Activation of Hsp90-eNOS and increased NO generation attenuate respiration of hypoxia-treated endothelial cells

Abstract

Hypoxia induces various adoptive signaling in cells that can cause several physiological changes. In the present work, we have observed that exposure of bovine aortic endothelial cells (BAECs) to extreme hypoxia (1–5% O₂) attenuates cellular respiration by a mechanism involving heat shock protein 90 (Hsp90) and endothelial nitric oxide (NO) synthase (eNOS), so that the cells are conditioned to consume less oxygen and survive in prolonged hypoxic conditions. BAECs, exposed to 1% O₂, showed a reduced respiration compared with 21% O₂-maintained cells. Western blot analysis showed an increase in the association of Hsp90-eNOS and enhanced NO generation on hypoxia exposure, whereas there was no significant accumulation of hypoxia-inducible factor-1α (HIF-1α). The addition of inhibitors of Hsp90, phosphatidylinositol 3-kinase, and NOS significantly alleviated this hypoxia-induced attenuation of respiration. Thus we conclude that hypoxia-induced excess NO and its derivatives such as ONOO⁻ cause inhibition of the electron transport chain and attenuate O₂ demand, leading to cell survival at extreme hypoxia. More importantly, such an attenuation is found to be independent of HIF-1α, which is otherwise thought to be the key regulator of respiration in hypoxia-exposed cells, through a nonphosphorylative glycolytic pathway. The present mechanistic insight will be helpful to understand the difference in the magnitude of endothelial dysfunction.

in tissues, hypoxia is the state of insufficient O₂, caused by inadequate transport or an excess consumption of O₂. Duration, frequency, and severity of hypoxia strongly influence whether the effect is detrimental or beneficial (20, 43). Interestingly, hypoxia-exposed cells do not always undergo cell death or diminish ATP levels (11), therefore leading to cell survival and normal cell function. Many factors that are activated upon hypoxia, such as the hypoxia-inducible transcription factor (HIF-1), heat shock protein 90 (Hsp90), nitric oxide (NO) synthase (NOS), and reactive oxygen species (11, 13), determine whether the cells survive after salvage or succumb to death (2). Particularly, the role of NO during and after hypoxia in the endothelium has been found to be important in maintaining oxygen metabolism (31). NO is a multifaceted endogenous factor and is involved in many pathophysiological processes in cells. In vitro studies have shown that NO production is increased in hypoxic conditions (9, 31). NO has been found to inhibit the mitochondrial electron transport chain (ETC) (8, 14, 15). Thus it is an endogenous modulator of cellular respiration in different pathogenic conditions. While NO restrains cytochrome-c oxidase (CcO) by competing for the oxygen binding site at the heme of the enzyme, its derivative, peroxynitrite (ONOO⁻), blocks the other ETC complexes mainly by S-nitrosation (25). Moreover, intermittent or acute hypoxia exposure is known to cause an increase in Hsp90 binding to endothelial NOS (eNOS), to facilitate Ser1177 phosphorylation (4, 13, 43). Thus, the excess NO in hypoxia that is caused by an increase in Hsp90-eNOS association may inhibit the ETC; yet, this hypothesis remains untested.

Hypoxia-induced HIF-1α can also attenuate cellular respiration (47). When there is a limited presence of oxygen, HIF-1α is induced and stabilized by inhibition of prolyl hydroxylation-dependent (PHD) binding of the ubiquitin ligase von Hippel-Lindau (pVHL) tumor suppressor (28). Moreover, Hsp90 has been reported to play a role in the stability of HIF-1α in low O₂ as well as heat-induced conditions (27, 28). Previous studies have shown that hypoxia-stabilized HIF-1α transcribes a set of genes that is related to glucose transporters (GLUT-1 and GLUT-3); hence the normal oxidative phosphorylation is slowly switched to glucose metabolism by an anaerobic lactate pathway (Warburg effect) in a hypoxic state (5, 6). Although this effect has been well established in cancer cells (47), recent reports have revealed that the cells that express high NOS, such as endothelial cells, show a different behavior in terms of HIF-1α stabilization. Mechanistic studies have established that even though PHD is inhibited at a low Po₂, the excess NO during hypoxia can induce PHD₂, which accelerates the HIF-1α (7). Thus, the role of HIF-1α in the regulation of hypoxia-treated endothelial cells (which have high abundance of NOS) is not known. Despite the two distinct possible mechanisms of inhibition of cellular respiration in hypoxia-treated endothelial cells, it is not known whether the respiration in hypoxia-exposed endothelial cells is regulated by a mechanism that is dependent on NO or HIF-1.

In the present work, we elucidate the mechanism of the regulation of oxygen consumption in hypoxia-exposed bovine aortic endothelial cells (BAECs). To measure cellular respiration, electron paramagnetic resonance (EPR) oximetry was used as a quantitative tool (24, 40). EPR oximetry is a highly sensitive technique that is accurate, requires only a microvolume of sample, and is capable of yielding high-resolution O₂ data similar to data obtained in high-resolution respirometry. In this work, we show that prolonged hypoxia exposure attenuates cellular respiration by the activation of the Hsp90-eNOS complex in BAECs. This activation was dependent on both duration and severity of hypoxia. Our results reveal that when BAECs undergo a prolonged hypoxia exposure, respiration is regulated by Hsp90-eNOS enhancement but not by HIF-1α-related factors, such as increased glycolysis.

MATERIALS AND METHODS

Materials

4,5-Diaminofluorescein diacetate (DAF-2DA) was purchased from Alexis Biochemicals (San Diego, CA). Dilithium phthalocyanine, acetonitrile, nitro-l-arginine methyl ester (l-NAME), and geldanamycin (GA) were obtained from Sigma-Aldrich (St. Louis, MO). Wortmannin was purchased from Millipore (Billerica, MA). Tetrabutyl ammonium perchlorate was purchased from ICN Biochemicals (Aurora, OH). The antibodies for Western blot analysis were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), Abcam (Cambridge, MA), and Cayman (Ann Arbor, MI).

Methods

Cell culture.

BAECs were obtained from Cell Systems (Kirkland, WA). The BAECs were cultured in MEM (GIBCO), 10% FBS, nonessential amino acid solution, and endothelial cell growth factor. Cells were grown in regular 150-cm² culture dishes, coated with attachment factor. Cells were trypsinized and used in experiments when the cultures reached 70–80% confluency.

