Differential Effects of Cytopathic Hypoxia on Human Retinal Endothelial Cellular Behavior: Implication for Ischemic Retinopathies

Loss of barrier integrity of retinal endothelial cells (RECs) is an early feature of ischemic retinopathies (IRs), but the triggering mechanisms remain incompletely understood. Previous studies have reported mitochondrial dysfunction in several forms of IRs, which creates a cytopathic hypoxic environment where cells cannot use oxygen for energy production. Nonetheless, the contribution of cytopathic hypoxia to the REC barrier failure has not been fully explored. In this study, we dissect in-depth the role of cytopathic hypoxia in impairing the barrier function of REC. We employed the electric cell-substrate impedance sensing (ECIS) technology to monitor in real-time the impedance (Z) and hence the barrier functionality of human RECs (HRECs) under cytopathic hypoxia-inducing agent, Cobalt(II) chloride (CoCl2). Furthermore, data were deconvoluted to test the effect of cytopathic hypoxia on the three key components of barrier integrity; Rb (paracellular resistance between HRECs), α (basolateral adhesion between HRECs and the extracellular matrix), and Cm (HREC membrane capacitance). Our results showed that CoCl2 decreased the Z of HRECs dose-dependently. Specifically, the Rb parameter of the HREC barrier was the parameter that declined first and most significantly by the cytopathic hypoxia-inducing agent and in a dose-dependent manner. When Rb began to fall to its minimum, other parameters of the HREC barrier, including α and Cm, were unaffected. Interestingly, the compromised effect of cytopathic hypoxia on Rb was associated with mitochondrial dysfunction but not with cytotoxicity. In conclusion, our results demonstrate distinguishable dielectric properties of HRECs under cytopathic hypoxia in which the paracellular junction between adjacent HRECs is the most vulnerable target. Such selective behavior could be utilized to screen agents or genes that maintain and strengthen the assembly of HRECs tight junction complex.


Introduction
The retina is a multi-layered structure that lines the back of the eyeball of most vertebrates. Because of the high metabolic activity of the retina, the retina has two independent vascular systems. These systems are the retinal vasculature networks, which supply the inner two-thirds of the retina, and the choroidal system, which supplies the outer onethird of the retina [1]. Retinal Endothelial Cells (RECs) form a simple layer of squamous cells lining the inner surface of retinal vasculature networks, where they form the inner blood-retinal barrier (iBRB) [1]. The integrity of the REC layer is critical for regulating substances that can pass through iBRB as well as for preventing harmful chemicals and plasma components from entering the retina [2]. When the integrity of the REC layer is compromised, the retinal functionality is disrupted, resulting in various retinal blinding diseases, including ischemic retinopathies (IRs) such as retinopathy of prematurity (ROP), diabetic retinopathy (DR), and neovascular age-related macular degeneration (AMD) [2].
IRs are visual diseases defined by ischemia in the first phase, followed by aberrant neovascularization in the second phase, leading to retinal detachment and blindness [3,4]. Despite the benefits of photocoagulation [5] and anti-VEGF therapies [6][7][8] in treating the neovascularization phase, there is still a need to find new targets that prevent the early stages of these ocular diseases [9][10][11]. In IRs, retinal ischemia is caused by damage to mature retinal vessels (as in many cases of diabetic retinopathy, retinal vein occlusion, or sickle cell retinopathy) or immature retinal vasculature (as in retinopathy of prematurity) [3]. However, the triggering mechanism of REC damage that causes retinal ischemia remains incompletely understood. Previous studies have reported mitochondrial dysfunction in several forms of IRs [12][13][14], suggesting that malfunctioned mitochondria damage REC by creating a cytopathic hypoxic environment, where cells are unable to use oxygen for energy production. Nonetheless, the contribution of cytopathic hypoxia to REC barrier failure has not been fully explored.
The hypoxia mimicking agent, cobalt chloride (CoCl 2 ), is frequently utilized to study a variety of ischemic diseases, including ischemic brain injury [15] and retinal ischemia [16]. CoCl 2 induces cytopathic hypoxia by its competitive inhibitory effect on multiple iron-dependent proteins linked with the mitochondrial electron transport chain (ETC) [17,18]. Furthermore, CoCl 2 inhibits prolyl hydroxylases (PHDs) that regulate the stability of hypoxia-inducible factors (HIFs), master regulators of cellular hypoxic response [19]. Accordingly, the use of CoCl 2 provides a suitable model to dissect the role of cytopathic hypoxia in compromising the REC barrier integrity.
Given the importance of preserving the REC of iBRB to the retinal functionality [20], the ability to dynamically evaluate how REC barrier function changes when subjected to cytopathic hypoxia would be a valuable tool. We have shown the usefulness of the Electric Cell-substrate Impedance Sensing (ECIS) system in real-time monitoring the effect of cytopathic hypoxia on the barrier integrity of retinal pigment epithelium an essential component of the outer BRB [21]. ECIS system is a biosensor multitasking system that continuously monitors changes in cells' behavior and models many important parameters that describe cellular barrier integrity [22]. By using alternating current (AC) instead of direct current (DC), ECIS has the capability of dissecting the components of RECs' total impedance (Z) into barrier resistance (R) and capacitance (C), which are frequently used to monitor barrier functionality and cellular spreading over the substrate, respectively. Another important feature of ECIS is the inclusion of multifrequency measurement along with the usage of AC [23]. This feature allows the mathematical modeling of the overall cellular measurements into three distinct parameters related to barrier integrity ( Figure 1); R b (paracellular resistance between RECs), α (basolateral resistance between RECs and the extracellular matrix), and C m (REC membrane capacitance). As a result, the application of ECIS Technology presents a valuable platform to non-invasively analyze the barrier functionality of RECs under ischemic conditions.
The goals of the current study were to determine the effect of cytopathic hypoxia on the behavior of RECs and to evaluate the quality of endothelial paracellular junctions, endothelial cell membrane, as well as the adhesion between endothelial cells and extracellular matrix under the condition of cytopathic hypoxia.

