Regulation of Mitochondrial Biogenesis in Diabetic Retinopathy

Abstract

Mitochondrial dysfunction plays a major role in diabetic retinopathy development and in its resistance to halt after the reversal of hyperglycemia (metabolic memory). Diabetes also upregulates many long noncoding RNAs, RNAs with >200 nucleotides with no reading frame, and several of them resist reversal after hyperglycemia cessation. Our aim was to investigate the role of LncRNA HOTAIR, a master regulator of chromatin dynamics, in mitochondrial biogenesis in diabetic retinopathy and in metabolic memory. Using retinal endothelial cells and Müller cells, incubated in high glucose (20 mM D-glucose), the effect of HOTAIR-siRNA on mitochondrial biogenesis was investigated by quantifying mitochondrial mass, copy numbers, and mtDNA replication, structure, and function. HOTAIR’s role in metabolic memory was investigated by analyzing mitochondrial biogenesis in HOTAIR-siRNA transfected cells incubated in high glucose for four days, followed by normal glucose (5 mM D-glucose) for four days. HOTAIR was upregulated in both retinal vascular and nonvascular cells, and HOTAIR-siRNA ameliorated decreases in mtDNA biogenesis and protected their mitochondria from structural/functional damage. Reversal of high glucose insult failed to ameliorate HOTAIR upregulation and impaired mtDNA biogenesis in both endothelial and Müller cells, but regulation of HOTAIR during high glucose incubation, which followed normal glucose, prevented a decrease in mitochondrial mass and mtDNA copies. Thus, HOTAIR has a major role in mitochondrial biogenesis and in the continued impaired biogenesis in both vascular and nonvascular cells. Regulating HOTAIR may provide a therapeutic option to inhibit the development/progression of diabetic retinopathy.

1. Introduction

Retinopathy remains a major diabetic complication, and it is the primary cause of vision loss. Through a cascade of pathophysiological events, retinal structure and function are damaged by chronic hyperglycemia, and, if not addressed on-time, they lead to retinal detachment-vision loss [1,2,3]. Retina has the highest oxygen-consuming tissue in the human body, and its cell function is intimately dependent on mitochondrial function [4,5]. In diabetes, retinal mitochondria are damaged and dysfunctional mitochondria accelerate cell death, leading to the formation of degenerative capillaries [6,7,8,9,10]. In addition to mitochondrial structural and functional damage, mitochondrial biogenesis is also impaired and mitochondrial DNA (mtDNA) copy numbers are reduced, leaving the cell with suboptimal number of mitochondria [7,11,12,13,14].

Mitochondria response to increased energy demand by producing new mitochondria, and they have their own small (~16.7 kb) circular DNA and mitochondrial DNA (mtDNA) biogenesis is a process via which cell increases its individual mitochondrial mass [15,16]. This complex biological process involves coordination between the mitochondrial and nuclear genomes and replication of the mtDNA [17]. Peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC1α) is the master regulator of mitochondrial biogenesis, and mitochondrial transcription factor A (TFAM) is the final effector of mtDNA replication and transcription, which binds on the noncoding region of mtDNA (D-loop, ~1 kbp) to initiate replication/transcription [18,19,20]. The replicative DNA polymerase gamma (POLG) and mitochondrial helicase (Twinkle) form a minimal replisome, which is essential for mtDNA replication [21]. In diabetic retinopathy, mtDNA replication/transcription are impaired [12,13,22].

Diabetes also results in aberrant expression of many noncoding RNAs, including micro RNAs and long noncoding RNAs (LncRNAs). LncRNAs are RNAs with >200 nucleotides, and these RNAs can interact with RNA, DNA or protein and regulate gene expression. They exert their cellular effects via complex, diverse, cell- and location-specific manner [23,24,25,26]. Diabetes results in aberrant expression of many LncRNAs, and their altered expression is associated with diabetic complications [26,27,28,29,30]. Among differentially expressed LncRNAs, LncRNA HOX antisense intergenic RNA (HOTAIR) is a master regulator of chromatin dynamics, and it suppresses gene expression by serving as a scaffold for the histone-modifying complexes polycomb-repressive complex 2 and lysine demethylase [31]. In diabetes, HOTAIR upregulation is seen in the retina, serum and vitreous, and is associated with inflammasome activation, hypoxia, endothelial dysfunction and angiogenesis [28,32,33,34,35]. However, whether HOTAIR has any role in mitochondrial biogenesis remains elusive.

