antigens (SAgs) are a family of bacterial and viral proteins that activate a high ... human colon cancer, were produced at Pharmacia (8, 9). E. coli HB101 was ...
Proc. Natl. Acad. Sci. USA Vol. 91, pp. 8945-8949, September 1994 Immunology
Monoclonal antibody-superantigen fusion proteins: Tumor-specific agents for T-cell-based tumor therapy MIKAEL DOHLSTEN*t, LARS ABRAHMStNt, PER BJ6RK*, PETER A. LANDO*, GUNNAR HEDLUND*t, GORAN FORSBERG*, THOMAS BRODIN*, NICK R. J. GASCOIGNE§, CECILIA F6RBERGt, PETER LINDf, AND TERJE KALLAND*t¶ *Pharmacia Oncology Immunology, Lund, Sweden; tThe Wallenberg Laboratory, Department of Tumor Immunology, University of Lund, Lund, Sweden; *Pharmacia Bioscience Center, Stockholm, Sweden; and §The Scripps Research Institute, Department of Immunology, La Jolla, CA 92037
Communicated by Hans J. Maller-Eberhard, April IS, 1994 (received for review October 18, 1993)
ABSTRACT The bacterial superantigen staphylococcal enterotoxin A (SEA) is an extremely potent activator of T lymphocytes when presented on major histocompatibility complex (MHC) class H molecules. To develop a tumor-specific superantigen for cancer therapy, we have made a recombinant fusion protein of SEA and the Fab region of the C215 monodonal antibody specific for human colon carcinoma cells. SEA as part of a fusion protein showed a >10-fold reduction in MHC class II binding compared to native SEA, and accordingly, the affinity of the FabC215-SEA fusion protein for the C215 tumor antigen was 100-fold stronger than to MHC class 1I molecules. The FabC215-SEA fusion protein efficiently targeted T cells to lyse C215+ MHC class II- human colon carcinoma cells, which demonstrates functional substitution of the MHC class fl-dependent presentation of SEA with tumor specificity. Treatment of mice carrying B16 melanoma cells expressing a transfected C215 antigen resulted in 85-99% inhibition of tumor growth and allowed long-term survival of animals. The therapeutic effect was dependent on antigenspecific targeting of the FabC215-SEA fusion protein, since native SEA and an antigen-irrelevant FabC242-SEA fusion protein did not influence tumor growth. The results suggest that Fab-SEA fusion proteins convey superantigenicity on tumor cells, which evokes T cells to suppress tumor growth.
We recently demonstrated that chemical conjugation of mAb against colon carcinomas with the bacterial SAg staphylococcal enterotoxin A (SEA) targeted T cells to lyse MHC class II tumor cells in vitro (6) and induced tumor-suppressive lymphokines (7). We have now expressed in Escherichia coli a recombinant fusion protein of SEA and a Fab fragment of a colon carcinoma-reactive mAb. Fab-SEA treatment of mice carrying tumors expressing the relevant antigen resulted in 85-99%o suppression of tumor growth in the absence of overt systemic side effects. We believe that tumor-specific SAgs should be of considerable interest as therapeutic agents against cancer.