Effect of Cytopathic Hypoxia on HREC Electrical Impedance
Given that IRs are related to retinal endothelial dysfunction, the role of cytopat hypoxia in impairing HREC barrier functionality was investigated in a real-time mann using ECIS ® instrumentation. In this experiment (Figure 2), various concentrations (0, 100, and 1000 µM) of cytopathic hypoxia-inducing agent (CoCl2) were applied to HRE after the impedance (Z) reached the plateau phase, where HRECs form stable and conf ent monolayer with mature tight junctions. Then the barrier integrity of HRECs was ev uated based on total Z over 25-h across a frequency range of 250 to 64,000 Hz. As indica in Figure 2A-D, CoCl2-treatment resulted in a dose-dependent reduction in overall Z HRECs across all measured frequencies, implicating an important role of cytopathic h poxia in impairing the barrier functionality of HRECs.
Next, since Z of cells consists of cell barrier resistance (R) and cell membrane capa tance (C), we determined whether cytopathic hypoxia affects one or both components Z. Generally when cells are subjected to alternating current (AC), both R and C are gen ated with Z as an endpoint. However, when cells are subjected to direct current (DC), C vanishes, and R is equivalent to Z. Taking advantage of ECIS as an AC applying syst with frequency-dependent amplitudes, simultaneous measurements of R and C were o tained at each specific frequency [22]. To determine the best fit frequency to be utilized subsequent evaluations of cytopathic hypoxia effects on Z components, the frequency d pendency spectra of Z components were first determined. Figure 3A,C,E illustrate the f quency dependence spectra of the Z, R, and C, respectively, for HRECs at the time (t 44.8 h after placing HRECs onto the ECIS array and just before starting any treatment. this time, HRECs were confluent, and the Z spectrum exhibited a distinctive frequency 16,000 Hz, where the maximum Z ratio was obtained between wells with HRECs vers those without cells ( Figure 3A). In addition, this frequency gives the broadest range comparing different post-CoCl2-treatment groups ( Figure 3B).