Diabetes Control and Complications Trial (DCCT) and follow up EDIC studies have shown that the participants with conventional glycemic control during DCCT continue to develop clinical features of retinopathy long after hyperglycemia cessation, and the benefits of early intensive glycemic control persist beyond the period of its institution, suggesting a ‘metabolic memory’ phenomenon [36,37,38,39]. Over three decades after the start of DCCT, intensive glycemic group still has a better visual quality of life [37,38,39], but the mechanism of this legacy effect remains obscure. Our studies have shown that mitochondria remain dysfunctional and mitochondrial biogenesis-mitophagy do not benefit from termination of hyperglycemia, and the vicious cycle of free radicals continues to self-propagate [8,13,22,40]. Furthermore, recently we have shown that upregulation of HOTAIR in the retina, initiated by hyperglycemia, does not cease, even when hyperglycemia is replaced by normal glycemia in rodents [34]. However, the role of HOTAIR in the continued impaired mitochondrial biogenesis remains unclear.

Our aim was to investigate the role of HOTAIR in regulation of mitochondrial biogenesis in diabetic retinopathy, and in its failure to cease after termination of hyperglycemic insult. Using human retinal endothelial cells (HRECs), the effect of HOTAIR regulation on mitochondrial biogenesis was determined. The role of HOTAIR in metabolic memory was investigated using HRECs, transfected with HOTAIR -siRNA and incubated in high glucose, followed by normal glucose. Diabetic retinopathy is now considered a complex neurovascular disorder, affecting not only vascular structure, but also neural tissue of the retina [41,42,43]. Measurements were also made in retinal Müller cells, the cells that expand radially across almost the whole width of the retina, ensheathing retinal neurons and microvascular cells [43,44].

2. Methods

Retinal cells: Primary human retinal endothelial cells (HRECs) were procured from Cell Systems Corporation (Cat. No. ACBRI 181; Cell Systems Corp, Kirkland, WA, USA), and primary human retinal Müller cells (RMCs) from Accegen (Cat No. ABT-TC133L), Accegen, Fairfield, NJ, USA). Incubation of HRECs was conducted in Dulbecco’s Modified Eagle Medium (DMEM) containing 1% heat-inactivated fetal bovine serum, 9% Nu-serum, 1 μg/mL endothelial cell growth supplement and 1% each of insulin, transferrin, selenium, glutamax and antibiotic/antimycotic, and RMCs in DMEM supplemented with 2% heat-inactivated fetal bovine serum, 8% Nu-Serum and 1% antibiotic/antimycotic [45,46]. Cells (HRECs and RMCs) from 5th to 6th passage were transfected with HOTAIR-siRNA (HTsi; Cat. No. 4392420, Invitrogen, Carlsbad, CA, USA) or scrambled control RNA (SC) using Lipofectamine RNAiMAX transfection reagent (Cat. No. 13778-030; Invitrogen). Transfection efficiency was calculated by measuring HOTAIR transcripts using SYBR green-based qRT-PCR [45].

Both HRECs and RMCs, untransfected or HTsi or SC transfected, were incubated in their respective incubation medium containing high glucose (HG, 20 mM D-glucose), and, after four days, the cells either remained in high glucose (HG, HG/HTsi and HG/SC groups, respectively) or were incubated in normal glucose (NG, 5 mM D-glucose) for four additional days (HG-NG, HG/HTsi-NG and HG/SC-NG groups, respectively). Each experiment had cells incubated in either normal glucose or in 20 mM L-glucose (osmotic/metabolic control) for eight continuous days (NG and L-Gl groups).

Gene transcripts: RNA (1 µg) isolated by TRIzol Reagent (Cat. No. 15596018, Invitrogen) was employed for cDNA synthesis using a High-Capacity cDNA Reverse Transcription kit (Cat. No. 4368814, Applied Biosystems, Waltham, MA, USA). Quantitative real-time PCR (qRT-PCR) was performed using PowerUp™ SYBR™ Green Master Mix (Cat. No. 4367659, Applied Biosystems) and gene-specific primers (Table 1). Relative fold change in gene transcripts was determined using the 2^−ΔΔCt method and β-actin as the housekeeping gene [45].

Table 1.

List of primers.

