A Single-Chain Immunotoxin against Carcinoembryonic Antigen · characterize a single-chain...

Vol. 4, 2825-2832, November /998 Clinical Cancer Research 2825

A Single-Chain Immunotoxin against Carcinoembryonic Antigen

That Suppresses Growth of Colorectal Carcinoma Cells’

Y. Akamatsu, J. C. Murphy, K. F. Nolan,

P. Thomas, R. J. Kreitman, 5-0. Leung, and

R. P. Junghans2

Biotherapeutics Development Lab, Division of Hematology-Oncology

[Y. A.. J. C. M., K. F. N., R. P. J.] and Department of Surgery [P. T.].

Harvard Medical School, Beth Israel Deaconess Medical Center,Boston, Massachusetts 02215 [P. T.J; Laboratory of Molecular

Biology, National Cancer Institute, National Institutes of Health,Bethesda. Maryland 20892 [R. J. K.]; and Immunomedics, Inc.,Moms Plain, New Jersey 07950 [5-0. L.]

ABSTRACTWe have engineered an anti-carcinoembryonic antigen

(CEA) single-chain immunotoxin derived from humanizedanti-CEA antibody (hMN14) and a truncated Pseudomonas

exotoxin (PE), PE4O. The purified anti-CEA immunotoxin(hMN14(Fv)-PE4O) was first measured for binding affinity

against a CEA-positive colorectal carcinoma cell line and

compared with its parental IgG and the monovalent Fabfragment. The Ka of sFv-PE4O, Fab, and IgG were 5 x i0�,6 x i0�, and 3 x i0� M1, respectively. There was no signif-icant affinity loss by conversion of Fab to the single-chain

Fv, but these monovalent forms were 5-6-fold reduced inaffinity compared with the parental IgG. In cytotoxicity

assays, the hMN14(Fv)-PE4O showed specific growth sup-pression of CEA-expressing colon cancer cell lines MIP-CEA (high CEA) and LS174T (moderate CEA) with IC5�s of

12 ng/ml (0.2 nM) and 69 nglml (1.1 nM). These IC5�s corre-lated inversely with the surface expression of CEA, such that50% killing was equivalent for each cell type when ex-pressed in toxin molecules bound/cell (3000-5000). The

presence of soluble CEA up to 1000 nglml did not affect the

cytotoxicity against CEA-expressing cells, with 50% sup-

pression only at 4000 ng/ml that correlated with the bindingKd of the single-chain Fv. The stability of the hMN14(Fv)-PE4O molecule at 37#{176}Cwas confirmed by bioassay and bylack of aggregation. Our hMN14(Fv)-PE4O may be clinicallyuseful for tumors with high CEA expression without affect-

Received 3/23/98; revised 8/17/98; accepted 8/26/98.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to

indicate this fact.

1 Supported by grants to R. P. J. from the Surgery Department of theformer New England Deaconess Hospital, the Skin Cancer Foundation,the American Cancer Society, and the National Cancer Institute, and bya Clinical Oncology Career Development Award to R. P. J. from theAmerican Cancer Society.2 To whom requests for reprints should be addressed, at Harvard Insti-tutes of Medicine, Room 403, Beth Israel Deaconess Medical Center,Boston, MA 02215. Phone: (617) 432-7004: Fax: (617) 432-7007:E-mail: [email protected].

ing normal tissues with low or absent CEA, even in patients

with high soluble antigen levels.

INTRODUCTION

CEA3 is a phosphoinositol-linked Mr 180,000220,000

glycoprotein expressed on a broad range of adenocarcinomas.

As an antitumor immunotherapy target, it has particular advan-

tages in terms of tissue expression and specificity, with high

expression on tumor cells and a combination of low expression

and a protected geometry in its normal tissue distribution ( 1, 2).

Although cohorectal carcinoma has been the prototypical malig-

nancy for testing anti-CEA therapies, based on its expression in

60-94% of patients with advanced disease, CEA is also cx-

pressed on tumors of #{176}-�60%of women with metastatic breast

cancer and >30% of patients with cancer of the lung. liver,

pancreas, head and neck, bladder, cervix, and prostate (1 , 3).

Approximately 150,000 people die each year from CEA-posi-

tive cancers, with an additional 50,000 eligible for adjuvant

therapies who are at high risk for recurrence after initial removal

of all macroscopic disease. A new therapeutic option that effec-

lively targets this antigen would have a very high clinical

relevance, with potential for major impact on the clinical and

financial consequences of cancer in this country.

Monoclonal antibodies specific to CEA have been studied

for diagnosis and therapy of CEA-positive human cancers. 5ev-

eral chemical immunoconjugates of anti-CEA whole IgG have

also been examined, and their specific cytotoxicities have been

shown (4-8). However, chemical conjugation methods can

modify antibody with adverse effects on antigen binding. In

addition, chemical conjugation yields a heterogenous mixture of

molecules joined via different positions on the antibody and

toxin in comparison with the structural uniformity of recombi-

nant immunotoxin. Potent single-chain immunotoxins derived

from PE have been made previously, against interleukin-2 re-

ceptor (9), transferrmn receptor ( 10), Le” family antigen ( 1 1),

and others, and their specific cytotoxicities have been shown.

Several sFv-immunotoxins are presently being evaluated in din-

ical trials (12).

The objective of the present study was to develop and

characterize a single-chain immunotoxin from hMN14, a hu-

manized anti-CEA monoclonal antibody ( 13), and to make a

preliminary evaluation of the potential of this immunotoxin

(hMN14(sFv)-PE4O) for future colon cancer treatment. To our

knowledge, this is the sole example of a recombinant immuno-

toxin against CEA to be reported to date.

3 The abbreviations used are: CEA, carcinoembryonic antigen; HPLC,high-performance liquid chromatography; PE, Pseudomonas exotoxin;sCEA, soluble CEA: sFv, single-chain Fv.

