This specificity distinguishes pegaptanib from other VEGF antagonists, which inhibit all isoforms. Pegaptanib binds to the heparin-binding domain of VEGF165,

which is not present in VEGF121, with high affinity (12 nM) [141]. Two possible mechanisms of inhibition have been suggested, steric interference and the preven-

tion of the interaction of the heparin-binding domain with heparan sulfates and NP-1. In steric interference, the aptamer would prevent the interaction of the recep- tor-binding domain with cell surface receptors. The other mechanism is based on the theory that the heparin-binding domain might increase the local concentration of VEGF165 at the cell surface by interacting with HS, thus enhancing the probability of receptor binding [141]. Pegaptanib might capture soluble VEGF165, preventing the interaction with cell surface proteoglycans. Additionally, the heparin-binding domain binds to NP-1, so pegaptanib may block the interaction between NP-1 and VEGF165, thereby diminishing VEGF165-induced signal transduction. Work of our own lab indicates that steric hindrance is not the main mechanisms of VEGF165. In an organ culture model, pegaptanib was not able to prevent the binding of antibodies to the receptor binding domain, suggesting that pegaptanib would not be able to prevent the binding of VEGF to its receptor, either [109]. While inhibitors like bevacizumab or ranibizumab inhibit the receptor binding and hence signal transduction itself, according to this model, pegaptanib would only inhibit the enhancement of receptor signal transduction. Pegaptanib is not as efficient in the treatment of wet AMD, which could be explained by this mode of action. Additionally, one has to keep in mind that a strong angiogenic potential of VEGF121 was shown in cancer [142], so it might have similar effects in the retina as well.

13.3.2Bevacizumab and Ranibizumab

In contrast to pegaptanib, both bevacizumab and ranibizumab are proteins. Bevacizumab is a humanized antibody, developed for intravenous use in cancer therapy [143], while ranibizumab, developed for intravitreal application, is an Fab fragment which has been affinity maturated [144]. A higher efficacy of ranibizumab could be confirmed in vitro [109]. Both proteins have been developed from the same molecule, a murine anti-VEGF antibody [15, 145]. Neither bevacizumab nor ranibizumab binds to murine VEGF, as a glycine is exchanged with a serine at AS88 [146, 147]. This should be kept in mind when assessing data about these agents that have been obtained in rodents. Both proteins bind to all available isoforms of VEGF (explicitly shown for ranibizumab for isoforms 165, 121 and 110) [148] and display good clinical performance, though bevacizumab has not been tested in clinical trials of high evidence so far [149]. They exert a complete neutralization of VEGF in vitro when used in clinical concentrations, but ranibizumab exerts a higher efficacy than bevacizumab when diluted. Additionally to their neutralizing effect on VEGF, they reduce intracellular VEGF protein content, suggesting autocrine mechanisms of VEGF regulation [109]. Both proteins increase the permeability of RPE cells [150] and have an effect on RPE proliferation [151],

with bevacizumab exerting the more profound effect. In vitro, bevacizumab, but not ranibizumab, is accumulated in the RPE and decreases phagocytotic function, while accumulation of bevacizumab does not extend the extracellular neutralization of VEGF [151, 152].While bevacizumab, ranibizumab, and pegaptanib have no direct toxic effect on ocular cells [153], they can interfere with physiological VEGF action. Bevacizumab neutralizes protection of retinal ganglion cells by VEGF after oxidative stress [154]. Also, in primates, bevacizumab treatment resulted in ultrastructural changes in the choroid, reduced endothelial cell fenestration, and photoreceptor damage [155]. Bevacizumab and ranibizumab bind primarily to the amino acids 81–96, of which only Ile83 is also involved in receptor binding. The neutralizing effect is most likely due to steric hindrance and not to a competition for the same binding determinants [156].

13.3.3VEGF-Trap Eye

VEGF-Trap is a recombinant soluble VEGF receptor protein combining the second Ig domain of VEGFR-1 fused with the third Ig domain of VEGFR-2, combined with an IgG Fc portion, and acts as an decoy receptor. The affinity of this recombinant receptor is very high for all isoforms of VEGF-A and additionally binds to PlGF [157]. VEGF-Trap forms 1:1 complexes with VEGF, in contrast to bevacizumab, which forms multimeric complexes [158]. VEGF-Trap Eye has the same chemical structure as aflibercept, which is used in oncology, but is differently processed, as it is more highly purified and the buffer formulation differs from its oncological counterpart [159]. Few studies on the action of VEGF-Trap Eye in the retina have been completed so far. As it exerts the same neutralizing effect as ranibizumab and bevacizumab, a similar influence on VEGF expression and on retinal cells can be expected.

