- Perfluorocarbons rocarbon emulsion. Oxygen is bound to fluorocarbons are not miscible with aqueous cell-free hemoglobin in the same way as it is bound to native hemoglobin. In solutions and must be prepared as emulsions contrast, oxygen readily dissolves in, before they can be used as blood substitutes. but does not bind to, perfluorocarbon The oxygen loading capacity of perfluorocar- liquids. The oxygen loading capacity of bons is linearly related to the partial pressure perfluorocarbons is linearly related to of oxygen in equilibrium with the emulsion. the partial pressure of oxygen in equi- Oxygen content (ml/dl) Thus, at a given partial pressure of oxygen, librium with the perfluorocarbon. At a Partial pressure of oxygen hemoglobin binds significantly more oxygen pO (mm Hg) given partial pressure of oxygen, hemo- 2 globin binds more oxygen than can be than can be dissolved in the perfluorocarbon dissolved in the perfluorocarbon. (Fig. 2). One perfluorocarbon, Fluosol DA,

1004 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN has been approved by the FDA as a blood (8). Similarly, the risk of transfusion-associ- tive, and difficulties with obtaining and pro- substitute for heart surgery: The fully oxy- ated infection with hepatitis C virus (HCV) is cessing sufficient amounts of these com- genated perfluorocarbon is infused at the time between 1 per 300,000 and 1 per 600,000, pounds must be overcome. of coronary angioplasty to provide oxygen compared with an incidence of 1 per 103,000 delivery during this surgical procedure. Even in the early 1990s before a test for HCV References though the perfluorocarbon reduced cardiac became available. As the safety of the donor 1. J. McCullough, Transfusion Medicine (McGraw-Hill, trauma and pain during surgery (5), its use blood supply continues to improve, there New York, 1998), pp. 91–93. has never been fully embraced by physicians. must be careful consideration of the advan- 2. E. C. Rossi et al.,inPrinciples of Transfusion Medicine, E. C. Rossi et al., Eds. (Williams & Wilkins, Baltimore, Currently, Oxygent, a perfluorocarbon blood tages of blood substitutes over donated blood, 1991), pp. 1–11. substitute developed by Alliance Pharmaceu- particularly given that new blood substitutes 3. S. E. Hill, Can. J. Anaesth. 48, 532 (2001). ticals, is in stage II/III clinical trials in the potentially carry unknown risks. Yet a short- 4. S. F. Rabiner et al., Ann. Surg. 171, 615 (1970). 5. R. M. Winslow, in Scientiﬁc Basis of Transfusion Med- United States. age of donor blood for transfusion still advo- icine, K. C. Anderson, P. M. Ness, Eds. (Saunders, cates for the development of readily available Philadelphia, 2000), pp. 588–598. Risk Versus Beneﬁt blood substitutes. However, there is still a 6. T. M. S. Chang, Blood Substitutes: Principles, Meth- ods, Products and Clinical Trials, Vol. 1 (Karger, New testing and screening procedures have long way to go before artificial blood can Basel, 1997) (www.medicine.mcgill.ca/artcell/ rendered the donor blood supply increasingly replace real blood in routine transfusions. bldsubcr.htm). safe. For example, the risk of transfusion- Most important, the intravascular dwell times 7. J. Creteur et al., Crit. Care Med. 28, 3025 (2000). 8. E. Vanvakas, in Evidence-Based Practice of Transfusion associated HIV infection is now estimated to of blood substitutes need to be increased, the Medicine (AABB Press, Bethesda, MD, 2001), pp. 101– be as low as 1 per 835,000 transfused patients cost of these products needs to be competi- 120.

VIEWPOINT A Bioartiﬁcial Liver—State of the Art Alastair J. Strain* and James M. Neuberger

End-stage liver disease is treated by liver transplantation, but donor organ are attempting to develop BAL devices in shortages remain a serious problem. This has prompted the design of which hepatocytes are optimally maintained bioartiﬁcial liver devices to “bridge” patients until they either recover or so that they carry out as many activities as receive a liver transplant. In these devices, patient plasma is circulated possible. extracorporeally through a bioreactor that houses liver cells (hepatocytes) sandwiched between artiﬁcial plates or capillaries. The Best Type of Cell for a BAL Bioreactor The healthy liver is able to regenerate itself dialysis supports the failing kidney (2, 3). Constituting 70% of the cellular content of after acute injury, but once damaged by The principal goal is to develop a bioartificial the liver, the primary hepatocyte is clearly the fibrosis and cirrhosis—caused by a variety liver (BAL) device in which patient plasma is cell of choice for a BAL bioreactor, particu- of chronic conditions such as alcohol abuse circulated extracorporeally through a bioreac- larly as it carries out most of the liver’s or infection with hepatitis virus B or C—it tor that houses metabolically active liver cells functions. However, when removed from the can no longer regenerate normally (1). Liv- (hepatocytes) sandwiched between artificial complex architecture of the liver, hepatocytes er transplantation is a routine treatment for plates or capillaries. rapidly lose liver-specific gene expression end-stage liver disease, but donor organ What must a BAL bioreactor accomplish? and become phenotypically unstable in cul- shortages remain a serious problem. Many The liver has a number of crucial functions ture. By manipulating tissue culture condi- patients with acute liver failure die while that are principally carried out by hepato- tions—providing an extracellular matrix and waiting for a transplant, and those with cytes. These cells synthesize many proteins, growth factors, and coculturing hepatocytes chronic disease often deteriorate so much including clotting factors; they produce bile with other types of liver cells—much can be that their survival rate after transplantation and regulate carbohydrate, fat, and protein done to maintain hepatocyte stability in vitro is low. Remarkably, “supporting” acute liv- metabolism; they detoxify the ammonia prod- (5). Indeed, cross-talk among hepatocytes er failure patients until their own liver re- uct of nitrogen metabolism and break down (which form the liver parenchyma), non- pairs itself may negate the need for a liver alcohol and drugs. In addition, the liver has parenchymal liver cells, and bile duct epithe- transplant (2, 3). This has generated interest Kupffer cells that are part of the immune lial cells is important for optimal hepatic in designing a “bridging” device that would system. The problem is deciding which liver activity (5) (Fig. 1). support or replace normal liver function until functions are the most important and should The enormous scale-up required to use the patient’s own liver recovered or a donor be carried out by the BAL bioreactor. Bio- BAL devices clinically is problematic: At liver became available for transplant. Alter- synthetic capacity is perhaps the least essen- least 1010 hepatocytes in a BAL bioreactor natively, a bridging device could support a tial because most proteins synthesized by the would be needed to support a patient’s failing failing liver long-term, in the same way that liver can be given exogenously to patients. liver. Given that the proliferative capacity of The generation and detoxification of ammo- primary human hepatocytes and the ability to nia is undoubtedly important. Although an cryopreserve them is limited, there is a seri- Schools of Biosciences and Medicine, University of increase in ammonia in the blood is toxic to ous numbers problem (5). In short, there is Birmingham, and Liver and Hepatobiliary Unit, Uni- versity Hospital Birmingham, Edgbaston, Birmingham the central nervous system, reduction of am- insufficient good-quality human liver tissue B15 2TH, UK. monia levels per se is not sufficient to alle- from which to derive hepatocytes for routine *To whom correspondence should be addressed. E- viate this toxicity (4). In response to this clinical use in a BAL device. One alternative mail: [email protected] panoply of different functions, researchers is to use hepatocytes from other species, par-

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1005 B ODYBUILDING:THE B IONIC H UMAN NEWS Not Blood Simple

After decades of setbacks, compounds that act like blood to deliver oxygen are in the final stretch of clinical trials Until March 1998, Baxter Healthcare Corp. Scientists began working on artificial thought it had a sure-fire winner. The Deer- blood in the 1960s, trying to see whether field, Illinois, company was in phase III trials iron-rich hemoglobin molecules extracted of HemAssist, an oxygen-carrying solution from red blood cells would be able to pick up designed to treat patients in shock from mas- oxygen molecules and deliver them to tis- might cause other problems.” Other experts sive bleeding. Analysts were excited because sues when transfused into a patient. Those stress that more needs to be known about how blood substitutes—more accurately known as early compounds ran into problems with kid- modified hemoglobin delivers oxygen to tissue oxygen therapeutics—could be stockpiled and ney toxicity. Work took off in the 1980s, with before an ideal substitute can be designed. given to anyone regardless of blood type. the realization that HIV, the virus that causes The novelty of blood substitutes has also But after enrolling about 100 people who AIDS, had contaminated the blood supply. raised questions about how to test them for ef- had been knifed, shot, or critically injured in “The urgency was such that industries tried ficacy and safety. “There is no golden standard car accidents, Baxter discovered that the pa- to bypass basic research and proceeded im- to compare blood substitutes to,” says Konrad tients given HemAssist were more likely to die mediately into development and clinical tri- Messmer, a professor of experimental surgery at Ludwig Maximilians University in Munich, BLOOD SUBSTITUTES IN ADVANCED CLINICAL TRIALS* Germany. (FDA uses surrogate endpoints, Product Source Use Half-life† Shelf life Side effects‡ such as a decreased need for transfused blood.) (Company) Another concern is the source of the Hemopure Modified Elective 24–36 2 years at Vasoconstriction, hemoglobin. Most products in advanced tri- (Biopure) bovine surgery hours room transient increase in als purify hemoglobin from collected human hemoglobin temperature liver and pancreas enzymes blood that has passed its expiration date. In PolyHeme Modified Trauma, 24 hours 1 year None reported contrast, Hemopure, made by Biopure Corp. (Northfield human urgent refrigerated in Cambridge, Massachusetts, is made from Laboratories) hemoglobin blood loss cow blood. Biopure says it uses only U.S. Hemolink Modified Heart 18–20 1 year Vasoconstriction, cattle bound for slaughter and controls their (Hemosol) human surgery hours refrigerated yellowing of skin feeding and care. The company says its man- hemoglobin ufacturing process removes any infectious on February 3, 2016 Oxygent Perfluoro- General 24–48 2 years Slight rise in body agents that might slip through, such as the (Alliance) chemical surgery hours refrigerated temperature, drop in emulsion platelet count prions implicated in bovine spongiform encephalopathy, or mad cow disease. Last * In the United States. † Dose-dependent. ‡ All transient. April, Hemopure was approved for anemia during surgery in South Africa. than those given regular blood. “They pretty als,” recalls blood substitute pioneer Thomas For products further upstream, investment much had a disaster,” says Reuven Rabinovici, Chang of McGill University in Montreal. capital is harder to come by now that the blood

chief of trauma and surgical critical care at This generation of compounds constricted supply is more safe from HIV. “Fifteen years Downloaded from Yale University School of Medicine. Baxter blood vessels and capillaries, which can ago, there was almost a bottomless well of cap- called off the trial. And although the problem limit circulation to oxygen-starved tissue. ital,” says Robert Winslow of Sangart Inc. in may have been faulty trial design rather than Half of the companies with compounds in San Diego. “The capital markets for this have HemAssist itself, the company soon ditched advanced trials still report vasoconstriction. almost dried up.” Sangart relies heavily on NIH its decade-long development effort. The reason is not entirely clear, but many re- funding and is planning a phase I trial in Swe- Most clinical testing involves disappoint- searchers suspect that these compounds pene- den for a hemoglobin molecule modified with ments, but blood substitutes have had a par- trate the lining of the blood vessels and scav- strands of polyethylene glycol, designed to pro- ticularly tortured history. Four decades after enge nitric oxide (NO), a molecule that would tect the molecule and prevent vasoconstriction. research began, the U.S. Food and Drug Ad- otherwise relax the muscular walls surround- The Defense Advanced Research Projects ministration (FDA) has approved only one ing blood vessels. Others believe that exces- Agency is also in the game, funding research blood substitute—for treating anemia in sive diffusion of hemoglobin within the vessel into freeze-dried blood products, for example. dogs. Longtime researchers say the main rea- is the key. In any case, scientists have reduced Any new products that reach the market son for the delay is the complexity of hemo- this problem—in animal studies and early hu- will have to compete with actual blood, globin and its sometimes surprising effects man trials—by modifying or clumping hemo- Winslow notes, which is now very safe. If on humans when it’s not ensconced in red globin molecules to make them larger. Mean- blood substitutes are approved in the United blood cells. Also contributing to the rocky while, Baxter is working on recombinant hu- States for human use, they’re likely to fetch a progress are the challenges of regulating a man hemoglobin that doesn’t scavenge NO. premium over blood, which can cost hospitals novel type of material, as well as ups and Harvey Klein, chief of transfusion med- $200 or more per unit. Biopure hasn’t set a downs in Wall Street enthusiasm. A handful icine at the National Institutes of Health’s price yet but estimates that each unit will sell of companies persevere, predicting an ever (NIH’s) Clinical Center in Bethesda, Maryland, for perhaps $700 to $1000. Whatever the cost, greater need for blood substitutes as growing says he is less worried about vasoconstriction researchers and clinicians want to have supplies numbers of elderly patients needing blood- than he is about the incomplete understanding of blood substitutes stocked for emergencies— intensive surgery such as hip repairs or re- of why the side effect occurs: “It’s telling us that including the discovery of new infectious

CREDIT: BIOPURE placements strain the existing donor supply. something is happening at a basic level that agents in donated blood. –ERIK STOKSTAD

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1003 B ODYBUILDING:THE B IONIC H UMAN has been approved by the FDA as a blood (8). Similarly, the risk of transfusion-associ- tive, and difficulties with obtaining and pro- substitute for heart surgery: The fully oxy- ated infection with hepatitis C virus (HCV) is cessing sufficient amounts of these com- genated perfluorocarbon is infused at the time between 1 per 300,000 and 1 per 600,000, pounds must be overcome. of coronary angioplasty to provide oxygen compared with an incidence of 1 per 103,000 delivery during this surgical procedure. Even in the early 1990s before a test for HCV References though the perfluorocarbon reduced cardiac became available. As the safety of the donor 1. J. McCullough, Transfusion Medicine (McGraw-Hill, trauma and pain during surgery (5), its use blood supply continues to improve, there New York, 1998), pp. 91–93. has never been fully embraced by physicians. must be careful consideration of the advan- 2. E. C. Rossi et al.,inPrinciples of Transfusion Medicine, E. C. Rossi et al., Eds. (Williams & Wilkins, Baltimore, Currently, Oxygent, a perfluorocarbon blood tages of blood substitutes over donated blood, 1991), pp. 1–11. substitute developed by Alliance Pharmaceu- particularly given that new blood substitutes 3. S. E. Hill, Can. J. Anaesth. 48, 532 (2001). ticals, is in stage II/III clinical trials in the potentially carry unknown risks. Yet a short- 4. S. F. Rabiner et al., Ann. Surg. 171, 615 (1970). 5. R. M. Winslow, in Scientiﬁc Basis of Transfusion Med- United States. age of donor blood for transfusion still advo- icine, K. C. Anderson, P. M. Ness, Eds. (Saunders, cates for the development of readily available Philadelphia, 2000), pp. 588–598. Risk Versus Beneﬁt blood substitutes. However, there is still a 6. T. M. S. Chang, Blood Substitutes: Principles, Meth- ods, Products and Clinical Trials, Vol. 1 (Karger, New testing and screening procedures have long way to go before artificial blood can Basel, 1997) (www.medicine.mcgill.ca/artcell/ rendered the donor blood supply increasingly replace real blood in routine transfusions. bldsubcr.htm). safe. For example, the risk of transfusion- Most important, the intravascular dwell times 7. J. Creteur et al., Crit. Care Med. 28, 3025 (2000). 8. E. Vanvakas, in Evidence-Based Practice of Transfusion associated HIV infection is now estimated to of blood substitutes need to be increased, the Medicine (AABB Press, Bethesda, MD, 2001), pp. 101– be as low as 1 per 835,000 transfused patients cost of these products needs to be competi- 120.

VIEWPOINT A Bioartiﬁcial Liver—State of the Art Alastair J. Strain* and James M. Neuberger

End-stage liver disease is treated by liver transplantation, but donor organ are attempting to develop BAL devices in shortages remain a serious problem. This has prompted the design of which hepatocytes are optimally maintained bioartiﬁcial liver devices to “bridge” patients until they either recover or so that they carry out as many activities as receive a liver transplant. In these devices, patient plasma is circulated possible. on February 3, 2016 extracorporeally through a bioreactor that houses liver cells (hepatocytes) sandwiched between artiﬁcial plates or capillaries. The Best Type of Cell for a BAL Bioreactor The healthy liver is able to regenerate itself dialysis supports the failing kidney (2, 3). Constituting 70% of the cellular content of after acute injury, but once damaged by The principal goal is to develop a bioartificial the liver, the primary hepatocyte is clearly the fibrosis and cirrhosis—caused by a variety liver (BAL) device in which patient plasma is cell of choice for a BAL bioreactor, particu- of chronic conditions such as alcohol abuse circulated extracorporeally through a bioreac- larly as it carries out most of the liver’s or infection with hepatitis virus B or C—it tor that houses metabolically active liver cells functions. However, when removed from the Downloaded from can no longer regenerate normally (1). Liv- (hepatocytes) sandwiched between artificial complex architecture of the liver, hepatocytes er transplantation is a routine treatment for plates or capillaries. rapidly lose liver-specific gene expression end-stage liver disease, but donor organ What must a BAL bioreactor accomplish? and become phenotypically unstable in cul- shortages remain a serious problem. Many The liver has a number of crucial functions ture. By manipulating tissue culture condi- patients with acute liver failure die while that are principally carried out by hepato- tions—providing an extracellular matrix and waiting for a transplant, and those with cytes. These cells synthesize many proteins, growth factors, and coculturing hepatocytes chronic disease often deteriorate so much including clotting factors; they produce bile with other types of liver cells—much can be that their survival rate after transplantation and regulate carbohydrate, fat, and protein done to maintain hepatocyte stability in vitro is low. Remarkably, “supporting” acute liv- metabolism; they detoxify the ammonia prod- (5). Indeed, cross-talk among hepatocytes er failure patients until their own liver re- uct of nitrogen metabolism and break down (which form the liver parenchyma), non- pairs itself may negate the need for a liver alcohol and drugs. In addition, the liver has parenchymal liver cells, and bile duct epithe- transplant (2, 3). This has generated interest Kupffer cells that are part of the immune lial cells is important for optimal hepatic in designing a “bridging” device that would system. The problem is deciding which liver activity (5) (Fig. 1). support or replace normal liver function until functions are the most important and should The enormous scale-up required to use the patient’s own liver recovered or a donor be carried out by the BAL bioreactor. Bio- BAL devices clinically is problematic: At liver became available for transplant. Alter- synthetic capacity is perhaps the least essen- least 1010 hepatocytes in a BAL bioreactor natively, a bridging device could support a tial because most proteins synthesized by the would be needed to support a patient’s failing failing liver long-term, in the same way that liver can be given exogenously to patients. liver. Given that the proliferative capacity of The generation and detoxification of ammo- primary human hepatocytes and the ability to nia is undoubtedly important. Although an cryopreserve them is limited, there is a seri- Schools of Biosciences and Medicine, University of increase in ammonia in the blood is toxic to ous numbers problem (5). In short, there is Birmingham, and Liver and Hepatobiliary Unit, Uni- versity Hospital Birmingham, Edgbaston, Birmingham the central nervous system, reduction of am- insufficient good-quality human liver tissue B15 2TH, UK. monia levels per se is not sufficient to alle- from which to derive hepatocytes for routine *To whom correspondence should be addressed. E- viate this toxicity (4). In response to this clinical use in a BAL device. One alternative mail: [email protected] panoply of different functions, researchers is to use hepatocytes from other species, par-

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1005 B ODYBUILDING:THE B IONIC H UMAN ticularly pigs. However, there is still an on- Designing a BAL Bioreactor three sets (bundles) of interwoven capillaries going debate about the risk of zoonoses, an- The basic BAL bioreactor consists of a col- that create a three-dimensional extracapillary imal infections transmitted by cells from oth- umn containing a collection of hollow-fiber space in which to house hepatocytes. One set er species (xenogeneic cells) (6). Porcine- capillaries through which patient plasma is of capillaries provides an efficient oxygen- derived hepatocytes are being used in clinical pumped (Fig. 2). Hepatocytes are inoculated ation supply, and the remaining two sets con- studies of BAL devices in the United States, into the extracapillary space (8, 18) either trol inflow and outflow of plasma, respective- but such studies remain on hold in the U.K. alone or attached to microcarrier beads (19). ly. This BAL bioreactor provides an opti- and most other European countries. In the secondary circuit, patient (or animal) mized compartment for hepatocytes that en- Using cultured hepatic cell lines over- plasma is separated, warmed, oxygenated, courages them to reorganize into functional comes the human hepatocyte shortage, al- and then perfused through the lumen of the cellular aggregates (20). The downside is that though it is still hotly contended whether the bioreactor capillaries (Fig. 2). This allows the Gerlach bioreactor is extremely complex hepatic lines currently available retain suffi- free exchange of molecules between hepato- to run. Another design by Chamuleau and cient functional capacity for use in BAL de- cytes and patient plasma: Hepatocytes extract colleagues (Amsterdam) incorporates a spi- vices. One solution may be to combine sev- oxygen and nutrients and detoxify chemicals rally wound polyester matrix sheet that in- eral different liver cell lines in the bioreactor in the plasma, and their metabolites pass into cludes an integral hollow-fiber compartment and thus exploit their functional complemen- the plasma. The ideal BAL design must en- for oxygenation (21). This design again cre- tarity. Alternatively, it should be possible to sure optimal ex vivo maintenance of hepato- ates an environment that in theory is more genetically engineer hepatic cell lines to have cytes. It has been assumed, but not yet prov- amenable to the optimal maintenance of the desired capabilities. For example, one en, that simple “flow-through” BAL systems hepatocytes, enabling them to retain many of recent study reported reversible immortaliza- achieve this. However, given that conven- their functions. Together with the Sussman tion of a human hepatocyte line using the tional monolayer cultures cannot optimally (18) and Demetriou (19) systems, the Cre-Lox genetic engineering system, which maintain hepatocytes, it is likely that hepato- Chamuleau and Gerlach BAL devices are allowed excision of the immortalizing gene, cytes will need to be induced to form cellular being tested in clinical trials. Other BAL thus removing the threat of tumor formation aggregates in which they reacquire their po- bioreactor designs under development in- (7). The safety aspects of using genetically larization (that is, the asymmetric localization clude a flat-bed capillary arrangement that altered and potentially tumorigenic liver cells of surface proteins that enables different parts incorporates a collagen gel system to support in the clinic still needs to be addressed. De- of the cell to carry out different functions). hepatocytes, and a series of flat stacking spite these reservations, one experimental A more sophisticated BAL bioreactor has plates or other radial flow elements (22–24). program is clinically testing a BAL device been developed in Berlin by Gerlach and Still other BAL bioreactors are under devel- containing the C3A hepatocyte line, a sub- colleagues (20). This bioreactor comprises opment in the laboratory (2, 3) but are not yet clone of the ubiquitous HepG2 hepatoblas- toma cell line (8). Fig. 1. Cellular archi- Liver stem cells, or even stem cells from tecture of the liver. Liv- other tissues, could potentially provide an er epithelial cells called alternative source of human hepatocytes. In hepatocytes (pink) are addition to bone marrow, stem cells reside in arranged in cords be- adult tissues such as the liver and central tween the capillaries nervous system, and have much greater plas- (sinusoids) of the liver. Oxygenated blood en- ticity than previously realized (9, 10). Inves- ters the liver from the tigators are trying to define the factors heart via the hepatic controlling the differentiation of stem cells artery (red) and from into the cell type of choice. The key will be the gut via the hepatic to determine conditions under which stem portal vein (blue), mix- cells can be clonally expanded in an undif- es in the sinusoids, and drains via the hepatic ferentiated state and then quickly central vein back to “switched” to the desired phenotype. Al- the heart. Sinusoidal though the differentiation of bone marrow– cells (purple)—includ- derived stem cells into hepatocytes has ing endothelial cells, been demonstrated in vivo (11–13), it has Kupffer cells, and stel- not been achieved in vitro as yet. However, late cells—line the sinusoids, thus separat- liver-derived hematopoietic (blood) cells ing hepatocytes from can give rise to cells that resemble bile duct blood. The hepatocytes epithelial cells (14). Embryonic stem (ES) are only exposed to cells offer another valuable option, partic- plasma, allowing ex- ularly as human ES cell lines are now change of plasma pro- available (15–17). Differentiation of hu- teins, nutrients, and metabolites. Bile (dark man ES cells into cardiomyocytes and neu- green arrows) is syn- ral progenitor cells has been reported, tak- thesized by hepato- ing us a step closer to the reality of using cytes and secreted into them as a viable source of human hepato- small channels (canal- cytes. Just as vital as finding a suitable iculi; pale green) be- source of hepatocytes is the need to devel- tween adjacent cells. The bile collects in ducts formed by biliary epithelial cells and drains from the liver into the gallbladder op a workable culture system where hepa- (or, if the gallbladder has been removed, bile drains directly into the gut). The challenge of a BAL tocytes (wherever they come from) can be bioreactor is to attempt to partly recreate the complex tissue architecture of the liver with all of its optimally maintained on a large scale. myriad functions. [Illustration: Nathalie Cary]

