The effect of variable domain orientation and arrangement on the antigen-binding activity of a recombinant human bispecific diabody

The effect of variable domain orientation and arrangement on the antigen-binding activity of a recombinant human bispecific diabody

BBRC Biochemical and Biophysical Research Communications 318 (2004) 507–513 www.elsevier.com/locate/ybbrc The effect of variable domain orientation an...

317KB Sizes 0 Downloads 2 Views

BBRC Biochemical and Biophysical Research Communications 318 (2004) 507–513 www.elsevier.com/locate/ybbrc

The effect of variable domain orientation and arrangement on the antigen-binding activity of a recombinant human bispecific diabodyq Dan Lu, Xenia Jimenez, Larry Witte, and Zhenping Zhu* Departments of Antibody Technology and Molecular and Cell Biology, ImClone Systems Incorporated, New York, NY 10014, USA Received 15 March 2004 Available online 22 April 2004

Abstract In recent years a variety of recombinant methods have been developed for efficient production of bispecific antibodies (BsAb) in various formats. Bispecific diabody (bDAb), a 55–60 kDa molecule comprising two non-covalently associated cross-over single chain Fv (scFv) polypeptides, represents one of the most promising as well the most straightforward approaches to BsAb production. Here we constructed a bDAb, using two human scFv, 11F8 and A12, directed against the epidermal growth factor receptor (EGFR) and the insulin-like growth factor receptor (IGFR), respectively, as the building blocks. A total of 8 scFv and diabody constructs were prepared comprising the same two variable heavy (VH ) and variable light (VL ) chain domains but arranged in different orientations. VH /VL orientation, i.e., VH –linker–VL or VL –linker–VH , showed significant effects on the expression and antigen-binding activity of scFv and monospecific diabody of both 11F8 and A12. Further, only 2 out of the 4 possible VH /VL orientations/arrangements in bDAb construction yielded active products that retain binding activity to both EGFR and IGFR. Both active bDAb preparations retained their original antigen-binding activity after incubation at 37 °C in mouse serum for up to 7 days, indicating excellent stability of the constructs. Taken together, our results underscore the importance of identifying/selecting optimal VH /VL orientation/arrangement for efficient production of active bDAb. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Diabody; Bispecific antibody; Antibody engineering; Variable domain; Domain orientation; EGFR; IGFR

Bispecific antibodies (BsAb) are immunoglobulinbased molecules that bind to two different epitopes on either the same or distinct antigens. Both laboratory and early clinical studies to date have demonstrated that BsAb may have significant potential application in cancer therapy by either targeting tumor cells with cytotoxic agents including effector cells, radionuclides, drugs, and toxins [1–3], or by targeting simultaneously two different tumor-associated antigens (or epitopes) for enhanced q Abbreviations: BsAb, bispecific antibodies; bDAb and mDAb, bispecific and monospecific diabody, respectively; EGFR, epidermal growth factor receptor; IGFR, insulin-like growth factor receptor; RT, room temperature; scFv, single chain Fv; V, variable; VH and VL, the variable domains of antibody heavy and light chains, respectively; VEGFR2, vascular endothelial growth factor receptor 2. * Corresponding author. Fax: 1-212-645-2054. E-mail address: [email protected] (Z. Zhu).

0006-291X/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2004.04.060

antitumor activity [4–6]. A major obstacle in the development of BsAb-based therapeutics has been the difficulty in producing the materials in sufficient quantity and quality for clinical studies via traditional methods, including the hybrid hybridoma and chemical conjugation [1,7]. In recent years, a variety of recombinant methods have been developed for efficient production of BsAb, both as antibody fragment [7–9] and full length IgG formats [10]. One of the most promising methods is the so-called “diabody” approach [11–13]. Diabody is a form of scFv dimer with a molecular size of approximately 55–60 kDa. The construction of scFv fragment, with a short linker (5–10 amino acid residues) between the variable heavy (VH ) and the variable light (VL ) domains, permits interchain, but not intrachain, pairing of the variable (V) domains, resulting in the formation of a bivalent antibody fragment known

