20 MAY 1988
`VoL. 240 • PAGES 957-IIIZ
`
`$3.00
`
`AMERICAN
`
`I
`
`BEQ 1027
`Page 1
`
`

`
`AMERICAN
`ASSOCIATION FOR THE
`ADVANCEMENT OF
`SCIENCE
`
`-~14fi*'*i.ihiiit511-
`
`M;i44Mik411~14fA-
`
`MaWMikiMDMii-
`
`Reports
`
`.
`
`.
`
`• CIENCE ISSN 0036-8075
`
`20 MAY 1988
`VOLUME 240
`NUMBER-4855
`
`963 This Week in Science
`
`965 Setting Priorities in Science
`
`967 Crisis in Biosystematics of Arthropods: J. H. OLIVER, JR.
`
`968 Fraud Allegations: H. H. WoRTIS, B. T. HUBER, R. T. WooDLAND • PRC
`Students Abroad: X. HAo; P. SUEDFELD • Estimates of Species Duration:
`M. P. RussELL AND D. R. LINDBERG; D. JABLONSKI
`
`973
`976
`977
`978
`980
`
`981
`
`982
`984
`986
`
`Testing the Limits at Mach 25
`A Change of Heart
`Rethinking Technology's Role in Economic Change
`Giving the Muse a Helping Hand
`Briefing: NSF Appoints Two Assistant Directors • Legislating Labs as Drug-Free ,
`Workplaces • Soviet Satellite in Trouble; Groups Call for Ban on Orbiting
`Reactors • National Science Board Elects Good and Day • Curien Returns as
`French Science Minister
`Biological Defense Defended
`
`I
`1
`• )
`
`Chromosomes: The Ends in View
`A Revolution of Ideas in Agricultural Origins
`Pluto's Orbital Motion Looks Chaotic
`
`991 Sampling Rare and Elusive Populations: S. SuDMAN, M. G. SIRKEN,
`C. D. CowAN
`996 Abrupt Climate Change and Extinction Events in Earth History: T. J. CROWLEY
`AND G. R. NoRTH
`1003 A Molecular Basis for MHC Class 11-Associated Autoimmunity: J. A. ToDD,
`H. AcHA-ORBEA, J. I. BELL, N. CHAo, Z. FRONEK, C. 0. JAcOB,
`M. McDERMOTI, A. A. SINHA, L. TIMMERMAN, L. STEINMAN, et al.
`
`1010 Fos-Associated Protein (p39) Is the Product ofthejun Proto-oncogene:
`F. J. RAuscHER III, D. R. CoHEN, T. CuRRAN, T. J. Bas, P. K. VoGT,
`D. BoHMANN, R. TJIAN, B. R. FRANZA, JR.
`
`1017 Aluminum-Induced Calcium Deficiency Syndrome in Declining Red Spruce:
`w. c. SHORTLE AND K. T. SMITH
`1018 Antarctic Owne Depletion Chemistry: Reactions ofN20 5 with H 20 and HCl on
`Ice Surfaces: M. A. ToLBERT, M. J. Rossi, D. M. GoLDEN
`1021 Evidence for Highly Reflecting Materials on the Surface and Subsurface of
`Venus: R. F. JURGENS, M.A. SLADE, R. S. SAUNDERS
`1024 Radioimmunotherapy with Alpha Particle-Emitting Immunoconjugates:
`R. M. MAcKLIS, B. M. KINSEY, A. I. KAssis, J. L. M. FERRARA,
`R. w. ATCHER, J. HINES, c. N. COLEMAN, s. J. ADELSTEIN, s. J. BURAKOFF
`
`J
`I
`=-------~----------~------------------------------~----1
`•
`SCIENCE Is published weekly on Friday, except the last week In December, and with an extra Issue In February
`by the American Association for the Advancement of Science, 1333 H Street, NW, Washington, DC 20005. Sec(cid:173)
`ond-class postage (publication No. 484460) paid at Washington, DC, and at an addifional entry. Now combined wtth The '
`Scientific Monthly® Copyright © 1988 by the American Association lor the Advancement of Science. The title SCI(cid:173)
`ENCE Is a registered trademark of the AAAS. Domestic individual membership and subscription (51 issues): $70. Do(cid:173)
`mestic institutional subscription (51 issues): $110. Foreign postage extra: Canada $32, other (surface mail) $32, air-sur- 1
