NATURE BIOTECHNOLOGY VOL 18 FEBRUARY 2000 http://biotech.nature.com 143

ANALYSIS

FDA, researchers consider first transgenic fish

Officials at the US Food and DrugAdministration (FDA; Rockville, MD) are cur-rently reviewing an application from marinebiotechnology firm A/F Protein (Waltham,MA) for genetically modified (GM) salmon. Ifapproved, the transgenic fish, which have beenengineered to grow faster and consume lessfood than their wild counterparts, will be thefirst to reach the marketplace. However, anFDA-funded study suggests that transgenicfish could destroy wild populations, and somescientists are skeptical about their futurebecause of ecological concerns.

While vitamin-packed fish is still severalyears away, GM salmon are already beingproduced to address the problem of wastefrom farm-raised fish—something that hasbecome an issue for coastal communities inNew England and the Pacific Northwest con-cerned about the impacts of aquaculture.

A/F researchers have taken a gene promot-er from the ocean pout and inserted it into theAtlantic salmon, causing it to express growthhormone from its liver, as well as its pituitarygland where the hormone is normally pro-duced. As a result, the fish grows faster andeats less, says A/F Protein CEO Elliot Entis.

A/F’s GM salmon, which are bred at itsexperimental facility in Prince EdwardIsland, Canada, grow from an egg to 8 lbs inonly 14 to 18 months, half the normal time.At one year, they are 4 to 6 times larger than anormal fish of the same age. Entis insists hisGM fish are no larger than normal ones atadulthood, however they reach that stagemuch more quickly and consume 20–25%less food. “Our fish don’t produce anythingthat the wild type doesn’t produce.”

However, researchers at Purdue University(West Lafayette, IN) have found transgenicfish don’t live as long, and have published astudy recently finding that GM fish could passon this negative trait and quickly eradicatewild fish stocks (PNAS, Nov 1999).

During the FDA-funded study, WilliamMuir and Richard Howard found that labo-ratory fish called the Japanese medaka(Oryzias latipes) that were genetically modi-fied to produce human growth hormonematured faster and carried more eggs thannon-GM relatives. Male individuals, becauseof their larger size for their age, attracted fourtimes as many mates as smaller rivals.However, only two-thirds of GM medakasurvived to reproductive age—30% moretransgenic fish died within the first few daysof development than the “wild” fish.

Using a computer model, Muir calculatedthat GM fish—bigger but less viable—couldpass on inferior traits to a wild population.The transgenic growth hormone gene, there-

fore, was dubbed the “trojan gene”. His resultsshowed 60 transgenic fish could lead to theextinction of a population of 60,000 fish in 40generations. “This ‘trojan gene’ was one ofthe unexpected things we found,” says Muir.

Moreover, Muir says the risk of GM fishescaping from aquaculture pens into the wildis a considerable one. Indeed, in Maine,home to a $65 million salmon aquacultureindustry, federal wildlife officials, environ-mentalists, and the aquaculture industry arebattling against declining wild stocks ofAtlantic salmon thanks in part to cross-breeding with escaped farm-raised fish of thesame species. The farm-raised fish have trou-ble finding their way back to local streams tospawn and are diluting the wild stock, whichhas been proposed for adding to the federalendangered species list.

A/F Protein’s Entis insists his fish don’tdie early and that Muir’s study “has no rele-vance” to his commercial operation. In anycase, Entis says his firm’s fish eggs are steril-ized and therefore would not pass on anyinferior traits.

However, Muir notes that A/F Protein’ssterilization technique—in which high pres-sure creates a triploidy instead of diploidychromosomes in the developing egg—is not100% reliable. “All you need is one egg,” says

Muir. “We consider aquaculture the onlyplace where biotechnology could have a riskin the environment.”

The FDA, which is developing a modelfor the potential ecological risks of GM fish,would not comment on when A/F Protein’sreview will be complete.

Canadian government researchers aredeveloping GM coho salmon and rainbowtrout using a different transgene but methodssimilar to those of A/F Protein. Their studieson reproduction of the transgenic fish foundthat adult GM fish are 1.5 to 2 times largerthan normal, even though they are raised inlaboratory conditions. They feed more effec-tively but don’t swim as well as their wildcounterparts, according to Robert Devlin,chief scientist at the Department of Fisheriesand Oceans in West Vancouver, BC.

Devlin says that very little data on sur-vivorship of transgenic fish has been pub-lished, but agrees that the commercial pro-duction of such fish will depend on whetherthey can be made sterile. The answer may beto raise them on land rather than in oceanpens, for example. “Right now, people aremaking scare scenarios without any infor-mation,” says Devlin. “We just need to becareful and move a bit slowly.”

