[CANCER RESEARCH 56. 3055-3061. July 1, 1996]

Minibody: A Novel Engineered Anti-Carcinoembryonic Antigen Antibody

Fragment (Single-Chain Fv-CH3) Which Exhibits Rapid,High-Level Targeting of Xenografts1

Shi-zhen Hu, Louise Shively, Andrew Raubitschek, Mark Sherman, Lawrence E. Williams, Jeffrey Y. C. Wong,John E. Shively, and Anna M. Wu2

Department of Molecular Biology ¡S-:.H.. A. M. W.], Divisions of Biology [L S., M. S.¡and Immunology [J. E. S.], Beckman Research Institute oj the City of Hope, and

Department of Radioimmunotherapv ¡A.RJ, Divisions of Diagnostic Radiology ¡L.E. W.¡and Radiation Oncalogv ¡J.Y. C. W.], City of Hope National Medical Center, Duarte.CA 91010

ABSTRACT

A novel engineered antibody fragment (VL-VH-CH3, or "minibody")

with bivalent binding to carcinoembryonic antigen (CEA) was producedby genetic fusion of a T84.66 (anti-CEA) single-chain antibody (scFv) to

the human IgGl CH3 domain. Two designs for the connecting peptidewere evaluated. In the 184.66/212 LD inin¡body,a two-amino acid linker

(generated by fusion of restriction sites) was used to join VH and CH3. Inthe T84.66/212 Flex minibody, the human IgGl hinge plus an additional10 residues were used as the connecting peptide. Size exclusion chroma-

tography of purified minibodies demonstrated that both proteins hadassembled into A/,.80,000 dimers as expected. Furthermore, analysis bySDS-PAGE under nonreducing conditions was consistent with disulfidebond formation in the hinge of the TX4.66 Flex minibody. Purified mini-bodies retained high affinity for CEA (A\. 2 x IO9 M~') and demon

strated bivalent binding to antigen. Tumor targeting properties wereevaluated in vivo using athymic mice bearing LS174T human colon carcinoma xenografts. IUI-labeled T84.66 minibodies demonstrated rapid,

high tumor uptake, reaching 17% injected dose/gram (%ID/g) for the LDminibody and 33%ID/g for the Flex minibody at 6 h following injection.Radioiodinated minibody also cleared rapidly from the circulation, yielding high tumorblood uptake ratios: 44.5 at 24 h for the LD minibody and64.9 at 48 h for the Flex minibody. Rapid localization by the T84.66/212Flex minibody allowed imaging of xenografts at 4 and 19 h after administration.

INTRODUCTION

Monoclonal antibodies directed against tumor-associated antigens

are finding wide applications as a means for targeted delivery ofdiagnostic radioisotopes or therapeutic agents (such as toxins, drugs,or therapeutic radioisotopes) to tumor cells in patients (1). The use ofmurine monoclonal antibodies in the clinic has required addressingseveral challenges, including the immunogenicity and poor biodistribution and pharmacokinetic properties of intact antibodies. Enzymaticdigestion of immunoglobulins can be used to obtain fragments such asFab and F(ab')2, which demonstrate more rapid tumor targeting, better

penetration, and faster blood clearance than their intact counterparts(2, 3). Furthermore, the lack of an Fc region and short residence in thecirculation contribute to lower immunogenicity of antibody fragmentsin vivo.

Genetic engineering has provided a means for production of defined antibody fragments as well as novel proteins such as immuno-toxins or other fusion proteins. One useful strategy has been produc-

Received 1/11/96; accepted 5/1/96.The costs of publication of this article were defrayed in pan by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

' This work was supported by N1H Grants CA43904 and CA42329. A. R.. L. E. W..

J. Y. C. W., J. E. S., and A. M. W. are members of the City of Hope Cancer Center(CA33572).

2 To whom requests for reprints should be addressed, at Department of Molecular

Biology. Beckman Research Institute of the City of Hope. 1450 East Duarte Road. Duarte,CA 91010.

tion of scFvs,3 comprised of the variable regions of the

immunoglobulin heavy and light chain, covalently connected by apeptide linker. These small (Mr 25,000) proteins generally retainspecificity and affinity for antigen in a single polypeptide and canprovide a convenient building block for larger, antigen-specific mol

ecules. However, due to their small size, scFvs themselves havelimited utility as tumor-targeting agents in vivo, generally demonstrating very rapid clearance (with half-lives less than 5 h) from thecirculation in animal models (4-6). Several groups have reported

biodistribution studies in xenografted athymic mice using scFv reactive against a variety of tumor antigens, including TAG-72, thec-erbB-2 receptor, and CEA (5-8), in which specific tumor localiza

tion has been observed. Furthermore, excellent penetration by a CC49anti-TAG-72 scFv was demonstrated by autoradiography in compar

ison to intact antibody, as well as larger mass fragments such asF(ab')2 and Fab (9). However, the short persistence of scFvs in the

circulation limits the exposure of tumor cells to the scFvs, placinglimits on the level of uptake. As a result, tumor uptake by scFvs inanimal studies has generally been only 1-5 %ID/g (4-6, 8) as

opposed to intact antibodies that can localize in tumors at30-40%ID/g and have reached levels as high as 60-70%ID/g (3, 10).

To obtain optimal tumor targeting, biodistribution, and clearanceproperties, antibody fragments of intermediate molecular weight canbe designed to have the proper combinations of high tumor uptake andrapid clearance from normal tissues. For example, dimers of scFvshave been generated by connecting two scFv by an additional peptide(11) or by a disulfide bridge formation through COOH-terminal Cys

residues (6, 12, 13), which can be further stabilized using chemicalcross-linking (6, 14). Others have fused scFvs to protein domains

capable of multimerization, e.g., leucine zippers (15), amphipathichelices (16), streptavidin ( 17), or the Kconstant domain ( 18) to inducedimer formation. Certain scFv constructs spontaneously form nonco-

valent dimers, or diabodies (19), especially when shorter linkers areused (7, 20). Recent biodistribution studies by our group (7) andothers (6, 20) demonstrate that scFv dimers show improved targetingas compared to their monomers in several antigen-antibody tumor

targeting systems, with tumor uptakes reaching the range of5-15%ID/g. This is likely the result of a combination of factors,including a longer biological half-life due to higher molecular weight

and a gain in avidity by these bivalent molecules.Larger fragments have been evaluated in murine studies, including

dimeric and trimeric forms of chemically cross-linked B72.3 Fab (Mr

110,000 or A/r 160,000; Ref. 14). These demonstrated enhancedtargeting of LS174T xenografts as compared to the correspondingdimeric and trimeric cross-linked scFv (14). I25l-labeled trimeric Fab

reached tumor uptakes of 8%ID/g at 24 h and T:B uptake ratios of 4.4at 48 h, 7.3 at 72 h, and 72.6 at 168 h. The 90Y-labeled form localized

3 The abbreviations used are: scFv. single-chain Fv; CEA. carcinoembryonic antigen;

%ID/g, percent injected dose per gram; T;B, tumorblood uptake ratio; HPLC. high-performance liquid chromatography; AUC. area under the curve.