`face via Amsterdam $85. First class, ainnaii, school-year, and student rates on request. Single copies $3.00 ($3.50 by mail); 1
`
`back issues $4.50 ($5.00 by mail); Biotechnology Issue, $5.50 ($6 by mail); classroom rates on request; Guide to Bio·
`technology Products and Instruments $16 ($17 by mail). Change of address: allow 6 weeks, gMng old and new ad-
`dresses and seven-digit account number. Authorization to photocopy material for internal or personal use under circum(cid:173)
`stances not falling within the fair use provisions of the Copyright Act is granted by AAAS to libraries and other users reg-
`istered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $1 per
`copy plus $0.10 per page is paid directly to CCC, 21 Congress Street, Salem, Massachusetts 01970. The identification
`code for Science is 0036-8075/83 $1 + .10. Postmaster: Send Form 3579 to Science, 1333 H Stree~ NW, Washington,
`DC 20005. Science is indexed in the Reader's Guida to Periodical Literature and in several specialized indexes .
`The American Association for the Advancement of Science was founded in 1848 and incorporated in 11!74. Its objects
`are to further the work of scientists, to facilttate cooperation among them, to foster scientific freedom and responsibility,
`to improve the effectiveness of science In the promotion of human welfare, and to increase public understanding and ap-'
`preciation of the importance and promise of the methods of science in human progress.
`
`•
`
`,
`
`•
`
`1
`
`•
`
`BEQ 1027
`Page 2
`
`

`
`14;. P. T. Jones, P. H. Dear, J. Foote, M.S. Neuberger,
`G. Winter, ibid. 321, 522 (1986).
`15. M. Potter, Adv. Immunol. 25, 141 (1977); R. M.
`Perlmutteretal., ibid. 35, 1 (1984); S. Rudikotfand
`M. Potter, Biochemistry 13, 4033 (1974); S. Rudi(cid:173)
`koff, Y. Satow, E. Padlan, D. Davies, M. Potter,
`Mol. Immunol. 18, 705 (1981).
`16. D. M. Segal et al., Proc. Nat!. Acad. Sci. U.S.A. 71,
`4298 (1974).
`17. L. Hendershot, D. Bole, J. F. Kearney, Immunol.
`Today 8, 111 (1987).
`18. M. S. Briggs and L. M. Gierasch,Adv. Protein Chem.
`38, 109 (1986).
`19. N. R. Movva, K. Nakamura, M. Inouye, J. Bioi.
`Chem. 255, 27 (1980).
`20. H. Inouye, W. Barnes, J. Beckwith,]. Bacteriol. 149,
`434 (1982).
`21. The rwo chains of the affinity-purified Fv protein
`were separated using SDS-PAGE and blotted onto
`an activated glass fiber sheet (22). The rwo bands
`corresponding to each of the rwo chains were each
`individually subjected to six cycles of gas-phase
`
`sequencing. In both sequenaror profiles, the NH2-
`terrninal amino acids expected from precise cleavage
`of both signals were seen exclusively.
`22. C. Eckerskom, W. Mewes, H. Goretzki, F. Lon(cid:173)
`speich, Bur. J. Biochem., in press.
`23. H. Metzger, B. Chesebro, N. M. Hadler, J. Lee, N.
`Otchin, in Progress in Immunology: Proceedings of the
`1st Congress of Immunology (Academic Press, New
`York, 1971), pp. 253-267.
`24. G. Schumacher, D. Sizmann, H. Haug, P. Buckel,
`A. BOck, Nucleic Acids Res. 14, 5713 (1986).
`25. J. R. Hobbs, Essays Med. Biochem. 1, 105 (1975).
