Ranibizumab prevents Müller cell edema by decreasing VEGF-A in diabetic retinopathy

Background: Diabetic macular edema (DME) is the most common cause of vision loss in patients with diabetic retinopathy. The efficacy of anti-VEGF therapy has been well demonstrated and become the standard of care in the management of DME. The present study is to explore the possible mechanism(s) of ranibizumab in protecting Müller cells from cellular edema in experimental diabetic retinopathy. Methods: Sprague-Dawley rats were rendered diabetes with intraperitoneal injection of streptozotocin. Intravitreal injection of ranibizumab was performed 8 weeks after diabetes onset. Four weeks later, the rats were killed and the retinas were harvested for examination. rMC-1 cells (rat Müller cell line) were treated with glyoxal for 24 hours, with or without ranibizumab. Cell viability was detected with CCK-8 assay. The expressions of inwardly rectifying K + channel 4.1 (Kir4.1), aquaporin 4 (AQP4), Dystrophin 71 (Dp71), vascular endothelial growth factor A (VEGF-A), glutamine synthetase (GS) and sodium-potassium-ATPase (Na + -K + -ATPase) were examined with Western blot. VEGF-A in the supernatant of cell culture was detected with ELISA. The intracellular potassium and sodium levels were detected with specific indicators. Results: Compared to the normal control, the protein expressions of Kir4.1, AQP4 and Dp71 were down-regulated significantly in diabetic rat retinas, which were prevented by ranibizumab. The above changes were recapitulated in vitro . As compared with the control, the intracellular potassium level in glyoxal-treated rMC-1 cells was increased, while the intracellular sodium level and Na + -K + -ATPase protein level remained unchanged. However, ranibizumab treatment increased Na + -K + -ATPase protein expression and decreased intracellular sodium, but not potassium level. Conclusion: Ranibizumab protected Müller cells from intracellular edema through up-regulation of Kir4.1,

(2 μL) of normal saline was injected. Four weeks after the injection, the rats were sacrificed and the eyes were enucleated for the following study.

Cell viability assay
Cell viability of rMC-1 cells was measured using the Cell Counting Kit-8 (CCK-8) assay.
Briefly, the rMC-1 cells, incubated with different doses of glyoxal (0.1-5 mM), were seeded on 96-well plates at a density of 10 4 cells per well treated with or without ranibizumab (0.125 mg/mL) for 1 to 36 hours. The cells were washed with phosphate-buffered saline (PBS), and then incubated with serum-free low glucose DMEM containing 10% of CCK-8 for 3 hours at 37°C. The absorbance was measured at 450 nm by using a microplate spectrophotometer (Tecan, Crailsheim, Germany). The cell viability was expressed as the percentage of the untreated control, which was defined as 100% for each experiment.

RNA extraction and real-time PCR
Total RNA was extracted from rMC-1 cells. Reverse transcription was performed and realtime PCR was carried out by using SYBR Green Real-Time PCR master mix (Toybo, Osaka, Japan). The primers were designed using the software Primer Premier Version 5.0 and were ordered from Shanghai DNA Biotechnology Co. Ltd. (Shanghai, China). The primer information was listed in Table 1.

Measurement of intracellular sodium and potassium levels
Intracellular sodium and potassium concentration were detected with the specific indicator SBFI AM (sodium) and PBFI AM (potassium). The stock solution (1 mM) was reconstituted in DMSO, stored in dark at -20°C. The rMC-1 cells were first planted on 96-well plates and treated with glyoxal (1 mM) with or without ranibizumab (0.125 mg/mL) for 24 hours. After washed with PBS, the cells were incubated with SBFI AM or PBFI AM (diluted to the final concentration 10 μM) at 37°C for 3 hours, followed by a brief wash with PBS. Fluorescence was measured using a plate-reader (excitation = 340 nm, emission = 500 nm). The intensity of fluorescence was normalized by the cell number.

Statistical analysis
The results were expressed as mean ± SE. Statistical analysis was carried out with the SPSS software, version 22.0 (IBM Company, Armonk, NY, USA), and one-way ANOVA with Dunnett's test was used. The p value of 0.05 or less was considered statistically significant.

Müller cell intracellular edema was detected in diabetic rat retina, which was alleviated by ranibizumab
In order to evaluated Müller cell intracellular edema in vivo, we adopted the published method by using semithin sections of the retina [19]. As shown in Fig. 1, Compared to normal control group, the fluid accumulation was detected between nuclei of the outer nuclear layer as strip-like morphology, indicating Müller cell apical processes swollen or dilated. The edema of Müller cells was alleviated after ranibizumab treatment (Fig. 1C).

