Designing antibodies for the inhibition of gastrin activity in tumoral cell lines
Version of Record online: 25 JAN 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 122, Issue 10, pages 2351–2359, 15 May 2008
How to Cite
Barderas, R., Shochat, S., Timmerman, P., Hollestelle, M. J., Martínez-Torrecuadrada, J. L., Höppener, J. W.M., Altschuh, D., Meloen, R. and Casal, J. I. (2008), Designing antibodies for the inhibition of gastrin activity in tumoral cell lines. Int. J. Cancer, 122: 2351–2359. doi: 10.1002/ijc.23395
- Issue online: 17 MAR 2008
- Version of Record online: 25 JAN 2008
- Manuscript Accepted: 9 NOV 2007
- Manuscript Received: 7 MAY 2007
- EU grant. Grant Number: COOP-CT-2004-512691
- pancreatic cancer;
- antibody libraries;
- antibody repertoire;
Gastrin and its derivatives are becoming important targets for immunotherapy of pancreatic, gastric and colorectal tumors. This study was conducted to design antibodies able to block gastrin binding to the gastrin/cholecystokinin-2 (CCK-2) receptor in order to delay tumor growth. The authors have used different gastrin molecules, combined with the diphtheria toxoid, to generate and select human single chain variable fragments (scFvs) as well as mouse monoclonal antibodies and scFvs against different regions of gastrin. There was a remarkable conservation in the antibody repertoire against gastrin, independently of the approach and the species. The germlines most frequently used in gastrin antibody formation were identified. Three different epitopes were identified in the gastrin molecule. The resulting mouse monoclonal antibodies and scFvs were analyzed for gastrin neutralization using Colo 320 WT cells, which overexpress the CCK-2 receptor. The gastrin neutralizing activity assay showed that N-terminal specific mouse monoclonal antibodies were more efficient to inhibit proliferation of Colo 320 WT cells than the anti-C terminal antibodies. Moreover, the human antigastrin scFvs obtained in this study inhibited significantly the proliferation of Colo 320 tumoral cells. These findings should contribute to a more rational design of antibody-based antigastrin therapies in cancer, including passive administration of human antibodies with blocking activity. © 2008 Wiley-Liss, Inc.
The peptide hormone gastrin is gaining importance as a therapeutic target because of its activity as a tumor growth factor. This hormone is initially produced as a pre-pro hormone that is proteolytically processed in several steps during its maturation process,1 including posttranslational modifications as amidation, sulfation or phosphorylation. Amidated gastrin17 (G17), the final molecule derived from preprogastrin, is a trophic factor for several cancers through autocrine, paracrine and endocrine mechanisms.2 Gastrin, together with its receptor CCK-2 (also known as gastrin/CCK-B receptor), is overexpressed in pancreatic,3, 4 gastric and colorectal cancers.5–7 Gastrin also plays a role as antiapoptotic agent through the upregulation of Bcl-2 and survivin,8 activation of the protein kinase B/AKT phosphorylation pathway and the transcriptional factor NFκB.9, 10
Gastrin neutralization is being used in different clinical trials as a therapy for several tumors by blocking the interaction with CCK-2. In fact, Gastrimmune®, a peptide composed of the first 9 amino acids of G17 coupled to diphtheria toxoid (DT),11 is a vaccine in clinical Phase II for pancreatic,12 colorectal and gastric cancer patients.13, 14 The antibodies generated in patients were capable of inhibiting partially the trophic effect of this hormone. Only vaccine responders showed an improved survival, demonstrating the potential benefits of this approach.12, 13 Thus, the possibility of using passive vaccination (with human antibodies) could be an alternative to active immunization in order to solve the problems associated with nonresponders to gastrin vaccination.
