WO2001012809A2 - Heterodimeric vegf variants used for inhibiting angiogenesis - Google Patents

Abstract

The present invention discloses heterodimeric VEGF variants with functional VEGF-receptor binding sites at one pole of the asymmetrical dimer, and mutations in the domains for binding to the VEGF tyrosine kinase receptors KDR and Flt-1 at the other pole. These molecules are potent inhibitors of VEGF-induced proliferation and tissue factor induction in endothelial cells and of vascular hyperpermeability.

Description

Title: Methods and means for inhibiting angiogenesis .

The present invention relates to the field of medicine, in particular human medicine, more in particular to the treatment of tumors, in particular solid tumors. The present invention also relates to the field of human gene therapy, more in particular to gene delivery vehicles encoding VEGF antagonists in order to treat diseases correlated with angiogenesis and/or VEGF, such as, but not intended to be limited to cancer. The present invention also relates to the field of molecular biology, in particular genetic engineering. The techniques of genetic engineering are employed herein to produce e.g. said VEGF antagonists and gene delivery vehicles encoding them. Angiogenesis or neovascularization, the sprouting of new blood vessels from pre-existing ones, is essential to many pathophysiological conditions like wound healing, rheumatoid arthritis, diabetic retinopathy, psoriasis, and tumor growth and metastasis, while being restricted to the menstruation cycle in the normal adult (1) . Angiogenesis is a complex process, involving proteolysis of the blood vessel wall, deposition of matrix proteins outside the blood vessel, proliferation and subsequent extravasation of endothelial cells and their adherence to the newly formed matrix to form a new lumen (13-15) . The process of angiogenesis is governed by a number of angiogenic but also anti-angiogenic factors (14). Among the anti-angiogenic factors are platelet factor 4, thrombospondin, angiostatin, endostatin, the N-terminal fragment of prolac m and soluble receptors for VEGF (16-20) . The angiogenic factors include angiopoietin I, Platelet Derived Growth Factor (PDGF) A and B, Hepatocyte Growth Factor/Scatter factor, acidic and basic Fibroblast Growth Factor (a/bFGF) , Transforming Growth Factors and β, and, most importantly, VEGF, also known as Vascular Permeability Factor (VPF) (14,15,21) .

VEGF is a member of a superfamily of homodimeric growth factors, encompassing PDGF-A and B, Placenta derived Growth Factor (P1GF) , VEGF-B, VEGF-C, VEGF-D and VEGF-E (12,22-26). These factors share a common motif, the cystine knot, which consists of eight spatially conserved cysteines that are involved in intra- and intermolecular disulphide bonds. Two cysteines, the second and fourth in the knot, interact intermolecularly, leading to the antiparallel conformation of the dimer (27,28). By itself, VEGF can stimulate endothelial cells to display a number of biological responses such as proliferation, migration, vascular permeability, production of proteases and induction of tissue factor, a key factor in the blood coagulation cascade, thereby creating the prime conditions under which angiogenesis can occur (28,29) . We recently observed in an in vivo tumor model that vascular permeability is important for the morphology of the vascular bed and the vascular density, while induction of endothelial proliferation by VEGF determines the tumor growth rate. VEGF exerts its action on endothelial cells by activating two tyrosine kinase receptors Fms-like tyrosine kinase (Flt-1) or human kinase domain receptor (KDR, the human homologue of the mouse Flk) , while neuropilin has recently been identified as an isoform-specific co-receptor (12,25,30). Because of its symmetry, the VEGF dimer has two receptor binding interfaces each situated at a pole of the molecule (see also Figure 1) . Each of the two binding interfaces is capable of contacting one receptor monomer, thus dimerizing the receptor, which subsequently results in receptor activation and induction of the VEGF response.

Despite some confusion in the literature about the functional roles of the KDR/Flk and Flt-1 receptors, there is consensus about the importance of KDR/Flk. Knocking out this receptor leads to lethality in an early embryonic stage, associated with a lack of endothelial cell differentiation (31) . Blocking the KDR receptor by a dominant-negative strategy completely shuts down angiogenesis in a tumor xenograft model (11) . Flt-1 knockout mice also die during embryogenesis, but this defect involves an inability to form tube-like structures (32) . Monocytes, expressing only Flt-1, can be triggered by VEGF to migrate and to produce tissue factor (33,34). Furthermore P1GF, which is a ligand for Flt-1 only, induces tissue factor expression in monocytes and in endothelial cells (34) . These findings suggest a role for Flt-1 in tissue factor expression and migration. Therefore it is probably of importance to prevent both KDR and Flt-1 receptors from activation by VEGF. However, VEGF-E, a parapox-virus encoded protein that is a ligand for KDR but not for Flt-1, is able to induce the whole frame of events required for angiogenesis, including tissue factor induction and migration, thus pinpointing at a role for KDR in these activities (23) . VEGF is not only very important as a mediator of angiogenesis, it has also become increasingly clear that this factor generates biological responses on other than endothelial cells. One of these functions involves a mechanism, by which tumor cells can escape the immune response of the patient. It has been shown that VEGF inhibits the development of immature dendritic cells (DCs) . Dendritic cells are major players in the immune anti-tumor response, as they are antigen-presenting cells. Ultimately, presentation of tumor antigens on DCs will lead to clearance of tumor cells by the host immune system. It has been convincingly demostrated that VEGF, by binding to the Flt-1 receptor on DC progenitor cells, inhibits antigen presentation and DC maturation. This inhibition is mediated by inhibition of nuclear factor κ_B (NF-κ_B) . Thus, by producing VEGF a tumor provides itself with a mechanism to escape the immune response.

Since our HD-VEGF is an antagonist, not only for KDR but also for Flt-1, this molecule is predicted to have multiple beneficial activities. It will not only inhibit angiogenesis by blockade of KDR, but will also suppress the immune suppressive effect which is mediated by the Flt-1 receptor.

A variety of human disorders and diseases are associated with aberrant endothelial proliferation and are associated with angiogenesis, or are at least associated with undesirable vascularisation and/or undesirable vascular permeability, thereby implicating a role for VEGF. These disorders, including tumor growth and in particular tumor metastasis (40) , diabetic retmopathy (41), retrolental fibroplasia, atherosclerosis, psoriasis (42), neovascular glaucoma, age-related macular degeneration, hemangiomas, thyroid hyperplasias, immune rejections of transplanted tissue such as corneal tissue, edema associated with brain tumors, Meigs ' syndrome, lung inflammation, nephrotic syndrome, pleural effusion, ascites associated with malignancies, peπcardial effusion asssociated with pericarditis, chronic inflammation and rheumatoid arthritis (43 are often accompanied by an increased expression of VEGF. Thus, the development of molecules that reduce or substantially inhibit the endothelial cell proliferating activity of the endogenous VEGF molecule has great potential for the therapeutic treatment of these aforementioned diseases. Especially inhibition of VEGF- induced vascular permeability, which might be an important determinant for metastasis, is likely to be very beneficial in a number of diseases, for As an example, the pivotal role of angiogenesis m tumor biology is widely recognized now. A growing tumor depends on a vascular bed not only for supply of oxygen and nutrients, but also for its capacity to metastasize (2) . Indeed, microvessel density is positively correlated with bad prognosis m a still increasing number of solid tumor types (3,4) and, conversely, inhibition of angiogenesis m a number of animal tumor models leads to inhibition of tumor growth or even regression of tumors (5-12) . Anti-angiogenic therapies are preferred above radiotherapy or cytostatic/cytotoxic therapies, because they are directed against vascular endothelium, consisting of normal cells that, unlike tumor cells, are not prone to mutations. This minimizes the chance of developing resistance, as compared to the chance of tumor cells becoming refractory to radiotherapy or cytostatic compounds. Since angiogenesis is rare in healthy adult tissues, anti -angiogenic therapies might be considered as tumor-specific, with few, if any, possible side-effects at distant sites m the body. A number of approaches that target the different stages in the angiogenic process have been proven to be successful m animal tumor models (reviewed in ref . 12) . An approach involving the production of molecules (VEGF variants) reducing the activation of VEGF receptors has been disclosed in W098/16551, incorporated herein by reference. The variants disclosed are mutants of VEGF in which the capability of dimerisation of VEGF subunits has been functionally deleted, thereby making a monomeric VEGF that can only bind one receptor subunit . Another approach involving the production of VEGF variants is disclosed in Siemeister et al . (38) incorporated herein by reference, in which variants are disclosed which are heterodimers having one functional receptor binding interface and one dysfunctional receptor binding interface. The dysfunctional interface is created through mutations in loop III of one subunit and loop II of the other subunit. However, the monomeric VEGF variants disclosed in W098/16551 will have strongly reduced affinity for the respective receptors than the wild type molecule since it is the interface between the subunits in a dimer that forms a contiguous receptor binding site. Since the application of these variants will typically involve a competition with wild type VEGF, chances are that an enormous excess of these variants is needed for successful reduction of receptor dimerisation and activation. The molecules of Siemeister have one normal functional interface and therefore have normal binding affinity at one pole of the dimer. The other pole however is still partially functional. As we show in here loop III mutation is not sufficient to substantially reduce receptor binding and activation.

The present invention is therefore based on, among other things, the finding that mutations in loop I of molecules of the VEGF family have a far more profound effect in receptor binding of any VEGF-like molecules than mutations in loop III, which are typically found in the same receptor binding interface.

