Methods and means for the production of Ig-like molecules
09758805 · 2017-09-12
Assignee
Inventors
- Cornelis Adriaan De Kruif (Utrecht, NL)
- Linda Johanna Aleida Hendriks (Utrecht, NL)
- Ton Logtenberg (Utrecht, NL)
Cpc classification
A61K39/395
HUMAN NECESSITIES
C07K16/2866
CHEMISTRY; METALLURGY
C07K2317/14
CHEMISTRY; METALLURGY
C07K16/2809
CHEMISTRY; METALLURGY
C07K2317/94
CHEMISTRY; METALLURGY
A61K2039/507
HUMAN NECESSITIES
C07K16/283
CHEMISTRY; METALLURGY
A61K39/3955
HUMAN NECESSITIES
C07K16/28
CHEMISTRY; METALLURGY
C07K16/2896
CHEMISTRY; METALLURGY
C07K16/26
CHEMISTRY; METALLURGY
A61P35/00
HUMAN NECESSITIES
International classification
C12N15/00
CHEMISTRY; METALLURGY
C07K16/26
CHEMISTRY; METALLURGY
A61K39/395
HUMAN NECESSITIES
C07K16/28
CHEMISTRY; METALLURGY
Abstract
The invention provides means and methods for producing one or more Ig-like molecules in a single host cell. Novel CH3 mutations enabling the production of monospecific and/or bispecific Ig-like molecules of interest are also provided.
Claims
1. A method of producing at least two different antibodies in a single host cell, the method comprising: a. providing a host cell comprising: (i) a first nucleic acid molecule encoding a 1.sup.st antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a positively charged amino acid residue, (ii) a second nucleic acid molecule encoding a 2.sup.nd antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a negatively charged amino acid residue, (iii) a third nucleic acid molecule encoding a 3.sup.rd antibody heavy chain, and (iv) a fourth nucleic acid molecule encoding a 4.sup.th antibody heavy chain; b. culturing the host cell and allowing for expression of the four nucleic acid molecules to produce the 1.sup.st antibody heavy chain, the 2.sup.nd antibody heavy chain, the 3.sup.rd antibody heavy chain and the 4.sup.th antibody heavy chain, wherein the CH3 domain of the 1.sup.st antibody heavy chain preferentially pairs with the CH3 domain of the 2.sup.nd antibody heavy chain, and the CH3 domain of the 3.sup.rd antibody heavy chain preferentially pairs with the CH3 domain of the 4.sup.th antibody heavy chain to produce antibodies containing the 1.sup.st and 2.sup.nd antibody heavy chains and the 3.sup.rd and 4.sup.th antibody heavy chains; and c. harvesting the at least two different antibodies from the culture.
2. The method of claim 1, further comprising providing the host cell with a nucleic acid molecule encoding a common light chain.
3. The method of claim 1, wherein the 1.sup.st antibody heavy chain comprises a substitution of the amino acid residue at position 366 in the CH3 domain according to the EU numbering system by a lysine (K) residue, and wherein the 2.sup.nd antibody heavy chain comprises a substitution of the amino acid residue at position 351 in the CH3 domain according to the EU numbering system by an aspartic acid (D) residue.
4. The method of claim 3, wherein the CH3-domain of the 1.sup.st antibody heavy chain further comprises a substitution of the amino acid residue at position 351 according to the EU numbering system by a lysine (K) residue.
5. The method of claim 4, wherein the CH3 domain of the 2.sup.nd antibody heavy chain further comprises amino acid substitution(s) selected from the group consisting of (i) a substitution of the amino acid residue at position 349 according to the EU numbering system by a glutamic acid (E) residue; (ii) a substitution of the amino acid residue at position 349 according to the EU numbering system by an aspartic acid (D) residue; (iii) a substitution of the amino acid residue at position 368 according to the EU numbering system by a glutamic acid (E) residue; (iv) a substitution of the amino acid residue at position 349 according to the EU numbering system by an aspartic acid (D) residue, and a substitution of the amino acid residue at position 368 according to the EU numbering system by a glutamic acid (E) residue; and (v) substitution of the amino acid residues at positions 349 and 355 according to the EU numbering system by aspartic acid (D) residues.
6. The method of claim 5, wherein the CH3 domain of the 2.sup.nd antibody heavy chain comprises the substitution of the amino acid residue at position 368 according to the EU numbering system by a glutamic acid (E) residue.
7. The method of claim 1, wherein the CH3 domain of the 1.sup.st antibody heavy chain comprises substitutions of the amino acid residues at positions 366 and 351 according to the EU numbering system by a lysine (K) residue, and the CH3 domain of the 2.sup.nd antibody heavy chain comprises a substitution of the amino acid residue at position 351 according to the EU numbering system by an aspartic acid (D) residue and a substitution of the amino acid residue at position 368 according to the EU numbering system by a glutamic acid (E) residue.
8. The method of claim 1, wherein the CH3 domain of the 3.sup.rd antibody heavy chain comprises the substitution of the amino acid residue at position 356 according to the EU numbering system by a lysine (K) reside and the substitution of the amino acid residue at position 399 according to the EU numbering system by a lysine (K) residue, and the 4.sup.th antibody heavy chain comprises the substitution of the amino acid residue at position 392 according to the EU numbering system by an aspartic acid (D) residue, and the substitution of the amino acid residue at position 409 according to the EU numbering system by an aspartic (D) residue.
9. The method of claim 1, wherein at least two of the variable regions of the 1.sup.st, 2.sup.nd, 3.sup.rd and 4.sup.th antibody heavy chains recognize different target epitopes.
10. The method of claim 1, wherein each of the variable regions of the 1.sup.st, 2.sup.nd, 3.sup.rd and 4.sup.th antibody heavy chains recognize different target epitopes.
11. The method of claim 9, wherein the variable regions of the 1.sup.st and the 2.sup.nd antibody heavy chains recognize different target epitopes, whereas the variable regions of the 3.sup.rd and the 4.sup.th antibody heavy chains recognize the same target epitope.
12. The method of claim 9, wherein the target epitope recognized by the variable regions of the 3.sup.rd and 4.sup.th antibody heavy chain is the same as one of the target epitopes recognized by the variable region of the 1.sup.st or the 2.sup.nd antibody heavy chain.
13. The method of claim 11, wherein the target epitope recognized by the variable regions of the 3.sup.rd and 4.sup.th antibody heavy chain is different from the target epitope recognized by the variable region of the 1.sup.st or the 2.sup.nd antibody heavy chain.
14. The method of claim 1, wherein the variable regions of the 1.sup.st and the 2.sup.nd antibody heavy chains recognize the same target epitope, whereas the variable regions of the 3.sup.rd and the 4.sup.th antibody heavy chains recognize a second target epitope which differs from the target epitope recognized by the 1.sup.st and 2.sup.nd variable regions.
15. The method of claim 9, wherein the at least two different target epitopes are located on the same target molecule.
16. The method of claim 15, wherein the target molecule is a soluble molecule.
17. The method of claim 15 wherein the target molecule is a membrane-bound molecule.
18. The method of claim 9, wherein the at least two different target epitopes are located on different target molecules.
19. The method of claim 18, wherein the different target molecules are expressed on the same cells.
20. The method of claim 18, wherein the different target molecules are expressed on different cells.
21. The method of claim 18, wherein the different target molecules are soluble molecules.
22. The method of claim 18, wherein one target molecule is a soluble molecule whereas the second target molecule is a membrane bound molecule.
23. The method of claim 1, wherein a. the CH3 domain of the 3.sup.rd and 4.sup.th antibody heavy chains are both wild type; b. the CH3 domain of the 3.sup.rd antibody heavy chain comprises a substitution of the amino acid residues at positions 356 and 399 according to the EU numbering system by a lysine (K) residue, and the CH3 domain of the 4.sup.th antibody heavy chain comprises a substitution of the amino acid residues at positions 392 and 409 according to the EU numbering system by an aspartic acid (D) residue; c. the CH3 domain of the 3.sup.rd antibody heavy chain comprises a substitution of the amino acid residues at positions 356 and 399 according to the EU numbering system by a lysine (K) residue, and the CH3 domain of the 4.sup.th antibody heavy chain comprises a substitution of the amino acid residues at positions 392, 409 and 439 according to the EU numbering system by an aspartic acid (D) residue; d. the CH3 domain of the 3.sup.rd antibody heavy chain comprises a substitution of the amino acid residues at positions 392 and 409 according to the EU numbering system by an aspartic acid (D) residue and a substitution of the amino acid residue at position 399 according to the EU numbering system by a lysine (K) residue, and the CH3 domain of the 4.sup.th antibody heavy chain comprises a substitution of the amino acid residues at positions 392 and 409 according to the EU numbering system by an aspartic acid (D) residue and a substitution of the amino acid residue at position 399 according to the EU numbering system by a lysine (K) residue; e. the CH3 domain of the 3.sup.rd antibody heavy chain comprises a substitution of the amino acid residues at position 356 and 357 according to the EU numbering system by a lysine (K) residue, and a substitution of the amino acid residues at position 439 and 370 according to the EU numbering system by an aspartic acid (D) residue, and the CH3 domain of the 4.sup.th antibody heavy chain comprises a substitution of the amino acid residues at position 356 and 357 according to the EU numbering system by a lysine (K) residue, and a substitution of the amino acid residues at position 439 and 370 according to the EU numbering system by an aspartic acid (D) residue; or f. the CH3 domain of the 3.sup.rd antibody heavy chain comprises a substitution of the amino acid residue at position 366 according to the EU numbering system by a tryptophan (W) residue, and the CH3 domain of the 4.sup.th antibody heavy chain comprises a substitution of the amino acid residue at position 366 according to the EU numbering system by a serine (S) residue, a substitution of the amino acid residue at position 368 according to the EU numbering system by an alanine (A) residue and a substitution of the amino acid residue at position 407 according to the EU numbering system by a valine (V) residue.
24. The method of claim 9, wherein at least one of the target epitopes is located on a tumor cell.
25. The method of claim 9, wherein at least one of the target epitopes is located on an effector cell.
26. The method of claim 25, wherein the effector cell is an NK cell, a T cell, a B cell, a monocyte, a macrophage, a dendritic cell or a neutrophilic granulocyte.
27. The method of claim 25, wherein the target epitope is located on a CD3, CD16, CD25, CD28, CD64, CD89, NKG2D or a NKp46 molecule.
28. A mixture of at least two different antibodies obtained by the method of claim 1.
29. The mixture of claim 28, wherein the at least two different antibodies bind to different epitopes on the same antigen.
30. The mixture of claim 28, wherein the mixture comprises at least one heterodimeric antibody.
31. The mixture of claim 28, wherein two of the at least two different antibodies are heterodimeric antibodies.
32. A recombinant host cell comprising nucleic acids encoding: (i) a first nucleic acid molecule encoding a 1.sup.st antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a positively charged amino acid residue, (ii) a second nucleic acid molecule encoding a 2.sup.nd antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a negatively charged amino acid residue, (iii) a third nucleic acid molecule encoding a 3.sup.rd antibody heavy chain, and (iv) a fourth nucleic acid molecule encoding a 4.sup.th antibody heavy chain, wherein expression of the 1.sup.st, 2.sup.nd, 3.sup.rd and 4.sup.th antibody heavy chains results in preferential pairing of the 1.sup.st antibody heavy chain with the 2.sup.nd antibody heavy chain, and preferential pairing of the 3.sup.rd antibody heavy chain with the 4.sup.th antibody heavy chain.
