Processing

Please wait...

Settings

Settings

Goto Application

1. WO2007137760 - DIMERIC MOLECULAR COMPLEXES

Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

[ EN ]

DIMERIC MOLECULAR COMPLEXES

SUMMARY OF THE INVENTION

The present invention provides dimeric molecular complexes comprising a first and second fusion protein, wherein each fusion protein comprises from its N to C terminus (a) a biological effector moiety, (b) a hinge region of an IgG molecule bound to the biological effector moiety and (c) a CH4 dimerization domain of an IgE molecule covalently bound to the hinge region, wherein the molecular complex comprises a disulfide bond between a cysteine residue in the hinge region of the first fusion protein and a cysteine residue in the hinge region of the second fusion protein.

Preferred biological effector moieties are a single chain antibody, an Fab fragment, an extracellular domain of a type I membrane receptor, a cytokine, a chemokine, an enzyme, a toxin or a detectable marker. More preferred biological effector moieties are a single chain antibody, an Fab fragment, a toxin or a detectable marker.

In one embodiment, the dimeric molecular complex of the invention comprises two fusion proteins each comprising identical biological effector moieties. In another embodiment, the two fusion proteins within the complex each comprise different biological effector moieties. In a preferred embodiment, the biological effector moieties are antigen binding sites, with either the same or different binding specificities.

Each fusion protein comprises a hinge region comprising amino acid residues 223 to 243 of SEQ ID NO:25, wherein positions 240-243 are occupied by the tetrapeptide VFLF. In preferred embodiments, the tetrapeptide VFLF is replaced with a tetrapeptide selected from the group consisting of DSEY, KSKY, DEEY and KRKY. Most preferred are embodiments where the tetrapeptide is DSEY or KSKY.

In another aspect, the invention provides dimeric molecular complexes comprising a first and second fusion protein, wherein each fusion protein comprises from its N to C terminus (a) a CH4 dimerization domain of an M2" IgE splice variant, (b) an amino acid linker which is covalently bound to the CH4 dimerization domain and (c) an extracellular domain of a type Il membrane receptor, wherein the molecular complex comprises a disulfide bond between cysteine residues within the C terminal M2" IgE splice variant CH4 dimerization domains of each of the two fusion proteins. In a preferred embodiment, the M2" IgE splice variant is SEQ ID NO:26, SEQ ID NO: 27, SEQ ID NO: 28, SEQ ID NO:29 or SEQ ID NO:30. In one embodiment, the type Il membrane receptor is a myeloid DAP12-associating lectin-1 (MDL-1) receptor.

In another aspect, the invention provides nucleic acid molecules which encode the fusion proteins which comprise the dimeric molecular complexes. In one embodiment, a nucleic acid molecule encodes the fusion protein of SEQ ID NO:1.

In another aspect, the invention provides pharmaceutical compositions including the subject dimeric molecular complexes.

In another aspect, the invention relates to a method of treatment using the subject dimeric molecular complexes. In preferred embodiments, methods of treatment involve administration of dimeric molecular complexes comprised of fusion proteins as described above, conjugated to
chemotherapeutic agents or to toxins. In a preferred embodiment, the biological effector moiety of at least one of the two fusion proteins comprises an antigen binding site.

In another aspect, the invention relates to a method of imaging a target area of the body using a dimeric molecular complex of the invention comprised of two fusion proteins, wherein the biological effector moiety of the first fusion protein comprises an antigen binding site and the biological effector moiety of the second fusion protein comprises a detectable signal.

BRIEF DESCRIPTION OF THE FIGURES

FIGS. 1A-D. Schematic drawings of various dimeric molecular complexes. In each case, disulfide bonds are formed between the cysteines in the hinge region of the fusion proteins comprising the complex. FIG. 1 A, a dimeric molecular complex in which the biological effector moiety is a single chain antibody. FIG. 1B, two monospecific dimeric molecular complexes in which the biological effector moiety is a Fab fragment; the complexes differ from each other in the sequence used to replace the "wild-type" tetrapeptide VFLF) within the hinge. FlG. 1C, a bispecific dimeric molecular complex in which the biological effector moieties are two different Fab fragments and the where each fusion protein further comprises a different M2" IgE splice variant at its C terminus (SEQ ID NOS:26 and 27). FIG. 1D, a dimeric molecular complex in which the biological effector moiety is an extracellular domain of a type I membrane receptor.

FIG. 2. Schematic of a dimeric molecular complex in which the biological effector moiety is an extracellular domain of a type Il membrane receptor.

FIG. 3. Schematic showing generation of the single chain antibody 19G9scFv. The VH and VL regions were amplified by PCR using primers which introduced restriction sites and sites for overlap extension of the GGGGSGGGGSGGGGSGGGGS linker (SEQ ID NO:8). Following 15 cycles of extension, the scFv was amplified with the original forward and back primers shown for 35 cycles and then cloned by restriction digestion into the pz613 bacterial expression vector. VH forward primer, SEQ ID NO.15; VH back primer, SEQ ID NO.16; VL forward primer, SEQ ID NO:17; VL back primer, SEQ ID NO:18.

FIG. 4A-B. Schematics showing construction of IgE CH4 dimerization domain constructs used to make the dimeric molecular complex containing the 19G9scFv single chain antibody. FIG. 4A1 construction of an IgE CH4 dimerization domain fused to an IgGI hinge. ATG001 primer, SEQ ID NO:9; ATG003 primer, SEQ ID NO:11 ; ATG004 primer, SEQ ID NO:12; ATG006 primer, SEQ ID NO:13. FIG. 4B, construction of an IgE CH4 dimerization domain fused to an IgG hinge with four amino acids mutated to create a more hydrophilic hinge (ATG019 primer, SEQ ID NO:14). Insert shows the sequence around the site of mutation including both the amino acid sequence, amino acids 252-281 of SEQ ID NO:1 and nucleotide sequence (nucleotides 783-869 of SEQ ID NO:2).

FIGS. 5A-B. Sequences of two 19G9scFv fusion proteins. FIG. 5A, amino acid sequence at bottom (SEQ ID NO:3) and nucleic acid sequence at top (SEQ ID NO:4) of "wild-type" 19G9scFv fusion protein contains the regions VH, bold italic; scFv linker, bold italic underline; VL and C-terminal extension, bold; IgG hinge, italic, with "wild-type" tetrapeptide in bold underline; IgE CH4, plain; C-terminal epitope tag, italic underline. FIG. SB, amino acid sequence at bottom (SEQ ID NO:1 ) and nucleic acid sequence at top (SEQ ID NO:2) of "mutant" 19G9scFv fusion protein contains the regions VH, bold italic; scFv linker, bold italic underline; VL and C-terminal extension, bold; IgG hinge, italic, with hydrophilic tetrapeptide substitution in bold underline; IgE CH4, plain; C-terminal epitope tag, italic underline.