Effect of Cytopathic Hypoxia on HREC Electrical Impedance
Given that IRs are related to retinal endothelial dysfunction, the role of cytopathic hypoxia in impairing HREC barrier functionality was investigated in a real-time manner using ECIS ® instrumentation. In this experiment (Figure 2), various concentrations (0, 10, 100, and 1000 µM) of cytopathic hypoxia-inducing agent (CoCl 2 ) were applied to HRECs after the impedance (Z) reached the plateau phase, where HRECs form stable and confluent monolayer with mature tight junctions. Then the barrier integrity of HRECs was evaluated based on total Z over 25-h across a frequency range of 250 to 64,000 Hz. As indicated in Figure 2A-D, CoCl 2 -treatment resulted in a dose-dependent reduction in overall Z of HRECs across all measured frequencies, implicating an important role of cytopathic hypoxia in impairing the barrier functionality of HRECs.
Next, since Z of cells consists of cell barrier resistance (R) and cell membrane capacitance (C), we determined whether cytopathic hypoxia affects one or both components of Z. Generally when cells are subjected to alternating current (AC), both R and C are generated with Z as an endpoint. However, when cells are subjected to direct current (DC), the C vanishes, and R is equivalent to Z. Taking advantage of ECIS as an AC applying system with frequencydependent amplitudes, simultaneous measurements of R and C were obtained at each specific frequency [22]. To determine the best fit frequency to be utilized in subsequent evaluations of cytopathic hypoxia effects on Z components, the frequency dependency spectra of Z components were first determined. Figure 3A,C,E illustrate the frequency dependence spectra of the Z, R, and C, respectively, for HRECs at the time (t) = 44.8 h after placing HRECs onto the ECIS array and just before starting any treatment. At this time, HRECs were confluent, and the Z spectrum exhibited a distinctive frequency of 16,000 Hz, where the maximum Z ratio was obtained between wells with HRECs versus those without cells ( Figure 3A). In addition, this frequency gives the broadest range for comparing different post-CoCl 2 -treatment groups ( Figure 3B). Three-dimensional plots of the log of normalized impedance (Z) across HRECs against time and the log of the alternating-current (AC) frequency applied to the ECIS electrode. The control vehicle or different CoCl2 treatment were added at the time (t) = 0, which was 44.8 h after culturing HRECs on the ECIS electrode. After treatment, Zt was tracked for 25 h, where Z0 at t = 0 was used to normalize all other Z measurements, such that each reading was calculated as the log of the ratio of Zt/Z0. The maximum value is when Zt = Z0 (hence the ratio becomes 1, and the log equals 0). Abbreviations: Z: impedance; Norm: normalized; Freq: frequency; Zt: the impedance at time t; Z0: the impedance at time 0.
Regarding the frequency dependence spectrum of the R ( Figure 3C), it showed that at a frequency of 4000 Hz, a maximum R ratio between wells with HRECs versus those without cells was obtained, providing the broadest possible range for the comparison between post-CoCl2-treatment groups ( Figure 3D). Therefore, this frequency was chosen for subsequent R analysis. Lastly, the frequency dependence spectrum of C ( Figure 3F) showed a frequency of 64,000 Hz to give a broadest possible range to compare the C among different CoCl2-treatment groups and thereby was utilized in the subsequent C analysis. Three-dimensional plots of the log of normalized impedance (Z) across HRECs against time and the log of the alternating-current (AC) frequency applied to the ECIS electrode. The control vehicle or different CoCl 2 treatment were added at the time (t) = 0, which was 44.8 h after culturing HRECs on the ECIS electrode. After treatment, Zt was tracked for 25 h, where Z0 at t = 0 was used to normalize all other Z measurements, such that each reading was calculated as the log of the ratio of Zt/Z0. The maximum value is when Zt = Z0 (hence the ratio becomes 1, and the log equals 0). Abbreviations: Z: impedance; Norm: normalized; Freq: frequency; Zt: the impedance at time t; Z0: the impedance at time 0.
Regarding the frequency dependence spectrum of the R ( Figure 3C), it showed that at a frequency of 4000 Hz, a maximum R ratio between wells with HRECs versus those without cells was obtained, providing the broadest possible range for the comparison between post-CoCl 2 -treatment groups ( Figure 3D). Therefore, this frequency was chosen for subsequent R analysis. Lastly, the frequency dependence spectrum of C ( Figure 3F) showed a frequency of 64,000 Hz to give a broadest possible range to compare the C among different CoCl 2 -treatment groups and thereby was utilized in the subsequent C analysis.