33. The recombinant host cell of claim 32, wherein the 1.sup.st antibody heavy chain comprises a substitution of the amino acid residue at position 366 in the CH3 domain according to the EU numbering system by a lysine (K) residue, and wherein the 2.sup.nd antibody heavy chain comprises a substitution of the amino acid residue at position 351 in the CH3 domain according to the EU numbering system by an aspartic acid (D) residue.
34. The recombinant host cell of claim 32, wherein the host cell further comprises a nucleic acid sequence encoding a common light chain.
35. A pharmaceutical composition comprising of the mixture of claim 28, and a pharmaceutically acceptable carrier.
36. A method for making a host cell which produces at least two different antibodies, the method comprising the step of introducing into the host cell nucleic acids encoding a 1.sup.st antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a positively charged amino acid residue, a 2.sup.nd antibody heavy chain comprising at least one substitution of a neutral amino acid residue in the CH3 domain by a negatively charged amino acid residue, a 3.sup.rd antibody heavy chain and a 4.sup.th antibody heavy chain, wherein the nucleic acids are introduced consecutively or concomitantly, and wherein expression of the 1.sup.st,2.sup.nd, 3.sup.rd and 4.sup.th antibody heavy chains in the host cell results in preferential pairing of the 1.sup.st antibody heavy chain with the 2.sup.nd antibody heavy chain, and preferential pairing of the 3.sup.rd antibody heavy chain with the 4.sup.th antibody heavy chain.
37. The method of claim 36, further comprising the step of introducing into the host cell a nucleic acid sequence encoding a common light chain.
38. The method of claim 36, wherein the 1.sup.st antibody heavy chain comprises a substitution of the amino acid residue at position 366 in the CH3 domain according to the EU numbering system by a lysine (K) residue, and wherein the 2.sup.nd antibody heavy chain comprises a substitution of the amino acid residue at position 351 in the CH3 domain according to the EU numbering system by an aspartic acid (D) residue.
39. The method of claim 1, wherein at least 90% of the antibodies produced comprising the 1.sup.st antibody heavy chain further comprise the 2.sup.nd antibody heavy chain, and at least 90% of the antibodies produced comprising the 3.sup.rd antibody heavy chain further comprise the 4.sup.th antibody heavy chain.
40. The method of claim 1, wherein the 1.sup.st and 2.sup.nd antibody heavy chains are IgG.
41. The method of claim 40, wherein the 1.sup.st and 2.sup.nd antibody heavy chains are IgG1.
42. The method of claim 1, wherein the 3.sup.rd and 4.sup.th antibody heavy chains are IgG.
43. The method of claim 42, wherein the 3.sup.rd and 4.sup.th antibody heavy chains are IgG1.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
(29)
(30)
(31)
(32)
EXAMPLES
Example 1
Amino Acid Substitutions to Create Various Different CH3-domains
(33) In order to have a wide variety of Ig-like molecules that differ in their CH3 domains such that pairing of CH3-domain comprising Ig-like molecules is preferentially promoted or inhibited, a number of amino acid substitutions that were known to promote heterodimer formation, as well as a number of alternative amino acid substitutions that were not previously reported nor tested but that were chosen to promote homodimer formation, were introduced into a construct vector (construct vector MV1057;
(34) TABLE-US-00003 TABLE 1 % bispecific AA substitu- product tions in CH3 construct # Will pair with reported — (wildtype) — — (wildtype) ~50% E356K, D399K 1 Construct 2 or 3 ~100% K392D, K409D 2 Construct 1 ~100% K392D, K409D, K439D 3 Construct 1 ~100% K392D, D399K, K409D 4 Construct 4 E356K, E357K, K439D, 5 Construct 5 K370D T366W 6 Construct 7 ~86.7% T366S, L368A, Y407V 7 Construct 6 ~86.7%
Example 2
Cloning of VH into Constructs with CH3 Mutations
(35) Several antibody VH regions with known specificities and known ability to pair with the human IGKV1-39 light chain were used for cloning into these constructs.
(36) As indicated earlier, all CH3 variants can be used in association with other antibody domains to generate full length antibodies that are either bispecific or monospecific. The specificity of the antibody as defined by the VH/VL combinations will not affect the heavy chain dimerization behaviour that is driven by the CH3 domains. Model VH/VL combinations were used throughout the studies, wherein all VLs are based on the germline human IGKV1-39 and VHs vary.
Example 3
Transfection and Expression of Full IgG in HEK293T Cells
(37) Transfection of the various plasmids encoding the recloned VH variants, and further encoding the common light chain huIGKV1-39, in HEK293T cells was performed according to standard procedures such that IgG could express (de Kruif et al Biotech Bioeng. 2010). After transfection, IgG expression levels in supernatants were measured using the ForteBIO Octet-QK system, which is based on Bio-Layer Interferometry (BLI) and which enables real-time quantitation and kinetic characterization of biomolecular interactions; for details see www.fortebio.com. When expression levels exceeding 5 μg/ml were measured, the IgG was purified using Protein A affinity purification.
Example 4
Purification of IgG
(38) Culture supernatants were purified using protein A columns (GE Healthcare/cat#11-0034-95/according to manufacturer's instructions) and eluted in 0.1 M citrate buffer pH 3.0 and immediately neutralized in an equal volume of 1.0 M Tris-HCL pH 8.0 or directly rebuffered to PBS using a desalting column. Alternatively one could purify IgG using protein A beads (sepharose beads CL-4B, GE healthcare cat #170780-01)
Example 5
Ag-specific ELISA's
(39) Antigen specific ELISAs were performed to establish binding activity against the antigens and capture ELISAs were carried out to demonstrate binding activity of the bispecific antibodies. Biotinylated second antigen was used for detection of the complex. (de Kruif et al Biotech Bioeng. 2010)
Example 6
SDS-PAGE
(40) The purified IgG mixtures were analysed by SDS-PAGE (NuPAGE® 4-12% bis-tris gel/Invitrogen/cat# NP0323BOX) under reduced and non-reducing conditions according to standard procedures, and staining of proteins in gel was carried out with colloidal blue (PageBlue™ protein staining solution/Fermentas/cat# RO571).
Example 7
Enzymatic Deglycosylation of IgG1
(41) As there is heterogeneity in the glycosylation of the IgGs, the proteins were deglycosylated in order to create a single product with a distinct mass, suitable for mass spectrometric analysis. One unit of N-glycosidase F (PNGase F; Roche Diagnostics, Mannheim, Germany) was incubated per 10 μg of IgG1, overnight at 37° C. Buffer exchange using 10 kDa MWCO centrifugal filter columns (Millipore) was performed to remove the original purification buffer (0.1 M citrate buffer pH 3.0/1.0 M Tris-HCL pH 8.0) and to rebuffer to PBS. Similar buffer exchange procedures were performed to remove the detached glycan chains, and to change the buffer to 150 mM ammonium acetate pH 7.5. Filters were washed with 200 μl 150 mM ammonium acetate pH 7.5, for 12 min 11,000 rpm and 4° C. After washing 50 μl deglycosylated IgG was loaded on the filter and 450 μl of 150 mM ammonium acetate pH 7.5 was added, subsequently followed by another centrifugation round of 12 min at 11,000 rpm at 4° C. In total the centrifugation was repeated 5 times, each time fresh 150 mM ammonium acetate pH 7.5 buffer was added to a total volume of 500 μl. After the last centrifugation step the remaining buffer exchanged deglycosylated IgG1, approximately 25 μl, was collected and transferred to an eppendorf tube, ready for mass spectrometric analysis.
Example 8
Native Mass Spectrometric Analysis
(42) Mass Spectrometry was used to identify the different IgG species in the purified IgG mixtures and to establish in what ratios these IgG species are present. Briefly, 2-3 μl at a 1 μM concentration in 150 mM ammonium acetate pH 7.5 of IgG's were loaded into gold-plated borosilicate capillaries made in-house (using a Sutter P-97 puller [Sutter Instruments Co., Novato, Calif., USA] and an Edwards Scancoat six sputter-coater [Edwards Laboratories, Milpitas, Calif., USA]) for analysis on a LCT 1 mass spectrometer (Waters Corp., Milford, Mass., USA), adjusted for optimal performance in high mass detection (Tahallah et al., RCM 2001). A capillary voltage of 1300 V was used and a sampling cone voltage of 200 V; however, these settings were adjusted when a higher resolution of the ‘signal-to-noise’ ratio was required. The source backing pressure was elevated in order to promote collisional cooling to approximately 7.5 mbar. To measure the IgG1's under denaturing conditions the proteins were sprayed at a 1 μM concentration in 5% formic acid.
Example 9
Data Processing and Quantification
(43) Processing of the acquired spectra was performed using MassLynx 4.1 software (Waters Corp., Milford, Mass., USA). Minimal smoothing was used, after which the spectra were centered. The mass of the species was calculated using each charge state in a series. The corresponding intensities of each charge state were assigned by MassLynx and summed. This approach allowed the relative quantification of all species in a sample. Alternatively, quantification of the peaks can be performed using area-under-the-curve (AUC) methods, known in the art. All analyses were repeated three times to calculate standard deviations of both the masses of the IgG's as well as their relative abundance.
Example 10
Mixtures of 2 or 3 Monospecific Antibodies from a Single Cell
(44) Several antibody VH regions with known specificities and known ability to pair with the human IGKV1-39 light chain (
(45) TABLE-US-00004 TABLE 2 VH specificity inserted in different constructs VH Merus VH Antigen mass desig- Cloned in Vector gene specificity (Da) nation construct # I IGHV Tetanus (A) 13703 MF1337 wildtype 1.08 II IGHV Thyroglobulin 12472 MF1025 4 3.23 (B) III IGHV Fibrinogen (C) 12794 MF1122 5 3.30
(46) TABLE-US-00005 TABLE 3 transfection schedule and results # different mono- Transfection AA BB CC Other specifics Transfection code Expected Calculated Experimental found found found molecules produced of and ratio species mass - 2LYS mass (%) (%) (%) (%) 1 Only A AA 146521 146503 100 vector I 1 Only G BB 144032 144087 100 vector II 1 Only H CC 144647 144656 100 vector III 2 Vector I M AA 146521 146518 51 45 4 and II (I:II = 1:1) BB 144032 144030 2 Vector I N AA 146521 146509 88 9 3 and III (I:III = 1:1) CC 144647 144633 U AA 146521 146522 47 48 5 (I:III = 1:5) CC 144647 144643 2 Vector II nd BB and III CC 3 Vector I, II O AA 146521 146525 66 4 30 and III (I:II:III = BB 144032 144032 1:1:1) CC 144647 144650 V AA 146521 146531 8 81 9 2 (I:II:III = BB 144032 144043 1:1:10) CC 144647 144654 nd = not done.