MATERIALS AND METHODS Protein Reagents. Recombinant SEA was expressed in E. coli and purified to homogeneity. The hybridomas secreting the C215 (IgG2a) and C242 (IgGl) mAbs, reacting with human colon cancer, were produced at Pharmacia (8, 9). E. coli HB101 was used as host during the DNA construction work (10). Cloning of the cDNAs Encoding the Murine Antibodies C215 and C242. PCR was performed on C215 hybridoma cells by using the total RNA isolation kit (Ginna/Biotecx Laboratories, Friendswood, TX). Reverse transcriptase (from Moloney murine leukemia virus, Superscript, GIBCO/BRL) and random hexamer primers were used to make cDNA. The primers Mhvp-7 (11) and Mushcp-21 (5'-CAATTTTCTTGTCCACCTTGGTGCT) were used to isolate the heavy (H)-chain DNA. Similarly, the primers Mkvp-7 (11) and
The therapeutic use of naked monoclonal antibodies (mAbs) in human epithelial cancer has met with limited success (1). The amount of mAb found to accumulate in the tumor is generally low, tumor cells display extensive heterogeneity in antigen expression, and the effector mechanisms responsible for tumor-cell destruction are insufficient (1, 2). Recent progress in the molecular understanding of T-cell recognition has provided insight into the role of T cells in the response against allogeneic and neoplastic tissues. The vigorous and destructive response against an allograft involves a high frequency of T lymphocytes that, via the T-cell receptor (TCR), recognizes peptide antigens presented in the context of major histocompatibility complex (MHC) molecules. In contrast, the frequency of T cells responding to a tumor is generally low and insufficient to interfere with tumor growth. This suggests that an attractive approach for immunotherapy would be to target a high frequency of T cells to the tumor area. We have therefore developed an antibody-based therapy that provides tumor cells with superantigenicity. Superantigens (SAgs) are a family of bacterial and viral proteins that activate a high frequency of T cells to cytokine release and cell-mediated cytotoxicity (3-5). Bacterial SAgs bind with high affinity to MHC class II molecules and subsequently interact with T cells expressing particular sequences in the variable (V) region of the TCR ,/ chain (TCR V(3) (3-5).
Muslcp-11 (5'-TGTGTCCCGGGATACAGTTGGTGCAGCATCAGCCC) were used to isolate the K-chain DNA. The C242 hybridoma cell line was used as source material for cloning of the cDNAs encoding the immunoglobulin H chain and the K chain. Polyadenylylated RNA was extracted from hybridoma cells, converted to double-stranded cDNA, and cloned into the phage A-based vector Uni-ZAP (Stratagene). Positive clones were used to prepare cDNA containing pBluescript SK(-) plasmids by phagemid excision and propagation. The resulting cDNA-containing plasmids were characterized by restriction enzyme mapping, and the nucleotide sequences of inserts from selected plasmids were determined. Assembly of the Fab-SEA Gene Fusion Segment. The following pieces were used: (i) A Not I-Mlu I fragment containing the ribosomal binding site and most of the signal peptide encoding a portion of the gene from E. coli outer membrane protein A as translation and secretion signals for Abbreviations: SEA, staphylococcal enterotoxin A; SAg, superantigen; MHC, major histocompatibility complex; CTL, cytotoxic T lymphocyte; TCR, T-cell receptor; mAb, monoclonal antibody; V, variable; C, constant; H, heavy; V8, (8 chain V region. ITo whom reprint requests should be addressed at: Pharmacia Oncology Immunology, Scheelevagen 22, S-223 63 Lund, Sweden.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
8945
8946
Immunology: Dohlsten et al.
Proc. Natl. Acad. Sci. USA 91
(1994)