3055

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

equally well, and activity did not accumulate in kidney or bone (14).CH2 domain-deleted (Mr 166,000) and CH1-CK domain-deleted (Mr

133,000) antibodies have been produced (21, 22). A comparison of thebiological properties of these chimeric domain-deleted forms of CC49revealed that the biodistribution of the CH1-deleted form was quite

similar to that of intact chimeric CC49 (23). High tumor uptake wasobserved for the CH1-deleted antibody (57.7%ID/g at 72 h), but blood

clearance was prolonged. As a result, T:B ratios increased slowly (3.4at 24 h and 11.4 at 72 h), and overall performance was similar to intactantibody (23). In contrast, the CH2-deleted form demonstrated more

modest uptake levels in the tumor (18.1%ID/g at 12 h) but achievedhigher T:B ratios due to faster clearance from the circulation (23). Theauthors concluded that CH2-deleted constructs were more suitable for

rapid imaging applications.We previously evaluated the tumor targeting of monomeric (Afr

27,000) and dimeric (MT 55,000) forms of scFv derived from theanti-CEA T84.66 antibody in athymic mice bearing xenografts ofLS174T, a CEA-positive human colorectal carcinoma line (7). Dimers

demonstrated rapid, high tumor uptake (13.8%ID/g at 1 h after administration) and rapid blood clearance, leading to high tumonnormalorgan ratios at early times. Although these results were excellentcompared to other single-chain antibodies, the absolute magnitudes of

tumor uptakes were still low compared to intact antibodies, due to thevery short persistence of scFvs in the circulation. Based on theobserved effect of molecular weight on clearance, we elected todesign a slightly larger molecule, comprised of the anti-CEA scFv

fused to the human IgGl CH3 domain. Since interactions betweenCH3 domains in intact IgGl promote and stabilize dimerization (withan association constant on the order of 10'°-1012M~'; Ref. 24), the

scFv-CH3 fusion, which we deemed the "minibody," was expected to

assemble into a chimeric, bivalent protein of intermediate molecularweight (Mr 80,000). Furthermore, the minibody was designed as asingle-chain protein to facilitate expression in bacterial and mamma

lian systems.

MATERIALS AND METHODS

Design of Minibodies. The high-affinity T84.66/212 scFv was joined to

the CH3 domain using two different peptide linkers. In the T84.66/212 LDminibody, a simple two-amino acid residue linker was used; in the second

design, T84.66/2I2 Flex, the human IgGl hinge region (including two potential disulfide bridges) and 10 additional amino acids, were used lo join the scFvwith CH3 domains (Fig. 1). Structures were displayed using the Insight II suiteof molecular modeling programs (BIOSYM, San Diego, CA), running on aSilicon Graphics IRIS workstation. The scFv was modeled based on the Fvregion of the KOL myeloma protein (25), and the human IgGl CH3 structurewas from Diesenhofer (26); coordinates were retrieved from the BrookhavenProtein Databank.

Plasmids and Gene Assembly. The T84.66/2I2 scFv construct, whichincluded an Xlw\ restriction site (encoding Leu-Glu) at the end of VH just

before the tandem stop codons, has been described (7). A plasmid containingthe human IgGl heavy chain constant regions was the gift of Dr. J. Schlom(National Cancer Institute), and pCIS2 (27) was supplied by Dr. C. Gorman(Genentech).

PCR methods were used to add the peptide linkers shown in Fig. 1 to theNH2 terminus of the human IgGl CH3 domain. To create the LD minibody, aSal\ restriction site (encoding Val-Asp) was added to the NH; terminus of CH3.

The scFv and CH3 domains were assembled by ligating the cohesive endsgenerated by Xho\ and Sail, respectively, and resulted in the two-amino acidlinker Leu-Asp (LD) at the junction. For the Flex minibody, PCR was used toadd the linker beginning with Leu-Glu (an Xhol site) and including the hingeand Gly-Ser sequence to the NH2 terminus of CH3. The resulting hinge-CH3was assembled with the anti-CEA scFv via the Xho\ restriction sites. Formammalian expression, an Xbal-Kpnl fragment containing the native mammalian signal sequence and NH,-terminal region of T84.66 VL (from the

LD minibody

LD

Flex minibody

LEPKSCDKTHTCPPCGGGSSGGGSG

Fig. 1. Schematic showing arrangement of VH. VL. and C,,3 domains in dimericminibodies; below each diagram is the sequence of the peptide linker used in eachconstruct. Potential disulfide bridges in the Flex minibody are also indicated. The smallblack dots indicate the approximate points of attachment of the scFv to the CH3 orhinge-CH3: Asp (the second residue in the LD linker); or Glu (the second residue in theFlex linker). The larger circles indicate the approximate location of His-Asn at residues433-434 in the CH3 domains (EU numbering system), implicated in catabolism of IgGl(see "Discussion").

original cDNA clone) was ligated to a Af/;nI-//indIIl fragment containing theremainder of V[, Virlinker-CM3, and the constructs were cloned into pClS2.