508

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513

as a diabody [11–13]. A bispecific diabody (bDAb) can be produced by coexpression of two “cross-over” scFv fragments in which the VH and the VL domains for the two antibodies are present on different polypeptide chains. Interchain pairing of the two cross-over polypeptide chains results in the formation of a scFv dimer, or bDAb, which is a divalent antibody molecule with monovalency to each of its target antigen [11,13]. A number of bDAb have been generated in the past years and proved to be capable of simultaneously binding to both of the target antigens, including soluble antigens and antigens expressed on the surface of either the same or two different cells [4,5,14,15]. Since there are 4 antibody V domains, two VH and two VL , involved in the construction/formation of a bDAb, it is plausible that the orientation/arrangement of individual VH and VL domains within the bDAb construct may have significant impact on the expression, antigen-binding activity, and protein stability of the resulting bDAb. In this report, we studied these issues with the construction of a bDAb in several different formats regarding the orientation/ arrangement of the V domains, using two human antibodies, one directed against the epidermal growth factor receptor (EGFR) and the other against the insulin-like growth factor receptor (IGFR), as the building blocks.

Materials and methods Antibodies and proteins. Fully human antibodies directed against human EGFR (clone 11F8) and IGFR (clone A12) were isolated from an antibody Fab fragment phage display library containing 3.7  1010 unique clones (Dyax, Cambridge, MA; also see [16]) as previously described [6,17]. Recombinant EGFR extracellular domain was pro-

duced at ImClone Systems (New York, NY). Recombinant IGFR extracellular domain was purchased from R&D Systems (Minneapolis, MN). Construction of various scFv and diabodies. The VH and VL genes of clone 11F8 and A12 were, respectively, amplified and assembled into scFv format by overlapping PCR, using the 15-amino-acid linker (Gly–Gly–Gly–Gly–Ser)3 , in two different orientations: VH –linker–VL or VL –linker–VH (Fig. 1), following a previously described procedure [5]. In constructing the monospecific diabodies (mDAb), the VH and VL genes of respective antibody were linked together via a 5-aminoacid linker, Arg–Thr–Val–Ala–Ala, also in two different orientations. Four different versions of bDAb, the anti-EGFR (11F8)  antiIFGR (A12) BsAb, were constructed in regard to the orientation/arrangement of various V domains (Fig. 1). In constructing a bDAb, four polypeptide chains were first generated: two consist of the VL of the 11F8 linked to the VH of A12 via the 5-amino-acid linker (Arg– Thr–Val–Ala–Ala), as 11F8VL –linker–A12VH or the reversed orientation, A12VH –linker–11F8VL ; and the other two consist of the VH of 11F8 linked to the VL of A12 via the same 5-amino-acid linker, as 11F8VH –linker–A12VL or A12VL –linker–11F8VH , followed by subcloning of two relevant polypeptides into a bi-cistronic vector for coexpression in Escherichia coli (Fig. 1). A 13-amino-acid long tag, the E tag, was fused to the C terminus of the second polypeptide for purification and detection purposes [18]. All sequences encoding the scFv and diabodies were verified by DNA sequencing. Expression and purification of the diabody. The scFv and diabodies were secreted from E. coli strain HB2151 containing the expression plasmid grown at 30 °C in a shaker flask, a periplasmic extract of the cells was prepared, and the soluble scFv and diabody were purified from the extract by anti-E tag affinity chromatography using the RPAS Purification Module (Amersham–Pharmacia Biotech, Piscataway, NJ) following a procedure previously described [4,18]. To examine the purity of the preparations, the purified scFv and diabodies were electrophoresed in an 18% polyacrylamide gel (Novex, San Diego, CA) and visualized by staining with Colloidal Blue Stain kit (Novex). Antigen-binding activity of the scFv and the diabodies. Various amounts of scFv or diabodies were added to EGFR or IGFR coated 96-well plates and incubated at room temperature (RT) for 1 h, after which the plates were washed 3 times with PBS containing 0.1%