the H chain. The introduction of the Mlu I site changed the second to the last codon from CAG to AAC. (ii) The H chain V region gene was modified by introducing flanking BssHII and Xma I sites by PCR. The last base of the former site is the first base of the gene and the latter spans codons 12-14 of the constant (C) portion. (iii) The gene encoding the first H chain C region was obtained from mAb C242 gene. The encoded protein is identical to this portion of a consensus murine IgG1 chain. A Xma I-Rsr II fragment was used from codon 12 including the first Cys codon of the hinge-encoding region followed by the tripeptide spacer Gly-Gly-Pro. (iv) The SEA gene included on a Rsr II-Sfi I fragment was taken from the plasmid pKP554. The SEA gene contained on this plasmid is identical to the sequence published by Betley and Mekalanos (12), except for two TAA stop codons. (v) The translation and secretion signals for the K chain were derived from a Sfi I-Mlu I fragment. This contained 40 bp from the staphylococcal protein A gene (13), including the ribosomal binding site and the first 10 bp of the signal peptide sequence, followed by 59 bases to compose a synthetic signal peptide. (vi) The K chain V region gene was modified by introducing flanking Dsa I and Kas I sites by PCR. The last base of the former site is the first base of the gene and the latter spans codons 4-6 of the C portion. (vii) The gene encoding the K chain C domain was obtained from the mAb C242 gene, with the same Kas I site as above introduced and ending with a TAA stop codon followed by a Xba I site. Expression and Purification of Fab-SEA Fusion Proteins. E. coli K-12 UL635 (ara-14, xyl-7, AompT, T4R) was used for expression of Fab-SEA. The fermentation was performed at 25°C at pH 7 in a 7-liter fermenter (Chemap, Mannedorf, Switzerland). At an OD600 of r50, transcription from the trc promoter was induced by addition of isopropyl P-Dthiogalactoside to 0.05 mM. The fermentation was interrupted at an D060 of -85 and the cells were separated by filtration. Medium containing Fab-SEA was incubated with Q-Sepharose FF (Pharmacia), the gel was removed by suction filtration, and the filtrate was further clarified by centrifugation at 20,000 x g. The Fab-SEA protein was then affinity-purified on a protein G-Sepharose column. The eluted fraction was further purified on a Mono S HR 5/5 column (Pharmacia) and fractions containing Fab-SEA were pooled and passed through a PD-10 column (Pharmacia) with 10 mM sodium phosphate (pH 7.4)-buffered saline (PBS) as
Cell Lines. The human B-cell lymphoma Raji, the human colon carcinomas SW620 and Colo2O5, the murine B16 melanoma (ATCC, MD), and CHO cells were cultured as described (6, 16). SEA-reactive T-cell lines (>99%o CD3+) were established from peripheral blood lymphocytes as described (6). Transfection with cDNA Encoding the C215 Antigen. The expression vector pKGE839 containing the GA733-2 cDNA (encoding the C215 antigen) and the neomycin-resistance gene (14) was transfected into B16 melanoma cells or CHO cells using Transfectam (Sepracor, IBF, Villeneuve-laGarenne, France) and electroporation, respectively. G418resistent clones were harvested and subjected to repeated cell sorting. The sorted CHO-C215 cells were kept as a stable cell line, and high-expressing clones were selected from the B16-C215 cells. Establishment of CHO cells expressing HLA-DR has been described (16). Cytotoxicity Assay. Cytotoxicity was measured at various effector/target cell ratios in 4-h 51Cr release assays as described (6, 16). Immunohistochenical Staining. Tumor tissues from colon cancer patients were snap-frozen in liquid nitrogen-cooled isopentane and 4-pm frozen sections were prepared. After drying and fixation in acetone, the sections were rehydrated and stained with FabC215-SEA and SEA followed by antimouse immunoglobulin-biotinylated antibody or rabbit antiSEA-biotinylated antibody and avidin-biotinylated peroxidase complex (17). The slides were developed in diaminobenzidine (Sigma) and mounted in DPX medium (Sigma). In Vivo Tumor Models. C57BL/6 (H-2d) mice were obtained from Bomholtgaard (Ry, Denmark). For induction of lung metastasis, 6- to 10-week-old C57BL/6 mice were inoculated i.v. into the tail vein with 75,000 B16-C215 cells in 0.2 ml of PBS containing 1% normal syngeneic serum. Treatment with Fab-SEA, Fab, SEA, or PBS (control) i.v. was initiated 1 day after tumor inoculation followed by additional doses on days 3, 5, and 7 as indicated. Mice were sacrificed after 3 weeks and lung metastases were counted. In the intraperitoneal tumor model, 3 x 103 B16-C215 cells were injected i.p. and therapy with FabC215-SEA or control vehicle was initiated i.v. 1 day later. The animals were observed daily and sacrificed when moribound, and the presence of tumor was confirmed by macroscopic examination.
