Abstract
Purpose: The carcinoembryonic antigen (CEA) is a glycoprotein that is overexpressed in nearly 50% of all human and veterinarian tumors. At present, anti-CEA antibodies are being tested in clinical studies as passive immunotherapeutics. This study aims to establish an active immunotherapy for the poorly immunogenic CEA glycoprotein by generating antigen surrogates.
Experimental Design: We used the monoclonal anti-CEA antibody Col-1 and the biopanning method to generate peptide mimics (mimotopes) of the Col-1 epitope. The peptide showing the highest specificity and mimicry was synthesized as an octameric multiple antigenic mimotope (MAM). Subsequently, immunogenicity of the selected mimotope was examined in BALB/c mice. We assessed antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity mediated by the induced antibodies on CEA-expressing HT29 tumor cells. Furthermore, after immunization, the BALB/c mice were transplanted s.c. with Meth-A/CEA tumor cells.
Results: When BALB/c mice were immunized with this MAM, they generated a specific humoral immune response against CEA. The mimotope-induced polyclonal and poly-isotypic antibodies induced antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity in vitro. Furthermore, when MAM-immunized mice were transplanted s.c. with Meth-A/CEA cells expressing human CEA, a suppressed tumor growth was observed.
Conclusion: From our results, we can conclude that the Col-1 epitope of the glycoprotein CEA can be translated into an immunogenic peptide mimic. The mimotope-induced antibodies recognize CEA and do effectively inhibit growth of CEA-positive tumors. Based on these finding, we suggest that the generated mimotopes are candidates for active immunotherapy of CEA-expressing tumors.
Mammals have a sophisticated immune system that clears invading pathogens associated with danger signals. In contrast to this perception, tumor cells are generally well tolerated by the immune system and elicit only weak and ineffective immune responses (1). Reasons for this phenomenon are that tumor-associated antigens represent well-tolerated self-molecules of low antigenicity per se due to a high content of non-protein compounds including carbohydrates.
The carcinoembryonic antigen (CEA) represents a tumor-associated antigen frequently expressed on adenocarcinomas of various tissue origin and constitutes a 180-kDa glycoprotein with ∼50% carbohydrates (2, 3). CEA was identified for the first time in 1965 in colorectal cancer (4, 5) and has since been found in 80% to 90% of all colorectal cancers as well as in up to 50% of all breast cancers, in non–small-cell lung cancers, in esophageal, pancreatic, and gastric malignancies, as well as in malignant tumors of the thyroid and female reproductive tract (6). Levels of CEA are up to 100 times higher in tissue samples derived from malignant tumors as compared with healthy tissues (7). Due to its high and specific expression, CEA is an attractive target for immunotherapy (6). Thus, many different immunologic approaches based on CEA have been pursued (8). Recently, the humanized anti-CEA monoclonal antibody (mAb) labetuzumab has entered clinical phase II studies (9) based on clinical experience that mAb therapy represents an important therapeutic strategy for the treatment of selected malignancies (10–13). However, passive antibody applications have several drawbacks. Not only the antibodies have to be engineered as human/mouse chimeric or fully humanized antibodies to prevent production of human anti-mouse antibodies, but also repeated administrations are necessary to obtain adequate serum levels for an appropriate antitumor effect. Therefore, active immunotherapy may be the overall preferable approach (14–16) due to the emergence and continuous induction of “natural,” polyclonal antitumor antibodies. It can be hypothesized that such antibodies would not be restricted to a single isotype antibody (sub)class but would instead initiate the whole spectrum of antibody-mediated immune functions against tumors.
To archive this goal, we decided to develop a mimotope-based vaccine. As indicated by previous studies (15, 17), mimotopes have the advantage to imitate antigenic structures without being identical, indicating that they may be ideal antigen surrogates in tumor immunology able to break tolerance toward self. Mimotopes can be created without detailed structural or sequential information, can be produced synthetically, and were revealed to induce the desired antibody specificity by immunizations (15, 16). Mimotopes may mimic discontinuous and even non-protein epitopes (17). Thus, in the present study, we aimed to develop a highly specific mimotope peptide cross-reacting with a motif of the CEA for active immunizations.
