Recombinant monovalent antibodies

Abstract

The invention relates to recombinant monovalent antibodies which are heterodimers of a first protein chain comprising the variable domain of the heavy chain of an antibody of interest and the CH2 and CH3 domains of an IgG immunoglobulin and a second protein chain comprising the variable domain of the light chain of said immunoglobulin of interest and the CH2 and CH3 domains of said IgG immunoglobulin. These antibodies can be used in particular as therapeutic agents in all cases where monovalent binding to a ligand such a cellular receptor is required.

Claims

1. A recombinant antibody derived from a parent antibody directed against an antigen of interest, wherein said recombinant antibody is an heterodimer of: i. a first protein chain consisting essentially of, from its N-terminus to its C-terminus: a. a region A having the structure of the variable domain of the heavy chain of an immunoglobulin, said region A comprising the CDRs of the heavy chain of said parent antibody; b. a region B consisting of a peptide linker and the CH2 and CH3 domains of an IgG immunoglobulin, wherein said peptide linker comprises one or more cysteine residues; ii. a second protein chain consisting essentially of, from its N-terminus to its C-terminus: a. a region A having the structure of the variable domain of the light chain of an immunoglobulin, said region A comprising the CDRs of the light chain of said parent antibody; b. a region B identical to the region B of the first polypeptide; wherein said first and second protein chains are devoid of a hinge region or any portion thereof and of a CH1 domain of an IgG immunoglobulin, and the first and second protein chains are linked by at least one inter-chain disulfide bond.

2. A recombinant monovalent antibody of claim 1, wherein the peptide linker is a peptide sequence of 1 to 16 amino acids.

3. A recombinant antibody of claim 1, wherein the CH2 and CH3 domains are those of an immunoglobulin of the IgG1 subclass, or of the IgG4 subclass.

4. A recombinant monovalent antibody of claim 1, wherein the region A consists of the variable domain of the heavy chain of the parent antibody.

5. A recombinant monovalent antibody of claim 1, wherein the region A consists of the variable domain of the light chain of the parent antibody.

6. A recombinant monovalent antibody of claim 1, wherein the parent antibody is the monoclonal immunoglobulin CD28.3, produced by the hybridoma deposited at Collection Nationale de Cultures de Microorganismes under Accession No. CNCM I-2582.

7. A polynucleotide selected from the group consisting of: (a) a polynucleotide comprising a sequence encoding the first protein chain of a recombinant monovalent antibody according to claim 1; and (b) a polynucleotide comprising a sequence encoding the second protein chain of a recombinant monovalent antibody according to claim 1.

8. An expression vector comprising a polynucleotide of claim 7.

9. A cell transformed with a polynucleotide (a) and a polynucleotide (b) of claim 7.

10. A method for preparing a recombinant monovalent antibody, wherein said method comprises culturing the transformed cell of claim 9, and recovering said recombinant monovalent antibody from said culture.

11. A medicinal product comprising the recombinant antibody of claim 1.

Description

LEGENDS OF THE DRAWINGS

(1) FIG. 1A: Nucleotidic and amino acid sequence of Mono28Fc, VH-CH2CH3 chain.

(2) Underlined: VH domain. Bold: linker. Double underlining: IgG1 CH2-CH3 domains.

(3) FIG. 1B: Nucleotidic and amino acid sequence of Mono28Fc, VL-CH2CH3 chain.

(4) Underlined: VL domain. Bold: linker. Double underlining: IgG1 CH2-CH3 domains.

(5) FIG. 1C: Molecular constructions allowing the expression of Mono28Fc after transfection into eukaryotic host cells.

(6) pCMV: promoter of the cytomegalovirus. Igk leader: signal sequence from the mouse immunoglobulin kappa light chain. VH: variable domain of the heavy chain of the CD28.3 antibody. VL: variable domain of the light chain of the CD28.3 antibody. CH2 and CH3 represent the corresponding domains of the IgG1 human immunoglobulin.

(7) FIG. 1D: Expression plasmids for the synthesis of Mono28Fc in eukaryotic cells.