1006 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN sufficiently advanced for testing in the clinic. (26). Despite the huge difficulties associated whom have acute liver failure. This study is There is an enormous literature (too ex- with designing and running clinical trials and important because it is the first large-scale tensive to review here) describing the testing the problems of selecting clinically relevant trial conducted in both U.S. and European of BAL devices in vitro and in animal models endpoints, some successful treatment regi- centers that attempts to demonstrate efficacy. (2, 3, 8, 18, 19). Many different parameters mens have been reported. The most promis- Although preliminary results are described as have been measured as indicators of hepato- ing clinical results are those of Sussman’s encouraging, a detailed analysis of the full cyte viability and function, including second- group with the C3A human hepatoma cell study awaits critical appraisal and peer re- ary endpoints such as albumin or clotting line, and of Demetriou and colleagues with view. Two other BAL systems (the Amster- factor production, urea synthesis, conjugation primary porcine hepatocytes, each using their dam and Berlin devices) have been tested of the bile pigment bilirubin, drug metabo- own BAL bioreactor (27, 28). Ideal patients clinically in preliminary, and as yet unpub- lism, and a variety of other metabolic pro- for treatment are those suffering from acute lished, clinical trials with small numbers of cesses. BAL bioreactors have been tested in a liver failure, who need to be “bridged” until patients. Both BAL devices contained pig number of different small and large animal they can receive a liver transplant or until hepatocytes and were used to treat patients models of liver failure. For example, Flendrig they recover (2, 3, 27, 28). with acute liver failure. et al. recently reported prolonged survival of Although the safety and feasibility of pigs whose livers were subjected to severe BAL devices were quickly established, dem- The Future ischemia by surgical manipulation of the onstrating efficacy has been much more dif- The greatest challenge for the BAL bioreactor blood supply (25). In a carefully controlled ficult. We await the outcome of randomized, is how best to maintain viable functional hepa- study, ischemic animals were treated with the controlled clinical trials. The first such trial, tocytes outside of the body. Perhaps we can Chamuleau bioreactor system housing 1.4 ϫ consisting of 24 acute liver failure patients look to the biomaterials scientists (see page 1010 porcine hepatocytes. Blood ammonia (12 treated with the Sussman BAL device 1014) (31) to provide us with biomaterials that and bilirubin levels decreased significantly containing C3A cells, and 12 controls), has could induce hepatocytes to retain their epithe- and the longest surviving animal was kept been completed in the U.K. (29). The safety lial polarization. In addition, tissue engineering alive for 63 hours (25). Thus, there is no and feasibility of the BAL bioreactor was strategies (see page 1009) (32) could be used to doubt that hepatocytes can and do function demonstrated, but there was no clinical sur- coax hepatocytes to reorganize into functional satisfactorily in various BAL systems, but vival benefit. This result could be at least three-dimensional aggregates in vitro. Interac- interpreting the data from experimental stud- partly explained by a higher than predicted tions among the different types of hepatic cell ies and extrapolating them to the clinic has survival in the control group. Clearly, in- populations are essential for the liver to operate proved difficult. creased patient numbers offer the only means appropriately. Coculture of hepatocytes with of resolving the question of efficacy. A larger nonparenchymal liver cells has been shown to Clinical Trials multicenter trial, again with the C3A cell line, be beneficial (5). The downside is that inclusion The first clinical use of a BAL bioreactor in is underway in the U.S. and U.K. (under the of other cell types would inevitably make BAL 1987 was to treat a single patient with acute direction of Vitagen). design, construction, and handling even more liver failure. The patient’s serum bilirubin Results of a second, more extensive clin- complex. levels fell, his neurological condition im- ical trial with Demetriou’s porcine hepato- Bile production by hepatocytes is not nor- proved, and he was discharged from hospital, cyte system were recently reported at the mally reproduced in a BAL bioreactor. It is thus demonstrating the potential value of the American Association for the Study of Liver possible that some of the net products of bile approach (26). This BAL device consisted of Diseases meeting in Dallas (30). This major production and metabolite excretion could be an adapted artificial kidney dialysis unit load- multicenter trial (supported by Circe Biomed- toxic to hepatocytes if not removed. Investi- ed with cryopreserved rabbit hepatocytes ical) includes more than 170 patients, most of gators tend to assume, indeed hope, that these potentially toxic intermediates, if made, are simply washed away or are sufficiently dilut- ed that they do not affect the hepatocytes. It is certainly hopelessly ambitious to engineer bile drainage into the current generation of bioreactors. Intriguingly, it has recently been reported that bile duct–like structures form in cultures of bile epithelial cells (33, 34). Even if these design challenges are overcome, we still face the cell supply problem. Looking into the future, ideally, a stem cell inoculum would be added to the BAL bioreactor of choice and allowed to expand before the appropriate molecular switches are activated, with the induction of the required hepatic phenotype and the assembly of hepatocytes and other types of liver cells into organized liver tissue. Finally, another issue that must be considered is ease of handling. In the acute liver Fig. 2. A BAL bioreactor. In the primary circuit, the patient is connected by venous catheters to a failure situation, a bioartificial liver machine plasma separator. In the secondary circuit, the plasma is then passed through a heat exchanger and must be available immediately without the oxygenator (some bioreactors include integral oxygenation ﬁbers) before going through the lumen of the capillaries in the bioreactor column (bottom right and top image). Up to 1010 human or need for lengthy preparation times. It must be porcine hepatocytes are housed in the extracapillary spaces of the bioreactor column. Blood cells are added back to the plasma before it is returned to the patient. [Photo courtesy of R. Chamuleau] CONTINUED ON PAGE 1009

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1007 B ODYBUILDING:THE B IONIC H UMAN

CONTINUED FROM PAGE 1007 4. A. S. Hazell, R. F. Butterworth, Proc. Soc. Exp. Biol. 20. J. Gerlach et al., Hepatology 22, 246 (1995). Med. 222, 99 (1999). 21. L. M. Flendrig et al., J. Hepatol. 26, 1379 (1997). “user friendly” for the critical care nursing 5. A. J. Strain, Gut 35, 433 (1994). 22. L. De Bartolo et al., Biotechnol. Prog. 16, 102 (2000). 6. R. A. Weiss, Philos. Trans. R. Soc. London Ser. B 356, 23. P. D. Hay et al., Artif. Organs 24, 278 (2000). and medical staffs, who are under inordinate 957 (2001). 24. M. Kawada et al., In Vitro Cell Dev. Biol. 34, 109 stress when dealing with acutely ill patients 7. N. Kobayashi et al., Science 287, 1258 (2000). (1998). and when decision times are short. These are 8. N. L. Sussman et al., Hepatology 16, 60 (1992). 25. L. M. Flendrig et al., Int. J. Artif. Organs 22, 701 the challenges faced collectively by bioengi- 9. J. M. W. Slack, Science 287, 1431 (2000). (1999). 10. D. J. Anderson, F. H. Gage, I. L. Weissman, Nature 26. K. N. Matsumura et al., Surgery 101, 99 (1987). neers, cell biologists, and clinicians alike as Med. 4, 393 (2001). 27. N. L. Sussman et al., J. Clin. Gastroenterol. 18, 320 we look to the future of the BAL bioreactor in 11. B. E. Petersen et al., Science 284, 1168 (1999). (1994). 12. N. D. Theise et al., Hepatology 32, 11 (2000). clinical medicine. 28. F. D. Watanabe et al., Ann. Surg. 225, 484 (1997). 13. E. LaGasse et al., Nature Med. 6, 1229 (2000). 29. A. J. Ellis et al., Hepatology 24, 1446 (1996). 14. H. A. Crosby, D. A. Kelly, A. J. Strain, Gastroenterology References and Notes 120, 534 (2001). 30. A. C. Stevens et al., Hepatology 34, 299A (2001). 1. A. J. Strain, A. Diehl, Eds., Liver Growth and Repair 15. M. Amit et al., Dev. Biol. 227, 271 (2000). 31. L. L. Hench, J. M. Polak, Science 295, 1014 (2002). (Chapman & Hall, London, 1998), pp. 3–27 and 558– 16. B. E. Reubinoff et al., Nature Biotechnol. 18, 399 32. L. Grifﬁth, G. Naughton, Science 295, 1009 (2002). 576. (2000). 33. A. Sirica et al., Hepatology 26, 537 (1997). 2. S. Nyberg et al., Am. J. Surg. 166, 512 (1993). 17. I. Kehat et al., J. Clin. Invest. 108, 407 (2001). 34. M. Auth et al., Hepatology 33, 519 (2001). 3. J. W. Allen, T. Hassanein, S. N. Bhatia, Hepatology 34, 18. H. O. Jauregui et al., Hepatology 21, 460 (1995). 35. Supported by the Wellcome Trust and the UK Bio- 447 (2001). 19. J. Rozga et al., Hepatology 17, 258 (1993). technology and Biological Sciences Research Council.

VIEWPOINT Tissue Engineering—Current Challenges and Expanding Opportunities Linda G. Grifﬁth1* and Gail Naughton2

Tissue engineering can be used to restore, maintain, or enhance tissues tissues assembled in vitro from cells and and organs. The potential impact of this ﬁeld, however, is far broader—in scaffolds; and (iii) in situ tissue regenera- the future, engineered tissues could reduce the need for organ replace- tion. For cellular implantation, individual ment, and could greatly accelerate the development of new drugs that cells or small cellular aggregates from the may cure patients, eliminating the need for organ transplants altogether. patient or a donor are either injected into the damaged tissue directly or are com- ranging from seconds to weeks and dimensions bined with a degradable scaffold in vitro Introduction ranging from 0.0001 to 10 cm. Coaxing cells to and then implanted. For tissue implanta- The field of tissue engineering exploits living form tissues in a reliable manner is the quintes- tion, a complete 3D tissue is grown in vitro cells in a variety of ways to restore, maintain, or sential engineering design problem that must be using patient or donor cells and a scaffold, enhance tissues and organs (1, 2). Tissue engi- accomplished under the classical engineering and then is implanted once it has reached neering conjures up visions of organs built from constraints of reliability, cost, government reg- “maturity.” For in situ regeneration, a scaf- scratch in the laboratory, ready to be transplant- ulation, and societal acceptance. fold implanted directly into the injured tis- ed into desperately ill patients. The potential Even though fewer than five engineered tissue stimulates the body’s own cells to pro- impact of this field, however, is far broader— sues have been approved by the Food and Drug mote local tissue repair. in the future, engineered tissues could reduce Administration (FDA), more than 70 companies Sources of cells for implantation include the need for organ replacement, and could are spending a total of $600 million per year to autologous cells from the patient, allogeneic greatly accelerate the development of new develop new products (2). There are still many cells from a human donor who is not immu- drugs that may cure patients, eliminating the technical challenges to overcome before we cre- nologically identical to the patient, and xeno- need for organ transplants altogether. ate “off-the-shelf ” tissues that represent the geneic cells from a different species. Each To engineer living tissues in vitro, cultured translation of scientific discoveries into treat- category may be further delineated in terms cells are coaxed to grow on bioactive degradable ments for millions of patients. The successful of whether the cells are adult or embryonic scaffolds that provide the physical and chemical large-scale production of engineered tissues re- stem cells (capable of both self renewal and cues to guide their differentiation and assembly quires an adequate source of healthy expandable differentiation into a variety of cell lineages), into three-dimensional (3D) tissues (3). The as- cells, the optimization of scaffolds, and the cre- or a mixture of differentiated cells at different sembly of cells into tissues is a highly orches- ation of bioreactors, which mimic the environ- stages of maturation (including rare stem and trated set of events that requires time scales ment of the body and that are amenable to progenitor cells). Some approaches use cell scale-up. Additional challenges include the pres- mixtures, whereas others rely on separation ervation of the product so that it has a long or enrichment of stem cells. 1 Department of Chemical Engineering, Division of shelf-life and the successful use of various ap- Although the prospect of using xenogene- Bioengineering and Environmental Health, and Bio- technology Process Engineering Center, Massachu- proaches to prevent tissue rejection. ic cells for tissue repair remains controversial setts Institute of Technology, Cambridge, MA 02139, because of the potential for transmitting ani- USA. 2Advanced Tissue Sciences, La Jolla, CA 92037, Biological Challenges: Cells and Their mal pathogens to humans, xenogeneic cells USA. Sources could perhaps temporarily support an ailing *To whom correspondence should be addressed at There are three principal therapeutic strat- tissue until either a human donor organ be- the Department of Chemical Engineering, Division of egies for treating diseased or injured tissues comes available for transplant, or the tissue Bioengineering and Environmental Health, Massachu- setts Institute of Technology 66-466, Cambridge, MA in patients: (i) implantation of freshly iso- repairs itself. For example, pig liver cells 02139, USA. E-mail: [email protected] lated or cultured cells; (ii) implantation of (hepatocytes) grown in extracorporeal biore-

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1009 B ODYBUILDING:THE B IONIC H UMAN

NEWS Wanted: Pig Transplants That Work

A shortage of human organs spurred researchers to look for Massachusetts, company running the tri- replacements from animals. They’re still struggling to cross the als, reported in November that the treat- species barrier successfully ment prolonged survival for a majority of the patients, buying them more In the early 1960s, Keith Reemtsma, a sur- Fritz Bach, a professor of time to find a human donor. geon at Tulane University in New Orleans, surgery at Harvard Med- At least as daunting as immune transplanted chimpanzee kidneys into six pa- ical School in Boston, rejection is the concern—fueled tients and launched the modern era of xeno- who suspects that other by mad cow disease and the appar- transplantation. One person survived for a antibodies and mecha- ent emergence of HIV from a pri- startling 9 months, sparking worldwide en- nisms may play a role in mate virus—that xenotransplanta- thusiasm for a potentially limitless supply of immune rejection. tion of any type could transmit virus- lifesaving organs and tissue. Foreign cells and tissues es harmless to pigs but deadly to hu- Since then, the field of xenotransplanta- are considered less risky to mans. For instance, all pigs carry tion—the transfer of organs, tissues, and cells transplant. Unlike whole or- Desperate measures. A porcine endogenous retrovirus from one species to another—has been hum- gans, they lack blood ves- baboon heart beat in Baby (PERV), but its potential effects on bled by a string of false starts and failed pro- sels and can evade some of Fae until she died 3 weeks people are unknown. Repeated test- cedures. Nearly 4 decades after Reemtsma’s the immune system’s revolt. after receiving it. ing, including a 1999 study of 160 transplants, no human recipient of a whole an- One convenient target for xenotransplant patients, has turned imal organ is known to have survived for as such xenotransplants is the brain and spinal up no sign of PERV. Still, no one can prove long as that early kidney patient. Twelve-day- cord, which are not subject to the same im- that PERV or another virus won’t jump to hu- old Baby Fae, for instance, received a baboon mune system scrutiny as the rest of the body. mans and cause disease, and health regulato- heart in 1984 but died 3 weeks later. Whole- In clinical trials since 1995, Diacrin, a compa- ry agencies have reacted accordingly. organ transplants into humans have been ny in Charlestown, Massachusetts, has im- The U.S. Food and Drug Administration largely abandoned in industrialized nations planted porcine neurons into the brains and (FDA) will permit xenotransplant trials on a due to massive immune rejection and other more recently spinal cords of more than 40 case-by-case basis, but it requests consent safety worries, although researchers hope people with neurological disorders; results so forms warning patients of troubling un- they may someday be practicable. Clinical tri- far are equivocal. A study on stroke victims known risks. “The patient consent form

als of cell and tissue xenotransplants, mean- was halted after two suffered seizures. Perhaps makes it sound like they’re on the verge of on February 3, 2016 while, are under way in Europe and the Unit- Diacrin’s greatest success emerged from au- unleashing some plague on the world,” says ed States, where they are tightly regulated. topsies of two volunteers whose deaths were Thomas Fraser, CEO of Diacrin. For in- Thus far, results have been mixed. stance, the FDA asks that consent forms Since abandoning chimpanzees as donors, warn subjects of potential risks to their fam- scientists have turned to pigs, whose organs re- ily, particularly during sexual contact, and semble those of humans in size and function. inform patients that they must submit to life- The one animal-to-human transplant routinely long monitoring (although the agency lacks

performed worldwide employs pig heart the legal means to enforce this). Downloaded from valves. Because the valves are chemically Although stringent, FDA offers more lee- treated to kill living cells, most nations do not way than many other countries. Italy and Ger- regulate their use as xenotransplants. many allow animal tissue to be used outside But living pig tissue is another story. Sci- the body but forbid implants. Japan bans entists have known for years about deadly xenotransplants outright, and rules in France immune responses generated by intact pig and England are so strict that study applica- organs. So-called hyperacute rejection oc- tions are rarely submitted. Norway and the curs when human antibodies swoop in on a Now, get to work. Pig nerve cell fibers (white Netherlands have moratoriums in place. “I porcine sugar called α-galactose (α-gal) that threads) survived in the brain of a patient with find it excessive,” says Henri Bismuth, a liver studs the surface of pig cells. The antibodies Parkinson’s disease. transplant surgeon at Paul Brousse Hospital in also trigger acute vascular rejection, where- Villejuif, France, of the French situation. The by the host attacks introduced porcine blood unrelated to the study: Millions of pig cells intergovernmental Council of Europe is work- vessels. T cells and possibly macrophages that had been implanted in the central nervous ing to establish common standards and in- also work to thwart the organ, as can happen system had survived long term. A Swedish clude countries such as Russia and China, in same-species transplants. study of porcine islet cell implants into the where cell xenotransplants may have been In January, researchers reported one ad- pancreas, however, found that few endured. performed with little government oversight. vance toward preventing hyperacute and vas- Another approach to using animal tissue is Xenotransplantation research continues cular rejection: They engineered piglets with to avoid transplants altogether. At hospitals in in labs worldwide, but many agree that stem just one gene for α-gal instead of the usual Germany, Italy, the Netherlands, Spain, Bel- cells and bioengineering hold at least as two (Science, 4 January, p. 25). Pigs with both gium, and the United States, patients with acute much promise. Annika Tibell, chair of trans- genes missing, which researchers hope to liver failure have been hooked up to a device plantation surgery at Huddinge Hospital in breed from these animals, could make organ that uses porcine liver cells to treat the patient’s Stockholm, Sweden, hopes to see “several xenotransplants more viable. But “there’s a blood. A phase II/III study of 171 patients re- different solutions to the organ problem.”

spectrum of opinion” on the subject, says cently ended; Circe Biomedical, the Lexington, –JENNIFER COUZIN UNIVERSITY MEDICAL LOMA LINDA CENTER;CREDITS: BOTTOM) TO H. JONATHAN (TOP DINSMORE

1008 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

VIEWPOINT Tissue Engineering—Current Challenges and Expanding Opportunities Linda G. Grifﬁth1* and Gail Naughton2

Tissue engineering can be used to restore, maintain, or enhance tissues tissues assembled in vitro from cells and and organs. The potential impact of this ﬁeld, however, is far broader—in scaffolds; and (iii) in situ tissue regenera- the future, engineered tissues could reduce the need for organ replace- tion. For cellular implantation, individual ment, and could greatly accelerate the development of new drugs that cells or small cellular aggregates from the may cure patients, eliminating the need for organ transplants altogether. patient or a donor are either injected into on February 3, 2016 the damaged tissue directly or are com- ranging from seconds to weeks and dimensions bined with a degradable scaffold in vitro Introduction ranging from 0.0001 to 10 cm. Coaxing cells to and then implanted. For tissue implanta- The field of tissue engineering exploits living form tissues in a reliable manner is the quintes- tion, a complete 3D tissue is grown in vitro cells in a variety of ways to restore, maintain, or sential engineering design problem that must be using patient or donor cells and a scaffold, enhance tissues and organs (1, 2). Tissue engi- accomplished under the classical engineering and then is implanted once it has reached neering conjures up visions of organs built from constraints of reliability, cost, government reg- “maturity.” For in situ regeneration, a scaf- scratch in the laboratory, ready to be transplant- ulation, and societal acceptance. fold implanted directly into the injured tis- Downloaded from ed into desperately ill patients. The potential Even though fewer than five engineered tissue stimulates the body’s own cells to pro- impact of this field, however, is far broader— sues have been approved by the Food and Drug mote local tissue repair. in the future, engineered tissues could reduce Administration (FDA), more than 70 companies Sources of cells for implantation include the need for organ replacement, and could are spending a total of $600 million per year to autologous cells from the patient, allogeneic greatly accelerate the development of new develop new products (2). There are still many cells from a human donor who is not immu- drugs that may cure patients, eliminating the technical challenges to overcome before we cre- nologically identical to the patient, and xeno- need for organ transplants altogether. ate “off-the-shelf ” tissues that represent the geneic cells from a different species. Each To engineer living tissues in vitro, cultured translation of scientific discoveries into treat- category may be further delineated in terms cells are coaxed to grow on bioactive degradable ments for millions of patients. The successful of whether the cells are adult or embryonic scaffolds that provide the physical and chemical large-scale production of engineered tissues re- stem cells (capable of both self renewal and cues to guide their differentiation and assembly quires an adequate source of healthy expandable differentiation into a variety of cell lineages), into three-dimensional (3D) tissues (3). The as- cells, the optimization of scaffolds, and the cre- or a mixture of differentiated cells at different sembly of cells into tissues is a highly orches- ation of bioreactors, which mimic the environ- stages of maturation (including rare stem and trated set of events that requires time scales ment of the body and that are amenable to progenitor cells). Some approaches use cell scale-up. Additional challenges include the pres- mixtures, whereas others rely on separation ervation of the product so that it has a long or enrichment of stem cells. 1 Department of Chemical Engineering, Division of shelf-life and the successful use of various ap- Although the prospect of using xenogene- Bioengineering and Environmental Health, and Bio- technology Process Engineering Center, Massachu- proaches to prevent tissue rejection. ic cells for tissue repair remains controversial setts Institute of Technology, Cambridge, MA 02139, because of the potential for transmitting ani- USA. 2Advanced Tissue Sciences, La Jolla, CA 92037, Biological Challenges: Cells and Their mal pathogens to humans, xenogeneic cells USA. Sources could perhaps temporarily support an ailing *To whom correspondence should be addressed at There are three principal therapeutic strat- tissue until either a human donor organ be- the Department of Chemical Engineering, Division of egies for treating diseased or injured tissues comes available for transplant, or the tissue Bioengineering and Environmental Health, Massachu- setts Institute of Technology 66-466, Cambridge, MA in patients: (i) implantation of freshly iso- repairs itself. For example, pig liver cells 02139, USA. E-mail: [email protected] lated or cultured cells; (ii) implantation of (hepatocytes) grown in extracorporeal biore-

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1009 B ODYBUILDING:THE B IONIC H UMAN actors are being tested clinically to see induced to form many different cell types. may still be too slow and the ultimate quality whether they can support patients with liver Although ES cells can be coaxed to assem- and stability of vessels suboptimal with this failure until a liver transplant can be per- ble into tissues as diverse as insulin-secret- approach. Interestingly, angiogenesis also can formed [see Viewpoint by Strain on page ing pancreatic islets (12) and blood (13), be induced using engineered skin products be- 1005 (4)]. they have not, as yet, been able to cure an cause the dermal fibroblasts that they contain Allogeneic cells have been used successfully animal model of disease. We still need bet- produce angiogenic growth factors (8, 18). The to treat skin ulcers (5), diabetes (6), and liver ter markers to identify stem cells and their need for preformed vascular beds or rapid an- disease (7). Patients with diabetic or venous progeny, better ways to expand them in giogenesis could be avoided altogether by ex- skin ulcers have been treated with two FDA- culture, and more research to see whether ploiting what may be a common property of approved living skin products engineered in the there is an immunological barrier to im- many stem and progenitor cells—their resis- lab. One product is composed of neonatal der- planting stem cells derived from allogeneic tance to low-oxygen conditions (19–23). mal fibroblasts obtained from human foreskins. donors (14). When small pieces of bone tissue are im- The neonatal fibroblasts are expanded in culture Adult bone marrow stem cells can be col- planted at the site of bone injury, existing mi- and seeded onto a thin scaffold composed of the lected from the circulation (after mobilization crovessels in the implant connect with blood polymer polylactide coglycolide (originally de- with cytokines) and used clinically to treat a vessels at the injury site. This has prompted the veloped for use in surgical sutures), which range of blood disorders. Recent reports that inclusion of endothelial cells (which form blood breaks down gradually in the presence of water marrow-derived stem cells can give rise to hepa- vessels) in cultures of the cells to be expanded, (8). The cells on their scaffold are cultured in tocytes, cardiac muscle cells, and lung tissue so that rudimentary tubelike vessels form within custom-designed bioreactors for several weeks suggest that efficient recruitment of bone mar- the assembling tissue (24). Even more ambi- until they form a tissue similar to the inner row stem cells to sites of injury or their injection tious is the goal to form fully vascularized tis- dermal layer of skin. This neo-dermis is then into these sites may provide a source of cells for sues for implantation that contain blood vessels frozen for shipment to physicians. The second tissue repair. At least one model of animal liver of sufficient size that they can be fused with the skin product has both dermal and epidermal disease has been cured by a bone marrow trans- patient’s own blood vessels during surgery. The layers. It is composed of dermal fibroblasts in a plant (15). The addition of bone marrow to complexities associated with organizing mil- collagen solution that forms a gel when heated engineered bone grafts improves healing of lions of cells into 3D structures such as blood to body temperature; the gel is coated with bone defects; concentrating and selecting for the vessels can be simplified using computer mod- several layers of human epidermal cells (kera- marrow stem cells that form bone improves eling, which translates the tissue’s 3D structure tinocytes). After transfer to the patient, this skin healing still further (16). into a 2D template. Composed of a degradable product is at least partially replaced by host skin polymer, the 2D template precisely guides cells cells as healing progresses. The dermal fibro- Engineering Challenges to their correct positions, the engineered tissue blasts in the skin products naturally secrete ex- Blood vessels of the microcirculation. One of finally being folded up to form the 3D structure tracellular matrix proteins and are able to re- the principal constraints on the size of tissues (25). spond to growth-regulatory molecules secreted engineered in vitro that do not have their own Scaffolds. Scaffolds are porous, degrad- by the host (9). These skin products can persist blood supply is the short distance over which able structures fabricated from either natu- for up to 6 months after implantation. oxygen can diffuse before being consumed (a ral materials (collagen, fibrin) or synthetic The success of engineered dermal implants few hundred micrometers at most). Once im- polymers (polyglycolide, polylactide, poly- for treating skin injuries and burns has not planted in the patient, cells in the engineered lactide coglycolide). They can be sponge- been as easy to replicate for organs such as tissue will consume the available oxygen within like sheets, gels, or highly complex struc- the liver and pancreas, partly because ex- a few hours, but it will take several days for the tures with intricate pores and channels fab- panding hepatocytes or pancreatic islet cells growth of new blood vessels (angiogenesis) that ricated using new materials-processing in culture is much more difficult than ex- will deliver oxygen and nutrients to the im- technologies. Virtually all scaffolds used in panding dermal fibroblasts or keratinocytes. plants. How can this problem be overcome? tissue engineering are intended to degrade There is an FDA-approved autologous cell Implanting cultured cells directly into the exist- slowly after implantation in the patient and product for the repair of articular cartilage ing vascular beds of the patient’s liver and be replaced by new tissue. (10). A small piece of cartilage is removed spleen looks like one promising strategy. Hepa- Many epithelial and connective tissues have from the healthy section of a patient’s in- tocytes injected directly into human liver show a simple macroscopic architecture consisting of jured knee. Cartilage cells (chondrocytes) engraftment and sufficient biochemical activity a number of thin layers. Bladder, intestine, and are isolated, expanded in culture, and are to ameliorate the symptoms of liver disease, blood vessels are composed of a layer of smooth then implanted at the injury site. In a vari- although this is clearly not a cure (7). Some muscle sandwiched between a layer of collage- ation on this approach, mesenchymal stem diabetic patients with pancreatic islet cells im- nous vascularized support matrix and an epithe- cells have been harvested from patient bone planted into the liver (a very vascular organ) lial lining. Such structures can be built by seed- marrow, expanded in culture, and then in- exhibited normal glucose tolerance for several ing the different cell types for each layer sequen- duced to differentiate into cells that can months after the procedure (6). tially ondegradable scaffolds made from syn- help to repair damaged bone, cartilage, ten- Unfortunately, cells implanted for the repair thetic fibers of polyglycolide or its derivatives don (see the News story by Pennisi on page of bone or tendon, for example, cannot exploit that are 10 to 20 micrometers in diameter. Poly- 1011), or ligament (11). Given that donor existing vascular beds. Inducing or speeding up glycolide, unlike polylactide, does not dissolve and patient cells are already being exploited angiogenesis by engineering a scaffold to slowly in solvents such as chloroform. Thus, 3D poly- therapeutically, why is there such intense release growth factors, such as vascular endo- glycolide scaffolds can be sculpted by dipping interest in adult and embryonic stem cells? thelial cell growth factor (VEGF) or fibroblast them in a solution of polylactide dissolved in Stem cells hold great promise for treating growth factor (FGF), may be the answer. For chloroform and shaping the wet fabric on a damaged tissue where the source of cells for example, controlled release of both VEGF and mold. When the chloroform evaporates, the repair is extremely limited or not readily platelet-derived growth factor (PDGF) from the polylactide serves as a solid glue to hold the accessible. Embryonic stem (ES) cells are same scaffold implanted into rats resulted in fabric in the desired shape (26, 27). A scaffold attractive because they can be expanded in blood vessel induction, maturation, and stabili- an undifferentiated state in vitro and can be zation (17). However, blood vessel formation CONTINUED ON PAGE 1012