(47) It was observed that transfections A, G and H resulted in formation of homodimers only, and 100% of bivalent monospecific AA, BB or CC was retrieved from cells transfected with any one of vectors I, II or III (
(48) Next, co-expression experiments of two vectors in a single cell were performed. Interestingly, transfections M and N show that wildtype and CH3 engineered Ig heavy chains can be co-expressed in a single cell together with a common light chain resulting in mixtures of two species of monospecific antibodies without the presence of undesired bispecific antibodies and with as little as 4-5% contaminating ‘other molecules’ present in the mixture. ‘Other molecules’ is defined as all molecules that do not have the mass of an intact IgG, and includes half molecules consisting of a single heavy and light chain pair. Importantly, the fraction ‘other’ does not include bispecific product. In transfection M, the ratio of AA:BB was close to 1:1 upon transfection of equal ratios of vector DNA. However, transfection N resulted in an almost 10:1 ratio of AA:CC. Therefore, this transfection was repeated with adjusted ratios of DNA (transfection U). Indeed, a 1:5 ratio of vector DNA I:III equalized the ratio of AA:CC antibody product in the mixture towards an almost 1:1 ratio. Thus, transfections M and U show that it is possible to express two different, essentially pure, monospecific antibodies in a single cell, without undesired by products (i.e., no abundant presence of AC or half molecules A or C) (
(49) Analogous to these results, also transfection of two different CH3 engineered Ig heavy chains (constructs 4 and 5) are expected to result in mixtures of two different monospecific antibodies only, without further undesired species present. It is reasoned that the CH3 modifications of construct 4 differ substantially from the CH3 modifications of constructs 5 such that heterodimerization does not occur. In that case, co-expression of CH3-engineered heavy chains of constructs 4 and 5, together with wildtype CH3 heavy chains in a single cell would results in 3 monospecific antibodies only.
(50) Indeed, this was observed to be the case as it was found that also a mixture of three pure monospecific antibodies could be obtained by expression of three different Ig heavy chains, designed to form homodimers over heterodimers, together with a common light chain in a single cell, with no contaminations present in the mixture (transfection O) (
(51) Transfections with altered vector DNA ratios (1:1:10, transfection V) demonstrated that ratios of AA:BB:CC in the mixtures can be steered towards desired ratios. Taken together, these experiments show that two or three essentially pure monospecific antibodies can be expressed in a single cell without undesired by products, offering advantages for large scale production of mixtures of therapeutic monospecific antibodies.
Example 11
Mixtures of 2 Bispecific Antibodies from a Single Cell
(52) Whereas use of CH3-engineered heavy chains for production of single bispecific antibodies has been reported elsewhere, this experiment was designed to investigate whether it is feasible to produce mixtures of 2 different bispecific antibodies from a single cell.
(53) Antibody VH regions with known specificities and known ability to pair with the human IGKV1-39 light chain (
(54) TABLE-US-00006 TABLE 4 VH specificity inserted in different constructs VH Cloned Antigen mass in Vector VH gene specificity (Da) construct # IV IGHV 3.23 Thyroglobulin 12472 1 (B) V IGHV 3.30 Fibrinogen (C) 12794 2 VI IGHV 1.08 Tetanus (A) 13703 2 VII IGHV 3.30 Fibrinogen (C) 12794 3 VIII IGHV 1.08 Tetanus (A) 13703 3 IX IGHV 1.08 Tetanus (A) 13703 6 X IGHV 3.23 Thyroglobulin 12472 7 (B)
(55) TABLE-US-00007 TABLE 5 Half # differrent Transfection molecules Full IgG Bispecific Other bispecifics code Expected Calculated Experimental found found found molecules produced Transfection of and ratio species mass - 2LYS mass (%) (%) (%) (%) 0 vector B Half B 144082 144066 40 60 IV 0 vector V C Half C 144651 144622 77 23 0 vector D Half A 146469 146459 23 77 VI 0 vector E Half C 144625 144643 76 24 VII 0 vector F Half A 146443 146468 64 36 VIII 0 vector P Half A 146691 146677 82 18 IX 0 vector X Q Half B 143818 143844 58 42 1 Vector I (1:1) BC 144367 144352 96 4 IV and V 1 Vector J (1:1) BC 144354 144382 96 4 IV and VII 2 Vector K(1:1:1) BC + 144367 + 144351+ 38 + 47 15 (A + IV, V S(2:1:1) AB 145276 145260 42 + 55 C) and VI BC + 144367 + 144371 + 3 (BB) AB 145276 145277 2 Vector L (1:1:1) BC + 144354 + 144346 + 16 + 60 24 (A + IV, VII T (2:1:1) AB 145263 145255 58 + 39 C) and BC + 144354 + 144385 + 3 (BB) VIII AB 145263 145292
(56) It was previously demonstrated that CH3-engineered Ig heavy chains encoded by constructs 1 and 2 are still able to form homodimers when expressed alone in single cells (WO2009/089004). However, WO2009/089004 further reports that CH3 domains that are engineered to comprise triple charge pair mutations, such as present in construct 3, are no longer capable of forming homodimers when expressed alone.
(57) In the present study, these findings were only partly confirmed. Indeed, the results of transfections B, C and D demonstrated the presence of full IgGs, in addition to a high proportion of unpaired half molecules, demonstrating some homodimerization of CH3 domains encoded by constructs 1 and 2. Transfections E and F also resulted in production of full IgGs in addition to unpaired half molecules, demonstrating that the triple charge mutations of construct 3 do not fully impair homodimerisation.
(58) It was furthermore demonstrated that also the ‘knob’ and ‘hole’ CH3 variants of constructs 6 and 7 form homodimers (18% homodimers for ‘knob-knob’ and 42% homodimers for ‘hole-hole’).
(59) CH3 variants that fully prevent homodimerisation when expressed alone are preferred, to prevent or minimize undesired byproducts (homodimers) upon co-expression with a second CH3 variant for heterodimerization.
(60) Interestingly, the present experiments demonstrate for the first time that also mixtures of bispecific antibodies can be expressed in single cells with virtually no homodimers in the mixture. Transfections K and L clearly show that the expected bispecific species BC+AB are indeed obtained (38%+47% in transfection K, and 16%+60% in transfection L). In both transfections a relatively high percentage of undesired half molecules was observed (15% half molecule A+half molecule C in transfection K, and 24% half molecule A+half molecule C in transfection L). The relatively high percentage of half molecules still present was attributed to low amounts of matching heavy chains of vector IV due to unbalanced expression of heavy chains in a matched pair. Therefore, transfections were repeated with an adjusted ratio of vector DNA, 2:1:1, in transfections S and T. This resulted in equal amounts of IgG heavy chains constituting a matched pair and pure mixtures of bispecific IgG without the presence of half IgG molecules and with as little as 3% homodimeric BB present. Ideally, this low proportion of contaminating monospecific product should be reduced to essentially zero. It is therefore desired to find additional CH3-mutants that would result in mixtures of bispecific antibodies with minimal contaminating monospecific antibodies present.
(61) The present study demonstrates for the first time that essentially pure mixtures of two bispecific antibodies recognizing 3 different target epitopes can be produced in a single cell, with minimal presence of monospecific antibodies in the mixture.
Example 12
Varieties of Mixtures
(62) As it was demonstrated that production of mixtures of 2 bispecific antibodies recognizing 3 epitopes from a single cell, or production of mixtures of 2 or 3 monospecific antibodies from a single cell is technically feasible, we next explored the feasibility of controlled production of a variety of other mixtures. A fourth antibody VH region with known specificity and known ability to pair with the human IGKV1-39 light chain will be used for recloning into vectors containing constructs 1-3 or 7 of Table 1, resulting in vectors I′, II′, III′ or X′ (the ‘ indicating a different specificity as compared to corresponding vector numbers). The resulting vectors I’-III′, X′ and IV-IX, each containing nucleic acid sequences encoding for the common human light chain as well as an Ig heavy chain with different CH3 region and different VH specificity, will subsequently be transfected into cells, in combination with other construct vectors to obtain a variety of mixtures of bispecific and/or monospecific antibodies. The variety of mixtures that will be obtained include mixtures of 2 bispecific antibodies recognizing 4 epitopes, 2 bispecific antibodies and one monospecific antibody, or mixtures of 1 bispecific and one monospecific antibody from a single cell. Table 6 depicts the transfection schedule and expected results.
(63) TABLE-US-00008 TABLE 6 Transfection Expected % Variety of code Expected monospecific Expected % mixture Transfection of and ratio species IgG Bispecific 2 BsAbs, 4 IV + V + IX + X′ ZA (1:1:1:1) BC + AD 0 50 + 50 epitopes 2 BsAbs, 4 IV + VII + IX + X′ ZB (1:1:1:1) BC + AD 0 50 + 50 epitopes 2 bsAbs + 1 IV + V + VI + wt′ ZC (2:1:1:2) BC + AB + 33 33 + 33 mAb DD 2 bsAbs + 1 IV + V + VI + II′ ZD (2:1:1:2) BC + AB + 33 33 + 33 mAb DD 2 bsAbs + 1 IV + V + VI + III′ ZE (2:1:1:2) BC + AB + 33 33 + 33 mAb DD 1 bsAb + 1 IV + V + wt′ ZF (1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IV + V + II′ ZG(1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IV + V + III′ ZH(1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IV + VII + wt′ ZI (1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IV + VII + II′ ZJ (1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IV + VII + III′ ZK (1:1:2) BC + DD 50 50 mAb 1 bsAb + 1 IX + X + wt′ ZL (1:1:2) AB + DD 50 50 mAb 1 bsAb + 1 IX + X + II′ ZM (1:1:2) AB + DD 50 50 mAb 1 bsAb + 1 IX + X + III′ ZN (1:1:2) AB + DD 50 50 mAb
(64) Although, theoretically, production of all mixtures should be feasible, it is known from previous work by others that large scale production of classical knob-into-hole variants is hampered by instability issues. Mixtures resulting from transfections ZA, ZB, ZL, ZM and ZN are thus expected to become problematic when transferred to larger scale production.
(65) Thus, the current set of constructs present in Table 1 would not allow production of all theoretical mixtures from single cells at a larger scale, as knob-into-hole variants are reported to be unstable, and it cannot be excluded that CH3 domains comprising a ‘knob’ or a ‘hole’ will dimerize with either charge variants or wildtype CH3 domains. It is thus desired to design new CH3-variants that are engineered to preferentially form homodimers or heterodimers only and which will not homo-or heterodimerize with constructs 1-5 of Table 1 as to allow for co-expression in single cells.