elution buffer. Determination of Kd for Binding to the C215 and MHC Class ntigen. 251 labeling of SEA, FabC215-SEA, and C215 mAb was done by the lactoperoxidase method (Amersham). The binding of the labeled protein (15-50 ,uCi/,ug; 1 Ci = 37 GBq) was as unlabeled protein. The Kd value for binding to the C215 antigen was determined by a sandwich RIA in which mAb KS1/4, recognizing an epitope on the GA733-2 protein distinct from that recognized by mAb C215 (14), was used to coat eight-well microplate strips. Tumor extract containing C215 antigen (14) was incubated with the mAb-coated microtiter wells, 125I-labeled protein was added at appropriate dilutions and incubated with the C215 antigen, and bound radioactivity was measured in a y counter. To determine the Kd value for binding to MHC class II, a binding assay using Raji cells and 125I-labeled SEA or 125I-labeled Fab-SEA fusion protein was used (6). TCR a-Chain Binding Assay. The cell binding assay was performed as described (15). Briefly, [35S]methionine-labeled Raji or Colo2O5 cells were coated with SEA or FabC215SEA (2.8 !.&M) and then applied to a microtiter well containing an immobilized truncated recombinant murine TCR V(3 chain. After incubation and washing (15), the bound cells were lysed and the radioactivity was determined. The number of bound cells was calculated from the specific activity of labeled cells. The SD was routinely 95% homogeneity as determined by SDS/PAGE and reverse-phase HPLC (data not shown). Under nonreducing conditions, the protein migrated as an 82-kDa band that dissociated into the light chain (26 kDa) and Fd-SEA fusion protein (52 kDa) under reducing conditions. Binding to the C215 Antigen, MHIC Class U, and TCR. Saturation curves and Scatchard plots for binding of the FabC215-SEA fusion protein and the hybridoma mAb C215 to the C215 antigen demonstrated Kd values of -2 nM and 1 nM, respectively (Fig. 2A and B). Inhibition experiments with iodinated mAb C215 as a tracer confirmed that the monovalent FabC215-SEA protein and the bivalent hybridoma mAb C215 display an -2-fold difference in binding affinities (data not shown). Immunohistochemistry was used
Immunology: Dohlsten et al.
Proc. Natl. Acad. Sci. USA 91 (1994)
chains was detected with FabC215-SEA presented on MHC class II+ Raji cells but not when presented on C215+ Colo2O5 cells (Fig. 3C). This difference was not due to insufficient amounts of FabC215-SEA exposed on Colo2O5 cells, since strong binding was seen with an immobilized anti-SEA mAb (Fig. 3D). Cytotoxic T-Lymphocyte (CTL) Targeting of FabC215SEA. The fusion protein targeted SEA reactive CTLs to lyse the MHC class II-C215+ human colon carcinoma cell line SW620 (18) at 10 pM, whereas native SEA showed no effect, even at 10 nM (Fig. 4A). No cytotoxicity was seen with native mAb C215 (data not shown). To -demonstrate that lysis of MHC class II- target cells was dependent on expression of the relevant tumor antigen, we used CHO cells transfected with a cDNA encoding the human C215 antigen. The CHOC215 transfectant was highly sensitive to FabC215-SEAinduced lysis at effector/target ratios of 3:1 and 10: 1, whereas the nontransfected parental CHO cell line remained resistant even at an effector/target ratio of 30:1 (Fig. 4B). No cytotoxicity against CHO-C215 or CHO was induced by SEA alone. To analyze whether the reduced MHC class II binding of the fusion protein would influence functional presentation on a MHC class II+ target cell, we examined cytotoxicity against MHC class II+ C215- Raji B-lymphoma cells. SEA targeted cytotoxic T cells against Raji cells efficiently at 1 pM, whereas similar effects were seen at 10 pM FabC215SEA protein (Fig. 4C). Since SEA-induced cytotoxicity is influenced by expression of adhesion molecules on the target cell (16), we determined the effect of the fusion protein against target cells, differing only in the expression of transfected MHC class II or C215 molecules. The FabC215-SEA protein was -10-fold more potent against the CHO-C215 compared to the CHO-DR target cells (Fig. 4D). In Vivo Tumor Therapy with FabC215-SEA. A syngeneic mouse tumor model for the analysis of the effects of FabC215-SEA was constructed by transfection of B16 mouse melanoma cells with a plasmid encoding the C215 antigen and selection of stable B16-C215 transfectants. Administration of FabC215-SEA fusion protein i.v. strongly reduced the number of B16-C215 lung metastases (Fig. SA). For two out of four tested B16-C215 clones (clones D2 and 7.B6) >99% suppression of the number of formed metastases was recorded and >50% of the animals became tumor free. Treatment of the two other clones resulted in :95% (clone 5.A12) and 85% (clone B5) suppression of tumor growth. The effect of FabC215-SEA was dose-dependent with maximal tumor inhibition at 640 pmol (50 ug) per mouse and strong
B1l iI I
o) ri I;
C, h
K aNS I
Rfil
pKP865
Iss III
S1E II
I.'.