A deeper understanding of gastrin antigenic structure and the corresponding antibody repertoire should be gained in order to develop a possible therapeutic replacement for active vaccination. We have generated a panel of human and mouse antigastrin antibodies and antibody fragments and characterized their sequences and gastrin binding properties. Additionally, we have analyzed their ability to inhibit gastrin binding to its receptor in order to evaluate their potential for functional neutralization.
For immunization and antibody selection, we used different gastrin molecules coupled to DT, the same carrier used in Gastrimmune. These molecules were G17 and gastrin 12 (containing the N-terminal 12 amino acids of G17) coupled either by the N-terminus or the C-terminus to DT. To get the highest possible antibody diversity, we prepared an extensive collection of human and mouse antibodies against gastrin using 3 different strategies (Fig. 1): (i) phage display human antibody libraries to isolate human single chain variable fragments (scFvs), (ii) generation of antigastrin mouse monoclonal antibodies (Mabs) and scFvs and (iii) generation of scFv libraries from the spleen of gastrin-immunized mouse.
Using these approaches we have obtained a full repertoire of human and mouse antibodies against gastrin, which enabled the mapping of the antigenic sites of G17 and the characterization of the antibody repertoire against gastrin. Moreover, the capacity of the different antibodies to inhibit gastrin-induced proliferation of Colo 320 WT cells, a human colorectal cancer cell line derived from the Colo 320 cell line, which overexpresses the CCK-2 receptor,15 was determined.
Material and methods
Peptides and antibodies
G17 (pEGPWLEEEEEAYGWMDF-NH2, where pE means pyroglutamic acid), biotinylated G17 (pEGPWLEEEEEAYGWMDFK(biotine)-NH2) and G17-C/G12-C (pEGPWLEEEEEAYGWMDFC-NH2 and pEGPWLEEEEEAYC-NH2) coupled to DT were provided by Pepscan Therapeutics (Lelystad, The Netherlands). G17-[125I] Tyr12- (81.4 TBq/mmol) was purchased from Perkin Elmer (USA). Mouse anti-c-myc Mab (clone 9E10) was purchased from Sigma. Horseradish peroxidase-conjugated anti-c-myc (clone 9E10) was purchased from Roche (Mannheim, Germany). Anti-6×His Mab was provided by NeoMarkers (Westinghouse Dr, Fremont, CA). Mabs anti-M13 and horseradish peroxidase-conjugated anti-M13 were from GE Healthcare (USA).
Mouse Mab production and sequencing of the variable regions
Antigastrin mouse Mabs were obtained according to established protocols using gastrin 12-DT and gastrin 17-DT coupled by the C terminus as immunogens. Selection was made by screening against G17 and gastrin 12 by peptide ELISA. From the positive hybridomas, the variable light chain was amplified by using 3 pairs of oligonucleotides.16 The amplification of the heavy chain was performed by using the mix of oligonucleotides previously described.17 When only a single PCR product was obtained, it was subcloned in ZeroBlunt vectors (Invitrogen), transformed into TOP10 cells and directly sequenced.
When more than 1 PCR product was obtained from the hybridomas, murine scFv phage libraries were built. Either 106 cells from the hybridomas (198CA8, 28CD5 or 119EB1) or 180 mg from the spleen of gastrin12-DT immunized mice were used for RNA extraction with Trizol and the RNeasy Mini Kit (Qiagen). For cDNA synthesis we used RT-PCR with oligo dT and the Superscript III enzyme (Invitrogen). The cDNA was used for PCR amplification of the heavy and light chains by using the oligonucleotide mixtures previously described.17 The length of the linker was reduced by 6 amino acids in order to have a linker of only 15 amino acids (G4S)3. After overlapping SOE-PCR, DNA was digested with SfiI, cloned into pAK100 and transformed into XL1 Blue cells (Invitrogen) by electroporation. For the expression and production of murine scFvs, DNA was subcloned into pAK400 after digestion with SfiI, as described.17
Production of gastrin-specific phages and human scFvs
Human scFvs against the N terminus of G17 were obtained from the Tomlinson libraries as described.18 For the C terminus, we used the Mehta human scFv libraries19 and DT-G17 as antigen (N-terminal coupling). Three rounds of selection were made by alternating the selection procedure. Selection was made either in ELISA plates or in solution, using streptavidin-coated magnetic beads with G17 coupled alternatively to DT or biotin at the N-terminal end of the molecule. Elution and production of phages were carried out essentially as described.20 Polyclonal and monoclonal gastrin-positive phages were tested by ELISA.21 Phagemid DNAs from individual positive colonies were sequenced according to standard protocols. The Mehta-derived human scFvs were subcloned into a modified pET28b, containing the pelB sequence and flag-6×His tags at the 5′ and 3′ of the insert sequence, respectively, and expressed after transformation into BL21 (DE3) E. coli cells (Invitrogen).