Example 13
Identification of Novel Charge Pair Mutants
(66) The objective of this study was to engineer the IgG CH3 region to result in the production of only heterodimers or only homodimers upon mixed expression of different IgG heavy chains in a single cell, wherein the novel engineered CH3 domains will not homo-or heterodimerize with known engineered CH3 domains, or with wildtype CH3 domains. Therefore, as a first step in identifying novel engineered CH3 domains that would meet the criteria, many interface contact residues in the IgG CH3 domain were scanned one by one or in groups for substitutions that would result in repulsion of identical heavy chains—i.e., reduced homodimer formation—via electrostatic interactions. The objective was to obtain a list of residues that, when substituted by a charged residue, would result in repulsion of identical chains such that these mutations may be used to drive homo-and/or heterodimer formation upon mixed expression of different IgG heavy chains, whereby the obtained full length IgGs are stable and are produced with high proportions. In a follow up, the identified substitutions will be used to generate bispecific antibodies or mixtures of bispecific or monospecific antibodies by engineering matched pairs of CH3 residues in one or more IgG heavy chains—CH3 regions. Additionally, newly identified charge mutant pairs may be combined with existing pairs, such that multiple nucleic acid molecules encoding different heavy chains, all carrying different and complementing CH3 mutations, can be used for expression in cells such that mixtures of monospecific antibodies only, or bispecific antibodies only, or mixtures of defined monospecific and bispecific antibodies can preferentially be obtained. The residues to be tested in the present study are contact residues as previously identified (Deisenhofer J., 1981; Miller S., 1990; Padlan, 1996, Gunasekaran, 2010). The rationale for this approach is that repulsive charges are engineered into each available pair of contacting residues. Samples are subsequently analyzed on non-reducing SDS-PAGE to identify pairs in which dimer formation is reduced, as visualized by the presence of bands of approximately 72 kD. All available pairs will be screened as single mutations or in combination with a single other mutation as the repulsive electrostatic interaction between one non-matching pair may or may not be sufficient to result in sufficient amounts of half-molecules for detection by this method, the mutations are also combined.
(67) Amino acid substitutions were introduced in construct vector MV1057 by Geneart according to the table 7 and expression of constructs was performed by transfection in HEK293T cells, according to standard procedures. IgG expression levels were measured in Octet. When production failed twice, the mutation was considered to be detrimental to expression and the mutation was not pursued further.
(68) TABLE-US-00009 TABLE 7 list of amino acid substitutions in the various constructs that were made (EU numbering) Effect on homodimer formation AA substitu- (− = no effect; +++ = max. inhibition; tions in CH3 construct # NT = not tested on gel) Q347K 8 − Y349D 9 +− Y349K 10 +− T350K 11 − T350K, S354K 12 +− L351K, S354K 13 +− L351K, T366K 14 ++ L351K, P352K 15 +− L351K, P353K 16 ++ S354K, Y349K 17 ++ D356K 18 − E357K 19 − S364K 20 ++ T366K, L351K 21 ++ T366K, Y407K 22 +++ L368K 23 NT L368K, S364K 24 ++ N390K, S400K 25 +− T394K, V397K 26 + T394K, F405K 27 +++ T394K, Y407K 28 +++ P395K, V397K 29 +− S400K 30 − F405K 31 +++ Y407K 32 ++ Q347K, V397K, 33 + T394K Y349D, P395K, 34 + V397K T350K, T394K, 35 NT V397K L351K, S354K, 36 + S400K S354K, Y349K, 37 +− Y407K T350K, N390K, 38 +− S400K L368K, F405K 39 ++ D356K, T366K, 40 +++ L351K Q347K, S364K 41 +++ L368D, Y407F 42 + T366K 43 + L351K, S354K, 44 + T366K Y349D, Y407D 45 + Y349D, S364K, 46 + Y407D Y349D, S364K, 47 + S400K, T407D D399K 48 +− D399R 49 +− D399H 50 +− K392D 51 +− K392E 52 +− K409D 53 +
(69) Supernatants containing ≧5 μg/ml IgG were analyzed in SDS-PAGE and IgG was purified using protein A. The proteins were stained using colloidal blue. Homodimers were visible as a band of approximately 150 kD. Smaller bands of approx 75 kD represented the presence of half molecules (see negative control: K392D, K409D). Blots are shown in
(70) The results of SDS-PAGE gels were analyzed and scored as presented in table 7, right hand column. A number of residues were considered promising for further testing in combination, including residues Q347, S354, Y349, L351, K360, T366, T394, and V397. The choice was based on high scores in the inhibition of formation of homodimers combined with the availability of contacting residues that can be modified without running into issues such as other non-complementary charges. For example, it is known that residues F405 and Y407 have multiple interactions at the CH3-CH3 interface, including interactions with residues that are already charged, which may be problematic after introduction of multiple charge mutations among these interacting residues (see Table A). New constructs were made in vector MV1057 (Table 8), and antibody VH regions with known specificities and known ability to pair with the human IGKV1-39 light chain were used for recloning into vectors containing these new constructs (see Table 9) such that combinations could further be tested. Table 10 depicts the transfection schedules and results.
(71) TABLE-US-00010 TABLE 8 AA substitu- tions in CH3 construct # L351K 61 T394K 62 L351D 63 T366D 64 S354D, Y349D 65 V397D 66 K360D 67
(72) TABLE-US-00011 TABLE 9 VH specificity inserted in different constructs VH Cloned Antigen mass in Vector VH gene specificity (Da) construct # XI IGHV 1.08 Tetanus (A) 13703 8 XII IGHV 1.08 Tetanus (A) 13703 17 XIII IGHV 1.08 Tetanus (A) 13703 43 XIV IGHV 1.08 Tetanus (A) 13703 61 XV IGHV 1.08 Tetanus (A) 13703 62 XVI IGHV 3.30 Fibrinogen (C) 12794 63 XVII IGHV 3.30 Fibrinogen (C) 12794 64 XVIII IGHV 3.30 Fibrinogen (C) 12794 65 XIX IGHV 3.30 Fibrinogen (C) 12794 66 XX IGHV 3.30 Fibrinogen (C) 12794 67
(73) TABLE-US-00012 TABLE 10 Transfection AA AC CC Half A Half C code Expected found found found found found other Transfection of (ratio) species (%) (%) (%) (%) (%) (%) XIII + XVI ZO (1:1) AC 0 69 7 24 0 0 ZT (3:1) AC 10 45 16 27 0 0 ZU (1:1) AC 5 61 10 13 0 0 ZV (1:3) AC 3 61 23 13 0 0 ZW (1:1) AC 0 88.3 2.4 7 0 2.3 XIV + XVII ZP AC 30 52 13 0 0 5 XII + XVIII ZQ AC 4 51 33 2 1 8 XV + XIX ZR AC 20 42 11 0 1 26 XI + XX ZS AC 34 41 15 0 0 10
(74) Combinations of CH3 variants were expressed, and analyzed in SDS-PAGE (data not shown) and in native mass spectrometry (MS). Results are summarized in Table 10. The ZO transfection resulted in the highest proportion of heterodimers in the mixtures (69% AC). Interestingly, in the ZO transfection, the AA homodimer was not present whereas the CC homodimer comprised a small proportion (7%). Mass spectrometric analysis unveiled that the remaining protein in the mixture consisted of half A molecules, probably resulting from unequal expression of the A and C heavy chains. The raw MS data from transfection sample ZO are shown in
(75) It can be envisaged that subtle variants of the newly found T366K/L351′D pair (e.g. by testing all permutations including new constructs T366R and L351E) may result in similar percentages of BsAbs.
Example 14
HADDOCK for Design of New CH3 Mutants to Drive Efficient Heterodimerization
(76) As described in example 13, the newly found charge pair T366K/L351′D increases the proportion of heterodimers in the mixture (69%) with a small fraction of undesired CC homodimers (7%) (L351D/L351′D) and a substantial fraction of half A molecules (24%) ‘contaminating’ the mixture. In this example, an in silico approach was used to generate further insight in amino acid residues involved CH3 interface interactions, to test complementary substitutions in opposing CH3 regions and to find novel CH3 pairs containing complementary substitutions that further increase efficient heterodimerization while preventing efficient formation of homodimers of the two heavy chains.
(77) HADDOCK (High Ambiguity Driven protein-protein DOCKing) is an information-driven flexible docking approach for the modeling of biomolecular complexes. HADDOCK distinguishes itself from ab-initio docking methods in the fact that it encodes information from identified or predicted protein interfaces in ambiguous interaction restraints (AIRs) to drive the docking process. (de Vries et al., 2010). The input for the HADDOCK web server consists of a protein structure file, which can be a crystal structure, NMR structure cluster or a modeled structure. After the docking or refinement, HADDOCK returns a so-called HADDOCK score, which is a weighted average of VanderWaals energy, electrostatic energy, buried surface area and desolvation energy. The HADDOCK score can be interpreted as an indication of binding energy or affinity, even though a direct translation to experimental data is often hard to achieve. In addition to this, HADDOCK provides structure files for the ‘top four’ structures that resulted from the docking run. These structure files can be downloaded and visualized, enabling the detailed analysis of the interactions of the individual residues.
(78) In this example, the interactions between the CH3-domains of the IgG1 heavy chains were studied. A high-resolution crystal structure of the Fc part of the IgG (structure 1L6X) was used as starting structure (http://www.rcsb.org/pdb/explore/explore.do?structureId=1l6x; Idusogie, E. E. et al., J. I. 2000 (164)4178-4184).
(79) In example 13, it was found that co-transfection of vectors XIII and XVI resulted in the formation of the CC homodimeric contaminant (Table 10). HADDOCK was used to search for additional mutations to the T366K/L351′D pair that prevent homodimerization.
(80) The HADDOCK output consists of a set of calculated energies, a HADDOCK score (which is a weighted average of the energies) and four structure files corresponding to the four lowest-energy structures found by the program. The HADDOCK-scores are used to compare different structures; the other energies are merely used to get an indication about what is happening in the structures (e.g. good electrostatic interactions, smaller buried surface, high Van der Waals energy). The lower the HADDOCK score, the better. For each mutation pair, the scores were calculated for the AA, AB and BB dimers.