(8) A: plasmid for the synthesis of the VH(Hc)-CH2-CH3 protein. B: plasmid for the synthesis of the VL(Lc)-CH2-CH3 protein. pCMV: promoter of the cytomegalovirus. Igk leader: signal sequence from the mouse immunoglobulin kappa light chain. Hc: VL variable domain of the heavy chain of the CD28.3 antibody. Lc: VL variable domain of the light chain of the CD28.3 antibody. CH2 and CH3 represent the corresponding domains of the IgG1 human immunoglobulin. BGH pA: signal for the initiation of the 3 polyadenylation of the mRNA molecule, from the bovine growth hormone. Zeocin, ampicillin: resistance genes for the corresponding antibiotic.

(9) FIG. 2: Western blot analysis of pSecVHFc and pSecVLFc expression.

(10) A: Supernatants from Cos cells transfected with the indicated plasmids were collected and reduced before analysis by 10 min. incubation at 100 C. with 10 mM DTT. B: no reduction. Molecular weights are indicated on the left sides.

(11) FIG. 3: Activity ELISA. Recombinant CD28 was immobilized on microtitration plates.

(12) A: Supernatants from control, transfected or co-transfected Cos cells were added at the indicated dilutions, washed and revealed with rabbit anti-VHNL antibodies plus anti-rabbit immunoglobulins-HRP. GFP: negative control; transfection with an irrelevant GFP plasmid. Sc28AT: positive control; transfection with a plasmid coding for a single-chain Fv against CD28. VLFc: transfection with the pSec-VLFc plasmid. VHFc: transfection with the pSec-VHFc plasmid. VH-(CH2-CH3)+VL-CH2-CH3: co-transfection with the pSec-VLFc and the pSec-VHFc plasmids. B: Binding ELISA on recombinant CD28 of purified Mono28Fc molecules at the indicated concentration. Revelation is as in A. Dots are means of triplicates.

(13) FIG. 4: Flow cytometry.

(14) CD28.sup.+ Jurkat T cells and CD28.sup. Raji B cells were incubated with purified Mono28Fc or with CD28.3 Fab fragments at 10 g/ml for 30 min. at 4 C., washed and revealed with rabbit anti-VHNH antibodies plus FITC-labeled goat anti-rabbit immunoglobulins (black profiles). As a control, cells were incubated with rabbit anti-VHNH antibodies plus FITC-labeled goat anti-rabbit immunoglobulins only (grey profiles). Cells were then washed, fixed and analyzed by Facs.

(15) FIG. 5: Activation assay.

(16) Human PBMC (10.sup.5/well) were cultivated in medium or in medium plus 10 g/ml Mono28Fc, sc28AT monovalent antibodies or with ANC28.1 superagonist antibodies for 3 days. 0.5 Ci .sup.3H-tymidine was added for the last 16 h of the culture. Incorporated radioactivity was evaluated on a scintillation counter after transfer on nitrocellulose membranes.

(17) FIG. 6: Pharmacokinetic in mice.

(18) A: Indicated proteins were injected i.v. into swiss mice and blood samples were collected after the indicated time points. CD28 binding activity was measured by ELISA. N=4 for each point, dots are means of the 4 measurements. B: Elimination half-lives (T.sub.1/2) were calculated from the curves in A.

(19) FIG. 7: Molecular constructions combining VH, VL with CH1-CH2-CH3 are non-functional.

(20) A: RT-PCR analysis of the VH and VL mRNA chains expression after transfection of Cos cells. B: Western blot analysis of supernatants (right panel) and lysates (left panel) of Cos cell transfected with pSecVH-CH1-CH2-CH3 and pSecVL-CH1-CH2-CH3 plasmids. Revelation was performed as in FIG. 2. C: Immunofluorescence analysis of Cos cells transfected with pSec-VH-CH1-CH2-CH3 and pSec-VL-CH1-CH2-CH3; revelation with rabbit anti-VH/VL antibody plus anti-PE. Magnification: 20. D: Activity ELISA of supernatants of Cos cell co-transfected with pSecVH-CH1-CH2-CH3 and pSecVL-CH1-CH2-CH3. Revelation was performed as in FIG. 3.

(21) FIG. 8: Amino acid sequence of Mono28Fc with IgG2 hinge and IgG4 CH2CH3 domains.

(22) A: VH-CH2CH3 chain: Underlined: VH domain. Bold: linker. Double underlining: IgG4 CH2-CH3 domains.