1010 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

CONTINUED FROM PAGE 1010

made in this way in the shape of a bladder and seeded with urinary epithelial cells and smooth muscle cells has been implanted into dogs. This artificial bladder acquired near-normal function (27). New methods are being developed to process fabrics for more demanding scaffolding applications. For example, using solid free-form fabrication techniques, complex 3D polymer structures have been built from a series of very thin 2D layers, starting with a computer model of the desired shape derived from an MRI (magnetic resonance imaging) or computerized to- mography image of the patient’s original tissue (28) (Fig. 1). Scaffolds can also be designed to release growth factors that induce cellular differentiation and tissue growth in vitro, or cell migration into the wound site in vivo. For example, a dual-release scaffold composed of PDGF encap- sulated in polylactide coglycolide microspheres together with a powder of the same polymer linked to VEGF promoted angiogenesis by re- Fig. 1. The 3DP printing process is a solid free-form fabrication method that builds complex 3D objects leasing VEGF quickly and PDGF slowly, thus from a series of 2D layers (61). First, a ﬁne powder is spread in a thin layer on a piston. A binder or “glue” mimicking the physiological production of these (green drops) is then “printed” into speciﬁc regions of the powder according to a computer program that has parsed the 3D object into thin 2D layers. The glue binds the particles together in precise regions, and growth factors (26). Scaffolds containing small, the 2D layer starts to take shape. The piston is then lowered, the cycle repeated, and the next layer degradable polymer beads that release nerve printed. This layer-by-layer process continues until the entire 3D object is fabricated, allowing the growth factor improve the viability of fetal neu- generation of gradients in composition, surface chemistry, and porosity. This provides 3D scaffolds that ral cells transplanted into rat brain (29). are suitable for growing composite tissue structures such as bone. [Illustration: Preston Morrighan] The fragile nature of proteins has motivated design of scaffolds that release naked plasmid and differentiation, finally degrading in responsive to the nanoscale spatial organization of DNA containing genes that encode growth fac- sponse to matrix remodeling enzymes released RGD peptides—such peptides more effectively tors (30). When a collagen scaffold fabricated to by the cells as tissue repair progresses (32). induce cell adhesion and migration when they release the gene for parathyroid hormone (a The discovery of adhesion domains in fi- are clustered rather than random (37, 38). The protein that regulates bone growth) was implant- bronectin and other extracellular matrix glyco- ability of fibrin scaffolds modified with the pep- ed at a bone injury site in the dog, new bone was proteins containing the amino acid sequence tide ligand L1Ig6, which binds to the cell sur- ␣ ␤ formed according to the “dose” of the gene (30). Arg-Gly-Asp (RGD) has enabled the design of face adhesion receptor integrin v 3, to promote Controlling the diffusion rates of genes and synthetic materials that can modulate cell adhe- angiogenesis is influenced by the supramolecu- proteins from scaffolds so that they are in the sion (33). Yet the process is not as simple as lar organization of the fibrin (39). physiological range is the next challenge. New identifying an adhesion peptide and incorporat- Design principles are emerging for modulat- bioactive materials, such as those that covalently ing it into a biodegradable material. Cell motil- ing the interactions of cells with growth factors. incorporate growth factors and other molecules ity is an adhesion-dependent process required Apart from hematopoietic cytokines, successful that regulate cell behavior, offer alternatives for for cell migration, angiogenesis, and regrowth use of growth factors for human tissue regener- enhancing scaffold performance. of severed nerve ends, among many other phys- ation has been notoriously difficult. Many Biomaterials. A crucial mainstay of tissue iological events. An engineering model that ingrowth factors, including the angiogenic factors engineering is the biomaterial from which scaf- corporates the biophysics of how cells bind to VEGF and FGF, are bound tightly to the extra- folds are fashioned (3) [see Viewpoint by extracellular matrix adhesion molecules and cellular matrix of normal tissues. A ligand for Hench and Polak on page 1014 (31)]. Many how they contract predicts that increasing the the epidermal growth factor receptor (EGFR) is biomaterials direct the growth of cells in cul- number of adhesion contacts between cells and immobilized within tenascin, a large extracellu- ture. However, tissue regeneration in vivo in- the extracellular matrix may not always be ad- lar matrix molecule (40). Presenting growth fac- volving the guided growth of nerve, bone, vantageous (34). If too few adhesive ligands tors as part of an extracellular matrix, rather than blood vessels, or corneal epithelia across criti- (such as RGD) are available, cells cannot get a just releasing them into the liquid medium, has cal injury sites requires that cells receive more strong enough grip to enable them to move, but improved nerve regeneration and growth of specific instructions. In vivo, the cells that re- if there are too many ligands, cells adhere so smooth muscle cells during engineering of arti- pair and regenerate damaged tissues are bom- firmly that they remain stuck in place. Thus, ficial arteries (41, 42). Using gel scaffolds that barded with molecular signals, both from the intermediate adhesion is required for optimal incorporate a complete compendium of growth “hostile” wound site and from healthy sur- cell migration (34). In vivo, bone scaffolds coat- factors and their correctly presented adhesion rounding tissues. The ideal biomaterial for a ed with adhesion proteins containing the RGD sites may be the next step (32). scaffold would selectively interact with the spe- motif promote maximal tissue ingrowth only at Tissue architecture. The correct molecular cific adhesion and growth factor receptors ex- intermediate values of ligand surface density and macroscopic architecture of cartilage, blood pressed by target cells in surrounding tissues (35); likewise, only at an intermediate density vessels, bone, and other tissue is essential for required for repair of damaged tissue. The scaf- do adhesion proteins on scaffolds induce neural proper tissue function. Connective tissue cells fold could guide migration of these target cells progenitor cells to extend neurites, a prerequisite grown on 3D scaffolds in vitro secrete biochem- into the injury site and stimulate their growth for nerve regeneration (36). Cells are also re- ically appropriate extracellular matrix molecules

1012 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

Fig. 2. A microfabricated bioreactor for perfusing 3D liver tissue engi- nels. (C) Hepatocytes seeded onto the scaffold of the bioreactor attach to neered in vitro (54, 55). (A) A cross section showing tissue aggregates the walls of the channels (four channels are shown) and reorganize to form growing attached to the inside walls of the narrow channels of the 3D structures that are reminiscent of liver cords. Bile canaliculi and tight silicon-chip scaffold. Culture medium flows across the top of the scaffold junctions can be seen with high-power microscopy (54, 55). Live cells are as well as through the narrow channels, enabling tissue aggregates to green and dead cells are red as visualized with the calcein AM/ethidium extract oxygen and nutrients. The design of the scaffold promotes homodimer stain. (D) Scanning electron micrograph showing vessel-like self-assembly of the cells into tissues. (B) A bioreactor containing a structures assembled from endothelial cells at the fluid-tissue interface in 0.2-mm-thick silicon-chip scaffold etched with 0.3-mm-diameter chan- the bioreactor channels. [Illustration: Preston Morrighan] yet fail to acquire the appropriate tissue archi- scaffold (the small-intestine submucosa stripped may provide a cheap in vitro system with greater tecture. The answer may lie in providing appro- of cells but with an intact extracellular matrix) control of variables for studying viral infection. priate physiological stresses during engineering that recruited endothelial cells (46). However, We speculate that the greatest impact of of the tissue in vitro. The first generation of translating vascular successes in animals to hu- tissue engineering in the coming decade will be bioreactors for culturing cells were designed mans is notoriously difficult because human and for designing in vitro physiological models to simply to pump nutrient liquid (culture medium) animal endothelia behave very differently (47). study disease pathogenesis and for developing through the assembling tissue. The next wave of molecular therapeutics. For example, engi- bioreactors designed for growing blood vessels Tissue-Engineered Model Systems neered 3D capillary beds could be used to study and cartilage subjected the nascent tissue to Tissue engineering can be applied to the the effects of a variety of insults in a high- compression, shear stresses, and even pulsatile development of drugs to treat many diseases that throughput manner that is reasonably physiolog- flow of culture medium. Such stresses pro- could be prevented or even cured if such drugs ical (54, 55). The creation of tissues containing foundly improve the mechanical properties of were available today. For example, the hepatitis hierarchical cell-cell interactions under appro- engineered vessels, cartilage, and cardiac mus- C virus has infected more than 170 million priate mechanical stresses (including perfusion cle (43). In a promising result for the ϳ500,000 people worldwide and is currently the leading shear as found in the microcirculation) will take patients who need heart bypass surgery, an ar- cause of liver failure in the United States (48– in vitro systems even closer to living tissues (see tificial artery engineered in the laboratory under 50). Infecting cultured liver cells with the virus Fig. 2). In Europe, human skin engineered in pulsatile conditions using endothelial cells seed- is extremely difficult, and the lack of small- vitro (56) has replaced animals in the regulatory ed on a polymer scaffold developed a burst animal models has hampered the development approval process for testing drugs and other strength of 2000 mmHg compared with 300 of drugs to combat this pathogen. Unfortunately, agents for their skin corrosiveness (57, 58). mmHg for unstressed tissues (although physio- human hepatocytes quickly lose many of their New microfabrication and microfluidics tech- logical responses to vasoactive factors were not liver-specific functions, including susceptibility nologies are beginning to transform the diagnos- completely normal) (44). A typical saphenous to viral infection and ability to metabolize drugs, tics field, allowing multiple chemical reactions vein graft used surgically has a burst pressure of when they are cultured. Chimeric animal models to take place on a small plastic chip (59, 60). ϳ1000 mmHg; normal systolic pressure exerted made by implanting human hepatocytes into Tissues engineered in vitro that can be easily on blood vessels in vivo is about ϳ120 mmHg. mice (51–53) may provide an alternative system manipulated have the potential to expand tissue A critical issue for engineering 3D tissues in for studying hepatitis C virus and other hepatitis engineering into areas such as cancer drug de- vitro is scale-up for clinical use. Hundreds or viruses. In the future, engineered liver tissue velopment. For example, a good in vitro model thousands of tissues must be grown and cryopreserved under sterile conditions. This has certainly been feasible for the allogeneic dermal fibroblasts that comprise the FDA-approved en- Clinical Endpoints gineered skin products. These cells can be ex- How are engineered tissues ultimately judged to be good enough for clinical application? panded quickly in culture without the need for With tissues such as kidney or bone marrow, analysis of urine or blood provides valuable external stresses, and are readily amenable to information about the success or failure of the tissue in both human patients and animal cryopreservation (45). If cells to be engineered models. But different methods are needed to quantitatively assess the long-term into tissues for implantation must first be de- outcome of engineered connective tissues, such as cartilage, tendon, and blood vessels. rived from the patient, then a separate culture The ultimate failure of these tissues (or their engineered counterparts) is defined as a system will be required for each patient, posing failure to perform mechanically, resulting in pain and disability. Metrics used in current clinical trials of engineered cartilage, for example, are a reduction in pain, appearance of formidable regulatory challenges and high costs. new tissue, and qualitative mechanical probing. Because connective tissues are relatively Thus, there is much interest in the in situ growth avascular, they take a long time to be remodeled and repaired, so that metrics must be of tissue from injected cells where mechanical obtained over the long term. In the absence of mechanistic models of long-term tissue stresses are applied naturally. Blood vessels repair, gene-expression changes obtained by microarray analysis, or imaging of 3D tissue with superior mechanical properties have been structure with MRI, may be valuable alternative strategies. grown in situ in an animal model using a natural

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1013 B ODYBUILDING:THE B IONIC H UMAN

of cancer metastasis that mimics the lodging of 13. D. S. Kaufman, E. T. Hanson, R. L. Lewis, R. Auerbach, 40. C. S. Swindle et al., J. Cell Biol. 154, 459 (2001). a single tumor cell in a capillary bed would J. A. Thomson, Proc. Natl. Acad. Sci. U.S.A. 98, 10716 41. B. K. Mann, R. H. Schmedlen, J. L. West, Biomaterials (2001). facilitate the development of antimetastatic 2, 84 (2001). 14. Committee of the Biological and Biomedical Applica- 42. A. H. Zisch, U. Schenk, J. C. Schense, S. E. Sakiyama- drugs. tions of Stem Cell Research; Board on Life Sciences, Elbert, J. A. Hubbell, J. Control. Release 72, 101 The increasingly intimate combination of Institute of Medicine, “Stem Cells and the Future of (2001). Regenerative Medicine” (National Academy Press, 43. A. J. Grodzinsky, M. E. Levenston, M. Jin, E. H. Frank, engineering and biology offers the prospect Washington, DC, 2001). Annu. Rev. Biomed. Eng. 2, 691 (2000). of sophisticated physiological in vitro models 15. E. Lagasse et al., Nature Med. 6, 1229 (2000). 44. L. E. Niklason et al., Science 284, 489 (1999). of many different human tissues. These phys- 16. J. E. Fleming, C. N. Cornell, G. F. Muschler, Orthop. Clin. N. Am. 31, 357 (2000). 45. G. K. Naughton, D. Applegate, in Methods of Tissue iological surrogates will ultimately allow ma- 17. T. P. Richardson, M. C. Peters, A. B. Ennett, D. J. Engineering, A. Atala, R. P. Lanza, Eds. (Academic jor advances in prevention, diagnosis, and Mooney, Nature Biotechnol. 19, 1029 (2001). Press, New York, 2002), pp. 1157–1175 molecular treatment of diseases that are cur- 18. R. S. Kellar et al., Circulation 104, 2063 (2001). 46. T. Huynh et al., Nature Biotechnol. 17, 1083 (1999). 47. R. M. Nerem, D. Seliktar, Annu. Rev. Biomed. Eng. 3, rently considered potential targets for tissue 19. D. L. Hevehan, E. T. Papoutsakis, W. M. Miller, Exp. Hematol. 28, 267 (2000). 225 (2001). engineering. Ultimately, this may result in a 20. Z. Ivanovic et al., Br. J. Haematol. 108, 424 (2000). 48. D. F. Mercer et al., Nature. Med. 7, 927 (2001). greater emphasis on treating different target 21. S. J. Morrison et al., J. Neurosci. 20, 7370 (2000). 49. C. Crabb, Science 294, 506 (2001). diseases, such as trauma and congenital de- 22. A. Storch et al., Exp. Neurol. 170, 317 (2001). 50. N. Fausto, Nature Med. 7, 890 (2001). 23. D. P. Lennon, J. M. Edmison, A. I. Caplan, J. Cell. 51. K. Ohashi et al., Nature Med. 6, 327 (2000). fects, with engineered tissue. Physiol. 187, 345 (2001). 52. M. Dandri et al., Hepatology 33, 981 (2001). With a scientific foundation firmly estab- 24. A. F. Black, F. Berthod, N. L’heureux, L. Germain, F. A. 53. M. Grompe, Hepatology 33, 1005 (2001). lished, we now need a robust infusion of Auger, FASEB J. 12, 1331 (1998). 54. M. Powers et al., Tissue Eng., in press. 25. S. Kaihara et al., Tissue Eng. 6, 105 (2000). biology-based engineering analysis and de- 55. M. J. Powers et al., Biotechnol. Bioeng., in press. 26. C. A. Vacanti, L. G. Cima, D. Rodkowski, J. Upton, in 56. L. P. Bernhofer, M. Seiberg, K. M. Martin, Toxicol. In sign to move the tissue-engineering field Tissue-Inducing Biomaterials, L. G. Cima, E. Ron, Eds. Vitro 13, 219 (1999). (Materials Research Society, Warrendale PA, 1992), from an era of phenomenological observation 57. Anonymous European Union Commission Directive vol. 252, pp. 323–330. and serendipity to one of commercially viable 27. F. Oberpenning, J. Meng, J. J. Yoo, A. Atala, Nature 2000/33/EC, 25 April 2000 (Official Journal of the products that will improve the lives of mil- Biotechnol. 17, 149 (1999.). European Communities), “Skin Corrosion, Rat Skin lions of patients. 28. L. Griffith, B. Wu, M. J. Cima, B. Chaignaud, J. P. TER and Human Skin Model Assay” (OJ L 136, 8 June Vacanti, Ann. N.Y. Acad. Sci. 831, 382 (1997). 2000), http://embryo.ib.amwaw.edu.pl/invittox/prot/ 29. M. J. Mahoney, W. M. Saltzman, Nature Biotechnol. l_13620000608en00010089.pdf (2000). References and Notes 19, 934 (2001). 58. http://iccvam.niehs.nih.gov (Interagency Coordinat- 1. R. Langer, J. P. Vacanti, Science 260, 920 (1993). 30. J. Bonadio, E. Smiley, P. Patil, S. Goldstein, Nature ing Committee on the Validation of Alternative 2. M. J. Lysaght, J. Reyes, Tissue Eng. 7, 485 (2001). Med. 7, 753 (1999). Methods, 2001). 3. L. G. Griffith, Acta Mater. 48, 263 (2000). 31. L. L. Hench, J. M. Polak, Science 295, 1014 (2002). 59. B. D. DeBusschere, G. T. Kovacs, Biosens. Bioelectron. 4. A. J. Strain, J. M. Neuberger, Science 295, 1005 32. J. A. Hubbell, Curr. Opin. Biotechnol. 10, 123 (1999). 16, 543 (2001). (2002). 33. M. D. Pierschbacher, E. Ruoslahti, J. Biol. Chem. 262, 60. M. Krishnan, V. Namasivayam, R. Lin, R. Pal, M. A. 5. R. A. Pollak, H. Edington, J. L. Jensen, R. O. Kroeker, 17294 (1987). Burns, Curr. Opin. Biotechnol. 12, 92 (2001). G. D. Gentzkow, Wounds 9, 175 (1997). 34. S. P. Palecek, J. C. Loftus, M. H. Ginsberg, D. A. 61. E. Sachs, M. Cima, P. Williams, D. Brancazio, J. Cornie, 6. E. A. Ryan et al., Diabetes 50, 710 (2001). Lauffenburger, A. F. Horwitz, Nature 385, 537 (1997). J. Eng. Ind. 114, 481 (1992). 7. S. C. Strom, Sem. Liver Dis. 19, 39 (1999). 35. K. Eid, E. Chen, L. G. Griffith, J. Glowacki, J. Biomed. 62. Research funded by NIH (National Institute of Gen- 8. J. Mansbridge, K. Liu, R. Patch, K. Symons, E. Pinney, Mater. Res. 57, 224 (2001). eral Medical Sciences and National Institute of Envi- Tissue Eng. 4, 403 (1998). 36. J. C. Schense, J. A. Hubbell, J. Biol. Chem. 275, 6813 ronmental Health Sciences), Defense Advanced Re- 9. A. Kern, K. Liu, J. Mansbridge, J. Invest. Dermatol. 117, (2000). search Project Agency ( Tissue-based Biosensors Pro- 112 (2001). 37. G. Maheshwari, G. L. Brown, D. A. Lauffenburger, A. gram), NSF (Engineering Research Center Program), 10. M. Brittberg, T. Tallheden, B. Sjogren-Jansson, A. Lin- Wells, L. G. Griffith, J. Cell Sci. 113, 1677 (2000). the DuPont–Massachusetts Institute of Technology dahl, L. Peterson, Clin. Orthop. 391 (suppl.), S337 38. D. J. Irvine, A. M. Mayes, L. G. Griffith, Biomacromol- Alliance, the Whitaker Foundation, and Therics, Inc. (2001). ecules 2, 84 (2001). We thank R. Langer, J. P. Vacanti, A. Grodzinsky, S. R. 11. M. F. Pittenger et al., Science 284, 143 (1999). 39. H. Hall, T. Baechi, J. A. Hubbell, Microvasc. Res. 62, Tannenbaum, and D. A. Lauffenburger for helpful 12. N. Lumelsky et al., Science 292, 1389 (2001). 315 (2001). comments.

VIEWPOINT Third-Generation Biomedical Materials Larry L. Hench* and Julia M. Polak*

Whereas second-generation biomaterials were designed to be either were implanted in patients in the United resorbable or bioactive, the next generation of biomaterials is combining States annually. A common feature of most of these two properties, with the aim of developing materials that, once the materials was their biological “inertness.” implanted, will help the body heal itself. The principle underlying the bulk of biomaterials development was to reduce to a mini- Initially, the choice of biomedical materials for materials selected proved to be either pathogen- mum the immune response to the foreign use in the body was dependent on those already ic or toxic. During the 1960s and 1970s a first body, and this is still valid 21 years later. Tens available off the shelf. Until an understanding of generation of materials was developed for use of millions of individuals have had their qual- the immune system developed, many of the inside the human body. The goal of all early ity of life enhanced for 5 to 25 years by use of biomaterials was to “achieve a suitable combi- implants made from such “inert” biomaterials. nation of physical properties to match those of Department of Materials and the Tissue Engineering the replaced tissue with a minimal toxic re- Second-Generation Biomaterials Centre, Imperial College of Science, Technology and Medicine, University of London, Prince Consort Road, sponse in the host” (1). In 1980 there were more The field of biomaterials began to shift in London SW7 2BP, UK. than 50 implanted devices (prostheses) in emphasis from achieving exclusively a bioin- *To whom correspondence should be addressed. E- clinical use made from 40 different materials mail: [email protected] and [email protected] (1), and some 2 to 3 million prosthetic parts CONTINUED ON PAGE 1016

1014 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN NEWS Tending Tender Tendons

What do a skate egg case and a tendon have in common? Cross-links that make in. Skate eggs emerge white and pliable from them strong—an observation that propelled Thomas Koob on his quest to build a a large gland in the reproductive tract; over the bionic tendon next few hours, they darken and become leathery. Made of collagenlike materials, the Gazing at an aquarium tank almost 20 years ment tendons, researchers tried polyesters egg cases are tough enough to persist in the ago, Thomas Koob made an observation that or other synthetic materials. But more re- ocean for months, and so are its enzymes. would eventually change the course of his recently, several research teams, including Koob was intrigued by the chemistry underly- search. While reflecting on the remarkable Koob’s and Banes’s, have gone after biolog- ing this transition and stability. Further study properties of a skate egg case, he realized it ically based materials that might be less suggested that these enzymes form bonds that might hold the secret to an artificial tendon. prone to immune rejection. are the cross-links between the case’s colla- Bidding farewell to his studies of human Koob and others are grappling with how genlike fibers. He wondered whether that pro- and shark reproduction, Koob, a biochemist at to replicate the complex structure of the ten- cess could be mimicked to make biosynthetic Shriners Hospital for Children in Tampa, Flori- don. A substitute must be strong, stiff, and tendons and began looking for a way to cross- da, joined the then-very-small ranks of scien- resilient, and it should be able to change its link human collagen fibers. tists trying to build a bionic tendon. He now properties—such as its stiffness—as real After much searching, Koob found a has plenty of competition: The field is quite tendons do. Contrary to the common view of chemical called NDGA, derived from a shrub “huge,” says Albert Banes, a cell biologist at them as elastic cables, “tendons are very called creosote, that could do just that in his the University of North Carolina, Chapel Hill, much alive,” explains Banes. lab. As he reported at a recent meeting,* the and another bionic-tendon pioneer. Viewed one way, the cells within a tendon collagen-NDGA combination is very compat- If any of these contenders are successful, look like boxcars along tracks of collagen. ible with rabbit tissue and does not trigger im- they will have a hot commodity. Tendons are “But it’s really more like a spider web. The mune reactions. (The Food and Drug Admin- the ropes that tie muscle to bone, istration requires this test for bio- enabling the former to maneuver materials to be used in humans.) the latter quickly and efficiently. For that reason, notes Summers, Some, such as the Achilles’tendon, this “natural” tendon material act as springs. Others get the body might be safer than a synthetic. moving and, for example, help turn Industrial strength. Cross-links between collagen strands impart Not that this new tendon mate- strength to tendons. rial is ready for use. Koob still

the open hand into a clenched fist. on February 3, 2016 But when they break, “our ability needs to find a way to glue in the to repair tendons is limited,” laments Richard cells are highly interactive and intertwined artificial tendon—something that can already Gelberman, an orthopedic surgeon at Wash- with the collagen,” says Banes. This network be done with a woven polyester tendon mate- ington University School of Medicine in St. includes molecular bridges that reach across rial, notes Bahaa Seedhom, a bioengineer at Louis, Missouri. He estimates that surgeons and set up cross-links between nearby colla- the University of Leeds, U.K., who helped try to fix some 300,000 tendons a year—and gen strands, thereby preventing the collagen develop the polyester substitute. And even that doesn’t include those in the legs and from slipping as the tendon is pulled by a though Koob’s material has the necessary

feet. Recovery can take months, notes Adam contracting muscle. strength and stiffness, he needs to demon- Downloaded from Summers, a comparative physiologist at the Collagen helps make the tissue strong, strate that it can withstand the multiple stress- University of California, Irvine, and at best but the tendon cells are equally important, es to which it will be subjected, notes Robert the restored tendon is about 60% functional. because they react to loads, activate genes, Ker, another Leeds tendon expert.