FIG. 6A-B. Characterization of a purified dimeric molecular complex comprising two fusion proteins containing the single chain antibody, 19G9scFv (sometimes referred to as 19G9scFv dimeric molecular complex). FIG. 6A, size exclusion chromatography (SEC) profile, which indicates pure protein is mainly a dimer (~90 kDa based on elution times of molecular weight standards). FIG. 6B, SDS-PAGE analysis of non-reduced 19G9scFv fusion protein (lanes 2 and 3, containing 2//L and 5μL of protein, respectively) and reduced 19G9scFv fusion protein (lanes 5 and 6, containing 2//L and 5//L of protein, respectively) is consistent with the calculated molecular weight of the dimeric complex (85.4 kDa) and of the 19G9scFv fusion protein monomer (42.7 kDa). Lanes 1 and 7 contain molecular weight markers.

FIG. 7. Graph showing ELISA data of purified 19G9 scFv dimeric molecular complex (open triangles) binding to BHK-RG-1 protein compared with binding of the corresponding 19G9 diabody (closed squares). Much of the 19G9 diabody sequence is similar to the 19G9 scFv sequence (amino acids 1-251 in SEQ ID NO:1). The linker between the VH and VL domains in the diabody, however, is only 5 amino acids in length (GGGGS).

FIG. 8. Diagram of vector pPEP MF-J DSEY2 used to express the Fab dimer complex, as shown schematically in FIG. 1B.

FIG. 9. In vivo bio-distribution of the 19G9scFv dimeric molecular complex as compared with other antibody formats (19G9 IgG, 19G9 monomeric scFv single chain antibody and 19G9 diabody). The Y-axis plots the average percent injected dose per gram (mean %ID per gram) for four tissues: blood (B), tumor (T), liver (L) and kidney (K). Values are shown for tissues harvested at 0.25, 3, 6 and 48 hrs following injection of 2//Ci of 111 In-labeled antibody per animal. The % ID per gram for the 48 hr tumor sample is indicated with an arrow. While the tumor is well labeled by both the 19G9 scFv dimeric molecular complex and the 19G9 IgG, a high percentage of the 19G9 IgG is also retained in the blood at 48 hr, in contrast to the dimeric molecular complex. The scFv and diabody do not label the tumor well and also result in an accumulation of label in the kidneys.

FIG. 10. Schematic showing generation of the plE-J_DSEY plasmid used to express the J_DSEY fusion protein, as described in Example 5.

FIG. 11. Binding of the J_DSEY dimeric molecular complex to the biotinylated antigen protein for antibody ABJ protein compared with the binding of the Fab form of ABJ. The J_DSEY dimeric molecular complex and the Fab form of ABJ have equivalent activity, with an EC50 that is approximately nanomolar. Previous studies comparing the monomeric Fab and dimeric IgG forms of ABJ in similar assays show that the monomeric and dimeric antibodies bind at similar concentrations. Antibody concentrations were calculated based upon molecular weights of 125 kD for the dimeric molecular complex, 50 kD for the Fab and 15OkD for the IgG. The secondary antibody used in the assay was HRP-anti-human Fab antibody (Jackson ImmunoReserch).

FIG. 12 Size exclusion chromatography (SEC) analysis. Both the J_DSEY dimeric molecular complex and the heterogeneous expression of J_DSEY and J_KSKY gave major peak at about 125 kD, suggesting the formation of dimeric complex. The J_KSKY dimeric molecular complex also showed a peak at about 125 kD although it is not the major peak.

FIG. 13. MALDI-TOF Mass analysis. The J_DSEY dimolecular complex (0.26mg/mL) was mixed with 10mg/rnL Sinapinic Acid MALDI Matrix in 50% Acetonitrile, 0.1% TFA at a ratio between 1 :1 and 1 :5 and Ipmole minimum protein was targeted per spot. MALDI-TOF mass analysis was performed using a Bruker UTT instrument calibrated linearly with Bruker Protein Standards II. Positive ion spectra were recorded with an accelerating potential of 25kV for 100 laser shots, accumulated for 15 events. The indicated molecular mass of 123,962.2 Da shown agrees with the molecular weight expected for the intact J_DSEY Fab dimeric molecular complex and is consistent with the SEC and calculated mass.

FIG. 14. Binding of the 25_DSEY dimeric molecular complex to biotinylated antigen protein for both antibody AB25 (same as antigen for ABJ above) compared with the binding of the Fab form of AB25 and the IgG form of AB25. The binding activity of the dimeric molecular complex (25_DSEY) was improved compared to that of the Fab, and equivalent to the binding found with the IgG form. The EC501S for the dimeric 25_DSEY Fab dimer and the IgG are both in the sub-nanomolar range. Previous studies comparing the monomeric Fab and dimeric IgG forms of AB25 in similar assays show that the dimeric IgG bound with greater avidity than did the monomeric Fab, consistent with these results. The antibody concentration was calculated based upon molecular weights of 125 kD for the dimeric molecular complex, 50 kD for the Fab and 15OkD for the IgG. The same HRP-anti-human Fab antibody as shown in Example 6 was used as a secondary antibody.

DETAILED DESCRIPTION OF THE INVENTION

The invention exploits the self-dimerization properties of the IgE CH4 domain to provide dimeric molecular complexes comprising biological effector moieties. A "biological effector moiety" is a polypeptide which comprises the biologically active portion of molecules such as antibodies, type I and type Ii membrane receptors, cytokines, enzymes, and the like. Useful biological effector moieties include single chain antibodies, Fab fragments, extracellular domains of type I or type Il membrane receptors, cytokines (including chemokines), active site domains of enzymes, protein hormones and peptide effector molecules.

The biological effector moieties in each of the two fusion proteins which comprise a dimeric molecular complex can be identical or can be moieties with two different functions (e.g., an antigen binding site and a toxin). Biological effector moieties which have a useful therapeutic or tissue specific targeting function are especially useful.