Materials and Methods
mAbs. Col-1, an immunoglobulin G (IgG)-2a mAb, was purchased from Zymed. The isotype control IgG2a was obtained from Sigma and used as a control antibody.
Phage library and biopanning. Three successive rounds of biopanning with the anti-CEA antibody Col-1 were done with two pIII-display peptide phage libraries: CL10 and LL9. CL10 expresses cysteine-flanked decapeptides circularized by disulfide bridges and LL9 comprises nine linear peptides. The libraries were applied separately or pooled for biopannings. They were kindly provided by Prof. Dr. Luca Mazzucchelli (University of Bern, Bern, Switzerland; ref. 18). Biopannings were done as outlined in ref. 19, with slight modifications. In short, ELISA plates (Nunc) were coated with 40 μg/mL Col-1 in bicarbonate buffer (pH 8.5). Wells were blocked with PBS/1% dry milk and incubated with an aliquot of the phage library (∼5 × 1010 phage particles) in PBS/1% dry milk/0.1% Tween 20 at 37°C and 4°C for 1 h each. Unbound phage particles were removed by extensive washing with PBS/0.5% Tween 20. Bound phage peptides were eluted with 0.1 mol/L glycine (pH 2.2) and immediately neutralized. Phages were amplified in Escherichia coli K91, precipitated from the bacterial culture supernatant with polyethylene glycol, and either immediately used for the next round of biopanning or stored at −20°C. For titer determination, aliquots of the eluate or amplificate were plated in serial dilution on Luria broth (LB) agar plates containing 75 μg/mL kanamycin.
Amplification, preparation, and titration of phages.E. coli K91 were grown in LB medium to an A600 of 0.8 and infected with eluted phages by incubation at room temperature for 15 min. The infected K91 culture was grown for 5 h at 37°C in LB selection medium containing 1 μg/mL kanamycin. Bacteria were spun down and phage particles were precipitated from the supernatant with 20% polyethylene glycol 8000 (Amresco) in 4 mol/L NaCl on ice for 1 h. Phage particles were centrifuged at 1,800 × g for 30 min and pellets resuspended in 10 mmol/L HEPES/150 mmol/L NaCl. Supernatants were subsequently used for the next biopanning round or stored at −20°C. For determination of phage titer, a colony forming units (cfu) test was done; aliquots of the eluate were used to infect E. coli K91 and plated in serial dilution on LB agar plates containing 75 μg/mL kanamycin.
Colony screening assay. Colony screenings for the selection of specific phage clones were done according to Barbas et al. (20). Immunoscreenings were done with Col-1 and the IgG2a control antibody diluted to 1 μg/mL in PBS/0.15% casein. Bound antibodies were detected with 125I-labeled antimouse IgG (IBL GmbH). Filters were washed, dried, and exposed to Biomax-MS films (Kodak) at −70°C. Positive clones were amplified as described above.
Phage sequence analysis and peptide alignment. Single-stranded phage DNA was prepared using Qiaprep Spin M13 kit (Qiagen). The amount of prepared DNA was checked with an EtBr2/0.7% agarose gel under UV illumination. DNA sequencing was done by VBC Biotech Service GmbH.
Specificity ELISA. ELISA plates (Nunc) were coated with Col-1 or the IgG2a antibody, 0.1 μg/mL overnight at 4°C in carbonate buffer (pH 9.6). Plates were washed with PBS/0.05% Tween 20 and nonspecific binding was blocked with PBS/1% dry milk for 2 h (37°C and 4°C). Phage clones were added at a concentration of 1 × 106 cfu/mL in PBS/0.1% dry milk overnight. Bound phage particles were detected with peroxidase-conjugated mouse anti-phage M13 mAb at a dilution of 1:1,000 (Amersham Pharmacia Biotech). The reaction was developed with 3,3′,5,5′-tetramethylbenzidine substrate (BD Biosciences). Absorbance was measured with an ELISA reader (Dynatech) at 450 and 630 nm.
Mimicry ELISA. ELISA plates (Nunc) were coated with 0.5 μg/mL purified human CEA (Sigma) in carbonate buffer (pH 9.6) at 4°C overnight (and blocked as described above). Phage particles in three different concentrations (5 × 1010, 1 × 1010, and 1 × 109 cfu/mL) and Col-1 antibody were incubated simultaneously at 37°C and subsequently at 4°C each for 1 h. After washing, bound phages were detected as described above.