Monomer scFvs were prepared and purified as previously described21 and analyzed by Coomassie blue staining of 10% SDS-PAGE gels and immunoblotting with anti-His6 antibody.21 Purified monomers were pooled and frozen at −20°C until use. Although the scFvs were monomers directly after size exclusion chromatography, in general they tended to dimerize during storage, as is often the case for scFvs.
Surface plasmon resonance analyses
The binding affinities of antigastrin antibodies were determined by surface plasmon resonance (SPR) using a Biacore 2000 (Biacore AB, Uppsala, Sweden). N- or C-terminal cysteine-extended G17 was immobilized on a CM5 sensor surface (Biacore) using the thiol coupling procedure. The reference surface was treated as the ligand surfaces except that peptide injection was omitted. The immobilization level was 1 RU for gastrin, and between 40 and 240 RU, for the Mabs and for the scFvs, respectively, to avoid mass transfer limitations. For the determination of the binding parameters, each Mab or scFv was passed above the reference and gastrin surfaces in 5–7 concentrations, in PBS containing 0.005% P-20, at a flow rate of 30 μl/min. The surfaces were regenerated by a 10 μl injection of 100 mM HCl. The BIA evaluation 4.1 software was used for data evaluation.
Gastrin Ala-scan peptide microarray
For alanine-scanning, G17 wild type and consecutive Ala substitution mutants (replacing each gastrin residue) were synthesized on an amino-functionalized solid support using standard Fmoc-peptide chemistry forming an ELISA microarray. The deprotected peptides on the solid support were washed with excess of water and sonicated in disruption buffer (1% SDS, 0.1% 2-mercaptoethanol in PBS, pH 7.2) at 70°C for 30 min followed by sonication in water for another 45 min.
Briefly, ELISA microarrays were pretreated with PBS for 30 min followed by precoating with PBS-Tween 0.1% (PBST) for 1 hr. Then, the microarrays were incubated with scFv (10 μg/ml, diluted in PBST) overnight at 4°C. After washing with PBST, they were incubated with peroxidase-labeled rabbit anti-mouse antibody for 1 hr at 25°C (1/1,000; Dako, Glostrup, Denmark), washed again with PBST and incubated with 2,2′-azino-di-3-ethylbenzothiazoline sulfonate (ABTS; 0.5 mg/ml in 0.1 M citricacid–sodium phosphate buffer (pH 4.0) containing 20 μl 30% H2O2). After 1 hr, Abs405 was measured using a CCD-camera (XC-77RR, Sony, Japan). Bound antibody was removed by sonication in disruption buffer as described earlier and used in subsequent ELISA-microarrays. Another Ala-scan was performed using the Biacore system. The cysteine-extended Ala substitution mutants were immobilized on CM5 surfaces by thiol coupling, and the different Mabs and scFvs were injected over these surfaces.