`26. D. Inbar, J. Hochman, D. Givol. Proc. Nat!. Acad.
`Sci. U.S.A. 69, 2659 (1972).
`27. J. Sen and S. Beychok, Proteins 1, 256 (1986).
`28. R. Huber and W. S. Bennett, Nature 326, 334
`(1987); M. Marquart and J. Deisenhofer, Immunol.
`Today 3, 160 (1982).
`29. C . .Yanisch-Perron et al., Gene 33, 103 (1985).
`30. A. Skerra and A. Pltickthun, in preparation.
`31. S. P. Fling and D. S. Gregerson, Anal. Biochem.
`155, 83 (1986).
`
`32. We monitored the selectivity of the cell fractionation
`using 13-lacramase as a periplasmic and 13-galacto(cid:173)
`sidase as a cytoplasmic marker enzyme. Under the
`conditions described, approximately 90% of roral13-
`lactamase activity and less than 0.5% of total 13-
`galactosidase activity were found in the periplasmic
`fraction.
`33. B. Chesebro and H. Metzger, Biochemistry 11, 766
`(1972).
`34. R. K. Scopes, Protein Purification: Principles and
`Practice (Springer, New York, 1982), pp. 241-242.
`35. G. Scatchard,Ann. N.Y.Acad. Sci. 51,660 (1949).
`36. We thank C. Eckerskom and F. Lottspeich for the
`NH2-terrninal protein sequencing and R. Glocks(cid:173)
`huber for the preparation of the phosphorylcholine
`affinity column. Supported by grant BCT0372 from
`the Bundesminisrerium fiir Forschung und Techno(cid:173)
`logie ro A.P., and by a predocroral fellowship from
`the Stifrung Volkswagenwerk and the Fonds der
`Chemischen Indusrrie ro A.S.
`
`7 January 1988; accepted 18 March 1988
`
`Escherichia coli Secretion of an Active Chimeric
`Antibody Fragment
`
`MARc BETTER, C. PAUL CHANG, RANDY R. RoBINSON,
`ARNOLD H. HORWITZ
`
`A chimeric mouse-human Fab protein that binds specifically to the human carcinoma
`cell line C334 7 has been expressed and secreted from Escherichia coli. This molecule,
`which contains functionally assembled kappa and Fd proteins, binds as effectively to
`sites on the surface of C334 7 cells as Fab fragments prepared proteolytically from
`whole chimeric or mouse antibody. The production in Escherichia coli of foreign
`heterodimeric protein reagents, such as Fab, should prove useful in the management of
`human disease.
`
`T HE BINDING-SELECTIVITY OF ANTI(cid:173)
`
`body molecules makes them suited
`for applications as diverse as affinity
`chromatography, diagnostic reagents, and
`therapeutics in the detection and treatment
`of human diseases. Monoclonal antibodies
`are especially useful for these purposes be(cid:173)
`cause they can be prepared with homoge(cid:173)
`neous recognition specificities targeted at
`virtually an unlimited number of antigenic
`determinants. The protein domains that
`confer these antigen recognition determi(cid:173)
`nants can be proteolytically separated from
`the remainder of the molecule and still retain
`their antigen-binding ability. This portion
`of an antibody (Fab) is roughly one-third
`the size of an intact immunoglobulin G
`(IgG) (about 48 kD) and exhibits monova(cid:173)
`lent antigen binding. The similar F(ab')z
`portion retains divalent antigen-binding ca(cid:173)
`pacity and contains both recognition do(cid:173)
`mains linked by two interchain disulfide
`bridges. Antibodies differ, however, in their
`susceptibility to proteolytic cleavage, and
`preparations can be heterogeneous. The rel(cid:173)
`atively simple structure of a Fab (5 disulfide
`
`Inremational Genetic Engineering, Inc. (INGENE),
`1545 17th Street, Santa Monica, CA 90404.