Cell viability.

The cell viability was determined by a NucleoCounter system (New Brunswick Scientific, Edison, NJ) composed of the NucleoCounter automatic cell counter, the NucleoCassette, a cell preparation lysing buffer and a stabilizing buffer, and NucleoView software. Two aliquots of the cell suspension, for nonviable count and the total cell count, were taken. For the total cell count, equal amounts of the lysing buffer and the stabilizing buffer were added to the cell suspension. Each sample was loaded into the NucleoCassette and placed into the NucleoCounter cell counter for analysis. The nonviable count was determined first, followed by the total cell count. Using the NucleoView software, the nonviable, total cell count, viable cell count, and viability were determined. Viability of BAECs in suspension was found to be about 90% to 95%.

Hypoxia.

Cell suspension was prepared and seeded in normoxic (21% O₂) conditions and then transferred to a hypoxic incubator (prefixed with desired O₂ tension) as described in Figs. 1A and 2A. BAECs were placed into a hypoxic atmosphere, using an incubator preadjusted for a desired lower value (Thermo Electron Forma Series II Water Jacketed CO₂ Incubator). The sensor in the incubator precisely measured the Po₂ inside the incubator and displayed the Po₂ along with CO₂. The hypoxic environment contained a gas mixture of 5% CO₂-94% N₂-1% O₂ (or 92% N₂-3% O₂ or 90% N₂-5% O₂) for various time periods. Following hypoxia, the cells were trypsinized for experiments. The confluence was observed to vary between 70% and 95%, depending on the severity of the hypoxia.

Fig. 1.

Hypoxia exposure and cellular respiration. Effect of different percentages of O₂. A: schematic illustration of hypoxia exposure of cells. The cells were seeded in normoxic conditions and taken to normoxic (21% Po₂) or hypoxic incubator with preset Po₂ (5%, 3%, or 1% Po₂) and cultured for desired time as indicated. At the end of hypoxic treatment, the cells were taken out and suspended in normoxic respiration medium, and electron paramagnetic resonance (EPR) oximetry was performed on a fixed cell density (5 × 10⁶ cells/ml). B: EPR spectra were obtained for both hypoxic and control cells, and the EPR line width data were converted into Po₂ data. As the percentage of O₂ reduces, the cells seem to require more time to consume O₂. C: data from B were converted into rate of oxygen consumption (dPo₂/dt) and plotted with respect to Po₂. Three distinct phases of respiration can be observed. The 1% O₂ cells have a substantial decrease in the maximum oxygen consumption rate.

Fig. 2.

Hypoxia exposure and cellular respiration. Effect of duration of hypoxia. A: experimental setup where bovine aortic endothelial cells (BAECs) were cultured at 1% O₂ for 2–24 h. B: EPR spectra were obtained for both hypoxic and control cells (5 × 10⁶ cells/ml), and the EPR line width data were converted into Po₂ data. At acute exposure of 1% O₂, no difference is seen in comparison to the control. C: data from B were converted into dPo₂/dt vs. Po₂. Intermittent hypoxia reveals no significant variation in the V̇o_2max; however, a clear attenuation is displayed at 24 h of prolonged hypoxia.

Mitochondrial membrane potential measurements by flow cytometry.

Rhodamine 123 (Molecular Probes, Eugene, OR) was used to measure mitochondrial membrane potential using the procedure described previously (38). Both 21% O₂-maintained cells (control) and hypoxia-exposed cells (1% O₂ for 24 h) were trypsinized and counted using the NucleoCounter system. Equal number density (5 × 10⁶/ml) cell suspensions were obtained by resuspension in MEM medium containing 10% FBS and 10 μg/ml rhodamine 123 and incubation for 10 min. After incubation, the cell suspensions were pelleted and resuspended in ice-cold MEM-10% FBS medium and subjected to flow cytometry, using BD FACSCalibur. Fluorescence values of all 10,000 cells were displayed in log normal distribution curves and converted to quadrant plots.

Glucose uptake by BAECs.

Both control and hypoxia-treated BAECs were incubated overnight with media containing a fluorescent, noncleavable glucose analog {2-[n-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino-2-deoxyglucose]; 2-NBDG; Invitrogen; 0.5 mg is dissolved in 15 ml MEM medium}(16). The cells were trypsinized, washed with MEM medium, counted, and finally suspended in the MEM medium. The relative fluorescence intensity was measured using the FACSCalibur flow cytometer, and histograms were analyzed using WinMDI software.

Drug treatments.

To understand the effect of Hsp90 and eNOS on hypoxic respiration, various inhibitors of each protein were used. BAECs were treated with GA (10 μM), l-NAME (0.5 mM), or wortmannin (1 μM) before or following hypoxia. In the first set of experiments, cells were placed in a 1% O₂ hypoxic environment for 24 h and treated with each respective inhibitor for 30 min. In the second condition, each drug was added to the cultured cells for 30 min and placed in 1% O₂ for 24 h. Following each experimental condition, the cells were trypsinized for experiments.

Oximetry probe.

Lithium phthalocyanine (LiPc) microcrystals were used as the oximetry probe. The probe was synthesized electrochemically, using the established procedure (23). The synthesized microcrystals were subjected to various physicochemical characterizations, such as X-ray diffraction, EPR, and microscopy, to ensure the purity of the material. These microcrystals were found to be in the pure λ-isoform (or equivalently known as X-form), which has been characterized to yield a Po₂-dependent EPR line width. For oximetry measurements, we used approximately 20–30 μg of LiPc.

Measurement of cellular respiration.

The oxygen measurements were performed using EPR oximetry (21, 24, 40). From the EPR line width, the Po₂ in the cell suspension was determined using the calibration curve. The EPR line width vs. Po₂ calibration curve was constructed using known ratios of premixed O₂ and N₂ gases. The slope of the calibration curve was 5.8 mG/mmHg. Although this calibration curve was constructed using gas mixtures, we have previously demonstrated that this curve is applicable in aqueous solutions as well (21). Thus, by measuring the EPR line width, the Po₂ in the solution can be obtained at any given time. LiPc measures the extracellular Po₂ in the cellular suspension. Since the LiPc microcrystals were comparatively larger in the oximetry measurements, the particulates remained in the bulk volume and there was no ingestion by the cells.