Target	Forward Primer (5′–3′)	Reverse Primer (5′–3′)
HOTAIR	ACGAAGGTGAAAGCGAACCA	CCCTCTGCCACGTTTGTTCC
D-loop	ATGGGGAAGCAGATTTGGGT	GCGTTTTGAGCTGCATTGCT
CytB	TCACCAGACGCCTCAACCGC	GCCTCGCCCGATGTGTAGGA
β-actin	AGCCTCGCCTTTGCCGATCCG	TCTCTTGCTCTGGGCCTCGTCG
CytB DNA	TCACCAGACGCCTCAACCGC	GCCTCGCCCGATGTGTAGGA
β-actin DNA	CTTTCCTGCCTGAGCTGACC	CCTAGAAGCATTTGCGGTGG

RNA Fluorescence in Situ Hybridization (RNA-FISH): HOTAIR expression was confirmed by RNA-FISH technique using aminoallyl-dUTP-Cy5-incorporated HOTAIR probes synthesized by asymmetric PCR, as reported previously [45]. Cells fixed on coverslips with paraformaldehyde (4% w/v) were dehydrated with 70–100% ethanol and air-dried. They were then incubated at 37 °C for three hours with the denatured HOTAIR-probes in hybridization buffer (10% dextran sulphate, 10% formamide and 4× saline-sodium citrate buffer, pH 7.0), washed with the hybridization buffer, followed by phosphate-buffered saline (PBS). The cells were mounted using DAPI containing mounting medium (Cat. No. H-1000, Vector Laboratories, Burlingame, CA, USA) and images were captured using a Zeiss microscope using 63× oil objective. The Arithmetic mean intensity (AMI) was determined using the Zeiss software module (Zen 2.6 Pro) by selecting the region of interest.

Chromatin immunoprecipitation (ChIP): Binding of POLG and of TFAM at D-loop was quantified by the ChIP technique using cells crosslinked in 1% formalin in PBS. After washing with PBS, they were lysed and chromatins were sheared by sonication. The clear lysate, collected by centrifugation at 12,000 rpm for 15 min at 4 °C, was immunoprecipitated with 3 μg each of POLG antibody (Cat. No. HPA056821, Sigma-Aldrich, St. Louis, MO, USA) or TFAM antibody (Cat. No. PIPA523776, Invitrogen); each experiment included IgG antibody control in place of POLG or TFAM. Using A/G plus agarose beads, the complex was collected. Beads were then washed with wash buffer and DNA was isolated employing the phenol-chloroform-isoamyl alcohol method. Using D-loop-specific primers, DNA samples were quantified for POLG or TFAM at D-loop, and qPCR values in each ChIP sample were normalized to the input control using the 2^−∆∆Ct method [40].

Mitochondrial mass and mtDNA copy numbers: Mitochondrial mass was determined by incubating cells on coverslips in CO₂ incubator for 30 min in the dark with 200 nM MitoTracker Green dye (Cat. No. M7514, Thermo Fisher Scientific, Waltham, MA, USA), the dye is insensitive to oxidative stress and membrane potential. They were imaged using a ZEISS ApoTome fluorescence microscope at 20× objective, and fluorescence intensity was calculated using the Zeiss software module by selecting the area of interest [11,47].

Copy numbers of mtDNA were measured by calculating the ratio of mtDNA-encoded cytochrome B (CytB) and nuclear DNA-encoded β-actin gene transcripts in the genomic DNA, isolated using a DNeasy Blood & Tissue Kit (Cat No. 69504, Qiagen, Valencia, CA, USA) [11,47].

Quantification of newly synthesized mtDNA: To measure newly synthesized mtDNA strands, functional assay of POLG was performed by the immunofluorescence technique by incorporating bromodeoxyuridine (BrDU, a thymidine analog) in mtDNA, as reported previously [22]. Briefly, after experimental incubations, the cells were incubated with 7 μM aphidicolin (Cat. No. 178273, Sigma-Aldrich) to inhibit nuclear DNA synthesis, followed by incubating them for six hours in fresh medium containing 15 μM BrDU and 7 μM aphidicolin. After washing with PBS, the cells were fixed with 2% paraformaldehyde, permeabilized with 0.1% TritonX-100, and incubated with 2 N HCl for 30 min at 37 °C to recover the BrDU epitope. They were washed 3×, blocked in PBS containing 5% goat serum and 0.1% Triton X-100 for 45 min, and then incubated overnight at 4 °C with mouse anti-BrDU antibody (Cat. No. B8434, Sigma-Aldrich; dilution 1:500); Texas Red-conjugated anti-mouse IgG (dilution 1:1000) was used as the secondary antibody. The cover slips were mounted using mounting medium containing DAPI. The fluorescence images were visualized and captured using ZEISS ApoTome fluorescence microscope at 63× oil objective and AMI was determined using the Zeiss software module [13].