Affinity determination of gastrin17 for CCK-2 receptor
To determine the affinity between G17 and the CCK-2 receptor expressed in Colo 320 WT, cells were resuspended by using a non-enzymatic cell dissociation solution (Sigma) followed by washing with PBS. Approximately, 5 × 105 cells were incubated in duplicate with different concentrations of G17-[125I] Tyr12- during 1 hr at room temperature. After the incubation step, cells were extensively washed with PBS, resuspended in 800 μl PBS and measured in a Wallac WIZARD 1470 automatic gamma counter (Perkin Elmer).
After the previous study, 2 concentrations of G17-[125I] Tyr12-, 40 and 10 nM, were used for the competition experiment with cold G17. Then, G17-[125I] Tyr12 was incubated with Colo 320 WT cells in the presence of 4-fold dilutions of 1 μM cold G17 in the same conditions as above mentioned. After extensive washing with PBS, cells were resuspended in PBS and the radioactivity was measured as before.
Colo 320 WT cells were plated onto 96-well tissue culture plates at a density of 104 cells per well in RPMI 1640 supplemented with 10% foetal bovine serum, antibiotics and G418 in duplicate. Once attached, cells were washed with PBS and cultured with different concentrations of antigastrin Mabs and scFvs (ranging from 0.05 to 6.7 μmol/l and 0.01 to 1.6 μmol/l, respectively) in RPMI 1640 containing 5 × 10−10 mol/l G17, G418 and antibiotics for 72 hr at 37°C and 5% CO2. After removing the medium, cell proliferation was scored by staining cells with 100 μl of the chromogenic dye 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma) added to the culture wells at a final concentration of 1 mg/ml in RPMI 1640. The cells were further incubated for 1 hr at 37°C and 5% CO2. Then, 100 μl of DMSO (Sigma) were added to each well. Absorbance was read at 570 nm. All the experiments were done at least thrice. Cell viability is represented as the ratio of absorbance given by scFv- or Mab-treated cells over absorbance of nontreated cells, expressed as a percentage of the inhibition of G17-induced proliferation.22
To assess the gastrin-specificity of the inhibition assay, scFvs 23CA8, LR28B4, LR28B5, TA4 and 28CD5 at 1.6 μM concentration and 2 Mabs (198CA8 and 119EB1) at 6.4 μM were preincubated with 2 nM concentration of gastrin12 during 1 hr and added to the cells in the same conditions as above. We used gastrin12 to avoid interference with G17-induced proliferation, since it cannot bind to the CCK2 receptor due to the lack of the 5 last amino acids that are responsible for the binding.
Production of human scFvs against gastrin
A first panel of human antibodies against G17 was prepared by using the Tomlinson I+J semi-synthetic libraries.18 The binding affinities for the human Tomlinson scFvs ranged between 1 and 10 μM. Then, we used the larger human Mehta libraries for the production of scFvs against the C-terminus of G17. From the Mehta libraries,19 we obtained 7 different scFvs that specifically recognized G17 by ELISA (data not shown). The alignment of the deduced amino acid sequences of the different antibodies is shown in Figure 2. The amino acid sequences of these human scFvs displayed a large variability not only in the CDRs but also in the framework as a consequence of the naïve origin of the Mehta libraries.
All the scFvs were expressed in E. coli as soluble products and monomers were isolated by size-exclusion chromatography. The affinities of the Mehta scFvs were compared from the fastest dissociation parameter, reflected by the kd1 values listed in Table I. Except for MBE7, kinetic data were best fit using the bivalent analyte model, presumably because the scFvs tend to dimerize when stored. The ka1, and apparent KD1 (KD1′) calculated as the ratio kd1/ka1 are also given, but were not used for ranking because they depend on the active concentration of the scFvs, a parameter difficult to determine with precision. The Mehta scFvs can be divided into 3 groups. MDA3 and MDE6 had a kd1 around 10−2 s−1, and should be the best binders. MBG11 had a kd1 >200 s−1, and is the worst binder. The other 4 scFvs showed intermediate kd1 values.