(81) Sets of mutation pairs from example 12 were run in HADDOCK to see whether the calculated energies would correlate to the experimental data. Table 11 presents all theoretical energies, which are visualized in
(82) TABLE-US-00013 TABLE 11 Electro- Desolva- Buried Construct HADDOCK VdW static tion surface combinations Score energy energy energy area Wildtype-wildtype −208.2 −62.8 −773 9.2 2505.8 1-2 (E356KD399K-K392DK409D) −225.8 −56.4 −862 3 2458.3 2-2 (K392DK409D-K392DK409D) −180.3 −67.9 −562.1 0.1 2312.5 1-1 (E356KD399K-E356KD399K) −176.7 −75.5 −469.3 −7.3 2349.6 1-3 (E356KD399K-K392DK409DK439D) −220.6 −67.9 −793.8 6.1 2499.8 3-3 (K392DK409DK439D-K392DK409DK439D) −150.1 −76.6 −387.6 4.1 2261.2 6-7 (T366W-T366SL368AY407V) −221.3 −65.8 −735.5 −8.3 2509.0 6-6 (T366W-T366W) 1916.9* 2072.3 −681.3 −19.2 2499.9 7-7 (T366SL368AY407V-T366SL368AY407V) −191.9 −55.0 −683.2 −0.2 2427.2 43-63 (T366K-L351D) −210.6 −64 −758.4 5.1 2456.5 43-43 (T366K-T366K) −191.7 −71.2 −634.1 6.3 2533.5 63-63 (L351D-L351D) −212.5 −60.4 −774 2.6 2445.6 *this value is unusually high due to high VanderWaals energy score, probably due to steric clash of T366W/T366′W
(83) With 2 wildtype CH3 domains, the HADDOCK scores are the same for AA, AB and BB because the A and B CH3 regions are identical. In most other cases, the AB pair has the lowest score, which is as expected. For the T366K/L351D pair the BB score is slightly better than the AB score (−210.6 vs.−212.5), but this difference is within the error of the calculations. Using HADDOCK, the structures of the heterodimers of these pairs were visualized. For example, the construct combinations 1-2, 1-1 and 2-2 are presented in
(84) Table 11 and
Example 15
366/351 Variations
(85) In example 13, it is hypothesized that alternatives for the T366K/L351′D mutant charge pair can be designed that may have similar results in terms of percentage of bispecific antibodies in the mixture. Alternatives may include substitutions T366R, T366D, T366E, L351E, L351K and L351R. The proportion of CC homodimers of L351D/L351′D may be diminished by creating variants of the 366/351 pair. All possible mutation pairs were run in HADDOCK and the resulting scores are presented in Table 12 and visualized in
(86) TABLE-US-00014 TABLE 12 Electro- Desolva- Buried Construct HADDOCK VdW static tion surface combinations Score energy energy energy area T366K-L351D −210.6 −64 −758.4 5.1 2456.5 T366K-T366K −191.7 −71.2 −634.1 6.3 2533.5 L351D-L351D −212.5 −60.4 −774 2.6 2445.6 T366K-L351E −216.9 −55.7 −854.7 9.8 2532.7 L351E-L351E −217.9 −65.5 −802.2 8 2532 T366R-L351D −210.5 −68.8 −760.8 10.4 2514.5 T366R-T366R −201.8 −77.4 −626.4 0.9 2608 T366R-L351E −225.8 −56.2 −874.8 5.4 2579.2 T366D-L351R −211.2 −71.3 −723.6 4.8 2455.6 T366D-T366D −198.1 −58.1 −713.4 2.1 2477 L351R-L351R −220.7 −75.5 −806.5 16.1 2552.2 T366D-L351K −223.9 −62.1 −810.1 0.3 2487.8 L351K-L351K −224.4 −75.6 −812.1 13.6 204.5 T366E-L351R −222.3 −69 −783 3.4 2557.2 T366E-T366E −201.9 −57.6 −741 4 2487.5 T366E-L351K −215.9 −58.4 −808.9 4.3 2486
(87) When looking at the HADDOCK scores, it was observed that some of the mutations have a similar ‘pattern’ when compared to T366K/L351′D. For most permutations the AA homodimer was found to have a higher HADDOCK-score than the AB heterodimer, but the BB homodimer appeared as favorable as the AB heterodimer Even though the 351 residue is known to be a ‘neighbor’ to itself on the other chain, i.e. residue 351 of chain A pairs with residue 351 of chain B at the CH3-CH3 interface, there is barely a negative influence of the identical charges when the BB dimer is formed. Looking at the L351D/L351′D structure this is explained by the asp artic acids bending away from each other and the stabilizing influence of at least the naturally occurring Arginine at position 355 and also some stabilization of negative charge by the naturally occurring Serine at position 354 (see
Example 16
Mutations Around T366K/L351′D
(88) In the series of HADDOCK analyses in this example, the T366K/L351′D or T366K/L351′E pair were taken as a starting structure. In order to identify additional mutations that would further increase the predicted percentage of bispecifics of these A and B chains, additional mutations on the B-chain were used to calculate the HADDOCK-scores and energies. When the structure of the CH3 domain is studied using a viewer for visualization of protein structures at a molecular level (YASARA, www.yasara.org), one can calculate the distances between individual residues. While doing so, it was observed that the two residues Y349 and L368 are neighboring residues that may contribute positively or negatively to dimer interactions and these have been mutated in this example—in addition to the L351D mutation—to study the result on dimer formation of the homo-and heterodimers (see
(89) In a next set of HADDOCK analyses, the T366K/L351′D pair was again taken as starting structure. In addition to the substitutions in the B chain that further increased heterodimerization (i.e. Y349D/E and L368E), additional mutations were added to the A-chain which already comprises the T366K substitution. As shown in
(90) As is apparent from
(91) A total of 14 best performing pairs, according to HADDOCK predictions, have been selected (see Table 13 and
(92) TABLE-US-00015 TABLE 13 Construct HADDOCK HADDOCK HADDOCK combinations Score AB Score AA Score BB Wildtype-wildtype −208.2 −208.2 −208.2 T366K-L351D −210.6 −191.7 −212.5 T366K-L351E −216.9 −191.7 −217.9 T366R-L351E −225.8 −201.8 −217.9 T366E-L351R −222.3 −201.9 −220.3 T366K-L351DY349E −215.9 −191.7 −190 T366K-L351DL368E −223.3 −191.7 −198.9 T366K-L351EY349E −214.5 −191.7 −187.5 T366KL351K-L351D −233.2 −205 −212.5 T366K- −207.5 −191.7 −179.5 L351DY349EL368E T366KL351K- −255.2 −205 −204.3 L351DY349D T366KL351K- −227.2 −205 −190 L351DY349E T366KL351K- −243.9 −205 −198.9 L351DL368E T366KL351K- −233.6 −205 −211.9 L351DR355D T366KL351K- −242.8 −205 −183.5 L351DY349DR355D T366D-L351KY349K −237.9 −198.1 −228.4
Example 17
In Vitro Expression of Bispecifics Using CH3 Mutants Based on HADDOCK Predictions
(93) The analysis in example 16 suggested that some CH3 variants with additional mutations around the T366K/L351′D pair would yield mixtures with higher proportions of the bispecific component and potentially lower proportions of the homodimeric component. These best performing pairs were selected for production and further analysis. In addition, the constructs T366R and L351E were also generated. Table 14 lists the constructs that were made and which were used for recloning antibody VH regions with known specificities and known ability to pair with the human IGKV1-39 light chain. Expression of the IgGs that contain the individual constructs was previously reported in example 13, and was repeated for the constructs as listed in Table 14. Aim was to assess which of the constructs homodimerize in the absence of a matching heterodimerization partner. Ideally, high percentages of half bodies would be formed and low percentages of homodimers. As a control, constructs containing previously reported charge mutations and constructs containing the previously reported knob-in-hole mutations were also used for expression as whole IgG by recombinant cells. Protein A purified supernatants were analyzed in SDS-PAGE; results were analyzed and scored as presented in Table 14
(94) TABLE-US-00016 TABLE 14 AA substitu- % half tions in CH3 Construct # % IgG molecule E356K, D399K 1 64.2 35.8 K392D, K409D 2 30.9 69.1 K392D, K409D, K439D 3 24.5 75.5 T366W 6 27.6 72.4 T366S, L368A, Y407V 7 58.6 41.4 T366K 43 32.9 67.1 L351D 63 89.8 10.2 T366D 64 89.6 10.4 T366K, L351K 68 34.7 65.3 L351D, L368E 69 83.7 16.3 L351E, Y349E 70 67.8 32.2 L351D, Y349E 71 79.7 20.3 L351D, R355D 72 100 — L351D, Y349E, L368E 73 79.3 20.7 L351D, Y349D 74 88.6 11.4 L351D, Y349D, R355D 75 89.9 10.1 L351K, L368K 76 56.6 43.4 L351R 77 100 — T366E 78 44.4 55.6 T366R 79 29.6 70.4 L351E 80 100 —
(95) The results of co-expression of a common light chain and two different heavy chains carrying the amino acid substitutions of constructs shown in Table 14 or heavy chains carrying the amino acid substitutions of previous constructs are presented in Table 15. Expression of two different heavy chains comprising the amino acid substitutions T366K and L351′D:L368′E respectively resulted in approximately 87% of the bispecific AB heterodimer in the mixture with no AA or BB homodimers present (combination nr. 3 of Table 15). About 12% half molecules (half A) comprising the T366K substitution was observed. Furthermore, it was found that the percentage of bispecific AB heterodimer increased when the additional amino acid substitution L351K was introduced in the first heavy chain. For example, co-expression of two different heavy chains comprising the amino acid substitutions T366K:L351K and L351′D:L368′E respectively resulted in approximately 92% of bispecific AB heterodimer whereas AA and BB homodimers are essentially absent in the mixture (combination nr. 12 of Table 15). Combinations 10 and 11 also resulted in favorable distributions of high percentages heterodimers and virtually absence of homodimers. The absence of homodimers is advantageous, because the fraction containing the intact IgG molecules is composed of AB heterodimer only. For purification and subsequent therapeutic application, the half molecules can be removed by standard approaches such as size exclusion chromatography. Hence, applying these newly identified charge mutants in the production process for generating bispecific antibodies provides advantages over known charge mutants and knobs-into-holes mutants where the presence of ‘contaminating’ homodimeric antibodies is not excluded. In addition, the T366K/L351′D:L368′E and T366K:L351K/L351′D:L368′E charge pairs have an additional advantage over the previously described E356K:D399K/K392′D:K409′D and E356K:D399K/K392′D:K409′D:K439′D charge reversal pairs, in that the previously described charge variants are based on the reversal of existing charges within the CH3-CH3 interface whereas the newly identified charge variants are adding additional charge pairs (charge-charge interactions) to the CH3-CH3 interface. The introduction of additional charge pairs in the CH3-CH3 interface may further increase the stability of the interface and thereby of the intact antibody. The same holds true for the mutations used in combinations nrs. 4, 5, 6, 9, 10, and 11, which also resulted in favorable proportions of bispecific heterodimer with exceedingly low proportions of AA and BB homodimers present in the mixtures.
(96) TABLE-US-00017 TABLE 15 Combination chain A*/ chain B**/ of 2 different mutations mutations % AA % AB % BB % half A % half B heavy chains (construct #) (construct #) found found found found found 1 T366E (78) L351R (77) 3 81 2 13 0 2 T366K (43) L351D (63) 0 88 3 9 0 3 T366K (43) L351D, L368E (69) 0 87 0 12 0 4 T366K (43) L351E, Y349E (70) 2 85 0 11 0 5 T366K (43) L351D, Y349E (71) 2 92 1 5 0 6 T366K (43) L351D, Y349E, L368E 0 96 1 4 0 (73) 7 T366K, L351K L351D (63) 0 77 12 10 1 (68) 8 T366K, L351K L351D, R355D (72) 0 79 8 10 1 (68) 9 T366K, L351K L351D, Y349D, R355D 1 93 2 4 1 (68) (75) 10 T366K, L351K L351D, Y349D (74) 1 95 1 3 0 (68) 11 T366K, L351K L351D, Y349E (71) 1 95 0 3 1 (68) 12 T366K, L351K L351D, L368E (69) 0 92 0 8 0 (68) 13 T366K (43) L351E (80) 0 70 10 18 2 14 T366R (79) L351E (80) 4 38 36 21 1 15 T366D (64) L351K, L368K (76) 3 92 2.5 2.5 0 16 T366D (64) L351R (77) 30 69 1 0 0 *chain A carries specificity of MF1337 (=tetanus toxoid); **chain B carries specificity of MF1122 (=fibrinogen)
Native MS
(97) Native MS was performed on all bispecific samples. The obtained graphs were analyzed to determine the relative ratio's of the present species in two ways: by peak height and by peak area. Peak area is the more scientifically correct way of analysis, but since all previous analyses for other studies were done based on peak height, both methods were included in the analysis, for comparison purposes. The differences between the methods were within the error of measurement, and therefore only the peak area values were used for future measurements. Two typical spectra are shown in
(98) A panel of ten combinations of 2 different heavy chains was selected from Table 15 for further analyses. These ten combinations included combinations 1, 2, 3, 4, 5, 6, 9, 10, 11 and 12 (Table 15). Selection of these ten was based on low percentages of homodimers present in the mixtures as determined by nMS, but also based on their overall physico-chemical properties, including production yields, SDS-PAGE, as well as the number of mutations present in the CH3 domain.