(23) B: VL-CH2CH3 chain: Underlined: VL domain. Bold: linker. Double underlining: IgG4 CH2-CH3 domains.

EXAMPLE 1: CONSTRUCTION OF THE MONOVALENT ANTIBODY MONO28FC

(24) The CH2-CH3 domains of a human IgG1 gene (NCBI Accession BC018747) was amplified using the following primers introducing NheI/XbaI sites: CH2CH3-5:

(25) TABLE-US-00001 CH2CH3-5': (SEQIDNO:6) 5-ATATGCTAGCCCAGCACCTGAACTCCTG-3; CH2CH3-3': (SEQIDNO:7) 5-ATATTCTAGATTATTTACCCGGAGA-3.

(26) The resulting fragment was introduced into the pSC-A vector (Stratagene, Amsterdam, The Netherlands), resulting in the pSC-A-CH2-CH3 vector.

(27) VH and VL domains corresponding to the CD28.3 antibody anti-human CD28 were amplified from the previously described CNCM 1-2762 scFv cDNA (VANHOVE et al., Blood, 102, 564-70, 2003) and NheI cloning sites were introduced by PCR with the following primers: VH:

(28) TABLE-US-00002 VH: Hc28.3-5': (SEQIDNO:8) 5'-ATATGCTAGCGGATCCGATATCGTCAAGCTGCAGCAGTCA-3'; Hc28.3-3': (SEQIDNO:9) 5'-ATATGCTAGCAGATGGTGCAGCCACAGTTGAGGAGACGGTGACCA T-3'; VL: Lc28.3-5': (SEQIDNO:10) 5'-ATATGCTAGCGGATCCGATATCGACATCCAGATGACCCAG-3'; Lc28.3-3': (SEQIDNO:11) 5'-ATATGCTAGCAGATGGTGCAGCCACAGTCCGTTTTATTTCCAGCTTG G-3'.

(29) The VH and VL fragments were cloned individually 5 to the CH2-CH3 domains into the NheI site of the pSC-A-CH2-CH3 vector, resulting in VH-pSC-A-CH2-CH3 and VL-pSC-A-CH2-CH3 plasmids. The nucleotidic and amino acid sequences of the resulting VH-CH2CH3 and VL-CH2CH3 constructs are indicated respectively on FIGS. 1A and 1B. They are also indicated as SEQ ID NO: 1 and 3.

(30) Each construct was then subcloned in the EcoRV restriction site of the pSecTag2B eukaryotic pCMV-based expression plasmid (Invitrogen, Cergy Pontoise, France), enabling a fusion at the N-terminus with the secretion signal from the V-J2-C region of the mouse Ig kappa-chain provided by the pSecTag2 vector. The constructs were proofread by sequencing. The resulting expression cassettes and the plasmids pSec-VH-Fc(CH2-CH3) and pSec-VL-Fc(CH2-CH3) containing these constructs are schematized respectively on FIGS. 1C and 1D.

EXAMPLE 2: EUCARYOTIC EXPRESSION OF MONO28FC

(31) COS cells were transfected separately with pSec-VH-Fc(CH2-CH3) (VH-Fc) or pSec-VL-Fc(CH2-CH3) (VL-Fc), or co-transfected with pSec-VH-Fc(CH2-CH3) and pSec-VL-Fc(CH2-CH3) or, as a control, transfected with a plasmid coding for an irrelevant green fluorescent protein (GFP), using the Fugene lipofection kit (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions. Cultures were maintained for 3 days at 37 C., divided one third, and put back into culture for an additional 3 days, after which time the cell supernatants were collected, electrophoresed in 10% polyacrylamide gels and blotted onto nitrocellulose membranes.

(32) Blots were revealed with rabbit anti-CD28.3VH/VL (1:5000 dilution) and an HRP-conjugated donkey antirabbit Ig antibody (Jackson Immuno-Research Laboratories) and developed by chemiluminescence (Amersham Pharmacia Biotech).

(33) The results are shown on FIG. 2. Immunoreactive proteins of the expected molecular weight (42 KDa for VL-CH2-CH3 and 44 KDa for VH-CH2-CH3 under reducing conditions) could be observed in the cell supernatant. A parallel analysis with non-reducing conditions indicated an apparent molecular weight compatible with the formation of both homodimers and heterodimers.