Expression and Purification. Sp2/0 murine myeloma cells were cotrans-fected by electroporation of 2 x IO6 cells with 15 fig of the vector pCIS2

carrying the LD or Flex minibody construct (linearized by digestion with Sail)and 15 fig of pSV2neo (linearized using £a>RI). G418-resistant coloniesscoring positive for anti-CEA activity were expanded to T-flasks, and culturesupernatants were collected. Minibodies were purified by dilution of superna-tants 2.5- to 3-fold in 30 nui Tris-HCl (pH 8.2) and application of the sampleto a 25-ml bed volume Fast Flow Q (Pharmacia) ion-exchange column (previously equilibrated with 30 mM Tris-HCl, pH 8.0). CEA-binding activity,

determined by a competition assay (see below) was eluted using 30 mMTris-HCl (pH 8.0) and 200 mM NaCI. Fractions showing high activity werefurther purified using an anti-idiotype affinity column as described previously

(7). Purified proteins were concentrated using Centriprep 10 and Centriconconcentrators (Amicon). The LD minibody was also expressed in Escherichiacoli as described previously for the T84.66 scFv (7). except that activity wasrecovered in the periplasm.

Characterization of Purified Minibodies. Purified proteins were analyzed by SDS-PAGE (28) under nonreducing or reducing (1 mM DTT) con

ditions. The concentration of purified protein samples was determined byamino acid composition analysis as described (7). Size exclusion HPLC wasperformed on two different systems; using tandem Superóse 6 or Superóse12HR columns (Pharmacia) as described previously (7) or using a Spectra-

Physics HPLC equipped with a TSK 2000 column.CEA-binding activity was determined using a competition ELISA in which

microtiter plate wells were coated with purified CEA (29); the sample wasadded, and the parental murine T84.66 antibody was added as competitor.Chimeric T84.66 (10) was used as the standard. ELISA results were thusexpressed as microgram equivalents of whole chimeric immunoglobulin. Affinities were determined by surface plasmon resonance analysis of the interaction of purified minibodies with immobilized CEA, using a PharmaciaBIAcore instrument as described previously (7). Coupling of CEA to biosensorchips was conducted at 40 /j.g/ml in 10 mM sodium acetate (pH 4.0) to densitiesof 1000-1300 resonance units. Association kinetics were determined by measuring the interaction of purified minibodies at concentrations from 30-500 nM

at a flow rate of 5 fil/min. Dissociation kinetics were determined at a higherflow rate (20 /xl/min) to minimize rebinding of dissociated minibody to thechip. Data were analyzed using the Kinetics Evaluation software packagesupplied by Pharmacia Biosensor.

Radioiodination. Fifty fig of purified 212 LD or 212 Flex minibody werelabeled with 1-2 mCi Na'21I (carrier-free; Nordion) using polypropylene tubes

coated with 10 jig lodogen (Pierce). Following a 2.5-3-min incubation at room

temperature, human serum albumin was added, and the sample was purified byHPLC on tandem Superóse 12 columns (Pharmacia). Immunoreactivity andvalency were determined by incubation of radiolabeled protein with a 20-foldmolar excess of CEA at 37°Cfor 15 min, followed by HPLC analysis on a

calibrated Superóse 6 column.

3056

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

t if« Q Q £^_ g- _l _l LL

Q •¿�§CM CM CM

.5? u. cj IS 15

* —¿�97

—¿�68

—¿�43

—¿�29

Non-reducing ReducingFig. 2. SDS-PAGE of affinity-purified T84.66/212 LD or Flex proteins. Left, nonre-

ducing conditions; right, reducing conditions. Size markers on the nonreducing gelinclude intact IgGl (M, 150,000). F(ab')2 (M, 110,000), and scFv (M, 27.000); molecular

weights (in thousands) of markers on the reducing gel are given at the right. B and Mindicate whether the protein was purified using bacterial or mammalian expressionsystems, respectively.

Biodistribution Studies in Tumor-bearing Mice. Groups of four or five7-8-week-old female athymic mice were injected s.c. in the flank with \0h

LS174T human colon carcinoma cells. When tumor sizes reached approximately 100-200 mg (6-7 days), 2-3 /¿Ciof '"l-labeled minibody (specificactivity, 15-25 /iCi//xg) was injected via the tail vein. At 0, 2, 6, 24. and 48 hafter injection, mice were euthanized and dissected; major organs wereweighed and counted in a gamma scintillation counter. Stability of the radio-labeled minibodies in vivo was evaluated by size-exclusion HPLC analysis ofserum taken from the mice at the time of sacrifice. Serum samples (100-200H-l)were fractionated on Superóse6, as described above, and radioactivity wasfollowed during elution.

Activities in liver, spleen, kidney, lung, stomach, bowel, carcass, tumor, andblood were determined and expressed as %ID/g. To quantitate the differencesin blood clearance times, the ADAPTII software package (30) was used toestimate two rate constants characteristic of each engineered fragment. Abi-exponential function was fitted, via the ID subroutine, to each bloodclearance curve (%ID/organ) after the latter had been corrected for the physicaldecay of the radiolabel. Significant differences in these values were examinedby comparing the 95% confidence intervals for the parameter estimates.

In estimating relative effectiveness of the constructs in a therapeutic application, radiation absorbed doses are proportional to %ID/g (31); thus, thetime-activity curves for tumor and blood uptake were integrated. For eachconstruct, this pair of curves was fitted by a respective sum of two exponentialfunctions, and integration was performed over the interval from zero to infinityusing ADAPTII (30) to give the pharmacokinetic or radionuclide-associatedAUC. Pharmacokinetic values refer to AUCs determined by first correctingblood and tumor uptake data for radiodecay. Such area values refer to ahypothetical situation whereby biodistribution experiments could be conductedwithout a radiolabel. These areas are larger than those observed because of thedecay factor.

Xenograft Imaging. Prior to conducting imaging studies, thyroid uptake ofradioiodine was blocked by pretreatment of the mice using 10 drops ofsaturated Kl/\00 ml drinking water for 24 h. Tumor-bearing mice wereinjected with 20-30 /¿Ciof ml-labeled minibody at the specific activities

given above. Immunoscintigraphy was conducted on lightly anesthetized miceusing a Toshiba 901/HG gamma camera equipped with a pinhole collimatorand interfaced to a Sun workstation for data storage and transmission.

Statistical Methods. To compare changes in %ID/g over time amongminibody constructs, two-way ANOVA was performed. Time (in hours),minibody group, and the interaction between these two factors were includedin the statistical model. Dependent variables compared using this modelincluded %ID/g for each organ, T:B ratios, and tumonliver ratios. If a significant interaction was detected, differences in mean %ID/g between minibodyconstructs at each time point were compared using the independent / test. Allsignificance testing was done at the P < 0.05 level. The SAS/STAT software(SAS, Inc., Cary, NC) was used.