Fig. 1. Schematic diagrams showing the expression constructs for various scFv and diabodies. ScFv and mDAb of 11F8 and A12 were constructed using both VH –VL and VL –VH orientations. A total of four bDAb variants were constructed representing all the possible V domain orientations/ arrangements between the two different sets of VL and VH domains. L, linker, equals 15 amino acids, (Gly–Gly–Gly–Gly–Ser)3 , in all the scFv constructs, and 5 amino acids, Arg–Thr–Val–Ala–Ala, in all the diabody constructs. E, the 13-amino-acid E-tag used for purification and detection. Note. Drawings are not to scale.

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513 Tween 20. The plates were then incubated at RT for 1 h with 100 ll of an anti-E tag antibody-HRP conjugate (Amersham–Pharmacia Biotech). The plates were washed, peroxidase substrate was added, and the absorbance at 450 nm was read following the procedure described previously [4,18]. Binding kinetics analysis of the scFv and the diabodies. The binding kinetics of the scFv and the diabodies to their respective targets were measured using a BIAcore biosensor (BIAcore 3000, Biacore, Uppsala, Sweden). EGFR or IGFR was immobilized onto a sensor chip and the antibodies were injected at concentrations ranging from 1.5 to 200 nM. Sensorgrams were obtained at each concentration and were evaluated using a program, BIA Evaluation 2.0, to determine the rate constants, the association rate (kon ), and the dissociation rate (koff ). The affinity constant, Kd , was calculated from the ratio of rate constants koff /kon . Stability of scFv and bDAb. The scFv or bDAb were added to PBS containing 10% mouse serum and incubated at RT or 37 °C. Aliquots of samples were removed at predefined intervals of incubation and assayed for efficiency for binding to their respective targets using ELISA described above.

Results and discussion Construction and expression of scFv and mDAb with opposite VH /VL orientations The VH and VL domains from two human Fab antibodies originally isolated from a phage display library, 11F8 and A12, directed against EGFR and IGFR, respectively, were used as the building blocks to construct the scFv and the diabodies including both mDAb and bDAb (Fig. 1A). Two versions of scFv and mDAb of both 11F8 and A12 were generated in either the VH – linker–VL or the VL –linker–VH orientation. The most commonly used 15-amino-acid linker, (Gly–Gly–Gly– Gly–Ser)3 , was used for the construction of scFv, whereas a shorter, 5-amino-acid linker, Arg–Thr–Val– Ala–Ala, representing the first 5 amino acids from the N-terminus of the human antibody j light chain constant domain, was chosen as the linker for the construction of the diabodies. This linker sequence was used successfully in our previous construction of a number of bDAb and was regarded as a preferred choice over the unnatural (non-human) linkers, such as the Gly–Gly– Gly–Gly–Ser sequence, based on the argument that the human sequence would potentially be less immunogenic in human therapy [13]. Both the scFv as well as the diabodies were expressed in E. coli and purified from the periplasmic extracts of the bacteria with affinity chromatography. The purity and identify (size) of these constructs were verified with SDS–PAGE. All the constructs yielded either a single (for scFv) or two (for the diabodies) major protein band(s) with expected molecular size/mobility (not shown). The V domain orientations did not have significant effect on the expression of both 11F8 scFv and A12 scFv (expression level ranged from 560 to 760 lg/L of overnight culture in shaker flasks). On the other hand, both 11F8 mDAb and A12 mDAb were expressed