Effect of Cytopathic Hypoxia on the Capacitance of HRECs
Given that the spreading of cells on substrates impacts their behavior, the ability of HRECs to spread over ECIS electrodes under cytopathic hypoxia was also evaluated. In this regard, the C of HRECs was measured because of the inverse relationship between the C and cells' spreading [23]. To do so, the frequency at which the maximum spreading of HRECs could be detected was chosen to be the one that gives the lowest C, which corresponds to 64,000 Hz in Figure 3E. This inverse relationship was confirmed in Figure 4A, where the C of HRECs at 64,000 Hz exhibited a decreasing tendency until it reached a plateau after 10-15 h of placing HRECs on ECIS electrode, reflecting a fully spread stage of HRECs.
Then the media was changed into serum-free media to prevent cell proliferation within the experimental period. After that, different concentrations of CoCl 2 were applied. As shown in Figure 4A, the C of HRECs increased instantly after CoCl 2 (1000 µM) treatment, whereas it took 8-12 h for the C of HRECs to rise in response to 100 µM of CoCl 2 . In contrast, CoCl 2 at 10 µM had no significant effect on the C of HRECs compared to the controls. Notably, 100 µM and 1000 µM of CoCl 2 resulted in significant increases in the C of HRECs both at the end (t = 25 h post-treatment, Figure 4B) and throughout the experiment (evaluated by the AUC, Figure 4C) in a dose-response manner. These data collectively demonstrate that only higher levels of cytopathic hypoxia affect how HRECs spread over the substrate. nt. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 5 of 1 Figure 3. Impedance, resistance, and capacitance ratios between HRECs and the cell-free medium against different frequencies were measured at 44.8 h after placing HRECs on ECIS electrodes and before adding CoCl2 (A,C,E, respectively) or at 17 h post-CoCl2 treatment , ranging from 0 to 1000 µM (B,D,F, respectively) measured at 62 h after placing HRECs on ECIS electrodes. Local maxi mums for impedance, resistance, and capacitance were at 16,000, 4000, and 64,000 Hz, respectively

Effect of Cytopathic Hypoxia on the Capacitance of HRECs
Given that the spreading of cells on substrates impacts their behavior, the ability o HRECs to spread over ECIS electrodes under cytopathic hypoxia was also evaluated. In this regard, the C of HRECs was measured because of the inverse relationship between the C and cells' spreading [23]. To do so, the frequency at which the maximum spreading of HRECs could be detected was chosen to be the one that gives the lowest C, which cor responds to 64,000 Hz in Figure 3E. This inverse relationship was confirmed in Figure 4A where the C of HRECs at 64,000 Hz exhibited a decreasing tendency until it reached a plateau after 10-15 h of placing HRECs on ECIS electrode, reflecting a fully spread stage of HRECs. Then the media was changed into serum-free media to prevent cell prolifera tion within the experimental period. After that, different concentrations of CoCl2 were ap plied. As shown in Figure 4A, the C of HRECs increased instantly after CoCl2 (1000 µM treatment, whereas it took 8-12 h for the C of HRECs to rise in response to 100 µM o CoCl2. In contrast, CoCl2 at 10 µM had no significant effect on the C of HRECs compared to the controls. Notably, 100 µM and 1000 µM of CoCl2 resulted in significant increases in the C of HRECs both at the end (t = 25 h post-treatment, Figure 4B) and throughout the experiment (evaluated by the AUC, Figure 4C) in a dose-response manner. These data . Impedance, resistance, and capacitance ratios between HRECs and the cell-free medium against different frequencies were measured at 44.8 h after placing HRECs on ECIS electrodes and before adding CoCl 2 (A,C,E, respectively) or at 17 h post-CoCl 2 treatment, ranging from 0 to 1000 µM (B,D,F, respectively) measured at 62 h after placing HRECs on ECIS electrodes. Local maximums for impedance, resistance, and capacitance were at 16,000, 4000, and 64,000 Hz, respectively.

Effect of Cytopathic Hypoxia on the Total Resistance of HRECs
To analyze the effect of cytopathic hypoxia on the barrier function of HRECs, the R parameter of the Z across the cell monolayer was measured at 4000 Hz, a frequency at which the R is maximum ( Figure 3C). As shown in Figure 5A, treating HRECs with CoCl 2 resulted in declines of R over time in a dose-dependent fashion. The decrease in R began with a 1000 µM concentration of CoCl 2 as it quickly reached a minimum level within 10 h of the treatment. The following change in the R was observed with 100 µM of CoCl 2 treatment, evident by a gradual and continuous drop in the R curve until it reached its minimum at 25 h post-treatment. The 10 µM CoCl 2 group was the last to have considerable losses in R compared to the control group. Interestingly, all concentrations of CoCl 2 showed significant reductions in R compared to the control group at the end of the experiment ( Figure 5B) as well as throughout the experiment ( Figure 5C) in a dose-dependent fashion, where each treatment group experienced R losses at rates significantly different from each other and from the control group. These results indicate that cytopathic hypoxia, at all concentrations, compromises the barrier integrity of HRECs.