EPR oximetry experimental setup.

The respiration studies have been carried out using an X-band (9.7 GHz) EPR spectrometer fitted with a TM₁₁₀ microwave cavity. A 50-μl microcapillary tube was used to hold the cells in the horizontal EPR cavity. In a typical experiment, the cell suspension of the desired cell density was maintained in respiration medium (in mM: 117.3 NaCl, 4.7 KCl, 1.3 MgSO₄, 1.2 CaCl₂, 1.2 KH₂PO₄, 25 NaHCO₃, and 20 glucose; pH 7.4) and saturated with room air (Po₂ ≈ 160 mmHg). The cell suspension was incubated for 10 min in a 37°C water bath. LiPc microcrystals (20 μg) were added to the cells and sampled into 50-μl capillary tubes. The tube was then closed off at both ends with tube sealing clay (Chase Scientific Glass, Rockwood, TN). While sealing, care was taken to ensure that there were no air gaps present inside the tube, since such a gap may act as an additional source of O₂. The tube was placed inside the horizontal microwave cavity, and EPR spectral acquisitions of the LiPc were immediately started. During measurements, the modulation amplitude was adjusted to always be less than one third of the line width to avoid modulation-induced broadening.

O₂ kinetics.

Quantitative EPR oximetry was performed using recently described procedures (40). Briefly, there are three phases of cellular respiration that can be analyzed from a single run of Po₂ vs. time using EPR oximetry: Po₂-dependent, Po₂-independent, and a steady-state respiration. These levels of cellular respiration were obtained by adopting the following equation:

(1)

From this equation, V̇o_2max, p₀, and p50 values were acquired (40). The V̇o_2max is defined as the oxygen consumption rate (V̇o₂) in coupled state, when oxygen is not limiting; p₀ is the equilibrium Po₂; and p50 is the concentration at which the V̇o_2max is reduced to 50%. This half-maximum value is analogous to the K_m value in enzymatic reactions and provides an indication of the oxygen affinity. Specifically, p50 is the inverse of the mitochondrial oxygen affinity to CcO in complex IV of the ETC. Since LiPc measures the extracellular Po₂ around each cell, the p₀ provides an indication of the potential intracellular O₂ content (40).

Western blot analysis.

Cells were washed twice with ice-cold PBS, trypsinized, and centrifuged at 1,500 rpm for 5 min. The cell pellet was homogenized in ice-cold RIPA buffer (1× Tris-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.004% sodium azide, 1× protease inhibitor, 1 mM PMSF, and 1 mM sodium orthovanadate) for 45 min in ice. The protein concentrations of the supernatants were measured by the bicinchoninic acid method and normalized to 25 μg per sample. The samples were resolved on 4–12% Bis-Tris polyacrylamide gels and transferred to polyvinylidene difluoride (PVDF) membrane at 45 V for 2 h. After blocking with 5% nonfat milk, blots were probed with a rabbit anti-Hsp90, anti-eNOS, or HIF-1α antibody (1:1,000 dilution). For HIF-1α determination, both 21% and 1% O₂-treated cells were lysed and denatured in respective Po₂-containing glove boxes. The denatured proteins were used to resolve in PAGE gels in normal 21% O₂ environments. Goat anti-rabbit horseradish peroxidase-conjugated antibody was used as the secondary antibody, and blots were developed with enhanced chemiluminescence.

Immunoprecipitation.

The total cell lysates were prepared as described in Western blot analysis and incubated with either anti-eNOS or anti-Hsp90 polyclonal antibody overnight at 4°C while rotating. To immunoprecipitate eNOS or Hsp90, the protein A/G agarose was added to the lysates and rotated at 4°C for 2 h. The immunoprecipitates were centrifuged at 10,000 rpm for 30 s at 4°C. The supernatant was carefully aspirated and discarded. The pellet was washed with 500 μl RIPA buffer three times and centrifuged at 10,000 rpm for 30 s at 4°C. After the final wash, the supernatant was removed and the pellet was suspended in 40 μl sample buffer. The samples were boiled at 98°C for 8 min and subjected to electrophoresis. The PVDF membrane was immunoblotted with anti-Hsp90 or anti-eNOS to determine the amount of association of Hsp90-eNOS.

Fluorescence microscopy.

NO production in BAECs was analyzed using fluorescence microscopic imaging with an inverted light Nikon TE2000-U microscope. DAF-2DA, a green fluorescence NO-specific probe, was used. The cells were suspended in serum-free medium, and a 10 μM concentration of DAF-2DA was added directly to the medium of the control and hypoxic cells. They were incubated at 37°C for 20 min and washed twice with PBS. The fluorescence microscopy measurements were immediately performed. MetaMorph software was used to calculate the average fluorescence intensity of individual cells.

DNA laddering.

BAECs were cultured in a 75-cm² flask using regular MEM medium supplemented with 10% FBS, nonessential amino acids, growth factor, and antibiotic. The cells were treated with either GA (10 μM) or wortmannin (1 μM), trypsinized, washed with the medium, and finally suspended in 200 μl PBS. The DNA from these cells is isolated using the Qiagen DNeasy kit. Finally, the extracted DNA was loaded and run on the 2% agarose gel containing ethidium bromide, and the bands were observed under ultraviolet illumination.

Curve fit and data analysis.

Data are presented as means ± SE. Statistical analysis was performed using Student's t-test and one-way ANOVA. The general acceptance level of significance was P < 0.05. The EPR spectra, collected during the cellular respiration measurements, were analyzed as formerly described (40). The correlation coefficient of 0.98 was set as the standard of acceptance of the results. The Po₂ data conversion, differentiation, and curve fit were carried out as described previously (40).