13.3.4siRNA

The protein-based VEGF antagonists as well as the pegaptanib aptamer sequester VEGF extracellularly in order to neutralize it. The siRNA approach, on the other hand, prevents the intracellular translation of VEGF mRNA by a sequence specific, posttranscriptional gene silencing method, designated RNA interference. In general, RNA interference functions by double stranded RNA homologues that are introduced into the cell and are processed by Dicer, a cellular RNase III. This generated srRNA duplex of 21 nt with 3¢ overhangs is incorporated in a multiprotein RNA-inducing silencing complex (RISC), which unwinds the duplex RNA, binding the homologue mRNA target in order to onset endonucleolytic cleavage [160]. Drawbacks of this system are the limited duration of the treatment, as siRNA is not replicated in mammalian cells, and the fact that the siRNA has to be incorporated

into the cells. For the treatment of AMD, two different approaches of siRNA have been developed, one to silence the expression of VEGF (Cand5, Bevasiranib) [161] and one to silence the expression of VEGFR-1 (Sirna-027) [162]. Both Sirna-027 and Bevasiranib reduce CNV lesion in vivo in animal models [161, 162]. However, the notion of specific mRNA silencing as the mechanism for these substances has been challenged by the assessment that any nonspecific double-stranded RNA would reduce experimental CNV, mediated by toll-like receptor (TLR)-3 signaling [163]. Also, a phase III clinical trial of Bevasiranib was terminated in March 2009 prior to enrollment (http://clinicaltrials.gov/ct2/show/NCT00557791) [159]. Nevertheless, an important function of VEGFR-1 for vascular cells in the retina has recently been proposed (pericytes can be ablated from mature retinal vasculature through VEGFR1-mediated signaling pathways, resulting in increased vascular leakage) [53]. Furthermore, specific downregulation of VEGFR-1 by Sirna-027 has been shown and first results of a phase one clinical trial has been published [162, 164].

13.3.5Small Molecule Tryrosine Kinase Inhibitors

In order to induce angiogenesis, VEGF binds to its receptor, a tyrosine receptor kinase, which autophosphorylates upon dimerization in order to induce a signaling cascade. The signaling is conducted by protein kinases, that transfer phosphate residues to proteins in order to regulate enzyme activation. Kinase inhibitors prevent this signal transduction because they prevent phosphorylation of target proteins. Several tyrosine kinase inhibitors have been developed in order to prevent VEGFR activation. These inhibitors are small chemical molecules, which have originally been developed for cancer therapy [57]. Due to their size and chemical properties, they do not necessarily need to be intravitreally injected but may be applied as eye drops. They are hydrophobic and can readily cross the cell membrane to act intracellularly. An oral application is also possible, but because kinases are present throughout the body, side effects might be an issue [165].

Tyrosine kinase inhibitors that are in clinical development generally are multityrosine kinase inhibitors (pazopanib, sunitinib, vatalanib) or multikinase inhibitors (sorafenib) [166]. Specific VEGFR-2 tyrosine kinase inhibitors are available (SU1498) but only experimentally used. These small kinase inhibitors compete with ATP, targeting the ATP binding site of a kinase. Tyrosine kinases inhibitors can be divided into three subgroups [166]. Sunitinib is a type I inhibitor, recognizing the active confirmation of a kinase, competing with its ATP-binding site [167]. Sorafenib is a type II inhibitor, binding to the ATP binding site of inactive kinases [167]. A third type of inhibitors binds covalently to cysteine residues of the kinase, blocking the binding of ATP to its kinase [168]. Generally, binding at the inactive side confers a higher specificity, as the inactive confirmation is unique to the kinase, while targeting the active site may be favorable in diseases with activating mutations [167]. All small molecule tyrosine kinase inhibitors prevent receptor

activation, hence disabling signal transduction in the target cell. They do not change the availability of extracellular VEGF, but might reduce VEGF concentration through inhibition of autoregulatory pathways.

13.3.6Other Inhibitors

Small molecule ligands for NP-1 have recently been developed, which attenuate the binding of VEGF to NP-1 and were able to reduce VEGFR-2 phosphorylation and migration inhibition of endothelial cells [169]. Also, VEGFR-1/NP-1 binding peptides have been developed [170]. These peptides inhibit VEGFR-1 signaling through their Arg-Pro-Leu motif targeting an extracellular ligand binding domain [171].

13.4Outlook on Anti-VEGF Therapeutics

13.4.1Specific VEGF Inhibition

As VEGF has a wide variety of physiological and protective functions in the retina, a continuing inhibition of VEGF might be deleterious for the retina. Especially with anti-VEGF strategies being approved for the treatment of diabetic macular edema or even being used in ROP, the preservation of physiological VEGF function may be considered more important in the future. The first attempt to do so by developing pegaptanib can be considered to have failed, as it basically has been pushed out of the market by ranibizumab, which has proven to be more effective, at least in exudative AMD [172]. However, this approach should be pursued, especially for the treatment of young patients, who will presumably be under VEGF suppression for a long time. To achieve this, the distinction between physiological and pathological VEGF regulation in the retina can be a rewarding aspect. For example, the MAPK ERK1/2 has been shown to be involved in the upregulation after a variety of stimuli, but not in physiological VEGF regulation in the RPE [120, 129, 173]. While the inhibition of an ubiquitous kinase like ERK1/2 is most likely to be associated with unwanted side effects [165], the approach of utilizing the differences between physiological and pathological VEGF secretion in the retina might be a promising aspect for the future. As elucidated above, many different stimuli induce VEGF expression in different pathways. The inhibition of specific hallmarks of pathological VEGF induction, e.g., HuR binding, and the enhancement of repressing mechanisms, e.g., E2F1 induced antiangiogenic shift, are not yet pursued possibilities of a highly specific VEGF regulation. Furthermore, VEGF secretion by different retinal cells are differently involved in pathological alteration of the retina. While Müller cells have been implicated to be responsible for VEGF secretion in diabetic retinopathy, RPE cells have been implicated to contribute to exudative AMD. A cell specific approach could differentially inhibit VEGF by the respective culprit cell type, sparing physiological function.