Effect of Cytopathic Hypoxia on the Total Resistance of HRECs
To analyze the effect of cytopathic hypoxia on the barrier function of HRECs, the parameter of the Z across the cell monolayer was measured at 4000 Hz, a frequency which the R is maximum ( Figure 3C). As shown in Figure 5A, treating HRECs with CoC resulted in declines of R over time in a dose-dependent fashion. The decrease in R bega with a 1000 µM concentration of CoCl2 as it quickly reached a minimum level within 10 of the treatment. The following change in the R was observed with 100 µM of CoCl2 trea ment, evident by a gradual and continuous drop in the R curve until it reached its min mum at 25 h post-treatment. The 10 µM CoCl2 group was the last to have considerab losses in R compared to the control group. Interestingly, all concentrations of CoC showed significant reductions in R compared to the control group at the end of the expe iment ( Figure 5B) as well as throughout the experiment ( Figure 5C) in a dose-dependen fashion, where each treatment group experienced R losses at rates significantly differe from each other and from the control group. These results indicate that cytopathic h poxia, at all concentrations, compromises the barrier integrity of HRECs.

Effects of Cytopathic Hypoxia on Components of Transendothelial Resistance of HRECs
Because the transendothelial resistance comprises three components: R b , α, and C m , we sought to determine whether cytopathic hypoxia has global or differential effects on these components in HRECs. To achieve this goal, the obtained data at 4000 Hz were first deconvoluted into three separate curves representing these components using the mathematical model of Giaever and Keese [22]. As shown in Figure 6A, adhesion of HRECs to their substrate (represented by α) and stable C m were achieved first at~2 h, followed by spreading of HRECs out and monolayer formation by~7 h, evident by reaching a plateau in the C curve of HRECs ( Figure 6B). R b values did not start to model until~5 h after HRECs were placed on ECIS electrodes and peaked at~10-12 h ( Figure 6A). These data demonstrate that the establishment of α, C m , and monolayer confluency for this specific cell type is necessary to achieve durable R b and a mature barrier function.
After dissecting transendothelial resistance of HRECs into its three components, the effect of cytopathic hypoxia on each of these components was then evaluated (Figure 7). Firstly, α curves in Figure 7A for 1000, 100, and 10 µM of CoCl 2 were terminated at (0-1 h), (10-15 h), and (20-25 h), respectively, which are time periods when the corresponding R b value becomes zero ( Figure 7C). ECIS can only model α values if the corresponding R b values are positive (not zero). During these experimental periods and before α curves ended, none of the CoCl 2 concentrations had a significant effect on α values ( Figure 7B). Secondly, Figure 7C shows the effect of cytopathic hypoxia on normalized R b over time. While all concentrations of CoCl 2 completely diminished the contribution of R b to cell resistance at the end of the experiment, there was no noticeable dose-response effect ( Figure 7D). However, monitoring the impact of CoCl 2 on R b behavior throughout the experiment with the calculation of AUC, dose-dependent reductions in R b were detected ( Figure 7E). Thirdly, Figure 7F displays C m over time after treatment with CoCl 2 , and again, we noticed that C m curves for 1000, 100, and 10 µM of CoCl 2 were ended abruptly at (0-1 h), (10-15 h), and (20-25 h), respectively. These terminations of C m curves were due to the fact that ECIS cannot estimate actual values for C m when the corresponding R b values are zero. Considering the AUC of C m across the aforementioned time intervals and before C m curves ended, none of the CoCl 2 concentrations significantly affected C m values ( Figure 7G). Collectively, the results in Figure 7 indicate that cytopathic hypoxia mainly affects the integrity of the paracellular junctions between HRECs as the R b was the only component of the three to respond in a dose-dependent manner.