RESULTS

The Effect of Hypoxia on BAEC Respiration

BAECs were cultured at various O₂ contents, namely, 21%, 5%, 3%, and 1% for 24 h, as shown in Fig. 1A. At the end of hypoxia exposure, the cells were trypsinized and resuspended in the respiratory medium for oxygen consumption measurements. The oxygen consumption of BAECs was followed by EPR oximetry. An equal density (5 × 10⁶ cells/ml) of cells was used for respiration measurements in each case presented in Fig. 1, B and C. EPR spectra were obtained at 15-s intervals for a period of 60 min. There was no permanent damage to the cells, noticed after 60 min of measurements, as determined by cell viability measurements described in materials and methods. The EPR spectrum of the LiPc, cosuspended in cell suspension within the sealed microtube, narrowed down with time due to the oxygen consumption by the BAECs. The EPR line width of each obtained spectrum was converted into Po₂ data by using a standard calibration curve and was plotted with respect to time. The acquired results are shown in Fig. 1B for the cells that were exposed to 5%, 3%, and 1% O₂ for 24 h, along with the cells that were constantly maintained at 21% O₂ (normoxia) as the control. These data were further transformed into rate of oxygen consumption (dPo₂/dt) vs. Po₂ plots (40), and the results are shown in Fig. 1C. It is evident from Fig. 1C that each curve demonstrates three phases of respiration: a maximum consumption rate (constant rate, V̇o_2max), which is independent of Po₂ (>15 mmHg); a Po₂-dependent consumption rate (curved portion <15 mmHg); and no consumption (the intercept in x-axis) at a residual Po₂ (equilibrium, p₀, at <1 mmHg). For quantitative analysis, the data in Fig. 1C were analyzed by fitting into an appropriate equation as described in materials and methods, and the relevant parameters were obtained. For the cells that were exposed to 5% O₂ for 24 h, there was no apparent difference in the maximum consumption rate (V̇o_2max) and the p50, compared with normoxia (21% O₂)-exposed cells (Fig. 1, B and C). However, in the cases of 3% and 1% O₂-exposed cells, the respiration was significantly attenuated (Fig. 1, B and C). Overall, comparing each Po₂ at 24 h, there is a significant decrease at 1% O₂ (Fig. 1C). The control cells reached a V̇o_2max of 4.07 ± 0.18 mmHg·min⁻¹·5 × 10⁻⁶ (p50 = 2.86 ± 0.25 mmHg, n = 7) and the 1% O₂-treated cells showed a significantly lower V̇o_2max of 2.44 ± 0.45 mmHg·min⁻¹·5 × 10⁻⁶ (p50 = 2.44 ± 0.33 mmHg, n = 3), demonstrating nearly a twofold reduction in the overall maximum rate (Fig. 1C and Table 1). This behavior reveals that exposure to extreme hypoxia, such as 1% O₂, attenuates cellular respiration, whereas moderate hypoxia, such as 5% O₂, does not significantly affect respiration.

Table 1.

Quantitative analyses of EPR oximetry measurements, in which 5 x 10⁶ cells were treated with various O₂ concentrations at a range of time points

Time	3% O2					1% O2
	V̇o2max, mmHg/min	p0, mmHg	p50, mmHg	1/p50, mmHg−1	n	V̇o2max, mmHg/min	p0, mmHg	p50, mmHg	1/p50, mmHg−1	n
2 h	5.72±0.73	0.35±0.06	3.90±0.41	0.26±0.03	3	4.22±0.57	0.40±0.09	2.86±0.33	0.35±0.04	3
4 h	4.65±0.18	0.49±0.13	3.27±0.13	0.30±0.01	3	3.76±0.19	0.44±0.07	2.60±0.24	0.39±0.04	3
8 h	4.43±0.46	0.44±0.09	3.42±0.34	0.30±0.03	3	4.08±0.24	0.44±0.16	2.99±0.12	0.34±0.01	3
16 h	3.45±0.34	0.45±0.16	2.70±0.16	0.37±0.02	3	2.92±0.52	0.45±0.10	3.82±0.76	0.27±0.05	3
24 h	3.01±0.69	0.45±0.09	2.19±0.28	0.47±0.05	3	2.44±0.45	0.46±0.03	2.44±0.33	0.42±0.06	3

Values are means ± SE. The control values are as follows (n = 7): V̇o_2max = 4.07 ± 0.18 mmHg/min; p₀ = 0.37 ± 0.03 mmHg; p50 = 2.86 ± 0.25 mmHg; and 1/p50 = 0.36 ± 0.03 mmHg⁻¹. EPR, electron paramagnetic resonance.

The effect of time of hypoxia exposure on respiration was studied. Since a substantial attenuation in respiration was seen only in extreme hypoxia (i.e., when the cells were exposed to 1% O₂; Fig. 1), these studies were restricted to 1% O₂. The time of hypoxia exposure was varied from 2 to 24 h as shown in Fig. 2A. Following, the cells were trypsinized, resuspended, and subjected to respiration measurements as described above. EPR data were acquired; the corresponding Po₂ values were determined and plotted with respect to time (Fig. 2B). From the Po₂ vs. time results, dPo₂/dt vs. Po₂ was established, as shown in Fig. 2C. The data were analyzed as previously described, and the results are summarized in Table 1. No significant changes in the respiration parameters were observed for the cells that were exposed for 2 and 8 h of 1% O₂ (Fig. 2C). However, as the hypoxia exposure time increased to 16 h or longer, a significant attenuation in the respiratory rate was observed (Table 1). At 2 h of 1% O₂, the maximum O₂ rate of 4.22 ± 0.57 mmHg·min⁻¹·5 × 10⁻⁶ cells is analogous to the control value. Similarly, corresponding p50 values of 2.86 ± 0.33 mmHg for control and p50 = 2.99 ± 0.12 mmHg for 2 h and 8 h at 1% O₂ exist (Table 1). As expected, the maximum rate of decline is exemplified in the cells cultured at 24 h of 1% O₂ (Fig. 2C and Table 1).