509

at a much higher level when constructed as the VL –VH orientation: 682  40 lg/L for 11F8 mDAb (VL –VH ) versus 216  22 lg/L for 11F8 mDAb (VH –VL ), and 1275  10 lg/L for A12 mDAb (VL –VH ) versus 372  203 lg/L for A12 mDAb (VH –VL ), respectively. Antigen-binding of scFv and mDAb with opposite VH /VL orientations The antigen-binding efficiency of the scFv and the mDAb of 11F8 and A12 was examined on immobilized recombinant receptors using an ELISA. As shown in Fig. 2, except for 11F8 scFv where both V domain orientations yielded a product that binds to EGFR with equal efficiency, the VL –VH orientation resulted in scFv (A12) and mDAb (for both A12 and 11F8) with much better antigen-binding activity than those in the VH –VL orientation. For example, the ED50 , the antibody concentration that yielded 50% of maximum binding, was approximately 0.08 and 0.4 nM for A12 scFv (VL –VH ) and A12 scFv (VH –VL ), respectively, and 0.03 and 0.2 nM for A12 mDAb (VL –VH ) and A12 mDAb (VH –VL ), respectively. The binding kinetics of the scFv and the mDAb to their respective target antigens were determined by surface plasmon resonance using a BIAcore instrument. As expected, both 11F8 and A12 in diabody format bind to their targets with much higher affinity than their respective scFv counterparts, due to significant improvement in the kon (for 11F8 mDAb only) and/or the koff (for both 11F8 mDAb and A12 mDAb) resulting from the bivalent binding of the diabodies (versus the monovalent binding of the scFv) (Table 1). Consistent with ELISA, both V orientations yielded an 11F8 scFv with similar binding kinetics, whereas the VL –VH orientation resulted in an A12 scFv of higher binding affinity. The V orientation did not show, however, any significant effect on the binding kinetics of both 11F8 mDAb and A12 mDAb (Table 1). Expression and dual antigen-binding of the bDAb constructed via various orientations/arrangements In constructing a bDAb with two sets of different VH and VL domains, there exist a total of 4 possible combinations regarding the orientation/arrangement of each individual VH and VL domains (Fig. 1). Using the VL and the VH domains from 11F8 and A12, all the 4 different bDAb constructs were generated. The constructs were expressed in E. coli and purified using an anti-E tag affinity chromatography with comparable yield with the exception of bDAb-1: 141  83, 565  75, 312  130, and 505  261 lg/L in shaker flasks for bDAb-1, bDAb-2, bDAb-3, and bDAb-4, respectively. When analyzed using electrophoresis all four purified products yielded two major protein bands of expected size (Fig. 3A): a

510

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513

Fig. 2. Antigen-binding efficiency of various 11F8 and A12 scFv and mDAb to their respective targets, EGFR and IGFR. Various amounts of scFv or diabodies were added to 96-well plates coated with EGFR or IGFR (1.0 lg/ml) and incubated at RT for 1 h, after which the plates were incubated with a mouse anti-E tag antibody-HRP conjugate for additional 1 h. The plates were washed, peroxidase substrate was added, and OD450 nm was read. Data shown represent means  SD of triplicate samples.

Table 1 Binding kinetics of various scFv, mDAb, and bDAB to their respective targets as determined by BIAcore analysisa Molecule

EGFR binding kon (105 M

11F8 scFv (VH –VL ) 11F8 scFv (VL –VH ) 11F8 mDAb (VH –VL ) 11F8 mDAb (VL –VH ) A12 scFv (VH –VL ) A12 scFv (VL –VH ) A12 mDAb (VH –VL ) A12 mDAb (VL –VH ) bDAb 3 bDAb 4

11.5  4.9 10.9  3.4 23.6  4.4 41.3  9.7 NB NB NB NB 1.1  0.1 12.2  0.5

1

s 1)

IGFR binding koff (10

4

s 1)