Dimeric molecular complexes are stable in vivo and bind to a target of interest with an affinity similar to that of the native molecule from which the biological effector moiety is derived.

Dimeric molecular complexes of the invention comprise two fusion proteins. In some embodiments, each fusion protein comprises the following, from N to C terminus: (a) a biological effector moiety, (b) a hinge region of an IgG molecule bound to the biological effector moiety; and (c) an IgE CH4 dimerization domain covalently bound to the hinge region. Dimerization of the CH4 dimerization domains of the first and second fusion proteins occurs by formation of disulfide bonds between cysteines in the hinge region of each fusion protein. See FIGs. 1A-D.

The effector molecule can be covalently attached directly to the native sequence of the IgGi hinge. In other embodiments, an amino acid linker can be employed. In such embodiments, the effector molecule is covalently attached to the linker, which is covalently bound to the IgGi hinge. The IgG hinge is preferably an IgGi hinge, preferably derived from human IgGi, although hinge regions of IgG1 from other species can be used (e.g., mouse, rabbit, etc.). In other embodiments, hinge regions from other IgGs may be used (e.g., IgG2, IgG3 and IgG4).

The IgGi hinge, as used herein, comprises amino acid residues 223 to 243 from within the first beta strand of the IgGi CH2 domain, where the numbering is based on the sequence of the Eu IgGi heavy chain (SEQ ID NO:25) as described in Edelman et al. (1969) Proc. Natl. Acad. Science USA, 63, pp. 78 - 85. This sequence contains a hydrophobic tetrapeptide, VFLF at positions 240-243 (corresponding to positions 277-280 in SEQ ID NO:3). In a preferred embodiment, this hydrophobic tetrapeptide is replaced by a tetrapeptide of hydrophilic amino acids, Asp, Ser, GIu, and Tyr (DSEY) (amino acid residues 277-280 in SEQ ID NO:1), to increase solubility. The hinge and N-terminal CH2 sequences of IgG2, IgG3Or IgG4 also contain the VFLF tetrapeptide and therefore hinge regions from these immunoglobulin molecules provide the same opportunities for substitution at these positions.

In another embodiment, modifications can be made in the hinge region to increase the ability to conjugate various molecules to the dimeric molecular complex. While lysine residues already present in the dimer can be used for conjugation, additional sites for conjugation can be provided by altering the same hydrophobic tetrapeptide (i.e. VFLF, at positions 277-280 in SEQ ID NO:3) that was changed to DSEY (resulting in SEQ ID NO:1) to, for example, the tetrapeptide, KSKY. The additional lysine residues provided by this substitution can be used to add polyglycols (e.g., PEG, POG) to a dimerized molecule to improve clearance. Alternatively, molecules such as toxins, detectable labels, radioactive molecules, etc., can be conjugated to fusion proteins of the dimeric molecular complex through these lysine residues. Methods of conjugation are well known in the art.

DIMERIC MOLECULAR COMPLEXES USING ANTIBODIES AS BIOLOGICAL EFFECTOR MOIETIES
scFv and Fab dimers

The dimeric molecular complexes of the invention can be formed using fusion proteins comprising antibodies as the biological effector moieties. In some embodiments, the biological effector moieties of each fusion protein are comprised of antigen binding sites. The antigen binding sites can be provided, for example, by a single-chain antibody (scFv) or an Fab fragment. The resulting dimeric molecular complex will comprise two antigen binding sites, which can be of the same or different specificities (i.e. forming monospecific or bispecific dimeric molecular complexes, respectively).

A monospecific dimeric molecular complex in which the biological effector moiety on both fusion proteins is a single chain antibody is shown schematically in FIG. 1 A. FIG. 1B is a schematic representation of monospecific dimeric molecular complexes in which the biological effector moiety on each fusion protein is a Fab fragment (i.e. "Fab dimer"). FIG. 1 C is a schematic representation of a bispecific dimeric molecular complex in which the biological effector moieties on the fusion proteins are Fab fragments with differing specificities.

The antibody-related dimeric molecular complexes of the invention have superior pharmacokinetic and biodistribution profiles when compared to those of monomeric single chain antibodies, diabodies, or full-sized dimeric IgGs. When compared to a whole IgG antibody, the molecular complexes of the invention provide a higher tumor to blood concentration ratio at an earlier time point after in vivo administration and show better penetration into a tumor. Such complexes also show a higher accumulation in the tumor with less accumulation in the kidney compared with both monomeric single chain antibodies or diabodies (See FIG. 9).

The antibody-related dimeric molecular complexes of the invention can be used diagnostically and therapeutically. For example, a dimeric molecular complex can be used for in vivo imaging, such as PET imaging, or for in vitro diagnostics. When conjugated to a therapeutic molecule, a dimeric molecular complex can be used to target the therapeutic molecule to a particular target.

In a preferred embodiment, the two fusion proteins within the dimeric complex each comprise single chain antibodies linked to the IgE CH4 domain via the hinge region and N-terminal amino acids including a portion of the first β sheet of the CH2 domain of human IgG1. The benefit of this attachment is that the disulfide bonds between the two fusion proteins will be located in the middle of the dimeric molecular complex, thus stabilizing the complex. Such dimeric molecular complexes typically have a molecular weight between 50 kD and 150 kD.

One example of an scFv-containing fusion protein used in a dimeric complex of the invention is SEQ ID NO:1 (FIG 5B), which contains the single chain antibody 19G9scFv.

The hinge region in this fusion protein, THTCPPCPAPELLGGPSDSEY (SEQ ID NO:5; amino acid residues 260-280 of SEQ ID NO:1) contains the hydrophilic tetrapeptide 'DSEY' at amino acid residues 277-280, in place of the "wild-type" tetrapeptide 'VFLF (amino acids residues 277-280 of SEQ ID NO:3).

This fusion protein is soluble and binds to antigen with an affinity similar to that of the parent immunoglobulin.

The linker between the VH and VL domains in the 19G9scFv fusion protein molecule consists of amino acid residues 125-144 of SEQ ID NO:1: GGGGSGGGGSGGGGSGGGGS (SEQ ID NO:8).

This particular fusion protein also comprises a C-terminal epitope-tag (GAPVPYPDPLEPRAA (SEQ ID NO:23)) at amino acid residues 391-405 of SEQ ID NO:1. Such epitope tags are often used to aid in purification of expressed molecules. In other embodiments of the invention, the dimeric molecular complexes of the invention are produced using fusion proteins which do not contain an epitope tag or with different epitope tags (e.g. ESSEEGGC (SEQ ID NO:24)).

