Transport and fate of aflibercept in VEGF-A165-challenged retinal endothelial cells

Biophysical differences in IgG1 Fc-based therapeutics relate to their cellular handling, interaction with FcRn and plasma half-life

Antibody-based therapeutics (ABTs) are used to treat a range of diseases. Most ABTs are either full-length IgG1 antibodies or fusions between for instance antigen (Ag)-binding receptor domains and the IgG1 Fc fragment. Interestingly, their plasma half-life varies considerably, which may relate to how they engage the neonatal Fc receptor (FcRn). As such, there is a need for an in-depth understanding of how different features of ABTs affect FcRn-binding and transport behavior. Here, we report on how FcRn-engagement of the IgG1 Fc fragment compare to clinically relevant IgGs and receptor domain Fc fusions, binding to VEGF or TNF-α. The results reveal FcRn-dependent intracellular accumulation of the Fc, which is in line with shorter plasma half-life than that of full-length IgG1 in human FcRn-expressing mice. Receptor domain fusion to the Fc increases its half-life, but not to the extent of IgG1. This is mirrored by a reduced cellular recycling capacity of the Fc-fusions. In addition, binding of cognate Ag to ABTs show that complexes of similar size undergo cellular transport at different rates, which could be explained by the biophysical properties of each ABT. Thus, the study provides knowledge that should guide tailoring of ABTs regarding optimal cellular sorting and plasma half-life.

Analysis of clinically approved antibody-based therapeutics reveals different structural designs, such as full-length IgG1 or Fc-fusions, entail distinct biophysical properties that affect FcRn binding, intracellular transport and plasma half-life.

Beovu, but not Lucentis impairs the function of the barrier formed by retinal endothelial cells in vitro

Because rare, but severe adverse effects, i.e. retinal vasculitis or retinal vein occlusion, have been observed after repetitive intravitreal injections of VEGF-A-binding single-chain variable fragment brolucizumab (Beovu), we investigated its possible impact on the barrier formed by immortalized bovine retinal endothelial cells (iBREC) in comparison to that of the VEGF-A-binding Fab fragment ranibizumab (Lucentis). As a measure of stability of the barrier formed by a confluent monolayer of iBREC, we determined the cell index over seven days by continuous electric cell-substrate impedance measurements: Beovu but not Lucentis indeed significantly lowered the cell index, evident about 1.5 days after its addition, pointing to barrier impairment. Early after addition of Beovu, amounts of the integrins α5 and β1-subunits of the fibronectin receptor-had changed in opposite ways, suggesting an effect on cell adhesion due to hindered dimer formation. After exposure for eight days to Beovu, levels of claudin-1-an essential part of the iBREC barrier-were significantly lower, less claudin-1 was located at the plasma membrane after exposure to the VEGF-A antagonist for five days. Beovu did not induce secretion of inflammatory cytokines or VEGF-A. Interestingly, polysorbate-80-component of Beovu-but not polysorbate-20-in Lucentis-slightly, but significantly lowered the cell index, also associated with reduced claudin-1 expression. In summary, our results indicate that Beovu changes the behavior of retinal endothelial cells, thus providing an alternative "non-immunological" explanation for the most relevant of observed side effects.

Impairment of the Retinal Endothelial Cell Barrier Induced by Long-Term Treatment with VEGF-A165 No Longer Depends on the Growth Factor's Presence

As responses of immortalized endothelial cells of the bovine retina (iBREC) to VEGF-A₁₆₅ depend on exposure time to the growth factor, we investigated changes evident after long-term treatment for nine days. The cell index of iBREC cultivated on gold electrodes—determined as a measure of permeability—was persistently reduced by exposure to the growth factor. Late after addition of VEGF-A₁₆₅ protein levels of claudin-1 and CD49e were significantly lower, those of CD29 significantly higher, and the plasmalemma vesicle associated protein was no longer detected. Nuclear levels of β-catenin were only elevated on day two. Extracellular levels of VEGF-A—measured by ELISA—were very low. Similar to the binding of the growth factor by brolucizumab, inhibition of VEGFR2 by tyrosine kinase inhibitors tivozanib or nintedanib led to complete, although transient, recovery of the low cell index when added early, though was inefficient when added three or six days later. Additional inhibition of other receptor tyrosine kinases by nintedanib was similarly unsuccessful, but additional blocking of c-kit by tivozanib led to sustained recovery of the low cell index, an effect observed only when the inhibitor was added early. From these data, we conclude that several days after the addition of VEGF-A₁₆₅ to iBREC, barrier dysfunction is mainly sustained by increased paracellular flow and impaired adhesion. Even more important, these changes are most likely no longer VEGF-A-controlled.