Oxygen consumption rate (OCR): OCR was measured using a Seahorse XF analyzer (Agilent Technologies, Santa Clara, CA, USA), as reported previously [45,47]. In brief, cells grown in a 96-well XF cell culture plate and incubated in their respective treatment media, were washed 2× with 100 μL assay medium (Seahorse XF DMEM supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose), and 180 μL assay medium was added to each well. After incubating at 37 °C for 30–45 minutes in a non-CO₂ incubator, OCR was determined using a Seahorse XF Cell Mito Stress Test Kit (Cat. No. 103015-100; Agilent Technologies) by injecting 1.5 μM oligomycin (ATP synthase inhibitor), 2.0 μM FCCP (uncoupler) and 0.5 μM rotenone/antimycin A (Complex I and III inhibitor) in ports A, B, and C, respectively. The data were analyzed using Wave software 10.0.1 (Agilent Technologies) respectively [45,46].

Electron microscopy: Cells (HRECs and RMCs), after experimental incubations, were fixed in fixing solution (0.1 M cacodylate buffer, 2% paraformaldehyde, and 2% glutaraldehyde) for two hours, washed with the 0.1 M cacodylate buffer, and incubated on ice for one hour in 0.1 M cacodylate buffer containing 2% osmium and 3% potassium ferrocyanide. They were then stained with 1% thiocarbohydrazide solution and incubated with 2% osmium at room temperature for 40 min, followed by 1% uranyl acetate at 4 °C overnight. After incubation in Walton’s lead solution (0.08 M lead nitrate, 0.12 M sodium citrate, and 0.16 M sodium hydroxide) for 30 min at 60 °C, followed by dehydration with 50–100% ethanol, cells were embedded in the resin to prepare 70–85 nm ultrathin sections by Leica ARTOS 3D ultramicrotome (Leica, Teaneck, NJ, USA). The sections, placed on glow discharge-treated silicon wafers, were imaged by a Zeiss Gemini300 scanning electron microscope (Carl Zeiss, Inc., Baden-Württemberg, Germany). A backscatter detector was employed to visualize mitochondrial morphology and cristae structure [45].

Statistical analysis: All the results are represented as mean ± SD, statistically analyzed using GraphPad Prism (version 10.3.1, San Diego, CA, USA). Significance of variance was determined using one-way ANOVA, and a p value < 0.05 was considered statistically significant.

3. Results

Endothelial cells: HOTAIR expression is increased in the retina from diabetic rats (streptozotocin-induced) and from human donors with established diabetic retinopathy [34]; to investigate how hyperglycemia affects retinal vasculature, HRECs were analyzed. HOTAIR transcripts were increased by ~75% in cells in high glucose (HG group), compared to cells in normal glucose (NG group) or 20 mM L-glucose (L-Gl group; Figure 1a), and RNA-FISH further confirmed it by showing significantly higher AMI of HOTAIR in the HG group (Figure 1b,c). Furthermore, there was no further increase in HOTAIR expression in cells incubated in high glucose for four days or eight days (data not presented). The glucose-induced increase in HOTAIR was ameliorated in HRECs transfected with HOTAIR-siRNA (HG/HTsi group), but not with scrambled control RNA (HG/SC group; Figure 1a–c). Figure 1d shows the transfection efficiency of HOTAIR-siRNA.

Figure 1.

HOTAIR expression in retinal endothelial cells. HRECs incubated in high glucose were analyzed for HOTAIR (a) transcripts by qRT-PCR using β-actin as housekeeping gene, and (b) expression by of RNA-FISH; scale bar = 10 μm. (c) AMI of HOTAIR was calculated using 10–15 cells/group/experiment. (d) Transfection efficiency of HOTAIR-siRNA. Values are presented as mean ± SD of 3–4 independent experiments, with each measurement made in triplicate. NG and HG = 5 mM and 20 mM D-glucose, respectively; HG/HTsi and HG/SC = HOTAIR-siRNA or scrambled RNA transfected cells in HG; HG-NG, HG/HTsi-NG and HG/SC-NG = untransfected, HOTAIR-siRNA or scrambled RNA transfected cells, respectively, in HG for four days, followed by four days in NG and L-Gl = 20 mM L-glucose. * p < 0.05 compared to NG and # p < 0.05 compared to HG.

To investigate the role of HOTAIR in mitochondrial biogenesis, mitochondrial mass and mtDNA copy numbers were quantified. Compared to normal glucose, high glucose, as expected [11], decreased the fluorescent intensity of MitoTracker green and reduced mtDNA copy numbers. Transfection of cells with HOTAIR-siRNA ameliorated a glucose-induced decrease in MitoTracker intensity and decline in mtDNA copy numbers. Consistent with mitochondria mass/copy numbers, HOTAIR siRNA also prevented a decrease in the newly synthesized mtDNA strands, as shown by the increased incorporation of BrDU in HG/HTsi group compared to the HG/SC group. Incubation of cells in 20 mM L-glucose had no effect on mitochondrial mass/mtDNA copies/newly synthesized mtDNA (Figure 2a–e).