RESULTS

Expression and Characterization of Minibodies. Initially, expression of the T84.66 Flex and LD minibodies was attempted in E.coli as described previously for the cognate scFv. However, activitylevels were 10-fold lower, and the Flex minibody proved susceptible

to proteolysis in the bacterial system. Mammalian expression of theminibodies using the nonsecreting Sp2/0 murine myeloma cell linewas investigated as an alternate method of production. Transfectionstudies yielded clones that secreted CEA-binding activity in the 5-10

fig/ml range for both the LD and Flex minibodies. Fractionation of theculture supernatants by ion-exchange and affinity chromatography

resulted in recovery of highly purified, active protein. In stabilitystudies, a sample of the Flex minibody retained 60% of its startingactivity after 3 months of storage in PBS at 4°C;a sample of the LD

minibody lost activity more rapidly, retaining only 7% after 2.5months of storage.

SDS-PAGE demonstrated that the two-step purification scheme

resulted in proteins that were greater than 95% pure (Fig. 2). Underreducing conditions, the T84.66/212 LD and Flex minibodiesshowed the expected molecular weights of Air 39,000 and MT41,000, respectively (Fig. 2, right). Following SDS gel electro-

phoresis under nonreducing conditions, the LD minibody (frombacterial or mammalian cultures) demonstrated migration expectedof the monomeric MT ~40,000 protein (Fig. 2, left). In contrast, the

T84.66/212 Flex minibody from mammalian cells migrated substantially more slowly, running slightly ahead of F(ab')2 (Mr

110,000), consistent with an Mr 80,000 covalently linked dimer(Fig. 2, left).

A kinetic analysis of the interaction of purified minibodies withimmobilized CEA by surface plasmon resonance allowed determination of the association and dissociation rate constants and calculationof the affinity constants. Slight differences were noted in the rateconstants, with the Flex minibody having faster on and off rates; butoverall, both minibodies demonstrated high affinity for CEA, withATAsof 2-3 X IO9 M~' (Table 1).

The immunoreactivity and valency of LD and Flex minibodies wereanalyzed following radioiodination by solution-phase incubation inthe presence of excess CEA. Size-exclusion HPLC analysis demonstrated that 90-95% of the radiolabeled LD (noncovalent dimers) and

Flex (covalent dimers) minibodies were shifted to high molecularweight complexes following incubation with CEA; very little unbound minibody was detected (Fig. 3). Furthermore, comparison of

Table 1 Kinetic analysis of minibody-CEA interactions

Interactions of purified minibodies (at 30-500 nMl with immobilized CEA werefollowed in real time by surface plasmon resonance as described in "Materials andMethods." Data were acquired and analyzed using the Kinetics Evaluation software

package provided by Pharmacia Biosensor. Molarities of the minibodies were calculatedbased on a M, 80,000 dimer.

^•assoc(/IIS) ^-diss (^) K^ (/M)

T84.66 LD minibodyT84.66 Flex minibody1

X IO53 X IO55

X IO"51.1 X IO"42

X IO92.7 X IO9

3057

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

the elution profiles with protein standards indicated that the two majorpeaks represented minibody bound to one or two CEA dimers(MT 360,000/dimer), respectively.

Murine Biodistribution Studies. Biodistribution and clearancestudies of I23I-radiolabeled LD and Flex minibodies were conducted

in xenograft-bearing athymic mice (Table 2). As can be seen in Table

3. these proteins showed significantly different clearance patterns.The T84.66 LD minibody had a significantly higher rapid kinetichalf-life (Ti/2a, 1.21 h) than the T84.66 Flex minibody (Tlria, 0.59 h);

the LD minibody also had a higher proportion of blood activity thatcleared with rapid kinetics (88% versus 57%). The minibodies showedno statistically significant difference in the slower kinetic half-lives(7",/2ß,4.8 and 5.3 h for LD and Flex, respectively).

Both minibodies demonstrated excellent tumor targeting, with theT84.66 Flex construct achieving higher tumor accumulation. Forexample, 6 h after administration, tumor uptake by the 123I-Flexminibody reached 32.9%ID/g, as opposed to the 123I-labeled LD

minibody value of 16.36%ID/g (Table 2). Activity was retained intumor, with values of 29.08%ID/g and 8.27%ID/g, respectively, at24 h after injection. Both minibody forms achieved high T:B uptakeratios, over 45:1 by 48 h after administration, as shown in Table 2.The LD minibody reached higher T:B ratios at an earlier times due toits faster blood clearance. For example, blood activity levels at 6 h

Table 3 Estimated values of blood half-limes for ¡heengineered fragments

212FI6X + CEA

212 LD + CEA

20 40 60 80 100

Retention Time (min)

Fig. 3. Size exclusion HPLC analysis of formation of minibody-CEA complexes.Radioiodinated engineered antibodies were incubated with excess CEA and analyzed bysize exclusion HPLC (see "Materials and Methods").

AntibodyfragmentT84.66/212LD

minibodyT84.66/212Flex

minibodyTma

(h)(95%CI)*1.21(1.15,

1.27)0.587(0.23,

0.94)SignificantAa°

(%ID)

(95%CI)30.19(28.25.

32.13)25.90(19.37,

32.43)NSrl/2p

(h)(95%CI)4.76(2.92,

6.60)5.27(3.60,

6.94)NSA„

(%ID)(95%CD4.088(2.15.6.03)

19.61(13.18,26.04)

Significant" The amplitudes of the two components are given by Aa and Aß.where the sum of Aa

and Aßis the total amount of activity in the blood compartment at the time of injection,expressed as %ID.

* 95% CI, 95% confidence interval of the estimate. Significant and NS. not significant,

indicate whether the values for the two forms differ significantly.

were 3.61%ID/g versus 16.55%ID/g, respectively, for LD and Flex,leading to T:B ratios of 4.46 versus 1.97. The difference in T:B ratiosincreased to 44.5 (LD) versus 14.21 (Flex) at 24 h, but by 48 h, T:Bratios for both minibodies were very high (Table 2).

Chemotherapy effects or radiation absorbed doses are proportionalto time integrals of the uptake values for various organs (31 ). Thus,estimation of the potential effectiveness of the minibody constructs intherapeutic applications requires integration of the time-activity

curves. Table 4 shows the calculated ratios of the AUCs for blood andtumor for unlabeled minibodies as well as if minibodies were radio-labeled with potential therapeutic radionuclides, such as I3II or '"Y.