The dimeric molecular complexes of the invention can also be made using Fab heavy chain fragments instead of single chain antibodies as the biological effector molecules in the fusion proteins which make up the dimeric complex (See Examples 5-8). When forming a dimeric molecular complex comprising Fab fragments, the CH 1 domain of the heavy chain can be directly linked to the hinge region as in a normal immunoglobulin molecule, further reducing the amount of non-native, immunogenic sequences in the molecule. When Fab light chains are combined with the dimeric molecular complex of Fab heavy chains, an Fab dimeric molecular complex is formed.

A Fab dimer which has the same affinity as an IgG but a lower molecular weight can be constructed by eliminating the Fc region. The lower molecular weight of such a dimer can result in increased tumor penetration. Moreover, such Fab dimers will not bind to cellular Fc receptors (FcR), thus reducing unwanted FcR-related interactions. For example, a dimeric molecular complex using such Fab dimers can be used to make toxin conjugates which will not bind cells with FcyR or the neonatal Fc receptor (FcRn). FIG. 8 shows a vector used to generate such an Fab containing dimeric molecular complex, comprising a light chain (SEQ ID NOS:19 and 21 , amino acid and nucleic acid sequences respectively) and a heavy chain and IgE domain (SEQ ID NOS:20 and 22, amino acid and nucleic acid sequences, respectively). Such a complex has an internal disulfide bond between the light and heavy chains of the Fab, as well as disulfide bonds between the hinge regions of the two fusion proteins. This molecule will not elicit any Fc receptor or C1q mediated activities in vivo.

When the Fc binding domain is included in an Fc fusion protein, the dimeric molecular complex generated can bind not only to the Fc receptor but also to other proteins {e.g., Fcyllla, Fcvllb, C1q, etc.) and can trigger the effector functions of these molecules. Binding to Fc receptors permits an antibody toxin conjugate to be directed to cells which express the Fc receptor. In vivo properties of an antibody that may be mediated by Fc receptor interaction include antigen dependent cellular cytotoxicity (ADCC), phagocytosis, complement dependent cytotoxicity, increased serum half-life, and decreased clearance, etc. The terms 'Fc fusion protein1 or 'Fc fusion' are widely used to refer to the practice of dimerizing proteins using the IgG hinge and the CH2 and CH3 domains. Such 'Fc fusions' retain the ability to bind Fc receptors.

Preferred embodiments of Fab containing dimeric molecular complexes comprise hinge regions wherein the "wild-type" tetrapeptide 'VFLF' within the hinge region is replaced by the more hydrophilic "mutant" tetrapeptide DSEY or with the tetrapeptide 'KSKY', which provides extra lysines for conjugation (See SEQ ID NOS:5 and 7, respectively).

In other embodiments of either scFv or Fab containing dimeric mo\ecu\ar complexes, tetrapeptide sequences other than the two described above can be used to replace the hydrophobic tetrapeptide VFLF, which is present in the "wild-type" dimerization domain (amino acid residues 277-280 in SEQ ID NO:3) of the fusion protein. Replacement of this tetrapeptide with different tetrapeptides in the first and second fusion proteins which make up the dimeric construct is of particular utility in helping to promote heterodimer formation. In preferred embodiments, replacements include: SESE or SDSD on one fusion protein with SKSK or SRSR on the second fusion protein. Additional sequence substitutions for the hydrophobic tetrapeptide that can be used to promote heterodimer formation in this manner include: SESY or SDSY with SKSY or SRSY; or DEEY, DDDY, DDEY, DEDY, EEEY, EDDY, EDEY, or EEDY with RRRY, RKRY, RRKY, RKKY, KKKY, KRRY, KRKY, or KKRY. In the human IgG1 structure, the two Phe residues in the hydrophobic tetrapeptide starting at Val24o (V2-JoFLF) point inward, toward the corresponding Phe residues of the other heavy chain in the IgG1 dimer. These Phe residues extend toward carbohydrate structures attached to CH2 and located between the two CH2 domains (Saphire (2002) J MoI Biol 319, pp. 9-18). The VaI and Leu residues in the VFLF tetrapeptide point outward, away from the carbohydrates and toward amino acid residues in CH2. Replacement of residues within the VFLF tetrapeptide with less hydrophobic residues (Ser, Thr, Asp, GIu, Asn, GIn, GIy, His, Lys, Arg, Cys or Ala) decreases the number of exposed hydrophobic side chains, thus improving the solubility of the dimeric molecular complex. Substitution with Lys or Cys also provides sites for chemical modifications. To promote heterodimer formation, one or both Phe residues within one fusion protein of the dimeric molecular complex are replaced by residues possessing a charge opposite that present in the second fusion protein in the dimeric molecular complex. Placement of oppositely charged residues pointing toward each other from each fusion protein comprising the heterodimer allows the charge interaction to promote heterodimer formation. In general, if X is any amino acid that is not hydrophobic, then the following combinations of sequences could be used to make heterodimers by substitution for the VFLF tetrapeptide in the two molecules of the heterodimer: 1) XEXY or XDXY with XKXY or XRXY; 2) XEXE, XEXD, XDXE or XDXD with XRXR, XRXK, XKXR or XKXK; or 3) XRXE, XRXD, XKXE or XKXD with XEXR, XEXK, XDXK or XDXR.

In other embodiments, the wild-type hydrophobic tetrapeptide sequence 'VFLF' and the adjacent proline residue (amino acid residues 277-281 of SEQ ID NO:3) can be replaced with amino acid sequences from the loop connecting CH3 and CH4 in the human IgE sequence (SEQ ID NO:45). This stretch of five amino acids (VFLFP) can be replaced with either LysThrSerGfy (amino acid residues 315 to 318 of SEQ ID NO:45 ) or ThrLysThrSerGly (amino acid residues 314 to 318 of SEQ ID NO:45). Similarly, residues 277 to 281 in SEQ ID NO:1 (DSEYP) or amino acid residues 263 to 267 in SEQ ID NO:20 (DSEYP) can also be replaced by these IgE-derived sequences in order to create dimeric molecular complexes composed entirely of native sequence from either human IgG or IgE.