Example 18
IgG Stability Analyses
(99) In this study, a series of CH3 mutation pairs that resulted in high proportions of bispecific heterodimers in the intact IgG fraction and very low amounts (<5%) of parental IgGs will be further analyzed for stability of the Fc part of the IgG molecule. The mutated CH3 domains that are used to promote the heterodimerization of the heavy chains may have unexpected destabilizing effects on the Fc region of the IgG, that may result in undesirable properties such as a reduction of in vivo half life, reduction in effector function and/or an increase in immunogenicity. The newly identified charge pairs will be compared to wildtype bispecifics and a bispecific containing previously identified charge mutations (chain A comprising construct 1 and chain B comprising construct 2). All bispecifics in this study will contain the same heavy and light chain variable regions, ensuring that the observed effects are caused by mutations in the Fc-part of the molecule and not by variation in the variable regions.
(100) A series of stability studies will be performed on these bispecifics. These studies include spectroscopic (UV-Vis absorbance, fluorescence and light-scatter) and microscopic (light and fluorescence microscopy with Nile Red staining) analyses that provide information on the aggregation state of the CH3 variants.
(101) The UV-Vis absorbance spectra will be recorded with a double beam, two monochromators Cary 300 Bio spectrophotometer at 25° C. The spectra will be monitored between 250 and 400 nm using a path length of 1 cm. The absorbance at wavelengths of 320 nm and longer provides information on the aggregation state of the IgG.
(102) Intrinsic fluorescence spectra will be monitored at 25° C. using a FluoroMax spectrofluorimeter. The fluorescence method will be optimized. The fluorescence emission will provide information on conformation and aggregation properties. 90° light-scattering spectra will be monitored at 25° C. using a FluoroMax spectrofluorimeter by running a synchronous scan (λ.sub.em=λ.sub.ex) between 400 nm and 750 nm with an integration time of 0.01 s. Excitation and emission slits will be optimized. For example, right angle light-scattering can distinguish between IgG samples that have no and 5% dimers.
(103) For fluorescence microscopy with Nile Red staining, just prior to measurements, Nile Red in ethanol will be added to the sample. The samples will be filled in a microscopy slide and analyzed by fluorescence microscopy. Particles will be counted. The lower size limit of the particles that can be observed by fluorescence microscopy is approximately 0.5 μm.
(104) Application of stress such as temperature, pH, mechanical stress or denaturants on proteins might result in a conformation change (e.g. unfolding) and/or aggregation. As it was previously reported that charge-engineered bispecific antibodies have reduced melting temperature of the modified CH3 (Gunasekaran 2010), these studies aim to discriminate between the novel charge mutants of the present invention and existing known charge mutants.
(105) Thermo-stability studies using the Octet are explored, both with Protein A biosensors and by using FcRn binding to IgG. To examine the thermal stability of CH3-engineered IgGs, the samples will be incubated at a concentration of 100 ug/ml (in PBS) at 4, 50, 55, 60, 65, 70 and 75° C. for 1 hour using a PCR machine. Following this the samples will be cooled down slowly during a period of 15 minutes to 25° C. and kept at this temperature for 2 hours, after which they will be stored overnight at 4° C. Precipitated antibodies will be removed by centrifugation, after which the total IgG concentration of soluble antibodies will be determined by Octet using the protein A Biosensor ( 1/10 dilution in PBS). Assays that measure binding of the CH3 engineered IgG to FcRn using the Octet are being explored. Either protein L biosensors are used to bind the light chain of IgG to the sensor, followed by incubation with FcRn in solution, or anti-penta-HIS biosensors are used to bind His-tagged FcRn protein, followed by incubation with the IgG of interest. These methods may be more sensitive than using the protein A Biosensor and can also be used for thermal stability studies. All samples will also be analyzed for serum stability. Briefly, (engineered) IgG samples will be incubated at 37° C. in human serum, control samples will be kept at 4° C. After 1, 2, 3 and 4 weeks, samples are centrifuged to remove precipitated IgG. Subsequently the sample is titrated in antigen-specific ELISA to determine the relative amounts of functional IgG. Purified control antibody freshly spiked in human serum will be used as a reference.
Example 19
Stability Analyses
(106) In previous experiments, high percentages of bispecific antibodies were obtained by co-expression of two different heavy chains comprising CH3 mutations, and a common light chain (example 17).
(107) A panel of eight combinations of 2 different heavy chains was selected from Table 15 for further analyses. These eight combinations included combinations 3, 4, 5, 6, 9, 10, 11 and 12 (Table 15). In this study, these eight combinations were analyzed, with a strong focus on stability of the Fc part of the IgG. As controls, wildtype bispecifics (i.e. without CH3 mutations) and/or bispecifics based on previously reported CH3 charge mutations were included. Note that for wildtype bispecifics, 2 heavy chains and the common light chain are co-expressed without means for preferential steering towards heterodimers. These ‘wildtype bispecifics’ thus represent a mixture of AA, AB and BB. All bispecifics in this study were designed to carry the same VH/VL-combinations, ensuring that the observed effects are caused by mutations in the Fc-part of the molecule and not by variation(s) in the Fab parts.
(108) It was hypothesized that the mutational pairs that were used to promote the heterodimeric pairing of the two different heavy chains could be associated with unexpected structural or otherwise destabilizing effects on the Fc region of the IgG. This could subsequently result in undesired issues that would hamper further clinical development, such as a reduction of in vivo half life, a reduced effector function and/or increased immunogenicity due to the presence of these mutations.
(109) Thermo Stability
(110) Application of stress such as increases or decreases in temperature might result in a conformation change (e.g. unfolding) and/or aggregation of proteins. To examine the thermal stability of CH3-engineered IgGs, the bispecific molecules from combinations 3-6 and 9-12 (Table 15), as well as wildtype bispecifics and bispecific molecules obtained when using constructs 1 and 2 (E356K:D399K/K392D′:K409D′ combination, also dubbed ‘charge reversal’ pair) were incubated at a concentration of 100 μg/ml (in PBS) at 4, 60, 62.5, 65, 67.5, 70 and 72.5° C. for 1 hour using a PCR machine. Following this the samples were cooled down slowly during a period of 15 minutes to 25° C. and kept at this temperature for 2 hours, after which they were stored overnight at 4° C. The next day, precipitated antibodies were removed by centrifugation (18,000 rpm; 4° C., 20 min), after which the total IgG concentration of soluble antibodies was determined by Octet using the protein A Biosensor ( 1/10 dilution in PBS). Results are shown in
(111) Freeze-thaw Stability
(112) To examine the stability of CH3-engineered IgGs upon repetitive freezing and thawing, the bispecific molecules from combinations 3-6 and 9-12 (Table 15), as well as wildtype bispecifics and bispecific molecules obtained when using constructs 1 and 2 (E356K:D399K/K392D′:K409D′ combination (charge reversal pair)) were exposed to ten subsequent freeze-thaw cycles by putting the samples at −80° C. for at least 15 minutes until they were completely frozen. Thereafter, samples were thawed at room temperature. When they were completely thawed, the freeze-thaw cycle was repeated. After 10 freeze-thaw cycles, precipitated antibodies were removed by centrifugation (18,000 rpm; 4° C., 20 min), after which the total IgG concentration of soluble antibodies was determined by Octet using the protein A Biosensor ( 1/10 dilution in PBS). The freeze-thaw stability test was repeated three times Results are shown in
(113) In Vitro Serum Stability
(114) To examine the stability of CH3-engineered IgGs in serum kept at 37° C., the bispecific molecules from combinations 3-6 and 9-12 (Table 15), as well as wildtype bispecifics and the charge reversal bispecific molecules were incubated at 37° C. in 10% human serum. Control samples were kept in human serum at 4° C. After 1, 2 or 5 days, precipitated antibodies were removed by centrifugation. Thereafter, the samples were titrated in a fibrinogen-specific ELISA, to determine the relative amounts of functional IgG. Purified control antibody freshly spiked in human serum was used as reference.
(115) Data of the fibrinogen ELISA show that all samples were quite stable in 10% human serum at 37° C. for 5 days. At the lower IgG concentration bispecific molecules from combinations 4 and 5 seem to be slightly less stable, especially at T=1 and T=2, but the difference is only minimal at the end-point of this experiment (see
Example 20
Further Stability Tests
(116) A further series of analytical methods was used to assess the stability of the variant IgGs. Bispecific molecules from combinations 3-6 and 9-12 (Table 15), as well as wildtype bispecifics (AA, AB, BB), the individual parental antibodies (AA and BB) and bispecific molecules obtained when using constructs 1 and 2 (E356K:D399K/K392D′:K409D′ combination (charge reversal pair)) were used as samples in these stability assays. All IgGs were diluted to 0.2 mg/ml and several stress conditions (2 days at 50° C., 2 weeks at 40° C., 5× freeze-thawing) were applied, aiming to be able to discriminate between the different samples. Of note, these high stress levels resulted in conditions in which one of the parental antibodies (the BB parental, carrying two 1122 Fabs) as used in all bispecifics became unstable. At 2 days at 50° C., aggregation of this protein was detected by UV absorbance. This suggested that this stress condition may not differentiate between instability of the Fab and the CH3 in the bispecific and data resulting from the 50° C. incubation should be used cautiously.
(117) The results are summarized in Table 16. Analytical methods that were used included: Fluorescence microscopy with Nile Red (Nile Red particles' in Table 16); to observe the amount of particles >0.5 μm after addition of Nile Red dye. UV spectrometry at 350 nm (UV 350 nm′); a change in absorption at wavelengths >320 nm gives information about the aggregation state of the protein. 90° Light scatter at 400 nm (LS 400 nm′); a sensitive technique to observe changes in protein aggregation, e.g. the difference between monomers and dimers of IgG. Intrinsic fluorescence; the fluorescence wavelength maximum and intensity of the aromatic residues in a protein change upon changes in the environment (e.g. unfolding) 1,8-ANS fluorescence spectroscopy; 1,8-ANS binds through electrostatic interactions to cationic groups through ion pair formation and changes in protein structure and/or conformation can be detected
UV-VIS Spectroscopy
(118) UV-Vis absorbance spectra were measured at 25° C. with a double beam, two monochromators Cary 300 Bio spectrophotometer from Varian in different quartz cuvettes (such as black low volume Hellma cuvettes with a pathlength of 1.0 cm and clear Hellma cuvettes of 0.2 cm×1.0 cm). The spectra were monitored between 220 and 450 nm using a pathlength of 1.0 cm. The absorbance around 280 nm provides information on the protein concentration. The region between 320 nm and 450 nm can provide information on the aggregation state of the samples.
