Sitagliptin and the Blood-Retina Barrier: Effects on Retinal Endothelial Cells Manifested Only after Prolonged Exposure

Abstract

Inhibitors of dipeptidyl peptidase-4 (DPP-4) are widely used to treat diabetes mellitus, but data concerning their effects on the barrier stability of retinal endothelial cells (REC) in vivo and in vitro are inconsistent. Therefore, we studied whether the barrier properties of immortalized endothelial cells of the bovine retina (iBREC) were affected by the inhibitors of DPP-4 sitagliptin (10-1000 nM) and diprotin A (1-25 μM). Their effects were also investigated in the presence of VEGF-A₁₆₅ because diabetic patients often develop macular edema caused by VEGF-A-induced permeability of REC. To detect even transient or subtle changes of paracellular and transcellular flow as well as adhesion of the cells to the extracellular matrix, we continuously monitored the cell index (CI) of confluent iBREC grown on gold electrodes. Initially, the CI remained stable but started to decline significantly and persistently at 40 h or 55 h after addition of sitagliptin or diprotin A, respectively. Both inhibitors did not modulate, prevent, or revert the persistent VEGF-A₁₆₅-induced reduction of the CI. Interestingly, sitagliptin and diprotin A increased the expression of the tight-junction protein claudin-1 which is an important component of a functional barrier formed by iBREC. In contrast, expressions of CD29-a subunit of the fibronectin receptor-or of the tetraspanin CD9 were lower after extended treatment with the DPP-4 inhibitors; less of the CD9 was seen at the plasma membrane after prolonged exposure to sitagliptin. Because both associated proteins are important for adhesion of iBREC to the extracellular matrix, the observed low CI might be caused by weakened attachment of the cells. From our results, we conclude that extended inhibition of DPP-4 destabilizes the barrier formed by microvascular REC and that DPP-4 inhibitors like sitagliptin do not counteract or enhance a VEGF-A₁₆₅-induced barrier dysfunction as frequently observed in DME.

Figure 1

Prolonged treatment with sitagliptin only weakly changed the TEER of unchallenged iBREC. (a) Structures of DPP-4 inhibitors used. (b) Confluent iBREC grown on porous membrane inserts were exposed to 10 or 1000 nM sitagliptin for up to three days. The transendothelial electrical resistance (TEER), determined as a measure of permeability at indicated time points, was only weakly and inconsistently reduced by the inhibitor. TEER values, normalized in relation to those measured immediately before addition of sitagliptin, are shown as means and standard deviations of data from at least four replicates. Statistical analyses were performed as described in Materials and Methods. ^∗∗p < 0.01 compared to control.

Figure 2

Treatment with sitagliptin reduced the cell index of unchallenged iBREC. Cells were cultivated on gold electrodes until confluency was reached and exposed to sitagliptin over three days. The cell index (CI) was determined continuously as a measure of barrier function. Sitagliptin (10-1000 nM) resulted in a persistent, concentration-dependent CI decline starting six to forty hours after addition. (a) CI values, normalized in relation to those measured immediately before addition of sitagliptin, are shown as means and standard deviations of data from at least five wells. (b) Statistical analyses of data gained at indicated time points after addition of sitagliptin were performed as described in Materials and Methods. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001 compared to control.

Figure 3

Sitagliptin did not modulate the barrier dysfunction induced by VEGF-A. Confluent monolayers of iBREC cultivated on gold electrodes were exposed to 10 or 1000 nM sitagliptin in combination with VEGF-A₁₆₅ over two days. The VEGF-A₁₆₅-induced decline of the CI was not affected by the DPP-4 inhibitor. Normalized CI values are presented as means and standard deviations from at least four wells. Normalization and statistical analyses of data recorded at 48 h after addition of sitagliptin and VEGF-A₁₆₅ were performed as described in Materials and Methods. ^∗∗∗p < 0.001 compared to control.

Figure 4

Tivozanib but not sitagliptin reverted the VEGF-A-induced barrier dysfunction of iBREC. Confluent monolayers of iBREC cultivated on gold electrodes were exposed to 50 ng/ml VEGF-A₁₆₅ (t = 0 h) for one day before (a, c) 10 nM tivozanib or (b, c) 10-1000 nM sitagliptin was added (t ≈ 24 h). The cell index (CI) was determined continuously as a measure of barrier function. In all experiments, CI values—presented as means and standard deviations from at least six wells—were normalized in relation to those measured immediately before addition of VEGF-A₁₆₅ (t = 0 h). Statistical analyses of data gained at indicated time points after addition of VEGF-A₁₆₅ were performed as described in Materials and Methods. (a, c) Inhibition of the VEGF receptor 2 completely reverted the VEGF-A₁₆₅-induced CI decrease. (b, c) Treatment with sitagliptin did not result even in partial reversion of the VEGF-A₁₆₅-caused CI reduction.