Synthesis of the multiple antigenic mimotope. The linear peptide mimotope DRGGLWKTP was selected for synthetic production of the octameric peptide (DRGGLWKTP-GG)4-(KKGGC)2-dithioacetylhexandiamin (piChem) as a so-called multiple antigenic peptide (MAP; ref. 21), which was further designated multiple antigenic mimotope (MAM). Antigenic accuracy was verified in an ELISA, where the MAM and, for control purposes, an irrelevant linear MAP [(QYIKANSKFIGITEL)]4-(K)2KGC2, which stands for a promiscuous T-cell epitope (22), were coated onto an ELISA plate (Nunc) at a concentration of 10 μg/mL by incubation overnight at 4°C. Col-1 and an IgG2a isotype control (Sigma) were added at a concentration of 1 μg/mL in PBS/0.1% dry milk. After washing, specific bound antibodies were detected as described in “Specificity ELISA”.
Immunization of BALB/c mice. Three groups (n = 8) of BALB/c mice (Institute of Laboratory Animal Science and Genetics, Austria) were immunized i.p. with 180 μg of MAM and control-MAP, both absorbed to Al(OH)3 as an adjuvant or were left naive. Mice were immunized on days 1, 15, 28, and 43. After 4 months without immunization, mice were again immunized with the respective antigen or the T-cell epitope on days 215, 232, and 261. Blood was taken from the tail vein on days 0 (preimmune serum), 12, 26, 41, 54, 225, 246, 274, and 288. All animals were treated according to European Union Rules of Animal Care, with permission number BMBWK-66.009/0153-BrGt/2005 from the Austrian Federal Ministry of Education, Science, and Culture.
Antibody isotype determination after mimotope immunization. ELISA plates (Nunc) were coated with CEA (0,5 μg/mL) or CEA-MAM or the control MAP (1 μg/mL) overnight at 4°C in carbonate buffer (pH 9.6). Unspecific binding was blocked with TBS/1% bovine serum albumin for 2 h (37°C and 4°C). Sera of immunized mice were added at a concentration of 1:100 in TBS/0.1% bovine serum albumin overnight. Bound antibodies were detected with peroxidase-labeled anti–total immunoglobulin or with unlabeled anti-IgM, anti-IgG1, anti-IgG2a, or anti-IgG2b mAb (Amersham Pharmacia Biotech) at a dilution of 1:500 for 2 h (37°C and 4°C). The latter antibodies were detected with peroxidase-conjugated anti-rat IgG (Amersham Pharmacia Biotech) by development with 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid). Absorbance was measured in an ELISA reader (Dynatech) at 405 and 490 nm and antibody levels were calculated according to a standard curve.
Immunofluorescence. HT29, a CEA-positive colon adenocarcinoma cell line, was plated at a concentration of 105 cells/mL on tissue adhesion slides (Marienfeld Superior, Laboratory Glassware). SW480 colon cancer cells, being CEA negative, were used as a negative control. Slides were incubated with the cells for 20 min, washed twice with PBS, and fixed with 4% paraformaldehyde for 30 min. Thereafter, slides were incubated with 50 mmol/L NH4Cl in PBS to quench followed by washing twice with PBS. Cells were incubated with pooled preimmune and seventh mouse immune serum. Pooled sera from the other groups were used as controls. Cells incubated with Col-1 served as positive control. Bound antibodies were detected with FITC-conjugated goat anti-mouse IgG (Alexa Fluor 488, Invitrogen, Molecular Probes). Nuclei were stained with 0.5 μg/mL Hoechst dye (Sigma) in PBS for 10 min. Cells were mounted in Immuno-Fluoro-Mounting Medium and viewed under an AxioVert 200 microscope (Carl Zeiss).