CREDITS: THOMAS J. KOOB In their first attempts to create replace- and change the properties of the tendon The ideal tendon material, all these ten- itself. Any re- don contenders recognize, would need to placement ten- contain or recruit cells that could help the don must ei- tendon develop the right set of properties and Inspiration. A skate egg case ther provide repair small tears as needed. Because his ma- is one of many natural “body” new cells or terial accumulates cells only on the outside, parts to inspire bionic ones. recruit new Koob is now working on making it less ones into the dense and more like a real tendon. Mean- artificial tissue while, Banes and those working on gel- if it is to function based substitutes have materials that already as well as the orig- come with their own tendon cells or, in one inal, says Banes, case, stem cells. These groups plan to add who thinks he has genes to these cells that would enhance their solved this problem. ability to work like natural tendon cells. At Koob is tackling a dif- this stage, however, no one can predict which ferent issue first: how to approach will yield a tendon that is almost as make a biological materi- good as the real thing. al strong enough to tem- –ELIZABETH PENNISI porarily take a tendon’s place as it heals. That’s * Society for Integrative and Comparative Biology where the egg cases came 2002 Annual Meeting, 2–6 January, Anaheim, CA.

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1011 B ODYBUILDING:THE B IONIC H UMAN

of cancer metastasis that mimics the lodging of 13. D. S. Kaufman, E. T. Hanson, R. L. Lewis, R. Auerbach, 40. C. S. Swindle et al., J. Cell Biol. 154, 459 (2001). a single tumor cell in a capillary bed would J. A. Thomson, Proc. Natl. Acad. Sci. U.S.A. 98, 10716 41. B. K. Mann, R. H. Schmedlen, J. L. West, Biomaterials (2001). facilitate the development of antimetastatic 2, 84 (2001). 14. Committee of the Biological and Biomedical Applica- 42. A. H. Zisch, U. Schenk, J. C. Schense, S. E. Sakiyama- drugs. tions of Stem Cell Research; Board on Life Sciences, Elbert, J. A. Hubbell, J. Control. Release 72, 101 The increasingly intimate combination of Institute of Medicine, “Stem Cells and the Future of (2001). Regenerative Medicine” (National Academy Press, 43. A. J. Grodzinsky, M. E. Levenston, M. Jin, E. H. Frank, engineering and biology offers the prospect Washington, DC, 2001). Annu. Rev. Biomed. Eng. 2, 691 (2000). of sophisticated physiological in vitro models 15. E. Lagasse et al., Nature Med. 6, 1229 (2000). 44. L. E. Niklason et al., Science 284, 489 (1999). of many different human tissues. These phys- 16. J. E. Fleming, C. N. Cornell, G. F. Muschler, Orthop. Clin. N. Am. 31, 357 (2000). 45. G. K. Naughton, D. Applegate, in Methods of Tissue iological surrogates will ultimately allow ma- 17. T. P. Richardson, M. C. Peters, A. B. Ennett, D. J. Engineering, A. Atala, R. P. Lanza, Eds. (Academic jor advances in prevention, diagnosis, and Mooney, Nature Biotechnol. 19, 1029 (2001). Press, New York, 2002), pp. 1157–1175 molecular treatment of diseases that are cur- 18. R. S. Kellar et al., Circulation 104, 2063 (2001). 46. T. Huynh et al., Nature Biotechnol. 17, 1083 (1999). 47. R. M. Nerem, D. Seliktar, Annu. Rev. Biomed. Eng. 3, rently considered potential targets for tissue 19. D. L. Hevehan, E. T. Papoutsakis, W. M. Miller, Exp. Hematol. 28, 267 (2000). 225 (2001). engineering. Ultimately, this may result in a 20. Z. Ivanovic et al., Br. J. Haematol. 108, 424 (2000). 48. D. F. Mercer et al., Nature. Med. 7, 927 (2001). greater emphasis on treating different target 21. S. J. Morrison et al., J. Neurosci. 20, 7370 (2000). 49. C. Crabb, Science 294, 506 (2001). diseases, such as trauma and congenital de- 22. A. Storch et al., Exp. Neurol. 170, 317 (2001). 50. N. Fausto, Nature Med. 7, 890 (2001). 23. D. P. Lennon, J. M. Edmison, A. I. Caplan, J. Cell. 51. K. Ohashi et al., Nature Med. 6, 327 (2000). fects, with engineered tissue. Physiol. 187, 345 (2001). 52. M. Dandri et al., Hepatology 33, 981 (2001). With a scientific foundation firmly estab- 24. A. F. Black, F. Berthod, N. L’heureux, L. Germain, F. A. 53. M. Grompe, Hepatology 33, 1005 (2001). lished, we now need a robust infusion of Auger, FASEB J. 12, 1331 (1998). 54. M. Powers et al., Tissue Eng., in press. 25. S. Kaihara et al., Tissue Eng. 6, 105 (2000). biology-based engineering analysis and de- 55. M. J. Powers et al., Biotechnol. Bioeng., in press. 26. C. A. Vacanti, L. G. Cima, D. Rodkowski, J. Upton, in 56. L. P. Bernhofer, M. Seiberg, K. M. Martin, Toxicol. In sign to move the tissue-engineering field Tissue-Inducing Biomaterials, L. G. Cima, E. Ron, Eds. Vitro 13, 219 (1999). (Materials Research Society, Warrendale PA, 1992), from an era of phenomenological observation 57. Anonymous European Union Commission Directive vol. 252, pp. 323–330. and serendipity to one of commercially viable 27. F. Oberpenning, J. Meng, J. J. Yoo, A. Atala, Nature 2000/33/EC, 25 April 2000 (Official Journal of the products that will improve the lives of mil- Biotechnol. 17, 149 (1999.). European Communities), “Skin Corrosion, Rat Skin lions of patients. 28. L. Griffith, B. Wu, M. J. Cima, B. Chaignaud, J. P. TER and Human Skin Model Assay” (OJ L 136, 8 June Vacanti, Ann. N.Y. Acad. Sci. 831, 382 (1997). 2000), http://embryo.ib.amwaw.edu.pl/invittox/prot/ 29. M. J. Mahoney, W. M. Saltzman, Nature Biotechnol. l_13620000608en00010089.pdf (2000). References and Notes 19, 934 (2001). 58. http://iccvam.niehs.nih.gov (Interagency Coordinat- 1. R. Langer, J. P. Vacanti, Science 260, 920 (1993). 30. J. Bonadio, E. Smiley, P. Patil, S. Goldstein, Nature ing Committee on the Validation of Alternative 2. M. J. Lysaght, J. Reyes, Tissue Eng. 7, 485 (2001). Med. 7, 753 (1999). Methods, 2001). 3. L. G. Griffith, Acta Mater. 48, 263 (2000). 31. L. L. Hench, J. M. Polak, Science 295, 1014 (2002). 59. B. D. DeBusschere, G. T. Kovacs, Biosens. Bioelectron. 4. A. J. Strain, J. M. Neuberger, Science 295, 1005 32. J. A. Hubbell, Curr. Opin. Biotechnol. 10, 123 (1999). 16, 543 (2001). (2002). 33. M. D. Pierschbacher, E. Ruoslahti, J. Biol. Chem. 262, 60. M. Krishnan, V. Namasivayam, R. Lin, R. Pal, M. A. 5. R. A. Pollak, H. Edington, J. L. Jensen, R. O. Kroeker, 17294 (1987). Burns, Curr. Opin. Biotechnol. 12, 92 (2001). G. D. Gentzkow, Wounds 9, 175 (1997). 34. S. P. Palecek, J. C. Loftus, M. H. Ginsberg, D. A. 61. E. Sachs, M. Cima, P. Williams, D. Brancazio, J. Cornie, on February 3, 2016 6. E. A. Ryan et al., Diabetes 50, 710 (2001). Lauffenburger, A. F. Horwitz, Nature 385, 537 (1997). J. Eng. Ind. 114, 481 (1992). 7. S. C. Strom, Sem. Liver Dis. 19, 39 (1999). 35. K. Eid, E. Chen, L. G. Griffith, J. Glowacki, J. Biomed. 62. Research funded by NIH (National Institute of Gen- 8. J. Mansbridge, K. Liu, R. Patch, K. Symons, E. Pinney, Mater. Res. 57, 224 (2001). eral Medical Sciences and National Institute of Envi- Tissue Eng. 4, 403 (1998). 36. J. C. Schense, J. A. Hubbell, J. Biol. Chem. 275, 6813 ronmental Health Sciences), Defense Advanced Re- 9. A. Kern, K. Liu, J. Mansbridge, J. Invest. Dermatol. 117, (2000). search Project Agency ( Tissue-based Biosensors Pro- 112 (2001). 37. G. Maheshwari, G. L. Brown, D. A. Lauffenburger, A. gram), NSF (Engineering Research Center Program), 10. M. Brittberg, T. Tallheden, B. Sjogren-Jansson, A. Lin- Wells, L. G. Griffith, J. Cell Sci. 113, 1677 (2000). the DuPont–Massachusetts Institute of Technology dahl, L. Peterson, Clin. Orthop. 391 (suppl.), S337 38. D. J. Irvine, A. M. Mayes, L. G. Griffith, Biomacromol- Alliance, the Whitaker Foundation, and Therics, Inc. (2001). ecules 2, 84 (2001). We thank R. Langer, J. P. Vacanti, A. Grodzinsky, S. R. 11. M. F. Pittenger et al., Science 284, 143 (1999). 39. H. Hall, T. Baechi, J. A. Hubbell, Microvasc. Res. 62, Tannenbaum, and D. A. Lauffenburger for helpful 12. N. Lumelsky et al., Science 292, 1389 (2001). 315 (2001). comments. Downloaded from

VIEWPOINT Third-Generation Biomedical Materials Larry L. Hench* and Julia M. Polak*

1014 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN

CONTINUED FROM PAGE 1014 ids hydrolyticly decomposes into CO2 and routes of repair are now available with the H2O. By 1984 clinical use of resorbable poly- use of these tailored biomaterials. ert tissue response to instead producing bio- mers as sutures was routine. Resorbable frac- Tissue engineering. Progenitor cells are active components that could elicit a con- ture fixation plates and screws in orthopedics seeded onto modified resorbable scaffolds. trolled action and reaction in the physiologi- and controlled-release drug-delivery systems The cells grow outside the body and become cal environment (2). The mechanism of were in their infancy (10). differentiated and mimic naturally occurring bonding of bioactive glasses (composed of The clinical success of bioinert, bioactive, tissues. These tissue-engineered constructs

Na2O-CaO-P2O5-SiO2) to living tissue (3) and resorbable implants has been a vital re- are then implanted into the patients to replace was shown to involve a sequence of 11 reac- sponse to the medical needs of a rapidly diseased or damaged tissues. With time the tion steps (4). The first five steps occurred on aging population. However, survival analyses scaffolds are resorbed and replaced by host the surface of the material and involved rapid of skeletal prostheses (7, 11) and artificial tissues that include a viable blood supply and ϩ ϩ ϩ ion exchange of Na with H and H3O heart valves (12) show that a third to half of nerves. The living tissue-engineered con- followed by a polycondensation reaction of prostheses fail within 10 to 25 years, and structs adapt to the physiological environ- surface silanols to create a high–surface area patients require revision surgery. Twenty ment and should provide long-lasting repair. silica gel, which provided a large number of years of research has had only small effects Clinical applications include repair of articu- sites for heterogeneous nucleation and crys- on failure rates (7), and continuing this path lar cartilage, skin, and the vascular system, tallization of a biologically reactive hydroxy- of trial-and-error experiments that require use although stability of the repaired tissues carbonate apatite (HCA) layer equivalent to of many animals and human clinical trials is needs improvement. the inorganic mineral phase of bone. The prohibitively expensive. Improvements of In situ tissue regeneration. This approach growing HCA layer on the surface of the first- and second-generation biomaterials are involves the use of biomaterials in the form material provided an ideal environment for limited in part because all man-made bioma- of powders, solutions, or doped micropar- six cellular reaction steps that included colo- terials used for repair or restoration of the ticles to stimulate local tissue repair. Bioac- nization by osteoblasts (the cells that make body represent a compromise (1). Living tis- tive materials release chemicals in the form bone), followed by proliferation and differen- sues can respond to changing physiological of ionic dissolution products, or growth fac- tiation of the cells to form new bone that had loads or biochemical stimuli, but synthetic tors such as bone morphogenic protein a mechanically strong bond to the implant materials cannot. This limits the lifetime of (BMP), at controlled rates, by diffusion or surface. artificial body parts. It also signals that we network breakdown, that activate the cells in By the mid-1980s bioactive materials had have reached a limit to our current medical contact with the stimuli. The cells produce reached clinical use in a variety of orthopedic paradigm that emphasizes replacement of tis- additional growth factors that in turn stimu- and dental applications, including various sues. It is time to consider a shift toward a late multiple generations of growing cells to compositions of bioactive glasses, ceramics, more biologically based method for the repair self-assemble into the required tissues in situ glass-ceramics, and composites. Synthetic and regeneration of tissues. along the biochemical and biomechanical hydroxyapatite (HA) ceramics began to be gradients that are present. routinely used as porous implants, powders, Third-Generation Biomaterials: Cell- For example, when a particulate of bio- and coatings on metallic prostheses to pro- and Gene-Activating Materials active glass is used to fill a bone defect vide bioactive fixation (5, 6). The presence of Third-generation biomaterials are being de- there is rapid regeneration of bone that sparingly soluble HA coatings led to a tissue signed to stimulate specific cellular responses matches the architecture and mechanical response (termed osteoconduction) in which at the molecular level. The separate concepts properties of bone at the site of repair. Both bone grew along the coating and formed a of bioactive materials and resorbable materi- osteoconduction and osteoproduction (13) mechanically strong interface. Bioactive als have converged; bioactive materials are occur as a consequence of rapid reactions on glasses and glass-ceramics were used as mid- being made resorbable and resorbable poly- a bioactive glass surface (2, 4). The surface dle-ear prostheses to restore the ossicular mers are being made bioactive. Molecular reactions release critical concentrations of chain and treat conductive hearing loss and as modifications of resorbable polymer systems soluble Si, Ca, P, and Na ions that give rise to oral implants to preserve the alveolar ridge elicit specific interactions with cell integrins both intracellular and extracellular responses from the bone resorption that follows tooth and thereby direct cell proliferation, differen- at the interface of the glass with its cellular extraction (7). A mechanically strong and tiation, and extracellular matrix production environment. tough bioactive glass-ceramic was used for and organization. Third-generation bioactive replacement of vertebrae in patients with spi- glasses and macroporous foams are being Genetic Control and Activation nal tumors (8). By the 1990s bioactive com- designed to activate genes that stimulate re- Rapid repair of bone requires differentiation posites, such as HA particles in a polyethyl- generation of living tissues. Two alternative as well as proliferation of osteoblasts. A syn- ene matrix, had become important in the repair and replacement of bones in the middle ear (9). Fig. 1. Confocal micrograph of primary human osteoblasts cultured on Another advance in this second generation a 45S5 monolith for 48 hours. Cells was the development of resorbable biomate- were stained with phalloidin (green) rials that exhibited clinically relevant con- for F-actin and propidium iodide trolled chemical breakdown and resorption. (red) for nuclei. A high level of actin In this manner, the interface problem is re- organization can be observed with ␮ solved, because the foreign material is ulti- parallel stress ﬁbers. Bar, 10 m. [Image supplied by J. Gough, Impe- mately replaced by regenerating tissues, and rial College] ultimately there is no discernible difference between the implant site and the host tissue (1). An example of this is the biodegradable suture, in which the polymer composed of polylactic (PLA) and polyglycolic (PGA) ac-

1016 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN chronized sequence of genes must be acti- Molecularly Tailored Resorbable invasive surgery. There are important economic vated in the osteoblasts so that they under- Polymers advantages to each of these new approaches that go cell division and then synthesize an Third-generation biomaterials that involve mo- may aid in solving the problems of caring for an extracellular matrix that is capable of min- lecular tailoring of resorbable polymers for spe- aging population. It should be feasible to design eralizing to become bone. Recent research cific cellular responses show great promise. By a new generation of gene-activating biomateri- shows that there is genetic control of the immobilizing specific proteins, peptides, and als tailored for specific patients and disease cellular response of osteoblasts to bioactive other biomolecules onto a material it is possible states. Tissue-engineered constructs based on a glasses. Seven families of genes are up- to mimic the extracellular matrix (ECM) envi- patient’s own cells may be produced that can be regulated within 48 hours of the exposure ronment (19) and provide a multifunctional cell- used to select an optimal pharmaceutical treat- of primary human osteoblasts to the ionic adhesive surface (20–22). Cell-specific recogni- ment. Perhaps of even more importance is the dissolution products of bioactive glasses tion factors can be incorporated onto the resorb- possibility that bioactive stimuli can be used to (14). The activated genes express numer- able polymer surface, including the adhesive activate genes in a preventive treatment to main- ous proteins that influence all aspects of protein fibronectin or functional domains of tain the health of tissues as they age. Only a few differentiation and proliferation of osteo- ECM components (23). Polymer surfaces can be years ago this concept would have seemed un- blasts: (i) transcription factors and cell- tailored with proteins that influence interactions imaginable. But we need to remember that only cycle regulators; (ii) signal transduction with endothelium (24), synaptic development 30 years ago the concept of a material that molecules; (iii) proteins involved in DNA (25), and neurite stimulation (26). would not be rejected by living tissues also synthesis, repair, and recombination; (iv) Cell transplants offer promise for treatment seemed unimaginable. growth factors and cytokines that influence of neurological disorders, such as Parkinson’s the inflammatory response to the material; disease. However, to restore function within References and Notes (v) cell-surface antigens and receptors; (vi) degenerating regions of the central nervous sys- 1. L. L. Hench, Science 208, 826 (1980). 2. ࿜࿜࿜࿜ , J. W. Wilson, Science, 226 630 (1984). extracellular-matrix components; and (vii) tem, transplanted cells must differentiate and 3. L. L. Hench, R. J. Splinter, W. C. Allen, T. K. Greenlee apoptosis regulators. extend axons that form synaptic contacts within Jr., J. Biomed. Mater. Res. 2, 117 (1971). Use of the dissolution products of resorb- the host tissue. Creating a local environment for 4. L. L. Hench, J. Am. Ceram. Soc. 81, 1705 (1998). able bioactive gel-glasses to stimulate cellu- transplanted brain cells that enhances this regen- 5. T. Yamamuro, L. L. Hench, J. Wilson, Eds., CRC Hand- lar repair at a molecular level offers promise erative process is essential. A third-generation book of Bioactive Ceramics, vol. 2, Calcium Phosphate and Hydroxylapatite Ceramics (CRC Press, Boca Ra- for creating scaffolds for bone tissue engi- biomaterial shows promise for achieving this ton, FL, 1990). neering. Under appropriate culture condi- microenvironment. Mahoney and Saltzman (27) 6. C. P. A. T. Klein, J. G. C. Wolke, K. deGroot, in An tions, differentiating embryonic stem (ES) have developed a technique that creates local Introduction to Bioceramics, L. L. Hench, J. Wilson, cells can be induced to form bone nodules and sustained levels of insoluble and soluble Eds. (World Scientific, London, 1993), chap. 11, p. 199. (15); these structures are multilayers of cells molecules directly at the site of cell transplan- 7. L. L. Hench, J. Wilson, Eds., Clinical Performance of embedded in a mineralized extracellular ma- tation in the brain. PLA/PGA copolymers were Skeletal Prostheses (Chapman & Hall, London, 1996), trix that contains type I collagen and osteo- used to incorporate nerve growth factor (NGF) chaps. 13 and 15. calcin. When a standard osteoblast differen- and release it at controlled rates. Cells were 8. T. Yamamuro, in CRC Handbook of Bioactive Ceram- ics, T. Yamamuro, L. L. Hench, J. Wilson, Eds., vol. 2, tiation medium is conditioned by exposure to assembled with the cell-adhesive/controlled- Calcium Phosphate and Hydroxylapatite Ceramics resorbable bioactive gel-glasses for 24 hours release microparticles to form transplantable (CRC Press, Boca Raton, FL, 1990), chap. 6, p. 89. and then applied to differentiating ES cells, a neo-tissues. NGF delivery by way of the syn- 9. W. Bonfield, in CRC Handbook of Bioactive Ceramics, dose-dependent increase in the numbers of thetic microenvironment increased levels of T. Yamamuro, L. L. Hench, J. Wilson, Eds., vol. 2, Calcium Phosphate and Hydroxylapatite Ceramics bone nodules formed compared with control NGF-induced biological activity over the course (CRC Press, Boca Raton, FL, 1990), chap. 16. cultures is seen (16). Bioactive scaffolds of 21 days in vivo. A similar approach with 10. G. Hastings, P. Ducheyne, Eds. Macromolcular Bioma- have been made that release optimal concen- molecularly tailored polymers has been used to terials (CRC Press, Boca Raton, FL, 1984). 11. B. M. Wrobelewski, P. A. Fleming, P. D. Siney, J. Bone trations of the ionic dissolution products as enhance directional regeneration of nerves (26). Jt. Surg. Br. Vol. 81, 427 (1999). they resorb in the presence of adherent hu- 12. F. J. Schoen, R. J. Levy, H. R. Piehler, J. Soc. Cardiovasc. man osteoblasts (Figs. 1 and 2). Molecular Implications for the Future Pathol. 1, 29 (1992). 13. J. Wilson, S. B. Low, J. Appl. Biomater. 3, 123 (1992). modifications of the three-dimensional A cellular and molecular basis for development 14. D. Xynos, A. J. Edgar, L. D. K. Buttery, L. L. Hench, J. M. resorbable inorganic scaffolds by chemisorp- of third-generation biomaterials provides the Polak, J. Biomed. Mater. Res. 55, 151 (2001). tion of surfactant proteins have been made to scientific foundation for molecular design of 15. L. D. K. Buttery et al., Tissue Eng. 7, 89 (2001). 16. R. Bielby, R. Pryce, L. L. Hench, J. M. Polak, Tissue Eng. enhance attachment and proliferation of lung scaffolds for tissue engineering and for in situ 6, 693 (2000). cells (17, 18). tissue regeneration and repair, with minimally 17. R. F. S. Lenza, J. R. Jones, W. Vasconcelos, L. L. Hench, J. Mater Sci. Mater. Med., in press. 18. H. Romanska, A. Bishop, J. M. Polak, personal Fig. 2. Scanning electron micrograph communication. of primary human osteoblasts cul- 19. B. S. Kim, D. J. Mooney, Trends Biotechnol. 16, 224 tured on porous 58S foam for 24 (1998). hours. Cells can be observed bridg- 20. B. D. Ratner, J Mol. Recognit. 9, 617 (1996). ing smaller pores with fine cellular 21. J. A. Hubbell, Biotechnology 13, 565 (1995). projections. In addition, calcium 22. J. J. Marler, J. Upton, R. Langer, J. P. Vacanti, Adv. Drug phosphate crystals can be observed Deliv. Rev. 33, 165 (1998). covering the material surface and 23. R. A. Quirk, W. C. Chan, M. C. Davies, S. J. B. Tendler, K. M. Shakesheff, Biomaterials 22, 865 (2001). within the porous structure. Bar, 50 ␮ 24. J. A. Hubbell, S. P. Massia, N. P. Desai, P. D. Drum- m. [Image supplied by J. Gough, heller, Biotechnology 9, 568(1991). Imperial College] 25. D. D. Hunter et al., J. Neurosci. 11, 3960 (1991). 26. N. Patel et al., FASEB J. 12, 1447 (1998). 27. M. J. Mahoney, W. M. Saltzman, Nature Biotechnol. 19, 934 (2001). 28. We gratefully acknowledge T. Ryder and M. Mobber- ley (Department of Histopathology, Charing Cross Hospital, London) for assistance with the scanning electron microscopy.