Figure 2.

Regulation of HOTAIR and mitochondrial biogenesis. (a) Relative mitochondrial mass was calculated by staining HRECs with MitoTracker Green, scale bar = 20 μm, and (b) its fluorescence intensity was quantified in 8–10 cells/group/experiment. (c) mtDNA copy numbers were quantified by the ratio of mtDNA-encoded CytB and nuclear DNA- encoded β-actin in total DNA. (d) Representative images from BrDU assay for analysis of newly synthesized mtDNA strands; scale bar = 10 μm. (e) BrDU fluorescence intensity calculated from 8 to 10 cells/group/experiment. Values in the graphs are mean ± SD of 3–4 different experiments, with each measurement made in triplicate * p < 0.05 vs. NG, and # p < 0.05 vs. HG.

Binding of POLG at the D-loop region initiates mtDNA replication, and that of TFAM stabilizes mtDNA [13,19]; the effect of HOTAIR-siRNA on the binding of POLG and of TFAM at D-loop was determined. As shown in Figure 3a, a decrease in POLG binding at D-loop was ameliorated by HOTAIR-siRNA, and the values in the HG/HTsi group and NG or L-Gl groups were similar. The IgG antibody control values were <1% of those from the POLG antibody. In accordance with POLG, the glucose-induced decreases in TFAM binding and CytB gene transcripts were also ameliorated by HOTAIR siRNA (Figure 3b,c).

Figure 3.

Role of HOTAIR in D-loop binding at POLG and at TFAM. Binding of (a) POLG and (b) TFAM at D-loop was measured by ChIP technique using IgG (^) as negative control. (c) CytB transcripts were quantified by qRT-PCR, and β-actin was used as the housekeeping gene. Values are represented as mean ± SD of three or more individual experiments, and each measurement was made in triplicate. NG = 5 mM D-glucose; HG, HG/HTsi and HG/SC = untransfected, HOTAIR-siRNA or scrambled RNA transfected HRECs in HG; HG-NG = cells four days in HG followed by four days in NG; HG/HTsi-NG and HG/SC-NG = HOTAIR-siRNA or scrambled RNA transfected cells four days in HG+ four days in NG; L-Gl= 20 mM L-glucose. * p < 0.05 compared to NG and # p < 0.05 compared to HG.

Decrease in mitochondrial mass-mtDNA copies compromise cellular functional integrity [48]; the effect of HOTAIR-siRNA on mitochondria function and structure was evaluated. Cells in high glucose had overall reduced oxygen consumption, and their basal and maximal respiration rates were also significantly lower compared to cells in normal glucose. However, HOTAIR-siRNA, but not scrambled RNA, prevented glucose-induced decreases in respiration rates (Figure 4a–c). In the same cell population, HOTAIR-siRNA also prevented glucose-induced mitochondrial structural damage; cells in the NG group had elongated mitochondria with regular cristae compared to cells in the HG group with rounded mitochondria and damaged cristae (Figure 4d).

Figure 4.

Effects of HOTAIR on mitochondrial function and structure of endothelial cells. (a) OCR was measured using Seahorse XF Analyzer using Cell Mito Stress Test Kit. (b) Basal and (c) maximal respiration were calculated from the values obtained after oligomycin and FCCP injection, respectively. Values are presented as mean ± SD from two to three cell preparations. * p < 0.05 vs. NG and # p < 0.05 vs. HG. (d) Representative electron micrographs of HRECs with the red arrow pointing at the mitochondrion; white scale bar = 400 nm.

Müller cells: Diabetic retinopathy is now recognized as a neurovascular disorder [41,42,43]; the effect of hyperglycemia on retinal Müller cells was examined. HOTAIR expression increased by over 75% in RMCs in the HG group compared to the NG group, as evidenced by both its transcripts and by RNA-FISH showing increased AMI of HOTAIR. Transfection of cells with its siRNA attenuated a glucose-induced increase in HOTAIR, and the values obtained from the HG/HTsi group were not different from those obtained from the NG or L-Gl groups (Figure 5a–c). Figure 5d is included to show the transfection efficiency of HOTAIR-siRNA.

Figure 5.

Effect of high glucose on HOTAIR expression in retinal Müller cells. (a) HOTAIR transcripts were quantified in RMCs by qRT-PCR and β-actin was used as housekeeping gene, and (b) its expression by RNA-FISH; scale bar = 10 μm. (c) AMI of HOTAIR calculated using 15–20 cells in each group. (d) HOTAIR-siRNA transfection efficiency in RMCs. Values are represented as mean ± SD of three different experiments, with each measurement made in duplicate/triplicate. NG and HG = RMCs in 5 mM D-glucose or 20 mM D-glucose; HG/HTsi and HG/SC = HOTAIR-siRNA or scrambled control RNA transfected, respectively; HG/HTsi-NG and HG/SC-NG = RMCs in HG for four days, followed by four days in NG, and L-Gl = 20 mM L-glucose. * p < 0.05 vs. NG and # p < 0.05 vs. HG.