`
`20 MAY 1988
`
`bonds) compared to an intact antibody (16
`disulfide bonds) and the therapeutic useful(cid:173)
`ness of Fab molecules make them attractive
`targets for production by microbial fermen(cid:173)
`tation after appropriate protein engineering.
`Here we discuss the expression of a mouse(cid:173)
`human chimeric Fab in Escherichia coli, that
`is, a molecule that contains the variable
`regions (antigen recognition domains) from
`a mouse monoclonal antibody and the CH 1
`and CK constant regions from a human IgGl
`antibody.
`Each protein chain of a Fab has two
`intrachain disulfide bonds that stabilize
`functional domains, and a single cysteine
`involved in interchain disulfide linkage.
`Escherichia coli has been used to produce
`individual immunoglobulin chains internal(cid:173)
`ly that are not properly folded (1, 2), or
`individually secreted chains (3); however,
`for E. coli to assemble the truncated heavy
`chain (Fd) and K into the correct heterodi(cid:173)
`meric molecule, both chains must be trans(cid:173)
`lated simultaneously and secreted. This op(cid:173)
`eration would then mimic the cognate
`immunoglobulin assembly process.
`The chimeric L6 antibody (4), directed
`toward a ganglioside antigen expressed on
`the cell surface of many human carcinomas
`
`(5), has been described. This antibody, pre(cid:173)
`pared from the culture supernatant of an
`Sp2/0 transfectoma cell line, is expressed
`from the eDNA copies of the chimeric L6
`IgG 'Y and K chain genes. The eDNA clones
`for these two chimeric genes were used as
`the starting point for expression of a Fab
`molecule in mammalian cells and bacteria.
`A termination codon was introduced into
`the chimeric heavy chain gene at amino acid
`228 by site-directed mutagenesis (6) (Fig.
`la) in a manner that introduced ;1 Bel I
`restriction site. A similar step introduced a
`restriction site, Sst I, into the coding region
`at the processing cleavage site of the native
`heavy chain leader peptide and the mature
`heavy chain. Site-directed mutagenesis was
`
`Table 1. Binding activity of bacterial Fab to
`human carcinoma cells. Target cells were incubat(cid:173)
`ed for 30 min at 4°C with each antibody or Fab,
`followed by incubation with fluorescein isothio(cid:173)
`cyanate (FITC)-labeled goat antibody against
`human K for the bacterial Fab, FITC-labeled goat
`antibody against mouse IgG for the L6 mouse
`antibody, FITC-labeled goat antibody against
`mouse K for L6 mouse Fab, or goat antibody
`against human IgG for the chimeric L6 antibody.
`We determined antibody bi,nding to the cell sur(cid:173)
`face by using a Coulter model EPIC-C cell sorter.
`FITC-labeled antibodies were obtained from
`TAGO.
`
`Antibody
`
`Binding ratio*
`
`C3347 cells
`L6+
`
`TSl cells
`L6-
`
`Mouse L6
`Sp2/0 chimeric L6
`Bacterial L6 Fab
`Mouse L6 Fabt
`
`95
`116
`54
`16
`
`l
`l
`l
`l
`
`*The binding ratio is the number of times brighter a rest
`sam.Ple is than a control sample treated with FITC(cid:173)
`conjugated second antibody. Quantitative differences in
`binding to C3347 cells probably reflect the relative
`activity of individual FITC-conjugates. Data shown are
`t Prepared by
`from one of rwo similar binding assays.
`enzymatic digestion of mouse L6 antibody with papain.
`
`REPORTS 104-1
`
`BEQ 1027
`Page 3
`
`

`
`also used to insert an Aat II restriction site
`into the chimeric K chain gene at the junc(cid:173)
`tion of the leader peptide and the mature
`processed K chain. Each coding sequence,
`Fd and K, was fused to the leader peptide
`segment of the bacterial pelB gene (pectate
`lyase) from Envinia carotovora ( 7) generating
`a gene fusion between the leader peptide
`segment of the pectate lyase gene and the
`mature coding sequence of the Fd and K
`genes. This bacterial leader sequence was
`chosen to direct membrane translocation in
`E. coli since pectate lyase can be secreted to
`high levels under the control of a regulated
`promoter. To ensure that both Fd and K
`were also translated in close physical prox(cid:173)
`imity, we assembled a plasmid that codes
`both genes in a single dicistronic message
`(Fig. 1b) . This operon was placed under the
`control of the inducible araB promoter from
`Salmonella typhimurium ( 8) and expressed in
`E. coli.