EXAMPLE 3: DETECTION OF MONO28FC BINDING ACTIVITY BY ELISA

(34) Recombinant human CD28 (R&D Systems, Abingdon, United Kingdom) was used at 1 g/mL in borate buffer (pH 9.0) to coat 96-well microtiter plates (Immulon, Chantilly, Va.) overnight at 4 C. These immobilized CD28 target molecules will bind only immunoreactive molecules with anti-CD28 activity.

(35) Reactive sites were blocked with 5% skimmed milk in PBS for 2 hours at 37 C. and supernatants from control cells transfected with the plasmid coding for GFP, from cells transfected with only one of the plasmids pSec-VH-Fc(CH2-CH3) or pSec-VL-Fc(CH2-CH3), and from cells co-transfected with pSec-VH-Fc(CH2-CH3) and pSec-VL-Fc(CH2-CH3) were added at different dilutions and reacted for 2 hours at 37 C. Bound Fc fusion proteins with anti-28 activity were revealed with successive incubations (1 hour, 37 C.) with rabbit anti-CD28.3VH/VL (1:2000 dilution; custom preparation at Agrobio, Orleans, France) and horseradish peroxidase (HRP)conjugated donkey antirabbit Ig antibodies (1:500 dilution; Jackson ImmunoResearch Laboratories, Bar Harbor, Me.). Bound antibody was revealed by colorimetry using the TMB substrate (Sigma, L'Isle d'Abeau Chesnes, France) read at 450 nm.

(36) The results are shown on FIG. 3 A.

(37) Supernatants from control cells (transfected with the plasmid coding for GFP) or from cells transfected with only one of the plasmids pSec-VH-Fc(CH2-CH3) or pSec-VL-Fc(CH2-CH3) did not contain any detectable level of immunoreactive molecule. This indicated that VH-Fc or VL-Fc homodimers cannot bind CD28. In contrast, supernatants from cells co-transfected with pSec-VH-Fc(CH2-CH3) and pSec-VL-Fc(CH2-CH3) contained dilution-dependant levels of immunoreactive molecules.

(38) Mono28Fc was purified from culture supernatants of COS cells co-transfected with pSec-VH-Fc(CH2-CH3) and pSec-VL-Fc(CH2-CH3) and maintained for 3 days at 37 C.

(39) Supernatants were passed through G-Protein Sepharose columns (Amersham) at a rate of 1 ml/min. The columns were rinsed with PBS and proteins were eluted with glycine buffer (pH 2.8), concentred by osmotic water retrieval using polyethylene glycol (Fluka, Riedel-de Haen, Germany) and dialysed extensively against PBS at 4 C.

(40) After purification, the Mono28Fc molecules were tested by ELISA as described above. The results are shown on FIG. 3B.

(41) These results show that 50% of the binding activity to CD28 could be reached at a concentration of 100 ng/ml, which represents 1.16 nM.

EXAMPLE 4: DETECTION OF MONO28FC BINDING ACTIVITY BY FLOW CYTOMETRY

(42) The binding of Mono28Fc was confirmed by flow cytometry using CD28+ Jurkat human T cells, which express CD28, or on Raji cells, a human B cell line that does not express CD28.

(43) Jurkat T cells or Raji cells were incubated for 1 hour at 4 C. with purified Mono28Fc proteins or with Fab fragments of CD28.3 (VANHOVE et al., Blood, 102, 564-70, 2003), at 10 g/ml for 30 min. As a control, cells were incubated with rabbit anti-VII/VH antibodies plus FITC-labeled goat anti-rabbit immunoglobulins only. Bound Fc fusion monomers were detected with a rabbit anti-CD28.3VH/VL and a fluorescein isothiocyanate (FITC)conjugated donkey anti-rabbit Ig antibody (dilution 1:200; Jackson ImmunoResearch Laboratories) for 30 minutes at 4 C. Cells were then analyzed by fluorescence-activated cell sorting (FACS).

(44) The results are shown on FIG. 4. Both mono28Fc and the Fab fragment of CD28.3 bind Jurkat T cells. In contrast, no binding of the mono28Fc protein could be observed on Raji cells, a human B cell line that does not express CD28. These data demonstrate mono28Fc that binds specifically to CD28.sup.+ cells.