Association for Cancer Research. by guest on September 2, 2020. Copyright 1998 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

https://bloodcancerdiscov.aacrjournals.org

2826 Anti-CEA Single-Chain Immunotoxin

MATERIALS AND METHODS

Cell Lines. Human colorectal cancer cell line MIP-lOl

(14) and MIP-CEA clone 8 (15) were obtained from Dr. P.

Thomas, and colon adenocarcinoma cell line LS 174T was ob-

tamed from American Type Culture Collection (Manassas, VA).

All cells were cultured in RPMI 1640 (BioWhittaker, Inc.,

Walkersville, MD) supplemented with 10% heat-inactivated

FCS, 100 units/ml penicillin G, 100 units/ml streptomycin sul-

fate, and 2 mM L-glutamine.

Cloning of Antibody Fragment of hMN14 Antibody.

Cloning experiments and propagation of plasmid were per-

formed in Escherichia co/i XL-! blue (Stratagene. Cambridge,

United Kingdom). Total RNA was extracted from 5 X l0�

hMNh4/7/g9 transfectoma cells (Immunomedics, Inc.) using an

RNA isolation kit (Stratagene). Amplification of variable re-

gions was achieved by reverse transcnption-PCR using the

following primers that also create the sFv linker (restriction sites

underlined). VH forward: 5 ‘-GCCGGATCCGGCTCTGGTG-

GCTCAGGATCGGAGGTCCAACTGGTGGAGAG-3’ incor-

porates a BamHI site; VL forward: 5’-GGCGCCGCCATATG-

GACATCCAGCTGACCCAG-3’ incorporates a NdeI site;

VH backward: 5 ‘ -CGCCCAAGCTTTAGTACTGGAGACG-

GTGACCG-3’ incorporates a Hindlll site; and VL backward:

5’-GCCGGATCCACCCGATCCTGAGCCACCTCGlTTGAT-

TfCCACCTFGG-3’ incorporates a BamHI site. The single-

chain linker, a 15-amino acid long bridge, was modified to

(GGSGS)1 instead of the canonical (GGGGS)3 (9). The serine

residue in the middle of the motif was introduced to increase

hydration and linker solubility ( 16). PCR products were digested

with appropriate enzymes and ligated into the NdeI-HindIII

digested pRK78 (17). DNA sequence was confirmed by using

Sequenase (Amersham Corp.. Arlington Heights, IL).

Preparation of Immunotoxin and Antibody Fragments.

The single-chain immunotoxin was obtained by sohubilization

and refolding of inclusion body proteins from the host E. co/i

BL2 1(�DE3), as described ( 1 8). Properly refolded proteins were

purified by sequential ion exchange chromatography on

Sepharose, Mono Q (Pharmacia, Uppsala, Sweden) followed by

size exclusion chromatography on a TSK 030005W (Toso-

Haas) column on a Dionex 500 HPLC apparatus (Dionex Corp.,

Sunnyvale, CA). Fab fragment was prepared from hMN14 IgG

by papain digestion using a Fab preparation kit (Pierce Chem-

ical Co., Rockford, IL). Fractions containing Fab fragment were

concentrated by Centricon 10 ultrafiltration (Amicon, Inc., Bev-

erly, MA) and dialysed against PBS. Purified proteins were

stored until use at -80#{176}Cto minimize aggregation and activity

loss.

Flow Cytometry Analysis. Cell surface CEA was deter-

mined by flow cytometry with hMN14 anti-CEA antibody and

a control irrelevant isotype-matched (IgG 1 ,K) humanized anti-

body (anti-Tad-H; Ref. 19). Cells (2 X 106) were incubated with

antibodies in 50 �i1 of binding buffer of RPMI 1640 containing

10% horse serum, 50 mM Hepes-NaOH (pH7.0), and 0.2%

sodium azide for 30 mm at 4#{176}Cwith mixing. The cells were then

washed with ice-cold PBS twice and incubated under the same

conditions with goat-antihuman IgG ‘y chain phycoerythrin con-

jugates (Tago Immunologicals, Burlingame, CA). After two

washes, the samples were analyzed on a Epics-Profile II flow

cytometer (Coulter Electronics, Hialeah, FL).

Binding Assays. Complete antibody (25 �g), Fab frag-

ment, and immunotoxin were radiolabeled in PBS using lodo-

beads Iodination Reagent (Pierce Chemical Co.) with 0.2 mCi of

[1 25I]Na. After a 30-mn incubation at room temperature, the

reaction mixture was applied to a PD1O column (Pharmacia),

and the protein peak was collected. Binding was initiated by the

addition of 50 p.1 of cell suspension (3 X i0� cells/ml) to 50 pi

of a premixed solution of ‘ 25I-labeled protein. For “cold com-

petition,” a 200-fold molar excess of unlabeled hMN14 IgG was

added. The samples were incubated for 2 h at 4#{176}Cwith mixing.

Cell pellets were counted directly after washing three times with

the ice-cold binding buffer. Measurements of 1251 radioactivity

were performed at an efficiency of 74% in a Gamma 5500

gamma counter (Beckman Instruments, Fullerton, CA). The

counts of cold competition and machine background were sub-

tracted to derive specific binding. For each concentration of

antibody, duplicate measurements were performed and the re-

sults of both measurements are presented.

Cytotoxicity Assays. The cytotoxic effect of

hMN14(Fv)-PE4O was assessed by measuring the inhibition of

protein synthesis relative to control. Cells (1.6 X l0�) were

seeded/well (100 p.1) of a 96-well plate. Twenty-four hours later,

wells in triplicate were treated with various concentrations (0.1-

1000 ng/ml) of toxin and BSA after the removal of old media

and incubated at 37#{176}Cfor 24 h. Wells were pulsed with 1 pCi

[3H]leucine for 6 h before harvesting. For cold competition

assays designed to prove specificity, excess amount (20 p.g) of

either hMN14 IgG or irrelevant antibody was added to each well

before the addition of immunotoxin to block specific binding by

immunotoxin. To assess the effect of sCEA on the cytotoxicity,

purified CEA (Calbiochem, San Diego, CA) was added in

various concentrations in the presence of 100 ng/ml immuno-

toxin. The immunotoxin-CEA mixtures were incubated in

growth medium for 0 mm, 15 mm and 2 h at 37#{176}Cbefore adding

to the cells. Each experiment was repeated more than three

times.