The present invention thus provides an isolated and/or recombinant nucleic acid which encodes a VEGF antagonist derived from VEGF, wherein said nucleic acid comprises at least one mutation which alters the receptor binding affinity for a VEGF receptor of the resulting (poly) peptide in a sequence which m VEGF encodes loop I, at approximately ammo acids 30- 50. An antagonist according to the present invention is a VEGF derived molecule having a mutation in its loop I like sequence which enables it to significantly reduce the activation of VEGF receptors m the presence of the wildtype dimer. The finding that mutations m sequences corresponding to loop I sequences in VEGF have a significant effect on receptor binding leads to at least two different approaches in preparing VEGF antagonists. First, one can enhance binding to a receptor monomer of the loop I-like binding site by making a mutation m the loop-I like sequence which enhances the binding affinity of such a sequence to a VEGF receptor (typically Flt-1 and/or KDR) . By enhancing the binding affinity one can create molecules that have a higher binding affinity than wild type VEGF and which can thereby really compete with wildtype VEGF m situ. Such molecules can be monomers (including for instance a functional deletion of the dimerisation site of VEGF subunits) . Such molecules may of course also be dimers m which the one receptor binding interface is enhanced in its affinity for a VEGF receptor and the other is dysfunctional in receptor binding. It must be borne in mind that natural VEGF is a homodimer having two receptor binding interfaces each comprising a loop I and loop III from one subunit and a loop II from the other subunit . One may thus enhance the binding m one interface by making a mutation m a loop I like sequence which enhances binding affinity and functionally delete the binding site of loop II on the same subunit and/or loop I and/or III at the other subunit . The second approach is then immediately clear, because one can also create a dysfunctional receptor binding interface in a dimenc VEGF like molecule by functionally deleting the loop I binding site m one subunit. The other interface may then be a wild type interface or an improved one as disclosed herein before. An isolated and/or recombinant nucleic acid molecule according to the invention is any such a nucleic acid molecule which encodes at least a VEGF loop I like sequence, wherein a loop I like sequence is a sequence based on VEGF loop I, located approximately at amino LO O M t t-¹ H

LΠ o LΠ o LΠ O LΠ en cα 3 CQ H- £ H- ø⁾ H tr 0 Φ ω ti φ a 0 3 P. 3 CQ 0J Φ 3 φ Φ ØJ m n M £ ØJ

C Φ e Φ en Φ cn cr Φ H- φ Hi a Φ 0 a H- H- Hi 0 H- ØJ Φ a a ØJ a Hi HI tr Φ "^ tr n tr a rr Λ 0 n a a tr n 0 tr cn g 3 ?r Λ tr - tr Hi HI H- n Φ H- en £ ju ø⁾ h- ' o c Φ α rt ØJ ^■i Φ TJ ØJ o Φ tr Φ φ Φ d H- CO ØJ ØJ φ H- en Φ £ ti rt φ rt Φ n h-¹ rt τj H- Φ tr a tα Φ TJ a t-¹ ti H- O ti CQ Φ rt c a rt a n a TJ tr Φ

ØJ a H- a rt 0J Φ rt a φ o 0 P rt o 0 H- a c P- a H- o n tr n rt H- 3 rr H- H- H ti n 0 o. £ en ø⁾ 3 0 LQ Φ Φ £ o 0 φ CQ CQ ω H- n 0 tr H- Φ φ cn rt E- 0 o a Φ

ΓT Φ a Φ ØJ H 0⁾ n H- α, Φ Φ ti H-¹ 0- Φ 0J H- Φ P- rt Φ a a ti P- ^ rt ti tr O

H- CD t-¹ rt ?r H- a <! en H- cn rt a a cn ØJ ØJ LQ φ ØJ øJ ØJ H-

^Q) l—¹ tr ^ H- 0J a <; tr φ tr ^ tr H- LQ to LQ 0J Hi tr H 0 rt o 3

Φ ^ φ a o rt ^» : Φ H- ti ØJ H- Φ H- ØJ O tr - ØJ 3 Φ 0 H- φ ii H- O c: 3 c

I-¹ P en o H- LQ rt φ CO a a H a a a a < o c ti CO 0 Hi ii a 0 rt Φ

^ o tQ 3 a a H- Hi rt o. φ en M c CQ φ Φ Hi Φ 0, rt a TJ ØJ rt CO o H- ØJ c rt 3 Hi H- H- ø H- φ H- H- o Ω Φ φ iQ Φ Φ H- H- O 0J rt øJ h-¹ H- O P- n Φ C ø⁾ 0⁾ a O a P ØJ a Λ a en Hi ^ CL rt Hi H C n O O a cn 3 ϋ H- rt LO

0 H- co ^Q) Φ a rt rt n o a LQ rt tr LQ C LQ 0 c φ Φ P rt 0 LQ n ø⁾ ØJ Φ 0 H- O ra P n Φ ø⁾ H- Φ rt øJ en Φ cn ti TJ øJ Λ a 3 a ^•< &, a 0 1 rt LQ H- H- rt O H- s: en LQ φ en a rt 0J ØJ Φ ti t-¹ c- o Φ øJ TJ CO o a LΠ en ^ a O o. H- a 0 0 tr H- 0 a H- n O rt H- n Φ 0 0J H- φ CO ϋ tr CO Φ rt 0J o

ΓT J en O a a Φ rt a rt rt Φ Φ Φ Hi a a ti cn a" H- 0 Φ a φ 0 r-⁴ H- ^ tr φ H rt 0J rt a ø^j a φ H- Φ øJ P, TJ Φ Φ rt φ ØJ cn c tr ØJ rt CQ

Φ C-. Φ P^J O O en H CO en 0 en H- H rt ti øJ φ <! o φ 0J Ch rt Φ 3 o co Φ H- a 0⁾ rt ØJ rt H - P < ØJ φ 0 ØJ 0J LQ a H- a tr a tr ti TJ 0 tr 0 W- o. H H- H- en tr a H- M a n ti tr o 0 n a LQ 3 Φ n rt Φ CQ ii ri

H- TJ a a Φ P rt 0 Φ n Hi H- s; Ω Φ CQ rt a O 0 LQ 0 ϋ H Φ O Φ Φ

P rt £ ø⁾ tr c a tr >n TJ ^• H- H- ¹- ØJ H Φ Φ 3 ≤ rt CO α rt H- n Φ sα Φ α rt t-^> a < a H- o rt H- <J CQ a H- Φ en Φ rt H- Φ cn H- tr φ TJ

W- ^' a t-¹ <{ o H- tr o c n TJ O ti H o a a ø rt n a a ØJ a a £ φ a 0J

P r-⁴ Φ CQ Φ Φ 0 0 Ό g PJ 0 O Ω rt ØJ tr φ ii 0 rt • φ LQ tr H- 3 H- • rt tr P-, rt ii to H- H- h-¹ a TJ H Φ rt TJ ^ H- t o i CQ 0 rt H- 0J rr C a CQ tr rt

? h-¹ H- o 1 φ h Φ 0 rt en φ φ • ti a H H- Hi a CO rt H 0 rt rt H- O H-

0J Φ 0J a tr H en H- H I-¹- a H- φ TJ O tr LQ rt rt c n ØJ P- tr Hi øJ a 0 n

Hi en ø⁾ £ 0 Φ H a n H- ØJ O Λ rt O H Φ ØJ CO ii Φ cn rt rt Φ a τ 3 C

Hi en rt øJ o tr rt a Φ 0 3 C C O tr ti <; < 0 rt α rt H- CO rt rt Φ H- TJ

H- Φ a H- Φ tr H- rt 3 φ a ø⁾ φ h-¹ φ H c? C Φ M Hi Φ tr ii O tr H- a ØJ 0J

P Λ o. α H fT ø⁾ Hi n ti ^ 0J a H- ra rt Ω a φ tr P o Φ ØJ \-> LQ H ϋ

H- C- 0 0J Φ rt Φ ø⁾ H- Hi rt H- en ti n rt a 0 <τl n tr H H- n cn 0 0 h- ^■ 0 Φ rt φ 3 φ tr 3- •0 a Φ tr Φ tr ØJ rt H- 0 0J a rt 0 TJ 0J 0 ØJ o- ^

^■ ; a Ό H- P ^• Φ 0 •0 en o Φ en H- Φ en ti a tr øJ Hi ti £ a C rt TJ 1 HI TJ HI n a n H t-^J rt a t-¹ O 0 rt φ <! Hi φ ϋ n ^ H- ti tr H HI 0 HI rt 0J

Hi Φ Φ 0 0 rt ø⁾ H- 0 φ n Φ 0 φ C 0 Φ n CQ Φ Ω a Φ øJ H H- O H H- 0 rt

0 ØJ P- 3 3 0 o rt 0⁾ rt a rt TJ a LQ a H- H- ti Φ φ α ^ LQ a a CO a ti s; 0 o H- H- rt 0⁾ t tr n C Φ n tr c a α Φ TJ ti H¹ H- H- tr H- ≤ ti

P^{^} Hi H- a H- a ¹ rt Φ CO Φ 0 H 0 o <! a rt P^J tr α 0J Φ a H- ? rt 0 H- rt H- Φ

H- α to O 0 Φ h¹- ti H- ti H- • H 0J t-¹ Φ TJ rt 0 0J H- Φ HI 3 φ Φ ^ Hi a •<: h^J CO

O 0⁾ w 0 Φ a H- a H- Φ a 0 ti CO a li HI 0J n Φ α H-

Ω t H- en rt 0J a tr LQ en a LQ Φ a 3 H- rt CQ D-. cn H- H- H- Φ 3 rt H- HI 1 P-

^ P φ tr h- ' en ^• C LQ a LQ F- n H- ω 0 • a H- < a a cn cn 0 ω tr a 0 rt (.;

P^J Φ Λ Φ ^• ; H- Φ £ φ tr ø⁾ n rt o H- 3 0 a φ H- cn CO rt h-¹ 0 Φ LQ ti en C Q, Λ ø⁾ rt 0 0 0J ØJ a tr ØJ H LQ rt ØJ H Φ rt TJ Φ

H 0⁾ CO

Φ en Φ ø⁾ rt £ Φ <! Hi a tr rt rt o H- a n ^• H C o tr tr 0J ØJ φ n 3 a 3 tr h-¹ Φ a 0 c H- Φ 0 H- tr H- H- 0J Φ a 0J ØJ 0J 0 ^• Ω n H- HI LO

Φ C n H- Φ φ P H- a rt 3 0 a φ O h-¹ CQ Hi TJ 3 a o rt TJ a HI <! cn

TJ O Φ a < n rt o. Ω TJ LQ ti a CQ 0J 0J ii H- o TJ Hi C Φ o H- M W 1 rr 0 3 φ φ ø⁾ H- rt tr ØJ 1 tr 0 Ω n tr φ a 0 ϋ CO H- a Ω Ω ^

0 Φ P^J ø⁾ C en a H- ø^j M H < H- ØJ 0⁾ rt H- ØJ o M H Φ tr CQ ^<τ\ ^ c

H- a H- n

H 0J n rt ? LQ O en Φ a <! TJ t-¹ . 0J H- LQ rt en a <! H- ø⁾ 0⁾ H- a Ω H- 0J cn H- ^• øJ rt tr