Effects of Cytopathic Hypoxia on Components of Transendothelial Resistance of HRECs
Because the transendothelial resistance comprises three components: Rb, α, and we sought to determine whether cytopathic hypoxia has global or differential effects these components in HRECs. To achieve this goal, the obtained data at 4000 Hz were f deconvoluted into three separate curves representing these components using the mat matical model of Giaever and Keese [22]. As shown in Figure 6A, adhesion of HREC their substrate (represented by α) and stable Cm were achieved first at ~2 h, followed spreading of HRECs out and monolayer formation by ~7 h, evident by reaching a plat  After dissecting transendothelial resistance of HRECs into its three components, the effect of cytopathic hypoxia on each of these components was then evaluated (Figure 7). Firstly, α curves in Figure 7A for 1000, 100, and 10 µM of CoCl2 were terminated at (0-1 h), (10-15 h), and (20-25 h), respectively, which are time periods when the corresponding Rb value becomes zero ( Figure 7C). ECIS can only model α values if the corresponding Rb values are positive (not zero). During these experimental periods and before α curves ended, none of the CoCl2 concentrations had a significant effect on α values ( Figure 7B). Secondly, Figure 7C shows the effect of cytopathic hypoxia on normalized Rb over time. While all concentrations of CoCl2 completely diminished the contribution of Rb to cell resistance at the end of the experiment, there was no noticeable dose-response effect ( Figure  7D). However, monitoring the impact of CoCl2 on Rb behavior throughout the experiment with the calculation of AUC, dose-dependent reductions in Rb were detected ( Figure 7E). Thirdly, Figure 7F displays Cm over time after treatment with CoCl2, and again, we noticed that Cm curves for 1000, 100, and 10 µM of CoCl2 were ended abruptly at (0-1 h), (10-15 h), and (20-25 h), respectively. These terminations of Cm curves were due to the fact that ECIS cannot estimate actual values for Cm when the corresponding Rb values are zero. Considering the AUC of Cm across the aforementioned time intervals and before Cm curves ended, none of the CoCl2 concentrations significantly affected Cm values ( Figure 7G). Collectively, the results in Figure 7 indicate that cytopathic hypoxia mainly affects the integrity of the paracellular junctions between HRECs as the Rb was the only component of the three to respond in a dose-dependent manner.   (F) Normalized C m vs. time curves. The ECIS is unable to calculate a real value for C m at time points when R b is at or below zero. (G) Areas under the normalized C m curves for the intervals when the C m value can be modeled in each group. CoCl 2 treatment is associated with no C m changes from control in each group along the interval for each their C m values can be calculated (t = 0-20 h for the 10 µM, t = 0-8 h for the 100 µM, and t = 0-1 h for the 1000 µM). Statistical analysis was performed using the ANOVA test followed by the Tukey post hoc test. Abbreviations: Norm: normalized; AUC: area under the curve; ns: no significance; *: p ≤ 0.05; ****: p ≤ 0.0001; n = 5-6/group.

CoCl 2 Compromises HRECs' Mitochondrial Bioenergetics without Causing Cytotoxicity
To verify that the adverse effects of CoCl 2 on HREC barrier function were associated with mitochondrial dysfunction, cellular oxygen consumption rates (OCRs) were analyzed using a Seahorse flux bioanalyzer (XFe96, Agilent, Santa Clara, CA, USA) and Mito-Stress test. First, as shown in Figure 8A, all tested concentrations of CoCl 2 dose-dependently and significantly reduced HRECs' basal OCRs, normalized by subtracting non-mitochondrial OCRs (corresponding to rotenone/antimycin-insensitive OCRs). Second, the ATP synthase inhibitor (oligomycin) was added to separate OCRs linked to ATP production by subtracting resultant OCRs from basal OCR. As shown in Figure 8B, only 100 and 1000 µM of CoCl 2 lowered ATP-linked OCRs in HRECs significantly and in a dose-dependent manner compared to control. Third, the addition of protonophore uncoupler (FCCP) stimulated OCRs to their maximal activities in both control and CoCl 2 -treated HRECs but to a significantly lower extent in 100 and 1000 µM CoCl 2 -treated cells ( Figure 8C). These results confirmed that all CoCl 2 tested concentrations impaired mitochondrial respiratory function under basal conditions, creating a cytopathic hypoxia-like condition. resistance have already occurred, particularly in 100 and 1000 µM CoCl2 groups ( Figure  5). Altogether, these data indicate that the disruption of paracellular barrier integrity of HRECs is an earlier event that occurs in response to cytopathic hypoxia long before any noticeable effect on cell viability.  To further ensure that the observed effect of CoCl 2 on barrier integrity was not a consequence of cell cytotoxicity, the LDH assay was performed at the end of the experiment. The LDH assay results in Figure 8D did not demonstrate any cytotoxicity at all tested concentrations of CoCl 2 within the 24 h experimental period, where losses to total resistance have already occurred, particularly in 100 and 1000 µM CoCl 2 groups ( Figure 5). Altogether, these data indicate that the disruption of paracellular barrier integrity of HRECs is an earlier event that occurs in response to cytopathic hypoxia long before any noticeable effect on cell viability.