Thermal Stability of the Immunotoxin. Thermal stabil-

ity was determined by incubating immunotoxin at 0. 1 mg/ml in

PBS at 37#{176}Cfor 8 and 24 h, followed by analytical chromatog-

raphy on a BIOSEP-SEC-S4000 column (Phenomenex, Tor-

rance, CA) on a Dionex 500 HPLC to distinguish the monomers

from larger aggregates. Bioassays for activity were performed as

above with the incubated toxin fractions.

RESULTS

Surface CEA Expression of Colon Cancer Cell Lines.

We first compared the surface expression of CEA among human

colorectal carcinoma cell lines, MIP-lOl, LS174T, and MIP-

CEA with hMN14 antibody. MIP-lOl (Fig. IA) showed no

detectable surface CEA expression as previously reported (15),

whereas CEA was detected on both LS174T and MIP-CEA cells

(Fig. 1, B and C), of which MIP-CEA was the higher expressing.

Preparation of hMN14(sFv)-PE4O Immunotoxin. Theinitial step was to choose an antibody to CEA. CEA is a member

of a family of related proteins, including nonspecific cross-

reactive antigen, biliary glycoprotein, and others, among which



.a

A.

C

00

‘aC.)

B.

C00

a)0

C. -

C00

‘a0

0 :c 10 20

�-,;- -v - . -- -I .- #{149}�i -c- rr,r...1 1 IS 10 ISS

t ‘I � ‘ ‘ ‘�‘ � � � ,-;,; � . ,- . ,

fluorescence

Fig. I Expression of CEA on colon carcinoma cell lines. Surfaceexpression of CEA on MIP-lOl (A), LS174T (B), and MIP-CEA (C)

was analyzed by flow cytometry. Broken line, control irrelevant anti-body; solid line, hMN14 antibody.

Clinical Cancer Research 2827

anti-CEA antibodies may be cross-reactive, depending on the

epitope recognized (2). From the many (>50) antibodies to

CEA presently available, MN14 was selected for these studies.

MN14 is a Primus class III antibody (i.e., it reacts exclusively

with CEA in the family of CEA-related proteins). This was also

available in a humanized format, which has advantages in terms

of reduced immunogenicity of the antibody moiety of the im-

munotoxin during human therapies (20).

VL and VH antibody segments were cloned and joined with

an intervening linker. The canonical (GGGGS)1 linker for VL

and VH joining incorporates a serine residue in the motif to

increase hydration and reduce the likelihood of invading and

disrupting the native hydrophobic cleft that joins VL and V� in

the parent antibody. Because a portion of sFvs are unstable (21),

we elected to use a linker with even greater hydration,

(GGSGS)3, on the possibility that this configuration would favor

the sFv stability even more. The product was expressed in E.

coli, purified and refolded in a well-behaved manner, with a

bOund. n.\l

Fig. 2 Affinity for CEA is preserved in hMN14(Fv)-PE4O. MIP-CEA

cells were incubated with ‘251-labeled sFv-PE4O or ‘251-labeled Fab or

25I-habehed IgG. as described in “Materials and Methods.” The data are

presented in Scatchard plot format. The monovalent affinity (-slope) is

comparable between the sFv-PE4O and Fab: both are less than that of thebivalent IgG. Similar results were obtained from at least two indepen-

dent experiments.

homogeneous appearance on SDS-PAGE (data not shown) and

on nondenaturing HPLC sizing chromatography (see below).

Affinity of hMN14(Fv)-PE4O. One of the risks of using

sFv is that it may lose affinity for the target antigen (2 1 ). This

is because these constructs use an artificial bridge (linker)

between VL and VH in lieu of the CL:CH 1 interaction that

normally stablizes an appropriate VL:VH juxtaposition for anti-

gen binding. To assess the affinity ofhMNl4(sFv), we tested its

binding activity against CEA-expressing target cells in compar-

ison with hMNh4 Fab and hMN14 whole IgG (Fig. 2). Immu-

notoxin, Fab, and whole antibody were labeled with 1251 and

incubated with MIP-CEA cells. The data for the specific binding

were analyzed by Scatchard plot. The measured affinity Ka

values of sFv, Fab, and IgG were 5 X l0�, 6 X l0�, and 3 X h0�

M1, respectively. Kds were 21 nM, 16 nM, and 3.4 nM. Although

monovalent forms were 5-6-fold reduced in affinity compared

with the parental IgG, there was no significant affinity loss

between the sFv and Fab. These data indicate that intact hMN 14

IgG binds bivalently, with an approximate 2-fold lower Bmax

(Fig. 2) and a net affinity enhancement of approximately 2.5-3

fold when normalized to binding sites per antibody molecule

(22). Data from this experiment indicate that MIP-CEA cx-

presses :��:5 X h0� CEA/cell.

Specific Cytotoxicity of hMN14(sFv)-PE4O. The cyto-

toxic activity of immunotoxin was assessed by measuring the

suppression of [3H]leucine incorporation by human colon can-

cer cell lines after treatment with serial dilutions of the recom-

binant protein. The immunotoxin inhibited protein synthesis of

all cell lines (Fig. 3, A-C; #{149},A, and �). The concentrations that

reduced the [3H]heucine incorporation by target cells to 50%

(IC50) were estimated and are shown in Table 1 . The suscepti-

bility to anti-CEA immunotoxin paralleled the CEA expression

of the tumor cell lines, with MIP-CEA the most sensitive and the

CEA-negative MIP-lOl the least sensitive. To examine the

specificity of the cytotoxicity and the role of antigen expression,



25�

0 I I 0 IlK) (XX)

A IT

-1k..--- lT�h5lNl4

---fr..- ll+t PC

IflhIflUIIOtI)\Ifl. ng nil

0.1 1 10 lOt) 10(X)

p rr-a---. lF+hMN14

-.-o-.-. lT+tP(�

imrnurn5o’un. ng ml

0. 1 It) 0(1 10(X)

immunotoxin, ng’ml


A. MIP-CEA B. LS174T C. MIP-lOl

Fig. 3 Specific cytotoxicity against colon carcinoma cell lines. Toxicity of hMNh4(Fv)-PE4O for MIP-CEA (A), LS174T (B), and MIP-hOl (C) werecompared. Cells were incubated with immunotoxin for 24 h, then assayed for protein synthesis activity by [3Hlleucine incorporation. #{149}.A. and U.