H- rt ^ o, en rt a ⁾ tr T >τl ØJ a tr rt π H- tr H H- O

01 0J P H- a- LQ i— ^■ Φ ØJ LQ n H- ?r φ rt o a Φ 0

H tQ 0 H- Φ ti Φ tr fi Φ LQ TJ

Φ H- a en a rt 0

0J 0J Hi

LO LO t t H

LΠ o LΠ o LΠ O π

rt H- 3 TJ TJ H h-¹ ra H- H- Φ Hi tr a 0 0 Φ a 0 φ a ra ra 0

Φ CO Ω, μ-¹ TJ Ω Λ <! 0 CO ii

3 φ H- rt

TJ 0 0⁾ £ φ φ ti Hi H- H rt φ a 0J a

TJ rt H- φ α Φ a rt rt rt ii Ω TJ Φ Φ Ω H- Φ H- £ Ω

Φ rt 0⁾ rt cn n φ 0 C 0J ^

CO tr rt H- - ØJ a \->

Φ Φ H- P, rt rt φ øJ H a ra 0 φ TJ 0 a Hi a ^ Φ rt Φ a ra 0 I-¹ Ω C Ω

CO TJ 0 0 ϋ 0J Φ øJ a tr T H. 0 Ch rt 0 tr TJ

CO ØJ 0 TJ H- tr H o rt

Ω 0⁾ <! φ α P Φ 0

Φ Ω H- TJ c H LQ H H H- ϋ

X Φ Ω a rt Ω H Φ cn cn

TJ μ- 0 LQ H- φ M rt Ω Ω tr

H Ω ϋ α • tr o O φ H-

Φ P- o Φ rt Φ 3 en

CQ 0J H- rt en tr TJ Fj^* ra Ω a tr Φ ii ii H- cn

H- H- LQ φ ØJ φ H- a H H- o ti a TJ Ω C0 0J tr LQ a ra rt α, H φ H- a C a

0 TJ 0 TJ a rt cn H-

< H- 0 LQ rt rt LQ Hi φ a rt CQ Φ 0 a rt H-

Ω rt tr rt ^ H- H ØJ c tr Ω rt 0 Φ 1 Ω a Ω φ 0J o O 0J tr a

H Φ H- O cn Ω H- C Φ H- rt ra i. g ti ^• Φ a a H-

• a

TJ j 0 P. Ω Ω <! C ti Φ TJ 0J c H- rt Φ ^

H φ a φ rt CQ a H- 0J rt a tr cn ii Φ LQ 0 Ω rt ii φ cn H- H- 3 a H- H- φ

CO H- 0 1 o CQ ØJ o. O φ 0 a rt TJ H- a c a H 0 Φ rt 0J Ω

< μ- 0⁾ Ω φ α Ω 0J Φ φ <! rt a c Φ Ω Q,

Ω φ CQ τ I—¹ s: 0 cn - rt Ω H- Φ φ tr Φ H 0

O rt ra 0J rt ra H- rt TJ li O rt H- Ω μ- H- TJ H cn ii p. H- _^—_^ tr 0 a ii Φ cn Φ 0 Φ H- a LQ 0 Hi

3 CO a CQ a H- < Φ

ØJ 0 H- 0J Ω a 0 rt H- ti

^• ii ti 0 Hi 0 P., ØJ

0J ti 0 < φ tr

Φ rt tr Oi M rt rt CO H

H- 0 r-⁴ H- Ω tr tr rt Φ a | φ Φ ØJ tr tr LQ φ a

0J rt H- ii <! o CO φ

encode one or two subunits of a VEGF antagonist. In the case of one the desire may be to produce a monomeric antagonist or to have the other subunit produced from a different vector to make a heterodimer. The vector having two subunits may encode both subunits of a heterodimer. The end result should be an antagonist having only one functional receptor binding site or interface. Thus the invention also provides an expression vector comprising at least one nucleic acid according to the invention. Preferably the binding site or interface has enhanced affinity for a VEGF receptor compared to wild type VEGF as explained herein before .

The antagonists according to the invention are preferably produced from a nucleic acid or an expression vector (wherein of course said nucleic acid may already be an expression vector) according to the invention, preferably in a host cell. Thus the invention also provides a cell comprising a nucleic acid or an expression vector according to the invention. Preferably said cell has the capability to produce a heterodimer having one functional and one dysfunctional receptor binding interface and thus has a nucleic acid or expression vector encoding a subunit having functional loop I and loop III like sequences, but dysfunctional loop II like sequences and a nucleic acid or expression vector encoding a subunit having dysfunctional loop I and loop III like sequences, but functional loop II like sequences, whereby the two nucleic acids or expression vectors may also be one and the same. Thus the invention also provides a cell comprising an expression vector encoding a VEGF analogue having a dysfunctional loop II and a functional loop I like sequence and an expression vector comprising a nucleic acid encoding a VEGF analogue having a functional loop II and a functional deletion of at least loop I, whereby dysfunctional loop I like sequences are preferably accompanied by dysfunctional loop III like sequences on the same subunit . The invention next to the nucleic acids, vectors and cells also provides the antagonists themselves. Thus the invention provides a proteinaceous VEGF antagonist which is a monomeric VEGF analogue comprising a funtional receptor binding sequence which in VEGF is located at loop I .

Furthermore the invention provides a VEGF antagonist as above, in which the binding affinity of loop I is enhanced in relation to wildtype VEGF through a mutation.

The invention also provides in yet another embodiment a heterodimeric proteinaceous VEGF antagonist comprising one functional receptor binding interface and one dysfunctional receptor binding interface, wherein said dysfunctional interface comprises a mutation in a sequence which in VEGF is located in loop I, preferably a heterodimeric VEGF antagonist as above in which said dysfunctional interface further comprises at least one mutation in a sequence which in VEGF is located in loop II or loop III. In another embodiment the invention provides a heterodimeric proteinaceous VEGF antagonist comprising one functional receptor binding interface and one dysfunctional receptor binding interface, wherein said dysfunctional interface comprises a mutation in a sequence which in VEGF is located in loop II and wherein the functional interface comprises a mutation enhancing receptor binding affinity in loop I. The invention further provides gene delivery vehicles encoding VEGF antagonists. Gene delivery vehicles are well known m the art and are the tool of gene therapy. They are defined as vehicles capable of delivering a nucleic acid or an expression vector of which such a nucleic acid can be a part to a target cell. Typically in this case such a target cell would be an endothelial cell. Suitable gene delivery vehicles are preferably of viral origin, more preferably of adenoviral origin. Suitable adenoviral gene delivery vehicles can be produced in packaging cell lines such as disclosed in ECACC deposit number 96022940 incorporated herein by reference. Thus the invention also provides a gene delivery vehicle comprising a nucleic acid or an expression vector according to the invention. The invention thus also provides a gene delivery vehicle encoding a VEGF antagonist according to the invention. In case a gene delivery vehicle encodes only one subunit of a heterodimeric antagonist a second gene delivery vehicle is needed encoding the other subunit. Thus the invention also provides a kit of parts for the inhibition of angiogenesis comprising at least two gene delivery vehicles each encoding a subunit of a heterodimeric VEGF antagonist according to the invention, preferably both vehicles are again of viral, preferably adenoviral, adeno-associated viral or retroviral origin. Of course methods for producing the VEGF antagonists according to the invention are also included. Thus in yet another embodiment the invention provides a method for producing a VEGF antagonist, comprising culturing a cell according to the invention under suitable conditions and harvesting the antagonist from the culture.

Also the invention provides a method for making gene delivery vehicles encoding VEGF antagonists. Typically these are made according to ECACC deposit number 96022940 incorporated herein by reference. Thus the invention provides a method for producing a gene delivery vehicle encoding a monomeric VEGF antagonist or at least one subunit of a heterodimeric VEGF antagonist , comprising inserting a nucleic ac d encoding said antagonist or said subunit in a defective adenoviral vector, transfecting said vector into a complementing cell, culturing said cell under suitable conditions and harvesting said gene delivery vehicle from the culture.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to the use of variant heterodimer VEGF molecules that bind to vegf receptors without activating them, thus preventing endogenous vegf from inducing the vegf response. The invention is particularly useful for the treatment of angiogenesis related diseases, such as, but not limited to, cancer. It has been documented that the exposed domains in PDGF are responsible for receptor binding. In PDGF three exposed domains termed loop 1, 2 and 3 can be recognized that are involved in receptor binding (27,35) . Alanine scanning mutagenesis in the corresponding domains in VEGF revealed that charged residues in two exposed domains of VEGF, in analogy to the PDGF situation called loop 2 and loop 3, are involved in binding to the receptors Flt-1 and KDR respectively (36,37). Mutations of different amino acid residues in loop 1 in VEGF to alanines did not abolish KDR or Flt-1 binding. In contrast, some of these mutations led to an even higher affinity for KDR or Flt-1. In the present invention we disclose changes in loop 1, loop 2 and loop 3 of VEGF with the corresponding domains of PDGF-B. Homodimers of the loop 1-mutated molecule were completely inactive in different in vivo and in vi tro assays, in contrast to the alanine-scanning data of Keyt et al . (36) . At higher concentrations, the loop 3 -mutated VEGF showed residual activity in proliferation assays and in the in vivo Miles vascular permeability assay. The loop 2 mutated-VEGF homodimer was incapable of inducing tissue factor expression and had little activity in the Miles assay.

DESCRIPTION OF THE INVENTION

The present invention discloses heterodimeric VEGF variants that retain at one pole the ability to bind both VEGF receptors KDR and Flt-1, but have lost the ability to bind these receptors at the other pole of the dimer, by swapping loop 1 in one and loop 2 in the other subunit . Thereby the ability to bind receptor monomers is retained while receptor dimerization is jeopardized. This protein potently antagonizes the effect of wtVEGF in proliferation assays, tissue factor induction assays and the Miles vascular permeability assay.

We disclose the identification of loop 1, ranging from aa 36 to 46, loop 2, encompassing aa 63 to 67, and loop 3, ranging from aa 84 to 90, as domains possibly involved in receptor binding, and important for biological activity in proliferation assays, tissue factor induction assays and permeability assays. This was facilitated by aligning the primary structures of VEGF and PDGF with the cysteines in the common cystine knot motif as a reference. The art suggest loop 3 as important for KDR binding and loop 2 for KDR and Flt-1 binding (36-38) . However, the importance of loop 1 as a receptor binding domain was not recognized in the art.

Loops 1 and 2 in different subunits of the VEGF dimer were swapped with the corresponding domains of PDGF-B to create a heterodimeric VEGF variant that only at one pole lost its capacity to bind both the KDR and the Flt-1 receptors (figure 1) . We used a loop 1 mutant and not a loop 3 mutant since we found in all our assays that substitution of loop 1 completely abolished VEGF activity while mutation of loop 3 yielded a molecule with residual activities at higher concentrations (see also figures 3-5) . The VEGF-LI/VEGF-L2 heterodimer was expressed in the baculovirus expression system.

Homodimerization of mutant subunits was prevented by mutating the second cysteine in the cystine knot in one chain and the LO LO t to H

LΠ o LΠ o LΠ o LΠ ^{^} < rt si < 3 C O ii rt TJ < < μ- Ω 3 ØJ ø

Φ M μ- φ M 0 C Φ tr ii t w rt 0 c Ω Ω rt Ω ra Ω a H cn CD μ- O Ω Ω CO <! rt rt rt

Φ >r| ra rt *τ| 0 TJ tr> cn * 'T) ØJ øJ μ- μ-

H 1 C tr - 3 ii to μ- 1 1 tr rt < <!