Typically, an IgE CH4 domain is used to dimerize fusion proteins of the invention. However, in order to further stabilize heterodimers, M2" IgE CH4 splice variants can be added at the C-termini of the fusion proteins comprising the dimeric molecular complex. In one fusion protein, an acidic form of the M2"

IgE CH4 splice variant (ESSEEGGC (SEQ ID NO:26)) is added to the C terminus, while on the second fusion protein an M2" IgE CH4 splice variant containing basic amino acids (ESSRRGGC (SEQ ID NO:27) is added. See FIG. 1C.

DIMERIC MOLECULAR COMPLEXES USING OTHER BIOLOGICAL EFFECTOR MOIETIES

A variety of biological effector moieties other than antibody-related molecules can be used in the fusion proteins which comprise the dimeric molecular complexes of the invention. In such proteins, an amino acid linker between the biological effector moiety and the hinge region may or may not be used. If included, the amino acid linker is preferably a poly GIy linker from 1 to 10 or more amino acids and may include other amino acids, including Ala, Ser, Thr and Asp. Depending on the intended use, dimeric molecular complexes may be comprised of fusion proteins where the two biological effector moieties are the same or different (i.e. homodimers or heterodimers, respectively). Heterodimers are stabilized in a manner similar to that described above.

A. Extracellular Domains of Type I Membrane Receptors

A dimeric molecular complex comprised of fusion proteins in which the biological effector moiety is an extracellular domain of a type I membrane receptor is shown schematically in FIG. 1D and such complexes are useful as biological effectors or to bind ligands. Type I membrane receptors useful in the invention include TNF receptors, ephrins, Ephs, VEGF receptors, IGF receptors, thrombospondin, thrombomodulin, PDGF receptors, IL-2R, TCR complex components, EGF receptors, TGF receptors, tissue factor, growth factor receptors, HGH receptor, IFN receptors, HER2, insulin receptor, etc.

S. Cytokines

In some embodiments, the biological effector moiety is a cytokine useful for modulating biological responses of cells. Cytokines useful in the invention include lymphokines such as macrophage activating factor (MAF), macrophage migration inhibition factor (MMIF), leukocyte migration inhibition factor (MCF), leukocyte migration inhibition factor (LMIF), a histamine releasing factor (HRF), or transfer factor (TF). Tumor necrosis factors, such as TNF-α (cachectin) and TNF-β (lymphotoxin) can be biological effector moieties. Interleukins, such as (L- 1, IL-2, IL-3, IL-4, (L-5, IL-6, IL- 7, IL-S, IL-9, IL-10, IL-11 , IL-12, IL-13, IL-14, and IL-15, IL-17, can be biological effector moieties. Interferons, such as IFN-α, IFN-β, IFN-γ, IFN-ω, and IF-τ, can be biological effector moieties.

Other useful cytokines include colony stimulating factors, chemokines, and stress proteins. Examples of colony stimulating factors include granulocyte colony stimulating factor (G-CSF), granulocyte macrophage colony stimulating factor (GM-CSF), macrophage colony stimulating factor (M-CSF), and multi-CSF (IL-3).

Examples of α-chemokines include IL-8, NAP-2 (neutrophil activating protein 2), PF-4 (platelet factor 4), and βTG (β-thromboglobulin). β-Chemokines include MCP-1 (monocytes chemoattractant protein 1), MCP-3, MIP-1α (macrophage inflammatory protein 1α), MIP-1β, and RANTES ("Regulated upon Activation Normal T Expressed and presumably Secreted chemokine"). Other useful chemokines include, e.g., CCL chemokines; CXC (SCYB) chemokines or CX3C chemokines; XC chemokines; and CC chemokines, such as CCL2, CCL7, CCL11 , CCL8, CCL13, CCL1 , CCL5, CCL16, CCL14, CCL15, CCL23, CCL18, CCL3 and CCL4.

Stress proteins include heat shock proteins (HSPs), glucose-regulated proteins (GRPs), ubiquitin, and superoxide dismutase.

C. Enzymes

In other embodiments, a biological effector moiety is an enzyme, e.g., a proteolytic enzyme (an amino peptidase, an aspartyl protease, a serine protease, a metailo protease, a cysteinyl protease, pepsin, trypsin, thrombin, lysozyme, Factor VII, Factor X, Factor IX). Other enzymes, such as glycosidases, esterases, hydrolases, nucleases, syntheases, isomerases, polymerases, kinases, phosphatases, reductases, including oxido-reductases, transferases, ligases, restriction enzymes, amidases, ATPases, carbohydrases, lipases, cellulases, dehydrogenases, and oxidases also can be used.

D. Toxins

Therapeutically useful toxins can also be used as biological effector moieties in the fusion proteins comprising the dimeric molecular complexes of the invention. There are numerous examples of such toxins, well known to those skilled in the art, such as the bacterial toxins Pseudomonas exotoxin A and diphtheria toxin, and the plant toxins ricin, abrin, modeccin, saporin, and gelonin.

E. Extracellular Domains of Type Il Membrane Receptors

A dimeric molecular complex also can be constructed wherein the biological effector moiety is an extracellular domain of a type Il membrane receptor. In this case, however, each of the two fusion proteins comprises, from N to C terminus, (a) a CH4 dimerization domain with a C terminal extension comprised of an M2" IgE CH4 splice variant (ESSRRGGC (SEQ ID NO:27)), (b) an amino acid linker (preferably 3-10residues in length) which is covalently bound to the CH4 dimerization domain and (c) an extracellular domain of the type Il membrane receptor. See FIG. 2. In some embodiments, the sequence of the M2" IgE splice variant can be modified to provide additional sites for chemical modification (e.g. ESSKKGGC (SEQ ID NO:28), ESSCRGGC (SEQ ID NO:29) or ESSRCGGC (SEQ ID NO-.30)).

Type Il membrane receptors comprise only -5% of transmembrane proteins, but include members with important biological effector functions such as hepsin protease, ectodysplasin, collagenous membrane proteins, macrophage scavenger receptors, MARCO protein, TNF ligand-like proteins,
asialoglycoprotein receptors, lymphocyte IgE receptor, Kupffer cell receptor, NKG2, NKR-P1 , Ly-49, CD69, CD72, LyB-2, collectins, CLEC5A, etc.

PRODUCTION OF DIMERIC MOLECULAR COMPLEXES

Fusion proteins for dimeric molecular complexes of the invention can be produced recombinantly or synthetically, or using a combination of the two approaches. For recombinant production, the invention provides nucleic acid molecules which encode fusion proteins of the invention (see below).