cytotoxicity by immunotoxin (IT): 0, A, and LI. antibody competition either with parental anti-CEA antibody (hMN14: solid lines) or nonspecificantibody (UPC: broken lines).

Table I Activities of anti-CEA(Fv)-PE4O on colon cancer cell lines

Expression and cytotoxicity data are from Fig. 1 and Fig. 3. Nonspecific IC�() is obtained from hMNl4 competition (see text).

Cell line

Surface CEA

antigen

sCEA

(ng/lO6cells/day)

IC3()

Specific

ng/ml nM

Nonspecific

ng/ml nM

MIP-CEA + + + 3.2k 1 2 0.2 I 100 17

LS174T ++ 5l2�’ 69 1.1 p1000 �15

MIP-lOl - 00h - - 500 8

a Results from Toth et al. (30).

#, Results from Thomas et al. (15).

assays were conducted by adding excess competitor hMN14

antibody to block all CEA binding sites for immunotoxin (Fig.

3, A-C; 0, & and LI). Cytotoxicity against MIP-CEA and

LS174T carcinoma cells was blocked by excess hMN14 but not

by nonspecific antibody, thus confirming that the inhibition of

protein synthesis by hMN14(sFv)-PE4O is due to specific bind-

ing to CEA. On the other hand, the cytotoxicity seen against the

CEA-negative MIP-lOl cell line at the high-immunotoxin con-

centrations is caused by nonspecific reaction because it could

not be blocked by hMNl4 antibody. The nonspecific cytotox-

icity was similar for MIP-lOl and MIP-CEA to a factor of “2,

but a nonspecific threshold was inapparent for LS 174T over the

concentration range studied. This suggests a greater native sen-

sitivity of the MIP cells than the LS174T cells to nonspecific

cytotoxicity, by nonspecific internalization of immunotoxin. It

is noted that the MIP lines have a common derivation: MW-

CEA was created by CEA gene transfection of the CEA-nega-

tive MIP-lOl cell line (15).

The Effect of sCEA on Cytotoxicity. Serum CEA up to

1000 ng/ml or more is sometimes observed in patient sera versus

the normal level of <5 ng/ml. In the presence of such high

sCEA, one could expect that an immunotoxin might be titrated-

out before arriving at the target tissue. To examine the effect of

high CEA levels on cytotoxic potency, we added free CEA in

various concentrations to 100 ng/ml immunotoxin and then

assessed the toxicity of the mixture against MIP-CEA. This

concentration of immunotoxin provides 77% of maximal sup-

pression of �3H]1eucine incorporation (Fig. 3A). Immunotoxin

was incubated in growth medium with sCEA for 0 mm, 15 mm,

and 2 h at 37#{176}Cbefore the addition to cells, and then incubated

with the cells an additional 24 h before labeling with

[3H]leucine. As seen in Fig. 4, no obvious change in protein

synthesis suppression was observed for CEA concentrations up

to 1000 ng/ml in the media, whereas sCEA at 5000 ng/ml

showed ““'60% reduction in net killing efficiency for this con-

centration of immunotoxin, from 77% down to ““'30% of max-

imal cytotoxicity. The duration of preincubation of CEA with

immunotoxin to ensure binding equilibrium before addition to

the target cells did not affect the toxicity profile. Averaging all

curves, the IC50 for sCEA inhibition of immunotoxin activity is



oluble (1i.�. ng ml


C

0

0�0

0000

Fig. 4 Specific cytotoxicity is resistant to high levels of sCEA. Immu-

notoxin (100 ng/ml) was preincubated for 0 mm, 15 mm, or 2 h withvarious concentrations of CEA before adding to the MIP-CEA cells.Cytotoxicity was assayed as in Fig. 3. At 100 ng/ml immunotoxin,

[3H]heucine incorporation is suppressed by 77% relative to the control.

estimated at 4000 ng/ml (20 nM) of CEA. The presence of CEA

itself did not affect the rate of protein synthesis (data not

shown).

Stability of Immunotoxin. The stability of immunotox-

ins at 37#{176}Cis an important factor in their usefulness as thera-

peutic agents. Loss of activity by immunotoxin is governed by

its tendency to aggregate at 37#{176}C,which has been documented

previously (21, 23, 24). Prior assays with 2-h preincubation in

media containing 10% serum showed no suggestion of activity

loss relative to unincubated hMN14(sFv)-PE4O (Fig. 4). How-

ever, albumin is typically added to proteins (e.g. , enzymes) to

reduce aggregation and improve stability, and serum is 5% by

weight albumin (50 mg/ml). Therefore, we chose to omit serum

in a further assay, as a more stringent test of the tendency of the

immunotoxin to aggregate. In Fig. 5, the thermal stability of

hMN14(Fv)-PE4O was determined by incubating for 8 and 24 h

at 37#{176}C,then measuring the amount of aggregation by HPLC

size analysis and activity loss by bioassay. No peak was detected

corresponding to aggregates even after the 24-h incubation, and

there was no loss of specific immunotoxin activity by bioassay.