0 f Φ φ Φ μ- ^^ rt TJ Hi O O μ- Φ μ- μ- μ-

P- μ¹ a øJ t CO n ii φ to t 0 a o rt rt μ- Hi ra μ- *> μ- 0 H cn cn rt a " ^<

3 øJ 0⁾ rt rt a ω ra rt 0J l-¹ tr¹ 0 φ a Ω φ tr ØJ LQ ι-> Φ rt to to LQ tr μ- ØJ ii P- rt cn o H Ω, μ- μ- ^^ ^^ μ- μ- a rt

O rt ra φ

^ a 0 n n Ω a CO

£ ii Φ LQ 0 Ω TJ - a »J> ^ ØJ < tr

Φ - tr 0 μ- cn cn μ- M μ- ØJ

H Ω ØJ £ 3 Ω TJ tr¹ tc a Ω LQ \->

Φ -T) ØJ rt rt a Φ TJ rt u £ μ> H 0J LQ ^ -r ti

1 CO tr 0 rt Φ μ- φ Ω 1 Φ φ

Ω c Φ ø^j Hi φ Ω, 0 s: tr rt tr tr¹ μ ØJ

0 3 0J CQ 0 rt li TJ ØJ φ μ- Φ to φ ra 0 c Φ μ- H CQ rt < rt Ω

TJ tr ØJ ØJ 3 P a r-⁴ a rt Φ Φ μ- £ ØJ 0

0 cn 3 C μ- P. ; 0 1 0J ii rt Φ tr a Ω,

Φ 3 C φ Ω ISI Hi μ- tr 0 ^ Φ 0 Ω μ-

H 0 ti tr Φ rt μ- μ- 0J a a φ ra

Φ P. Φ 3 tr a LQ Ω Ω Φ μ- _^~. μ- a Ω μ- α C øJ tr C rt c 3 Ω, rt co rt C

^ 3 rt Φ rt rt μ- H μ- tr rt φ 0J tr tr ti CO φ r ø CO tr tr φ < 0J o H, rt Φ Φ ØJ CO c H. ^• P CQ Φ rt μ- ØJ rt 0J rt φ a CO rt Φ tr rt LO rt φ øJ ≤ < μ- Ω,

0J rt en H TJ Φ Φ a μ- &. a ØJ P M μ- 0 tr ØJ tr ti CL • 0 rt CO 0 Ω rt a a Φ Hi Φ 0 a w øJ rt •n ra ra < φ Ω, O X M £ > | LQ 0J 1 . M μ"

TJ H rt μ- Ω rt £ ^ 0 CQ XA ØJ Ω rt rt H Φ Hi >τ] to μ- M a ra tr to Ω H Ti

O tr O rt a φ 1 < o rt μ- o 0 rt a 1 φ Ω tr rt H π M 1 tr Ω N ≤ cn μ- rt Lr¹ μ- 0 Φ 0J to Ω Hi φ φ μ- a_^ C <! H LO a <: 0⁾ μ- rt rt n ^ 0 a tr μ- 0 w LQ H tr μ- tr¹ 1 O £ øJ • c rt Ω, cn c Ω C ØJ <! to μ- cn rt Ω a ^• C tr

Ω v ØJ a φ P rt > μ- Ω 0

1 a <

Φ ØJ tr øJ P- Φ LQ Hi td μ- CO rt rt £ n a μ- rt ØJ a X ^*\ o Ω <! cn μ- μ- Φ rt to rt tr Ω Ω Ω 3 ii •n φ φ a 0 μ- cn μ- Φ rt Φ Φ 1 • £ Ω, co a tr¹ øJ rt μ- < ra H¹ μ- TJ øJ μ- cn

CO t ra t < M ra £ o a H φ cn Ω 0 μ-

C ^_^ ra Φ μ- Ω TJ rt Ω, 0 Ω 0J Hi LQ

Φ n ØJ rt rt rt Ti ϋ 0 3 c rt Ω rt

*> a

^ 0 φ ^• 1 0 Hi < μ- Ω rt Φ μ- 0J μ-

Hi • ii n μ-¹ M a φ Φ C a Hi

ØJ tr¹ μ- o 0 »J μ- < Ω c en ^ Ω LQ n μ-

Ω H ω a Hi n Hi W ^ rt rt μ- ^•< Ω rt O Ω, μ- tr" φ Ω φ rt ø^j rt cn 0J

O rt c 3 £ - H n. ω £ tr t—¹ tr to ii tr Ω ω tr 1 a

0J H tr Φ 0J cn rt

Φ ii rt tr £ Φ Hi rt Φ μ- φ μ- rt o ti 0 rt tr rt a tr Φ tr μ, φ

LO O to to μ¹

Lπ σ LΠ o LΠ o LΠ rt μ- tr TJ CQ to Ω <! Ω TJ ØJ CO ØJ H tr tr 0 * ii 2, £ n CO »J> μ- Φ tr Hi rt TI — Ω rt

0 a μ- 0 Ω W " ii a μ- ϋ 0 μ- 0 a μ- Φ 0 øJ to TJ 1 a X μ- μ- tr μ- Hi 0 tr Hi a ra 0J 0J C Ω cn 0 rt rt Φ φ 0 P rt ra CO CO cn Φ Hi TJ CQ CQ φ a μ- 3 φ μ-

PJ t-¹ μ- a a cn TI rt X μ- TJ tr rt Ω TJ μ- tr¹ Ω O c φ tr C 0J CQ TJ LQ

O μ- rt a P, o Φ μ- TJ øJ μ- LO ra μ- tr r-⁴ 0 a 0 to μ- Ω Ω φ μ S C X

0 a μ- μ- 3 μ- 3 øJ rt a cn a < Φ 0 a Hi tr Hi Φ rt φ μ- »i Φ ϋ

Φ TJ LQ < a LO ØJ 0 a μ- ii Φ 1 LQ M CQ ra 0 CD ØJ μ- Φ ^ Φ rt % Φ

CO Φ LQ a e Φ rt øJ Ω TJ ^ Ω TJ Φ μ- ϋ φ a Ω φ Ω- Ω cn φ Φ w ra to rt μ- ra ^• tr ii T\ ii tr 3 μ X tr 0 ra ti> ^j n >J φ 0 ^• Φ a α 0J o Φ to φ Φ rt øJ <; Hi Ω 3 a s: ω ^■ cn øJ X Hi rt φ rt CQ μ^> CQ 0J to tr φ rt Φ < 0 φ rt Ω Φ — ^ - — '

TJ rt Ω TJ ii 0 φ μ 0 Φ ^ ØJ μ- a φ φ μ- M μ cn Φ tr φ rt ØJ ØJ φ TJ tr tr Φ φ 0 X 0 rt 00 a 0 a Hi 0 Ω cn ti Φ a tr ra rt tr

X Φ φ ØJ H Ω TJ rt 3 Ω tr TJ — a o CO o a TI rt TJ rt φ Ω Φ o o > rt øJ ti μ- φ Φ 0 0 φ ii CO J Φ rt μ tr O φ < μ> C CO P-

Φ li w LQ 3 TJ LO a rt 0 a 0 ØJ LO D ii μ- O Φ Hi ii M 0J < rt μ- 3 CX a Φ Q Φ Φ rt rt tr TJ Hi 0 X a ^ Ω 0 C i ω 3 Ω rt M 0J Φ ra μ- μ- rt Pd α ti 0 Hi Φ 0 TJ μ- H T\ Hi Ω Φ cn tr rt Φ TI μ- Ω ii tr a rt

- rt ti μ. Hi t ii TJ 3 c a - rt Ω øJ tr¹ φ tr ØJ 1 0 TI Φ c μ- rt ii ra CO 0 μ Ω 3 0 μ- rt φ rt C- Φ < H rt μ- tr Ω a 1 tr rt rt 0 rt 0 Φ C - tr 3 0 ^•< 0 ØJ CQ rt tr ii J £ tr H TJ øJ Φ o μ- to CO tr¹ α Φ Φ øJ r 3 ii n ¹ 3 tr Φ a

0 Ω ii μ- 3 ra Hi rt μ- Φ φ t μ- Φ £ cn tr Φ Hi £ a tr* rt μ- rt ^ Ω 0 <-• 0 0 TJ μ- tr" < φ rt o s Φ TJ ØJ tr ^•< T) H- rt 0 Φ tr μ- TI TJ tr H ØJ a P, ii rt to M tr ii ØJ TJ Hi rt Ω μ- μ- to tr a tr a £ ØJ μ μ- tr tr 0 μ- O ^ \_^ Ω 0 TJ Ω < ii ttf ii O φ Ω a rjo Φ Φ Φ Φ tr CO Φ Φ CQ 0 tr 3 3 rt Ω J 3 Φ rt M μ- 0J

Φ ii tr LQ Φ. LO o a μ- μ- a rt tr 3 μ- φ Φ φ Φ Ω O ii μ- Ω o

,— _^ CO μ- cn o Hi <-• Ω rt tr H- o a ii μ μ- O cn tr¹ 3 TI ii ^

LO TJ P μ^j £ rt a rt Φ ØJ μ- ϋ Ω ^•< 0 CO a c to μ- φ μ- n 0 £ Φ ØJ 0 rt 0 tr rt o Hi Φ rt 0 μ- a • 3 ØJ rt μ- rt Hi

- a tr ra 0 rt Φ tr Hi r-¹ H- Φ Φ Ω CQ a £ cn £ φ tr μ- a 0 0

LO CO μ- 0 X - ϋ Φ o ^•< Φ P- 0 LQ rt øJ μ- H £ ØJ ti μ- φ α μ^J

^J μ- o rt μ- H rt 0 α Φ ti . tr ra a tr tr ØJ CQ CQ c øJ P-

~- — tr φ TJ o ØJ ra £ cn tr tr TJ Ω X rt CO tr Φ Ω μ- Φ ra tr μ- Ω o

• ii vo Φ Φ 0 Φ tr _^-_^ TJ tr φ Φ TJ φ CQ CQ ØJ rt 0 Φ rt Φ cn

Φ P LO rt LQ o ^ tr 3 to ØJ to o Φ ii rt a μ. ii ra μ- ^• Hi μ- X

M φ tr φ cn TJ £ 0 3 ii -J CO <! £ Φ φ £ μ- TJ a μ- <! Ω

CQ Hi LQ £ Φ ti a 0 . — LQ ^ φ TJ φ Φ rt LQ tr a Φ 0J CQ a ØJ rt rt 01 Φ

TJ 0 ØJ tr ^• H Φ μ- μ- > φ LO P, 0 Φ tr μ- μ- Φ tr Ω Ω tr 0 cn tr μ- rt CO

Φ μ rt μ- μ- tr P, rt 3 Φ CQ α LΠ CO a Φ 0 a a μ- μ- rt rt cn φ cn μ- CO

Ω H- Ω C Ω Φ 0 μ- • φ Ω a ^ — P- μ- Ω a ι_ι. tr Hi μ- 0J ØJ cn o μ- tr < tr Φ O 0 Ω ii c μ> 0 rt ^>< < TJ cn Φ ØJ μ- μ- <! ^• Ω ^• C a 0 ø^j μ- Φ a a μ- TJ rt ω - cn 0 3 μ- ra M μ. Ω rt Ω φ rt μ- Φ Hi a μ- rt Hi P Φ ^ φ LΠ 0 tr¹ 0J 0 rt Ω Φ μ- rt 3 0 ØJ ØJ μ- Hi £