Eric Niiler

apparently brought on by a massive immunesystem response. Other unusual findingsincluded severe damage to blood progenitorcells in the bone marrow, possibly due to asuperimposed infection by another virus.

Current efforts to assess the safety ofgene-therapy procedures are complicatedbecause modified adenovirus-based vectorsdiffer extensively from one another withrespect to structural variability, differencesin the way they are delivered to patients,and other differences that reflect the varietyof transgenes that have been introducedinto these vectors in efforts to treat variousdiseases.

The most recent version of the UP aden-ovirus vector has had several of its genesdeleted, rendering it unable to replicate with-out a helper virus, according to its developersJames Wilson and UP colleagues. Its generalsafety characteristics are good, based on ani-mal tests, including in several strains of miceand primates. However, at high vector doses,some animals develop complications, includ-ing blood clotting abnormalities and mild tomoderate liver damage, and several of them

The US Food and Drug Administration (FDA;Rockville, MD) and the National Institutes ofHealth (NIH; Bethesda, MD) have respondedto widespread concerns about the risks ofgene therapy with public meetings, reviewsand investigations, and a series of administra-tive changes to deal with gene therapy morebroadly. During December, members of theNIH Recombinant DNA Advisory Committee(NIHRAC), outside experts, NIH and FDAofficials, and members of the news media andpublic met to review safety findings from bothpreclinical testing and gene therapy clinicaltrials. The meeting also provided a forum forairing renewed concerns about the effective-ness of federal oversight for such research,whose duties are shared by FDA and NIH.

The action follows the death lastSeptember of Jesse Gelsinger, an 18-year-oldparticipant in a phase I gene therapy clinicaltrial at the University of Pennsylvania (UP;Philadelphia, PA) to evaluate the safety of agenetically engineered adenovirus vector(Nat. Biotechnol, 17, 1153). Gelsinger diedfollowing an acute respiratory system col-lapse and subsequent multiorgan failure,

Investigation of gene therapy begins

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om

144 NATURE BIOTECHNOLOGY VOL 18 FEBRUARY 2000 http://biotech.nature.com

ANALYSIS

In contrast to the vigor of public companieslike CuraGen, Celera, and Affymetrix, whoseshare prices are currently appreciating rapidly,smaller private companies that serve thegenomics sector appear to be on shakier foot-ing. The demise in December of MolecularApplications Group (MAG), a bioinformaticssoftware company in Palo Alto, CA, could por-tend trouble ahead for other specialty firms.

Debbie Yu, a Silicon Valley venture capi-talist and MAG’s caretaker president, whobroke up the company and sold off its partsto Affymetrix (Santa Clara, CA) and Celera(Rockville, MD), says, “bioinformatics soft-ware can still be an attractive business, butthe tools have to be really effective and value-enhancing.” She says that many of the com-panies’ problems are due to both what theysell and how they sell it.

These companies face the especially-formi-dable twin hurdles of narrow markets andhigh development costs. Cordell Brown,General Manager of Hitachi Genetic Systems(Alameda, CA), whose DNASIS suite ofsequence analysis programs were a mainstay ofthe company until the mid-90s, says his com-pany’s emphasis today has shifted to equip-ment such as chip spotters and fluorescence

imagers. Market saturation plus competitionfrom the NCBI’s (US National Center forBiotechnology Information) free tools anddatabases equaled an environment in which“we couldn’t make it on software alone.”

Roy Whitfield, president of IncytePharmaceuticals in Palo Alto, CA, concurs.“Even if specialty companies do everythingright and dominate the field, the numbersaren’t there,” he says, citing chemical informat-ics. In that sector, a couple of dozen companiesa decade ago, such as BioCAD and Hypercube,have today sifted down to two dominantfirms—MDL Information (San Leandro, CA)and Molecular Simulations (San Diego, CA).

It is the financiers of the first generationof bioinformatics companies who are toblame, according to Lion Bioscience’s(Heidelburg) Christian Marcazzo, productmanager for its SRS bioinformatics database-querying software. They “grievously under-estimated the scientific and technologicalhurdles necessary to bring these products tomarket.” Marcazzo says that from concept toproduct to IPO in 15 months—as if theywere e-commerce companies—“is not a real-istic expectation in bioinformatics, as theventure capitalists are now discovering.”