DISCUSSION

CEA is an antigen expressed on the surface of human

cancers of epithelial cell origin, especially in gastrointestinal

carcinomas (1, 2, 25), and it has accordingly been of interest to

target CEA in immunotherapies. Tumor cells typically express

quantitatively much higher levels of CEA, averaging 35-fold

higher than normal colonic mucosa (26). This should enhance

discrimination between normal and tumorous expression of the

protein, as shown by our data relating cellular sensitivity to level

of CEA expression (Figs. 1 and 3). Furthermore, the expression

of CEA on normal cells of the colonic epithelium is on lumenal

surfaces that should be less accessible to attack by a blood-borne

immunotoxin (1, 2).

In this study, we engineered an anti-CEA sFv with a

humanized variable region using hMNl4 as the parental anti-

body. The therapeutic interval of a particular toxin construct is

usually limited by host immune response against the toxin

moiety, which antibody humanization will not change. How-

ever, antibody humanization in the present setting has the ad-

vantage of avoiding concurrent antiglobulin responses, thus

allowing the subsequent use of the hMN14 antibody in other

therapeutic modifications. Our Scatchard analysis indicated that

sFv retains its specific binding activity against CEA with no

obvious affinity loss relative to its two-chain counterpart, Fab

(Fig. 2), and is comparable in affinity to other anti-CEA sFvs

obtained by recombinant phage display technology (27-29).

It is noted that some immunotoxins may lose >70% of

their initial cytotoxic potency due to aggregation during 8-h

incubation at 37#{176}C.To ameliorate this common problem, strat-

egies were devised to include interchain disulfides in some Fv

toxin constructs (21, 23, 24). In contrast, the hMN14(Fv)-PE4O

was stable with prolonged incubation at 37#{176}Ceven without

disulfide-stabilization. The stability of the immunotoxin de-

pends on the structure of the antigen-binding domain, indicating

that the sFv of hMN14 antibody has a suitable character to serve

as a single-chain im.munotoxin. An additional possibility is that

the more hydrated linker we applied may foster improved sta-

bility by not invading the hydrophobic cleft that binds V� and

VL to form an appropriate antigen binding site; however, direct

comparisons with the canonical linker were not made from

which to draw a conclusion.

Although membrane-expressed CEA does not internalize

actively, our hMN14(Fv)-PE4O is able to kill target cells. This is

similar to Tac (interleukin-2 receptor a) targeting in which

antigen is not actively internalized, but immunotoxin neverthe-

less kills cells in an antigen-dependent manner (9). This mani-

festation of specific cytotoxic activity is thought to be due to the

nonspecific bulk clearing of membrane surface and associated

proteins via on-going cellular endocytic activities, in which

binding of immunotoxin to surface-bound antigen increases the

probability that it will also be internalized. hMN14(Fv)-PE4O

showed specific cytotoxicity to MIP-CEA but not to the CEA

nonexpressing parental line, MIP-lOl (Fig. 3, A and C). Protein

synthesis in MIP-lOl was inhibited at high immunotoxin con-

centrations (“'500 ng/ml) that was comparable, to a factor of

two, with the nonspecific killing of MIP-CEA. This residual

killing is presumably a measure of nonspecific cellular uptake

from bulk fluid phase of the medium, which would not be

inhibited by excess hMN14 antibody (Fig. 3C).

Another CEA-expressing target cell, LS174T, was also

specifically killed by hMN14(Fv)-PE4O, but not as efficiently as

MIP-CEA (Fig. 3, A and B). This lower activity is plausibly

explained by the approximate 10-fold lower surface CEA on

LS174T versus MIP-CEA (comparing histograms of Fig. 1, B

and C). By our Kd measurements, we estimate similar numbers

of toxin molecules bound/cell at the respective IC50 values

(5,000 for MIP-CEA and 3,000 for LS174T), thus suggesting a

common final threshold of cell binding to mediate cytotoxicity.

A similar observation correlated the lower sensitivity of Tac-

expressing cell lines to anti-Tac(Fv)-PE4O when they had lower

antigen expression, but which were constant in sensitivity when

normalized to estimated toxin molecules bound/cell (Ref. 9;

R. P. J., calculations not shown). The relation of antigen expres-

sion to specific cytotoxicity suggests that maneuvers to increase



A.

0.10 Oh

0.08

AU 0.06

0.04

0.02

0.00 � � � � � � � I � � f-f

0 2 4 6 8 10 12 14

0.10 8h

0.08

AU 0.06

0.04

0.02

0.00 �� t �I ir(� I �tI�fl� � I I�

0 2 4 6 8 10 12 14

0.10 � 24h

0.08

AU 0.06

0.04

0.02

0.00 iTIEJ � II I � � H � � � 1 ‘I � � � � � � � I I

0 2 4 6 8 10 12 14

Minutes

&

B.

125%

I(X)% � � “

. �

7�.

.i0’5�

2::. :�:�:; E� �, �..

0 I 1 10 IfS) 0(X)

immunotoxin. ngiml


Fig. 5 hMNI4(Fv)-PE4O is stable

at 37#{176}C.Immunotoxin was incu-

bated at 100 p.g/ml at 37#{176}Cin PBS

for 8 and 24 h and analyzed by size

exclusion chromatography (A). Themonomer peak chutes at 10-1 1 mm.IgG chutes at ‘-9 mm (data notshown). Cytotoxic activities of heat-

treated immunotoxin (B). Curvedifferences are not statistically sig-

nificant.

tumorous CEA expression, such as IFN-’y treatment (30, 31),

may be productively applied to enhance the toxicity profile

against CEA-expressing cancers. But this analysis also suggests

a potentially important future direction to improve this agent for

therapy: it implies that a higher affinity version of this antibody

would be effective at still lower concentrations than seen here-

without increasing nonspecific toxicity-thus, significantly en-

hancing the therapeutic index.

A further possible explanation for the lower sensitivity of

LS h74T to specific killing that could be considered is the

difference in the rate of sCEA production. The CEA shedding

rate for LS174T is extremely high, 512 ng/106 cells/day, com-

pared with 3.2 ng/106 cells/day for MIP-CEA (Table 1). This

means 7 1 ,000 molecules of sCEA are produced by one cell/hour

for LS 174T compared with 450 molecules for MIP-CEA. Al-

though total accumulated sCEA in our assays at 24 h (-100

ng/mh) would be far below the inhibitory concentrations in Fig.