^• CQ μ- μ- <! to a. . TJ 0 μ- P μ- T) CD a Φ μ- ϋ ØJ Ω <: Hi Hi μ- rt

^• μ- Hi ti 0 øJ 0 Ω 0 a a Φ X ; ra rt μ- ra φ ØJ rt tr a Ω φ μ- 3 Hi rt 0 O μ-¹ LO TJ cn 0 Φ øJ t LQ a < rt ^ μ- rt μ Ω tr φ r tr CO Ω φ < ti tr μ- TJ c a - Hi a rt a C Φ £ < φ rt tr ØJ CO 0J Φ 0 ØJ LQ CQ φ H ,-_^ Ω a ii Hi μ- Φ φ ^ Φ a 0 £ tr

Φ rt μ φ rt 3 3 rt a 1 Ω rt P μ- TI 0 Φ Hi a . 3 rt μ rt Φ

0 LQ a μ- tr φ μ- rt σ . _^ Φ tr 0 TJ a Φ tr £ 0 rt rt Φ rt 0 ^> a a ^ 0 - — f- ii O Φ rt α <! rt μ- 0 Ω μ- . > tr a TJ Φ < Φ ϋ TI μ- a rt CO 0 LQ LO μ^> øJ μ- Ω φ tr a Hi rt tr CQ μ- CQ ii X a H μ- ØJ ØJ TJ 0 rt μ- 00 a CQ H 3 TI a ØJ tr μ- s: 0J rt O TJ Ω o

TJ rt ti o t-¹ o cn TJ rt 0J tr ra -J LQ a Φ 0 rt rt μ- r Φ μ- ts o 0J μ- n £ rt rt μ- φ ϋ rt ii TI P-

Φ Hi øJ Hi LO tr ii φ μ- ØJ o rt μ- tr øJ Φ a øJ Φ Φ μ-

Φ to Hi a μ* Φ Φ >τ) P t Ω CQ < 0J rt μ- CO 3 μ- o 0J μ- rt ti LQ φ TJ Hi 0J a μ rt T! W ØJ rt Φ a cn Φ

0J , μ a o o H- μ- O rt CO Φ μ- 1 tr Ω ra ra μ- ii a c Φ o P P μ> Hi o μ- φ £ o ^• TI O

O* φ TJ l£> ii ϋ a Φ a μ- μ- a rt

CQ ra LO O ØJ a a 0

3

co O to to μ^>

LΠ o cn o LΠ o LΠ

s to

0⁾

P

Ω,

TI rt

1 μ> øJ ii

Φ μ a

TJ ii

0

LQ li

Φ ra ra

•

to to

LΠ o LΠ o Lπ LΠ

ØJ Ω Hi ØJ øJ μ- c 0 £ Φ £ ii £-5 0J £ rt σ rt cn CO ra TJ TJ Φ tr TJ rt rt £ o 3 ØJ tύ

ØJ 0 a Ω a o TJ Hi 0 a tr LΠ Φ Φ CQ Φ tr O μ- μ- C ii tr 0 rt ii tr 0 ØJ a C- Ω Φ rt TJ μ P. rt CO £ ii C o Φ _^ TJ f- CQ Φ o tr 0 0 0 τ 0 μ- cn Φ rt rt Ω tr øJ μ- rt a φ y-⁴ ii a 0 rt 0J Hi Ω 0 φ φ ra cn TJ øJ Φ CQ ØJ øJ μ- 0⁾ o tr J rt <! Φ LQ M Q< 3 ti ii ^• 0 c £ • rt μ- TJ t ii μ- TJ μ- a <: 0 c tr øJ ØJ 0 C- Ω 0 0J — rt øJ Ui c CD ti 0 μ- Hi tr ^^ • TJ Hi tr tr ii a rt μ- Ul

LQ φ Ω Φ rt tr TI tr c • φ a Φ TJ μ. Hi rt φ 0 ^•—• Φ φ Φ ii Φ rt rt Φ tr TJ Φ CD ØJ Φ l H» Ω, μ- rt ii 3 tr C μ. ϋ o _^ ϋ μ. rt 0 0J £ μ- o TJ o Φ rt μ X N tr o o £ μ- rt rt ØJ o 3 ØJ Φ 0J μ- Ω μ- φ CO rt øJ 0J Hi ii Φ rt 3 ØJ cn 0 rt μ ØJ μ Hi rt CO tr μ- rt — ' TJ φ rt ϋ a, Ω rt^¬ rt CO £ tr a ii tr < CO 0 3 ^• £ o a rt Φ μ- øJ φ 0 μ- ØJ — ØJ μ- 0 rt tr øJ

Φ μ- 0J rt øJ Φ ϋ 0 Ul Φ LQ CO 0J Hi a 3 P 0 rt • tr o ^• Φ ii rt TJ μ- a TJ tr rt 0J Φ c rt <! tr £ μ- φ rt 0 P- a μ- 3 μ- a μ- P- Φ ØJ tr TJ ii P, < 0 < φ μ- rt Hi a Φ φ ØJ ØJ Φ 3 μ. μ- o 0 c H 3 0J Ω ØJ μ- c ra X ra 0 _^ øJ rt 3 ϋ <J a c C ØJ cn Φ 3 Hi P rt tr μ- φ a rt Φ < a a

TJ Ω Ω Ω a_ CO P* ra μ- Φ rt TJ 0J rt rt μ- Ω φ H 0J Φ rt rt ii rt 0 TJ ra LQ tr μ- TI <! TI μ- μ- μ- < Ω μ- Φ øJ μ- a c μ tr ØJ a CO ^•< μ- μ- ØJ rt Ω CO

^• P μ- • LQ rt a øJ 0 ØJ ØJ tr ii rt o Ω - o Φ a rt Φ CQ Ω LQ tr O TJ £ ra LQ ω ii ØJ Φ a Hi < a a 0 Hi ØJ Φ a Φ Φ o •- CD 0 μ- ii 0J 0 μ C w a > CO TJ rt M rt 0 Ω P- øJ TJ 3 ØJ c 3 a a TJ 0

0 rt . — .. CQ a ØJ 0J TJ Ω ØJ μ- μ- ii rt ØJ tr a 0 < Φ f- φ o μ- tr 0 TJ TJ tr to t 0 0 ϋ LQ μ TJ a φ 3 μ- -¹ ≤ ØJ o ra X rt Ul ra μ- 0 μ- o Φ to Φ μ- rt Hi rt Φ 0 rt øJ CQ Φ <! 0 Φ H P. Φ Ω TJ tr μ- φ rt a a TJ a μ>

LQ a μ- CO tr μ- 0 rt μ- μ- a μ- 0 o ØJ £ o TJ ii 0 3 0J o μ Φ CL LQ μ- £ o ra φ øJ Ω c Φ en LQ ØJ rt TJ ØJ o H- μ- c Φ ϋ μ- Ω C Ω μ- μ- to £

Ω tr to o rt H Ω 0 ui μ- Ω μ LQ ^• rt a CO rt μ CO ra rt rt a øJ

0J 0 ØJ tr rt a a 0J o 0J rt^¬ Hi Hi μ- tr φ a CD ØJ 0 φ tr LQ 3 o CO μ- ^■ Φ ØJ ØJ μ- ØJ rt - φ a LQ μ O ii O Φ 3 Φ CO ii μ M φ 0

Φ t CO rt <! ii 0 C Ω Φ a Hi μ- μ- μ. μ- rt rt t; a- £ Ω ii μ- rt TJ 3 ϋ cn 0 < Φ Φ rt φ a ra a tr P- μ- tr tr Ω 0 CO μ- ø⁾ TI a øJ 0

Φ ii — < w ØJ ra rt cn li ii μ- rt rt CD c 0 ØJ Φ φ Φ μ- £ rt a CQ rt rt a μ.

ØJ ra 0⁾ o a øJ μ- øJ TJ ø Ω tr rt rt φ ii ϋ a φ C < 0 φ rt - Φ

Φ a Ω to rt a a ϋ φ rt < φ Φ Φ X TJ a. c 0J ØJ ϋ

< LQ cn TJ ii ø⁾ LQ Ω LQ Ω μ- tr ØJ ra ØJ Ω øJ ^•< Ω Φ ra Ω tr ii w Hi rt £ P-

ØJ Φ 0 φ ØJ LQ μ- φ Φ μ- N Φ a Ω rt Φ a Ω ϋ a . rt rt μ- U ØJ 0 Φ Φ

P 3 ra Ω a Φ 0 rt ØJ φ ϋ TJ 0 tr cn rt 3 TJ Φ o tr £ μ- Φ 0J SJ Ω

Ω 0 φ φ LQ rt øJ o o μ- ra øJ C 0J o 0 ØJ ØJ o a φ Ω Ω Φ φ Hi TJ 0 φ tr rt ^ TJ ØJ tr TJ Hi CQ LQ rt rt rt rt a rt rt P ii o rt

0 £ rt TJ a φ tr ^• a ØJ £ 0 ø^j Φ tr μ- O O ØJ Φ Φ μ ØJ Hi tr 0 Φ φ rt rt <! 0J øJ a rt φ - tr TJ 0J ii a 0J tr μ. ra a μ- ii rt rt ra < μ- tr Ω μ- φ μ- CO μ- Φ 0 Φ C a μ- H rt μ- μ-

LQ rt ØJ μ- ra < a Φ 0 3 3 Hi CQ 0 Ω μ- μ- rt tr a a rt Φ 3 0 a μ- tr φ Hi H μ ra Ω ra ii c rt σ C rt TJ 0 ØJ tr Φ t-⁴ c Hi a a Φ tr 0 O Φ

0 0 Φ 0 LQ TJ Ω f- Φ rt rt Φ Φ

LQ ra a H- ii μ- <j μ- Φ a øJ Ul rt TI ti 0J Φ TI C Hi P- 3 Φ øJ øJ μ. Φ ØJ O rt tr 0 ra 0J tr Φ TJ Φ φ Φ ii rt φ ii 3 0 0 P rt rt φ Ω 0 ø^j o μ- P, tr CQ μ- H¹ tr ØJ Hi φ a rt < Φ μ- • φ ii ØJ ii 0 ii Φ Φ a Φ rt TJ cn H- φ tr μ- Φ Ω rt rt μ- Φ 0