|scFv||KD′ or KD1′ (10−8 M)||ka or ka1 (104 M−1 s−1)||kd or kd1 (10−3 s−1)|
Production and characterization of Mabs against gastrin and sequencing of the variable regions
Fifty-seven antigastrin mouse monoclonal antibodies were obtained by hybridoma technology. Nine Mabs were specific for the C-terminus and 48 for the N-terminus of G17 based on the reactivity against G17 and gastrin 12 as differential ELISA antigens (data not shown). On the basis of the highest ELISA values against G17, 8 Mabs to the N terminus and 4 to the C terminus were selected for kinetic characterization. The highest apparent affinities (KD1), ranged between 10−8 and 10−10 M (Table II). The lowest dissociation parameters kd1 were displayed by 23CA8, 145FH7, 198CA8, 225CB8, 189DB3 and 243BA5 (N-terminus) and 28CD5, 119EB1 and 131FE1 (C-terminus). All but 145FH7 and 131FE1 were selected for further studies.
|KD1′ (10−9 M)||ka1 (106 M−1 s−1)||kd1 (10−3 s−1)|
To prepare the scFvs, hybridoma cells from the 7 selected Mabs were used for amplification of the cDNA corresponding to the variable regions of the IgGs by RT-PCR. The amino acid sequences of the variable regions corresponding to Mabs 23CA8, 225CB8, 189DB3 and 243BA5 were determined after cloning in Zero Blunt vectors and direct DNA sequencing. Since the remaining Mabs 198CA8, 28CD5 and 119EB1 contained a number of aberrant immunoglobulin chains, 3 scFv phage display libraries were built. Phages were selected after 3 rounds of panning using G17-DT as antigen. The sequences of their variable regions were obtained by direct sequencing (Fig. 3).
Construction of scFv libraries from gastrin-immunized mouse spleens and alignment of sequences
In addition, we used spleen B cells from a mouse immunized with gastrin12-DT for scFv library construction. After mRNA purification and cDNA amplification, the cDNA was cloned into pAK100, yielding a phage antibody library with a size of 6 × 106. After 3 rounds of selection using G17 and gastrin12-DT, 3 murine antigastrin recombinant scFv antibodies (LR10A6, LR28B4 and LR28B5) were obtained and sequenced.
The alignment of the amino acid sequences of all the mouse antibodies is shown in Figure 3. Interestingly, the CDR amino acid sequences corresponding to Mab 243BA5 and the spleen scFv LR28B4 were almost identical. Only occasional changes due to the oligonucleotides used during the PCR amplification of the variable regions were detected in both ends. This result indicates a remarkable limitation in the antibody repertoire against gastrin.
Purification of antigastrin scFv, kinetic characterization and epitope mapping of gastrin
All human and murine scFvs, except those corresponding to Mabs 189DB3 and 225CB8, were produced as soluble proteins in E. coli and purified to homogeneity as monomers (Fig. 4a). All the purified scFvs recognized gastrin17 by ELISA (Fig. 4b) and G17DT using Western blot (data not shown). Regarding affinity determinations for murine scFvs (Table III), the kd measured for 23CA8, 243BA5 and LR28B4 was lower than the kd1 values measured for the best Mehta scFv (MDE6; Table I), indicating a higher affinity for these murine scFvs. Furthermore, the observed stability of the complex is unlikely to result from avidity effects as for other scFvs (Langmuir instead of bivalent analyte model was used for evaluation). The kd values measured for scFvs 23CA8 and 243BA5 (2.9 and 1.7 × 10−3 s−1, respectively) were very similar to the kd1 values measured for the cognate Mabs (4.3 and 0.8 × 10−3 s−1, respectively). For 6 of the 8 scFvs, the KD′ or KD1′ ranged between 15 and 500 nM (Table III).
|scFv||KD′ or KD1′ (10−8 M)1||ka or ka1 (104 M−1 s−1)1||kd or kd1 (10−3 s−1)1|
Fine epitope mapping was carried out by Ala-scan using PEPSCAN microarrays and SPR. Results are shown in Figure 4c. Two linear epitopes were found in the molecule at the N- and C-terminal ends of the gastrin. The N-terminal epitope was composed of the first 6 amino acids (pEGPWLE), whereas the C-terminal epitope is mainly formed by the sequence (WMDF). Human scFv TA4 was the only scFv that recognized the negatively charged region in the middle of the molecule (WLEEEEEA). Some antibodies recognized mixed epitopes containing residues in both the C and the N-termini. Similar epitopes were identified by either human or mouse antibodies, showing no differences in antibody-gastrin 17 recognition between both species.