As can be seen in Table 4, AUC(tumor):AUC(blood) ratios werehigher for the Flex minibody than for the LD minibody. Enhancementwas approximately a factor of 1.35 for the pharmacokinetic case.When decay was included, the AUC ratio was only better by factorsof 1.25 and 1.14 for 131I and ""Y labels, respectively, since the

advantage of high T:B ratios at later times was lost. Thus, for radio-immunotherapy with 9"Y, one would expect about a 14% improve

ment in tumor to marrow dose using the Flex minibody as comparedto the LD minibody.

Stability in Vivo. Concurrent with the biodistribution studies, serum samples taken from the mice at times from 0 to 12 h followinginjection of the T84.66 Flex minibody were analyzed by size exclusion HPLC. Fig. 4A demonstrates that the radioactivity in the serumwas exclusively associated with a Mr 80,000 species throughout thistime interval. There was no evidence of degradation of the protein, norwas there any indication of aggregation or association with serumproteins during the course of the experiments. Similar results wereseen for the noncovalent LD minibody at 0, 2, and 6 h, with a minorcomponent of low molecular weight (Fig. 4B).

Table 2 Biodtstribution of ~ l-labeled T84.66 minibodies in alhymic mice bearing LSI 74T xenograßs

Groups of four or five mice were analyzed at each time point. Tumor and normal organ uptake are expressed as %ID/g. SDs are given in parentheses. Statistically significantdifferences in the biodistribution curves (assessed by the interaction term) were found for all organs (P = 0.03 for bone; P < 0.002 for all other organs). Blood values were significantlydifferent at each time point (P < 0.05) except 48 h (P = 0.06. marginally significant) after injection. Tumor uptakes were significantly different at all time points (P < 0.05 except

0 and 2 h. noi significant).

Organ(%ID/g)TumorBloodLiverSpleenKidneyLungStomachIntestineBoneRatios'1T:BTumor:

LiverTumorTumor

weight01.66(0.07)44.77

(6.04)7.29(1.00)6.54

(0.83)9.13(1.81)15.17(0.91)1.53(0.53)1.60(0.46)3.05

(0.73)0.040.230.09

(0.02)214.89(4.26)18.51

(2.14)5.79(0.44)7.45

(0.86)6.27(0.75)8.26(0.63)23.74(4.17)2.62(0.32)2.95

(0.27)0.802.570.14(0.06)212

LD"616.36(6.13)3.61

(0.70)1.29(0.21)1.82(0.35)1

.49(0.30)2.33(0.35)10.57(1.87)1.25(0.28)0.90(0.15)4.4612.660.09

(0.04)248.27(2.41)0.19(0.07)0.16(0.06)0.09

(0.02)0.12(0.04)0.15(0.05)0.88

(0.62)0.15(0.05)0.05

(0.04)44.5051.330.21

(0.03)484.06

(0.54)0.09(0.02)0.15(0.04)0.00(0.00)0.07

(0.02)0.02(0.02)0.14(0.02)0.07

(0.02)0.00(0.00)46.9726.690.19(0.05)02.32(1.11)52.55

(3.05)9.43(0.57)1

1.20(0.74)12.65(1.22)19.81

(2.89)1.48(0.48)1.44(0.28)3.94

(0.66)0.040.250.17(0.08)218.64(4.54)27.82

(3.06)6.01(1.39)7.24(2.71)7.85

(0.97)12.27(1.90)11.39(1.87)2.91

(0.35)3.36(0.88)0.693.100.16(0.06)212

Flex"632.90(11.18)16.55(4.16)3.81

(0.65)3.90(0.51)4.97

( 1.47)8.29(1.59)10.54(3.71)2.42

(0.58)2.44(0.76)1.978.640.21

(0.07)2429.08

(7.32)2.14(0.30)0.78

(0.23)0.99(0.32)0.79(0.17)1.37(0.11)6.47

(2.43)0.62(0.16)0.45

(0.08)14.2137.140.15(0.04)4811.10(2.77)0.18(0.07)0.17(0.08)0.17(0.08)0.11

(0.04)0.17(0.07)0.45

(0.20)0.07(0.02)0.00

(0.05)64.8963.420.37(0.18)

" Time given in hours; SD in parentheses.h The ratios presented are the averages of the T:B and tumorliver ratios for the individual mice.

3058

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

Table 4 AUC analysis of minibodies: ratio of AUC(Tumor):AUC(Blood)"

Antibody fragment Pharmacokinetic «»Y

T84.66/212 LD minibodyT84.66/212 Flex minibody

4.886.62

4.505.65

3.814.37

"Calculations are described in "Materials and Methods."

Radioimmunoimaging Using Minibodies. Use of ' ''I as the ra-

diolabel allowed imaging of tumor-bearing mice using a conventional

nuclear medicine camera, due to the 159 keV pure y emission of thisradionuclide. In addition, the 13.2-h half-life of 123Iis well matched

to the rapid distribution kinetics exhibited by the minibodies. Threeathymic mice bearing LS 174T xenografts of 1.2 , 0.6, or 0.15 g wereinjected with 2-3 f/.Ci of I23l-radiolabeled T84.66 Flex minibody and

were imaged immediately and at 4 and 19 h after injection. Fig. 5demonstrates localization of activity in all three tumors as early as 4 hafter injection. By 19 h, activity in normal tissues had cleared almostto background, allowing distinct images of the tumors to be obtained(Fig. 5).

DISCUSSION

The minibody (VL-VH-CH3) resembles a F(ab')2 antibody fragment

in several ways, including its intermediate size (Mr 80,000 versus M,110,000 for F(ab')2) and ability to bind two antigen molecules. The

LD and Flex minibodies were expected to exist as stable dimers dueto the extensive interactions between the human CH3 domains. In theFlex version of minibody, the human IgGl hinge region was alsoincorporated into the molecule, anticipating that formation of disulfidebridges between the subunits would further stabilize the dimer. Thus,the single-chain design and self-assembly into homodimers provideda simple approach for generating a F(ab')2-like molecule, obviating

the need for enzymatic digestion of intact antibodies and repurifica-

tion.Noncovalent LD and covalent Flex minibody dimers were obtained

in high yield through mammalian expression, and the purified proteinsretained high affinity and were capable of bivalent binding to CEA.Furthermore, in the T84.66 Flex minibody, formation of disulfidebridges between the subunits could be demonstrated. Thus, the myeloma system proved superior to bacterial expression for producingfunctional bivalent engineered fragments.