X μ- 3 M a LQ Ω φ a φ a ii P, Ω P- μ- Φ a μ- 3 f- a rt < μ

ØJ Ω 0 μ> Ω a LQ Φ Ω CO ^•—• TJ Φ φ <J to Λ Φ rt φ £ £ rt φ ra

Ω TJ o TJ φ ^•< ii TJ < rt μ- C Φ ii tr 0J rt Φ - φ ϋ Ω tr rt φ Φ μ- rt rt μ- rt TJ ii 0 Φ M 0 μ- O rt O Φ t . μ- <

H ^• ii ϋ 0 TI μ-

<! Ω ra n tr a 0 r H Φ tr TJ ØJ Ω o a ; a a ØJ Ω M en rt Hi 0

^•< φ C- - C Φ ϋ 0 tr A Φ φ rt TI TJ Ω Ω w cn Ω tr tr ØJ μ- ¹ a ØJ a rt en μ, Φ C ii Φ TJ 0 Ω rt Φ y μ- TI 0 f- Ω 3 0 0

Ω rt Φ ar CQ 3 μ- ii tr ui ii μ- CO rt μ- μ> o 3 0 tr t ra φ £ ra φ φ 0 LQ ii Λ φ φ ØJ φ »i c μ- a _^ rt TJ 3 Φ Ω 3 μ- Φ μ TJ μ-

Φ - φ Ω C o ii μ- μ- Ω φ ^• Φ TJ o TI φ a rt Ω

0J 0J Ω TJ Φ Φ rt φ a o a Ω CO ØJ ra φ TJ a φ 0J O a μ- φ tr £ ¹ 0J ti μ C cn μ LQ Φ μ- Ω rt ii rt co

Φ TJ α rt rt Φ c rt ØJ 0J μ- μ^j

3 tr t rt rt o o rt Ω tr P. r 0 Φ Hi a φ φ

Ω P^ Φ 0 ii ϋ a φ Φ ØJ μ- tr 0J μ- C Φ rt ii Φ rt ii a P. rt O LQ

VEGF-C has been implicated m lymphangiogenesis but might also be involved m tumor angiogenesis since it is also found to be produced by some tumor cells.

In conclusion, we have created m the baculovirus expression system a heterodimeric VEGF variant with one wild-type and one mutated receptor binding interface. This molecule acts as a potent antagonist for VEGF receptors KDR and Flt-1, thereby not only antagonizing the activity of VEGF, but possibly also the activities of VEGF-B, C, D and E. The heterodimer inhibits wild-type VEGF activity m a number of m vi tro and in vivo assays and might be helpful as a therapeutic anti-angiogenic agent, not only m tumor treatment, but also in other angiogenesis-dependent disorders like diabetic retmopathy.

EXAMPLES

Example 1 Construction of recombinant baculoviruses All enzymes for DNA manipulations and culture media for growth of insect cells were purchased from Life Technologies (Breda, The Netherlands) unless otherwise stated. The isolation and cloning m pBluescπpt of the complete coding region of VEGF_16S cDNA has been described previously (28) . Using the helper phage M13K07, single stranded template DNA for mutagenesis was prepared according to standard protocols. Mutagenesis was performed using the Amersham Sculptor Kit (Amersham, Buckinghamshire, England). Oligonucleotides to introduce the C2S and C4S mutations were described previously (28) . The VEGF- CyslSer mutant was created by mutagenesis using oligonucleotide

5 ' -cgcagctactcccatccaatc-3 ' . The receptor binding domain loop 1 was substituted for the corresponding loop from PDGF-B m two sequential steps with the 45-mer Bl (5'- accctqqtqqacatctcccqqcσcctc taqatqaqatcgaqtac-3 ' ) and the 45- mer B2 (5 ' -cgqcqcctcataqatcgcaccaacqccaacttcaaqccatcctqt-3 ' ) , using the Bl mutant as a template. This procedure resulted m plasmid pBS-VEGF-Ll. The loop 2 mutation (exchanging loop 2 from VEGF by the corresponding region of PDGF-B) was introduced using the 51-mer 5¹- qqqqqctqctqcaataaccqcaacqtqcaqtqccqccccactqaqqaqtcc-3 ' , resulting in pBS-VEGF-L2. When required, C2S and C4S mutations were combined with the loop 1 or loop 2 mutations to yield clones pBS-VEGF-C2SL2 or pBS-VEGF-C4SLl . Inserts were isolated as Sall-Xbal fragments and cloned in Sail -Xbal -cut vector pFastBacI (Invitrogen) . Recombination of transfer vector and baculovirus genome was allowed to take place in the E.coli strain DH10-BAC. Positive clones, identified by blue-white screening, were used for isolation of recombinant baculoviral DNA, which was subsequently transfected into SF9 insect cells using Insecticin reagent (InVitrogen) . After 5 days, conditioned media were assayed for VEGF (mutant) content by western blotting, using a rabbit anti-VEGF antibody. Viral stocks were generated by infecting SF9 cells at an M.O.I, of 0.1 and collecting conditioned media 5 days later. To confirm the presence of the desired mutations, DNA was isolated from virus stocks and PCR-sequenced using the Amersham cycle sequencing kit .

Example 2 Production of mutant VEGFs in the baculovirus expression system SF9 cells were seeded at a density of 10^s cells/cm² in Grace Insect medium, supplemented with 10% fetal bovine serum and penicillin/streptomycin. Cells were allowed to adhere to the culture plastic for 2 hrs before infection with recombinant viruses with an M.O.I, of 5. For generation of the heterodimer, SF9 cells were co-infected with VEGF-C2SL2 and VEGF-C4SL1 baculoviruses , both at an M.O.I, of 5. 3 days afterwards, medium was collected and mixed with 50 : 1 of heparin-Sepharose slurry/ml (Pharmacia, Uppsala, Sweden) . VEGF or its mutants were allowed to bind for 2 hrs at 4°C, after which the beads were washed 3 times with phosphate-buffered saline (PBS) . Finally, bound proteins were eluted using PBS containing 1.5 M NaCl . After removal of the beads by centrifugation, proteins were aliquoted and stored at -80°C. Concentration of the VEGF proteins was measured by ELISA (Santa Cruz) . Because mutant VEGFs were not efficiently recognized in this moAb based sandwich ELISA, concentrations of these were determined by western blotting using a known amount of wtVEGF_1S5 as a standard and a rabbit polyclonal anti -VEGF antiserum.

Example 3 Procoagulant assay- Expression of TF by HUVECs in response to VEGF treatment was measured in the procoagulant assay as described elsewhere (28) . In short, HUVECs were incubated in 6 well dishes (Costar) for 5 hr with the agent of study in PMB medium (EMEM, containing 15% newborn calf serum, 50 μg/ml polymyxin B, 2 mM L-glutamine and 40 μg/ml gentamycin) . Endothelial cells were collected, washed and suspended in 100 μl Veronal buffer (7 mM Veronal , 4 mM Na- acetate, 130 mM NaCl , pH 7.4) . After mixing cells with 100 μl 20 mM CaCl₂ and 100 μl pooled normal human plasma at 37°C, the time needed for fibrin clot formation was recorded. Experiments were always performed in duplicate.

Example 4 Proliferation assay

The proliferative response of HUVE cells towards VEGF or its mutants was measured using a BrdU incorporation kit (Boehringer Roche) . HUVE cells were plated at a density of 5000 cells/well in flat bottom 96-well dishes in EMEM containing 10% pooled human serum and 20% new born calf serum (NCS, Life Technologies, Breda, The Netherlands) . The next day, medium was replaced by 100 μl EMEM containing 0.5% NCS and factors to be tested. After one day of incubation, BrdU was added and cells were cultured further for one day. Subsequently, the cells were fixed and incorporated BrdU was quantified by immunodetection. Example 5 Miles vascular permeability assay

Anesthesized Hartley guinea pigs were shaved and injected intramuscularly with 0.2 ml of Phenergan" (2,5% promethazine, Rhone-Poulenc) to reduce mast-cell induced background permeability. After 20 minutes, the animals were injected intracardially with 2 ml of a 0.5% Evans Blue solution in 0.9% NaCl. After another 20 minutes, 20 ng of VEGF (mutants) were injected intracutaneously in the flank in a volume of 50 :1. After extravasation of the Evans Blue dye (10-20 minutes) the animal was photographed.

Example 6

Increasing- affini ty of the HD-VEGF variant for the respective receptors :

For HD-VEGF to be useful as an antagonist, its affinity for receptors should be equal to or higher than the affinity of wtVEGF. Binding of HD-VEGF to KDR is mediated by the unmutated LI domain in the VEGF-C2SL2 subunit (see the model in figure

1) . Since VEGF-L2 homodimers have a slightly decreased affinity for KDR, as is demonstrated by the receptor competition studies (figure 3), it is of importance to compensate for this decrease in affinity. To this end, a phage display experiment was performed, utilizing a phage library that displays at random 10 -mer peptides in the piII protein. In short, a panning was performed on immobilized KDR receptor. Phages that bound to or in the neighbourhood of the VEGF-binding site on the receptor, were specifically eluted using high concentrations of wtVEGF. In three consecutive rounds of panning, an enrichment of phage binding was observed, which is indicative of specific binding. From the last panning experiment, approximately 100 phages were grown individually and tested for their capacity to bind to immobilized KDR. From this set, sixty phages were identified that bound specifically to the VEGF binding domain of KDR. Sequence analysis revealed that this group consisted of eleven independent phages, of which the 10 -mer sequences and the relative occurance are illustrated in table I .

10- -mer sequence occurance

1 , SRSWA WEWL 19/60

2. TRDWYYDFLV 21/60

3. TLYYWEWSV 1/60

4. KLVY EEVFH 2/60

5. DLIYWESVTL 4/60

6. DFWYWEMLGM 6/60

7. ATLRFRWNN 1/60

8. LREWQWWEWL 1/60

9. LTPVWRPHPG 1/60

10 ITWFTKGGMR 1/60

11 ELWYWETVLV 1/60

Table I: amino acid sequences of 10-mers, displayed by phages that specifically bind to or near the VEGF binding site of the KDR receptor. Notice (in bold) the consensus YWE which is found in a large part of the poulation.

When these peptides bind to the VEGF binding site on KDR, hey should be able to inhibit binding of wtVEGF. At least one peptide, peptide 6, is able to reduce binding of wtVEGF to the KDR receptor to background levels (figure 6) .