Consistent with results from HRECs, HOTAIR-siRNA, but not scrambled RNA, also ameliorated a decrease in mitochondrial mass, mtDNA copies, and the synthesis of newly synthesized mtDNA strands (BrDU intensity), observed in high-glucose medium (Figure 6a–d). Binding of POLG and TFAM at D-loop also decreased, further confirming subnormal mtDNA replication and stability in Müller cells. POLG binding was similar in the HG/HTsi, NG, and L-Gl groups, and was significantly higher compared to the HG group (Figure 7a,b). Consistent with our previous results from HRECs showing impaired mitochondrial potential in high-glucose conditions [45], RMCs in the HG group also had impaired mitochondrial potential with a decreased ratio of aggregates to monomers, and their mitochondria presented partial cristolysis (Figure 8a–c).

Figure 6.

Effect of HOTAIR regulation on mitochondrial biogenesis. RMCs were analyzed for (a) relative mitochondrial mass by MitoTracker Green staining, scale bar = 20 μm, and (b) represents AMI, calculated from 10 to 15 cells/group/experiment. (c) mtDNA copy numbers were calculated from the ratio of CytB and β-actin in genomic DNA. (d) A representative image representing BrDU staining of newly synthesized mtDNA strands; scale bar = 10 μm, and (e) BrDU fluorescence intensity, calculated using 10–15 cells/experimental condition/experiment. Each measurement was made in triplicate in three or more different experiments, and values are presented as mean ± SD from three or more independent experiments. * p < 0.05 compared to NG and # p < 0.05 compared to HG.

Figure 7.

Effects of HOTAIR on interactions between D-loop- and POLG and TFAM. Using RMCs incubated in different experimental conditions, binding (a) POLG at D-loop and (b) TFAM at D-loop were analyzed by ChIP assay using IgG as the antibody control (^). (c) CytB transcripts were measured by qRT-PCR, and β-actin was used as the housekeeping gene. Values are presented as mean ± SD from three to five individual experiments, and each measurement was made in triplicate. NG = 5 mM D-glucose; HG, HG/HTsi and HG/SC = untransfected, HOTAIR-siRNA or scrambled RNA transfected RMCs in HG; HG-NG = cells four days in HG followed by four days in NG; HG/HTsi-NG and HG/SC-NG = HOTAIR-siRNA or scrambled RNA transfected cells four days in HG+ four days in NG; L-Gl = 20 mM L-glucose. * p < 0.05 compared to NG and # p < 0.05 compared to HG.

Figure 8.

Effect of HOTAIR on mitochondrial functional and structural stability of Müller cells. (a) Mitochondrial membrane potential was determined by JC-1 staining, scale bar = 20 μm, and (b) the ratio of aggregates (polarized, red) to monomers (depolarized, green) was calculated using intensities of red and green fluorescence. (c) Representative electron microscopy images of RMCs with mitochondrion marked by red arrow; white scale bar = 400 nm. Each measurement was made in triplicate in 3–5 different experiments, and the values in the graphs are presented as mean ± SD. NG and HG = RMCs in 5 mM D-glucose or 20 mM D-glucose; HG/HTsi and HG/SC = HOTAIR-siRNA or scrambled RNA transfected cells in HG; HG-NG, HG/HTsi-NG and HG/SC-NG = RMCs untransfected, HOTAIR-siRNA or scrambled RNA transfected, respectively, in HG for four days, with four additional days in NG; L-Gl = 20 mM L-glucose. * p < 0.05 compared with NG and # p < 0.05 compared with HG.

Reversal of high glucose insult by normal glucose: Since retinal mitochondrial biogenesis remains compromised and HOTAIR continues to be upregulated after the termination of hyperglycemic insult [12,34], HOTAIR’s role in the continued impaired mitochondrial biogenesis was also investigated. HOTAIR expression continued to be upregulated in cells that were in normal glucose for four days, which followed four days of high glucose (HG-NG group), but when cells transfected with its siRNA were incubated in normal glucose for four days, after four days of high glucose (HG/HTsi-NG group), HOTAIR expression was significantly lower than that observed in cells in HG-NG group. Scrambled RNA transfected cells incubated in high glucose for four days with four additional days of normal glucose (HG/SC-NG) had HOTAIR values similar to those obtained in the HG-NG group (Figure 1a–c). Consistent with HOTAIR expression, MitoTracker green intensity, mtDNA copy numbers, BrDU incorporation, and interactions between D-loop and POLG and D-loop and TFAM were also significantly higher in the HG/HTsi-NG group vs. the HG-NG and HG/SC-NG groups, (p < 0.05, Figure 2 and Figure 3). Furthermore, overall OCR, basal and maximal oxygen consumption rates were not different from those obtained from cells in NG group (Figure 4).