4, the immunotoxin bound to LS174T could be thought to have

less opportunity to be internalized due to its higher chance to be

shed from the cell before endocytosis. However, we consider

this explanation less likely. The steady state surface expression

should yield the same net internalization for a stochastic endo-

cytosis process, regardless of the synthetic and shedding rates,

under certain assumptions. (See Ref. 32 for a more detailed

kinetics analysis of protein shedding and expression.) Finally,

the sufficiency of the rationale of lower surface CEA expression

to explain the lower drug sensitivity of LS174T (see above)

would seem to obviate any need to invoke such more complex

arguments as involve shedding.

Coincubation with sCEA in the cytotoxicity assay did not

affect the activity of the immunotoxin up to 1000 ng/ml, but

immunotoxin activity was suppressed approximately 60% at

5000 ng/ml sCEA. sCEA in patient sera only infrequently

reaches 1000 ng/ml (5 nM) and it is, therefore, unlikely to be an

important factor in the efficacy of this immunotoxin in vivo. In

our assay, the CEA on cells (““' 10-20 fmol) is far less than the

CEA in the supernatant (1000 ng/ml 500 fmol) versus total

immunotoxin (100 ng/ml = 200 fmoh). The Scatchard plot

analysis (Fig. 2) allows us to estimate the surface density of

CEA molecules on the MIP-CEA, ““'5 X l05/cell, which corre-

sponds to a CEA concentration of - 100 p.M on the cell surface,

and ‘� 10 p.M for LS174T relative to sCEA concentrations of 10

flM or less. This might seem to provide a basis for a selective

advantage of CEA on cells to acquire anti-CEA immunotoxin

and resist sCEA competition. However, this is unlikely to be a

suitable explanation. We have previously argued that monova-

lent antibody binding will normally partition between cellular

and soluble antigen in their respective proportion to total antigen

present (33). The fact that there is little impact on cellular

toxicity under this condition despite the large ratio excess of

sCEA suggests a different and more likely rationale, as follows.

Expressed in terms of Kd, the inhibition pattern of sCEA

becomes fully understandable. We assume provisionally that

sFv affinity for sCEA equals that determined for cellular CEA




(Fig. 2). sCEA at 1000 ng/ml (5 nM) is still well below the sFv

Kd (21 flM) and will not appreciably reduce the free sFv toxin

that can bind to tumor CEA, and correspondingly has little or no

effect on cytotoxicity. At the sCEA IC50 of 4000 ng/ml (20 nM),

however, sCEA also equals the Kd for the immunotoxin; our

Scatchard analysis indicates that the immunotoxin should be

half-saturated with sCEA and, correspondingly, the free immu-

notoxin available for cell CEA binding should be reduced by

half. Because MN14(Fv)-PE4O at 100 ng/mh (2 nM) is below the

binding Kd (21 nM) and in the linear range for immunotoxin

activity on MIP-CEA (Fig. 3), this change in free immunotoxin

concentration at the sCEA IC50 directly reduces both the cellular

binding of toxin and drug potency in parallel. Similarly, cyto-

toxicity by anti-Tac(Fv)-PE4O against Tac-expressing tumor

cells was resistant to soluble Tac antigen until exceeding the

anti-Tac sFv Kd (0.3 nM) and reducing free immunotoxin (34)#{149}4

Finally, this correspondence of expected results with observa-

tion ultimately supports our assumption of comparable affinities

of sFv for cellular and sCEA.

Logically, as the affinity of immunotoxin for antigen in-

creases, the tumor targeting efficacy increases, but the suscep-

tibihity to soluble antigen binding increases also. Ultimately,

when soluble antigen greatly exceeds the Kd of the sFv-toxin,

there will be virtually no free toxin. However, we (33) and

others (35) have previously shown that antibody in the setting of

saturating antigen still achieves tumor targeting by an exchange

partition between cell-bound and soluble antigen. Of particular

interest in the present study is the fact that sCEA is a long-hived

protein in serum (t112 ““4d; 36), far exceeding the typical half-

life of sFv-toxins (t112< 1 h; 37). Parallels may be drawn with

our prior study of soluble Tac antigen, which has an abbreviated

half-life that is markedly prolonged by binding to long-surviv-

ing anti-Tac antibody (32). Binding to sCEA predicts a marked

prolongation of survival of sFv-toxin in vivo that dramatically

changes its pharmacokinetics, potentially altering its therapeutic

profile in ways that may not be fully predictable a priori, with

either decreased (38) or increased (39) biological activity and

systemic toxicity. For hMN14(Fv)-PE4O, sCEA binding should

not be a major factor in the drug pharmacokinetics, but this

feature will be important to consider in any Phase I anti-CEA

studies with sFv-toxins of high affinity, defined as having a Kd

for CEA that is lower than commonly encountered concentra-

tions of sCEA in vivo (e.g., < 1 nM).

By criteria of cytotoxicity, specificity, affinity, and stabil-

ity, our hMN14(Fv)-PE4O displays a satisfactory in vitro profile.

This agent may be clinically useful for tumors with elevated

CEA expression without affecting normal tissues with no or low

CEA, even for patients with high serum CEA levels.

ACKNOWLEDGMENTS

We arc particularly grateful to Dr. Glenn D. Steele, Jr. (Chairman

of the Surgery Department of the former New England Deaconess

Hospital) for initial seed money and support that made this and allied

4 We note that there is an error in the reported unit definition in this

reference (34) that underrepresents the mass of soluble Tac antigen by..-10-fold.

projects possible. We arc grateful to Drs. Robert Sharkey. David Gold-

cnbcrg, and Hans Hansen (Garden State Cancer Center and Immuno-

medics, Inc., Belleville, NJ) for providing hMNh4. We thank Drs. Ira

Pastan and Ellen Vitetta for reviewing the manuscript: and Daniel Hagg

and Dr. Gang Zhcng for excellent technical assistance.