When biodistribution and tumor targeting characteristics of mini-

bodies were evaluated in the athymic mouse/LSI74T xenograft system, the two versions demonstrated different blood clearance properties. The bulk of the T84.66 LD minibody cleared with rapid kinetics,whereas the T84.66 Flex minibody was more evenly split betweenrapid- and slow-clearing components. The longer overall residence

time in the circulation probably contributed to the higher tumoruptakes observed for the Flex minibody, which reached levels as highas 32.9% ID/g at 6 h after administration. Clearance properties of theminibodies demonstrated a balance between being slow enough toallow sufficient exposure to the tumor for specific localization andrapid enough that background activities in normal tissues rapidlydeclined, resulting in excellent tumonnormal organ ratios. For example, the T84.66/212 Flex minibody achieved a T:B ratio of approximately 65 by 48 h following injection; the comparable LD minibodyT:B ratio was 47.

These results represent a substantial improvement over publishedresults on tumor targeting of engineered antibody fragments, including lower molecular weight species (scFv, monomers, and dimers) aswell as larger fragments (such as CH1-CK- or CH2-deleted antibodies).

Our previous studies on targeting by monomeric and noncovalentdimeric forms of T84.66 scFv used the same experimental system and

LS174T xenografts of similar size (100-200 mg), allowing directcomparison with the present study (7). The Mr 55,000 divalent (dia-body) form of 125I-radiolabeled T84.66/212 scFv reached maximum

tumor uptake of 13.2%ID/g at l h following administration (7),whereas the highest tumor uptake reached by the Flex minibodyconstruct described herein was 32.9%ID/g at 6 h (Table 2). Tumoruptake levels at 24 h were significantly higher for the T84.66 Flexminibody (29.08 ± 7.32%ID/g) than for the corresponding scFvdimer (3.89 ±0.63%ID/g; Ref. 7). Those scFv dimers and the presentminibodies used the identical antigen-binding domains and had es

sentially the same affinities for antigen; furthermore, both boundantigen bivalently. A major difference between the two forms was thehigher molecular weight of the minibody (MT80,000) as compared tothe diabody (Mr 55,000 ), which contributed to a longer residence timein the circulation and, thus, greater exposure to tumor.

Tumor targeting by the Flex minibody also represented an improvement over other engineered antibody fragments, such as CH2 domain-deleted or CH1-C „¿�-deletedCC49 (anti-TAG72) antibodies in a similar

LS174T model (23), although experiments may not be directly comparable due to differences in the nature of the target antigen, antigenshedding, antibody affinity, and other factors. Nevertheless, the anti-CEA Flex minibody reaches 2-3-fold higher tumor uptake levelswhile maintaining equivalent T:B ratios with the CH2 domain-deleted

CC49 at the 12, 24, and 48 h time points (Table 2; Ref. 23). Additionally, when compared to CHl-CK-deleted CC49, although maxi

mum tumor uptakes by the Flex minibody were lower (29.08%ID/gversus 42.0%ID/g at 24 h; 11.0%ID/g versus 44.8%ID/g at 48 h, forT84.66 and CC49 fragments respectively), the T:B ratios were muchhigher (14.21 versus 3.4 at 24 h; 64.89 versus 6.3 at 48 h, respectively;Table 2; Ref. 23). A comparison with a recent biodistribution study onradioiodinated, chemically cross-linked dimers and trimers of B72.3

Fab (14), also in LS174T xenografted mice, showed that the T84.66Flex minibody reached 4-8-fold higher tumor uptakes and over8-fold higher T:B uptake ratios at the 24-h time point, compared to the

Fab dimers or trimers (Table 2; Ref. 14). The Flex minibody performed as well as or better than conventional F(ab')2 fragments of

CC49 (5) and the anti-CEA antibody A5B7 (3), again in LS174T-

xenografted nude mice. The T84.66 Flex minibody demonstratedhigher tumor uptake and similar T:B ratios compared with CC49F(ab')2 and substantially higher tumor uptakes than the A5B7 F(ab')-.

(3, 5).

B.T84.66LD

>.">

oO

O 20 40 60 80 100 120 140 160

Retention time (min)

Fig. 4. Stability of radioiodinated T84.66 Flex and LD minibodies in vivo. Serumsamples were obtained from mice at the time of sacrifice in the biodistribution studies.Samples were chromatographed on a Superóse 6 column equipped with a radioactivitymonitor. Results of analysis of the T84.66/212 minibodies are shown for the time pointsindicated. The major peak eluting at approximately 68 min corresponds to a molecularweight of Mt 80,000.

3059

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

Fig. 5. Radioimmunoscintigraphy of threeathymic mice bearing xenografts of LS174T. Tumor-bearing mice were injected with 2-3 ¿iCiof'"l-labeled T84.66 Rex minibody, and images

were acquired immediately and at 4 and 19 h later.Tumor sizes were: Mouse 1. 1.2 g; Mouse 2, 0.6 g;and Mouse 3, 0.15 g. A small amount of residualactivity can be seen in the stomach.