HIF-1α Stability in Hypoxia-Treated Endothelial Cells

The observed attenuation of respiration in 1% O₂-exposed BAECs could be either HIF-1α-induced glycolysis (47) or an Hsp90-eNOS-induced, NO-triggered mechanism (13, 43). Thus, experiments were carried out to determine whether HIF-1α is stabilized in our experimental conditions. BAECs were cultured in normoxia (21%) and exposed to hypoxia (1% O₂) for 24 h. The cells were trypsinized and lysed in the respective O₂ environments, and Western blot analysis was carried out. Figure 3A shows a set of representative Western blots of whole cell lysates probed for HIF-1α and corresponding β-actin. Figure 3B shows the quantitative plots of HIF-1α blot density, normalized to corresponding β-actin. There was no significant difference in the expression of HIF-1α at each condition, revealing that the accumulation of HIF-1α is prevented in hypoxia-exposed BAECs cells. This observation correlates with earlier reported studies. Previously, hypoxia was found to activate eNOS and increase the production of NO (26, 43). On the other hand, NO has been found to induce PHD₂, which will effectively increase HIF-1α degradation, even if PHD₁ is inhibited (7). Thus the observed attenuation in respiration is not due to increased HIF-1α accumulation and subsequent increase of glycolysis (nonoxidative phosphorylation). Furthermore, we determined the levels of glucose transporters, namely, GLUT-1 and GLUT-2, which are known to be transcribed by HIF-1α. Figure 3 shows Western blots and the corresponding quantitative analyses. There was no significant change in these proteins in 21% and 1% O₂-treated cells. Glucose uptake of normoxia and hypoxia-treated cells was also determined using 2-NBDG, a fluorescent tagged glucose. The 2-NBDG uptake by these cells was found to be the same (Fig. 3C), indicating that the functional levels of GLUT-1 and -2 are the same both in control and in hypoxia-treated cells. These results together prove that the observed attenuation of respiration is not due to higher glycolysis in 1% O₂-exposed cells. Considering these facts, additional experiments were carried out to determine whether the activation of eNOS is responsible for the observed attenuation in the hypoxia-exposed BAECs.

Fig. 3.

Hypoxia-inducible factor (HIF)-1α and glucose transporter (GLUT)-1 and GLUT-2 levels in hypoxia-exposed endothelial cells. BAECs were treated either at 21% O₂ or at 1% O₂ for 24 h and were then trypsinized and analyzed for HIF-1α, GLUT-1, and GLUT-2. A: representative Western blots. In the case of HIF-1α, jurkat cell lysate treated with 0.1 mm CaCl₂ was loaded as reference. B: quantitative plots of 3 independent measurements, normalized with respect to β-actin. There was no significant change in the levels of HIF-1α, GLUT-1, and GLUT-2 in 1% O₂-exposed cells compared with 21% O₂-maintained cells. AU, arbitrary units. C: FACS analyses of glucose uptake. Shown are histograms of both control cells (black lines) and hypoxia-exposed cells (red lines) that were incubated with (solid lines) and without (broken lines) 2-NBDG.

Increased Production of NO in Hypoxia-Treated Cells

Observing that there was no significant accumulation of HIF-1 in hypoxia treated BAECs, NO generation was measured with fluorescence microscopy at each time exposure to 1% O₂ hypoxia. DAF-2DA staining, which yields green fluorescence by reacting with NO and its derivatives, was used to quantify NO generation on hypoxia exposure. BAECs were cultured in glass cover slides in normoxia (21% O₂) and hypoxia (1% O₂) for 24 h and treated with DAF-2DA for 20 min. Figure 4, A and B, illustrates the fluorescence images of control and hypoxic cells stained with DAF-2DA. When compared with the control BAECs, a distinct intensification in fluorescence can be seen in the hypoxia-exposed cells. The quantitative determination of fluorescence intensity was carried out and is illustrated in Fig. 4C. On the basis of the intensity measurements, the cells that were exposed to 24 h of 1% hypoxia had an average intensity of 722.8 ± 23.6 arbitrary units (AU; n = 12), whereas the control BAECs had an average intensity of 186.5 ± 6.9 AU (n = 24), demonstrating nearly a fourfold increase (Fig. 4C). These data confirm that NO generation is greatly augmented in hypoxic cells versus normal BAECs. Further studies were carried out to determine whether the higher NO results in nitration of proteins in 1% O₂-exposed cells. Figure 4D shows the Western blots of nitrotyrosine proteins in whole cell lysates, probed with 3-nitrotyrosine-specific antibodies. Higher tyrosine-nitrated proteins are observed in hypoxia-treated samples than in 21% O₂-maintained cells, indicating that the protein nitration in mitochondrial proteins may be responsible for the observed attenuation of respiration. However, treating hypoxia-exposed cells with l-NAME (NOS inhibitor) reduced the nitrosylation of proteins (Fig. 4D).

Fig. 4.

Fluorescence microscopic analysis of nitric oxide (NO) production and 3-nitrotyrosine. Both control cells (A) and 24 h 1% O₂ hypoxia-treated cells (B) were stained with 4,5-diaminofluorescein diacetate (DAF-2DA). The cells were incubated at 37°C for 20 min, and images were taken under identical conditions (×20 magnification). C: quantitative fluorescence image intensity averaged from intensities from individual cells (n = 24 for control; n = 12 for 1% O₂ for 24 h). *P < 0.05 vs. control. D: representative Western blots of tyrosine-nitrated cellular protein, probed with 3-nitrotyrosine antibody, in normoxic (21% O₂), hypoxic (1% O₂), and hypoxic + 100 μM nitro-l-arginine methyl ester (l-NAME)-treated cells. Molecular mass (in kDa) is indicated at right. E: flow cytometry of rhodamine 123 (Rhod123)-stained normoxia- and hypoxia-treated cells.

Hsp90-eNOS Association in Hypoxic BAECs

The interaction of Hsp90 and eNOS has been reported to be enhanced in stressful conditions such as heat shock and hypoxia, in which this association leads to an increase in NO production (13, 22). To determine if Hsp90-eNOS association is an important factor in the attenuation of respiration in hypoxia-exposed cells, the induction and association of Hsp90 and eNOS and its activity were studied. BAECs were exposed to 5%, 3%, and 1% O₂ for the same times as used in EPR oximetry studies of respiration (Fig. 1). At the end of hypoxia exposure, the cells were trypsinized, lysed, and used for either Western blot analysis or immunoprecipitation (IP). Two sets of experiments were carried out: In the first set, eNOS was immunoprecipitated and Hsp90 and eNOS were immunoblotted (IB); in the second set, IP of Hsp90 and IB of eNOS and Hsp90 were performed. In both cases of 5% and 3% O₂-exposed cells, there was no change in the expression of Hsp90 or eNOS at each time of treatment (data not shown). These results are consistent with the respiration measurements that no significant changes occurred in the respiration by exposing the cells to 5% or 3% O₂. On the other hand, a significant difference in the expression of Hsp90 and eNOS, as well as the IP of Hsp90, obtained from IB of eNOS, was observed for cells treated at 1% O₂. As seen in Fig. 5A, the induction of both Hsp90 and eNOS increased up to 16 h of 1% O₂ hypoxia exposure, followed by a moderate decline at 24 h. Likewise, there is an increased expression and association of Hsp90 to eNOS, between 8 and 16 h of exposure to 1% O_2, and a slight decrease at 24 h (Fig. 5B). This increased association may contribute to the increased NO production in hypoxia-exposed cells. Such an increased NO production may cause both reversible and irreversible inhibition of the ETC, resulting in attenuation of respiration. Also in the case of hypoxia-treated cells, the confluence was less, indicating that the cells are still actively proliferating. This is in agreement with previous reports that such active proliferation increases Hsp90/eNOS association in endothelial cells (36). Thus it is difficult to decide whether the observed activation of Hsp90/eNOS association is due to hypoxia alone. However, when compared with control (which had also not reached 100% confluency at the time of the experiment), the reduced respiration of hypoxia-treated cells is likely due to hypoxia. Moreover, it is also possible that detaching the cells from culture dishes may promote Hsp90/eNOS and trigger apoptosis. However, the DNA laddering studies did not show any significant fragmentation of DNA (Fig. 7B), confirming that there was no apoptosis due to detachment of cells.