`Examination of culture supernatants or
`extracts of the peri plasmic space (9) of E. coli
`by enzyme-linked
`immunosorbent assay
`(ELISA) for chimeric K with antibody
`against human K (Cappel), or Fab produc(cid:173)
`tion with antibody against human Fd (Cal(cid:173)
`biochem) and antibody against human K,
`revealed that about 90% of the secreted K
`chain accumulated in the culture medium.
`This was a surprising observation that al(cid:173)
`lowed simple purification of this material
`from induced bacterial cultures. Approxi(cid:173)
`mately 2 mg/liter of material reactive as Fab
`in an ELISA is secreted into culture super-
`
`natants of MC106l (p1Tl06). Immunoblot
`analysis with antibody against human K
`revealed that under nonreducing gel condi(cid:173)
`tions, the predominant reactive species had a
`molecular size of about 48 kD. Under gel
`conditions where disulfide linkages were
`reduced, the predominant species had a mo(cid:173)
`lecular size of about 23 kD. These observa(cid:173)
`tions are consistent with the predicted mo(cid:173)
`lecular sizes of the processed chimeric Fab
`(K, 23.3 kD; Fd, 24.7 kD), and suggest that
`the material is properly assembled.
`For purification ofFab, bacterial superna(cid:173)
`tants were concentrated, filtered, and loaded
`on an SP disk equilibrated with 10 mM
`phosphate buffer, pH 7.5. Fab was eluted
`with O.W NaCl and purified by S Sepha(cid:173)
`rose column chromatography, where it was
`eluted as a single peak with a linear 0 to
`0.12Ai NaCI gradient. The immunologically
`reactive material was more than 90% pure as
`determined by SDS-polyacrylamide gel
`electrophoresis (PAGE) followed by Coo(cid:173)
`massie blue staining (Fig. 2). Purified mate(cid:173)
`rial has a molecular size of about 48 kD
`under nonreducing gel conditions and
`about 24.5 and 23 kD under reducing gel
`conditions. The 23-kD band is immunologi(cid:173)
`cally distinguishable by using antibody
`against human K.
`Purified bacterial Fab was tested for bind(cid:173)
`ing to L6 antigen-containing cells (Table
`1). Bacterial F ab bound specifically to the
`human colon carcinoma cell line C3347;
`cells from the T cell line T51 served as a
`negative control. Bacterially produced Fab
`
`also exhibited characteristic binding inhibi(cid:173)
`tion of FITC-labeled mouse L6 antibody to
`the surface of antigen-positive C3347 colon
`carcinoma cells (Fig. 3). We tested bacterial(cid:173)
`ly produced chimeric Fab, proteolytically
`prepared Fab from L6 mouse antibody, L6
`chimeric antibody, and Fab prepared from
`Sp2/0 cells transfected with the truncated
`chimeric Fd and the chimeric K gene. All
`Fab preparations have essentially identical
`binding inhibition profiles. The proteolyti(cid:173)
`cally produced Fab contains a significant
`proportion of degraded, low molecular size
`peptides, whereas chimeric Fab from bacte(cid:173)
`ria or Sp2/0 cells is homogeneous (Fig. 2).
`Protein engineering allowed the expres(cid:173)
`sion in E. coli of a functional chimeric Fab
`that has binding specificity for a human
`carcinoma cell marker. The finding that E .