The expression of Kir4.1 was down-regulated in rat retina with diabetes progression
The examination of protein expression of Kir4.1 in diabetic rat retinas showed that, compared with the control, the Kir4.1 level in diabetic rat retinas was decreased by 21.0% in 6-week (n = 6, p > 0.05, Fig. 2A) and 46.7% in 12-week (n = 4, p < 0.05, Fig. 2B), respectively.
The decreased expression of Kir4.1 in 12-week diabetic rat retina was also confirmed with immunofluorescence. As shown in Fig. 2C, in normal control, Kir4.1 is mainly expressed in the inner limiting membrane (ILM) and co-localized with GS, a specific marker for Müller cells. However, in diabetic retinas, the distribution of Kir4.1 was largely disrupted, extending from ILM to the outer limiting membrane (OLM), with weak immunostaining especially in ILM and around retinal blood vessels.

Ranibizumab increased the expressions of Kir4.1 and AQP4 in diabetic rat retina
To test the effect of ranibizumab on the expressions of Kir4.1 and AQP4, western blot was performed in 12-week diabetic rat retinas treated with or without ranibizumab. As showed in Fig. 3A, after ranibizumab treatment, the protein level of Kir4.1 was up-regulated by 47.5% (n = 7, p < 0.05) compared with that in diabetic rat. Similarly, the protein level of AQP4 in diabetic group was decreased significantly by 43.3% (n = 7, p < 0.05) compared to that in normal control group, which was up-regulated by 30.9% (n = 7, p < 0.05) after ranibizumab treatment (Fig. 3B). The protein expressions of GS and GFAP in the Müller cells in diabetic retinas were also evaluated and the data showed that the GS expression in diabetic retinas was decreased by 23.7% (n = 7, p < 0.05, Fig. 3C), while GFAP was increased by 222.7% (n = 7, p < 0.05, Fig. 3D), as compared with the control, indicating the activation of Müller cells with decreased function in metabolizing glutamate. However, ranibizumab has no effect on the expressions of GS and GFAP.
To further confirm the effect of ranibizumab on Kir4.1 and GFAP, we performed double immunostaining of both proteins in diabetic rat retinas treated with or without ranibizumab. As shown in Fig. 3E, in normal control, Kir4.1 was mainly expressed in the ILM and around the vessels, which co-localized with GFAP, another marker of Müller cells.
However, in 12-week diabetic rat retinas, the decreased expression of Kir4.1 with its altered distribution was detected, attenuated staining pattern especially in ILM and around vessels. While GFAP immunostaining in Müller cells was increased in 12-week diabetic rat retinas with its characteristic radial immunostaining pattern. Ranibizumab treatment increased the expression of Kir4.1 as well as maintained its distribution to nearly normal level, but showed no effect on GFAP (Fig. 3E).

Ranibizumab decreased VEGF-A and increased the protein expressions of Kir4.1, AQP4 and Dp71 in glyoxal-treated rMC-1 cells
To further confirm above observation, we adopted glyoxal-treated rMC-1 cells to mimic diabetic condition. As shown in Fig. 4 Fig. 4C). The changes of Kir4.1 was also confirmed with WB, which showed that the protein level was decreased by 27.2% and 51.0%, separately, at 12 and 24 hours after glyoxal treatment (Fig. 4D).
To study the effect of ranibizumab on rMC-1 cells, glyoxal-treated rMC-1 cells were treated with or without ranibizumab and the changes of VEGF-A, Kir4.1, AQP4, Dp71 and GS were examined at both mRNA and protein levels. Although the cell viability was decreased in a time-dependent manner with glyoxal treatment (Fig. 4B), VEGF expression was increased at both 12 and 24 hours ( Fig. 5A and B). The mRNA level of VEGF-A was increased by 54.4% (n = 6, p < 0.05) and 26.4% (n = 6, p < 0.05) at 12 and 24 hours in glyoxal-treated group (Fig. 5A). VEGF-A protein level was increased by 44.4% (n = 3, p < 0.05) and 78.9% (n = 3, p < 0.05), separately, at the same time points (Fig. 5B). VEGF-A level in the supernatant of cell culture was decreased significantly after ranibizumab treatment (n = 4, p < 0.05, Fig. 5C).

To study whether the increased VEGF-A in glyoxal-treated rMC-1 cells could decrease
Kir4.1 expression, we treated rMC-1 cells with recombinant human VEGF-A (rh-VEGF-A). In  (Fig. 8B). Since ranibizumab has no effect on cell viability (Fig. 8C), the increased Kir4.1 by ranibizumab further confirmed the causal effect of VEGF-A on Kir4.1. We also detected the changes of AQP4 and Dp71 under the treatment of rh-VEGF-A and found no significant change for these 2 proteins (Data not shown).