Gastrin antibody diversity
The alignment of the human Mehta scFv amino acid sequences with the different heavy and light chain germlines allowed us to determine the diversity of the antibody panel together with the phylogenetic tree (Fig. 5a). The different human Mehta antibodies resulted from the pairing of different VH-VL germlines with only the repetition of the pairing VH3 and Vλ3 observed in 2 cases (MDA3 and MBC12). The VH3 segment is the most common heavy chain germline family (6 of 7), with VH3-23 and JH3b being the most prevalent. The light chain germline was very different in all cases, with only the repetition of VL3.1 germline and the JL segment JL2/JL3a observed in 2 cases. The CDR3 of the heavy chain was observed to be the most variable region, not only in amino acid composition but also in length, which varies between 5 and 21 amino acids.
The alignment of the nucleotide sequences of the murine antibodies together with the germlines VH, VL, D (diversity) and J (joining) databases was performed by using the VBase tool (http://vbase.mrc-cpe.cam.ac.uk/). The results are shown in Figure 5b together with the phylogenetic tree that indicates the relationship between all the antibody sequences. VH4 (VH4–61) was found to be the most common heavy chain germline. It was observed in 5 of 10 antibody sequences. The light chain germline was restricted to the Vk family with Vk A2 appearing in 5 cases and A29 and Vk O11 in 2 cases. The murine germlines VH 4-61 and Vk A2 appear to be preferred for G17 binding. The 189DB3 antibody, using VH3 and Vk2 A19 germlines, differed most from others.
Gastrin neutralization by antibodies against different epitopes
A crucial goal of our study was to develop antibodies capable of inhibiting the proliferation of tumor cells. This ability was investigated in an in vitro assay, using a tumoral cell line, Colo 320 WT, which overexpresses the CCK-2 receptor. To study the significance of the inhibition of proliferation mediated by antibodies, we first determined the affinity of G17 for its receptor CCK-2, expressed in Colo 320 WT cells. By incubating Colo 320 WT cells with 40 and 10 nM concentration of G17-[125I] Tyr12- and 4-fold dilutions of nonradioactive 1 μM G17, we determined that the IC50 was 5 nM (Fig. 6).
For the gastrin neutralization studies, maximal proliferation of Colo 320 WT cells was obtained with 5 × 10−10 M G17 (Fig. 7a). Other gastrin concentrations above this started to be toxic for the cells. This gastrin concentration was used to analyze the ability of 2 Mabs (198CA8, 243BA5) recognizing the N-terminal end of G17 and 2 (28CD5, 119EB1) directed against the C-terminal to inhibit gastrin-induced proliferation (Fig. 7b). The Mabs 198CA8 and 243BA5 showed a significant inhibition of the G17-induced proliferation of 73 and 84%, respectively. The anti-C terminal antibodies showed a lower (37%) inhibition of the gastrin-induced proliferation effect at the same concentrations.