28.7  12.8 21.9  9.1 13.5  7.9 19.2  8.5 NB NB NB NB 45  0.72 58  0.6

Kd (nM) 2.6  1.8 2.2  1.2 0.56  0.28 0.51  0.31 NB NB NB NB 41  2.7 4.8  0.74

kon (105 M b

NB NB NB NB 2.0  0.9 5.3  1.1 3.9  0.78 3.9  0.93 0.56  0.15 2.6  0.74

1

s 1)

koff (10

4

NB NB NB NB 19  0.4 13  2.2 2.6  1.0 2.1  0.5 9.5  3.9 10  3.6

s 1)

Kd (nM) NB NB NB NB 9.2  4.0 2.4  0.25 0.69  0.34 0.55  0.18 19  12.6 4.1  1.9

a All numbers are determined by BIAcore analysis and represent means  SE of 4–6 separate determinations. Kd values are calculated as the ratios of koff /kon . b NB, no binding.

larger band representing the second polypeptide with the E-tag fusion, and a smaller band corresponding to the first polypeptide, which is non-covalently associated with the E-tag-containing polypeptide to form the diabody (Fig. 1). An ELISA was employed to test the dual antigenbinding efficiency of the various bDAb constructs to

immobilized EGFR and IGFR. bDAb-4 demonstrated the best antigen-binding activity to both antigens, with ED50 of 0.15 and 0.4 nM to EGFR and IGFR, respectively. bDAb-3 also retained excellent reactivity to IGFR (ED50 , 0.4 nM) but with significantly reduced binding to EGFR (ED50 , 2 nM) (Figs. 3B and C). bDAb-1 and bDAb-2 only reacted weakly to EGFR

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513

Fig. 3. Expression and dual antigen-binding efficiency of various bDAb to EGFR and IGFR. (Top panel) The 4 bDAb constructs were expressed in E. coli, purified with an anti-E-tag affinity chromatography, and analyzed on an 18% polyacrylamide gel. The two major protein bands in each lane represent the two polypeptides constituting the diabodies—only the larger (top band) polypeptide contains the Etag. (Middle and bottom panels) Binding of the bDAb to EGFR and IGFR. Various amounts of bDAb were added to 96-well plates coated with EGFR or IGFR (1.0 lg/ml) and incubated at RT for 1 h, after which the plates were incubated with a mouse anti-E tag antibodyHRP conjugate for additional 1 h. The plates were washed, peroxidase substrate was added, and OD450 nm was read. Data shown represent means  SD of triplicate samples.

(ED50 were 20 and 40 nM, respectively), and totally lost their binding to IGFR (Figs. 3B and C). bDAb-3 and bDAb-4 also demonstrated markedly different antigenbinding kinetics (Table 1). The overall binding affinities