Fig. 5.

Immunoblot analysis of endothelial NO synthase (eNOS) and heat shock protein 90 (Hsp90) in hypoxia-treated BAECs. A: immunoblots (IB) of Hsp90 and eNOS in the immunoprecipitates of eNOS in control (Con) and 2–24 h 1% O₂ hypoxia-treated BAECs. The quantitative plots were obtained from 3 sets of immunoprecipitation and immunoblotting (n = 3). The expression of Hsp90 and eNOS is enhanced during hypoxia in comparison to the control. The bar graphs show increased expression of eNOS on exposure to hypoxia. Also, the ratio of Hsp90 to eNOS increases with an increase in hypoxia time, indicating that the association of Hsp90 with eNOS increases with an increase in hypoxia time. B: immunoblots of eNOS and Hsp90 in the immunoprecipitates of Hsp90 in control and 2–24 h 1% O₂ hypoxia-treated BAECs (n = 3). Similar to A, the bar graphs show that Hsp90 increases slightly on exposure to hypoxia and the association between eNOS and Hsp90 increases. However, there is a decrease in expression at 24 h. *P < 0.05 vs. control.

Mitochondrial integrity was determined to infer whether the observed attenuation in respiration is due to physical or functional changes of mitochondria. Rhodamine 123 uptake, a measure of mitochondrial membrane potential (38), was determined for control and 1% O₂-treated cells, and the obtained results are included in Fig. 4E. The rhodamine fluorescence measured, using flow cytometry, was not different for the samples, and, indeed, both samples showed >99% positive staining for rhodamine 123. This result indicates that the observed attenuation in respiration in hypoxic cells is not due to physical changes and is indeed due to functional changes in mitochondria. This is most likely due to the nitration of mitochondrial proteins, as observed in Fig. 4D.

Alleviation of NO-Induced Inhibition of Respiration in Hypoxia-Exposed Cells

To ensure the role of Hsp90 and eNOS in the observed attenuation of respiration in 1% O₂-exposed cells, the effects of various inhibitors of Hsp90, phosphatidylinositol 3-kinase (PI3K), and NOS on the respiration of hypoxia exposed cells were studied. GA was used as the Hsp90 inhibitor, l-NAME as the NOS inhibitor, and wortmannin as the PI3K inhibitor. Wortmannin is a cell-permeable irreversible inhibitor of PI3K that blocks the catalytic activity of PI3K without influencing upstream signaling events. Two sets of experiments were carried out with these inhibitors. In the first set of experiments, BAECs were treated for 30 min with 10 μM GA, 1 μM wortmannin, or 0.5 mM l-NAME and were returned to regular culture medium and maintained for 24 h at 1% O₂. The cells were trypsinized and resuspended in regular respiration medium for oxygen consumption measurements. The oxygen consumption of BAECs was followed by EPR oximetry as described above. Briefly, EPR spectra were obtained at 15-s intervals for a period of 90 min for 5 × 10⁶ cells/ml. The Po₂ data were obtained, plotted with respect to time (Fig. 6A), and further transformed into dPo₂/dt vs. Po₂ data (Fig. 6B) (40). In Fig. 6, A and B, it is clear that the cells treated with GA have an enhanced maximum respiration rate in comparison to the hypoxic cells. The V̇o_2max was recovered from 1.96 ± 0.35 mmHg·min⁻¹·5 × 10⁻⁶ cells to 3.23 ± 0.43 mmHg·min⁻¹·5 × 10⁻⁶ cells; n = 2 (Table 2). Similarly, the addition of l-NAME also significantly increased the V̇o_2max value; however, there was no relevant difference with the treatment of wortmannin (Fig. 6, C and D). Although GA and l-NAME recovered the maximum O₂ rate, there was no significant change in the p50 or mitochondrial affinity between all of the inhibitors. A complete analysis of the data was performed as described in materials and methods, and the relevant parameters have been summarized in Table 2. A second set of experiments was done in which the cells were cultured at 1% O₂ for 24 h and then treated for 30 min with each drug treatment before trypsinization. There was no significant difference between the control cells and the BAECs treated with the inhibitors, implying that the NO inhibition at this stage was not effective to recover respiration (data not shown).

Fig. 6.

Influence of inhibiting Hsp90 and eNOS on respiration. BAECs were treated with geldanamycin (GA; 10 μM), l-NAME (0.5 mM), or wortmannin (1 μM) for 30 min and placed in a 1% O₂ hypoxic environment. EPR spectra were acquired and transferred into Po₂ data. A: to inhibit Hsp90, GA was used, where GA-treated cells consume O₂ faster than 1% O₂-exposed cells alone. B: data from A were converted into dPo₂/dt vs. Po₂. There is a clear restoration of respiration close to the control value in the presence of GA. C: to inhibit NOS and phosphatidylinositol 3-kinase (PI3K), BAECs were treated with l-NAME and wortmannin, respectively. l-NAME-treated cells consume O₂ close to the control value, whereas there is no significant difference with wortmannin treatment. D: there is a clear restoration of respiration close to the control value in the presence of l-NAME; yet, the O₂ consumption of wortmannin-treated BAECs is similar to the 1% O₂ rate.