`coli can be engineered to secrete a foreign
`heterodimeric molecule builds on the earlier
`report that intrachain disulfide bonds can
`form correctly in proteins secreted into the
`periplasmic space of E. coli (10). One useful
`application for bacterially produced Fab
`molecules will be in tumor imaging in vivo
`(11) 12) [tumor marker-specific Fabs pro(cid:173)
`duced proteolytically from whole antibodies
`have already been used for this purpose (13-
`15)]. A great advantage of the engineered
`bacterial Fab is that the protein heterogene(cid:173)
`ity that results from nonspecific cleavages
`and differences in the susceptibility of anti-
`
`a
`
`B Sp L6' L6
`
`HC
`
`t
`
`t
`
`226
`lie Cys Asp Lys Thr His Thr Cys Pro Pro Cys Pro Ala
`ATC TGT GAC AAA ACT CAC ACA TGC CCA CCG TGC CCA GCA
`
`a
`
`LC
`LJ
`
`t
`
`Fig. 1. (a) Strategy for introduction of a stop
`codon and Bel I restriction site at amino acid 228
`in the chimeric L6 heavy chain eDNA clone. An
`M l3mp 19 subclone containing a portion (V H,
`CHI, CH2) of the chimeric L6
`-y gene,
`piNGl400, was used as a substrate for site(cid:173)
`directed mutagenesis with the primer shown.
`Arrows indicate the cysteine residue involved in
`disulfide linkage with light chain (LC) and those
`involved in linkage with heavy chain (HC) in an
`intact IgG. (b) Map of plasmid p1Tl06 and a
`schematic view of the relevant Fab operon placed
`under the control of the S. typhimurium araB
`promoter (ParaB). The 300-bp pelB leader se(cid:173)
`quence cassette including the 22-amino acid pelB
`leader and 230 bp of upstream DNA was derived
`from pSSl004 (7). Gene fusions were construct-
`ed with pelB and both Fd and K genes. Erwinia
`carotovora DNA sequences 5' to the -48 position
`were removed by means of an Nde I restriction
`site, and the coding sequences for these genes
`were assembled [in the araB expression plasmid
`piT2 (11)] into a single transcription unit with 70
`bp separating the two genes. The pectate lyase
`ribosome binding site is located adjacent to each
`gene. In addition, the DNA sequence from -48
`to the pelB:antibody gene junction is a direct
`repeat preceding the Fd and K genes. The 28-bp
`trpA transcriptional terminator (Pharmacia) was cloned at the end of the transcription unit (Tm).
`Restriction endonuclease sites are as indicated: P, Pst I; Bg, Bgl II; Be, Bel I; and H, Hind III. The solid
`area denotes the approximate location of the pelB leader sequence.
`
`3' TGA GTG TGT ACT AGT GGC AGG GG 5'
`
`Bell
`
`p
`
`Tm
`
`b
`ParaB
`
`-p
`
`Bg
`
`Be
`
`H
`
`,
`' ' '
`'\
`\
`
`'
`
`1042
`
`kD
`-94
`
`-68
`
`-43
`
`-30
`
`-20
`
`-14
`
`-94
`-sa
`-43
`-30
`
`-20
`
`-14
`
`b
`
`iii
`
`.... , .. - """"
`-... -,_.
`
`""
`~~
`
`Fig. 2. SDS-PAGE comparison of bacterially (B)
`and Sp2/0-produced (Sp) Fab to papain-pro(cid:173)
`duced mouse L6 and chimeric L6* Fab. Mouse
`L6 and chimeric L6 antibodies were digested with
`papain (18), and Fab was purified by S-Sepharose
`chromatography. Each protein was examined by
`SDS-PAGE on a 10% gel under nonreducing
`conditions (a) and on a 12% gel under reducing
`conditions (b).
`
`SCIENCE, VOL. 240
`
`BEQ 1027
`Page 4
`
`

`
`-o- Fab-E. coli
`+ Fab-Sp2/0
`+ Fab-L6'-papain
`+ Fab-L6-papain
`
`Fig. 3. Binding inhibition of bacteri(cid:173)
`al Fab. Bacterial Fab, Sp2/0 Fab, and
`proteolytically produced chimeric L6
`Fab (L6*) and mouse L6 Fab were
`used to inhibit FITC-labeled mouse
`L6 antibody binding to the surface of
`antigen-positive C3347 colon carci(cid:173)
`noma cells.