Discussion
The novel finding of the present study is that cytopathic hypoxia compromises the barrier integrity of HRECs in distinct ways, with the paracellular junction between adjacent HRECs being the most vulnerable target. This conclusion was based on real-time monitoring and dissecting of HREC barrier dysfunction induced by cytopathic hypoxia across the paracellular junction between HRECs (evaluated by R b ), the HRECs-substrate basolateral adhesion (evaluated by α), and the barrier to flow through the cell membrane of HRECs (evaluated by C m ). The following experimental results support this conclusion: (i) the R b parameter of the HREC barrier was the parameter that declined first and most significantly by the cytopathic hypoxia-inducing agent and in a dose-dependent manner; (ii) during the period when R b began to fall to it is minimum, other parameters of the HREC barrier, including α and C m , were unaffected, and (iii) intriguingly, the compromised effect of cytopathic hypoxia on R b was associated with mitochondrial dysfunction but not with cytotoxicity. Using mathematical modeling of ECIS data, our study is the first to demonstrate these temporal correlations between the three components of the HREC barrier under the condition of cytopathic hypoxia. ECIS instrumentation is the only technology currently available that enables modeling each of these critical cellular barrier components. In terms of modeling endothelial barrier function in vitro, the R b denotes paracellular permeability, governed by tight junction complexes. The component α reflects changes in cell basal adhesion, controlled by integrins, while C m describes the changes in membrane composition as a function of capacitance.
Maintaining the endothelial barrier integrity is essential for the proper functioning of tissues and organs. Therefore, it is crucial to know which barrier parameters contribute to the endothelial barrier's strength, and this will vary depending on endothelial cell type. For example, lymphatic endothelial cells are leaky because they are poor in cell-cell tight junction (TJ) structures, yet they demonstrated high barrier resistance on ECIS, which is derived mainly from α of cell-substrate adhesion [24,25]. On the other hand, brain endothelial cells are enriched in TJs and exhibit high R b and α. In agreement, our study shows that HRECs have high barrier resistance on ECIS, which is derived from establishing robust α and C m followed by a stable R b [23]. Since the endothelial barrier of the retina is compromised in IRs, our study is the first to dissect which parameters of the HREC barrier are more sensitive to ischemic insults. The ECIS modeling software clearly indicates that the paracellular resistance (R b ) is the only barrier parameter affected by the cytopathic hypoxia-inducing agent in a dose-dependent manner (Figure 7).
There are several mechanisms through which cytopathic hypoxia induces HREC dysfunctions that can be applied to changes in R b seen in our model. One such mechanism refers to the upregulation of vascular endothelial growth factor (VEGF) by CoCl 2 and activation of its cognate receptor (VEGFR2) [26,27]. Activation of VEGFR2 phosphorylates a critical paracellular TJ protein that regulates R b , Zonula occludens (ZO)-1, and results in subsequent dissociation of ZO-1 from TJ complexes [28]. In agreement, VEGF receptor inhibitor (SU1498) improved blood-brain barrier maintenance under hypoxia [29]. Additionally, our previous study showed that CoCl 2 reduces expression and disrupts the distribution of ZO-1 in retinal epithelial cells [25]. Therefore, it is plausible that phosphorylation of ZO-1 by the cytopathic hypoxia-driven VEGF causes remodeling of HRECs paracellular junctions and changes in R b seen in our model.
The second mechanism that may contribute to the breakdown of the HREC paracellular barrier under cytopathic hypoxia is the generation of reactive oxygen species (ROS), particularly from mitochondrial dysfunction, that damages adherens junction proteins (AJPs). VE-cadherin and β-catenin are the major AJPs regulating paracellular permeability that are affected by ROS generation [30]. In addition, previous studies have shown that CoCl 2 stabilizes HIF-1α expression [31][32][33], and increased HIF-1α levels are necessary to increase NADPH oxidase (Nox)2 expression and the subsequent ROS production [34], which promotes internalization of VE-cadherin along with β-catenin from the cell-cell contacts [35]. This internalization of AJPs results in adherens junction disassembly and thus may explain the compromise of R b barrier integrity in HRECs subjected to cytopathic hypoxia.
An additional explanation for the breakdown of the HREC paracellular barrier under cytopathic hypoxia is the decreased ATP-linked OCR ( Figure 8C), which describes damaged ETC and impaired ATP synthesis [36]. ATP depletion has been shown to cause tight junction disassembly by interfering with specific protein-protein interactions in the atypical protein kinase C (aPKC) signaling pathway. Under normal conditions, aPKC phosphorylates partitioning-defective (Par)3 at serine 827, which is vital in establishing and maintaining TJ assembly [37]. However, under ATP-depleting conditions, aPKC is unable to phosphorylate Par3 stabilizing the association of Par3 with aPKC, thereby preventing its release and localization to the tight junction [38]. Collectively, reduced ATP production with aberrant aPKC signaling may contribute to the loss of tight junctions and lower R b values in HRECs subjected to cytopathic hypoxia.