The antiproliferative capacity of the panel of anti-G17 scFvs was tested under similar conditions, at scFv concentrations of 1.6, 0.4 and 0.1 μM (Fig. 7c). All the scFvs were able to inhibit the proliferation of the Colo 320 WT cells up to 78%, except for the MBE7 human scFv that showed a background inhibition. Murine scFvs were the best inhibitors, with inhibition values between 56 and 78%. These values were comparable to those obtained with the anti-N terminal Mabs at the same concentration 1.6 μM. For the anti-C terminal scFvs, 28CD5 and 119EB1, the inhibition capacity was larger than that of the corresponding Mabs. The TA4 human scFv that recognizes the polyE region inhibited cell proliferation up to 55%. To confirm that this inhibition was gastrin-specific, both, scFvs and Mabs were made to compete with 2 nM gastrin 12 (Fig. 7d). A substantial decrease in the inhibition ability (<40%) was observed in those antibodies preincubated with gastrin 12, indicating the specificity of the inhibition. As expected, in those cases where anti-C terminal antibodies were used (28CD5 and 119EB1) no significant differences were observed, since these molecules did not bind gastrin12 (Fig. 4c).
Gastrin is becoming an important target for the development of new therapeutic vaccines in pancreatic, gastric and colorectal cancer.12–14 The vaccine currently in clinical trials (Phase II) is based on Gastrimmune, which contains the first 9 N-terminal residues of G17. This vaccine induces antibodies that seem to neutralize gastrin by blocking its binding to the CCK2 receptor.14 The blocking of the CCK2 receptor is known to inhibit tumor cell proliferation.14, 22 The use of this vaccine has resulted in a significant increase of survival time in patients with the 3 types of tumors. Unfortunately, there is always a percentage of nonresponders to the vaccine, probably due to the limited immunogenicity of the molecule or individual disease conditions.12 To overcome these problems, 2 different approaches can be considered. First, passive administration of gastrin-specific human antibodies might overcome the low response of some patients to the vaccine and, second, immunization with more immunogenic molecules. Both approaches were taken in consideration for this study.
We have generated a panel of antigastrin Mabs and scFvs from in vivo (hybridoma technology) or in vitro (human naïve or mouse immunized libraries) selection. Since gastrin can be considered a small immunogen, the goal was to get as much antibody diversity as possible for gastrin. The antigastrin antibody repertoire and the antigenic structure of gastrin were defined from their sequences and gastrin-binding properties. We used 2 human naïve scFv libraries that differ in size and composition, being the Mehta libraries 2 orders of magnitude larger than the Tomlinson ones. As expected, scFvs derived from the Mehta libraries showed, on average, a significantly higher affinity than those derived from the Tomlinson library. However, antibody fragments with the slowest dissociation rates were obtained from immune animals, either from Mabs or spleen cells.
The use of G17 coupled to DT at the C-terminus, such as in Gastrimmune®, preferentially elicited antibodies to the N terminus. When G17-DT was used for phage selection, all the recovered antibodies were directed towards the N-terminal region. After mouse immunization, the relative antibody ratio was 5–1 (N to C). Therefore, in order to generate antibodies against the C-terminal region the use of N-linked DT-G17 seems to be more efficient. In summary, to elicit antibodies against both termini of the molecule it was necessary to use 2 different gastrin/DT conjugates. This result will be especially relevant if, as suggested in some publications,23 the 4 C-terminal residues of G17 (WMDF), also known as tetragastrin, are involved in the binding of gastrin to the CCK-2 receptor and in its functionality.
Regarding antigastrin antibody amino acid sequences, a higher variability was observed for human antibodies than for mouse antibodies. In this case, a remarkable finding was the limited variability of the antibody repertoire against gastrin. Independently of the strategy used for the production of the antibodies, either conventional hybridoma technology or scFv libraries from spleen cells, the mouse antibodies generated were very similar in sequence, if not identical. This rather extraordinary limitation in the mouse antibody repertoire is probably due to the small size of gastrin (17 amino acids) and the acidic nature of the hormone, with a cluster of acid residues (poly E) in the middle of the molecule. These 2 characteristics of the structure of gastrin might also influence the antibody response in cancer patients. In any case, this result suggests that our strategy was effective enough to collect most, if not all, the potential antibody sequences against gastrin.