`
`80
`
`60
`
`40
`
`20
`
`~
`" 0
`:;:: :c :;:
`"' " '6
`" iii
`
`. 5
`
`0 +-~-+-~........,=---~r?.-~.,.-~~~,.....,.,,---~~~...,
`o-2
`o-1
`o
`101
`102
`1
`1
`10
`[Fa b) (llgiml)
`
`bodies to protease cleavage will be obviated;
`a consistent, homogeneous preparation can
`be produced. Of additional interest is the
`relative ease with which the Fab eDNA
`genes can be modified before expression in
`bacteria. For example, modifications of the
`primary structure of either the Fd or K chain
`(or both) that are useful for subsequent
`conjugation of imaging or therapeutic
`agents or fusion to other peptides (16) can
`be introduced by site-directed mutagenesis
`
`techniques. We found that E. roli can prop-
`erly assemble a functional two-protein unit
`with a complex pattern of intra- and inter-
`ffi .
`h . d"
`lfid
`link
`d th
`c am
`lSU
`e
`ages an
`at su Clent
`quantities of this material may be prepared ~
`for eventual use as a human diagnostic and
`.
`therapeutic reagent.
`
`REFERENCES AND NOTES
`
`1. M. A. Boss, J. H. Kenten, C. R. Wood, J. S.
`Emtage, Nucleic Acids Res. 12, 3791 (1984).
`2. S. Cabilly et al., Proc. Nat/. Acad. Sci. U.S.A. 81,
`
`Technical Comm·ents
`
`Carcinog~nic Risk Estimation
`
`In their widely publicized and popular(cid:173)
`ized article "Ranking possible carcinogenic
`hazard," Bruce N. Ames et a/.. (17 Apr.
`1987, p. 271) conclude that "analysis on the
`levels of synthetic pollutants in drinking
`water and of synthetic pesticide residues in
`foods suggests that this pollution is likely to
`be a minimal carcinogenic hazard relative to
`the background of natural carcinogens" and
`thus that the "high costs of regulation" of
`such environmental carcinogens are unwar(cid:173)
`ranted. These conclusions
`reflect both
`flawed science and public policy.
`Although Ames et a/.. challenge the valid(cid:173)
`ity of animal carcinogenicity data for quanti(cid:173)
`tative estimation of human risk, they never(cid:173)
`theless use such extrapolations, based on the
`percentage Human Exposure dose/Rodent
`Potency dose (HERP), for ranking carcino(cid:173)
`genic hazards. Apart from the fact that
`HERP rankings are based on average popu(cid:173)
`lation exposures excluding sensitive sub(cid:173)
`groups, such as pregnant women, the de(cid:173)
`rived potencies of Ames eta/.., doses induc(cid:173)
`ing tumors in half the tumor-free animals,
`are misleading. Potencies for "synthetic pol-
`
`lutants," such as trichloroethylene, are de(cid:173)
`rived from bioassays in which lowest doses
`are large fractions of the maximally tolerated
`dose (MTD), whereas potencies for more
`extensively studied "natural carcinogens,"
`such as afiatoxins, are generally derived from
`titrated doses, orders of magnitude below
`the MTD. Since dose-response curves are
`usually flattened near the MTD (1), poten(cid:173)
`cies derived from high-dose testing yield
`artificially low risk estimates; HERPs for
`"synthetic" carcinogens are thus substantial(cid:173)
`ly underestimated compared with many
`"natural carcinogens."
`Compounding this misconception, Ames
`et a/.. maintain that carcinogenic dose-re(cid:173)
`sponse curves rise more steeply than linear
`curves and that tumor incidences increase
`more rapidly than proportional to dose. At
`high doses, dose-response curves are usually
`less steep than linear curves (1), as also
`recognized elsewhere by Ames and his col(cid:173)
`leagues (2). Thus at MTD doses, large fur(cid:173)
`ther dose increases may induce only small
`increases in tumor incidence, perhaps re(cid:173)
`flecting competition between transforma-
`
`20 MAY 1988
`
`3273 (1984).