Companies like MAG also hurt them-selves with inappropriate marketing strate-gies, adds Yu. “Software is too often soldunder license agreements, but unless these

are at least 6 figures, it’s a long row to hoe toget to a big enough revenue base.” Pharmacustomers don’t like buying expensive soft-ware solutions, she continues, so a modularapproach such as the creative one Spotfire(Cambridge, MA) has adopted is a betterbusiness model for specialty companies.“Spotfire is maximizing every possible distri-bution channel,” she observes.

Meanwhile, Whitfield believes that con-tent provision and data processing have farmore potential for growth and revenue, andare thus more interesting from a businessstandpoint, than the tools segment.“Content, data processing, and tools arethree quite different businesses, and havedifferent sustainabilities,” he says.

Doug Brutlag, a Stanford biochemistryprofessor and chief scientific officer ofDoubleTwist (Oakland, CA), says the internetis one way for small bioinformatics companiesto broaden their base. “The old paradigm ofsoftware and database distribution to the cus-tomer is on the decline, and the modern one ofproviding services on the Web, so that the cus-tomer comes to you, is taking over,” he says.DoubleTwist, which provides application ser-vices for gene discovery, is in the process ofreinventing itself along these lines. “Each sci-entist has different needs, and software may below on the list.”

Potter Wickware

died during such tests. This vector wasadministered to about 20 individuals duringthe Gelsinger trial, most of whom—exceptfor Gelsinger—developed a range of moder-ately adverse symptoms.

Other factors appear to complicate theclinical use of adenovirus-based gene vec-tors, according to Wilson and otherresearchers. For instance, the doses at whichthere are toxic effects or potential therapeuticeffects may be separated only narrowly, andthere may be thresholds where adverse effectsabruptly appear—complicating how vectorsmight be used and perhaps undermining thereliability of results from tests in animals.Moreover, such viruses can sometimes pro-voke or otherwise disrupt cytokine-deter-mined inflammatory responses, according toLinda Gooding of Emory University(Atlanta, GA), one of several experts on aNIH–FDA working group that is reviewingadenovirus-related adverse effects.

Equally if not more problematic forwould-be gene-therapy procedures, thesevectors are not so reliable in delivering genesto where they are targeted. After the aden-ovirus vector was applied through a catheteronto the liver of Gelsinger and others in the

trial, it spread widely through other organsand also, at least early on, into immune sys-tem cells, based on the post mortem analysisof his tissues—distributing quite differentlyfrom how it behaved during animal experi-ments, according to Wilson.

FDA officials routinely review gene-transfer and gene-therapy proposals, whichare also subject to oversight by officials inthe NIH Office of Biotechnology Activities(OBA; formerly, Office of RecombinantDNA Activities). In the aftermath of theGelsinger death, FDA officials began a for-mal investigation of the UP clinical trial.Kathryn Zoon, Director of the FDA Centerfor Biologics Evaluation and Research, saidthere was “preliminary evidence of protocoldeviations.” Although Zoon would notdescribe these violations, they apparentlyconcern escalating doses and enrolment ofpatients. For instance, the slot that Gelsingerfilled was originally for a woman, accordingto the protocol. Another apparent violationrevolves around ammonia levels inGelsinger’s blood, which were higher thanthose stipulated by the protocol.

However, other agency officials indicatedthat there was extensive communication

between them and members of the UP teamduring the course of the clinical trial.Moreover, on several occasions, officialsexplicitly approved requests from the UPteam to move ahead and use higher doses ofthe adenovirus vector—the highest of whichproved fatal in Gelsinger’s case.

In addition to the specific investigation,FDA and NIH officials renewed their genetherapy review-harmonizing efforts, andboth agencies notified researchers in thisfield of their obligations to report adverseeffects promptly. The proposals embedded inthese formal notices led industry representa-tives to renew long-held objections to theduplicative regulatory hoops through whichresearchers in this field need to jump. Forexample, the Biotechnology IndustryOrganization (BIO; Washington, DC) issueda white paper in December urging NIHRAC,OBA, and FDA to protect information fur-nished to them and, in particular, avoid dis-closing proprietary information. It alsoadmonished officials to keep FDA “the onlyagency with regulatory authority,” whilemaintaining NIHRAC in its “role as an edu-cational advisory body.”

Jeffrey L. Fox

MAG’s demise signals trouble for bioinformatics firms

Potter Wickware is a freelance writer workingin Mill Valley, CA.

© 2000 Nature America Inc. • http://biotech.nature.com©

200

0 N

atu

re A

mer

ica

Inc.

• h

ttp

://b

iote

ch.n

atu

re.c

om