Examination of the 3D-structure of the VEGF homodimer shows that in the structure the first nine amino acids are not seen. The explanation for this is that this sequence is flanked at the C-terminus by a very flexible sequence (Gly-Gly-Gly) . It is also known from literature that this region can be missed for receptor binding, a VEGF mutant which ranges from aa 9 to 110 contains all receptor bidning information. This knowledge can than be used to incorporate the 10-mer sequences: receptor- binding 10-mers are cloned N-terminally from this glycine stretch. This will generate a flexible extra binding site for KDR to VEGF. In this way, affinity of VEGF for the receptor can be increased. Upon identification of the mutants with increased affinity, these will be incorporated into HD-VEGF to generate a more efficient antagonist.

Similar ways can of course be employed to identify mutants with increased affinity for the Flt-1 receptor.

Example 7

In vivo activi ties of VEGF '-C4SL1 -wtVEGF '-heterodimer ε

DNA constructs - Mutations in VEGF to obtain VEGF-C4SL1 were described previously. In short, this mutant contains a combination of mutations Cys60Ser, preventing its homodimerization (28) and a swap mutation of loop 1, by which binding to the KDR/Flk-1 receptor is lost. The sequence was based on the VEGF165 isoform. VEGF-C4SL1 was cloned as an

EcoRI-Xbal (blunt) fragment behind the CMV promoter in vector pIRES-neo (Clontech) using the EcoRl and BamHl (blunt) sites, leading to plasmid pCMV-VEGF-C4SLl-IRES-neo .

Cell cul ture - All media and antibiotics were obtained from Life Technologies (Breda, The Netherlands) . Culture of MV3 cells has been described. Cells were maintained in Dulbecco's Modified medium (DMEM) supplemented with 10% fetal calf serum and penicillin/streptomycin. Transfections were performed with Fugene 6 (Boehringer Roche, Germany) according to the manufacturer's instructions. Two days after transfection, cells were placed in selective medium (same medium with 1 mg/ml G418) . Surviving clones were isolated and grown individually in medium containing 400 μg/ml G418. Total RNA was isolated from the transfected cell lines using Trizol (Life Technologies) and samples of 5 μg were subjected to northern blotting to allow for the detection of recombinant RNA. Also, in order to check for protein production, conditioned media were subjected to heparin- Sepharose chromatography and eluted proteins were electyrophoresed , followed by SDS-PAGE and western bloting.

Animal experiments - MV3 cells or transfectants were trypsinized, washed 3 times in phosphate buffered saline (PBS) and resuspended in PBS at 10⁷ cells/ml. 100 μl cell suspension was injected on the flank in the subcutaneous of nude male

Balb/c mice of 6-7 weeks of age. Experiments were performed in duplicate, using 5 mice per experiment. Tumor volumes were measured weekly. At the end of each experiment, mice were anesthesized, bled and tumors were removed for further analysis.

Generation of stably transfected MV3 cell lines - Transfection of MV3 cells with pCMV-VEGF-C4SLl-IRES-neo yielded a number of clones of which two were subcloned and used for further experiments. Northern analysis of total RNA from these clones identified both as high expressors of the recombinant RNA (not shown) . Growth of the different cell lines in vi tro or in vivo after injection in the subcutaneous space of immunodeficient nude mice is depicted in figure 7. Although both VEGF-C4SL1- expressing clones had in vitro growth rates, comparable to that of the parental line (figure 7A) , in vivo a strong reduction of tumor growth was observed for both cell lines (figure 7B) .

MV3 cells produce readily detectable levels of endogenous VEGF. It is expected that these enodgenously produced molecules heterodimerize with the recombinant C4SL1 variant, to yield a dimer with at one side a defective KDR binding domain. These heterodimers should thus act as KDR- but not Flt-1 antagonists. We hypothesize that these heterodimers are responsible for the tumor growth inhibition that is observed in this experiment. CITED LITERATURE

1. Folkman, J. & Klagsbrun, M. (1987) Science 235, 442-447.

2. Folkman, J. & Klagsbrun, M. (1987) Nature 329, 671-672.

3. Takahashi, Y. , Kitadai, Y. , Bucana, C, Cleary, K. & Ellis L. (1995) Cancer Res . 55, 3964-3968. 4. Viglietto, G., Maglione, D., Rambaldi, M. , Cerutti, J. , Romano, A., Trapasso, F., Fedele, M. , Ippolito, P., Chiappetta, G. , Botti, G. , Fusco, A. & Persico M. (1995) Oncogene 11, 1569-1579.

5. Asano, A., Yukita, A., Matsumoto, T., Kondo, S. & Suzuki H. (1995) Cancer Res . 55, 5296-5301.

6. Benjamin, L.E. & Keshet, E. (1997) Proc . Natl . Acad . Sci . USA 94, 8761-8766.

7. Borgstrόm, P., Hillan, K.J., Sriramarao, P. & Ferrara, N. (1996) Cancer Res . 56, 4032-4039. 8. Cheng, S-Y., Huang, S., Nagane, M. , Ji , X-D., Wang, D., Shih, C.C-H., Arap, W. , Huang, C-M. & Cavenee W.K. (1996J Proc . Natl . Acad . Sci . USA 93, 8502-8507.

9. Harris, A.L. (1997) Lancet (sup II) 349, 13-15.

10. Jin Kim, K. , Li , B., Winer, J. , Armanini , M. , Gillett, N., Philips, H.S. & Ferrara N. (1993) Nature 362, 841-844.

11. Millauer, B., Longhi , M.P., Plate, K.H., Shawver, L.K., Risau, W., Ullrich, A. 6_ Strawn L.M. (1996) Cancer Res 56,

1615-1620.

12. Leenders, W.P.J. (1998) Int J Exp Pathol 79, 339-346. 13. Bischoff, J. (1997) J^" Clin Invest 99, 373-376.

14. Hanahan D., Folkman J. (1996) Cell 86, 353-364.

15. Klagsbrun M. (1991) J^" Cell Biochem 47, 199-200.

16. Gengrinovitch, S., Greenberg, S.M., Cohen, T. , Gitay- Goren, H., Rockwell, P., Maione, T.E., Levi, B-Z. & Neufeld, G. (1995) J Biol Chem 270, 15059-15065.

17. Kim Lee Sim, B., O'Reilly, M. , Liang, H., Fortier, A., He, W., Madsen, J. , Lapcevich, R. & Nacy CA. (1997) Cancer Res 57, 1329-1334. 18. O'Reilly, M. , Boehm, T., Shing, Y., Fukai , N. , Vasios, G., Lane, W., Flynn, E., Birkhead, J., Olsen, B. & Folkman J.

(1997) Cell 88, 277-285.

19. d'Angelo, G. , Struman, I., Martial, J. & Weiner R.I. (1995; Proc . Natl . Acad . Sci . (USA) 92, 6374-6378.

20. Kendall, R.L., Wang, G. & Thomas, K.A. (1996) Biochem Biophys Res Comm 226, 324-328.

21. Suri, C, McClain, J. , Thurston, G., McDonald, D.M., Zhou, H., Oldmixon, E.H. & Yancopoulos, G.D. (1998) Science 282, 468-470.

22. Neufeld, G., Cohen, T., Gengrinovitch, S., Poltorak, Z. (1999) The FASEB J 13, 9-22.

23. Meyer, M. , Clauss, M. , Lepple-Wienhues, A., Waltenberger, J. , Augustin, H.G., Ziche, M. , Lanz, C, Buttner, M. , Rziha, H.-J. & Dehio, C. The EMBO J 18, 363-374.

24. Joukov, V., Pajusola, K. , Kaipanen, A., Chilov, D., Lahtinen, I., Kukk, E., Saksela, 0., Kalkkinen, N. & Alitalo, K. (1996) The EMBO J 15, 290-298.

25. Olofson, B., Pajusola, K. , Kaipanen, A., von Euler, G. , Joukov, V., Saksela, 0., Orpana, A., Petterson, R. , Alitalo,

K. & Eriksson, U. (1996) Proc Natl Acad Sci USA 93, 2576-

2581.

26. Park, J., Chen, H. , Winer, J. , Houck, K. & Ferrara N. (1994) J Biol Chem 269, 25646-25654. 27. Andersson, M. , Ostman, A., Backstrom, G., Hellman, U. ,

George-Nascimento, C, Westermark, B. & Heldin C-H. (1992) J. Biol . Chem. 267, 11260-11266.

28. Potgens, A.J., van Altena, M.C., Lubsen, N.H. , Ruiter, D.J. S de Waal, R.M.W. J Biol Chem 269, 32879-32885. 29. Claffey P.C., Senger D. & Spiegelman B.M.(1995) Biochim Biophys Acta 1246, 1-9.

30. Soker, S., Fidder, H., Neufeld, G. & Klagsbrun, M. (1996) J Biol Chem 271, 5761-5767. 31. Shalaby, F., Rossant, J. , Yamaguchi, T.P., Gertsenstein, M. , Wu, X-F., Breitman, M.L. & Schuh A.C. (1995) Nature 376, 62-66.

32. Fong, G.-H., Rossant, J. , Gertsenstein, M. & Breitman, M.L. (1995) Nature 376, 66-70

33. Barleon, B., Sozzani, S., Zhou, D., Weich, H.A., Mantovani, A. & Marme, D. (1996) Blood 87, 3336-3343.

34. Clauss, M., Weich, H. , Breier, G. , Knies, U. , Rockl , W. , Waltenberger, J. & Risau, W. (1996) J Biol Chem 271, 17629- 17634.

35. Andersson, M. , Ostman, A., Kreysing, J. , Backstrom, G. , van der Poll, M. & Heldin, C.-H. (1995; Growth Factors 12, 159-164.

36. Keyt, B.A., Nguyen, H.V., Berleau, L.T., Duarte, CM., Park, J. , Chen ,H. & Ferrara, N. (1996) J Biol Chem 271,

5638-5646.

37. Fuh, G. , Li, B., Crowley, C, Cunningham, B. & Wells, J.A. (1998) J Biol Chem 273, 11197-11204

38. Siemeister, G., Schirner, M. , Reusch, P., Barleon, B., Marme, D. & Martiny-Baron, G. (1998) Proc Natl Acad Sci USA

95, 4625-4629.