EXAMPLE 5: MONO28FC HAS NO AGONIST ACTIVITY ON HUMAN T CELLS

(45) To verify that mono28Fc binds to CD28 and does not induce activation of the target T cell, we compared the biological effect of Mono28Fc with those of the superagonistic antibody ANC28.1 (WAIBLER et al., PLoS ONE, 3, e1708, 2008), or of sc28AT, a monovalent anti-CD28 ligand without Fc domain (VANHOVE et al., Blood, 102, 564-70, 2003).

(46) Human PBMC (10.sup.5/well) were cultivated in culture medium without additive (control), or in culture medium with 10 g/ml of mono28Fc, of sc28AT, or of ANC28.1 for 3 days. 0.5 Ci .sup.3H-tymidine was added for the last 16 h of the culture. Incorporated radioactivity was evaluated on a scintillation counter after transfer on nitrocellulose membranes. The results are shown on FIG. 5.

(47) As expected, ANC28.1 induced a robust proliferation of the target cells. In contrast, Mono28Fc, as well as sc28AT did not induce any response in this assay.

EXAMPLE 6: PHARMACOKINETICS OF MONO28FC IN MICE

(48) Recombinant proteins fused with an Fc fragment and immunoglobulins usually present an extended half-life in vivo because they are recognised by the FcRn receptor presented on endothelial and epithelial cells allowing the recycling of that molecules back in the circulation. To determine if our Mono28FC molecule also presents an extended half-live, we followed the distribution in mice of Mono28Fc in comparison with monovalent Fab 28.3 antibody fragments and native IgG CD28.3 antibodies.

(49) Each protein tested (288 g per injection) was injected into the tail vein of male Swiss mice. Blood samples (2 L) were collected at different times from the tail vein. The proteins were quantified by measuring the CD28 binding activity in blood samples by ELISA. The data were analyzed by Siphar software (Simed, Utrecht, The Netherlands) with the use of a 2-compartment model. Significance was evaluated with an non-parametric ANOVA test followed by a Bonferroni's Multiple Comparison Test.

(50) The results are shown on FIG. 6.

(51) The distribution half-live (T.sub.1/2) was of 2.51.1; 5.10.3 and 5.41.2 hours for IgG, Fab and Mono28Fc, respectively. The elimination half-live (T.sub.1/2) was of 11919; 396 and 8326 hours for IgG, Fab and Mono28Fc, respectively (FIG. 6). The data reveal a significant increase of the elimination half-live of Mono28Fc, as compared with Fab fragments, whereas no statistical difference is pointed out when Mono28Fc is compared with a divalent IgG.

EXAMPLE 7: COMPARISON OF MONO28FC WITH A CONSTRUCTION COMPRISING THE CH1-CH2-CH3 IG HEAVY CHAIN DOMAINS

(52) The human IgG1 CH1-CH2-CH3 cDNA was given by Dr. S. Birkle (Univ. Nantes, France). It was inserted into the pcDNA3.1 into the HindIII/BamHI restriction sites, resulting in the pcDNA3.1-CH1-CH2-CH3 plasmid. VH and VL domains corresponding to the CD28.3 antibody anti-human CD28 (NUNES et al., Int Immunol, 5, 311-5, 1993) were amplified as described in Example 1 above, digested with the NheI enzyme and inserted separately into the NheI site of the pcDNA3.1-CH1-CH2-CH3 plasmid. The VH-CH1-CH2-CH3 and VL-CH1-CH2-CH3 cassettes were then excised by EcoRV/XbaI digestion and inserted into the EcoR V digested pSecTag2B vector (Invitrogen), as disclosed in Example 1.

(53) After transfection in Cos cells, messenger RNA molecules corresponding to the two chains were equally synthesised (FIG. 7A). The analysis of proteins by western blotting revealed the synthesis of some corresponding molecules, although clearly more abundant within the cell (FIG. 7B, left panel) than in the supernatant (FIG. 7B, right panel) for the light chain (VL-CH1-CH2-CH3). By immunohistology, the synthesis of both heavy and light chains by transfected Cos cells could be confirmed (FIG. 7C). By ELISA, no CD28 binding activity could be detected in the supernatant (data not shown) nor in transfected cell lysates (FIG. 7D).