In accordance with HRECs, glucose-induced upregulation in HOTAIR in RMCs also did not benefit from the termination of glucose insult, and mitochondrial biogenesis continued to be subnormal with reduced mitochondrial mass, mtDNA copies, and newly synthesized mtDNA strands. Furthermore, their mitochondrial function and structure rem00ained compromised with impaired mitochondrial potential and damaged cristae. However, if the cells were transfected with HOTAIR-siRNA before incubation in high glucose, which was then followed by normal glucose (HG/HTsi-NG), mitochondrial biogenesis was significantly better than RMCs in the HG, HG/SC, HG-NG, and HG/SC-NG groups (Figure 5, Figure 6, Figure 7 and Figure 8).

4. Discussion

Oxidative stress–mitochondrial dysfunction is considered to have a central role in the pathogenesis of diabetic retinopathy, and the damaged mitochondria continues to fuel into the vicious cycle of free radicals [8]. Sustained damage of retinal structural and functional stability by free radicals does not allow for the progression of diabetic retinopathy to halt, even when hyperglycemia is terminated [8]. LncRNAs, RNAs that can regulate physiological function by interacting with protein, DNA, or RNA, are also aberrantly expressed in the retina, and many of them resist reversal after cessation of hyperglycemia [34]. Here, our results show that HOTAIR, which is considered be a key regulator of chromatin dynamics and genes involved in many pathophysiological processes including apoptosis, cellular metabolism, and diabetes, is upregulated in both retinal vascular and nonvascular cells, and its upregulation has a significant role in impaired biogenesis of mtDNA. Results from retinal endothelial cells and Müller cells show that regulation of HOTAIR by its siRNA ameliorates decreases in mtDNA biogenesis, experienced by both vascular and nonvascular cells of the retina in hyperglycemic milieu, and protects their mitochondria from structural and functional damage. In addition, although metabolic memory phenomenon associated with continued progression of diabetic retinopathy after the reversal of hyperglycemia has been well documented in the retina and its vascular cells [22,49,50,51], here we show that Müller cells also experience a similar metabolic memory phenomenon to that experienced by vascular cells; the termination of high-glucose insult fails to ameliorate HOTAIR upregulation and mtDNA biogenesis in both vascular and nonvascular cells, but regulation of HOTAIR during high-glucose incubation, which is followed by normal glucose, prevents a decrease in mitochondrial mass and mtDNA copies, suggesting the importance of HOTAIR in mitochondrial stability–metabolic memory phenomenon.

Mitochondrial biogenesis, a process by which cells increase their individual mitochondrial mass, is essential to meet the specific energy and metabolic demands [17,52]. Master regulator PGC1α coordinates the activity of transcription factors nuclear respiratory factors 1 and 2 (NRF1 and 2), and TFAM is the major mitochondrial transcription factor essential for cellular bioenergetics [53,54]. Mitochondrial DNA, a double-stranded DNA, requires TFAM for promoter recognition by mitochondrial RNA polymerase for accurate initiation from L-strand and H- strand promoters. In addition, TFAM binds and bends mtDNA without sequence specificity, initiating transcription and subsequent mtDNA replication [53,55]. Binding of TFAM also introduces specific structural alterations, allowing for the packaging of histone-free mtDNA into protective nucleoids [56]. Our data here demonstrate that the inhibition of HOTAIR upregulation by its siRNA ameliorates a decrease in mitochondrial mass and mtDNA copies and also helps in the formation of newly synthesized mtDNA strands.

Mitochondrial biogenesis is a self-renewal process where new mitochondria are generated from existing ones; mtDNA replicates to meet energy demand and maintain normal functions, and POLG is an essential enzyme for accurate mtDNA replication. It synthesizes both heavy and light strands of mtDNA, and its 5′–3′ polymerase and 3′–5′ exonuclease activities help it in correcting errors during DNA replication to maintain the integrity of mtDNA [57,58]. Furthermore, TFAM also regulates mtDNA replication and transcription of mtDNA-encoded proteins essential for electron transport chain functioning [53,59]. The results presented here show that HOTAIR-siRNA prevents a decrease in the binding of POLG, and also the binding of TFAM, at D-loop. This results in improving the stability of mtDNA, resulting in ameliorating downregulation of mtDNA-encoded CytB, an integral member of complex III of the electron transport chain system.