The large panel of antibodies raised against gastrin enabled us to get a clear delineation of the antigenic structure of gastrin. Two preferential binding sites were found in the amino and carboxy termini of G17. A few human antibodies required residues in both termini for an optimal binding to gastrin, suggesting a folding of the gastrin molecule that brings together both ends of the molecule. Remarkably, human scFv TA4 recognized preferentially the polyE region, an unusually acidic sequence, which represents a large portion of the gastrin sequence and might play an important role in the binding of gastrin to its receptor. Although scFv TA4 possesses a low affinity (1 μM), it has shown promising inhibition effects of the gastrin-induced proliferation of Colo 320 WT that is now being further investigated.
Gastrin 17 and pentagastrin share the same 5 amino acids at the C-terminus and the same receptor. Both are known to enhance tumor proliferation of adenocarcinoma cells.24 Therefore, it would be expected that antibody binding to the C terminus of gastrin would be more effective in blocking G17/CCK-2 receptor activity than those binding to the N terminus. The availability of antibodies to both ends enabled us to study the respective neutralizing activity by using a gastrin-sensitive cell line such as Colo 320 WT. Contrary to our assumptions, the anti-N terminal Mabs 198CA8 and 243BA5 consistently inhibited the proliferation better than the anti-C terminal Mabs. These differences could be explained by the different affinity values of anti-N terminal Mabs, which possesses approximately between 10 and 50 times higher affinity than anti-C terminal Mabs. Taking into account that the affinity of G17 to the CCK-2 receptor is about 5 nM, the superior neutralizing ability of the N-terminal antibodies could be explained by this higher affinity. Still, combinations of anti-N terminal and C-terminal remain to be studied in order to get a more effective inhibition of the proliferation.
With the scFv fragments, we got a quite similar inhibition of the proliferation than when we used the whole Mabs. Mouse scFvs derived from the monoclonal antibodies gave more consistent inhibition than those obtained from the human naÿve phage display libraries, probably due to their superior affinity and slowest dissociation rates. Although the Mabs recognizing the N-terminal sequence pEGPWLE gave the strongest gastrin inhibition, the scFvs directed to both termini showed a similar inhibition potential.
It should be remarked that in most of the cases, Mabs or scFvs, the affinity of the antibodies for gastrin is similar or lower than the nanomolar affinity that we have determined for the binding of gastrin to its CCK-2 receptor in Colo 320 WT cells, making more physiologically relevant the significance of the neutralization effect induced by these antibodies.
In summary, we have designed a strategy that led to the recovery of a large collection of antibodies, describing a comprehensive repertoire, against a therapeutic peptide like gastrin. The combination of different antibodies covering different epitopes of the gastrin molecule could be an interesting therapeutic tool for those tumoral processes susceptible to the trophic effect of gastrin such as gastric, colon or pancreatic cancer. Moreover, these results lead to the design of sequence-specific antibodies that might improve considerably the antigastrin antibody arsenal.
The authors thank the Monoclonal Antibody Unit staff of the CNIO for their collaboration in the production of the mouse Mabs. They also thank Mr. Ronald Boshuizen (Pepscan) for his practical assistance in peptide synthesis and Dr. Frank Schmitz (St. Josef-Hospital, Ruhr-University of Bochum) for kindly providing the Colo 320 WT cells. The authors thank Dr. Juan Carlos Murciano (CNIC, Madrid, Spain) for his assistance with the G17 and CCK-2 receptor binding experiments. Dr. Rodrigo Barderas is recipient of a Postdoctoral Contract of the FIS supported by Spanish Ministry of Health.
- 10Gastrin activates nuclear factor κB (NFκB) through a protein kinase C dependent pathway involving NFκB inducing kinase, inhibitor κB (IκB) kinase, and tumour necrosis factor receptor associated factor 6 (TRAF6) in MKN-28 cells transfected with gastrin receptor. Gut 2003; 52: 813–19., , , , , , , , , .