It is possible to produce a fusion protein of the invention using chemical methods to synthesize the amino acid sequence of the fusion protein. Methods include direct peptide synthesis using solid-phase techniques (Merrifield, J. Am. Chem. Soc. 85, 2149-2154, 1963; Robergβ et al., Science 269, 202-204, 1995). Protein synthesis can be performed using manual techniques or by automation. Automated synthesis can be achieved, for example, using Applied Biosystems 431 A Peptide Synthesizer (Perkin Elmer). Optionally, fragments of a fusion protein can be separately synthesized and combined using chemical methods to produce a full-length fusion protein. See WO 01/98340.

NUCLEIC ACID MOLECULES

Nucleic acid molecules of the invention can comprise any nucleotide sequence which encodes the desired fusion protein. Nucleic acid molecules of the invention include single- and double-stranded DNA (including cDNA) and mRNA. Many kits for constructing fusion proteins are available from companies such as Promega Corporation (Madison, Wl), Stratagene (La JoIIa, CA), CLONTECH (Mountain View, CA), Santa Cruz Biotechnology (Santa Cruz, CA), MBL International Corporation (MIC; Watertown, MA), and Quantum Biotechnologies (Montreal, Canada; 1 -888-DNA-KITS).

Methods which are well known to those skilled in the art can be used to construct nucleic acid molecules of the invention. These methods include in vitro recombinant DNA techniques and synthetic techniques. Such techniques are described, for example, in Sambrook et a/. (1989) and in Ausubel et a/., CURRENT PROTOCOLS IN MOLECULAR BIOLOGY, John Wiley & Sons, New York, N.Y., 1989.

In some embodiments, the nucleic acid molecules are expression constructs which contain the necessary elements for the transcription and translation of an inserted coding sequence encoding a fusion protein. Fab dimer expression constructs can include a coding sequence for the light chain with a C-terminal cysteine. An expression construct can be present in a vector suitable for introducing fusion proteins of the invention into a cell.

Fusion proteins of the invention can be recombinant^ expressed in a variety of host cells. These include, but are not limited to, microorganisms, such as bacteria transformed with recombinant bacteriophage, plasmid, or cosmid DNA expression vectors; yeast transformed with yeast expression vectors, insect cell systems infected with virus expression vectors {e.g., baculovirus), plant cell systems transformed with virus expression vectors (e.g., cauliflower mosaic virus, CaMV; tobacco mosaic virus, TMV) or with bacterial expression vectors (e.g., Ti or pBR322 plasmids), or animal cell systems, particularly mammalian systems, including human systems. See WO 01/98340, which is incorporated herein by reference in its entirety. The choice of vector components and appropriate host cells is well within the capabilities of those skilled in the art.

In some embodiments of this invention, both fusion proteins of a dimeric molecular complex are expressed in the same cell, preferably from the same plasmid, e.g., as a dicistronic operon (Skerra et al, Protein Eng. 4, 971 , 1991). A signal sequence can be included to direct the fusion proteins to the desired cellular location. Expression of the two fusion proteins of a dimeric molecular complex from the same plasm id leads to an increased amount of the bispecific dimer being formed, as equivalent amounts of each component are being produced within the cell.

Optionally, a fusion protein can comprise a moiety which can be used as a detection or purification tag, such as peptides comprising at least five histidine residues, or the commonly used c-myc and FLAG tags.

DIAGNOSTIC METHODS

The dimeric molecular complexes of the invention can also be useful for diagnostic purposes. For example, the complex can comprise two fusion proteins, one protein comprising a biological effector moiety designed to bind to an analyte of interest and the other protein comprising a biological effector molecule that is, or is bound to, a detectable label which can easily be quantified, e.g. an enzyme, a fluorescent protein, a radionuclide, etc.

THERAPEUTIC METHODS

Dimeric molecular complexes of the invention can be provided in a pharmaceutical composition for administration to a mammal, preferably a human. Complexes composed of antibody fragments (either mono- or bi-specific) are particularly useful in tumor therapy. For example, one fusion protein of the complex can comprise a molecule which binds to a tumor marker and the other fusion protein can comprise a molecule which binds to a T-cell epitope, a toxin, or a radionuclide binding peptide or protein to bring a killing function close to the tumor cell.

For preparing suitable pharmaceutical compositions comprising dimeric molecular complexes of the invention, one skilled in the art can use known injectable, physiologically acceptable sterile solutions. For preparing a ready-to-use solution for parenteral injection or infusion, aqueous isotonic solutions, such as, e.g., saline or corresponding plasma protein solutions are readily available. The
pharmaceutical compositions may be present as lyophylisates or dry preparations, which can be reconstituted with a known injectable solution directly before use under sterile conditions. A pharmaceutical composition can be supplemented with known carrier substances or/and additives (e.g., serum albumin, dextrose, sodium bisulfite, EDTA, etc.).

Pharmaceutical compositions of the invention can be administered by different routes of application known to one skilled in the art, particularly by intravenous injection or direct injection into target tissues. For systemic application, intravenous, intravascular, intramuscular, intraarterial, intraperitoneal, oral, or intrathecal routes can be used. More local administration can be effected subcutaneously, intracutaneous^, intracardially, intralobally, intramedullarly, intrapulmonarily or directly in or near the tissue to be treated (connective-, bone-, muscle-, nerve-, epithelial tissue). Depending on the desired duration and effectiveness of the treatment, compositions may be administered once or several times, also intermittently, for instance on a daily basis for several days, weeks or months and in different dosages.

The dosage will depend on age, condition, sex and extent of the disease in the patient and can vary from 0.1 mg/kg to 200 mg/kg, preferably from 0.1 mg/kg or 100 mg/kg/dose, in one or more dose administrations daily, for one to several days.

All patents, patent applications, and references cited in this disclosure are expressly incorporated herein by reference. The above disclosure generally describes the present invention. A more complete understanding can be obtained by reference to the following specific examples, which are provided for purposes of illustration only and are not intended to limit the scope of the invention.