\mlal
Nto
FIG. 1. Outline of the expression vector pKP865 encoding the FabC215-SEA fusion protein. The orientation of the genes is indicated by arrows. The secretion of the H-chain-SEA fusion is directed by an OmpA-derived signal peptide, and K-chain secretion is directed by a synthetic signal peptide (signal peptides are hatched). The
transcription of the bicistronic mRNA is initiated from the trc promoter. The transcript ends at two phage fd transcription terminators (indicated by solid boxes). To keep the basal level of expression low, the lacI gene was included on the plasmid and the replication part from pBR322 was used. Km, kanamycin-resistance gene; CK and VK, K chain C and V region genes, respectively; CH1 and VH, H chain first C region and V region, respectively.
to confirm that the FabC215-SEA fusion protein retained the staining pattern of native mAb C215. FabC215-SEA (Fig. 3A) and mAb C215 (data not shown) showed homogeneous and strong staining of human colon carcinoma cells. In contrast, SEA stained exclusively cells in the stroma, probably reflecting infiltrating MHC class II' leukocytes (Fig. 3B). The MHC class U binding of the fusion protein was analyzed against MHC class II+ Raji B-lymphoma cells. Scatchard analysis demonstrated that the Kd value for binding of FabC215-SEA was 238 nM, whereas native SEA showed a Kd value of -.14 nM (Fig. 2 C and D). Competition experiments with fluorescein isothiocyanate-labeled SEA as a tracer confirmed that the fusion protein displayed an -.'.15-fold lower binding affinity compared to native SEA (data not shown). The FabC215-SEA protein may interact with three distinct structures, the C215 antigen, MHC class II molecules, and the TCR complex. To examine the TCR interaction, we analyzed binding of the FabC215-SEA protein to a TCR Vf33 chain. Binding to immobilized TCR V83 1.0
0.8
1.2
A.
i
0.5
0.
0
10
5
15
0.8
B. 4km
E O.
m
0
T (nM)
Kd=2.3 aM r= -0.97
It
0.018
C.
50
1.0
-
D.
;
2.
0.5 B (nM)
0 0.112a LO
25
T (nM)
Kd=1.3 nM r=40.98
0 I .5
0.5 1.0 B (nM)
0
0.012
0.112 0
A
0.006
Kd=288 DM r. -0.95 0
0.5
400 T (nM)
1.0 1.5 2.0 B (M)
mgo
80(
r= 8 *\
0
0.1A6
A
50
100
T (nJ)
1M
Kd=13.7 nM
0 2.5
8947
II
0
1
B (nM)
2
FIG. 2. Binding characteristics of FabC215-SEA (A), mAb C215 (n), and SEA (e) to the C215 antigen (A and B) and MHC class II+ Raji cells (C and D). (A and B) Representative Scatchard plots and saturation curves (Insets) for FabC215-SEA (A) and intact mAb C215 (B) from a C215 antigen RIA. (C and D) Corresponding diagrams for FabC215-SEA (C) and SEA (D) in the Raji cell assay. The Kd value and the regression coefficient, r, are shown. Each point represents the mean of triplicate samples. B, bound; F, free.
Immunology: Dohlsten et al.
8948 A
C 15 12 96
T0
13
0-
Raji
,'ii 800
B~~~~~~~~~~~~~~~a
r_F-If Raji
control
60'
T7CR 3
:::1
l'
n0 *8 40 20
2~~~~~ti
_
-
:::
_
''
_ _ _
_ _
O20
:::
_
_
A, lllll _
0
LL.-
Raji control
|: ': ''
_
lsC
_
-
80
SW620
Colo205 a-SEA
FIG. 3. (A and B) Immunohistochemical analysis of binding to human colon carcinoma tissue. Tissue sections were stained with FabC215-SEA (A) or SEA (B) at 100 nM followed by secondary reagents. (C) Binding of a soluble TCR VP chain to FabC215-SEA presented on MHC class II+ Raji and C215+ Colo2O5 cells. Radiolabeled cells were treated with FabC215-SEA or SEA and compared for binding to TCR V133 chains immobilized on plastic microtiter wells. Control refers to cell binding to uncoated wells. (D) Binding of an immobilized anti-SEA mAb (a-SEA) to FabC215-SEA presented on radiolabeled Raji and Colo2O5 cells. Control refers to cell binding to uncoated wells. In C and D, data for the cells bound are presented as cpm x 10-3.