0 hours 4 hours 19 hours

Based on the data from Tables 2 and 3, tumor targeting andblood clearance of the two minibodies (LD and Flex) were observed to be significantly different. Because of their similar behavior in plasma (Fig. 4), these variations cannot be attributed todifferences in stability in vivo. Nor are the slight differences inmolecular weight or affinity likely to be the definitive causes of thedisparities in blood clearance and tumor retention properties. Apossibly important factor would be the differences in spacing androtational freedom of the scFv units. From molecular modelingstudies, in the LD minibody the scFv units were found to be joinedto the two CH3 domains at sites approximately 40Àapart (indicated schematically by black dots in Fig. 1), whereas in the Flex

minibody, disulfide bond formation in the hinge limited the separation of the scFvs to approximately 23Â (Fig. 1). Thus, the twominibodies probably had slightly different shapes and properties insolution, although this was not examined experimentally. Alternately, as can be seen in Fig. 1, the scFv domains in the LDminibody could have been more likely to mask key residues on theCH3 domain (Fig. 1, larger circles) that are known to affectcatabolism of murine and human IgGl (32, 33). Accessibility ofthese sites in the Flex construct could have contributed to longerserum persistence. Finally, the presence or absence of disulfidebridges and/or sulfhydryl groups might also have influenced thepersistence in the circulation or uptake by peripheral tissues by one

3060

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

RAPID TARGETING BY ENGINEERED ANTI-CEA ANTIBODY FRAGMENTS

version or the other. These possibilities can be further tested bysite-directed mutagenesis of the hinge and CH3 domains of the

minibodies.In summary, a simple engineered antibody fragment has been

designed, produced, and demonstrated to have excellent tumor targeting and biodistribution properties in xenografted nude mice. In particular, rapid, high, and persistent tumor uptake of the T84.66 Flexminibody, coupled with moderately rapid clearance from the circulation, leads to high absolute levels of activity in LS174T xenografts aswell as excellent tumor: normal organ ratios. These conditions are metshortly after infusion of I21l-radiolabeled minibodies, allowing image

acquisition using conventional gamma cameras at early times (4-19

h). Furthermore, these engineered fragments are likely to demonstratereduced immunogenicity if administered in a clinical setting, due tothe use of the human IgGl CH3 domain, and short exposure to theimmune system while in the circulation. As a result, 123I-labeled

anti-CEA minibodies are an excellent candidate for clinical radioim-

munoimaging trials.

ACKNOWLEDGMENTS

We are grateful to Dr. Arthur Riggs tor his ongoing support of this work.Cheryl Clark. Millie Martinez. Tulan Do. and Ming Liu provided experttechnical support. Dr. Yijae Kao and Randall Woo of the City of HopeRadiopharmacy performed the radioiodinations. Dr. Tamara Odom-Maryon

and Ken Clarke compiled the statistical analysis. We (hank Karla Carlson forassistance with the tables and figures.

REFERENCES

1. Waldmann. T. A. Monoclonal antibodies in diagnosis and therapy. Science (Washington DC). 252: 1657-1662. 1991.

2. Buchegger. F.. Haskell. C. M. Schreyer. M.. Scazziga. B. R.. Randin. S., Carrel, S.,and Mach, J. P. Radiolabelcd fragments of monoclonal antibodies against carcino-embryonic antigen for localization of human colon carcinoma grafted into nude mice.J. Exp. Med.. 158: 413-427. 1983.

3. Vogel. C-A.. Bischof-Delaloye. A.. Mach. J-P.. Peflegrin. A.. Hardman. N.. Delaloye.B., and Buchegger. F. Direct comparison of a radioiodinated intact chimeric anti-CEAMAb with its F(ab'): fragment in nude mice bearing different human colon cancer

xenografts. Br. J. Cancer. 68: 684-690. 1993.

4. Colcher. D.. Bird. R.. Roselli. M.. Hardman. K. D.. Johnson. S.. Pope. S.. Dodd.S. W., Pantoliano. M. W.. Milenic. D. E.. and Schlom. J. In vivo tumor targeting ofa recombinant single-chain antigen-binding protein. J. Nati. Cancer Inst.. H2: 1191-

1197, 1990.5. Milenic. D. E.. Yokola. T.. Filpula, D. R.. Finkelman. M. A. J.. Dodd. S. W., Wood,

J. F.. Whitlow, M.. Snoy. P.. and Schlom. J. Construction, binding properties,metabolism, and tumor targeting of a single-chain Fv derived from the pancarcinomamonoclonal aniibody CC49. Cancer Res.. 51: 6363-6371. 1991.

6. Adams. G. P.. McCartney, J. E.. Tai. M-S.. Oppermann. H.. Huston. J. S.. Stafford.W. F.. Bookman. M. A.. Fand. I.. Houston, L. L.. and Weiner. L. W. Highly specificm vivo tumor targeting by monovalent and divalent forms of 741F8 anti-c-erbB-2single-chain Fv. Cancer Res., 53: 4026-4034. 1993.

7. Wu. A. M., Chen, W.. Raubitschek, A. A.. Williams. L. E.. Fischer. R.. Hu, S-z.,Odom-Maryon. T., Wong. J. Y. C., and Shively. J. E. Tumor localization of anti-CEAsingle chain Fvs: improved targeting by non-covalent dimers. Immunotechnology. 2:21-36. 1996.

8. Berhaar. J. J.. Chester. K. A.. Keep. P. A.. Robson. L.. Pedley. R. B.. Boden. J. A..Hawkins, R. E.. and Begent, R. H. J. A single chain Fv derived from a filamentousphage library has distinct tumor targeting advantages over one derived from ahybridoma. Int. J. Cancer. 61: 497-501. 1995.

9. Yokota, T., Milenic. D. F... Whitlow. M.. and Schlom. J. Rapid tumor penetration ofa single-chain Fv and comparison with other immunoglohulin forms. Cancer Res.. 52:3402-3408. 1992.

10. Williams. L. E., Primus. F. J.. Wong. J. Y. C.. Wu. A. M.. Odom-Maryon. T.,

Johnson. D. K.. Hefta. L. J. F.. Shively. J. E.. and Raubitschek. A. A. Biodislributionof an In-Ill- or Y-90-labeled chimeric anti-CEA monoclonal antibody (C-TX4.66)following its large-scale production in a bioreactor. Tumor Targeting, in press. 1996.

11. Mallender. W. D.. and Voss, E. W., Jr. Construction, expression, and activity of abivalent bispecific single-chain aniibody. J. Biol. Chem.. 269: 199-206. 1994.

12. Cumber. A. J.. Ward. E. S.. Winter. G.. Parnell, G. D.. and Wawrzynczak. E. J. Comparative stabilities in vitro and in vivo of a recombinant mouse antibody FvCysfragment and a bisFvCys conjugate. J. Immunol.. 14V: 120-126. 1992.

13. Kipriyanov. S. M.. Dübel.S.. Breitling. F.. Konterinann. R. E.. and Little. M.Recombinant single-chain Fv fragments carrying C-terminal cysteine residues: production of bivalent and biotinylated miniantibodies. Mol. Immunol.. 31: 1047-1058.1994.