Complement-dependent cytotoxicity and antibody-dependent cellular cytotoxicity. The complement-dependent cytotoxicity (CDC) and antibody-dependent cellular cytotoxicity (ADCC) of the antibodies induced by mimotope vaccination were measured with the CytoTox 96 Nonradioactive Cytotoxicity assay (Promega). CEA-overexpressing HT29 cells were used as positive target cells. CEA-negative colon cancer cells (SW480) served as a negative control. The number of both target cells was optimized in preceding experiments to 2 × 105/mL. For CDC and ADCC reactions, pooled fifth and seventh immune sera were diluted 1:50 in CytoTox 96 assay medium. Diluted serum from the animals immunized with the control MAP or the naïve group (both diluted 1:50) or the Col-1 antibody served as controls. Spleen cells of naïve BALB/c mice used as effector cells were prepared by mashing the spleen and lysing the erythrocytes with ammonium chloride. To evaluate ADCC data, we subtracted CDC from ADCC values obtained in the same assay. All assay procedures and readouts were done as described by the manufacturer. Assays were done in triplicates. The results of the cytotoxicity assays were calculated as follows:
.
Tumor cell injection and histopathology. Meth-A/CEA cells, a CEA-transfected sarcoma cell line established from mice with the human CEA transgene (23), were cultured in RPMI 1640 with 10% heat inactivated FCS (PAA Laboratories), 2 mmol/L l-glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, nonessential amino acids, and 1 mmol/L sodium pyruvate (Invitrogen). Cells were loosened with Na-EDTA and 107 tumor cells were washed thrice in 1-mL PBS, taken up in 50 μL, and injected into the shaved right flank of each mouse. Experimental groups consisted of four to six mice. Tumor development was controlled daily by serial measurements of tumor size; the tumor volume was calculated according to the following equation: tumor volume (mm3) = d2 × D / 2, with d as the shortest and D as the longest diameter. Animals were euthanized when the tumor reached a volume of more than 300 mm3. Tumor sections were fixed in 10% buffered formalin, processed, and embedded in paraffin. Four-micrometer sections were H&E stained and examined under a light microscope (Olympus BH2). Micrographs were taken at magnifications of ×100 (middle pictures) and ×400 (bottom pictures) using an Olympus digital camera.
Statistical analysis. Statistical comparisons were done with the nonparametric Mann-Whitney U test using the SPSS 15.0 program (SPSS, Inc.) and P < 0.05 was considered significant.
Results
Biopanning and colony screening. Four rounds of biopanning with the anti-CEA antibody Col-1 resulted in increased phage titers from the first to the fourth round, indicating amplification of phage clones (termed COL1-COL7) expressing specific peptide ligands (Table 1). The phage clones with the best specificity and binding capacity to Col-1 were amplified and DNA sequencing was done. Deduced amino acid sequences were subjected to the multiple algorithm of the Vector NTI Suite and a phylogenetic tree was created indicating the degree of similarity between the peptides (Fig. 1A). In almost all the peptides, the amino acid motif GGL was seen.
Monitoring of Col-1–specific phage titers during four rounds of biopanning with libraries CL10 and LL9
. | Rounds . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | 1 . | 2 . | 3 . | 4 . | ||||
Libraries | ||||||||
CL10 (cfu/mL) | 1 × 104 | 5 × 105 | 2 × 106 | 4 × 107 | ||||
LL9 (cfu/mL) | 3 × 104 | 3 × 106 | 3 × 106 | 5 × 107 |
. | Rounds . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | 1 . | 2 . | 3 . | 4 . | ||||
Libraries | ||||||||
CL10 (cfu/mL) | 1 × 104 | 5 × 105 | 2 × 106 | 4 × 107 | ||||
LL9 (cfu/mL) | 3 × 104 | 3 × 106 | 3 × 106 | 5 × 107 |
Selection and characterization of the CEA mimotope. A, amino acid sequences of peptide inserts identified with the anti-CEA antibody Col-1 in biopanning procedure were subjected to the alignment algorithm of the Vector NTI suite. Their origin from the circular (CL10) or linear (LL9) phage library is indicated in parentheses. The phylogenetic tree indicates the degree of similarities of the peptides to each other. B, in a sandwich assay, phage clones were bound by coated anti-CEA antibody Col-1 and detected by a peroxidase-labeled rabbit anti-phage antibody to determine specificity of peptides. No phage binding occurred to isotype control antibody. C, mimicry analysis of CEA-peptides was done by an ELISA competition assay with coated CEA antigen and Col-1 as the detection antibody. Simultaneous incubation was done with titrated specific or control phages (solid columns, 5 × 1010 cfu/mL; horizontally hatched columns, 1 × 1010 cfu/mL; open columns, 1 × 109 cfu/mL concentration of phage clones) and the competition for Col-1 binding was measured. D, the synthetically produced octameric MAM of the COL2 mimotope was checked for specific recognition by the Col-1 antibody in ELISA. The MAM or an irrelevant control MAM was coated and incubated with the CEA-specific Col-1 (solid columns) or with the isotype control (open columns).