511

of bDAb-4 to both EGFR and IGFR were comparable to those of the parent scFv, and were 10-fold and 4fold higher than those of bDAb-3 for binding to EGFR and IGFR, respectively. On the other hand, mainly due to a much reduced kon rate, bDAb-3 showed a 2- to 8-fold lower affinity than A12 scFv to IGFR, and 16to 18-fold lower affinity than 11F8 scFv to EGFR (Table 1). Although several previous reports have suggested that the V domain orientation/arrangement might have significant effects on the expression and antigen-binding activity of the resulting bDAb, there were no systematic studies thus far performed/reported in detail to support the notion. Various bDAb have been constructed using all the different V domain orientations/arrangements with various successes—the two formats that were most frequently used were “VH1 –VL2 /VH2 –VL1 ” and “VL1 – VH2 /VL2 –VH1 ” (i.e., as bDAb-1 and bDAb-4 as depicted in Fig. 1). While active bDAb have been generated using both the formats [4,5,11,13–15], at least in two cases the VH1 –VL2 /VH2 –VL1 format led to the expression of inactive molecules [5,13, and our unpublished observations]. In this study, using the V domains of two human antibodies as the building blocks we observed that only 2 out of the 4 possible V domain orientations/arrangements resulted in the production of active bDAb—the VH1 –VL2 /VH2 –VL1 format again, interestingly, failed to yield an active molecule. The exact mechanisms underlined the different behavior between the bDAb constructs may only be elucidated by crystallographic structural analysis of each individual bDAb protein [19,20]. In the present study both the active bDAb, bDAb-3, and bDAb-4, share an identical polypeptide (A12VL –11F8VH ) partnered with a second polypeptide comprising of the same V domains but arranged in opposite orientation (A12VH –11F8VL in bDAb-3 versus 11F8VL –A12VH in bDAb-4). The only difference between the active bDAb-3 and the inactive bDAB-1, as well as between the active bDAb-4 and the inactive bDAb-2, is the orientation of A12 VL and 11F8 VH within one of the polypeptides (Fig. 1). Since both polypeptides within each bDAb constructs were well expressed and are associated with each other (non-covalently) as expected as confirmed by the SDS–PAGE analysis (Fig. 3A), these observations strongly indicate that a free N-terminus of A12 VL domain is critical in reconstituting (or maintaining) dual antigen-binding activity of the bDAb. It is possible that the free N-terminus of A12 VL is important in initiating and/or maintaining correct and stable folding of A12 VL /VH domains to form the IGFR-binding surface, which then lead to subsequent stable association of 11F8 VL /VH domains to reconstitute the EGFR-binding activity. Requirement of a free N- or C-terminus of a component scFv for maintaining antigen-binding activity of a bDAb has been previously observed. For example, while

512

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513

anti-CD3 scFv constructed in both VH –VL and VL –VH orientations were active, the scFv lost their CD3 binding activity when they were fused through a peptide linker via either N- or C-terminus to an anti-CD19 scFv to form single chain bispecific diabodies [21]. Interestingly, a bDAb, constructed using the V domains form the same anti-CD3 and the anti-CD19 antibodies via the CD3VH –CD19VL /CD19VH –CD3VL orientation, was able to bind both of its antigens [21]. Since correct assembly/folding of each cognate VL /VH domains within a diabody depends on both the intrinsic properties of the interacting V domains as well as the extrinsic contributions from the partnering V domains, a single VL /VH domain orientation/arrangement is unlikely to serve as the universal choice in construction of active bDAb from any two sets of VL and VH domains from two different antibodies. In supporting this notion, we constructed another bDAb, in all the 4 V domain orientations/arrangements (as depicted in Fig. 1), using the

same V domains of 11F8 but paired with those from a different antibody, clone 1121, a human antibody directed against vascular endothelial growth factor receptor 2 (VEGFR2) [22]. While the construct in the same format of bDAb-4 (i.e., 1121VL –11F8VH /11F8VL – 1121VH ) showed excellent dual antigen-binding activity, construct in the format of bDAb-3 retained well the reactivity to VEGFR2 but not that to EGFR (P25-fold reduction)—the reverse held true for a construct in the format of bDAb-2. In contrast to 11F8/A12 bDAb, the construct in the format of bDAb-1 (11F8VH –1121VL / 1121VH –11F8VL ) also demonstrated good reactivity to both EGFR and VEGFR2, albeit with a moderately (3to 10-fold) reduced efficiency (Lu and Zhu, unpublished data). It is particularly noteworthy that over more than a half dozen bDAb we constructed, each directed against two different targets, the “VL1 –VH2 /VL2 –VH1 ” orientation/arrangement has always yielded a bDAb with good retention of dual antigen-binding activity

Fig. 4. Stability of bDAb when incubated at 37 °C in mouse serum. The bDAb were incubated at 37 °C with PBS containing 10% mouse serum for up to 7 days. Aliquots of samples were removed at 1, 3, and 7 days postincubation and assayed for efficiency for binding to both EGFR and IGFR by an ELISA described in the legend to Fig. 3. Data shown represent means  SD of triplicate samples.