Conclusions
Our results support that HRECs have discernible dielectric properties under cytopathic hypoxia, in which the disturbance of the paracellular barrier between adjacent HRECs is a prodromal index of impending endothelial dysfunction that occurs prior to any loss in cell viability and cell-substrate adhesion as well as prior to any changes in membrane composition. Future studies aimed at finding gene targets manipulating paracellular barriers of retinal endothelial cells subjected to ischemic insults could utilize the ECIS technology as a powerful tool in screening such genes. and, therefore, drug development.

ECIS Experiment and Modeling
The effects of cytopathic hypoxia on retinal endothelial cellular behaviors were assessed by monitoring the overall cellular impedance (Z) using (ECIS ® Zθ (theta)) technology (Applied Biophysics Inc., Troy, NY, USA) followed by mathematical modeling as previously described [21,39]. Briefly, a 96-well array (96W20idf PET; Applied Biophysics Inc.) was first coated with 100 µM cysteine (50 µL/well; Applied Biophysics) for half an hour, followed by aspiration. Then, the array was coated with 0.02% gelatin (50 µL/well; Sigma, Burlington, MA, USA) for another half an hour, followed by aspiration. Next, human retinal endothelial cells (HRECs) obtained from Cell Systems (Kirkland, WA, USA) were seeded in Microvascular Endothelial Cell Growth Medium-2 BulletKit (Lonza, Walkersville, MD, USA; Catalog #: CC-3202 EGM-2 MV). After HRECs became confluent and formed a mature monolayer (as indicated by a capacitance below 20 nF), the culture media were replaced by media free of serum and growth factors for 10-12 h before applying different concentrations of CoCl 2 (15862-1ML-F, Sigma, St. Louis, MO, USA). Thereafter, AC of 1 µA was subjected to HRECs cultured onto the electrode surfaces embedded in the bottom of each well to measure the overall Z with respect to time and frequency. Nine multifrequency measurements in a range from 250 Hz to 64,000 Hz were used. The Z value at each time point was normalized to the baseline Z acquired before the addition of CoCl 2 and then plotted as a function of time.
Additionally, the ECIS system was used to dissect the overall Z into two parameters (resistance and capacitance) across the HREC monolayer. Following this, we further used the ECIS software to mathematical model transendothelial resistance across the HREC monolayer into three important components as previously described [40]. These components are R b (an indicator for the integrity of the paracellular junctions between HRECs measured in Ω·cm 2 ), alpha (α, an indicator of the basolateral attachment of HRECs to their extracellular matrix measured in Ω·cm 1/2 ), and C m (cell membrane's capacitance which indicates changes in the HREC membrane morphology measured in µF/cm 2 ). The data were collected either at the end or throughout the experiment by calculating the area under the curve (AUC).

Assessment of HREC Viability
The effect of different concentrations of CoCl 2 on the viability of HRECs was assessed by lactate dehydrogenase (LDH) Cytotoxicity Assay (CyQUANT™; Invitrogen-C20300, Waltham, MA, USA). In this assay, HRECs were cultured in 96-well plates (1 × 10 4 /well), and after cells became confluent, the culture media were replaced by media free of serum and growth factors for 10-12 h before applying different concentrations of CoCl 2 (0, 10, 100, and 1000 µM) for 24 h. After that, the amount of LDH released into the medium was determined per the manufacturer's instructions.

Mitochondrial Bioenergetic Profiles
Seahorse bioanalyzer (XFe96, Agilent Technologies, Santa Clara, CA, USA) was used to measure the mitochondrial bioenergetic profile of HRECs by determining the oxygen consumption rate (OCR) as previously described [12,21]. HRECs (40,000 cells/well) were cultured in 96-XF tissue culture microplate (Agilent Technologies) using full media for 24 h. As HRECs reached confluency, the culture media were replaced by media free of serum and growth factors for 12-16 h, followed by CoCl 2 -treatment (0, 10, 100, and 1000 µM). At the end of the experiment, the culture media were replaced by XF assay media, then the Mito Stress test (Agilent Technologies) was performed. After the baseline was established, sequential injections of oligomycin (Olig, 1 µM), carbonyl cyanide-4 trifluoromethoxy phenylhydrazone (FCCP, 1 µM), and rotenone/Antimycin (1 µM each) were used to determine basal OCR, ATP-linked OCR, and maximal OCR.

Statistical Analysis
The two-tailed Student t-test or one-way analysis of variance (ANOVA) followed by the Tukey post-hoc test was used to determine differences between experimental groups. Graphical representations of p values are * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.