EXAMPLE 1

Production of scFv dimeric molecular complexes

This example demonstrates the production of two dimeric molecular complexes with the structure shown in FIG. 1 A. Two constructs were prepared. In both constructs, a poly-Gly linker was added to the N-terminus of the wild-type IgG1 hinge region. In a "wild-type" construct, this was followed by the native sequence of some of the first beta strand of CH2 of human IgGI . In the "mutant" construct, mutations were introduced to eliminate hydrophobic residues present in the "wild-type" beta strand which would otherwise be exposed to solvent and could potentially hinder solubility. Two hydrophobic residues, VaI240 and He242 within the heavy constant region of IgG1 (SEQ ID NO:25), have hydrophobic interactions with the IgG CH2 domain, and two phenylalanines, Phe241 and Phe243, interact with the carbohydrate groups. These hydrophobic amino acids VaI, Phe, Leu, and Phe at positions 240-243 of SEQ ID NO:25 were replaced by the hydrophilic residues Asp, Ser, GIu, and Tyr, respectively. Two cysteine groups in the hinge region result in covalent links between each fusion protein upon dimerization of the CH4 regions.

Generation of the single chain antibody

A single chain antibody termed 19G9scFv was generated from VH and VL chains which were amplified by PCR using primers which introduced restriction sites and sites for overlap extension of the 16-amino acid linker GGGGSGGGGSGGGGSGGGGS (SEQ ID NO:5). After 15 cycles of extension the scFv was amplified with the original forward and back primers and cloned by restriction digest into the bacterial expression vector pz613. See FlG. 3.

VH forward primer: GCGGCCCAGCCGGCCATGGCCCAGGTTCAGCTGGTGCAGTC (SEQ ID NO:15);

VH back primer:
CCACCGGAGCCGCCGCCGCCAGAACCACCACCACCAGAACCACCACCACCTGAAGAGAC GGTGACC (SEQ ID NO:16);
VL forward primer:
GGCGGCGGCGGCTCCGGTGGTGGTGGATCCGAAATTGTGTTGACGCAGTC
(SEQ ID NO:17);
VL back primer: GCGGCCGCTTTGATCTCCACCTTGGTCC (SEQ ID NO:18).

Generation of the CH4 dimeπzation domain

The CH4 domain of IgE was cloned from mRNA purchased from Invitrogen (clone 2132581) using PCR primers "ATG001" (lgECH4_bk_a): CCGTCAGTCT TCCTCTTCCC CCCGCGTGCT
GCCCCGGAAG (SEQ ID NO:9) and "ATG003" (lgECH4_for_a): CGGATACGGC ACCGGCGCAC CTTTACCGGG ATTTACAGAC (SEQ ID NO: 11 ). The lgd hinge region and the first beta strand of IgG1 CH2 was introduced by PCR using primer "ATG002": TTCCTCTTCC CCCCGCGTGC
TGCCCCGGAA G (SEQ ID NO:10). Λ/ofi and SgrA1 sites were introduced by primers ATG001 and ATG003, respectively. These sites permitted direct cloning by restriction digestion of the IgGi hinge-IgE CH4 region fusion into the single chain vector ("19G911). See FIG. 4.

The amino acid sequence of the "wild-type" 19G9scFv fusion protein, sometimes referred to as 19G9M1 (SEQ ID NO:3) is shown in FIG. 5A. The amino acid sequence of the "mutated" fusion protein, referred to as 19G9M2 (SEQ ID NO:1), which contains the sequence DSEY at amino acid residues 277-280 in the dimerization domain, is shown in FIG. 5B.

EXAMPLE 2

Expression, purification, and function ofscFv dimeric molecular complexes

The "mutant" 19G9scFv dimeric molecular complex construct (19G9M2; SEQ ID NO:1) described in Example 1 was cloned into the mammalian expression vector pPEPIpoly and expressed in CHO cells. Purification was carried out using epitope-tag affinity chromatography, and the purified complex was tested for binding, antigen specificity, and affinity using an ELISA. See FIG. 6A. The molecular weight of the purified complex (reduced and non-reduced) was analyzed in an SDS-PAGE system. 4%-12% SDS-PAGE was used, with a running buffer of MES SDS at 40 minutes at 12OmA, 200V. See FIG. 6B.

FIG. 7 shows the results of binding of the 19G9M2 dimeric molecular complex to BHK-RG-1 protein compared with the binding of the corresponding diabody (the dimeric single chain antibody made from the 19G9 VH and VL linked by a shortened five amino acid linker). The diabody and the IgG of 19G9 have equivalent activity. The EC50 is roughly nanomolar. The antibody concentration was calculated based upon molecular weights of 80 kD for the dimeric molecular complex and 50 kD for the diabody. The secondary antibody used in the assay was HRP-anti-E tag antibody.

EXAMPLE 3

Demonstration of disulfide bond formation in the hinge region

FIG. 6B is a Coomassie blue-stained SDS-polyacrylamide gel demonstrating that the dimeric molecular complex preparation described in Examples 1 and 2 comprises predominantly disulfide dinners (lanes 2 and 3) with only a small contamination by monomer. Upon reduction, all the disulfide dimers collapse to monomers (lanes 5 and 6). This result demonstrates that the hinge region from the IgG1 functions properly and that disulfide bonds form in the hinge after the IgE CH4 domain dimerizes. Lanes, from left to right: 1 , 10 μ\ marker; 2, 2 μ\ 19G9 M2; 3, bμ\ 19G9 M2; 4, blank; 5, 2μ\ reduced 19G9 M2; 6, 5 μ\ reduced 19G9 M2; 7, ΛQ μ\ marker.

EXAMPLE 4
In vivo biodistribution of the 19G9 dimeric molecular complex

Methods for radiolabeling and chelator attachment were adapted from Nikula et al. (1995), Nucl. Med. Biol., 22, 387-390. Purified 19G9scFv dimeric molecular complex, 19G9 IgG, 19G9scFv and the 19G9 diabody were characterized by SEC and SDS-PAGE prior to use. Labeling was performed using 111In, in buffers and equipment which were rendered metal-free by repeated rinsing with 1OmM EDTA solution and Chelex- treatment prior to filtration. The chelator p-SCN-CHX-A"-DPTA was purchased from Macrocyclics Inc. The buffer used for conjugation contained 5OmM carbonate, 15OmM NaCI, pH 6.5. Radiolabeling was performed in a buffer containing 5OmM NaAc, 150 mM NaCI, pH 6.5.

111In -radiolabeled antibody solutions (PBS) of the four antibody groups described above were injected into 8 LNCaP tumor bearing mice per group (2/vCi per animal). Two animals per group were weighed, sacrificed and tissues harvested at 15 min, 3 hr, 6 hr and 48 hr following injection of labeled antibody. The tumor, tissue or blood sample was weighed and the total radioactivity determined. The mCi per gram of sample was compared with the total mCi per animal weight to determine the percentage of the injected dose per gram tissue and the average calculated for each group and time point (mean % ID per gram). The results are shown in FIG. 9.