(119) 90° Light-scattering
(120) The 90° light-scattering spectral method was developed to study protein aggregation and was performed as described in Capelle, 2005; Demeule, 2007a. 90° light-scattering spectra were monitored at 25° C. using a FluoroMax spectrofluorimeter (Spex, Instruments S.A., Inc. U.K.) by running a synchronous scan (λ.sub.em=λ.sub.ex) between 400 nm and 750 nm with an integration time of 0.01 s. Different slits settings were tested in order to find the optimal conditions. After optimization, the same slit settings were used for all measurements.
(121) Steady-state Fluorescence Emission
(122) The fluorescence emission of tryptophan, tyrosine and phenylalanine residues gives information on the local environment of these fluorophores. Changes or differences in hydrophobicity and/or rigidity are measured. Typically, a more hydrophobic and rigid environment leads to an increase in the fluorescence intensity and a blue shift of the emission maximum. Intrinsic fluorescence spectroscopy can provide information on the current state of the protein and monitor changes in the physical and chemical properties. More information on the fluorescence of tyrosine and tryptophan can be found in the book of Lakowicz [Lakowicz, 2006].
(123) The fluorescence emission and excitation spectra were recorded at 25° C. in different quartz cuvettes. The samples were excited at different wavelengths. Integration times and slit settings were optimized. After optimization, the same integration times and slit settings were applied for all samples.
(124) Fluorescence Microscopy with Nile Red Staining
(125) The Nile Red staining method was developed to visualize protein aggregates and was performed as described in Demeule et al., 2007b.
(126) The microscopy observations were performed on a Leica DM RXE microscope (Leica Microsystems GmbH, Wetzlar, Germany) equipped with a mercury lamp. The images were acquired with a Sony NEX-5 camera and its firmware. The objectives were 10×, 20× and 40×. For microscopy investigations slides with a fixed distance of 0.1 mm between the slide and the cover glass were used. The size of the 4×4 grids is 1 mm×1 mm and corresponds to 0.1 μl.
(127) 1,8-ANS Fluorescence Spectroscopy
(128) 1-anilinonaphthalene-8-sulfonic acid (1,8-ANS) is an uncharged small hydrophobic fluorescent probe (Mw 299.34 Da) used to study both membrane surfaces and proteins.
(129) 1,8-ANS is essentially non-fluorescent in water and only becomes appreciably fluorescent when bound to membranes (quantum yields ˜0.25) or proteins (quantum yields ˜0.7). This property of 1,8-ANS makes it a sensitive indicator of protein folding, conformational changes and other processes that modify the exposure of the probe to water. References on 1,8-ANS can be found on the Internet home page of Molecular Probes, www.probes.com.
(130) The fluorescence emission spectra of 1,8-ANS were recorded using a FluoroMax spectrometer. A direct comparison of the 1,8-ANS fluorescence between IgGs will not be performed. Each IgG can have different number of 1,8-ANS binding sites and can therefore not be compared. In principle, the lower the 1,8-ANS fluorescence, the less 1,8-ANS molecules are bound to the antibody. The changes in the 1,8-ANS fluorescence intensity and emission wavelength due to stress will be evaluated.
(131) Taken together, these data indicate that the various IgG samples are remarkably stable. Severe stress conditions (e.g. 2 days at 50° C.) were needed to generate measurable differences between the tested samples. Under these conditions, samples of combinations #9 and #10 seem to aggregate more than other samples.
(132) The most discriminating factors for stability between the proteins are the freeze-thaw cycles and increased temperature. Taking into account the very stringent stress factor of incubating at 50° C., the T366K/L351E,Y349E (combi.#4) and T366K,L351K/L351D,Y349E (combi.#11) variants are the two most stable proteins within panel, closely followed by T366K,L351K/L351D,Y349D (combi.#10) and T366K,L351K/L351D,L368E (combi.#12).
Example 21
Native MS on Ratio Experiments; Transfection Ratio's from 1:5 to 5:1
(133) To become more knowledgeable about the behavior of the CH3 mutated IgGs in skewed transfection mixtures, in particular about the T366K:L351K/L351D′:L368E′ combination (from now on dubbed KK/DE or DEKK), a more elaborate ratio experiment was conducted.
(134) Previously used antibody VH regions with known ability to pair with the common light chain IGKV1-39 were used for recloning into constructs 1, 2, 68 and 69, resulting in vectors I-V of Table 16. Vectors I-V, each containing nucleic acid sequences encoding the common human light chain as well as an Ig heavy chain with different CH3 region and different antigen specificity, were subsequently transfected into cells with different transfection ratios as indicated in Table 17. Results are shown in
(135) TABLE-US-00018 TABLE 17 VH Merus VH Antigen mass desig- Cloned in Vector gene specificity (Da) nation construct # I IGHV Fibrinogen 12794 MF1122 69 (L351D, 3.30 (A) L368E) II IGHV RSV (C) 13941 MF2729 69 (L351D, 3.23 L368E) III IGHV Tetanus (B) 13703 MF1337 68 (T366K, 1.08 L351K) IV IGHV Fibrinogen 12794 MF1122 1 (E356K, 3.30 (A) D399K) V IGHV Tetanus (B) 13703 MF1337 2 (K392D, 1.08 K409D)
(136) TABLE-US-00019 TABLE 18 Transfection nr vectors ratio 1 I and III 5:1 2 I and III 3:1 3 I and III 1:1 4 I and III 1:3 5 I and III 1:5 6 II and III 5:1 7 II and III 3:1* 8 II and III 1:1 9 II and III 1:3 10 II and III 1:5 11 IV and V 5:1 12 IV and V 3:1 13 IV and V 1:1 14 IV and V 1:3 15 IV and V 1:5 *due to a technical error, this sample has not been measured.
(137)
(138) The observed findings for the different ratios of DE:KK are in contrast to the control charge reversal CH3 mutations in vectors IV and V.
(139) Taken together, the DEKK combination of mutations offers a clear benefit over the charge reversal CH3 mutations, in that one of the chains of the heterodimer does not form homodimers.
Example 22
Varieties of Mixtures Using the DEKK Combination
(140) As it was demonstrated that the DEKK combination of mutations drives the formation of bispecific IgG molecules (‘AB’) with high purity, we next explored the feasibility of controlled production of more complex antibody mixtures from one cell, such as ‘AB and AA’ or ‘AB and AC’ mixtures. Previously used model Fabs were incorporated in vectors that contain either the ‘DE construct’ or the ‘KK construct’ and various combinations of these vectors were co-expressed to create mixtures, to demonstrate the versatility of the technology. Model Fabs MF1337 (tetanus toxoid), MF1122 (fibrinogen) and MF1025 (thyroglobulin) were chosen based on their overall stable behaviour, good expression levels and mass differences between the IgGs containing these Fabs (see Table 18)
(141) TABLE-US-00020 TABLE 19 Specificity Fab name IgG mass Δ-mass MF1122 Tetanus (A) (MF)*1337 146747.03 +1842.05 Fibrinogen (B) (MF)1122 144904.98 0 Thyroglobulin (C) (MF)1025 144259.87 −645.11 *MF = Merus Fab, designations such as MF1337 and 1337 are both used interchangeably.
(142) TABLE-US-00021 TABLE 20 Transfection schedule: Heavy Heavy Heavy Tr. Expected species Observed species Tr. # chain 1 chain 2 chain 3 ratio (%) (%) 1 1337-KK 1122-DE 1025-DE 2:1:1 AB (50%) AC (50%) AB (43%) AC (57%) 2 1337-DE 1122-KK 1025-KK 2:1:1 AB (50%) AC (50%) AB (40%) AC (54%) AA (6%) 3 1337-KK 1122-DE 1025-KK 1:2:1 AB (50%) BC (50%) AB (54%) BC (46%) 4 1337-KK 1122-KK 1025-DE 1:1:2 AC (50%) BC (50%) AC (66%) BC (33%) CC (1%) 5 1337-KK 1337-DE 1122-DE 2:1:1 AA (50%) AB (50%) AA (57%) AB (43%) 6 1337-KK 1122-KK 1122-DE 1:1:2 AB (50%) BB (50%) AB (75%) BB (25%) 7 1337-KK 1337-DE 1025-DE 2:1:1 AA (50%) AC (50%) AA (46%) AC (54%) 8 1337-KK 1025-KK 1025-DE 1:1:2 AC (50%) CC (50%) AC (60%) CC (40%) 9 1337-KK 1122-DE 1:1 AB (100%) AB (>98%) 10 1337-KK 1025-DE 1:1 AC (100%) AC (>98%) 11 1122-KK 1025-DE 1:1 BC (100%) AC (>98%)
(143) SDS-PAGE analysis demonstrated that most samples consisted of predominantly full IgGs and in some cases half bodies were present at small percentages. Furthermore, many of the samples showed two bands at ca. 150 kDa on non-reduced gels, reflecting the presence of two distinct IgG species in the sample. Also on the reduced gels, two heavy chain bands were visible in some samples (data not shown).
(144) Native MS was performed on all samples and the percentages of observed species were calculated based on peak height (% of observed species in Table 19). Results are presented in
(145) The IgG samples were further tested in a sandwich ELISA to confirm the functional presence of the desired specificities. Coating of ELISA plates was done with fibrinogen or thyroglobulin and detection was performed with fluorescein-labelled thyroglobulin or—tetanus toxoid. The detection antigens were labelled with fluorescein (Pierce NHS-fluorescin Antibody Labeling kit, cat. #53029) according to the manufacturer's instructions. Fluorescein-labeled antigens could subsequently be detected by a FITC-conjugated anti-fluorescein antibody (Roche diagnostics, cat. #11426346910). Results of the bispecific ELISA (0D450 values) are summarized in
Example 23
Improved Mixtures of Two Bispecific Antibodies Recognizing 4 Different Epitopes (AB and CD) from a Single Cell
(146) In example 12 it was hypothesized that mixtures resulting from transfections ZA or ZB are expected to become problematic when transferred to larger scale production, as knob-into-hole variants are reported to be unstable and it cannot be excluded that CH3 domains comprising a ‘knob’ or a ‘hole’ will dimerize with charge-engineered CH3 domains. As it was demonstrated in the above examples that novel charge pair mutants have been found that preferentially drive heterodimerization with virtually no formation of homodimers, CH3 domain-comprising polypeptide chains comprising these novel charge pair mutants can be expressed in cells together with previously known charge-engineered CH3 domain-comprising polypeptide chains or potentially with SEED bodies, and are likely to result in the preferential formation of two bispecific molecules only.
(147) From the above examples it was clear that the DEKK combination of mutations is excellent for the production of one bispecific (AB) or two bispecifics (AB plus AC) by clonal cells where dimerization of the heavy chains is driven by the CH3 domains. However, using only one vector set of complementary CH3 mutations limits the number of possibilities of mixture-varieties that can be produced. It would be possible to produce more complex mixtures of IgGs and/or bispecifics, such as ‘AB and CD’ or ‘AB and CC’ mixtures if a second ‘orthogonal’ vector set could be used in combination with DEKK. When combining two vector sets, an important requirement is that the heavy chains expressed from the two different sets of CH3 engineered vectors cannot make ‘crossed’ dimers, which is that the heavy chains produced by one of the vector sets dimerize into full IgG with heavy chains expressed by the other vector set.