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1017 B ODYBUILDING:THE B IONIC H UMAN NEWS New Prospects for Putting Organs on Ice

After a lull, scientists are again exploring vitrification and other groups have recently shown that they can vit- techniques for deep-freezing tissues and organs rify corneas, getting much less damage than from freezing. But the scientists have yet to In the movie Vanilla Sky, Tom Cruise has his space between cells, and this ice destroys show whether these corneas function in vivo. broken body frozen in the hope that he’ll fragile structures such as ducts and blood Roger Gosden and colleagues at McGill someday be revived and healed. In reality, vessels. Sometimes pieces of organs, such as University in Montreal have had success even cryonics, as this practice is known, remains pancreatic islet cells, work adequately after with conventional freezing with a small organ, the most speculative science fiction. Re- freezing. But larger, more complex organs they reported last month in Nature. They froze searchers have failed for decades to deep- such as kidneys don’t function properly rat ovaries, fallopian tubes, and attached blood freeze and thaw most tissues—let alone or- when sufficiently ravaged. vessels in liquid nitrogen and transplanted gans or animals—without damaging them, As a way around this problem, some re- them into genetically identical rats whose often seriously. But recently, a few cryobiol- searchers turned to vitrification, or “ice-free” ovaries had been removed. The animals ovu- ogists have celebrated successes with ovaries cryopreservation. The idea is to fill the organ lated. Some ice damage occurred, so vitrifica- and complex tissues like vascular grafts. with a viscous fluid that turns into a glassy tion might be even more successful, suggests The work has sparked (not crystalline) solid at Gosden, now at the Jones Institute for Repro- hope that donated organs low temperatures. This ductive Medicine in Norfolk, Virginia. can eventually be banked reduces the problems of Several groups are tackling problems that for longer than the current ice, but toxicity of the thwart both vitrification and conventional few days, which would cryopreservant can still freezing. Pegg’s group at York, for instance, is buy time for distributing Image not damage organs. And ice working on how to thaw tissues and avoid ice them and finding recipi- crystals tend to form crystal formation. The team has developed a ents who are good im- available for as vitrified tissue—espe- technique that Pegg says can evenly and munological matches. And online use. cially large pieces—is quickly heat pingpong ball–sized clumps of as progress is made engi- warmed. Partial success cells embedded in gelatin to simulate large tis- neering artificial livers, came in the 1980s, when sue. Taylor’s group, meanwhile, is collaborat- bladders, and other tissues Red Cross scientist Greg ing with Carnegie Mellon scientists on dosing containing living cells, the Fahy showed that rabbit tissues with iron compounds that are then ex- need for better preserva- kidneys could withstand cited with magnets to generate heat. on February 3, 2016 tion techniques is expected the high concentrations Taking a cue from nature, researchers are to grow (see Viewpoint on of cryoprotectants need- also using natural antifreeze proteins to help p. 1009). “Each company ed to vitrify these organs. mop up crystals formed during freezing and at some point will need to They worked when reim- thawing. Many organisms, from carrots to store and transport their Image not planted, but Fahy had on- fish to beetles, produce proteins that latch product,” says Mike Taylor ly cooled them to –3°C. onto and isolate growing ice crystals. The of Organ Recovery Sys- available for Many experiments later, one drawback is that at temperatures well be-

tems in Charleston, South online use. it’s clear that “vitrifica- low 0°C, this system can backfire by causing Downloaded from Carolina. tion is very, very compli- ice crystals to form spikes that disrupt cells. Currently, transplant cated,” says cryobiologist Several companies are developing improved surgeons perfuse whole David Pegg of the Uni- synthetic versions of these “ice blockers”; organs with a special solu- versity of York, U.K., and for example, Taylor’s company hopes to pro- tion that enables them to in most labs vitrification duce smaller molecules that attach to ice be banked just above 0°C Vanquishing ice. Damaging ice crys- of large organs has been crystals at the base as well as the face, which for up to a few days. Ideal- tals form in a rabbit jugular vein that’s on hold. could prevent spike formation. ly, they would like to pre- frozen (top), but they’re absent from a Recently, cryobiolo- Fahy, meanwhile, has never given up try- serve organs at –196°C, vein that was vitrified. gists have had better luck ing to vitrify large organs. “He has soldiered the boiling point of liquid studying vitrification on a the way on this for years and years,” Pegg nitrogen—so cold that molecular motion vir- smaller scale. For example, fine-tuning the says. Now at a company called 21st Century tually stops and tissues cease to decay. Half a solutions that are injected into the organs, as Medicine in Rancho Cucamonga, California,

(18 MARCH 2000) century ago, that seemed within easy reach. well as heating and cooling rates, minimized Fahy has tested hundreds of vitrification solu- Scientists found that blood and sperm could injury to 2- to 3-centimeter-long pieces of rab- tions and patented the most promising ones. survive such deep freezing if mixed with glyc- bit veins, Taylor’s team at Organ Recovery “Greg has always been tantalizingly close to erol. The glycerol lowers the cells’ freezing Systems reported in the 18 March 2000 issue getting it to work,” says cryobiologist William point and keeps them from getting lethally of Nature Biotechnology. The vessels retained Rall of the National Institutes of Health. The salty when they do freeze and water diffuses 80% of their function when dosed with drugs company Fahy works for receives funding out. And since 1972, embryos have been that cause them to contract, compared to 20% from the Life Extension Foundation, which

ET AL., NATURE BIOTECHNOLOGY frozen with liquid nitrogen and later success- following simple freezing. When implanted in supports cryonics. Fahy says his work is strict- fully implanted. rabbits, the grafts appeared to work normally. ly limited to cryopreserving organs. But he Organs, however, don’t hold up so well “We’re the first to demonstrate [that vitrifica- adds, “If I’m successful, perhaps it will re- below the freezing point. Water leaked from tion works better than freezing] in reasonably move some of the [cryonics] controversy.”

CREDITS: Y. C. SONG cells during freezing forms ice crystals in the complex tissue,” Taylor says. Two other –JOCELYN KAISER

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1015 B ODYBUILDING:THE B IONIC H UMAN VIEWPOINT The Bionic Man: Restoring Mobility William Craelius*

Bionics engineers are making increasingly bold and successful use of their commands from their brains (or spinal tools to restore mobility to persons with missing or nonfunctional limbs. cords) are harnessed either noninvasively These tools include the latest materials, minielectronics and megacom- with the electroencephalogram (EEG) or puters, advanced robotic mechanisms, and algorithms. With crucial help with implanted electrodes. Paralyzed pa- from their pioneering users, they are learning how and where the residual tients fitted with a brain-implanted chip sensorimotor system can be tapped in order to transmit its intents to have learned to move cursors and select replacement or reactivated body parts. letters on a computer (8), as well as to direct robotic arm movements using tech- When humans replace a missing body part to alternate regions (4). This procedure trans- nology known as brain-computer interface with an artificial one, they begin an inti- fers hand-control signals to healthy pectoral (BCI) (9). These systems operate by train- mate relationship with a partner they barely muscles, where they can be conveniently ac- ing an algorithmic filter to associate specif- know. The key to success for such a rela- cessed and deciphered. ic movement requests with specific neuro- tionship may be no different than that The second requirement is more challeng- nal signal patterns, recorded either directly found in marriage manuals: communica- ing owing to the complexity of human move- from neurons or indirectly from EEGs or tion. Unfortunately, for bionic parts, com- ment control. Each action originates with a other noninvasive signals. The filter then munication is the weakest link in the chain few neurons in the motor cortex that trigger a directs the appropriate action after recog- of components that includes electronics, large neural network that coordinates the ac- nizing, or decoding, the volition . computing, actuators, mechanisms, and tivities of several effector muscles after re- How much more function can be restored materials, all of which are adequate for the ceiving and processing feedback information by advanced bionic systems? Complex activi- application. This situation is perhaps best from thousands of tactile, positional, and vi- ties, such as walking, could probably not be illustrated by the problem of hand restora- sual senses. Transforming this tangled mesh achieved by noninvasive methods such as EEG, tion. NASA, for example, has developed a of millions of electrical pulses into graceful owing to their poor resolution of brain activity. robotic hand that approximates human movements is a routine accomplishment of The difficulty in extracting volitional requests dexterity, moving up to 22 joints indepen- our sensorimotor system that bionics engi- from the EEG is underscored by the limi- dently (1). Available prosthetic technology, neers can only envy. Its artistry can be appre- tations of present technology, which cannot however, can control only one joint at a ciated by comparing the elegance and adapt- decipher from the brain more than 25 bits on February 3, 2016 time, leaving amputees with little hope of ability of human motions with that of the of information (three characters) per using the advanced hand. This glaring mis- most advanced walking robots, whose latest minute. This rate is many thousands of match between machine and human capa- achievement is the slow climbing of stairs times too slow to control even the simplest bilities clearly reflects the inadequate lines (5). Restoration of lost sensorimotor function movements. Finer resolution and hence of communication. On the bright side, bion- by robotic assistance is correspondingly chal- ics researchers are opening new vistas for lenging, but is steadily progressing, as exem- restoring lost human functions as they plified by robotic devices such as the steadily bridge the gap between human and RoboWalker, an active exoskeleton that as- Downloaded from machine. sists walking by motor-impaired individuals Bionics can restore lost mobility to in- (6). dividuals if (i) they can express cognitive The human sensorimotor system, though control over relevant motor functions still far too complex to duplicate, is being somewhere in their residual anatomy and invaded by increasingly more versatile (ii) a device can pick up and decipher that bionic interfaces. Our bionic potential was cognition. The first requirement is ade- recently demonstrated by a monkey in quately satisfied by many individuals who Brooklyn, whose brain signals, monitored have lost function either through paralysis by electrodes, controlled a three-dimen- or amputation, and who can both sense and sional (3D) robotic arm located in North imagine manipulating the nonfunctional or Carolina, while he watched it on the Inter- absent joints. These individuals express net (7). This feat and similar ones were motor control over their lost parts, includ- taught to the bionic animals (rats as well as ing legs, joints, and individual fingers, primates) through the presentation of food through nervous and/or muscular activity rewards, and did not require actual move- directed to their residual limbs, and these ment of the animals’ own limbs, because expressions can be registered by appropriate they quickly learned that movement was technology (2, 3). Patients who do not meet unnecessary to be rewarded. Cognitive con- this requirement because of damaged residual trol of artificial limbs, at least for primitive Fig. 1. Bionic restoration of hand mobility. This muscles may soon have a surgical option, but important motions such as grasping, user, having nerve damage as a result of spinal whereby their hand motor nerves are rerouted thus can be achieved with a bionic brain- cord injury, can grasp objects when his forearm machine interface (BMI) for individuals muscles are stimulated with Bions. The control computer in his shirt pocket reads his inten- with paralyzed or missing limbs. Department of Biomedical Engineering, Rutgers Uni- tions to grasp and activates the wire coils versity, Piscataway, NJ 08855–0909, USA. *E-mail: Humans with motor disabilities can al- around his arm, to deliver electrical pulses to [email protected] ready surpass animal performance when the Bions (brown rods).

1018 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN more functionality can be achieved by di- the brain in control of movement, alterna- switches on his walker that control a stimulator rect recording with cortically implanted tive bionic approaches bypass the brain, chip implanted in his spinal cord (13, 14). The electrodes if a sufficient number of them directly communicating with peripheral strictly peripheral approach, however, requires can be permanently located in the brain. nerves and muscles. These peripheral-ma- FES to micromanage all actions, much like that The minimum number of individual neu- chine interfaces (PMIs) operate by func- of a puppet master. For spinal cord–injured rons required to transform thoughts into a tional electrical stimulation (FES) of mus- persons, this detail is a welcome remedy for reasonable range of motions probably ex- cles and peripheral nerves and can be pro- immobility; more natural and graceful control ceeds 1000 (10). This is achievable with grammed for specific movement patterns. may eventually be possible through devel- microwires in the brain; however, the re- A successful example is FreeHand, one of opments of hybrid brain-machine interfaces cording electronics for 1000 channels is the first commercially available and Food (HBMIs), whereby volitional EEG signals currently too bulky to fit in the cranium. and Drug Administration–approved FES recorded directly from the brain can control For now, the analog electronics of a versa- devices, pioneered by Peckham (11, 12). the muscles (9, 10). tile BCI or BMI would need to be located FreeHand restores grasping to patients Bionic coordination of muscle movements externally and receive brain signals by with upper-limb paralysis or weakness by has been advanced by a new device called wireless transmission. Near–real time oper- giving them control over extrinsic hand Bion (Fig. 1) (15). Bions are single-channel ation of such a device is possible, because muscles through movements of their oppo- stimulators about the size of a long grain of earlier experiments showed that volitions site shoulder that generates radio waves to rice that can be injected into muscles with a could be deciphered from monkeys by sim- activate electrodes in the forearm. 12-gauge needle and controlled by an exter- ply computing the weighted sum of firing Bionic technologies can be adapted for renal radio-frequency coil. Their excellent lon- rates of several cortical neurons (7). storing some degree of almost any lost function. gevity and functionality in vivo was demon- Whereas brain interfaces tap into the Their broader applications are highlighted by a strated in clinical trials with users for over 1 head of the sensorimotor system, putting paraplegic who now ambulates by operating year. Bions can independently control each of the many muscles involved in coordinated movements, given the appropriate motor commands. Direct control over muscles is desirable because human muscles, in contrast to robotic actuators, respond to their natural controller, i.e. neurons, in a highly nonlinear and unpredictable manner, which is not yet understood. When leaving the brain out of the control loop, bionics engineers must somehow de- code volition at the periphery. The most common approach is to train users to execute specific muscle activities to produce surface electromyographic (EMG) patterns recogniz- able by the decoder. This approach can restore a limited number of activities, such as 1D grasping; however, it does not adequately resolve volition having more than one degree of freedom (16). An alternative to control by EMG, developed by Sam Phillips and others in our laboratory, registers volitions by the entire 3D pattern of forces in the residuum (residual limb), known as residual kinetic imaging (RKI) (17, 18). Such patterns can be discriminated by using filters that can be readily trained and retrained as needed. This adaptibility lessens the requirement for precise placement of sensors, an important practical consideration for amputees whose residuum is constantly changing and who must don and doff their prosthesis daily. A key advantage of the RKI method is that it is biomimetic: The original motor pathways can be used to control robotic replacement parts, such as fingers (Fig. 2). Each approach to bionic restoration of movement is specialized for particular user characteristics (Table 1). For example, BCI is invaluable for severely paralyzed pa- Fig. 2. Biomimetic Dextra hand prosthesis. The silicone “smart sleeve” fits snugly over the residual tients, for whom simple communication limb and registers 3D forces produced by muscle activity within the hard socket. The pocket computer allows the user to retrain the robotic hand for optimal performance. The hand can flex with the outside world is a primary goal (8). and extend all five digits in response to commands from the natural motor pathways of the user. [Figure provided by D. Curcie] CONTINUED ON PAGE 1021

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1019 B ODYBUILDING:THE B IONIC H UMAN

Table 1. Strategies for bionic restoration of movement. Bionic approach (with electrodes Actuator types (with example Candidate users either implanted or external) references) BMI (brain-machine interface) Completely paralyzed persons who can beneﬁt from mechanical Motors, i.e., robotic arms controlled assist devices by monkeys (7) BCI (brain-computer interface) Completely paralyzed persons who wish to simply communicate Computer screen, i.e., moving a cursor (8) PMI (peripheral- machine interface) Amputees and persons with intact central nervous system but Motors, i.e., prosthetic hand (23) weak muscles HBMI (hybrid brain-machine interface) Spinal cord–injured persons with intact limb muscles Muscles, i.e., direct brain control of FreeHand (9) CBI (computer-brain interface) Parkinson’s disease Muscles (19, 20)

CONTINUED FROM PAGE 1019 risk is so severe that some bionic systems, tic sockets for limbs, are glaring anachro- such as the Jarvik 2000 heart (21), have nisms relative to other bionic components. The converse of BCI, computer-brain inter- eliminated all digital electronics. Such pre- Users who subject themselves to brain im- face (CBI), can treat conditions such as Par- cautions may not be an option for bionics plantation of hundreds of electrodes and kinson’s disease, wherein thalamic neuronal that restore mobility, wherein digital pro- wearing of transmitters may rightly expect activity is substantially impaired, causing cessing is fundamental, and hence new anti- a more versatile and responsive attachment tremors. The first commercially available de- interference strategies may be required. Bat- of their arm or leg. To help meet these vice for this purpose, Activa Tremor Control tery energy density and recharging issues will expectations, much fruitful work is focused (19, 20), operates from a small computer, become limiting especially as orthotic and on how to integrate prosthetic structural located in the chest cavity, that rhythmically prosthetic devices gain functionality and de- components, such as the titanium pylon, stimulates the thalamus to simulate the oper- mand more power. Maximal energy density with bone, i.e., osseointegration (25). The ation of the diseased neurons. of implantable batteries is about 1.1 W⅐hours/ natural feel, or “osseoperception,” of the Persons with an intact central nervous sys- cm3 (lithium ion); however, external battery environment provided by such attachments tem who have completely lost function of packs that can use anode air-cells have higher can complete a sensory link that is crucial specific muscles can be aided by robotic de- densities. The slight risk of explosion by the to bionic restoration of function. vices controlled either directly from the brain latter cells is one willingly borne by users of with HBMI systems or from muscles or pe- powered prosthetic devices. The need for References and Notes ripheral nerves with PMI systems. Strictly more frequent recharging may require more 1. “Compact Dexterous Robotic Hand,” U.S. patent peripheral devices, such as the RoboWalker convenient options than radio-frequency number 6,244,644 (12 June 2001). 2. R. Abboudi, C. Glass, N. A. Newby, W. Craelius, IEEE mentioned earlier, can be controlled from transmission, such as optical recharging (22). Trans. Rehabil. Eng. 7, 121 (1999). muscle activation or movement patterns. Computer requirements for a typical 3. V. S. Ramachandran, Proc. Natl. Acad. Sci. U.S.A. 90, Such interfaces, possibly in combination with PMI hand controller include a relatively 10413 (1993). surgical reinnervation (4), could benefit para- high analog I/O throughput of many kilo- 4. T. A. Kuiken, D. S. Childress, W. Z. Rymer, Brain Res. plegics, amputees, and those with weak mus- baud but minimal CPU power and memory. 676, 113 (1995). 5. http://world.honda.com/robot/ (accessed 15 Novem- cles as a result of stroke, spinal cord injury, or The main processing task is decoding: de- ber 2001). neuromuscular disease. riving motor commands from volitions. The 6. www.yobotics.com/ Should progress continue at its present pace, task involves a mathematical operation 7. J. Wessberg et al., Nature 408, 361 (2000). human-machine communication could soon known as pseudoinversion that decodes sig- 8. P. R. Kennedy, R. A. E. Bakay, M. M. Moore, K. Adams, J. Goldwaithe, IEEE Trans. Rehabil. Eng. 8, 198 (2000). lose its distinction as the number one obstacle nals from several sources in near–real time, 9. J. R. Wolpaw et al., IEEE Trans. Rehabil. Eng. 8, 164 to bionics. It is relevant therefore to revisit other all of which contain a portion of the code (2000). technical problems in bionics. First, as noted for a complex motion, such as finger flex- 10. N. Nicolelis, Nature 409, 403 (2001). earlier, the size of electronic devices limits the ion (23). There are several alternative ap- 11. B. T. Smith et al., IEEE Trans. Biomed. Eng. 45, 463 (1998). functionality that may be implanted inside the proaches to decoding, none of which is 12. L. D. Maloney, Design News 90 (6 March 2000). brain or elsewhere. Analog recording electron- clearly superior, including neural networks, 13. P. Rabischong, Bull. Acad. Natl. Med. 183, 531 ics alone for a minimal brain interface (125 pattern-recognition algorithms, and hybrid (1999). electrodes) would occupy board space of at filters. 14. Reuters, “Chip helps paraplegic walk,” 20 March 2 2000, Strasbourg, France. least 60 cm , even with very large scale inte- Although actuators are not yet opti- 15. G. Loeb, R. A. Peck, W. H. Moore, K. Hood, Med. Eng. gration electronic technology. Sufficient minia- mized for bionic use, several commercially Phys. 23, 9 (2001). turization could be reached within a decade, available motors and servos are adequate 16. B. Hudgins, P. Paarker, R. N. Scott, IEEE Trans. Biomed. however, as long as “Moore’s law” is not re- for some advanced applications, including Eng. 40, 82 (1993). pealed. The latter is a remarkably accurate pre- a computer-modulated knee (24) and a 17. S. L. Phillips, W. Craelius, Robotica, in press. 18. S. L. Phillips et al., Sixth World Biomaterials Congress, diction from the year 1965 that transistor den- multifinger hand (23). The large sizes and Hawaii, #1343, 15 to 20 May 2000. sity on integrated circuits would double every power inefficiencies of present actuators, 19. A. Benabid et al., Arch. Med. Res. 31, 282 (2000). year (with a slight revision in 1975 to doubling however, limit functional expansion of ro- 20. J. Rosengren, IEEE Spectrum 35, 62 (1998). every 1.5 years). botic assist devices. 21. R. Jarvik, V. Scott, M. Morrow, E. Takecuchi, Artif. Organs 23, 487 (1999). A complication of shrinking size is the A final challenge for bionics is estab- 22. K. Goto, T. Nakagowa, O. Nakamura, S. Kawata, IEEE difficulty of hermetically sealing small ob- lishing a convenient physical interface be- Trans. Biomed. Eng. 48, 830 (2001). jects and protecting them from corrosion in tween effector devices and the body. Al- 23. D. J. Curcie, J. Flint, W. Craelius, IEEE Trans. Neural bodily fluids. Another issue for electronics though limb replacement as depicted by Syst. Rehabil. Eng. 9, 69 (2001). 24. J. Michael, Clin. Orthop. Relat. Res. 361 43 (1999). is susceptibility, especially of digital sys- Hollywood will likely remain a fantasy, 25. R. Branemark, P. I. Branemark, B. Rydevik, R. R. Myers, tems, to electromagnetic interference. This current interfaces, consisting of bulky plas- J. Rehabil. Res. Dev. 38, 175 (2001).