Selection and characterization of the CEA mimotope. A, amino acid sequences of peptide inserts identified with the anti-CEA antibody Col-1 in biopanning procedure were subjected to the alignment algorithm of the Vector NTI suite. Their origin from the circular (CL10) or linear (LL9) phage library is indicated in parentheses. The phylogenetic tree indicates the degree of similarities of the peptides to each other. B, in a sandwich assay, phage clones were bound by coated anti-CEA antibody Col-1 and detected by a peroxidase-labeled rabbit anti-phage antibody to determine specificity of peptides. No phage binding occurred to isotype control antibody. C, mimicry analysis of CEA-peptides was done by an ELISA competition assay with coated CEA antigen and Col-1 as the detection antibody. Simultaneous incubation was done with titrated specific or control phages (solid columns, 5 × 1010 cfu/mL; horizontally hatched columns, 1 × 1010 cfu/mL; open columns, 1 × 109 cfu/mL concentration of phage clones) and the competition for Col-1 binding was measured. D, the synthetically produced octameric MAM of the COL2 mimotope was checked for specific recognition by the Col-1 antibody in ELISA. The MAM or an irrelevant control MAM was coated and incubated with the CEA-specific Col-1 (solid columns) or with the isotype control (open columns).
Mimotope characterization. An ELISA was done to confirm the colony screening results in a separate test system. Seven independent phage clones were specifically bound by the antibody Col-1 but were not recognized by an isotype control (Fig. 1B). No reactivity was observed with the wild-type phage as control. In this “Specificity ELISA,” the phage-displayed peptide COL2 had the highest reactivity to the antibody Col-1.
To prove the mimicry potential of selected phage mimotopes with the original antigen CEA, a competitive ELISA was done (Fig. 1C). When the Col-1 antibody was added to wells simultaneously with titrated phage concentrations, the amount of antibody binding to the original CEA antigen could be reduced in a dose-dependent manner. A control phage displaying an irrelevant peptide did not compete with the plate-bound CEA for Col-1 binding, even at the highest dose. Thus, the mimotopes displayed distinct competition potential with CEA, with clone COL4 being the best circular mimotope candidate and COL2 the best linear. This assay evidenced that the selected mimotopes were true mimics of the Col-1 epitope on the CEA antigen.
Synthetic production of MAMs. The sequence DRGGLWKTP of linear mimotope clone COL2 was selected for synthetic production of the octameric multiple antigenic mimotope (MAM; (DRGGLWKTP)4-KKGGC)2-dithioacetylhexandiamin. The correct fold of the MAM was controlled via Col-1 binding analysis in ELISA. Figure 1D shows that coated MAM is specifically recognized by Col-1 but not by the isotype control antibody. As an additional control, an irrelevant, linear control MAP was coated and remained negative.
Immune response induced by mimotope immunization. To evaluate the MAM immunization experiments in BALB/c mice, sera were screened for reactivity toward the MAM or toward the control MAP. As depicted in Fig. 2A, increasing antibody titers were observed only toward the CEA-MAM but not toward the control-MAP. Interestingly, the mice immunized with the CEA-MAM developed antibodies recognizing also the human CEA, which were mainly of the IgM and IgG2b subclass, whereas only limited amounts of CEA-specific IgG1 and IgG2a were detected (Table 2).
Induction of CEA-specific antibodies by CEA-MAM immunizations. A, sera of immunized mice were tested for binding to CEA-MAM (rhombus) and to the irrelevant control MAP (square). Murine preimmune serum (PIS) to the seventh mouse immune serum (Mis) taken during the immunization period were evaluated for antigen-specific total immunoglobulin (Ig) and the values were calculated according to the standard curve (ng/mL). B, in immunofluorescence staining, sera from CEA-MAM immunized mice revealed membrane staining of CEA-positive HT29 cells similar to the positive control Col-1, whereas no binding was observed with sera taken before immunization.