(148) To test for such potential formation of ‘crossed’ dimers, an in silico analysis was performed using HADDOCK to obtain further insights whether possible pairing between wildtype CH3 domains and CH3 domains containing DE-or KK-mutations would occur. Similarly, potential pairings between wildtype CH3 domains and CH3 domains containing E356K,D399K or K392D,K409D mutations were analyzed, as well as potential pairings between wildtype CH3 domains and CH3 domains containing knob-into-hole mutations and any combination of the above. Combinations of CH3-mutants that were analyzed in HADDOCK are listed in Table 22 and the resulting HADDOCK scores are summarized in
(149) TABLE-US-00022 TABLE 22 CH3 variants analyzed in HADDOCK, with one letter codes for assigned for each CH3-variant carrying heavy chain. One letter code CH3 combination Mutations in HADDOCK DEKK Chain 1: T366K, L351K A Chain 2: L351D, L368E B Wildtype (WT) Chain 1: none C* Chain 2: none D* Charge reversal Chain 1: K392D, K409D A/C** (CR) Chain 2: E356K, D399K B/D** Knob-into-hole Chain 1: T366W C (KIH) Chain 2: T366S, L368A, Y407V D *Wildtype chains are designated ‘C’ and ‘D’ for matters of consistency; **The charge reversal variants are designated ‘A and B’ when combined with knob-into-hole variants, and are designated ‘C and D’ when combined with DE/KK variants.
(150)
(151) In contrast, when looking at the predictions for co-expressing DEKK with wildtype, it can be seen that the HADDOCK scores for AC and AD are lower than the HADDOCK score for CD, which indicates that AC and AD are very likely contaminants when trying to produce a mixture of AB and CD by co-expression of vectors encoding for CH3 combinations of DEKK together with vectors encoding wildtype CH3. Lastly, the predictions for co-expressing either DEKK or charge reversal variants together with the knob-into-hole variants results in undesired bispecific variants with relatively low HADDOCK scores, i.e. a high likelihood that these undesired species will be produced upon co-expression.
(152) It is thus concluded that combining the CH3 combinations of DEKK with charge reversal CH3 combinations (E356K,D399K/K392′D,K409D′) is ideally suited for obtaining essentially pure ‘AB and CD’ and/or ‘AB and CC’ mixtures of antibodies.
(153) Next, mixtures of 2 bispecifics recognizing 4 targets/epitopes (AB and CD) and mixtures of one bispecific and 1 monospecific antibody recognizing 3 targets/epitopes (AB and CC) were created by putting the above into practice. These mixtures were created using 4 different VHs that are all capable of pairing with the common light chain IGVK1-39, but the individual VH/VL combinations all have different specificities. To enable native MS analysis, the mass difference between the (expected) species has to be sufficient, i.e. >190 Da. Four individual VHs have been selected and the masses of these were such that the expected species upon co-transfection could be identified and separated by nMS. Furthermore, the mass differences between the 4 selected VHs are also large enough to identify most of the possible contaminants in the mixtures, in addition to the two desired species. Selected VHs are listed in Table 21.
(154) TABLE-US-00023 TABLE 23 VH (target) Mass as wt IgG A (RTK1) 146736.78 B (Tetanus toxoid) 146106.20 C (Fibrinogen) 144904.98 D (RTK2) 145421.37
(155) The 4 different VHs were cloned into vectors containing the ‘DE’ or ‘KK’ constructs or the charge reversal constructs, and several co-transfections were performed as indicated in Table 22. NB: as always, all vectors also contained the nucleic acid encoding the common light chain IGKV1-39. As previously indicated, when combining two vector sets, an important requirement is that the heavy chains expressed from the two different sets of CH3 engineered vectors cannot make ‘crossed’ dimers, which is that the heavy chains produced by one of the vector sets dimerize into full IgG with heavy chains expressed by the other vector set. To test for such potential formation of ‘crossed’ dimers between heavy chains containing charge reversal mutations and heavy chains containing DE or KK mutations, control transfections were performed.
(156) TABLE-US-00024 TABLE 24 1.sup.st VH/ 2.sup.nd VH/ Expected Tr. # construct # construct # species 1 D/68 A/68 mismatch ‘KK’ with ‘KK’; Mostly half-bodies expected 2 D/68 A/69 match ‘KK’ with ‘DE’; AD product expected 3 D/68 A/1 Expected mismatch ‘KK’ with ‘E356K:D399K’ 4 D/68 A/2 Expected mismatch ‘KK’ with ‘K392D:K409D’ 5 D/69 A/68 match ‘DE’ with ‘KK’; AD product expected 6 D/69 A/69 mismatch ‘DE’ with ‘DE’; mixture of half- bodies, AA, AD and DD expected 7 D/69 A/1 Expected mismatch ‘DE’ with ‘E356K:D399K’ 8 D/69 A/2 Expected mismatch ‘DE’ with ‘K392D:K409D’ 1.sup.st VH/ 2.sup.nd VH/ 3.sup.rd VH/ 4.sup.th VH/ Expected Tr. # construct # construct # construct # construct # species 9 A/68 B/69 C/1 D/2 AB and CD 10 A/68 A/69 C/1 D/2 AA and CD 11 A/68 B/69 C/1 C/2 AB and CC
(157)
(158) All purified protein samples obtained from transfections #1-11 were analyzed on SDS-PAGE, and three control samples were included (
(159) In
Example 24
Single Dose Pharmacokinetic Study in Mice
(160) To study the pharmacokinetic (pK) behavior of bispecific antibodies carrying the DEKK combination of mutations in their CH3 regions, in this study the pK parameters for three different IgG batches were determined and compared.
(161) The three IgG batches included 1) wildtype anti-tetanus toxoid parental antibody 1337:1337 (two MF1337 Fabs on a wildtype Fc backbone); 2) wildtype anti-tetanus toxoid parental antibody 1516:1516 (two MF1516 Fabs on a wildtype Fc backbone); 3) CH3 engineered bispecific anti-tetanus toxoid antibody 1516:1337 that carries the DEKK combination of mutations in its Fc region (MF1516 Fab on DE-side, MF1337 Fab on KK-side).
(162) The parental antibodies 1337:1337 and 1516:1516 were chosen as specificities to be included in the DEKK-bispecific product, as it was known based on previous studies that no pre-dose serum response against these antibodies was present in several mice strains. NB: the presence of a pre-dose serum response would of course invalidate the study. In addition, there is sufficient mass difference between the parental antibodies to enable the identification of 1337:1337 (wt Fc), 1516:1337 (DEKK Fc) and 1516:1516 (wt Fc) species by nMS. The three IgG batches were prepared as previously described, but the DNA used for transfection was made using an endo-free maxiprep kit to ensure that the amount of endotoxins is as low as possible. The batches were subsequently tested for protein concentration, aggregate levels, endotoxin levels and percentage bispecific product. It was demonstrated that the acceptance criteria for subsequent use of the IgG batches in a pK study were met, i.e. the IgG concentration after gel filtration was >0.3 mg/ml, aggregate levels were <5%, endotoxin levels were <3 EU/mg protein and the DEKK batch contained >90% bispecific IgG.
(163) Native mass spectrometry of the gel filtrated samples showed that the expected species were present in high percentages. In sample 1516:1337 a small amount of the DE:DE homodimer is detected, which is estimated to be ca. 2% (
(164) For comparison of pK parameters between the three batches, 3 groups of female C57BL/6J mice (Harlan, The Netherlands) were dosed at 1 mg/kg human IgG (5 ml/kg immunoglobulin solution/kg body weight). At dosing time, the animals were between 7-8 weeks of age and had a body weight of about 18-20 grams. Blood samples were collected pre-dose and at 15, 60 minutes, and 2, 4, 8, 24, 48, 96, 168, 268 and 336 h after dosing. Serum samples were prepared and stored at <−20° C. until analysis. Each group consisted of 3 subgroups of 4 mice, i.e. 12 mice/group. From each mice 6 time points were sampled. The welfare of the animals was maintained in accordance with the general principles governing the use of animals in experiments of the European Communities (Directive 86/609/EEC) and Dutch legislation (The Experiments on Animals Act, 1997). This study was also performed in compliance with the Standards for Humane Care and Use of Laboratory Animals, as issued by the Office of Laboratory Animal Welfare of the U.S. National Institutes of Health under identification number 45859-01 (expiration date: 30 Apr. 2015).
(165) Mice of Group 1 received the full length monospecific IgG 1516:1516 antibody (triangles); Mice of Group 2 received the full length monospecific IgG 1337:1337 antibody (squares); Mice of Group 3 received the full length bispecific IgG 1516:1337 antibody (diamonds), with DEKK engineered CH3 regions (1516 on the DE-side and 1337 on the KK-side).
(166) An ELISA assay was applied for the quantitative analysis of monoclonal human antibodies in mouse serum using a quantitative human IgG ELISA (ZeptoMetrix, NY USA; ELISA kit nr. 0801182). Briefly, the ELISA assay is based on the principle that the human monoclonal antibody binds to anti-human lgG coated in a 96-wells ELISA plate. Bound antibody was subsequently visualized using a polyclonal antihuman lgG antibody conjugated with horseradish peroxidase (HRP). The optical density (OD) of each well is directly proportional to the amount of antibody in the serum sample. Results are shown in
REFERENCES
(167) Deisenhofer J., Biochemistry 1981 (20)2361-2370; Miller S., J. Mol. Biol. 1990 (216)965-973; Padlan, Advances in Protein Chemistry 1996 (49) 57-133 Ellerson J R., et al., J. Immunol 1976 (116) 510-517; Lee and Richards J. Mol. Biol. 1971 (55)379. Gunasekaran et al J. Biol. Chem. 2010 (285)19637-19646 De Vries Nature Protocols 2010 (5)883 Kabat et al, (1991) Mammalian Cell Biotechnology: a Practical Approach (M. Butler, ed., IRL Press, 1991 Merchant Nature biotechnology 1998 (16)677 Ridgeway Protein Engineering 1996 (9)617-621. Davis J H. Et al., Protein Engineering, Design & Selection 2010 (23)195-202 Papadea and Check. Crit Rev Clin Lab Sci. 1989; 27 (1):27-58. Tissue Culture, Academic Press, Kruse and Paterson, editors (1973) Ionescu et al., J. Pharm. Sci. 2008 (97)1414) Current protocols in Protein Science 1995, coligan J E et al., Wingfield P T, ISBN 0-471-11184-8, Bendig 1988. Capelle, M. A. H., Brugger, P., Arvinte, T. Vaccine 23 (2005), 1686-1694. Demeule, B., Lawrence, M. J., Drake, A. F., Gurny, R., Arvinte, T. Biochim. Biophys. Acta 1774 (2007a), 146-153. Demeule, B., Gurny, R., Arvinte, T., Int. J. Pharm 329 (2007b), 37-45. Lakowicz, J. R., Principles of fluorescence spectroscopy; Second Edition Kluwer Academic/Plenum Publishers, New York, Boston, Dordrecht, London, Moscow, (2006) ISBN 0-306-46093-9.