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1021 B ODYBUILDING:THE B IONIC H UMAN NEWS Part Man, Part Computer: Researcher Tests the Limits Kevin Warwick plans to connect a computer chip to the nerves in his arm Warwick and others in the cybernetics to see if a computer can read and communicate signals directly from his community envision a world in which hu- nervous system. Are computer-controlled humans next? mans are able to expand their senses to hear ultrasonic sounds or see infrared wavelengths. By the end of this month, Kevin Warwick the robot by simply moving his finger. “It’s tremendously exciting. Can we in the fu- hopes to be a cyborg. If all goes as planned, in Animal experiments suggest that goal is ture link extra memory into our brains? Why late February the University of Reading, U.K., realistic. Several groups, including those of shouldn’t we do something like that?” he asks. professor of cybernetics will have surgery to neuroscientist Miguel Nicolelis at Duke Uni- Such questions are premature, says Peter connect nerves in his arm with wires leading versity in Durham, North Carolina, and An- Fromherz of the Max Planck Institute for Bio- to a “smart card”–sized collection of micro- drew Schwartz at the Neurosciences Institute chemistry in Martinsried, Germany. “War- processors. The wires will pick up signals in San Diego, have been able to program wick is a very interesting person. But what from his central nervous sys- he’s doing is scientifically crazy,” he says. tem and relay them via a radio For 15 years, Fromherz has been working transmitter to an external com- on experiments that join single neurons and puter that will record the pat- computer chips, with the goal of enabling re- terns. Warwick hopes the de- searchers to build better computer-enhanced vice will pick up discrete sig- prosthetic devices. In a paper published nals from the nerves depending l November in the Proceedings of the Na- on his movements, his sense of Image not tional Academy of Sciences, his team de- touch, and even his mood, and available for scribed a nerve-computer circuit consisting then send those signals back online use. of snail neurons and a computer chip. The to his nerves to see if they circuit was able to send a discrete signal can mimic the movement from a computer chip to a neuron, from that or the sensation. Warwick’s neuron to a second neuron in a network, and wife plans to have a similar from the second neuron back to the comput- implant so the two can try er chip. On a cellular scale, Fromherz’s work on February 3, 2016 to communicate through is state of the art, but it is a long way from computer-mediated signals. computer-synthesized emotions, he notes. For decades, science fic- One small step? Kevin Warwick, here with his first computer chip Others think Warwick’s experiment tion writers have imagined be- implant, plans to become the first computer-enhanced person. should not be allowed to proceed. (It does not ings who are part human, part require permission from a formal ethics machine—Robocop, for instance. In the robot arms to crudely mimic the movement board, as Warwick is experimenting on him- nonfiction realm, cochlear implants have re- of a monkey’s arm based on patterns in the self.) Political scientist Langdon Winner of

stored hearing to deaf patients, and comput- animal’s brain waves. Warwick is curious to Rensselaer Polytechnic Institute in Troy, Downloaded from ers that sense brain waves have enabled par- see whether the computer chip can “play New York, for one, calls the experiment alyzed patients to communicate (Science, 29 back” such signals, triggering his arm to “profoundly amoral. Enhancing one’s October 1999, p. 888). But as far as Warwick move involuntarily or tricking him into information-processing ability by connect- knows, this is the first time anyone has at- thinking his finger has been stuck. ing chips to the nervous system marks a tempted to computer-enhance the nervous The bundle of nerves that runs down the very fundamental change in what human be- system of a healthy human. Some scientists arm, called the median nerve, also commu- ings are.” Should it become possible, he doubt Warwick will succeed, and others call nicates with the body’s limbic system—for says, “then it is a matter for theologians, the effort unethical. But Warwick says the example, making our palms sweat when politicians, and citizens to address.” potential benefits of computer enhancement we’re nervous. Warwick will attempt to And that’s the debate Warwick hopes to outweigh the risks, and he is eager to debate record the signals produced by shock or spark, much as the birth of Dolly fueled the just how far the technology should go. anger, then have a colleague try to send them debate over cloning. Although he sees great Warwick first made headlines in 1998 back to the implant when he is calm—and benefits from computer enhancement, left when he had a much simpler device implanted unsuspecting—to see if they have an effect. unchecked, the technology also has great po- into his arm. Like a built-in security badge, If the initial experiments go well, War- tential for harm, he says. For example, if the this computer chip allowed sensors in his lab wick’s wife, Irena, will receive her own im- computer is able to prompt Warwick to move building to detect his location and movements. plant a few weeks later. The pair will then at- his arm involuntarily, then that suggests that With the new device, Warwick is aiming for tempt to send nerve-mediated messages— a computer could someday remotely control much more. He hopes the computer will detect emotional or otherwise—through their com- a person instead of the other way around—a patterns of electrical signals that correspond to puter connection. If Warwick cuts his finger troubling prospect, he concedes. “If you’re movements or sensations: the bending of his while slicing a bagel, for instance, the chip creating superhumans,” Warwick says, “that index finger or the pain of a pinprick, for ex- should record the signal from his medial could mean the end of humanity.” And even ample. In one experiment, Warwick hopes to nerve and then send it to a computer that com- this cyborg hopeful thinks that’s a question send those signals to a robot. By trial and er- municates with his wife’s implant, adding that humans, not computers, should decide.

ror, he hopes he can learn to remotely control new meaning to the phrase “I feel your pain.” –GRETCHEN VOGEL CREDIT: N. NICOLAOU

1020 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN VIEWPOINT Will Retinal Implants Restore Vision? Eberhart Zrenner

A number of research groups are developing electrical implants that can be tissues—such as cochlear implants for the attached directly to the retina in an attempt to restore vision to patients hearing impaired [see the Viewpoint by suffering from retinal degeneration. However, despite promising results in Rauschecker and Shannon on page 1025 (2)] animal experiments, there are still several major obstacles to overcome or pacemakers for individuals with heart before retinal prostheses can be used clinically. disease—restoring vision with electrical devices implanted into the retina is much Vision is an enormously complex form of in- cells, and ganglion cells. Visual information more difficult. The transformation of visual formation processing that depends on a remark- from the retina’s 130 million photoreceptors is scenes into the electrical “images” carried able neuroprocessor at the back of the eye called compressed into electrical signals carried by 1.2 by the optic nerve to the brain requires that the retina. Seeing is initiated when light passing million highly specialized ganglion neurons, numerous sensory neurons are stimulated in through the pupil of the eye is focused by the whose axons form the optic nerve. The optic parallel and in a spatially correct order to lens onto the retina’s sensory neuroepithelium nerve transmits visual information via the lat- enable three-dimensional objects to be ac- (Fig. 1). This results in the projection of a eral geniculate nucleus to the primary visual curately encoded. reduced, upside-down image of the object onto cortex of the brain. the roughly 130 million photoreceptor cells Blindness can result when any step of the The Evolution of a Concept (rods and cones) in the outermost layer of the optical pathway—the optics, the retina, the In 1956, Tassiker described in a patent (3) how retina. The cones, providing chromatic (color) optic nerve, visual cortex, or other cortical a small, flat, light-sensitive selenium cell placed images of high spatial resolution, and the rods, areas involved in the processing of vision— behind the retina of a blind patient transiently required for achromatic vision with less spatial sustains damage. In Germany, 17,000 pa- restored the patient’s ability to perceive the resolution in dim light, transform local lumi- tients become blind every year for whom sensation of light. Later attempts to restore vi- nance and color patterns of the projected image there is no effective treatment or cure; about sion by coupling electrodes to the surface of the into electrical and chemical signals. These sig- 50% of all blindness is caused by damage to visual cortex of blind patients (4, 5) did not nals then activate a complex circuit of retinal the retina (1). Blinding diseases, such as ret- provide useful images because of limited spatial neurons: horizontal cells, bipolar cells, amacrine initis pigmentosa or age-related macular de- resolution and the fading of phosphenes (sen- generation (the most common form of blind- sations of light). Subsequent human trials with

ness in the elderly), cause progressive degen- cortical implants have been more promising on February 3, 2016 University Eye Hospital Tu¨bingen, Department of eration of the outer retina. Although there are (6–8), but diminished neuronal excitations and Pathophysiology of Vision and Neuro-Ophthalmolo- gy, Schleichstr. 12, D - 72076 Tu¨bingen, Germany. many examples of electrical devices that can stable spatial resolution are still unsolved prob- E-mail: [email protected] support or replace the function of defective lems, even with 100 narrowly spaced intracor- Downloaded from

Fig. 1. An object (in this case a face) is projected by the cornea and lens currents into the remaining neural cells (horizontal cells, bipolar cells, onto the retina in an upside-down manner and is transformed into an amacrine cells, and ganglion cells) of the retinal inner layer. In contrast, electrical image by the photoreceptor cells (rods and cones) of the outer the epiretinal implant has no light-sensitive areas but receives electrical retina. With a subretinal implant, the rods and cones are replaced by a signals from a distant camera and processing unit outside of the body. silicon plate carrying thousands of light-sensitive microphotodiodes, Electrodes in the epiretinal implant (small black knobs) then directly each equipped with a stimulation electrode. Light from the image stimulate the axons of the inner-layer ganglion cells that form the optic directly modulates the microphotodiodes, and the electrodes inject tiny nerve [adapted from A. Stett].

1022 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN tical “needle” electrode arrays (9). Taking a between the pigmented epithelium and outer contacting the inner or outer retina, reveal different approach, a Belgian group has at- layer of the retina. Light falling on the retina that injecting a charge of about 0.4 nanoCou- tempted to attach a cuff comprising a few generates currents in the photodiodes that then lombs (nC) per electrode (typically equiva- electrodes around the optic nerve of a blind activate the microelectrodes, resulting in stim- lent to a current of ϳ 10 ␮A) is sufficient to patient. The patient was able to localize sin- ulation of retinal sensory neurons. In contrast, excite retinal neurons (12, 22). With a dis- gle bright spots of light, but high spatial the epiretinal device is implanted onto the in- tance of 50 to 150 ␮m between electrodes resolution cannot be expected with such a nermost layer of the retina that contains the implanted in rat outer retina, ganglion cells stimulation arrangement (10). ganglion cells. The epiretinal implant is essen- can receive electrical information in a spatial- In the early 1990s, many researchers tially a readout chip that receives electrical sig- ly organized manner (22) even in animals that switched their efforts to developing a prosthesis nals containing image information from a dis- have lost their photoreceptor cells through that could be implanted directly into the retina tant camera and processing unit, and is coupled retinal degeneration (24). Coating the surfac- (see the News story by J. Cohen on page 1026). to the ganglion cells and their axons. In re- es of microphotodiodes with glycoproteins A retinal implant requires the creation of com- sponse to stimulation by the external image such as laminins may improve their biocom- plicated subminiature electrode arrays and elec- receiver system, the epiretinal implant gener- patibility (25). trical circuits consisting of materials that must ates electrical impulses that travel via the gan- Work in vivo demonstrates that in principle be stable and relatively inert so that their impact glion cell axons of the optic nerve to the brain. there are two surgical approaches for safe intro- on remaining retinal tissue is minimal. Next, duction of the subretinal implant: gaining access electrical parameters for stimulation of retinal Subretinal Implants to the retina through the vitreous humor of the nerve cells must be determined and electronic Chow and colleagues (11–17) in Chicago and eye (ab interno) (16, 26), and gaining direct circuits developed to accommodate the large our group in Tu¨bingen (18–29) are develop- access to the subretinal space through a scleral brightness and contrast variances of the envi- ing subretinal implants. A thin plate (ϳ50 to incision (ab externo) (27). The retina of cats and ronment. Before retinal implants can be tested 100 ␮m thick and 2 to 3 mm in diameter) pigs remains intact for more than 2 years after in patients, surgical techniques for implanting, carries hundreds to thousands of light-sensi- implantation of a subretinal device (14, 17, 28). removing, and fixing these electronic prosthe- tive microphotodiodes equipped with micro- In rats, the implant remains fixed in a stable ses in the eye must be developed. In addition, electrodes of gold or titanium nitride ar- subretinal position and continues to work for up suitable animal models for testing retinal pros- ranged in arrays (11, 16, 18, 19, 21). Light to 16 months. Implanted subretinal devices theses must be found, ethical questions ad- emanating from visible objects is converted show some damage over time due to accumu- dressed, and regulatory matters considered. by the microphotodiodes into tiny currents at lation of silicon oxide on their surfaces (20) and There is also the question of which patients are each of hundreds of microelectrodes. These disintegration of the gold electrodes. Such prob- best suited to receive retinal implants. Retinal currents are then “injected” into whichever lems can be addressed by encapsulating the implants under development in the United neurons remain of the retinal network, the device in an inert polymer. Action potentials in States, Germany, and Japan could potentially middle and inner retina thus taking over the the visual cortex of the brain evoked by subreti- enter early clinical trials for functional testing information-processing part of vision. Sub- nal electrodes have been recorded in the pig within the next few years. retinal prostheses have a number of advan- (Fig. 2) (29). Following acute electrical stimu- tages—the microphotodiodes directly replace lation of the cat retina with subretinal electrodes, Two Kinds of Retinal Implant damaged photoreceptor cells; the retina’s re- a spatial resolution of at least 1° was obtained as There are two kinds of retinal implant under maining intact neural network is still capable measured by multi-electrode (30) and optical development: subretinal and epiretinal (Fig. 1). of processing electrical signals; positioning recording (31) from the visual cortex. The subretinal device is implanted between the and fixing of the microphotodiodes in the In vivo experiments also reveal weaknesses pigment epithelial layer and the outer layer of subretinal space is relatively easy; no external in subretinal implant prototypes. For example, the retina, which contains the photoreceptor camera or external image processing is re- the current generated by a single microphoto- cells. In the subretinal device, thousands of quired; and eye movements can still be used diode with its small light-sensitive area is not light-sensitive microphotodiodes equipped with to locate objects. sufficient to stimulate adjacent neurons with the microelectrodes are assembled on a very thin Experiments in several animal models, ambient light available from the environment. plate and are placed in the subretinal space where recordings are made with electrodes Thus, an active subretinal implant supported by an external energy source—such as transpupil- Fig. 2. (Top) Cortically evoked po- Light-evoked potentials in pig lary infrared illumination of receivers close to tentials recorded with epidural the chip or electromagnetic transfer—is now surface electrodes from the visual under development. When illuminated, sub- cortex of the pig. Stimulation with a short light flash (8 ms) evokes a retinal microphotodiodes are able to transfer positive electrical deflection with these external currents via their microelec- a peak potential about 90 to 100 trodes to the retinal neurons in a retinotopi- ms later. The amplitude of the cally accurate manner. Only with the aid of an peak is small in low light and large external energy source will the light from the in bright light (measured in cd/ normal environment be sufficient to modulate m2). (Bottom) With electrical Light stimulation the stimulating current at each individual stimulation of the outer retina by Electrically evoked potentials in pig a wire-bound subretinal multielec- electrode. trode array (in lieu of light stimulation), an electrical cortical re- Epiretinal Implants sponse similar to that evoked by The epiretinal implant has no light-sensitive moderate light stimulation is elic- elements (Fig. 1). A very tiny field sensor, like ited. The charge released with a 3-V electrical pulse of 400-ms du- a camera, is positioned either outside the eye or ration is about 20 nC and 200 within an intraocular plastic lens that replaces ␮C/cm2, respectively, per elec- the natural lens of the eye and is introduced trode [modified from (29)]. Electrical stimulation using techniques developed for cataract sur-

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1023 B ODYBUILDING:THE B IONIC H UMAN gery. Foil-bound wires connect the field sensor and biocompatibility of the epiretinal implant What Spatial Resolution Can Be at the anterior of the eye with an electrode array has been demonstrated in the cat, where the Achieved? implanted on top of the inner retina. This array implant provided a cortical resolution of about a The best spatial resolution provided by an elec- is attached to the inner retinal membrane that 1° visual angle (30). trical excitation pattern at the retinal level can be separates the neural layer from the vitreous The subretinal and epiretinal approaches calculated from in vitro and in vivo experiments body of the eye’s posterior chamber. Epiretinal both have their advantages and disadvan- (Fig. 3) (20, 23). If a portrait (Fig. 3, middle implants are being investigated by Eckmiller tages. Whereas the subretinal implant uses row) or an optotype image (Fig. 3, bottom row) and his colleagues (32–34), Humayun et al. the remaining neural network of the retina, is pixelated in a 40 by 40 grid, the pattern shown (35–39), Rizzo and co-workers (40–45), and the epiretinal implant does not and thus in column 2 of Fig. 3 corresponds to the number more recently by Abrams’s group from Detroit must provide additional processing to pre- and distance of electrodes on a microphotodiode and a Japanese consortium lead by Tano et al. pare the visual information. On the other array (3 mm by 3 mm) of a subretinal implant, from Osaka. hand, the information-transfer characteris- where electrodes 70 ␮m apart evoked discern- Unlike the subretinal implant, the epiretinal tics of the epiretinal implant are more ame- ible patterns in vivo; the stimulation current is implant does not use the remaining network of nable to external control. assumed to be near threshold so that many little the retina for information processing. Thus, the Fixing the subretinal implant in the subreti- excitation dots with grosser distributions are epiretinal sensor has to encode visual informa- nal space is relatively easy because the pigment achieved. If the current is increased (for exam- tion as trains of electrical impulses that are then epithelial cells of the retina “pump out” this ple, by increasing the illumination), each point conveyed by the electrode array directly into the space such that the implant is sequestered like “a becomes larger and the current waves (top row) axons of ganglion cells, which unite to form the peanut in a vacuum package.” In contrast, fixing as well as the points merge, resulting in the optic nerve. The visual information has to be the epiretinal implant is very difficult and carries images shown in column 3. Very strong stimu- translated into a spatiotemporal stimulation pat- the additional risk of stimulating cellular prolif- lation produces patterns with even more conflu- tern of electrical impulses that can be under- eration. The two-dimensional signal pattern of ent points, resulting in the images in column 4. stood by the brain’s visual cortex. This spatio- retinal photoreceptor cells is a mirror image of This is, of course, only valid if one assumes that temporal stimulation pattern is first conveyed to the outer world so that correct retinotopic stim- each electrode in the array can be connected the electrode array positioned on the inner reti- ulation by the subretinal implant, which replaces appropriately with a retinal neuron, a question nal membrane, which is stabilized either by the photoreceptor cells, can be achieved. The that remains to be addressed. Certainly the very slight mechanical pressure, or by attachment epiretinal implant stimulates both the axons of high spatial resolution of natural photoreceptor through cellular contacts or microtacks. As in ganglion cells, which may be far away, and their cells cannot be achieved because this resolution the subretinal implant, specific surgical tech- cell bodies, which are nearby, resulting in a is based on highly specialized pre- and post- niques have been developed for implanting the more disordered stimulation pattern that has to synaptic structures that ensure high gain and epiretinal sensor (34, 41), and the parameters of be corrected electronically. However, the sub- high-fidelity transmission to second-order corti- the current needed for optimal stimulation have retinal implant needs intact optics, whereas the cal neurons. been defined (35, 41, 45). Long-term stability epiretinal implant does not. Will Retinal Implants Work in Patients? Humayun et al. (35, 38), as well as Rizzo and Wyatt (42, 45), have stimulated the retina of blind patients with epiretinal electrodes that were transiently inserted into the eye through a scleral opening. Both groups reported a sensation of light patterns by the patients, but perception of geometric patterns was reported in only a few instances. Limited perception was achieved only at high charge densities at the site of the epiretinal implant (45). During the available intraoperative stimulation time (up to 4.5 hours), the perceptual responses elicited by epiretinal electrodes did not often meet expectations in relation to the pattern of stimulation (45). These results are still far from true object recognition. However, they do demonstrate the feasibility of generating perception of light patterns in blind people. It is also reassuring to learn (28, 36, 39) that ganglion cells and other cells of the inner retina are still present in patients with retinitis pigmentosa, even after many years of blindness. Consequently, even in these patients, Fig. 3. (Top) Example of a proﬁle of electrical excitation produced by light-sensitive areas of a there may be enough neural cells available subretinal implant, activated by electrodes spaced 70 ␮m apart. (Bottom) If a regular three- for efficient stimulation by subretinal or dimensional image (the face or optotype) is transformed into a two-dimensional excitation proﬁle, epiretinal implants. the image would be represented by a two-dimensional array of 40 by 40 small excitation spots Chow and colleagues (46) have reported (pixels), sized according to the width of the local electrical wave (see second column). When the light intensity is increased, each of the excitation spots enlarges and a more homogeneous picture implantation of a passive subretinal device emerges (see third column). A further increase in luminosity causes a merging of excitation spots, (without an external energy supply) into the resulting in a blurred picture (see fourth column) (20, 23). eyes of patients using the method of Peyman et

1024 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN al. (16). Their study was designed to examine (ARVO). Encouragingly, 1999 saw the number 20. E. Zrenner et al., Ophthalmologe 98, 357 (2001). biocompatibility of the device but not its func- of presentations rise to 33, and the numbers 21. F. E. Gekeler et al., Invest. Ophthalmol. Vis. Sci. 42, tion. The results in six patients (three received continue to increase. Successful testing of each S815 (2001). 22. A. Stett et al., Vision Res. 40, 1785 (2000). the chip in 2000, and another three in 2001) major step in retinal implant development— 23. A. Stett, T. Hermann, Invest. Ophthalmol. Vis. Sci. 41, have not been released as yet. production of implant prototypes, surgical pro- S859 (2000). cedures, long-term stability and biocompatibil- 24. A. Stett et al., Invest. Ophthalmol. Vis. Sci. 40, S734 The Future ity of implant material, electrical testing in vitro (1999). 25. E. Guenther, B. Troeger, B. Schlosshauer, E. Zrenner, It is indeed feasible to elicit action potentials in as well as in animal models, recording electrical Vis. Res. 39, 3988 (1999). the visual cortex using electrical impulses gen- activity in animal brains—is providing essential 26. H. G. Sachs et al., Invest. Ophthalmol Vis. Sci. 40, erated by subretinal or epiretinal devices, but a data about the resolution required for blind S734 (1999). number of obstacles remain to be overcome. We patients to regain mobility in a world that very 27. K. Kobuch et al., Invest. Ophthalmol. Vis. Sci. 39, S903 (1998). need to know whether the encoding of orienta- much depends on visual information. A number 28. K. Kohler, J. A. Hartmann, D. Werts, E. Zrenner, Oph- tion and movement perception as well as feature of international groups (44) are tackling the thalmologe 98, 364 (2001). localization is maintained at the level of the remaining problems associated with epiretinal 29. H. Schwahn et al., Graefe’s Arch. Clin. Exp. Ophtha- visual cortex. How can the long-term stability of and subretinal implants, and we await the out- lmol. 239, 961 (2001). 30. R. Eckhorn et al., Ophthalmologe 98, 4, 369 (2001). these implants, whose surfaces do deteriorate come of clinical trials to determine the value of 31. U. Eysel, in preparation. after long-term implantation, be achieved? Will refined nanotechnology for treating blinding 32. R. Eckmiller, Ophthalmic Res. 29, 281 (1997). retinal neurons tolerate long-term electrical eye diseases. 33. ࿜࿜࿜࿜ ,R.Hu¨nermann, M. Becker, Invest. Ophthal- stimulation without themselves being altered mol. Vis. Sci. 39, S990 (1998). morphologically or functionally? What type of References and Notes 34. P. Walter et al., Retina 19, 546 (1999). 35. M. S. Humayun et al., Invest. Ophthalmol. Vis. Sci. 40, image can be perceived by blind patients 1. H.G. Krumpazsky, V. Klauss, Ophthalmologica 210,1 (1996). 143 (1999). through an epiretinal implant or the light-sensi- 2. J. P. Rauschecker, R. V. Shannon, Science 295, 1025 36. M. S. Humayun et al., Arch. Ophthalmol. 114,40 tive microphotodiodes of a subretinal implant? (2002). (1996). The patient group best suited to test such 3. G. E. Tassiker, U.S. patent 2,760,483 (1956). 37. M. S. Humayun, E. J. de Juan, J. D. Weiland, R. 4. G. S. Brindley, W. S. Lewin, J. Physiol. 196, 479 Greenberg, IEEE Int. Solid-State Circuits TP 12.7 questions by implantation of retinal prosthe- (1968). (1999). ses may be patients with hereditary retinal 5. W. H. Dobelle, W. G. Mladejowski, J. Physiol. 243, 38. M. S. Humayun et al., Vision Res. 39, 2569 (1999). degenerative diseases who are stricken with 553 (1974). 39. A. Santos et al., Arch. Ophthalmol. 115, 511 (1997). 6. W. H. Dobelle, ASAIO J. 46, 3 (2000). 40. J. F. Rizzo, J. L. Wyatt, Neuroscientist 3, 251 (1997). blindness yet still have intact middle and 7. E. M. Schmidt et al., Brain 119, 507 (1996). 41. J. F. Rizzo, J. Loewenstein, J. L. Wyatt, Retinal Degen- inner retinal layers. In addition, a second 8. R. A. Normann, E. M. Maynard, K. S. Guillory, D. J. erative Diseases and Experimental Therapy, J. G. Hol- group of patients with certain forms of mac- Warren, IEEE Spectrum 33, 54 (1996). lyﬁeld et al., Eds. (Kluwer, Dordrecht, Netherlands, ular degeneration, including the age-related 9. R. A. Normann et al., Vision Res. 41, 1261 (2001). 1999), pp. 463–469. 10. C. Veraart et al., Brain Res. 813, 181 (1998). 42. J. F. Rizzo et al., Invest. Ophthalmol. Vis. Sci. 40, S783 type, could benefit from retinal prostheses 11. A. Y. Chow, Invest Ophthalmol. Vis. Sci. 34 (suppl.), (1999). even at advanced stages of disease. We need 835࿜࿜࿜࿜ (1993). 43. A. E. Grumet, J. L. Wyatt Jr, J. F. Rizzo III, J. Neurosci. to instigate long-term clinical trials to ensure 12. , V. Y. Chow, Neurosci. Lett. 225, 13 (1997). Methods 101, 31 (2000). 13. A. Y. Chow et al., Invest. Ophthalmol. Vis. Sci. 39, 44. J. F. Rizzo et al., Ophthalmology 108, 13 (2001). that the peripheral vision still present in pa- S565 (1998). 45. J. F. Rizzo, J. Wyatt, personal communication. tients with macular degeneration is not en- 14. A. Y. Chow et al., IEEE Trans. Neural Syst. Rehabil. 46. A. Y. Chow, personal communication. dangered by surgical procedures and the ef- Eng. 9, 86 (2001). 47. We thank all of our colleagues including A. Stett, H. 15. N. S. Peachey, A. Y. Chow, J. Rehabil. Res. Dev. 36, Haemmerle, H. Schwahn, and F. Gekeler for lively fects of the retinal implants themselves. 371 (1999). discussions. The subretinal group of the consortium is In 1993, there were only two papers on 16. G. Peyman et al., Ophthalmic Surg Lasers 29, 234 supported by the German Federal Ministry of Educa- retinal implants among the thousands presented (1998). tion and Research (BMBF) grant 01 IN 50 2A-D and 17. M. T. Pardue et al., Exp. Eye Res. 73, 333 (2001). 01KP0008-0012, the Pro-Retina e.V. Germany, the at the annual meeting of the Association for 18. E. Zrenner et al., Ophthalmic Res. 29, 269 (1997). “Fortu¨ne”–Program of Tu¨bingen Medical Faculty, and Research in Vision and Ophthalmology 19. E. Zrenner et al., Vision Res. 39, 2555 (1999). the E. and K. Hochbaum Foundation.

VIEWPOINT Sending Sound to the Brain J. P. Rauschecker1 and R. V. Shannon2

The cochlear implant, a microelectrode array that directly stimulates the (CIs) (1, 2). The CI is a microelectrode array auditory nerve, has greatly beneﬁted many individuals with profound implanted in the cochlea that directly stimu- deafness. Deaf patients without an intact auditory nerve may be helped by lates the auditory nerve. With more than the next generation of auditory prostheses: surface or penetrating audi- 40,000 patients worldwide, the success of tory brainstem implants that bypass the auditory nerve and directly these devices is nothing short of miraculous: stimulate auditory processing centers in the brainstem. Most adults are able to converse on the phone, and most children are able to be edu- Partial or total hearing loss has many dif- by loss of the sensory hair cells in the fluid- ferent causes. Defects in either the outer ear filled, snail-shaped inner ear, or cochlea, that CONTINUED ON PAGE 1027 or middle ear (composed of the tympanic transduce sound waves into electrical impuls- membrane, ear drum, and auditory ossicles) es, which are then transmitted to the brain result in a conductive hearing loss that can (Fig. 1). Profoundly deaf individuals who still 1Department of Physiology and Biophysics, George- usually be remedied by insertion of a hearing have an intact auditory nerve have profited town University Medical Center, Washington, DC 20007, USA. 2Department of Auditory Implants and aid, which amplifies sound vibrations. Pro- from the dramatic advances made over the Perception, House Ear Institute, Los Angeles, CA found deafness, on the other hand, is caused past 30 years in the field of cochlear implants 90057, USA.