REFERENCES

1. Schwartz, M. K. Cancer markers. In: V. T. DeVita, Jr., S. Hellman,and S. A. Rosenberg (eds.), Cancer: Principles and Practice of Oncol-

ogy, pp. 53 1-542. Philadelphia: Lippincott, 1993.

2. Nap, M., Hammarstrom, M. L., Bormer, 0., Hammarstrom, S.,

Wagener, C., Handt, S., Schrcyer, M., Mach, J-P., Buchegger, F., von

Kleist, S., Grunert, F., Seguin, P., Fuks, A., HoIm. R., and Lamerz, R.

Specificity and affinity of monoclonal antibodies against carcinoembry-

onic antigen. Cancer Rcs., 52: 2329-2339, 1992.

3. Landis, S. H., Murray, T.. Bolden. S., and Wingo, P. A. Cancer

statistics, 1998. CA Cancer J. Chin., 48: 6-29, 1998.

4. Watanabe, Y., Miyazaki, H., and Osawa, T. Effect of a monoclonal

anti-carcinoembryonic antigen antibody-ricin A-chain conjugate on hu-

man carcinoembryonic antigen producing tumor cells. J. Pharmacobio-

dyn., 7: 593-603, 1984.

5. Avila, A. D., Matco de Acosta, C., and Lage, A. A carcinoembryonic

antigen-directed immunotoxin built by linking a monoclonal antibody to

a hemolytic toxin. Int. J. Cancer, 43: 926-929, 1989.

6. Richardson, V. J., Ford, C. H., Tsaltas, G., and Gallent, M. E.Doxorubicin anti-carcinoembryonic antigen immunoconjugate activity

in vitro. Eur. J. Cancer Clin. Oncol., 25: 533-540, 1989.

7. Manosroi, J., von Klcist, S., Manosroi, A., and Grunert, F. Thermo-

stability and antitumor activity on colon cancer cell lines of monoclonal

anti-CEA antibody-saporin immunotoxin. J. Korean Mcd. Sci., 7: 128-135, 1992.

8. Tsai, L-C., Chen, Y-L., Lee, C., Chen, H-M., Chang, Z-N., Hung,

M-W., Chao, P-L., and Lin, J-Y. Growth suppression of human cob-rectal carcinoma in nude mice by monocbonal antibody C27-abrin A

chain conjugate. Dis. Colon Rectum, 38: 1067-1074. 1995.

9. Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann, T. A.,FitzGerald, D. J., and Pastan, I. A recombinant immunotoxin consisting

of two antibody variable domains fused to Pseudomonas exotoxin.

Nature (Lond.), 339: 394-397, 1989.

10. Batra, J. K., Jinno, Y., Chaudhary, V. K.. Kondo, T., Wilhingham,

M. C.. FitzGerald, D. J., and Pastan, I. Antitumor activity in mice of an

immunotoxin made with anti-transferrin receptor and a recombinant

form of Pseudomonas exotoxin. Proc. Natl. Acad. Sci. USA, 86: 8545-

8549, 1989.

1 1. Brinkmann, U., Pai, L. H., FitzGerald, D. J., Willingham, M.. and

Pastan, I. B3(Fv)-PE38KDEL, a single-chain immunotoxin that causes

complete regression of a human carcinoma in mice. Proc. NatI. Acad.

Sci. USA, 88: 8616-8620, 1991.

12. Pai, L. H., and Pastan, I. Immunotoxin therapy. In: V. T. DeVita,Jr., S. Hellman, and S. A. Rosenberg (eds.), Cancer: Principles andPractice of Oncology, pp. 3045-3057. Philadelphia: Lippincott-Raven,

1997.

13. Sharkey, R. M., Juweid, M., Shcvitz, J., Bchr, T., Dunn. R.,Swaync, L. C., Wong, G. Y., Blumenthal, R. D., Griffiths, G. L.. Siegel,

J. A., Leung, 5-0., Hansen. H. J., and Goldcnberg, D. M. Evaluation ofa complementarity-determining region-grafted (humanized) anti-carci-

noembryonic antigen monocbonal antibody in preclinical and clinical

studies. Cancer Res., 55 (23 Suppl.): 5935-5945, 1995.

14. Niles, R. M., Wilhelm, S. A., Steele, G. D., Jr., Burke, B., Chris-

tensen, T., Dexter, D., O’Brien. M. J., Thomas, P., and Zamcheck. N.Isolation and characterization of an undifferentiated human colon car-

cinoma cell line (MIP-lOl). Cancer Invest., 5: 545-552, 1987.

15. Thomas, P., Gangopadhyay. A., Steele, G., Jr.. Andrews, C..

Nakazato, H., Oikawa, S., and Jessup, J. M. The effect of transfection of

the CEA gene on the metastatic behavior of the human colorectal cancer

cell line MIP-lOl. Cancer Lett., 92: 59-66, 1995.




16. Wu, X-C., Ng, S-C., Near, R. I., and Wong, S-L. Efficient produc-tion of a functional single-chain antidigoxin antibody via an engineered

Bacillus subtilis expression-secretion system. Biotechnology, I 1: 71-

76, 1993.

17. Kreitman, R. J., Baibon, P., Chaudhary, V. K., FitzGerald, D. J. P.,and Pastan, I. Recombinant immunotoxins containing anti-Tac(Fv) andderivatives of Pseudomonas exotoxin produce complete regression in

mice of an interhcukin-2 receptor-expressing human carcinoma. Blood,

83: 426-434, 1994.

I 8. Buchner, J., Pastan, I., and Brinkmann, U. A method for increasing

the yield of properly folded recombinant fusion proteins: single-chain

immunotoxins from renaturation of bacterial inclusion bodies. Anal.

Biochem., 205: 263-270, 1992.