EXAMPLE 5

Production of Fab fusion proteins and expression of Fab dimeric molecular complexes

This example demonstrates the production of Fab dimeric molecular complexes with the structure shown in FIG. 1 B (i.e. a monospecific dimeric molecular complex in which the biological effector moiety is a Fab fragment). This construct does not have any linker sequence between the C-terminus of the CH 1 domain of the Fab and the N-terminus of the IgG1 hinge region and the sequence in this region is the same as in an IgG.

In the IgGi hinge region, four hydrophobic amino acids, VaI, Phe, Leu and Phe, (amino acid residues 263-266 of SEQ ID NO:20) located at the C-terminus of the hinge region were replaced by the hydrophilic residues Asp, Ser, GIu, and Tyr respectively, to avoid potential difficulty in solubility. The two cysteine groups present in the hinge region result in covalent links between each fusion protein upon dimerization of the CH4 regions. Another cysteine is also located at the C-terminus of both CH 1 and CL to allow these domains to be linked by a covalent disulfide bond.

(a) Insertion of IgGI hinge region and IgE CH4 domain into IgGI expression vector

DNA coding for the IgGI hinge region and the IgE CH4 domain was amplified by PCR using primers which introduced Pcil and Fsel restriction sites, and this region was then cloned by restriction digestion into the cloning vector pCR2.1_TOPO. to produce pCR2.1_hinge_lgE CH4 to create the heavy chain of the Fab fusion protein of the J_DSEY dimeric molecular complex.

ATG47 forward primer: CTCACACATGTCCACCGTGCCCAGCACCTGAAC (SEQ ID NO:31); ATG46 reverse primer: TGTGGCCGGCCCTATTTACCGGGATTTACAGACACCGCT (SEQ ID NO:32);
The VH and CH domains of a Fab antibody (ABJ), was amplified by PCR using primers which introduced a Pcil restriction site, and cloned by restriction digestion (EcoRV and Pcil) into
pCR2.1_hinge_lgE CH4.

ATG51 forward primer: GTTGAAATT AAACGTACGGTGGCTGC (SEQ ID NO:33);
ATG56 reverse primer: GACATGTGTGAGTTTTATCGCAGCTTTTCGGTTCCACTTTTTTATCC (SEQ ID NO:34);
The entire heavy chain complex was amplified by PCR using primers which introduced an Xma restriction site, and then cloned by restriction digestion (BIpI and Xmal) into an IgG-, expression vector, plE_ABJ to produce the vector plE-JJ3SEY. See FIG.10

ATG75 forward primer: CAGGTGCAATTGGTTCAGAGCG (SEQ ID NO:35);
ATG76 reverse primer: CGACTCCCGGGTTACTATTTACCGGGATTTACAGACAC (SEQ ID
NO-.36).

(b) Generation of a second heavy chain Fab fusion protein

A second Fab dimeric molecular construct with a different (more basic) IgG1 hinge domain was generated from the J_DSEY dimer heavy chain and IgE CH4 domain (SEQ ID NO:20) by replacing the amino acids Asp, Ser, GIu and Tyr at positions 263 to 266 of SEQ ID NO:20 with Lys, Ser, Lys and Tyr (KSKY) using the QuikChange Il Site-Directed Mutagenesis Kit provided by Stratagene and an appropriate set of primers.

ATG85 forward primer: GGGGGACCGTCAAAAAGCAAATACCCGCCGCGTGC(SEQ ID
NO:37);
ATG86 reverse primer: GCACGCGGCGGGT ATTTGC I I I I I GACGGTCCCCC (SEQ ID
NO:38);
producing the vector plE-J_KSKY.

EXAMPLE 6

Expression and purification of the J_DSEY and J_KSKY Fab dimeric molecular complexes

The J_DSEY and J_KSKY Fab heavy chain fusion protien constructs described in Example 5 were each cloned into the mammalian expression vector plE-ABK and expressed in CHO-K1 cells. Purification of each expression product was carried out using Protein L affinity chromatography, and the purified complexes were tested for binding, antigen specificity and affinity using an ELISA. The molecular weight of the purified complexes was analyzed in an SDS-PAGE system under reducing conditions (4%-12% SDS-PAGE, with a running buffer of MOPS SDS at 60 minutes at 12OmA, 200V) and were shown to possess the predicted molecular weight.

EXAMPLE 7

Production of another heavy chain Fab fusion protein

Another Fab dimeric molecular complex using an antibody to the same antigen (AB25) was produced.

The heavy chain variable region of AB25 (SEQ ID NO:42) was cloned by PCR using primers which introduced a Mfel restriction site, and inserted by restriction digestion (Notl and BIpI) into the plE-J_DSEY expression vector.

ATG 105 forward primer:
GCGCGGCCGCGCCACCATGAAACACCTGTGGTTCTTCCTCCTGCTGGTGGCAGCTCCCA GATGGGTCCTGTCCCAGGTGCAATTGGTTCAGAGCGGCGCG (SEQ ID NO:39);
ATG106 reverse primer: CCTTTGGTCGACGCTGAGCT (SEQ ID NO:40);
The light chain variable region of AB25 (SEQ ID NO:44) was inserted by restriction digestion (EcoRI and BsiWI) into the plE-J_DSEY expression vector.

EXAMPLE 8

Expression and purification of the 25_DSEY Fab dimeric molecular complex

The 25_DSEY Fab dimeric molecular complex construct described in Example 7 was expressed in CHO-K1 cells. Purification was carried out using Protein L affinity chromatography, and the purified complex was tested for binding, antigen specificity, and affinity using an ELlSA. The molecular weight of the purified complex was analyzed in an SDS-PAGE system under reducing conditions and was shown to possess the predicted molecular weight. As expected, the binding activity of the dimeric molecular complex (25_DSEY) was improved compared to that of the Fab, and equivalent to the binding found with the IgG form. The EC5Q1S for the dimeric 25_DSEY Fab dimer and the IgG are both in the sub-nanomolar range. Previous studies comparing the monomeric Fab and dimeric IgG forms of AB25 in similar assays show that the dimeric IgG bound with greater avidity than did the monomeric Fab, consistent with these results. See FIG. 14.