Table 2.

Quantitative analyses of EPR oximetry measurements, in which 5 x 10⁶ cells were incubated with each inhibitor for 30 min before 1% O₂ hypoxia for 24 h

Drug Treatment	V̇o2max, mmHg·min−1·5 × 10−6 cells	p0, mmHg	p50, mmHg	O2 Affinity, mmHg−1	n
Geldanamycin	3.23±0.43*	0.61±0.29	2.89±0.40	0.35±0.09	2
Radicicol	3.03±0.35	0.62±0.23	2.86±0.38	0.35±0.05	3
l-NAME	2.88±0.47*	0.64±0.17	2.35±0.32	0.43±0.05	2
Wortmannin	2.23±0.31	0.62±0.25	3.00±0.29	0.33±0.08	2
None	1.96±0.35	0.59±0.19	2.43±0.35	0.41±0.07	3

Values are means ± SE. l-NAME, nitro-l-arginine methyl ester.

^*P < 0.05, significant tests compared with the 24-h 1% O₂ cells.

Finally, experiments were also carried out to determine whether there was any significant difference in Hsp90 binding to eNOS upon treatment with these inhibitors. BAECs were treated with each inhibitor for 30 min before 24 h of 1% O₂ and returned to regular medium. Cells were trypsinized and lysed, and the protein estimation was carried out. Normalized cell lysates were used to measure the association of Hsp90 and eNOS for each condition, and the results are summarized in Fig. 7. In comparison to the control, there is an increased association of Hsp90 to eNOS for the cells that were exposed to 1% O₂ for 24 h. The addition of GA slightly reduced the Hsp90 immunoblotted from the IP of eNOS (Fig. 7). Knowing that the PI3K/Akt pathway can interact with Hsp90, BAECs were treated with wortmannin. Wortmannin blocked the association of eNOS and Hsp90. l-NAME, a NOS inhibitor, also exhibited the same effect of suppressing the Hsp90 and eNOS interaction (Fig. 7).

Fig. 7.

Western blot analysis of the inhibition of extreme hypoxia. BAECs were treated with GA (10 μM), l-NAME (0.5 mM), or wortmannin (1 μM) for 30 min and placed in a 1% O₂ hypoxic environment. A: immunoblots of Hsp90 and eNOS obtained in the immunoprecipitates of eNOS. The bar graphs represent the ratio of Hsp90 to eNOS in respective samples that were treated with GA; wortmannin (Wort) reduced the association (**P < 0.001). The quantitative plots were obtained from 3 sets of immunoprecipitation and immunoblotting (n = 3). B: DNA laddering analysis in GA- and wortmannin-treated control and hypoxia-treated cells.

DISCUSSION

The primary finding of the present work is that prolonged hypoxia exposure of BAECs attenuates respiration by a mechanism involving Hsp90 and eNOS. More importantly, this reduced respiration in endothelial cells appears to be independent of HIF-1α, which is otherwise thought to be a key factor influencing respiration. Earlier reports have demonstrated reduced respiration, caused by HIF-1, CcO, and AMP kinase, in hypoxia-treated cells such as hepatocytes and carcinomas (10, 12, 30, 47). Although hypoxia has been previously shown to increase eNOS mRNA and protein expression, as well as augment basal and bradykinin-stimulated NO production in BAECs (20), to our knowledge, the present work is the first to reveal the influential role of Hsp90 and eNOS in posthypoxic cellular respiration. A number of experiments were carried out, in which BAECs were exposed to various percentages of O₂ in the range of normoxia (21%) to extreme hypoxia (1%) for different durations. Under these conditions, the expression of eNOS and Hsp90 and their association and NO production were determined. Together, these results prove the proposed mechanism that the activation and association of Hsp90 and eNOS increase NO under hypoxia, leading to an observable attenuation of respiration. Overall, no accumulation of HIF-1α was found upon exposure of BAECs to hypoxia. However, a significant upregulation of the association of Hsp90 with eNOS occurred (Fig. 5), and increased NO generation (Fig. 4) was observed. The exposure of endothelial cells to hypoxia increases eNOS activity due to Hsp90-assisted increase in phosphorylation at Ser1177 by PI3K/Akt pathway, and an increase in NO generation occurs. Such an upregulation of NO in cells is known to induce adaptation of cells to the reduced oxygen content. In endothelial cells, eNOS phosphorylation at Ser1177 is necessary for this hypoxia-induced eNOS activation and NO production (13). Other factors have also been explored and shown to contribute to this attenuated respiration, such as CcO, ATP utilization, HIF-1, and AMP kinase; however, the role of Hsp90 and eNOS in cellular respiration in a hypoxic environment remains unclear (10, 12, 30, 47).

The mechanism of NO-induced inhibition of respiration has been well elucidated by various groups (1, 8, 14, 15). NO has been found to exert two distinct types of inhibition on cellular ETC, namely, irreversible inhibition due to chemical modifications caused in the ETC complexes by ONOO⁻ and reversible inhibition at CcO of complex IV. Moreover, CcO has also been recognized as the mitochondrial enzyme that reduces NO₂⁻ to NO (9). The reversible inhibition of CcO is due to direct competition of NO with O₂ at the O₂ binding site of CcO (32). In our model expression described in materials and methods, the V̇o_2max decrease is attributed to the overall inhibition, and the increase in p50 is attributed to the reversible inhibition at CcO. Although there was a relevant change in the V̇o_2max for the hypoxia-exposed BAECs, there was no considerable difference in the p50 values (Po₂ at which the V̇o_2max is half). Such a behavior has been previously observed for BAECs, where eNOS was activated by various stimulators (32). We have found that there is no CcO inhibition at low Po₂ values (even in eNOS-activated conditions), because there is not an adequate amount of O₂ at a low Po₂ to generate NO by NOS and hence there is no reversible inhibition at CcO (unpublished observation). It appears that a similar effect seems to be caused by the excess flux of NO generated during hypoxic treatment and its reactions with ETC complexes, meaning that the irreversible damage is responsible for the observed attenuation of respiration. However, the data in Fig. 4 show higher fluorescence intensity, which is likely due to nonspecific staining of NO and derivatives such as ONOO⁻. This argument is further supported with the results of various inhibitors used in the present work. When the cells were treated with l-NAME and GA during hypoxic exposure (even as low as 1% O₂), the V̇o_2max was unaffected compared with the cells maintained in normoxia (21% O₂) (i.e., the attenuation of respiration is prevented; Fig. 6). Yet, the addition of these agents following hypoxia and immediately before respiration measurements did not eliminate the attenuation of respiration (data not shown). While l-NAME indiscriminately inhibits all of the isoforms of NOS, the GA binds to Hsp90, preventing the eNOS association with Hsp90. Thus it appears that the increased binding of eNOS with Hsp90 during hypoxia is essential for the observed attenuation of respiration (Fig. 2) in the hypoxia-treated cells. This association leads to the increased phosphorylation of eNOS and generation of NO, which can potentially induce PHD₂ for HIF-1 degradation and will inhibit the ETC. Further results of Western blotting (in terms of time course) also supported the proposed mechanism of hypoxia-induced NO upregulation. Between 8 and 16 h of hypoxia, there is a clear increase in expression of both Hsp90 and eNOS, as well as an enhanced association of Hsp90 and eNOS (Fig. 5). This directly correlates to the trend in the attenuation of respiration observed in similar experimental conditions (shown in Fig. 2).