Figure 5

Prolonged treatment with diprotin A affected the barrier function of unchallenged iBREC. The CI was determined continuously during exposure of confluent iBREC to 1-25 μM diprotin A. Similar to sitagliptin, this inhibitor of DPP-4 also induced a significant decline of the CI although this change was further delayed, starting about 55 h after its addition. (a) CI values, normalized in relation to those measured immediately before addition of diprotin A, are shown as means and standard deviations of data from at least five wells. (b) Statistical analyses of data from indicated time points after addition of diprotin A were performed as described in Materials and Methods. ^∗∗p < 0.01 and ^∗∗∗p < 0.001 compared to control.

Figure 6

Diprotin A did not enhance the detrimental effect of VEGF-A on the iBREC barrier. Confluent iBREC were exposed to 1-25 μM diprotin A and VEGF-A₁₆₅, and the CI was determined continuously. Diprotin A did not influence the CI decrease caused by VEGF-A₁₆₅. Normalized CI values are presented as means and standard deviations from at least six wells. Normalization and statistical analyses of data recorded 72 h after addition of diprotin A and VEGF-A₁₆₅ were performed as described in Materials and Methods. ^∗∗∗∗p < 0.0001 compared to control.

Figure 7

Prolonged treatment of iBREC with sitagliptin increased expression of claudin-1 and changed the plasma membrane localization of VEcadherin. (a) After exposure of confluent iBREC to 10-1000 nM sitagliptin for two days, cells were harvested for preparation of cell extracts, followed by Western blot analyses. Expression of claudin-1 was higher after treatment with sitagliptin at all concentrations used, but only 1 μM sitagliptin led to a significantly higher level of claudin-5; expression of VEcadherin was not changed. Signals were normalized as described in Materials and Methods; n ≥ 8 for each condition. (b) Immunofluorescence staining showed that plasma membrane-localized claudin-5 was not affected by the inhibitor of DPP-4, but staining of VEcadherin at the plasma membrane appeared more diffuse when cells had been exposed to sitagliptin. Scale bar: 10 μm.

Figure 8

Prolonged treatment of iBREC with diprotin A increased expression of TJ protein claudin-1. After exposure of confluent iBREC to (a, b) 1-25 μM diprotin A for three days or (b) 10-1000 nM sitagliptin for two days, cells were harvested for preparation of cell extracts, followed by Western blot analyses. (a) Expression of claudin-1 was significantly higher after treatment with 10 and 25 μM diprotin A, but those of claudin-5 or VEcadherin were not changed. (b) Expression of caveolin-1 was also not affected by treatment with sitagliptin or diprotin A. Signals were normalized as described in Materials and Methods; n ≥ 6 for each condition.

Figure 9

Prolonged treatment of iBREC with sitagliptin decreased expression of CD29 and changed the amount of plasma membrane-localized CD9. (a) Proteins were immunoprecipitated from whole cell extracts with antibodies specific for CD9 or CD29 and analyzed by Western blot with an antibody binding to CD9. Precipitates obtained with both antibodies (αCD9 and αCD29) contained the tetraspanin CD9 but not samples derived from precipitation with an isotype-matched control antibody, indicating that CD9 and CD29 are present in the same protein complex. (b) Confluent iBREC exposed to 10-1000 nM sitagliptin for two days were harvested for preparation of cell extracts followed by Western blot analyses. Expression of CD29 and CD9 was lower after treatment with sitagliptin, although the differences were statistically significant only for CD29. Signals were normalized as described in Materials and Methods. (c) Prominent CD9-specific staining was observed at the plasma membrane of control cells (yellow arrows). This staining was more diffuse and less intense (yellow arrowheads) after treatment of the cells with the DPP-4 inhibitor for two days. Scale bar: 10 μm.

Figure 10

Prolonged treatment of iBREC with diprotin A decreased expression of CD9. After exposure of confluent iBREC to 1-25 μM diprotin A for three days, cells were harvested for preparation of cell extracts to be analyzed by Western blot. Expression of CD9 was significantly lower after treatment with diprotin A. Signals were normalized as described in Materials and Methods; n ≥ 5 for each condition.