ibound and sacrificed 30-45 days after tumor inoculation (Fig. 5F).
DISCUSSION Current strategies for immunotherapy include attempts to improve tumor-specific T-cell activation by vaccination with tumor cells transfected with genes encoding cytokines or costimulatory molecules, such as B7 (19, 20). These approaches have proven highly effective in experimental animal models, but they may be limited by the scarcity of T-cell epitopes expressed on human tumor cells. In contrast, numerous tumor-associated epitopes have been defined on human tumors by using mAbs. This led us to use tumorreactive mAbs for targeting of the extremely potent bacterial SAg SEA. In this paper we have used gene-transfer experiments to show that the specificity in Fab-SEA targeting against MHC class II- tumor cells is exclusively defined by the antigen reactivity of the mAb. This result is in conformity with earlier observations that SAg may interact with TCR
C
Raji
60
/ FabSEA
30
_
Raii X-SEA
reduction in the number of B16-C215 lung metastases at 64 pmol per mouse (Fig. SB). The treatment was well tolerated as no change in body weight or any other overt side effects were noted in animals treated with these doses of FabC215SEA. Systemic immune activation was not sufficient to inhibit tumor growth since treatment with SEA was without effect (Fig. 5C). The therapeutic effect was dependent on antigen-specific targeting, since a similar fusion protein based on the irrelevant mAb C242 did not mediate significant anti-tumor effects (Fig. 5D). The FabC242-SEA fusion protein showed a similar MHC class II-dependent binding and T-cell activation as FabC215-SEA (data not shown). Moreover, treatment with the FabC215 fragment alone failed to induce a significant anti-tumor effect (Fig. 5E). The FabC215-SEA fusion protein also mediated strong antitumor effects when the tumor was growing i.p. The survival of mice carrying B16-C215 cells i.p. was prolonged after treatment with 64 or 6.4 pmol of FabC215-SEA i.v. (Fig. 5F). A substantial fraction of the treated mice were long-term survivors (>90 days), whereas all untreated mice were mor50
::
_ _
|||l
(1994)
Colo2O5 TCR A
_
(C
Proc. Natl. Acad Sci. USA 91
40-
10
.5 x
0 0
i-14
20
~~SEA
1--
-13-12 -
-14 -13 -12 -11
-1 0
I0-
-9
-14
-13
-12
-1 1
-10
P
'HO-C215
'HO-DR
3
10
Effector/target ratio
'HO i Tv 10 ~9 0
log[compound]
FIG. 4. FabC215-SEA fusion protein directs CTLs against MHC class II-C215+ tumor cells. (A) Effects of the FabC215-SEA fusion protein (o) or SEA (A)
against the MHC class II- human colon carcinoma cell line SW620 in a 4-h Cr release assay at an effector/ target ratio of 30:1. (B) Effect of FabC215-SEA (4 nM) at various effector/target ratios against CHO-C215 transfectants or untransfected control CHO cells (dashed line). (C) FabC215-SEA and SEA targeting of CTLs against MHC class II+ Raji cells at an effector/ target ratio of 30:1. (D) Dose-response curves of FabC215-SEA-induced CTLs targeted against CHOC215, CHO-DR, and CHO cells at an effector/target ratio of 30:1.
Immunology: Dohlsten et al.
Proc. Natl. Acad. Sci. USA 91 (1994)
1I
PBS
Il 1 °''!i iFt1-ah s} 90. _._ ...
120
60-
~~~~~~~~~~~~~~~~~~~.... 40 -
0.)
30}-.
:. l3 . 11 .-
(
1