Blocking of VEGF-A is not sufficient to completely revert its long-term effects on the barrier formed by retinal endothelial cells

The VEGF-A-induced functional impairment of the barrier formed by retinal endothelial cells (REC) can be prevented and even - at least temporarily - reverted by trapping the growth factor in a complex with a VEGF-binding protein or by inhibiting the activity of the VEGF receptor 2 (VEGFR2). In an approach to emulate the clinically relevant situation of constant exposure to effectors, we investigated (1) whether prolonged exposure to VEGF-A₁₆₅ for up to six days results in a different type of disturbance of the barrier formed by immortalized bovine REC (iBREC) and (2) whether alterations of the barrier induced by VEGF-A₁₆₅ can indeed be sustainably reverted by subsequent treatment with the VEGF-A-binding proteins ranibizumab or brolucizumab. As a measure of barrier integrity, the cell index (CI) of iBREC cultivated on gold electrodes was monitored continuously. CI values declined shortly after addition of the growth factor and then remained low for more than six days over which considerable amounts of both extra- and intracellular VEGF-A were measured. Interestingly, the specific VEGFR2 inhibitor nintedanib normalized the lowered CI when added to iBREC pre-treated with VEGF-A₁₆₅ for one day, but failed to do so when cells had been exposed to the growth factor for six days. Expression of the tight junction (TJ) protein claudin-5 was unchanged early after addition of VEGF-A₁₆₅ but higher after prolonged treatment, whereas decreased amounts of the TJ-protein claudin-1 remained low, and increased expression of the plasmalemma vesicle-associated protein (PLVAP) remained high during further exposure. After two days, the characteristic even plasma membrane stainings of claudin-1 or claudin-5 appeared weaker or disordered, respectively. After six days the subcellular localization of claudin-5 was similar to that of control cells again, but claudin-1 remained relocated from the plasma membrane. To counteract these effects of VEGF-A₁₆₅, brolucizumab or ranibizumab was added after one day, resulting in recovery of the then lowered CI to normal values within a few hours. However, despite the VEGF antagonist being present, the CI declined again two days later to values that were just slightly higher than without VEGF inhibition during further assessment for several days. At this stage, neither the supernatants nor whole cell extracts from iBREC treated with VEGF-A₁₆₅ and its antagonists contained significant amounts of free VEGF-A. Treatment of VEGF-A₁₆₅-challenged iBREC with ranibizumab or brolucizumab normalized expression of claudin-1 and claudin-5, but not completely that of PLVAP. Interestingly, the characteristic VEGF-A₁₆₅-induced relocalization of claudin-1 from the plasma membrane was reverted within one day by any of the VEGF antagonists, but reappeared despite their presence after further exposure for several days. Taken together, barrier dysfunction induced by VEGF-A₁₆₅ results from deregulated para- and transcellular flow but the precise nature or magnitude of underlying changes on a molecular level clearly depend on the time of exposure, evolving into a stage of VEGF-A₁₆₅-independent barrier impairment. These findings also provide a plausible explanation for resistance to treatment with VEGF-A antagonists frequently observed in clinical practice.

Type of culture medium determines properties of cultivated retinal endothelial cells: induction of substantial phenotypic conversion by standard DMEM

Contradictory behavior of microvascular retinal endothelial cells (REC) - a reliable in vitro model to study retinal diseases - have recently been reported which might result from cultivating the cells in standard DMEM not optimized for this cell type. Therefore, we studied DMEM's effects on phenotype and behavior of immortalized bovine REC. Cells were cultivated in endothelial cell growth medium (ECGM) until a confluent monolayer was reached and then further kept for 1-4 days in ECGM, DMEM, or mixes thereof all supplemented with 5% fetal bovine serum, endothelial cell growth supplement, 90 μg/ml heparin, and 100 nM hydrocortisone. Within hours of cultivation in DMEM, the cell index - measured to assess the cell layer's barrier function - dropped to ~5% of the initial value and only slowly recovered, not only accompanied by stronger expression of HSP70 mRNA and secretion of interleukin-6, but also by lower expressions of tight junction proteins claudin-5, claudin-1 or of the marker of cell type conversion caveolin-1. Altered subcellular localizations of EC-typic claudin-5, vascular endothelial cadherin and von Willebrand factor were also observed. Taken together, all experiments with (retinal) EC cultivated in common DMEM need to be interpreted very cautiously and should at least include phenotypic validation.