D. Lu et al. / Biochemical and Biophysical Research Communications 318 (2004) 507–513

[5,13–15]. This V domain orientation/arrangement may, therefore, represent the preferred format (or the first choice) for bDAb construction. Stability of bDAb in mouse serum The product stability of the two active bDAb, bDAb3 and bDAb-4, as well as that of the 11F8 and A12 scFv preparations, was examined after incubation in the presence of 10% mouse serum at either RT or 37 °C for up to 7 days. Both bDAb preparations retained full dual antigen-binding activity to EGFR and IGFR after incubation for 1–7 days at RT (not shown) or 37 °C (Fig. 4), suggesting excellent stability of the diabody constructs. Under the same condition, both 11F8 scFv and A12 scFv also retained full binding activity to their respective target antigens (not shown). In summary, here by using the V domains from two human antibodies directed against EGFR and IGFR as the building blocks, we constructed a bDAb in 4 different formats and examined the effects of orientation/ arrangement of individual VH /VL domains on the expression and antigen-binding activity of the resulting bDAb. Although all 4 constructs were secreted from E. coli as non-covalent dimer (diabody), only two orientations/arrangements yielded active bDAb with dual antigen-binding activity. Taken together with those of our previous observations, these results underscore the importance of identifying/selecting optimal VH /VL orientations/arrangements for efficient production of active bDAb.

References [1] Y. Cao, M.R. Suresh, Bispecific antibodies as novel bioconjugates, Bioconjug. Chem. 9 (1998) 635–644. [2] A.B. van Spriel, H.H. van Ojik, J.G.J. van de Winkel, Immunotherapeutic perspective for bispecific antibodies, Immunol. Today 21 (2000) 391–397. [3] D.M. Segal, G.J. Weiner, L.M. Weiner, Introduction: bispecific antibodies, J. Immunol. Methods 248 (2001) 1–6. [4] D. Lu, H. Kotanides, J. Jimenez, Q. Zhou, K. Persaud, P. Bohlen, L. Witte, Z. Zhu, Acquired antagonistic activity of a bispecific diabody directed against two different epitopes on vascular endothelial growth factor receptor 2, J. Immunol. Methods 230 (1999) 159–171. [5] D. Lu, X. Jimenez, H. Zhang, Y. Wu, P. Bohlen, L. Witte, Z. Zhu, Complete inhibition of vascular endothelial growth factor (VEGF) activities with a bifunctional diabody directed against both VEGF receptors, fms-like tyrosine kinase receptor and kinase insert domain-containing receptor, Cancer Res. 61 (2001) 7002–7008. [6] D. Lu, H. Zhang, D. Ludwig, A. Persaud, X. Jimenez, D. Burtrum, P. Balderes, M. Liu, P. Bohlen, L. Witte, Z. Zhu, Simultaneous blockade of both the epidermal growth factor receptor and the insulin-like growth factor receptor signaling pathways in cancer cells with a fully human recombinant bispecific antibody, J. Biol. Chem. 279 (2004) 2856–2865.