14. King. D. J.. Turner. A.. Farnsworth, A. P. H.. Adair. J. R.. Owens. R. J.. Pedley. R. B..Baldock. D.. Proudfoot. K. A.. Lawson. A. D. G.. Beeley. N. R. A.. Millar. K..Millican. T. A.. Boyce. B. A.. Antoniw. P., Mountain. A.. Begent. R. H. J.. Shochat.D.. and Yarranton. G. T. Improved lurnor targeting with chemically cross-linkedrecombinanl antibody fragments. Cancer Res., 54: 6176-6185, 1994.

15. Kostelny, S. A.. Cole. M. S.. and Tso. J. Y. Formation of a hispecific antibody by theuse of leucine zippers. J. Immunol.. 148: 1547-1553. 1992.

16. Pack. P.. and Pluckthun. A. Miniantibodies: use of amphipathic helices to producefunctional, flexibly linked dimeric Fv fragments with high avidity in Escherichia coli.Biochemistry. 31: 1579-1584. 1992.

17. Dübel,S.. Breitling. F.. Kontcrmann. R.. Schmidt. T.. Skerra. A., and Little. M.Bifunctional and multimene complexes of streptavidin fused lo single chain antibodies (scFv). J. Immunol. Methods. 17«:201-209. 1995.

18. McGregor. D. P.. Molloy. P. E.. Cunningham. C.. and Harris. W. J. Spontaneousassembly of bivalent single chain antibody fragments in Escherichia coli. Mol.Immunol.. 31: 219-226, 1994.

19. Holliger. P.. Prospero. T.. and Winter. G. "Diahodies": small bivalent and bispecific

antibody fragments. Proc. Nati. Acad. Sci. USA, 90: 6444-6448. 1993.20. Whitlow. M.. Bell. B. A.. Feng, S-L.. Filpula, D.. Hardman. K. D.. Hubert. S. L..

Rollence, M. L.. Wood. J. F.. Scholl. M. E.. Milenic. D. E.. Yokota. T.. and Schlom.J. An improved linker for single-chain Fv with reduced aggregation and enhancedproteolytic stability. Protein Eng., fi: 989-995. 1993.

21. Mueller, B. M., Reisfeld, R. A., and Gillies. S. D. Serum half-life and tumorlocalization of a chimeric antibody deleted of the C,,2 domain and directed against thedisialoganglioside GD2. Proc. Nati. Acad. Sci. USA. «7:5702-5705, 1990.

22. Shu. L.. Qi. C-F.. Schlom. J.. and Kashmiri. S. V. S. Secretion of a single-gene-

encoded immunoglobulin from myeloma cells. Proc. Nati. Acad. Sci. USA, 90:7995-7999, 1993.

23. Slavin-Chiorini. D. C.. Kashmiri, S. V. S., Schlom. J., Calvo. B.. Shu. L. M.. Schott,

M. E.. Milenic, D. E.. Snoy, P.. Carrasquillo. J., Anderson. K., and Horan Hand. P.Biological properties of chimeric domain-deleted anticarcinoma immunoglobulins.Cancer Res.. 55 (Suppl.): 5957s-5967s. 1995.

24. Schiffer. M.. Chang. C-H.. Naik. V. M.. and Stevens. F. J. Analysis of immunoglob

ulin domain interactions. Evidence for a dominant role of salt bridges. J. Mol. Biol..203: 799-802. 1988.

25. Marquât.M.. Diesenhofer. J.. Huber. R., and Palm, W. Cryslallographic refinementand atomic models of the intact immunoglobulin molecule Kol and its antigen-binding fragment at 3.0 A and 1.9 A resolution. J. Mol. Biol.. 141: 369-391. 1980.

26. Diesenhofer. J. Crystallographic refinement and atomic models of a human Fcfragment and its complex with fragment B of protein A from Staphylococcus aureusat 2.9- and 2.8-À resolution. Biochemistry. 20: 2361-2370. 1981.

27. Gorman, C. M.. Gies. D. R.. and McCray. G. Transient production of proteins usingan adenovirus transformed cell line. DNA Prot. Eng. Tech.. 2: 3-10. 1990.

28. Laemmli. U. K. Cleavage of structural proteins during assembly of the head ofbacteriophage T4. Nature (Lond.). 227: 680-685. 1970.

29. Coligan. J. E.. Lautenschleger. J. T., Egan, M. S., and Todd. C. W. Isolation andcharacterization of carcinoemhryonic antigen. Immunochemistry. 9: 377-386. 1972.

30. D'Argcnio, D. Z.. and Schumilzky, A. A program package for simulation and

parameter estimation in pharmacokinetic systems. Compul. Prog. Biomed.. 9: 115-

134. 1979.31. Siegel. J. A., Wessels, B. W., Watson. E. E., Stabin, M. G.. Vriesendorp. H. M.,

Bradley. E W.. Badger. C. C.. Brill. A. B.. Kwok. C. S.. Stickney. D. R.. Eckerman.K. F., Fisher. D. R.. Buchsbaum. D. J.. and Order. S. E. Bone marrow dosimelrv andtoxicity for radioimmunotherapy. Antibodies Immunoconj. Radioimmunopharma-ceut.. 3: 213-233. 1990.

32. Kim. J-K. Tsen. M-F.. Ghclie. V., and Ward. E. S. Identifying amino acid residuesthat influence plasma clearance of murine IgGl fragments by site-directed rnulagen-esis. Eur. J. Immunol.. 24: 542-548. 1994.

33. Kim. J-K.. Tsen. F-F.. Ghetie. V.. and Ward. E. S. Catabolism of the murine IgGamolecule: evidence that both CH2-CH3 domain interfaces are required for persistenceof IgGl in the circulation of mice. Scand. J. Immunol.. 40: 457-465. 1994.

3061

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from

1996;56:3055-3061. Cancer Res   Shi-zhen Hu, Louise Shively, Andrew Raubitschek, et al.   Rapid, High-Level Targeting of Xenografts

3) Which ExhibitsHAntibody Fragment (Single-Chain Fv-CMinibody: A Novel Engineered Anti-Carcinoembryonic Antigen

  Updated version

  http://cancerres.aacrjournals.org/content/56/13/3055

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/56/13/3055To request permission to re-use all or part of this article, use this link

Research. on January 14, 2020. © 1996 American Association for Cancercancerres.aacrjournals.org Downloaded from