Induction of CEA-specific antibodies by CEA-MAM immunizations. A, sera of immunized mice were tested for binding to CEA-MAM (rhombus) and to the irrelevant control MAP (square). Murine preimmune serum (PIS) to the seventh mouse immune serum (Mis) taken during the immunization period were evaluated for antigen-specific total immunoglobulin (Ig) and the values were calculated according to the standard curve (ng/mL). B, in immunofluorescence staining, sera from CEA-MAM immunized mice revealed membrane staining of CEA-positive HT29 cells similar to the positive control Col-1, whereas no binding was observed with sera taken before immunization.
CEA-specific antibody quantities in sera after the seventh CEA-MAM immunization
Antibody subclass . | Mean ± SD (ng/mL) . |
---|---|
lgM | 28.57 ± 9.63 |
lgG1 | 5.57 ± 1.27 |
lgG2a | 9.14 ± 2.12 |
lgG2b | 25.50 ± 6.22 |
Antibody subclass . | Mean ± SD (ng/mL) . |
---|---|
lgM | 28.57 ± 9.63 |
lgG1 | 5.57 ± 1.27 |
lgG2a | 9.14 ± 2.12 |
lgG2b | 25.50 ± 6.22 |
Immunofluorescence. To further show that the induced antibodies were able to recognize CEA on the cell surface, immunofluorescence staining was done. Only sera of mice immunized with the CEA mimotope revealed membrane staining of CEA-expressing HT29 (Fig. 2B) similar to the staining pattern observed with the positive control Col-1. However, no antibody binding was observed with our CEA-negative control SW480 cells or when carrying out the staining with sera of mice immunized with a control mimotope or the naïve group (data not shown).
Functional analysis of the induced antibodies. We evaluated the effector function of the serum antibodies induced by CEA mimotope immunizations against CEA-positive target cells (HT29) and CEA-negative cells (SW480) and compared to the results to the Col-1 antibody, and to serum of the control mouse groups. We could clearly show effective CDC reaction mediated by the induced antibodies on HT29 cells (Fig. 3A). The concentration dependency and specificity could be proved because sera of control MAP-immunized mice or naïve mice and even the Col-1 antibody did not show any cytotoxic capacity. We even observed an increase of cytotoxic potential during the immunization course as the fifth immune serum achieved 10.29% cytotoxicity, whereas the seventh immune serum revealed 41.18% of cell lysis. Moreover, the induced antibodies were able to show an effective ADCC reaction (Fig. 3B). Only sera from the CEA MAM-immunized mice showed 9.56% cytotoxicity for the fifth and 8.82% for the seventh mouse immune sera against the HT29 cell line. Neither the group immunized with the irrelevant mimotope nor the naïve control group were able to trigger an antibody-dependent cytotoxic reaction. In addition, no reactions were observed against SW480 cells.
Cytotoxicity effects induced by mimotope immunizations against CEA-positive tumor cells. Cell killing activity was determined against CEA-positive HT29 and CEA-negative SW480 cells. A, for CDC reactions, mouse immune sera from CEA-MAM immunized animals taken after the fifth (1) and seventh (2) immunizations were used. Sera of the mice immunized with the control mimotope (3), sera of the naïve mice (4), and the control antibody Col-1 (5) served as controls. B, for ADCC reactions, sera from CEA-MAM immunized animals after the fifth (1) and seventh immunizations (2). Sera from mice immunized with the control-MAP (3), the naïve mice (4), and the Col-1 antibody (5) were used as controls.
Cytotoxicity effects induced by mimotope immunizations against CEA-positive tumor cells. Cell killing activity was determined against CEA-positive HT29 and CEA-negative SW480 cells. A, for CDC reactions, mouse immune sera from CEA-MAM immunized animals taken after the fifth (1) and seventh (2) immunizations were used. Sera of the mice immunized with the control mimotope (3), sera of the naïve mice (4), and the control antibody Col-1 (5) served as controls. B, for ADCC reactions, sera from CEA-MAM immunized animals after the fifth (1) and seventh immunizations (2). Sera from mice immunized with the control-MAP (3), the naïve mice (4), and the Col-1 antibody (5) were used as controls.