Ranibizumab decreased intracellular osmotic pressure by sodium efflux
decreasing the osmotic pressure, we detected the intracellular potassium and sodium level with their corresponding indicators (PBFI and SBFI). After treatment with glyoxal (1mM) for 24 hours, the intracellular potassium level is increased significantly (n=10, p < 0.05), while the intracellular sodium level remained relatively unchanged (n=10, p > 0.05) compared with that in normal control group (Fig. 9). However, when treated with ranibizumab, intracellular sodium level, but not potassium, was decreased significantly (n=10, p < 0.05). This result indicated that, besides up-regulation of Kir4.1, decreasing intracellular osmotic pressure might be another mechanism for ranibizumab to prevent the cellular edema of Müller cells in DR. To further explore the possible reasons, we performed the western blot to detect the protein expression of Na + -K + -ATPase in glyoxal-treated rMC-1 cells with or without ranibizumab treatment. The data in Fig. 9C showed that, compared with that in normal control, the expression of Na + -K + -ATPase in glyoxal-treated group remained unchanged, while ranibizumab treatment increased the expression of Na + -K + -ATPase by 20.6% (n=4, p < 0.05). The detailed mechanisms need further exploration.

Discussion
Diabetic macular edema (DME) is the main cause of blindness in patients with DR [20].
Anti-VEGF therapy has been an effective treatment improving both microstructure and functions of retina in DME patients. Further studying the underlying mechanisms of anti-VEGF therapy on DME and exploring other effective treatments are of great importance. In this study, we found that increased VEGF-A, and decreased Kir4. The ranibizumab effect is independent of the gliotic state of Müller cells since ranibizumab showed no effect on both GS and GFAP expressions in Müller cells. Besides directly binding VEGF-A, ranibizumab could also decrease the intracellular sodium level to reduce the osmotic pressure, consequently preventing the cellular edema.
The pathogenesis of DME is complex. Breakdown of inner BRB, and dysfunction of Müller cells and RPE were all involved in the pathogenesis of DME. Müller cell, like a pump, drains the ion and water into vitreous body and retinal blood vessels with normal distribution and function of Kir4.1 and AQP4. It is reported that the distribution of Kir4.1 is altered 6-month diabetic rat retina, which is globally decreased especially in the OLM and around blood vessels [21]. Other study found Kir4.1 is absent in the perivascular areas and in ILM in 3month diabetic rats [22]. However, most studies focused on the distributions of these channels with immunofluorescence, the protein expression levels were rarely reported in DR. In this study, we found the protein levels of Kir4.1 and AQP4 were decreased Ranibizumab is a recombinant, humanized neutralizing antibody fragment, directly binding all isoforms of VEGF-A. It was verified as a powerful treatment to decrease macular edema in many clinical trials [24]. Besides its binding VEGF-A, the detailed mechanisms clearing the accumulated fluid were rarely reported. We hypothesized that, except for its effect on BRB, anti-VEGF reagents might enhance the "pumping" ability of Müller cells to transport water and ion out of retina via retinal vasculature, maintaining the homeostasis of the retina. In this study, we found ranibizumab, through binding VEGF-A, protected Müller cells from edema by up-regulating Kir4.1, AQP4 and Dp71.
Ion accumulation is considered as the initial step for intracellular edema. The downregulation of Kir4.1 could weaken efflux of potassium, causing potassium accumulation and increasing intracellular osmotic pressure. Water driven by osmotic pressure entered Müller cells through AQP4, leading to cell swelling. We found that the expression of Kir4.1, not AQP4 or Dp71, was decreased by rh-VEGF (Data not shown), indicating that the downregulation of AQP4 and Dp71 in diabetic retinopathy might be regulated by other factors, but not VEGF.
What beyond our expectation is that the intracellular potassium level was not decreased even though Kir4.1 was up-regulated by ranibizumab. However, the intracellular sodium level was decreased significantly by ranibizumab. The possible explanation is that the activity of Na + -K + -ATPase was inhibited by glyoxal, which could not be reversed by ranibizumab. In this study, although we did not detect the activity of Na + -K + -ATPase, we did find that ranibizumab increased the protein expression of Na + -K + -ATPase, which might partially interpret the decrease of intracellular sodium level by ranibizumab. The detailed mechanism for ranibizumab to decrease intracellular sodium level needs further study.
Although anti-VEGF therapy is effective to treat DME, a weak correlation was reported between gain of visual acuity and the anatomical improvement [25][26][27][28]. The loss of retinal neurons, especially cones, could result in the decreased visual acuity, which cannot be improved even after the anatomical recovery. Further the loss of retinal neurons might induce the gliotic reaction of Müller cells with overexpression of GFAP and downregulation of GS [29,21]. In our study, cell viability and expression of GFAP and GS is not influenced by ranibizumab in vivo and in vitro, indicating that ranibizumab has no effect on the gliotic reaction. Thus, it is of importance to develop combo-therapy to treat DME, e.g., to

Consent for publication
Written informed consent for publication was obtained from all participants.

Availability of data and material
All the data supporting our findings are provided in the manuscript.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.
ARRIVE Guidelines .pdf