Mitochondrial homeostasis is maintained by coordinated efforts of mitochondrial biogenesis, dynamics, and mitophagy, and cellular function also depends on mitochondria mass/mtDNA copy number [60,61]. Our results show that HOTAIR-siRNA, in addition to improving impairments in mitochondrial biogenesis that retinal cells experience in hyperglycemic milieu, also helps in maintaining their structural and functional integrity by preventing loss of their cristae and restoring their membrane permeability and oxygen consumption rate.

We recognize that HOTAIR is a master chromatin regulator, serving as a scaffold for histone-modifying complexes polycomb-repressive complex 2 (PRC2) and lysine demethylase 1 (LSD1) [31,62,63], and, in diabetic retinopathy, LSD1 and Ezh2 (catalytic component of PRC2) are activated in the retina [49]. Thus, there remains a possibility that upregulation of retinal HOTAIR could be altering mitochondrial biogenesis by possibly modifying histones at the promoters of genes encoding for proteins associated with mitochondrial biogenesis, e.g., POLG or Twinkle helicase.

As mentioned above, metabolic memory remains a major challenging issue for many diabetic patients; despite readjusting glycemic control to close to normal levels, the persistent adverse effects of prior hyperglycemia continue to damage their retina [8]. We have shown previously that mitochondrial dysfunction, brought up by prior hyperglycemia, does not benefit from near normoglycemia, which follows it, and the damaged mitochondria continues into the futile cycle of free radicals, and this is observed both in in vivo and in vitro models of diabetic retinopathy [7,11,12,13]. Moreover, over 350 genes, mainly implicated in mitochondrial damage, cytokine production, stress response, and cell death, and over 1400 LncRNAs, including LncNEAT1, LncMeg3, LncCytB, and HOTAIR, continue to be differentially expressed in the retina of diabetic rats after reversal of prior poor glycemic control (blood glucose > 400 mg/dL) with good glycemic control (blood glucose < 150 mg/dL) [34]. The data presented here show that the inhibition of HOTAIR upregulation in retinal vascular and nonvascular cells during high glucose exposure, which is followed by normal glucose, ameliorates impairments in mitochondrial biogenesis. This prevents a decrease in mitochondrial mass and mtDNA copy numbers and helps to keep mitochondrial functional and structural integrity.

Although histopathology of diabetic retinopathy is seen mainly in the retinal microvasculature, and historically this disease is considered as a microvascular complication of diabetes, diabetes damages the entire retina including vascular cells, glial cells, neurons, and pigment epithelium [41,42,43]. The results presented here clearly demonstrate that HOTAIR upregulation has a significant role in the impaired mitochondrial biogenesis in endothelial cells and in Müller cells, seen in hyperglycemic milieu, and both these cells experience a similar benefit from regulating HOTAIR upregulation. Also, in addition to endothelial cells of the retina, Müller cells also experience metabolic imprinting from prior glucose exposure and the regulation of HOTAIR prevents imprinting harmful effects of prior high glucose on mitochondrial biogenesis in both vascular and nonvascular cells.

In summary, HOTAIR overexpression in diabetes impairs mitochondrial biogenesis, and mitochondria mass and mtDNA copy numbers are decreased in both retinal vascular and nonvascular cells, leading to compromised retinal cellular health. Furthermore, retinal nonvascular cells also experience metabolic memory phenomenon with sustained damage to their mitochondria, and HOTAIR plays a significant role in the continued impaired mitochondrial biogenesis in endothelial cells and Müller cells—the cells that surround retinal capillaries and provide metabolic support [43,44]. Thus, our results from in vitro models show the functional involvement of HOTAIR in mitochondrial biogenesis; future validation of these results in in vivo models could potentially provide a therapeutic option to inhibit development of diabetic retinopathy and prevent its continued progression after the removal of hyperglycemic insult.

Author Contributions

J.K. participated in conducting the experiments and acquiring and analyzing the data. R.A.K. designed research studies, provided reagents, analyzed the data and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

R.A.K. has full responsibility for the integrity of the data and the accuracy of the data analysis. The author confirms that all data generated or analysed during this study are included in this published article. The raw data supporting the conclusions of this article will be made available by the authors on request.

Conflicts of Interest

J.K. and R.A.K. have no conflicts of interest.

Funding Statement

The study was supported in parts by grants from the National Institutes of Health (R01 EY 017313 and R01 EY 033516) and The Thomas Foundation grant to R.A.K., and an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology, Wayne State University.