Tumor challenge. Evidence for in vivo functionality of the MAP vaccination came from a challenge of mice with Meth-A/CEA tumor cells. We examined tumor growth in mice immunized with the CEA MAM compared with a group immunized with a control MAP or naïve animals, after transplant challenge with 107 live tumor cells per mouse. The mice receiving an irrelevant mimotope or no treatment showed accelerated tumor growth (Fig. 4B and C). Until day 7, a tumor size of 300 mm3 was reached and animals had to be sacrificed. In contrast, mice immunized with the CEA mimotope showed no or only very small (0-13.5 mm3) tumor formation. Only one mouse in the CEA immunized group developed a tumor with a size of 87.5 mm3, possibly due to the fact that this mouse hardly revealed any antibody titer toward CEA (Fig. 4A). Although tumor sizes were comparable on day 1, validating the tumor transplants, statistical comparisons revealed a significant difference between MAM-immunized animals and the control mice from day 4 until sacrifice on day 7 (P = 0.01 or P = 0.006, respectively). Differences in vascularization were observed macroscopically. Mice from the control groups presented blood vessel supplying the tumor (Fig. 4B and C, macroscopic pictures) whereas only very small blood vessels could be verified around the tumor transplants from CEA mimotope–immunized mice (Fig. 4A). The histologic results supported these observations. Mice immunized with the CEA mimotope showed a very limited number of tumor cells at the site of tumor transplantation. Instead, necrotic regions with an accumulation of eosinophilic and neutrophilic granulocytes were found with signs of inflammation and a tendency of encapsulation (Fig. 4A, magnifications of ×100 and ×400). Control immunizations were not able to inhibit tumor growth, resulting in proliferation of tumor cells with invasive tumor growth similar to the situation in the naïve animals (Fig. 4B and C, magnifications of ×100 and ×400).
Murine Meth-A/CEA tumor transplant model. The in vivo outcome after tumor challenge was assessed in BALB/c mice immunized with the CEA-MAM (A), in mice injected with an irrelevant control mimotope (B), and in naïve animals (C). After transplanting Meth-A/CEA tumor cells, the tumor size was controlled on a daily basis until a tumor volume of 300 mm3 was reached. The diagrams show the effects of the treatment through the volume of tumor development (y-axis) during the time course of 1 wk (x-axis). Differences in tumor transplant sizes and vascularization was observed among the three animal groups macroscopically (top picture). Sections of the Meth-A/CEA tumors were stained with H&E and shown at ×100 (middle picture) and ×400 (bottom picture) magnifications, indicating that mimotope vaccination inhibits the settling of Meth-A/CEA cells through inflammation, whereas sham or nontreated animals reveal flourishing tumor cell proliferation.
Murine Meth-A/CEA tumor transplant model. The in vivo outcome after tumor challenge was assessed in BALB/c mice immunized with the CEA-MAM (A), in mice injected with an irrelevant control mimotope (B), and in naïve animals (C). After transplanting Meth-A/CEA tumor cells, the tumor size was controlled on a daily basis until a tumor volume of 300 mm3 was reached. The diagrams show the effects of the treatment through the volume of tumor development (y-axis) during the time course of 1 wk (x-axis). Differences in tumor transplant sizes and vascularization was observed among the three animal groups macroscopically (top picture). Sections of the Meth-A/CEA tumors were stained with H&E and shown at ×100 (middle picture) and ×400 (bottom picture) magnifications, indicating that mimotope vaccination inhibits the settling of Meth-A/CEA cells through inflammation, whereas sham or nontreated animals reveal flourishing tumor cell proliferation.
Discussion
CEA represents an interesting target for antitumor immunotherapy because it is specifically and highly expressed on many different malignancies. Thus, different therapeutic strategies targeting CEA including CEA-pulsed dendritic cells as well as vaccines using anti-idiotypic antibodies, recombinant CEA, DNA, or CEA-associated peptides are under investigation (8, 24). A further prominent example is the development of the anti-CEA antibody labetuzumab, which has recently entered phase II clinical trials (9, 25–27).