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1025 B ODYBUILDING:THE B IONIC H UMAN

CONTINUED FROM PAGE 1025 ly, this frequency range turns out to be largely patients who become deaf before or after unnecessary for speech transmission, perhaps acquiring speech (pre- and postlingual deaf- cated in mainstream classrooms. For some because of a peculiar psychophysical phe- ness, respectively). Whereas the postlingually profoundly deaf individuals, however, even nomenon called “residue pitch.” This effect deaf “reconnect” almost immediately after electrical stimulation of the inner ear with a was already known to Helmholtz in the 19th receiving their CI, success in prelingually CI is impossible owing to an absence or century as the phenomenon of the “missing deaf individuals depends very much on the destruction of the auditory nerve. Instead, an fundamental” (6). As long as sounds contain age of implantation. As with corrections of auditory prosthesis consisting of a microelec- enough overtones or harmonics, higher audi- early visual defects, such as cataract or stra- trode array that directly stimulates one of the tory processing centers in the brain, such as bismus, the maxim is “the earlier the better” auditory processing centers of the brainstem, the auditory cortex, are able to “reconstruct” (2, 8). Thus, another fundamental and en- bypassing the cochlea and auditory nerve, the missing fundamental frequency. The ad- couraging lesson learned from CI research is might restore hearing to these patients. Such vantage for CIs (and modern telephones for about the plasticity of neural representations auditory brainstem implants (ABIs) have that matter) is that frequencies below ϳ1000 of auditory information in the brains of young been under development since the late 1970s, Hz need not be transmitted for a clear under- children—that is, in response to sound, the pioneered by physicians and researchers at standing of speech, which helps to save valu- stimulated auditory cortex can recruit neu- the House Ear Institute in Los Angeles (3), able bandwidth. Thus, despite the relatively rons from adjacent regions of the brain and but have had only limited success. The next crude CI signal (in comparison to the original can form new neuronal connections. Studies step in ABI evolution is already under way: speech signal), delivered by a discrete and in congenitally deaf cats confirm the mallea- Whereas conventional ABIs stimulate the limited number of stimulating electrodes, bility of the auditory cortex, which is moul- surface of the ventral cochlear nucleus in the most implant listeners are capable of excel- ded by auditory experience from an early age brainstem, the microelectrode array in the lent language understanding. Although much (9). If environmental sounds transmitted via a new generation of ABIs penetrates into the of the information coming from the cochlea is microphone and CI are used to stimulate the depths of the ventral cochlear nucleus, direct- important for sound quality, music apprecia- central auditory pathways of young deaf cats, ly stimulating its neurons (nerve cells) (Fig. tion, or higher-order cues for the identifica- the animals soon begin to respond with ap- 2). This new approach has become feasible tion of sound objects, only a coarse represen- propriate behaviors to these sounds, and their owing to the high-tech development of mate- tation of the auditory information from the auditory cortex begins to develop normal ac- rials and electronics by researchers in the cochlea is required for speech understanding tivation patterns. Much less plasticity is ob- field of neural prosthetics (including CIs), as (7). Processing by the auditory cortex fills in served in congenitally deaf animals that are well as successful stereotactic studies carried much of the missing information, just as the exposed to sound at an older age (9). These out with permanent depth electrodes in cats, visual cortex fills in the blanks left by our results are very much in tune with the visual- an animal with an auditory system similar to blind spot or by illusory contours. deprivation literature (10) and indicate the our own (4). Obviously, a difference exists between existence of a sensitive period during early Early CIs were designed primarily to replace the lost function of the cochlea, with Fig. 1. The cochlear im- little regard to the way the brain processes plant (CI). The auditory and adapts to auditory information. Thirty system is composed of the years of research with CI patients have re- outer ear, the middle ear vealed new insights into the contributions of comprising the tympanic membrane and auditory both the cochlea and the brain in auditory ossicles, and the inner ear perception. These insights will prove to be composed of the snail-like invaluable for the next generation of brain- cochlea containing sensory stem auditory prostheses (and perhaps other hair cells bathed in fluid. neural prostheses as well). In CI devices, These sensory hair cells ac- multiple stimulating electrodes embedded in tivate the fibers of the cochlear (auditory) nerve, a silicone matrix are inserted into the cochlea which emanate from the of the inner ear, so that the electrodes are spiral ganglion cells and situated at different locations along the co- project to the cochlear nu- chlea’s basilar membrane. The mechanics of cleus of the auditory brain- the cochlea are such that the basilar mem- stem. The neural pathway brane is tonotopically organized, that is, high- then leads to the higher auditory processing cen- frequency sounds activate hair cells at the ters of the brain: the infe- base of the basilar membrane, whereas low- rior colliculus, medial frequency sounds activate hair cells near the geniculate nucleus, and pri- apex of the basilar membrane. The sound, mary auditory cortex. A CI received by an external microphone, is first consists of an external mi- analyzed by a microprocessor, and an electri- crophone, which collects sound waves, and a speech cal signal representing the information from processor, which converts each frequency region is transmitted to the the sound waves into elec- appropriate electrode at each tonotopic loca- trical impulses that are tion (Fig. 1). Thus, CIs can access most of the then transmitted to a re- frequency range of the cochlea, with the ex- ceiver implanted under the ception of the lowest frequencies represented skin. The receiver sends the electrical impulses to a microelectrode array implanted within the cochlea. The electrodes directly by hair cells in the thinnest, narrowest wind- stimulate the correct populations of auditory nerve fibers so that electrical signals are propagated to the ing part of the cochlear apex, which are in- appropriate (tonotopic) regions of the cochlear nucleus of the brainstem and then on to higher auditory accessible to most CI electrodes (5). Curious- processing centers. [Illustration: Nathalie Cary/Katharine Sutliff]

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1027 B ODYBUILDING:THE B IONIC H UMAN postnatal development of the central auditory that they will wake up without hearing. stimulate distinct neural populations. system, especially the auditory cortex (11, Their only hope is that an auditory prosthe- Improvements in ABI performance re- 12). Other examples supporting the existence sis such as an ABI that stimulates the brain- quire that the microelectrode array penetrates of a sensitive period include early musical stem directly may provide at least some deep into the ventral cochlear nucleus. Mc- training in young children, which seems to limited hearing capability. Creery et al. (4) have successfully implanted lead to a higher incidence of absolute pitch Existing ABI technology (implanted in such penetrating electrode arrays in cats and and an enhanced sensitivity of the auditory ϳ200 patients so far) stimulates the surface have demonstrated the ability of the electrode cortex to complex harmonic sounds (13). The of the ventral cochlear nucleus in the auditory arrays to evoke tonotopically localized neural application of CIs is also creating renewed brainstem, the next stage of auditory process- activation in the next auditory relay station of interest in and a new tool for the study of ing after the cochlea (18). The strategy is the brainstem, the inferior colliculus. The cortical plasticity in the adult auditory sys- similar to that used in cochlear implants: An hope is that the improved tonotopic selectiv- tem. Some studies have shown remarkable external speech processor and receiver trans- ity provided by penetrating microelectrodes perceptual plasticity in adapting to auditory mit sound waves in the form of electrical will result in improved speech understanding patterns shifted in frequency (as they might impulses to an array of platinum electrodes in human patients. After 12 years of basic be by a CI) (14), a result that is reminiscent that are implanted on the surface of the brain- research, the Food and Drug Administration of the perceptual learning in vision encoun- stem’s ventral cochlear nucleus (Fig. 2). The in the United States has approved the first tered when using inverting and displacing microelectrode array is inserted through an clinical trial of the penetrating ABI in NF2 goggles (15). Other studies have not shown opening in the mastoid bone behind the ear patients, and the first candidate is expected to such plasticity and again suggest that audito- and is advanced into the lateral recess of the undergo surgery in 2002. ry speech-pattern recognition may not be as fourth ventricle of the brain, which is adja- Another reason for the limited success of malleable in adults (16). To effectively de- cent to the ventral cochlear nucleus. The ven- the current ABIs may be that direct stimula- sign and program auditory prostheses, it is tral cochlear nucleus is part of the main- tion of the cochlear nucleus bypasses impor- important to understand the critical elements stream (lateral lemniscal) auditory system tant intrinsic neuronal processing—that is, of auditory pattern recognition and the extent that transmits sound frequency information to the cochlear nucleus may already contain to which central plasticity can correct or com- higher auditory centers (inferior colliculus, specialized neural circuitry that extracts in- pensate for distortions and omissions of in- medial geniculate nucleus, and auditory cor- formation on sound modulation and period- coming sound information. tex) and is tonotopically organized (19–21). icity, and information on sound onset and CIs are not an option for those individuals Unfortunately, its tonotopic map is oriented offset. If penetrating electrodes do not result whose deafness is caused by lesions beyond the at a shallow angle to the surface. The effect of in a significant improvement in speech rec- cochlea. This includes a patient group with electrodes simply placed along the surface of ognition, it could imply that such specialized neurofibromatosis type 2 (NF2), a genetic dis- the ventral cochlear nucleus is therefore tan- processing in the cochlear nucleus is critical ease that occurs in about 1 in 40,000 births tamount to stimulation with a single elec- for speech understanding and cannot be re- (17). NF2 produces Schwann cell tumors along trode. The multiple microelectrode array of- created by prosthetic electrical stimulation. afferent nerve tracts as they enter the brainstem ten does not evoke different pitch sensations Such a failure would not bode well for a and spinal cord. One of the defining symptoms in ABI patients, whereas such pitch differ- future auditory prosthesis that directly stim- of NF2 is the growth of bilateral tumors along ences are routine in CI patients. It is therefore ulates the inferior colliculus or auditory cor- the eighth cranial nerve (composed of the audi- not surprising that the range of frequency tex. Both structures, particularly the inferior tory and vestibular nerves). Removal of the information and the frequency resolution of colliculus, would be more surgically accessi- tumors almost always necessitates a transection these surface-electrode ABIs are not satisfac- ble than the cochlear nucleus, which is locat- of the auditory nerve and thus results in total tory and usually do not lead to an understand- ed deep in the brainstem. Regarding direct deafness. If the tumors are not removed, they ing of speech even after months and years of stimulation of the auditory cortex, attempts to produce compression of the brainstem that is practice (22–24). Another factor contributing directly stimulate the visual cortex have been ultimately fatal. Faced with the tragic choice to the lack of tonotopic resolution in surface- disappointing, with patients seeing only between total deafness and premature death, electrode ABIs may be the high current need- phosphenes (spots of light) or other discon- most patients, usually young people in the ed for stimulation: Current fields spread nected and meaningless sensations (25). prime of life, opt for surgery, knowing full well broadly from each electrode and may not However, with the gains made by research on both the visual and auditory cortex, it may be Fig. 2. Auditory brainstem im- Low acoustic Surface possible to design new electrode arrays that array plant (ABI). Both panels show a frequencies DCN DCN stimulate the auditory cortex appropriately section through the brainstem Medium acoustic (26, 27). frequencies that reveals the dorsal and ven- Neuroimaging techniques, such as function- tral regions of the cochlear nu- High acoustic cleus (DCN, VCN). Depicted are frequencies al magnetic resonance imaging, will be valu- two types of ABI in which the able in the presurgical planning of stereotacti- microelectrode array is either cally placed ABIs that must penetrate deep into placed on the surface of the co- small brainstem structures (28, 29), particularly chlear nucleus (right panel), or VCN VCN if neuroimaging can illuminate tonotopic orga- penetrates the VCN (left panel). nization in individual patients (30). Stereotactic The tonotopic organization of the VCN is such that iso-fre- neurosurgical interventions could eventually quency regions run at a shallow help to widen the use of auditory prostheses for angle almost parallel to the VCN binaurally deaf patients with other etiologies, surface. This explains why pene- Penetrating such as an absent or malformed inner ear or a array trating ABIs provide more selec- basal skull fracture. Long-term stereotactic tive tonotopic stimulation than Cable electrode implants are technically feasible and surface ABIs. [Illustration: Nathalie Cary] Stump of can be highly beneficial for patients with chron- cochlear nerve Cable ic debilitating diseases. For example, deep brain

1028 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN stimulation of the subthalamic nucleus of Par- middle turn. If one assumes that CI electrodes 18. D. E. Brackmann et al., Otolaryngol. Head Neck Surg. kinson’s patients with a stereotactically placed stimulate fibers en passant, they may stimulate 108, 624 (1993). electrode results in alleviation of many of the low frequencies too, even though the electrodes do 19. J. K. Moore, K. K. Osen, Am. J. Anat. 154, 393 (1979). not extend all the way into the apical turn. 20. In contrast to the VCN, the dorsal cochlear nucleus motor symptoms of this disease (31). 6. H. L. F. Helmholtz, Die Lehre von den Tonempfindun- (DCN) contains neurons with complex response Clearly, CIs and ABIs not only are of gen als physiologische Grundlage fu¨r die Theorie der properties already at this early phase of processing benefit to profoundly deaf individuals but Musik ( Vieweg, Braunschweig, Germany, 1863). and may be involved in early stages of spatial hearing 7. R. V. Shannon, F.-G. Zeng, J. Wygonski, V. Kamath, M. (21). also continue to provide insight into informa- Ekelid, Science 270, 303 (1995). 21. E. D. Young et al., Philos. Trans. R. Soc. London Ser. B tion processing in the auditory nervous sys- 8. M. A. Svirsky, A. M. Robbins, K. I. Kirk, D. B. Pisoni, R. T. 336, 407 (1992). tem. And certainly, each new understanding Miyamoto, Psychol. Sci. 11, 153 (2000). 22. However, speech perception in combination with lip- 9. R. Klinke, A. Kral, S. Heid, J. Tillein, R. Hartmann, reading is markedly improved (23, 24). achieved in basic scientific research will re- Science 285, 1729 (1999). 23. D. Matthies et al., J. Laryngol. Otol. Suppl. 27,32 sult in improvements to the technology of 10. T. N. Wiesel, Nature 299, 583 (1982). (2000). auditory prostheses, and increased benefits 11. J. P. Rauschecker, Trends Neurosci. 18, 36 (1999). 24. S. R. Otto et al., J. Neurosurg., in press. for patients. 12. J. K. Moore, Guan, J. Assoc. Res. Otolaryngol. 2, 297 25. G. S. Brindley, W. S. Lewin, J Physiol. 196, 479 (1968). (2001). 26. R. A. Normann, E. M. Maynard, P. J. Rousche, D. J. 13. C. Pantev et al., Nature 392, 811 (1998). Warren, Vision Res. 39, 2577 (1999). References and Notes 14. S. Rosen, A. Faulkner, L. Wilkinson, J. Acoust. Soc. Am. 27. Q. Bui, K. D. Wise, D. J. Anderson, IEEE Trans. Biomed. 1. G. E. Loeb, Annu. Rev. Neurosci. 13, 357 (1990). 106, 3629 (1999). Eng. 47, 281 (2000). 2. J. P. Rauschecker, Science 285, 1686 (1999). 15. I. Kohler, The Formation and Transformation of the 28. A. R. Guimaraes et al., Hum. Brain Map. 6, 33 (1998). 3. B. J. Edgerton, W. F. House, W. Hitselberger, Ann. Perceptual World [U¨ber Aufbau und Wandlungen der 29. T. D. Griffiths, S. Uppenkamp, I. Johnsrude, O. Jo- Otol. Rhinol. Otolaryngol. 91 (suppl.), 117 (1984). Wahrnehmungswelt (International Univ. Press, Vi- sephs, R. D. Patterson, Nature Neurosci. 4, 633 4. D. B. McCreery, R. V. Shannon, J. K. Moore, M. Chat- enna, 1951)], R. M. Rohrer, transl. (International Univ. (2001). terjee, IEEE Trans. Rehabil. Eng. 6, 391 (1998). Press, Vienna, 1964). 30. C. M. Wessinger et al., J. Cognit. Neurosci. 13,1 5. The cell somata of nerve fibers coming from the 16. Q.-J. Fu, R. V. Shannon, J. Galvin, in preparation. (2001). apical turn of the human cochlea are not located in 17. D. G. R. Evans et al., J. Med. Genet. 29, 841 31. M. R. DeLong, T. Wichmann, Ann. Neurol. 49, 142 the center of the apical coil, but more toward the (1992). (2001).

VIEWPOINT Repairing the Injured Spinal Cord Martin E. Schwab

Certain cell, molecular, and bioengineering strategies for repairing the cord in paraplegic or quadriplegic patients injured spinal cord are showing encouraging results (either alone or in show that complete anatomical separation of combination) in animal models, with limited recovery of mobility being the spinal cord is very rare. Instead, bridges reported in some cases. of nerve tissue (tracts) connecting regions above and below the lesion often persist, Our spinal cord is a finger-thick strand of the injury was sustained in the thoracic/lum- mostly at the peripheral edges of the spinal nervous tissue that is tightly enclosed in the bar region or neck region of the spinal col- cord (4). This type of anatomically incom- bony vertebrae of the spinal column. The umn, respectively. plete spinal cord lesion has been recreated in spinal cord receives sensory information Destruction of the spinal cord can be com- rats and other animals, either by microsurgi- from the skin, the muscles, the joints, and pared to a bomb exploding in a computer cen- cal transection of defined regions of the spi- other tissues of the body. It transmits this ter, and repairing the spinal cord is as compli- nal cord or by crushing the cord with metal information in the form of electrical im- cated as trying to rebuild all of the computer rods of different weights. These animal mod- pulses to the brain, along millions of nerve connections. In the last few years, there has els are valuable tools with which to test var- fibers that are grouped together in bundles. been encouraging progress in animal models, ious spinal cord repair strategies. The motor commands that are subsequently with sufficient regeneration of the damaged There are four principal strategies for re- generated in the brain are sent to the spinal spinal cord to enable some recovery of motor pairing spinal cord lesions: (i) promoting the cord along fast-conducting nerve fibers, ability. When the spinal cord is injured, the first regrowth of interrupted nerve fiber tracts, which terminate in local spinal motor cir- phase of injury involves mechanical tissue de- using nerve growth stimulatory factors or cuits. From here, the electrical impulses that struction. It is followed by a second phase of molecules that suppress inhibitors of neuro- will direct coordinated muscle contraction tissue loss, which is principally caused by a nal extensions (neurites); (ii) bridging spinal reach the muscles via the peripheral nerves. A severe local disturbance of the blood supply (1, cord lesions with scaffolds that are impreg- sharp blow to the spinal column can cause 2). There have been attempts to minimize this nated with nerve growth factors, which pro- dislocation of individual vertebrae and severe secondary damage with neuroprotective agents, mote axon growth and reduce the barrier damage to the spinal cord, including its com- but, so far, only high-dose methylprednisolone caused by scar tissue; (iii) repairing damaged plete severance. Clinically, the result of an (a synthetic corticosteroid) given within the first myelin (the insulating sheath that surrounds incomplete or complete spinal cord lesion is hours after injury is in use clinically (2, 3). axons) and restoring nerve fiber impulse con- either paraplegia (paralysis of the lower Within several weeks of the injury, macro- ductivity in the lesion area; and (iv) enhanc- body) or quadriplegia (paralysis of the body phages migrating from the bloodstream have ing central nervous system (CNS) plasticity from the neck down), depending on whether cleared the tissue debris at the lesion site, re- by promoting compensatory growth of sulting in fluid-filled cysts surrounded by scar spared, intact nerve fibers above and below tissue (Fig. 1). Whether this inflammatory re- the lesion. action leads to additional damage of spinal cord Brain Research Institute, University of Zu¨rich, and tissue that is still intact remains a matter of Regeneration of Nerve Fiber Tracts Department of Biology, Swiss Federal Institute of Technology Zu¨rich, Winterthurerstrasse 190, 8057 debate. Crushed or transected nerve fibers in the CNS Zu¨rich, Switzerland. Remarkably, images of the lesioned spinal of the adult often react with a spontaneous,

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1029 B ODYBUILDING:THE B IONIC H UMAN

NEWS The Confusing Mix of Hype and Hope

Despite promising leads and publicity galore, restoring vision through ply tested the effects of surgically delivering bionic eyes remains a distant dream temporary electrical stimulations to the retina. Even so, Humayun is enthused that On 3 December 1999, media around the a plastic one—lends itself to hyperbole. Re- he has helped prove wrong several once- world reported that music icon Stevie Won- searchers work at the extreme edges of bio- common beliefs, such as that there is no safe der had said he might be able to see someday, logical knowledge, attempting feats that way to place an electronic device on a retina thanks to an artificial retina being developed were earlier the things of myths. Many also without damaging it. at Johns Hopkins University in Baltimore, have commercial ties and hope to profit from But John Wyatt, an electrical engineer at Maryland. Wonder made the remark at a fu- their research. the Massachusetts Institute of Technology, neral, where, according to press accounts, Restoring vision, in particular, has a says his work on a retinal implant with Har- the mourners jumped up and cheered. Many unique appeal, says ophthalmologist Eber- vard neuroophthalmologist Joseph Rizzo has researchers in this small, high-risk field had hart Zrenner, who is working on an artifi- shown how difficult it is to create meaning- a decidedly different reaction. Two weeks cial retina with colleagues at the University ful vision. In their experiments, what pa- later, their disbelief turned to tients “saw” often did not correlate with the dismay when ophthalmologist stimulation pattern, and none has recognized Mark Humayun of Hopkins’s shapes or letters, as reported by Humayun Wilmer Eye Institute appeared and colleagues. “Let’s aim for small success- with Wonder on the television es for now,” implores Wyatt. “To impart a newsmagazine 20/20. coarse level of vision that would expand a “There is the possibility that blind person’s autonomy is an ambitious but Stevie Wonder could see?” host plausible goal.” Barbara Walters asked. Overstating the promise comes at a steep “Certainly, there is the possi- price, says Wyatt, noting that the blind and bility,” said Humayun, caution- their families can suffer mightily. Harvard’s ing that he had yet to complete Don Eddington, a biophysicist who develops his exams of Wonder’s eyes. “We cochlear implants, says hype also “reduces believe that something could be the credibility of this work in the minds of available for patients in the next serious scientists.” Eddington says this, in on February 3, 2016 2 to 3 years.” turn, “not only affects the individuals work- Humayun and his colleague, Wonderland. A TV interview raised hopes—false, some say— ing in retinal prostheses but neural prosthe- ophthalmologist Eugene de Juan that Stevie Wonder might see, thanks to an artificial retina. ses in general.” Jr., now both at the Doheny Reti- Humayun strongly denies that he hyped na Institute at the University of Southern Eye Hospital in Tübingen, Germany (see anything. And Gerald Chader, chief scien- California (USC) in Los Angeles, have yet to Viewpoint on p. 1022). And success would tific director of the Foundation Fighting implant their artificial retina in a single pa- offer a unique triumph, he notes. You could Blindness near Baltimore and a former of-

tient. “It was basically just hype,” charges ri- say: “I’m the one who, like Jesus, made a ficial at the National Eye Institute (NEI), Downloaded from val Alan Chow, a pediatric ophthalmologist blind man see again.” agrees: “I think [Humayun] tried to be very in Chicago, Illinois. Chow’s company, Opto- A dozen research groups are attempting evenhanded.” bionics, is developing a similar artificial reti- to create artificial vision by either stimulat- Humayun points out that the story be- na now being tested in six patients—the ing the visual cortex in the brain, the optic came news because of what Stevie Wonder, only clinical trial of a permanently implant- nerve, or, the most popular approach, the not Humayun, said. And although he ed device under way. Chow also emphasizes retina. The retina is a collection of rods, agrees that Wonder is not an ideal candi- that Wonder, who became blind shortly after cones, ganglia, and bipolar cells sandwiched date for this device, he doesn’t think he birth, would not be a good candidate for together. Retinal devices are silicon chips provided false hope to the public or to prosthetic eyes, as he likely has substantial that surgeons can place on either side of that Wonder, as no one knows enough to state damage to his retinas. sandwich—the “subretinal” or the “epireti- definitively what will or won’t work. The But Humayun and others have accused nal” space. The hope is that the retinal cells USC group hopes to launch clinical trials Chow, too, of overstating his results. “He will transmit the signals to the optic nerve. of its device later this year. hasn’t shown any data, and it remains a Chow’s device, which he invented with his The Stevie Wonder publicity had an up- mystery why,” says Humayun, who attend- brother Vincent, resembles a fly’s eye and side, too, Chader argues. It sparked a flow of ed a Vitreous Society meeting last Novem- has solar cells that convert light into electri- philanthropic donations and, he suspects, ber where Chow claimed that his six pa- cal signals. The USC group and others are may have influenced NEI’s decision to fund tients had shown “substantial visual func- placing arrays of electrodes on their chips, a multimillion-dollar consortium effort tion improvements.” Before discussing de- which receive electronic signals from a that includes the USC researchers as well as tails, Chow says he wants a scientific jour- miniature camera mounted on glasses that Wyatt, Rizzo, and others. nal to publish his data. the person wears. Others use cameras to send When and if this research will pan out is Welcome to the contentious, competitive signals through wires implanted on the optic anyone’s guess. Says Zrenner, “That’s like world of research on the bionic eye. Bionics nerve or in the visual cortex. someone asking us when will we arrive at —be it restoring vision by implanting a sili- To date, however, results have been the moon when we’ve just begun to build a

con chip or replacing a damaged heart with mixed. Most human experiments have sim- rocket.” –JON COHEN CREDIT: IMAGES ALEXANDER SIBAJA/GETTY

1026 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN stimulation of the subthalamic nucleus of Par- middle turn. If one assumes that CI electrodes 18. D. E. Brackmann et al., Otolaryngol. Head Neck Surg. kinson’s patients with a stereotactically placed stimulate fibers en passant, they may stimulate 108, 624 (1993). electrode results in alleviation of many of the low frequencies too, even though the electrodes do 19. J. K. Moore, K. K. Osen, Am. J. Anat. 154, 393 (1979). not extend all the way into the apical turn. 20. In contrast to the VCN, the dorsal cochlear nucleus motor symptoms of this disease (31). 6. H. L. F. Helmholtz, Die Lehre von den Tonempfindun- (DCN) contains neurons with complex response Clearly, CIs and ABIs not only are of gen als physiologische Grundlage fu¨r die Theorie der properties already at this early phase of processing benefit to profoundly deaf individuals but Musik ( Vieweg, Braunschweig, Germany, 1863). and may be involved in early stages of spatial hearing 7. R. V. Shannon, F.-G. Zeng, J. Wygonski, V. Kamath, M. (21). also continue to provide insight into informa- Ekelid, Science 270, 303 (1995). 21. E. D. Young et al., Philos. Trans. R. Soc. London Ser. B tion processing in the auditory nervous sys- 8. M. A. Svirsky, A. M. Robbins, K. I. Kirk, D. B. Pisoni, R. T. 336, 407 (1992). tem. And certainly, each new understanding Miyamoto, Psychol. Sci. 11, 153 (2000). 22. However, speech perception in combination with lip- 9. R. Klinke, A. Kral, S. Heid, J. Tillein, R. Hartmann, reading is markedly improved (23, 24). achieved in basic scientific research will re- Science 285, 1729 (1999). 23. D. Matthies et al., J. Laryngol. Otol. Suppl. 27,32 sult in improvements to the technology of 10. T. N. Wiesel, Nature 299, 583 (1982). (2000). auditory prostheses, and increased benefits 11. J. P. Rauschecker, Trends Neurosci. 18, 36 (1999). 24. S. R. Otto et al., J. Neurosurg., in press. for patients. 12. J. K. Moore, Guan, J. Assoc. Res. Otolaryngol. 2, 297 25. G. S. Brindley, W. S. Lewin, J Physiol. 196, 479 (1968). (2001). 26. R. A. Normann, E. M. Maynard, P. J. Rousche, D. J. 13. C. Pantev et al., Nature 392, 811 (1998). Warren, Vision Res. 39, 2577 (1999). References and Notes 14. S. Rosen, A. Faulkner, L. Wilkinson, J. Acoust. Soc. Am. 27. Q. Bui, K. D. Wise, D. J. Anderson, IEEE Trans. Biomed. 1. G. E. Loeb, Annu. Rev. Neurosci. 13, 357 (1990). 106, 3629 (1999). Eng. 47, 281 (2000). 2. J. P. Rauschecker, Science 285, 1686 (1999). 15. I. Kohler, The Formation and Transformation of the 28. A. R. Guimaraes et al., Hum. Brain Map. 6, 33 (1998). 3. B. J. Edgerton, W. F. House, W. Hitselberger, Ann. Perceptual World [U¨ber Aufbau und Wandlungen der 29. T. D. Griffiths, S. Uppenkamp, I. Johnsrude, O. Jo- Otol. Rhinol. Otolaryngol. 91 (suppl.), 117 (1984). Wahrnehmungswelt (International Univ. Press, Vi- sephs, R. D. Patterson, Nature Neurosci. 4, 633 4. D. B. McCreery, R. V. Shannon, J. K. Moore, M. Chat- enna, 1951)], R. M. Rohrer, transl. (International Univ. (2001). terjee, IEEE Trans. Rehabil. Eng. 6, 391 (1998). Press, Vienna, 1964). 30. C. M. Wessinger et al., J. Cognit. Neurosci. 13,1 5. The cell somata of nerve fibers coming from the 16. Q.-J. Fu, R. V. Shannon, J. Galvin, in preparation. (2001). apical turn of the human cochlea are not located in 17. D. G. R. Evans et al., J. Med. Genet. 29, 841 31. M. R. DeLong, T. Wichmann, Ann. Neurol. 49, 142 the center of the apical coil, but more toward the (1992). (2001).