513

[7] P. Carter, J. Ridgway, Z. Zhu, Toward the production of bispecific antibody fragments for clinical applications, J. Hematother. 4 (1995) 463–470. [8] A. Pl€ uckthun, P. Pack, New protein engineering approaches to multivalent and bispecific antibody fragments, Immunotechnology 3 (1997) 83–105. [9] A. Todorovska, R.C. Roovers, O. Dolezal, A.A. Kortt, H.R. Hoogenboom, P.J. Hudson, Design and application of diabodies, triabodies, and tetrabodies for cancer targeting, J. Immunol. Methods 248 (2001) 47–66. [10] P. Carter, Bispecific human IgG by design, J. Immunol. Methods 248 (2001) 7–15. [11] P. Holliger, T. Prospero, G. Winter, “Diabodies”: small bivalent and bispecific antibody fragments, Proc. Natl. Acad. Sci. USA 90 (1993) 6444–6448. [12] M. Whitlow, D. Filpula, M.L. Rollence, S.L. Feng, J.F. Wood, Multivalent Fvs: Characterization of single-chain Fv oligomers and preparation of a bispecific Fv, Protein Eng. 7 (1994) 1017– 1026. [13] Z. Zhu, G. Zapata, M.R. Shalaby, B. Snedecor, H. Chen, P. Carter, High level secretion of a humanized bispecific diabody from Escherichia coli, Bio/technology (NY) 14 (1996) 192–196. [14] D. Xiong, Y. Xu, H. Liu, H. Peng, X. Shao, Z. Lai, D. Fan, M. Yang, J. Han, Y. Xie, C. Yang, Z. Zhu, Efficient inhibition of human B-cell lymphoma xenografts with an anti-CD20  antiCD3 bispecific diabody, Cancer Lett. 177 (2002) 29–39. [15] Y. Gao, D. Xiong, M. Yang, H. Liu, H. Peng, X. Shao, Y. Xu, C. Xu, D. Fan, L. Qin, C. Yang, Z. Zhu, Efficient inhibition of multidrug-resistant human tumors with a recombinant bispecific anti-P-glycoprotein  anti-CD3 diabody, Leukemia 18 (2004) 513–520. [16] H.J. De Haard, N. va Neer, A. Reurs, S.E. Hufton, R.C. Roovers, P. Henderikx, A.P. de Bruine, J.-W. Arends, H.R. Hoogenboom, A large non-immunized human Fab fragment phage library that permit rapid isolation and kinetic analysis of high affinity antibodies, J. Biol. Chem. 274 (1999) 18218–18230. [17] D. Burtrum, Z. Zhu, D. Lu, D.M. Anderson, M. Prewett, D.S. Pereira, R. Bassi, R. Abdullah, A.T. Hooper, H. Koo, X. Jimenez, D. Johnson, R. Apblett, P. Kussie, P. Bohlen, L. Witte, D.J. Hicklin, D.L. Ludwig, A fully human monoclonal antibody to the insulin-like growth factor I receptor blocks ligand-dependent signaling and inhibits human tumor growth in vivo, Cancer Res. 63 (2003) 8912–8921. [18] Z. Zhu, P. Rockwell, D. Lu, H. Kotanides, B. Pytowski, D.J. Hicklin, P. Bohlen, L. Witte, Inhibition of vascular endothelial growth factor-induced receptor activation with anti-kinase insert domain-containing receptor single-chain antibodies from a phage display library, Cancer Res. 58 (1998) 3209–3214. [19] O. Perisic, P.A. Webb, P. Holliger, G. Winter, R.L. Williams, Crystal structure of a diabody, a bivalent antibody fragment, Structure 2 (1994) 1217–1226. [20] J.A. Carmichael, B.E. Power, T.P. Garrett, P.J. Yazaki, J.E. Shively, A.A. Raubischek, A.M. Wu, P.J. Hudson, The crystal structure of an anti-CEA scFv diabody assembled from T84.66 scFvs in V(L)-to-V(H) orientation: implications for diabody flexibility, J. Mol. Biol. 326 (2003) 341–351. [21] S.M. Kipriyanov, G. Moldenhauer, M. Braunagel, U. Reusch, B. Cochlovius, F. Le Gall, O.A. Kouprianova, C.W. Von der Lieth, M. Little, Effect of domain order on the activity of bacterially produced bispecific single-chain Fv antibodies, J. Mol. Biol. 330 (2003) 99–111. [22] D. Lu, J. Shen, M.D. Vil, H. Zhang, X. Jimenez, P. Bohlen, L. Witte, Z. Zhu, Tailoring in vitro selection for a picomolar affinity human antibody directed against vascular endothelial growth factor receptor 2 for enhanced neutralizing activity, J. Biol. Chem. 278 (2003) 43496–43507.