Previously, several studies have correlated the magnitude of the Hsp90-eNOS association and higher NO generation to protective effects; yet, none of them reported respiration measurements, especially for hypoxia-treated cells (37). Castello et al. (9) found that NO production began when the oxygen concentration dropped below a 2% dissolved O₂ concentration and maximized when O₂ was not present. Shi et al. (43) have suggested that the advantages of chronic hypoxia are more closely related to how much Hsp90 associates with eNOS than to the magnitude of eNOS phosphorylation at Ser1177 alone (43). The generation of NO in central and peripheral neurons is increased during chronic hypoxia. Because of increased production of NO, the animals may develop a tolerance to the low Po₂ environment (39). While nearly all of these studies have proposed NO upregulation in hypoxia-exposed cells, its role in cellular respiration in the posthypoxic phase was never considered before. For example, the relationship of eNOS and Hsp90 was reported to decline in the hypoxic pulmonary artery injury (33, 35). Other studies have established that NO production is correlated to eNOS activity at the posttranslational level and not from the eNOS protein expression alone (33, 35). Thus, activation of Hsp90 increases posttranslational modifications of eNOS in hypoxia-exposed cells, even though the eNOS protein was observed to be the same (Fig. 5). Furthermore, these results can be compared with an ischemia-reperfusion model in the heart. Through adaptation by the cells to a low O₂ environment, an increase in cardiac tolerance to all critical consequences of O₂ deprivation is established (44, 46). Hsp90 is considered a target to augment NO formation, significantly lessening myocardial reperfusion injury. Overexpression of Hsp90 can protect the myocardium from hazardous effects of ischemia-reperfusion through the endothelial NO pathway (29). In the event of very high concentrations of NO, it has been proven to be detrimental. For example, extreme amounts of NO have been reported to mediate pathophysiological events in hypoxia-induced brain injuries (45). Furthermore, hypoxia activates the expression of several genes. Prabhakar et al. (39) have shown that acute hypoxia at 12 h activates the neuronal NOS gene and increases posttranscriptional neuronal NOS protein.

This work further demonstrates that there was no observable accumulation of HIF-1α (Fig. 3), therefore showing that HIF-1α does not play any role in the observed attenuation of respiration in BAECs. Recently, HIF-1 has been reported to downregulate mitochondrial O₂ consumption during hypoxia through the activation of pyruvate dehydrogenase kinase (38). HIF-1 is vital in mediating cellular responses to hypoxia (3, 28). It manages oxygen consumption, angiogenesis, glycolysis, cell proliferation, and cell survival (17, 41). We observed no significant difference in the expression of HIF-1α. HIF-1α- subunits are stabilized when prolyl hydroxylation-dependent binding of the ubiquitin ligase pVHL is inhibited. HIF-1α-subunits can interact with Hsp90, where the PAS B domain is necessary and HIF-1α stabilization occurs. Any disturbance in the function of Hsp90 defers HIF-1α accumulation (28). In its active state, the PI3K/Akt pathway is necessary for the expression of Hsp90 to protect HIF-1α from degradation in renal cell carcinoma (48). HIF-1α was also found to be stabilized by reactive oxygen species (42). It appears that the higher NO generated during hypoxia in endothelial cells enhances HIF-1 degradation (19, 34).

In summary, we have used EPR oximetry as a tool to measure changes in cellular respiration when BAECs were exposed to various O₂ concentrations. EPR oximetry reports the extracellular Po₂. Thus the mitochondrial Po₂ could be potentially different from the measured extracellular Po₂ due to oxygen diffusion barrier (18). Indeed, the gradient has been found to be ∼50 μM under some conditions. At a moderate level of hypoxia (5% O₂), no changes in cellular respiration were observed. However, at an extended hypoxic state of 1% O₂, the overall maximum O₂ consumption rate is decreased, whereas the direct NO competition with O₂ at CcO (p50 did not significantly differ) is not present. The stability of HIF-1α is hindered by hypoxia-induced NO. Additionally, the interaction of Hsp90 and eNOS was shown to peak around 8–16 h, followed by a moderate decline in association at 24 h. Together, these results show that the oxygen consumption of BAECs is attenuated when placed in a hypoxic environment (around 1% or less) for a considerable amount of time. The mechanism of such a reduced respiration seems to be involved with the Hsp90-eNOS pathway (Fig. 8). Overall, our results represent a new mechanism of cellular adaptation and modifications of cellular respiration that occur during hypoxia. This finding may provide insight into the different magnitude of endothelial dysfunction in various ischemic tissues. The variability could be due to the variation in the magnitude of Hsp90 association with eNOS, depending on individual cases.

Fig. 8.

Schematic representation of Hsp90-eNOS during normal and low oxygen. During normoxia (21% O₂), Hsp90, PI3K, and eNOS are dissociated. This allows normal oxygen consumption to occur (left). At an extreme hypoxic level (∼1% O₂), there is an augmented association (right). This leads to an increase in NO production and attenuation in the overall oxygen consumption rate.

GRANTS

We acknowledge support from National Institutes of Health Grants R21-EB-004658 and R01-HL-078796 (to G. Ilangovan) and F31-GM-078772-01 (to T. Presley).