`3. 0. Zemel-Dreasen and A. Zamir, Gene 27, 315
`(1984).
`4. A. Y. Liu et al., Proc. Nat/. Acad. Sci. U.S.A. 84,
`3439 (1987).
`5. I. Hellstrom et al., Cancer Res. 46, 3917 (1984) ..
`6. W. Kramer et al., Nucleic Acids Res. 12, 9441
`(1984).
`7. S. P. Lei, H. C. Lin, S. S. Wang, J. Callaway, G.
`Wilcox,]. Bacteriol. 169, 4379 (1987) .
`8. S. Johnston et al., Gene 34, 137 (1985).
`9. N. Yanagida, T. Uozumi, T. Beppu, J. Bacteriol.
`166, 937 ( 1986).
`10. H. M. Hsiung, N. G. Mayne, G. W. Becker, Biotech-
`nology 4, 991 (1986).
`11. E. Haber, Annu. Rev. Med. 37, 249 (1986).
`12. J. R. Zalcberg,Am.J. Clin. Oncol. 8, 481 (1985).
`13. B. Delaloye et al.,J. Clin. Invest. 77, 301 (1986).
`14. J. F. Eary et al., in preparation; P. G. Abrams et al.,
`Second International Conference of Monoclonal Anti(cid:173)
`body Immunoconjugants for Cancer (University of
`California, San Diego Cancer Center, San Diego,
`1987), p. 12.
`15. R. L. Wail!, D. w. Parker, G. W. Philpott,]. Nucl.
`Med. 24, 316 (1983).
`16. E. s. Vitetta et al., Science 238, 1098 (1987).
`17. L. Masson and D. S. Ray, Nucleic Acids Res. 14,
`5693 (1986).
`18. A. Johnstone and R. Thorpe, Immunochemistry in
`Practice (Blackwell, Oxford, 1982), pp. 48-56.
`19. We th~ K. E. Hellstrom and c<;>·workers for help
`perfornung Fab assays, R. Kodon for the pelB gene
`cassette, R. Railbar and B. Haselbeck for technical
`assistance, and R. Wall and G. Wilcox for comments
`on the manuscript. This work was supported by
`lNGENE.
`
`13 January 1988; accepted 18 March 1988
`
`tion and cytotoxicity (3); linear extrapola(cid:173)
`tions from high-dose tests thus underesti(cid:173)
`mate low-dose risks.
`For Ames et a/.., the term "carcinogen"
`heterogeneously includes direct and indirect
`influences, including promoting and modi(cid:173)
`fying factors and mutagens. Caloric intake is
`considered "the most striking rodent carcin(cid:173)
`ogen." However, no correlations have been
`established between food intake and tumor
`incidence among animals eating ad libitum,
`despite wide variations in caloric intake and
`body weight (4), nor have correlations been
`established between obesity and most hu(cid:173)
`man cancers. In the statement by Ames et
`a/.., "at the MTD a high percentage of all
`chemicals might be classified as 'carcino(cid:173)
`gens'," toxicity and carcinogenicity are con(cid:173)
`fused. However, among some 150 industrial
`chemicals selected as likely carcinogens and
`tested neonatally at MTD levels, fewer than
`10% were carcinogenic (5). Many highly
`toxic chemicals are noncarcinogenic, and
`carcinogen doses in excess of the MTD often
`inhibit tumor yields. While Ames et a/..
`revive the discredited theory that chronic
`irritation causes cancer, most irritants are
`noncarcinogenic, and there is no correlation
`between nonspecific cell injury and carcino(cid:173)
`genic potency ( 6).
`Ames et a/.. classify ethanol as carcinogen(cid:173)
`ic, " [one of the two] largest identified causes
`
`TECHNICAL COMMENTS 1043
`
`BEQ 1027
`Page 5