In the present study, we aimed to develop an active vaccine strategy inducing a long-lasting antibody response with high immunogenicity simultaneously circumventing or breaking tolerance mechanisms toward “self.” Using the Col-1 antibody, we selected epitope mimics from two phage display peptide libraries as potential antigen surrogates of CEA based on previous experience from other laboratories and ours, proving mimotopes to be suitable tools for the generation of specific active antitumor responses (15, 17, 28–31).
After biopanning, successful peptide selection was ascertained by the increase of total and specific phage titers from subsequent rounds of the selection process. The specificity of the single mimotopes and the potential of the phage-displayed mimotopes to interfere with Col-1 binding to CEA in a dose-dependent manner were proved, indicating molecular mimicry of the original antigen. A lack of sequence similarities between the peptides and natural antigen (i.e., CEA) suggested the Col-1 epitope to be of conformational nature and exhibiting similar electron clouds as the peptides. In this context, it is notable that the Col-1 antibody recognizes an epitope on the CEA molecule, which is expressed exclusively on premalignant or malignant cells in a large number of human carcinomas (32–34).
Because small peptides are not immunogenic per se and may even induce tolerance, a suitable mimotope presentation had to be chosen for vaccine formulation. The sequence DRGGLWKTP of the linear mimotope clone COL2 was selected, enabling a straightforward synthesis of an octameric MAP (21) with a tyrosine spacer as the sole carrier allowing a dense display of eight peptide copies. The resulting product with eight linear epitope mimics of Col-1 was termed MAM. It is technically of importance that mimotopes have to adapt the same structure in the vaccine as displayed by the phage during selection. Correct folding could be confirmed by specific Col-1 binding. Mimotopes were revealed to represent pure B-cell epitopes, with T cells only providing bystander help in vaccination (35). Thus, we focused on antibody determination and on investigation of their potential in purified. CEA MAM had the potential to elicit the production of not only antibodies recognizing the mimotope itself but also human CEA as well as CEA expressed on HT29 colon cancer cells. Similar to the in vitro function of the anti-CEA antibody labetuzumab (26), the antibodies induced by mimotope immunization also had distinct cytotoxic functions against CEA-positive tumor cells. Interestingly, their killing potential was significantly higher compared with the Col-1 IgG2a antibody directed toward the same structure in in vitro and in vivo experiments, possibly due to epitope-spreading and the poly-isotype nature of the actively induced antibodies. To prove the biological relevance for the in vivo situation, we transplanted MAM-immunized mice with live CEA-expressing tumor cells (23). Whereas mice receiving an irrelevant mimotope or being left naive showed considerable tumor growth, reaching a tumor size of 300 mm3 on day 7 after tumor transplant, mice immunized with the CEA mimotope had significantly reduced tumor formation. Moreover, differences in tumor vascularization were noted. Mice of the control groups had a dense blood vessel supply of the transplanted tumor, whereas an only very scarce vascularization was seen around tumor transplants in mice immunized with the CEA mimotope. Histologically, the necrotic regions in immunized mice showed no tumor cells but signs of inflammation and a tendency of encapsulation. In contrast, in the control mice, tumor growth was not inhibited, showing proliferation of tumor cells with invasive tumor growth. Thus, the specificity of CEA mimotope immunizations was shown because treatment with an irrelevant mimotope or no treatment remained completely ineffective in suppression of tumor growth.
The present study demonstrates the successful use of the antibody Col-1 for peptide selection from a phage display library resulting in the selection of true B-cell epitope mimics for the antigen CEA. Based on the immunization experiments, we suggest that MAMs are capable of recruiting T-cell bystander help, induce memory phenomena, and launch functional antitumor immune reactions in vitro and in vivo. Thus, our future working hypothesis is that vaccination with the CEA mimotope is likely to be most effective at lower tumor load, applied as prophylaxis or for treatment of minimal residual disease, which might be ideal situations for a clinical study in human tumors overexpressing CEA.
Grant support: Fund of the City of Vienna for Innovative Interdisciplinary Cancer Research and Fund of the Mayor of the City of Vienna grant 023484, BioLife Science (S. Gruber), and Austrian Science Fund grant 18238 (R. Knittelfelder).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Conflict of interest: O. Scheiner and C. Zielinski are stake holders of BioLife Science GmbH, Vienna, Austria.