VIEWPOINT Repairing the Injured Spinal Cord Martin E. Schwab

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1029 B ODYBUILDING:THE B IONIC H UMAN but short-lived and ultimately abortive, attempt ly manipulating the downstream signaling path- (15). Injections and pumps can be used to at repair called regenerative sprouting. In vitro ways induced by these inhibitory signals in deliver neurotrophic factors to the lesion site. studies have revealed that spontaneous regen- growing neurites (12, 13). One way to deliver However, other strategies have also been erative sprouting is short-lived because the such agents is by subcutaneous pumps that tried, such as implanting cells engineered to adult CNS (particularly the myelin sheaths of inject antibodies or other blocking molecules secrete these factors directly into the lesion adult nerves) produces specific inhibitory pro- directly into the cerebrospinal fluid that bathes site, or using gene therapy with viral con- teins that block neurite outgrowth (5). Since this the spinal cord. Such pumps are routinely used structs containing neurotrophic factor genes discovery, there have been numerous attempts in the clinic for infusing drugs, e.g., into chronic to locally produce the factors where they are to ablate or neutralize these inhibitory factors in pain patients. needed (15). order to promote regenerative sprouting of the Neurotrophic factors are small proteins lesioned spinal cord. Ten years ago, in vivo secreted by a variety of different cell types Bridging Cysts and Scars studies showed that preventing myelin forma- that enhance nerve fiber outgrowth during In anatomically incomplete spinal cord le- tion or using a monoclonal antibody (mAb embryonic development. Local application of sions, regenerating nerve fibers are able to IN-1) that neutralized the activity of the myelin neurotrophic factors such as neurotrophin-3 bypass the injury site by using persisting protein Nogo-A (5, 6) promoted regeneration (NT-3) to lesioned fiber tracts of the spinal tissue bridges (Fig. 1). But sprouting nerve of corticospinal fibers (which connect the cere- cord can directly promote sprouting, and, in fibers often do not seem eager to cross this bral cortex and the spinal cord) in the lesioned some cases, can induce long-distance regen- stretch of foreign territory, in particular be- spinal cords of adult rats (2, 7, 8). These obser- eration of lesioned fibers (14, 15). The spec- cause scar tissue at the injury site contains vations were confirmed by immunizing le- ificities of neurotrophic factors for particular scar-associated neurite growth-inhibitory sioned animals with CNS myelin, lysing myelin nerve cell populations during embryonic de- molecules such as the chondroitin sulfate pro- with antibody and complement in vivo, or en- velopment are well established. However, as teoglycans (16). Enzymatic digestion of hancing myelin clearance by injecting activated yet, such specificities have not been well chondroitin sulfate proteoglycans enhances macrophages (9–11). Functional improvements worked out for the adult CNS. Indeed, the the regenerative growth of CNS fibers in a observed in such animals strongly indicated that regenerative effects of nerve growth factor brain lesion rat model (17). Scar formation is nerve fibers allowed to regrow in an adult spi- (NGF) on peripheral nerves, for example, a natural reaction of the lesioned CNS tissue, nal cord can still recognize and connect with have turned out to be clinically useless be- and attempts to prevent it have not been their correct targets. cause NGF affects pain-sensitive neurons, re- successful. The complex interplay between Current efforts are aimed at blocking sulting in hyperalgesia (increased sensitivity inflammatory cells and astrocytes, the glial Nogo-A as well as other growth-inhibitory mol- to pain). Although more than 30 neurotrophic cells of the nervous system that are the prin- ecules by the in vivo application of antibodies factors are known, fewer than six of them cipal contributors to scar formation, is still against these molecules, by blocking the recep- have been investigated as potential treatments poorly understood. tors to which they bind, or by pharmacological- for lesioned spinal cords in animal models Bridges that form a growth-permissive scaffold within the lesion site should greatly facil- Fig. 1. Spinal cord le- itate regenerative axon growth. Many types of sions in human and cells, tissues, or artificial materials have been rat. (A) Magnetic res- implanted as bridges into the injured spinal cord onance image of a hu- (18). The success of these experiments is lim- man spinal column ited, often because astrocytes “wall off ” the showing the spinal cord lying within the foreign object, thus greatly restricting access of central canal. Bone regenerating fibers to the implants. Important fragments from a exceptions to this rather deceiving picture are burst fracture of the implants made of Schwann cells (the glial cells middle vertebral bod- that produce the myelin sheaths surrounding ies have contused the peripheral nerves) or precisely placed peripher- spinal cord, resulting in a fluid-filled cyst al nerve grafts, which are invaded by regener- (dark) at the site of ating axons and can serve as bridges across tissue destruction. (B) even anatomically complete lesions (18, 19). In Longitudinal histologi- a spinal cord lesion rat model, implanted olfac- cal section through tory nerve glial cells align themselves with the the lesion site of an lesion site, migrate long distances into the an- injured rat spinal cord. Cysts and scar tissue terior and posterior spinal cord stumps, and are prominent at the guide regenerating axons through the lesion injury site. The corti- (20). The migratory behavior of these special- cospinal tract (black), ized glial cells, which are also known to pro- the bundle of nerve fi- duce a variety of growth factors and growth- bers connecting the permissive extracellular matrix molecules, may cerebral cortex to the spinal cord, is fully in- be part of their success. However, these glial terrupted by the le- cells may also damage intact nervous tissue, sion, depriving the because they are foreign to the spinal cord and lower spinal cord of its may promote an immunological reaction, thus cortical input for mo- negating the possibility of implanting them in tor control. (C) Strat- human patients. egies to repair the injured spinal cord include implanting scaffolds to bridge scars and cysts, treating the lesion site with antibodies or other agents that counteract the effects of neurite growth- Several labs are currently exploring the inhibitory factors, applying neurite growth-promoting molecules or nerve attractants by injection potential use of other cell types as bridging pump, or gene delivery to increase nerve fiber sprouting and nerve regeneration. material, including neural stem cells and

1030 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN cells genetically engineered to allow manipula- of adult rats (21–23). Repair of the myelin ery after spinal cord injury was bleak, and tion of their behavior or their ablation once their sheath and subsequent restoration of impulse this field of research was labeled a lost cause. bridging function is complete (14, 18). Neural conductivity in nerve fibers that have survived Now, at least from the research perspective, stem cells can be obtained by expanding a small the lesion would enable patients to make much the situation has changed radically. Some of biopsy of brain or bone marrow tissue in cul- better use of their surviving nerve fiber tracts. the new strategies to repair spinal cord inju- ture. These cells could then be induced to form ries, either alone or in combination, offer the growth-permissive astrocytes, which could be Enhancing CNS Plasticity possibility of clinically effective therapies for implanted into the spinal cord lesion cavity After spinal cord injury, the remaining intact paraplegic and quadriplegic patients in the (Fig. 1). If such cells misbehave, for example, nerve fibers react to the imbalance of the neural not too distant future. by inducing formation of dense scars or tumors, circuitry by sprouting compensatory fibers. Al- they could be ablated by activation of an inbuilt though pronounced before and just after birth, References suicide gene. For repair of lesioned spinal cord this plasticity (that is, the ability of one group of 1. R. J. Dumont et al., Clin. Neuropharmacol. 24, 254 (2001). tissue, fibrin or hydrogels loaded with growth nerve fibers to take over the role of another 2. M. E. Schwab, D. Bartholdi, Physiol. Rev. 76, 319 factors that attract regenerating neurites and injured group) is much less prominent in the (1996). keep scar formation under control show prom- adult CNS (24). Antibodies against neurite 3. M. S. Bracken et al., J. Neurosurg. 89, 699 (1998). ise. The principal obstacle to be overcome with growth-inhibitory factors have been shown to 4. B. A. Kakulas, J. Spinal Cord Med. 22, 119 (1999). all of these bridges is how to integrate them into enhance compensatory sprouting of nerve fi- 5. P. Caroni, M. E. Schwab, Neuron 1, 85 (1988). 6. M. S. Chen et al., Nature 403, 434 (2000). the spinal cord tissue without inducing scar bers in the partially lesioned brain stem and 7. T. Savio, M. E. Schwab, Proc. Natl. Acad. Sci. U.S.A. formation. spinal cord of the adult rat. This resulted in an 87, 4130 (1990). almost complete functional restoration of fine 8. L. Schnell, M. E. Schwab, Nature 343, 269 (1990). Myelin Repair 9. D. W. Huang et al., Neuron 24, 639 (1999). paw movements in these animals, as demon- 10. J. K. Dyer et al., Exp. Neurol. 154, 12 (1998). In all fast-conducting nerves, the nerve fibers strated by the food pellet reaching test (25, 26). 11. O. Rapalino et al., Nature Med. 4, 814 (1998). are surrounded by a myelin sheath. Local loss of Several of the interventions that promote regen- 12. C. Bro¨samle et al., J. Neurosci. 20, 8061 (2000). 13. P. A. Brittis, J. G. Flanagan, Neuron 30, 11 (2001). myelin often occurs in surviving bundles of eration of damaged nerve fibers (neurotrophic 14. P. J. Horner, F. H. Gage, Nature 407, 963 (2000). nerve fibers at the site of the spinal cord lesion factors, antibodies) will almost certainly also 15. S. Lacroix, M. H. Tuszynski, Neurorehabil. Neural Re- (2, 4), leading to disruption of electrical impulse enhance compensatory growth of nonlesioned pair 14, 265 (2000). conductivity. The adult spinal cord seems un- nerve fibers. In addition, the flow of electrical 16. J. W. Fawcett, R. A. Asher, Brain Res. Bull. 49, 377 (1999). able to efficiently replace either lost myelin or impulses in a particular neural circuit is 17. L. D. F. Moon et al., Nature Neurosci. 4, 465 (2001). loss of oligodendrocytes, the glial cells that pro- known to strengthen connections, or even to 18. M. B. Bunge, Neuroscientist 7, 325 (2001). duce myelin in the CNS. Neural stem cells, induce sprouting and the formation of new 19. H. Cheng et al., Science 273, 510 (1996). 20. A. Ramon-Cueto et al., J. Neurosci. 18, 3803 (1998). obtained from adult animals or human brain connections (24). Here, specific rehabilita- 21. N. Uchida et al., Proc. Natl. Acad. Sci. U.S.A. 97, biopsies (and in the future possibly from adult tion strategies may come into play. The ef- 14720 (2000). bone marrow) and expanded in culture, could be fects of activity on the molecular machinery 22. S. Liu et al., Proc. Natl. Acad. Sci. U.S.A. 97, 6126 (2000). primed with the correct growth factor cocktail to of nerve fiber growth are currently under 23. M. Sasaki et al., Glia 35, 26 (2001). become oligodendrocytes, which could then be investigation, as are the effects of combining 24. O. Raineteau, M. E. Schwab, Nature Rev. Neurosci. 2, implanted into lesion sites to promote myelin growth factors with specific training pro- 263 (2001). 25. M. Thallmair et al., Nature Neurosci. 1, 124 (1998). production. Successful remyelination by such grams for treating spinal cord injuries. 26. O. Raineteau et al., Proc. Natl. Acad. Sci. U.S.A. 98, cells has been shown to occur in the spinal cord A few years ago, the prognosis for recov- 6929 (2001).

www.sciencemag.org SCIENCE VOL 295 8 FEBRUARY 2002 1031 B ODYBUILDING:THE B IONIC H UMAN

NEWS The Quest to Reverse Time’s Toll

Scientists and entrepreneurs are optimistic to varying degrees about partnership with Cynthia Kenyon, a nema- prospects for extending the human life-span tode researcher at the University of Califor- nia, San Francisco, called Elixir, to see if Tom Johnson has a way to get rich quick, if they can make an aging retardant from the he wants to: Go into the business of grind- gene product. ing up his worms for snake oil. Johnson Roth of NIA is looking for a guaranteed studies aging at the University of Colorado, blockbuster. He’s involved in a company Boulder; his work with nematodes has re- called GeroTech, the goal of which is to vealed that changing a single gene can dra- make a “calorie restriction mimetic”—a matically increase the worm’s life-span. drug to make your body think you’re starv- Some years ago a clinic in Guadalajara, ing even while you’re eating normally. Mexico, proposed that he supply them with Roth says he already has encouraging re- extracts from long-lived nematodes to results from restricting monkeys’ calories: juvenate elderly patients. “I get two or three One of his subjects has lived to 38 instead similar offers a year,” says Johnson. of the usual 25. Aging research is hot these days. And so The biggest player in the antiaging re- is a commercial “antiaging” movement that search industry is Geron Corp. in Menlo is being fueled by demographics as millions Park, California, which has numerous of baby boomers hit their 50s. “Aging is patents on telomerase, the enzyme that re- probably the process that humans have most stores telomere length. Telomeres shorten as tried to conquer for our entire history,” says cells age and lose their ability to divide. Johnson. The difference today is “we now Calvin Harley, chief scientific officer, says know that we can alter aging.” Geron’s researchers have shown that they Scientists have known for decades that can transplant aging human skin cells, the a very low calorie diet extends life in telomerase in which has been activated, onto some species. Recently they have begun immunocompromised mice, where they be- to directly manipulate the mechanisms of have like young cells. The company is also aging with gene alterations that substan- working on both gene and drug therapies for on February 3, 2016 tially extend the life-span of fruit flies, ne- telomerase activation in age-related diseases. matodes, and mice. Using markers gener- Seeking youth. Oscar Wilde’s The Picture of Dorian Despite the heady possibilities, some re- ated by the Human Genome Project, re- Gray represents an eternal human goal. searchers caution that longevity probably searchers expect to come up with a flood comes with a downside. “We evolved in a of new candidates for genes influencing the gerontologist did not have some commer- high-hazard environment with disease, food processes of aging. These discoveries, and cial affiliation,” says pioneering aging re- shortages, droughts, accidents—enormous the promise of more to come, have spawned searcher Leonard Hayflick, now a consul- selective pressure to get the job done fast,”

a host of new companies dedicated to revers- tant to Genentech in South San Francisco. observes George Martin of the University Downloaded from ing aging’s toll. One way they hope to make money is by of Washington, Seattle. “You have genes George Roth, chief of molecular physiol- looking at those who survive to extreme old to make you develop rapidly, have progeny ogy and genetics at the National Institute age. Centenarians seem to be healthier than early. There was no need to develop robust on Aging (NIA), counts about 40 recent most old people: Witness Frenchwoman quality-control mechanisms allowing you to scientist-run start-ups, most of them looking Jeanne Calment, the oldest person on record, live beyond reproduction.” for life-extending genes—and many of who died in 1997 at 122. Last year, Thomas Mechanisms that serve the youthful or- which have already fallen by the wayside. Perls of Beth Israel Deaconess Medical Cen- ganism may precipitate aging or disease lat- There’s also bustling commerce in antiaging ter in Boston, director of the New England er on, notes Martin, who is also editor-in- products ranging from vitamin supplements Centenarian Study, with molecular geneti- chief of the Science of Aging Knowledge to potentially dangerous hormone prepara- cists Louis M. Kunkel and Annibale A. Puca Environment (Science’s SAGE KE, sageke. tions that purport to reverse diseases of ag- of Children’s Hospital, Boston, founded a sciencemag.org). Human growth hormone, ing. Some merchants try to give their prod- company called Centagenetix that is scan- for example, a favorite of antiaging huck- ucts a veneer of plausibility by citing recent ning the genomes of his subjects to find sters, affects metabolism and physiological research with telomeres, antioxidants, or genes that set them apart. indicators such as muscle mass. But it also hormones. Scientists such as zoologist Leonard Guarente of the Massachusetts creates undesirable side effects, such as Steven Austad of the University of Idaho in Institute of Technology is coming at the raised glucose levels, and may even shorten Moscow are wary. When groups interested problem from a different starting point: life. Other mechanisms are also two-edged. in life extension ask him for advice, “I tend yeast. Several years ago he found a gene Telomerase is implicated not just in cell re- to avoid them like the plague. … I don’t want called Sir2 that, when its expression was in- newal but cancer. Even the cancer-fighting to end up on their Web site.” creased, extended yeast’s life-span. Guarente p53 gene, when up-regulated, has been believes different organisms may share some found to hasten aging in rats (Science, 4 Jan- Researchers as entrepreneurs universal survival-promoting mechanisms— uary, p. 28). “It’s difficult to retard aging But nowadays everybody’s in the game. “I and Sir2 may be one of them. He’s so opti- without affecting something else,” says Aus-

would be surprised if any legitimate bio- mistic that he has founded a company, in tad. “Genes evolve in a certain context be- CREDIT: CORBIS

1032 8 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org B ODYBUILDING:THE B IONIC H UMAN cause they do something well at that level of expression. If you soup them up, there’s al- Cracking the Secrets of Aging ways a price to pay.” It’s hard to fight senescence, scientists admit, because “we still don’t know how to define aging per se,” says former National Institute on Aging head Robert Butler, now director of On the edge the U.S. branch of the International Longevity Center in New York City. Theories abound, S. Jay Olshansky, a biodemographer at the however, including the accumulation of cell damage (from oxygen free radicals and envi- University of Illinois, Chicago, cautions ronmental toxins) and glycation (stiffening of tissue with formation of sugar-protein that researchers should be careful of the bonds). Another is the progressive shortening of telomeres, the tips of chromosomes, as company they keep. “There is so much cells divide. Scientists are still not sure whether aging is a unitary phenomenon—the money to be made,” he says, that he and result, perhaps, of a predictable pattern of altered gene expression—or “a set of process- others are “really concerned” that some es that happen to occur in rough synchrony,” says Richard Miller of the University of scientists “have either crossed the line or Michigan Institute of Gerontology in Ann Arbor. are about to cross the line.” Meanwhile, the life-extending effect of calorie restriction offers researchers tantalizing That line is not always clear, and rep- clues. It’s been replicated in a number of species, including spiders: Those fed about two- utable scientists may differ on where it thirds of their regular diet lived longer, and dogs and rodents experienced later onset of should be drawn. For example, the Life Ex- diseases of aging, including arthritis and cancer. Scientists speculate that calorie restric- tension Foundation (LEF), based in Fort tion triggers a mechanism that probably evolved at times of food shortage, changing a va- Lauderdale, Florida, sells a vast assortment riety of gene functions that affect parameters such as glucose levels and body tempera- of products from vitamins to a product called ture so the body goes into a conservation mode. Some believe it works by slowing down pregnenolone that is said to contain “brain- energy utilization and its “wear and tear” on cells. boosting nutrients that function via a variety Even if researchers can figure out ways to slow the aging process, opinion varies on the of mechanisms to correct the molecular dev- ultimate limits to human life-span. S. Jay Olshansky, a biodemographer at the University astation that aging inflicts on the brain.” It of Illinois, Chicago, says that “anything past 130 is ridiculous.” But William Haseltine of had a long-running battle with the Food and Human Genome Sciences in Rockville, Maryland, thinks stem cells will ultimately prove Drug Administration over its claims. (A the route to virtual immortality. In an interview last year with Fool.com, Haseltine pre- court eventually ruled in LEF’s favor.) dicted that it will one day be possible to “reseed the body with our own cells that are LEF also funds research, including tests made more potent and younger, so we can repopulate the body.” of possible longevity compounds on mice Nematode man Tom Johnson of the University of Colorado, Boulder, is also optimistic. by Richard Weindruch of the University of There are no “aging” genes, he notes, because natural selection washes its hands of an or- Wisconsin, Madison, and Stephen Spindler ganism once it has passed the age of procreation. “The very absence of an evolutionary of the University of California (UC), reason to die makes it relatively easy to manipulate life-span,” he says. “If humans are as Riverside. Spindler, who is still at UC, is malleable as worms, we could see life-spans of 350.” now chief technical officer of a LEF- But the picture is obscured because there are no “biomarkers” for aging—such as a bio- founded company called LifeSpan Genet- chemical that predictably changes in level with age—as there are for, say, cancer or dia- ics, which uses gene chips to study the betes. Gray hair, which correlates about 0.7 with age, is still “as good as you could do” for ability of compounds to mimic genomic ef- a marker, says gerontologist Leonard Hayflick. At this point, the most reliable biomarker fects of calorie restriction. for aging is death. –C.H. Former NIA researcher Richard Cutler has an association with a company that Olshansky and Austad, at least, find distinctly problemat- oxidants can “significantly extend our health ly with his scientific colleagues, however, in ic. Cutler, who studies antioxidants, com- span,” although Cutler readily acknowledges distancing himself from the 7-year-old Amer- pounds that prevent damage to cells from that antioxidant supplements would benefit ican Academy of Anti-Aging Medicine, byproducts called only a minority of founded by physician Ron Klatz. “Practical free radicals, now people with oxidative immortality”—that is, living maybe a few works at Kronos, a stress from circum- hundred years if we want—is the group’s ulti- reputable nonprofit stances such as infec- mate goal, according to Klatz, who told in Phoenix, Arizona, tious diseases. “Younge- Science that “antiaging medicine has been devoted to uncover- vity makes a strong sell, smeared by the gerontology establishment, ing some of the se- that’s true,” he says. [which is] perfectly comfortable with the con- crets of aging. He al- “I’ve been asked [by cept that aging is immutable.” The group, so advises a compa- Kronos] to quit. I proba- which claims to have 10,000 physician and ny called Youngevity bly will.” He told scientist members, refers the public to antiag- in Carrollton, Texas, Science he will ask ing clinics and products and offers physicians which touts products Youngevity to use more certification in “antiaging medicine.” Olshan- containing a mix of accurate wording. Cut- sky says he goes to the annual convention in “miracle minerals” ler’s boss at Kronos, Las Vegas “to collect information on the latest captured from the Mitch Harman, was dis- antiaging quackery.” waters of Vilcabam- tressed when he learned If it’s “antiaging,” he notes, it’s quackery ba, Ecuador, whose that Cutler’s name was by definition. Some potions and elixirs may denizens “live their still on the Youngevity indeed have some effect on aging’s manifesta- lives in a state of Web site: “I’m trying to tions, but they don’t touch the still-mysterious youthfulness.” The do legitimate research processes at the core. Figuring out those pro- Youngevity Web site and get tarred with the cesses will drive research—and many com- cites his research as Special genes? Oldest documented human, same brush.” mercial endeavors—for years to come.

CREDIT: STR/AP evidence that anti- Madame Calment, at 122. Cutler stands firm- –CONSTANCE HOLDEN