19. Junghans, R. P., Waldmann, T. A., Landolfi, N. F., Avdaloric,

N. M., Schneider, W. P., and Queen, C. Anti-Tac-H, a humanized

antibody to the interleukin 2 receptor with new features for immuno-

therapy in malignant and immune disorders. Cancer Res., 50: 1495-

1502, 1990.

20. Junghans, R. P., Sgouros, G., and Schcinbcrg, D. B. Antibody-

based immunotherapies for cancer. In: B. A. Chabner and D. L. Longo(eds.), Cancer Chemotherapy and Biotherapy: Principles and Practice,

pp. 655-689. Lippincott: Philadelphia, 1996.

21. Brinkmann, U., Reiter, Y., Jung, S-H., Lee, B., and Pastan, I. Arecombinant immunotoxin containing a disulfide-stabilized Fv frag-

ment. Proc. Nail Acad. Sci. USA, 90: 7538-7542, 1993.

22. Junghans, R. P. Cruel antibody fictions! Cellular antigen enumer-

ation by “saturation” binding, example CD3. Immunol. Today, in press,

1998.

23. Reiter, Y., Brinkmann, U., Webber, K. 0., Jung, S-H., Lee, B., and

Pastan, I. Engineering interchain disulfide bonds into conserved frame-

work regions of Fv fragments: improved biochemical characteristics of

recombinant immunotoxins containing disulfide-stabihized Fv. Protein

Eng., 7: 697-704, 1994.

24. Benhar. I., and Pastan, I. Characterization of Bl(Fv)PE38 andBh(dsFv)PE38: single-chain and disulfide-stabilized Fv immunotoxins

with increased activity that cause complete remissions of established

human carcinoma xcnografts in nude mice. Chin. Cancer Res., 1: 1023-

1029, 1995.

25. Cournoycr, D., Beauchemin, N., Bouchcr, D., Bcnchmol, S., Fuks,

A., and Stanners, C. Transcription of genes of the carcinoembryonicantigen family in malignant and nonmalignant human tissues. Cancer

Res., 48: 3153-3157, 1988.

26. Boucher, D., Cournoycr, D., Stanners, C. P., and Fuks, A. Studies

on the control of gene expression of the carcinoembryonic antigen

family in human tissue. Cancer Res., 49: 847-852, 1989.

27. Ayala, M., Duc#{241}as, M., Santos, A., V#{225}zqucz, J., Mcn#{233}ndcz, A.,

Silva, A., and Gavibondo, J. V. Bacterial single-chain antibody frag-

ments, specific for carcinoembryonic antigen. Biolechniques, 13: 790-799, 1992.

28. Abraham, R., Buxbaum, S., Link, J., Smith, R., Venti, C., andDarsley, M. Screening and kinetic analysis of recombinant anti-CEAantibody fragments. J. Immunol. Methods, 183: 119-125, 1995.

29. Verhaar, M. J., Chester, K. A., Keep, P. A., Robson, L., Pcdley,

R. B., Boden, J. A., Hawkins, R. E., and Begent, R. H. J. A single chainFv derived from a filamentous phage library has distinct tumour target-ing advantages over one derived from a hybridoma. Int. J. Cancer, 61:

497-501, 1995.

30. Toth, C. A., and Thomas, P. The effect of interferon treatment on 14human colorectal cancer cell lines: growth and carcinoembryonic anti-

gen secretion in vitro. J. Interferon Res., 10: 579-588, 1990.

31. Guadagni, F., Witt, P. L., Robbins, P. F., Schlom, J., and Greiner,

J. W. Regulation of carcinoembryonic antigen expression in differenthuman colorectal tumor cells by interferon--y. Cancer Res., 50: 6248-

6255, 1990.

32. Junghans, R. P., and Wahdmann, T. A. Metabolism of Tac (IL2R):

physiology of cell surface shedding and renal elimination, and suppres-

sion of catabolism by antibody binding. J. Exp. Med., 183: 1587-1602, 1996.

33. Junghans, R. P., Carrisquillo, J. A., Waldmann, T. A. Impact of

antigenemia on the bioactivity of infused anti-Tac antibody: implication

for dose selection in antibody immunotherapies. Proc. Nath. Acad. Sci.USA, 95: 1752-1757, 1998.

34. Kreitman, R. J., and Pastan, I. Targeting Pseudomonas exotoxin tohematohogic malignancies. Semin. Cancer Biol., 6: 297-306, 1995.

35. Sharkey, R. M., Goldenberg, H., Lee, R. E., Ballance, C., Pawlyk,D., Varga, D., and Hansen, H. J. Murine monochonal antibodies againstcarcinoembryonic antigen: immunological, pharmacokinetic, and target-

ing properties in humans. Cancer Res., 50: 2823-2831, 1990.

36. Mai, M., and Takahashi, Y. Prediction of recurrence of gastromn-testinal cancer from standpoint of biological malignancies: tumor

marker doubling time and its half life period line. Hum. Cell, 6: 82-87,

1993.

37. Kreitman, R. J., Chaudhary, V. K., Waldmann, T. A., Hanchard, B.,Cranston, B., FitzGerald, D. J., and Pastan, I. Cytotoxic activity of

recombinant immunotoxins or diphtheria toxin toward lymphocytes

from patients with adult T-cehh leukemia. Leukemia (Baltimore), 7:

553-562, 1993.

38. Berson, S. A., and Yalow, R. S. Insulin in blood and insulinantibodies. Am. J. Med., 40: 676-690, 1966.

39. Finkelman, F. D., Madden, K. B., Morris, S. C., Holmes, J. M.,Boiani, N., Katona, I. M., and Maliszewski, C. R. Anti-cytokinc anti-

bodies as carrier proteins. Prolongation of in vivo effects of exogenous

cytokines by injection of cytokinc-anti-cytokine antibody complexes.

J. Immunol., 151: 1235-1244, 1993.



Date post:	16-Jul-2020
Category:	Documents
Upload:	others
View:	5 times
Download:	0 times

A Single-Chain Immunotoxin against Carcinoembryonic Antigen · characterize a single-chain...

Documents