39. Muller, Y.A. , Li , B., Christinger, H.W., Wells, J.A. , Cunningham, B.C. & de Vos, A.M. (1997) Proc Natl Acad Sci USA 94, 7192-7197. 40. Plate, K.H., Breier, G., Weich, H.A., Risau, W. (1992) Nature 359, 845-848

41. Aiello, L.P., Avery, R.L., Arrigg, P.G., Keyt, B.A., Jampel, H.D., Shah, S.T., Pasquale, L.R., Thieme, H., Iwamoto, M.A. , Park, J.E. (1994) N. Engl . J. Med. 331, 1480- 1487

42. Detmar, M. , Brown, L.F., Claffey, K.P., Yeo, K.T., Kocher, 0., Jackman, R.W., Berse, B., Dvorak, H.F. (1994) J. Exp/ Med. 180, 1141-1146 43. Fava, R.A. , Olsen, N.J., Spencer-Green, G. , Yeo, K.T., Yeo, T.K., Berse, B., Jackman, R.W., Senger, D.R., Dvorak, H.F., Brown, L.F. (1994) J. Exp. Med. 180, 341-346

44. Gabrilovich D., Ishida T., Oyama T., Ran S., Kravtsov V., Nadaf S., Carbone D. Vascular endothelial growth factor inhibits the development of dendritic cells and dramatically affects the differentiation of multiple hematopoietic lineages in vivo. Blood 1998, 92:4150-4166

45. Oyama T., Ran S., Ishida T., Nadaf S., Kerr L. , Carbone D., Gabrilovich D. Vascular endothelial growth factor affects dendritic cell maturation through the inhibition of nuclear factor-kappa B activation in hemopoietic progenitor cells. J Immunol 1998, 160:1224-1232

Legends to figures

Figure 1 - Model of binding of VEGF to its receptors KDR and Flt-1, and proposed broad antagonism of the heterodimeric variant. 1A) wtVEGF dimerizes KDR receptors via binding domains involving loop 1 and 3 (Ll/3) on opposite sides of the symmetrical dimer, while loop 2 (L2) is involved in binding Flt-1 and KDR receptors. IB) The heterodimeric protein can only bind monomeric receptors and thus will not give rise to receptor activation. Occupance of the VEGF- binding sites on the receptors will confer antagonistic behaviour upon this molecule.

Figure 2 - Western blots of a reducing (A) and non-reducing (B) 12% PAGE of recombinant baculovirus-derived VEGF mutants. All proteins migrate as doublets of ~23 kDa on reducing gels, while wtVEGF (lanes 1) migrates as a 45 kDa dimer in non- reducing gels. Note that VEGF-C2SL2 (lanes 3) and VEGF-C4SL1 (lane 4) migrate as monomers on non-reducing gels, while after co-infection of SF9 cells with these recombinant baculoviruses , a 45 kDa protein is secreted (lane 5), indicating efficient heterodimerization of VEGF-C2SL2 and VEGF-C4SL1.

Figure 3 - A) Induction of proliferation of HUVEC by VEGF and VEGF mutants. Experimental procedures are explained in the text. Note that homodimeric VEGF-L2 activity similar to that of wtVEGF

homodimeric VEGF-L1 ( Q ) and the heterodimeric VEGF- C2SL2/C4SL1 ( 1 ) do not display detectable activity.

At high concentrations, VEGF-L3 displayed significant activity. B) Specific inhibition of wtVEGF induced HUVEC proliferation by VEGF-C2SL2/C4SL1. HUVE cells were incubated with increasing amounts of mutant proteins prior to activation with 10 ng wtVEGF. The observed increase of proliferative activity after addition of VEGF-L2 is due to the proliferative activity of this molecule itself. Values on the Y-axis represent extinctions, measured after immunodetection of incorporated BrdU.

Figure 4 - A) Induction of procoagulant activity by VEGF and VEGF mutants. HUVE cells were incubated with VEGF or mutants thereof, and tissue factor expression was measured in a procoagulant assay as described in the text. VEGF activities are displayed as shortening of blood coagulation time, which is a measure for the amount of tissue factor expression.

Mutation of both loops 1 (VEGF-L1) , 2 (VEGF-L2) and 3 (VEGF- L3 ) led to a complete loss of activity. The heterodimeric VEGF mutant was also completely inactive. In figure B, HUVE cells were incubated with increasing amounts of VEGF-C1S or VEGF-C2SL2/C4SL1 prior to activation with 20 ng/ml wtVEGF. Note that VEGF-C1S, which is completely inactive in all assays, had no effect on wtVEGF-induced tissue factor expression, while only a 4-fold excess of VEGF-C2SL2/C4SL1 sufficed to almost completely block TF expression.

Figure 5 - Vascular permeability induction in the Miles assay. Guinea pigs were injected in the heart with an Ξvans ' Blue solution, prior to intradermal injection with samples to be tested. Figure A) shows the activities of VEGF and VEGF- loop mutants. Although VEGF-L2 and VEGF-L3 had reduced activities as compared to wtVEGF, these mutants were significantly more active than VEGF-L1. Nonrelevant spots were not omitted from this figure for technical reasons. B) Inhibition of wtVEGF by mutant proteins was examined by injecting these proteins 10 min before intracardial injection with Evans Blue. Samples tested are as indicated in the lower part of the figure.

Figure 6 Peptide 6 (nr. 22-037) inhibits binding of wtVEGF to predimerized KDR in a specific manner. Peptides 1 and 2 (22- 038 and 22-039 respectively) do not inhibit, these peptides probably do not bind to the exact VEGF binding site but to a KDR sequence, located nearby the VEGF binding site. Thus, peptides do bind, but are incapable of inhibiting wtVEGF binding. Selection of the corresponding phages from the panning experiments can be explained by steric hindrance: VEGF displaces the large phage, but not the small peptide.

Figure 7A) Expression of VEGF-C4SL1 has no effect on proliferation in vitro. Cells were seeded at 10,000 cells/well in 6-wells dishes. At day 1, 2, 3 and

4 cells were trypsininzed, washed in PBS and counted in a Coulter Counter. B) Cells were injected at 106 in a volume of 100 μl PBS in the flank of male nude mice. Tumor volumes were estimated weekly by measuring the tumor in three dimensions.

Claims

1. An isolated and/or recombinant nucleic acid which encodes a VEGF antagonist derived from VEGF, wherein said nucleic acid comprises at least one mutation which alters the receptor binding affinity for a VEGF receptor of the resulting (poly) peptide in a sequence which in VEGF encodes loop I, at approximately amino acids 30-50.

2. An isolated and/or recombinant nucleic acid according to claim 1, which comprises at least a mutation in the sequence which in VEGF encodes amino acids 36-46.

3. An isolated and/or recombinant nucleic acid according to claim 1 or 2 , in which the binding affinity for a VEGF receptor is enhanced through said mutation.

4. An isolated and/or recombinant nucleic acid according to claim 1, 2 or 3 , in which sequences encoding other VEGF receptor binding sites, in particular loop II encoding sequences are absent or functionally deleted.

5. An isolated and/or recombinant nucleic acid according to claim 1 or 2 , wherein the receptor binding affinity through a loop I-like sequence for a VEGF receptor is significantly reduced and wherein binding affinity for a VEGF receptor is provided through at least one sequence which in VEGF encodes loop II.

6. An isolated and/or recombinant nucleic acid according to claim 5, further comprising a functional deletion of the sequence encoding the receptor binding site which in VEGF is located at loop III.

7. An expression vector comprising at least one nucleic acid according to any one of claims 1-4.

8. An expression vector comprising at least one nucleic acid according to any one of claims 1-6.

9. An expression vector comprising at least one nucleic acid according to any one of claims 1-4 and at least one nucleic acid according to claim 5 or 6.

10. A cell comprising a nucleic acid according to anyone of claims 1-6, or an expression vector according to claim 8 or 9.

11. A cell comprising an expression vector according to claim 7 and an expression vector comprising a nucleic acid encoding a VEGF analogue having a functional loop II and a functional deletion of at least loop I .

12. A cell comprising a nucleic acid according to any one of claims 1-4 and a nucleic acid encoding a VEGF analogue having a functional loop II and a functional deletion of at least loop I .

13. A cell according to claim 11 or 12 in which said nucleic acid encoding a VEGF analogue encoding a functional loop II further comprises a functional deletion of a sequence encoding loop III.

14. A proteinaceous VEGF antagonist which is a monomeric VEGF analogue comprising a funtional receptor binding sequence which in VEGF is located at loop I.

15. A VEGF antagonist according to claim 14, in which the binding affinity of loop I is enhanced in relation to wildtype VEGF through a mutation.

16. A heterodimeric proteinaceous VEGF antagonist comprising one functional receptor binding interface and one dysfunctional receptor binding interface, wherein said dysfunctional interface comprises a mutation in a sequence which in VEGF is located in loop I.

17. A heterodimerio VEGF antagonist according to claim 16 in which said dysfunctional interface further comprises at least one mutation in a sequence which in VEGF is located in loop II or loop III.

18. A heterodimeric proteinaceous VEGF antagonist comprising one functional receptor binding interface and one dysfunctional receptor binding interface, wherein said dysfunctional interface comprises a mutation in a sequence which in VEGF is located in loop II and wherein the functional interface comprises a mutation enhancing receptor binding affinity in loop I .

19. A heterodimeric proteinaceous VEGF antagonist according to claim 16, 17 or 18, further comprising a functional deletion of a cysteine located at the second and/or fourth position of cysteine in the cysteine knot of the unmodified VGEF.

20. A gene delivery vehicle comprising a nucleic acid according to anyone of claims 1-6, or an expression vector according to any one of claims 7-10.

21. A gene delivery vehicle encoding a VEGF antagonist according to any one of claims 14-19.

22. A kit of parts for the inhibition of angiogenesis comprising at least two gene delivery vehicles each encoding a subunit of a heterodimeric VEGF antagonist according to claim 16, 17, 18 or 19.

23. A gene delivery vehicle according to any one of claims 20-21, which is of viral, preferably adenoviral, adeno- associated viral or retroviral origin.

24. A kit of parts according to claim 22 in which a gene delivery vehicle is of viral, preferably adenoviral, adenoassociated viral or retroviral origin.

25. A method for producing a VEGF antagonist, comprising culturing a cell according to claim 11, 12 or 13 under suitable conditions and harvesting the antagonist from the culture .

26. A method for producing a gene delivery vehicle encoding a monomeric VEGF antagonist or at least one subunit of a heterodimeric VEGF antagonist , comprising inserting a nucleic acid encoding said antagonist or said subunit in a defective adenoviral vector, transfecting said vector into a complementing cell, culturing said cell under suitable conditions and harvesting said gene delivery vehicle from the culture .

27. A proteinaceous substance having affinity for a Kinase domain Receptor (KDR) , comprising at least 10 amino acid residues of which three consecutive residues are YWE.

28. A proteinaceous substance according to claim 27, comprising said three consecutive residues at positions 4-6 of said at least 10-mer.

29. A proteinaceous substance according to claim 27 or 28, which is an antagonist.

30. A proteinaceous substance according to claim 27, 28 or 29, which is a homodimer.

31. A proteinaceous substance according to any one of claims 27-29, which is a heterodimer.