Regulation of autophagy in human and murine cartilage: Hypoxia-inducible factor 2 suppresses chondrocyte autophagy




We have previously demonstrated that the transcription factor hypoxia-inducible factor 1 (HIF-1) promotes the onset of autophagy in chondrocytes. The overall goal of this study was to test the hypothesis that another HIF family transcription factor, HIF-2, modulates the induction of autophagy by chondrocytes.


Expression of HIF-1, HIF-2, and light chain 3 (LC3) in human and murine articular cartilage was visualized by immunohistochemistry. Suppression of HIF-2 was achieved using small interfering RNA technology. Assessments of autophagic flux and lysosomal activity, as well as ultrastructural analysis, were performed in chondrocytes in cell culture.


HIF-2 was expressed abundantly by cells in human and murine articular cartilage and in the cartilage of mineralizing vertebrae from neonatal mice. Protein levels were reduced in articular cartilage from older mice, in end-plate cartilage from mice, and in chondrocytes from human osteoarthritic (OA) cartilage. HIF-2 was robustly expressed in the prehypertrophic cells of mouse growth cartilage. When HIF-2α was silenced, the generation of reactive oxygen species was found to be elevated, with a concomitant decrease in catalase and superoxide dismutase activity. Suppression of HIF-2 was associated with decreased Akt-1 and mammalian target of rapamycin activities, reduced Bcl-xL expression, and a robust autophagic response, even under nutrient-replete conditions. In these silenced chondrocytes, HIF-1 expression was elevated. Decreased HIF-2 expression was associated with autophagy in OA tissues and aging cartilage samples. The autophagic response of chondrocytes in HIF-2α–knockout mouse growth plate showed an elevated autophagic response throughout the plate.


Based on these observations, we conclude that HIF-2 is a potent regulator of autophagy in maturing chondrocytes. Our data suggest that this protein acts as a brake on the autophagy-accelerator function of HIF-1.

Among the skeletal elements of the body, cartilage comprises a group of highly diverse tissues that perform a wide array of functions, including support, growth, and locomotion. Articular cartilage is present at the ends of long bones, where it serves as a load-bearing surface. Although cell distribution within the tissue is sparse, the highly glycosylated proteoglycan molecules of the extracellular matrix bind water, and by rapidly changing the cartilage hydration state, accommodate biomechanical forces. The network of oriented collagen fibers endows the tissue with much of its mechanical strength. During aging and progression of osteoarthritis (OA), there is a net loss of matrix macromolecules (in particular, the proteoglycan aggrecan), disruption of tissue architecture, and a concomitant limitation of joint function (1).

In the axial and appendicular skeleton, much of the bone growth is regulated by the activity of chondrocytes residing in a specialized transient cartilage, the epiphyseal growth plate. The maturing cells in this cartilage undergo a series of phenotype changes that include secretion of a unique set of proteins into the avascular extracellular matrix, up-regulation of alkaline phosphatase, and the release and subsequent mineralization of matrix vesicles (2). We have recently shown that prior to deletion from the growth plate, the mature hypertrophic chondrocyte becomes glycolytic and undergoes functional and immunohistochemical changes that are characteristic of autophagy (3). In addition, we recently demonstrated that hypoxia-inducible factor 1 (HIF-1), a transcription factor that responds to the tissue oxemic state, promotes chondrocyte autophagy (4). We have suggested that autophagy is an intermediate step in the chondrocyte life cycle that permits the cells to assume a mature phenotype prior to elimination by apoptosis (4, 5).

A second HIF isoform, HIF-2, has also been shown to be present in cartilage (6). In contrast to HIF-1, which serves to metabolically adapt chondrocytes to their microenvironment and sensitize them to apoptogens, HIF-2 is cytoprotective. Thus, up-regulation of HIF-2 lowers intracellular levels of reactive oxygen species (ROS) by promoting the activities of the dismutating proteins catalase and superoxide dismutase (SOD) (7). In addition, the small size of HIF-2–knockout animals suggests that there may be an increased rate of chondrocyte apoptosis that serves to impede normal long bone growth.

The overall goal of this study was to test the hypothesis that HIF-2 modulates the induction of autophagy by chondrocytes. We found that HIF-2α was expressed by cells in a number of cartilage types. Furthermore, this expression was decreased with maturation, aging, and onset of disease. Our finding that in cartilage, a decrease in HIF-2 expression was accompanied by an increase in HIF-1 expression, supports the idea that the activities of these 2 isoforms can be viewed as opposing arms of a rheostat that functions to modulate autophagic activity, thereby allowing chondrocytes to complete their life cycle.



Antibodies against HIF-2α (catalog no. NB 100-122) and anti–beclin 1 were obtained from Novus Biologicals (Littleton, CO), light chain 3 (LC3) was from Abgent (San Diego, CA), tubulin, lamin A/C, Bcl-xL, and Bcl-2 were from Santa Cruz Biotechnology (Santa Cruz, CA), HIF-1α was from R&D Systems (Minneapolis, MN), and Akt-1, phospho–Akt-1, mammalian target of rapamycin (mTOR), phospho-mTOR, S6 kinase (S6K), and phospho-S6K were from Cell Signaling Technology (Danvers, MA). Cell culture reagents were obtained from Fisher Scientific (Malvern, PA). Transfection reagents were purchased from Invitrogen (Carlsbad, CA).

Mammalian protein extraction reagent (M-PER), protein A/G–Sepharose beads, and horseradish peroxidase–labeled secondary antibody were purchased from Pierce (Rockford, IL). Alexa Fluor 594–labeled and fluorescein-labeled secondary antibodies (SouthernBiotech, Birmingham, AL) were used in the immunohistochemical studies. Reagents for Western blotting were from Bio-Rad (Hercules, CA). All other reagents were from Sigma-Aldrich (St. Louis, MO).

Cell culture.

N1511 mouse chondrocytes and derived cell lines were maintained in culture in α-minimum essential medium (α-MEM) at 37°C in an atmosphere of 5% CO2, 95% air. Details of the cell culture system have been described previously (7). For hypoxia studies, chondrocytes were maintained at 2% O2 in an Invivo2 Hypoxia Workstation (Ruskinn, Cincinnati, OH). In some studies, cells were treated for 4 hours with 200 nM bafilomycin A1 (LC laboratories, Woburn, MA).

In vivo immunolocalization of HIF-2α, LC3, and HIF-1α in cartilage.

Expression of HIF-2α, LC3, or HIF-1α was assessed in normal and OA human proximal femoral cartilage obtained from the National Disease Research Interchange (Philadelphia, PA) tissue bank, in the distal tibial and proximal femoral cartilage of newborn and 18- and 30-month-old mice, in mouse neonatal mineralizing vertebral cartilage and intervertebral end-plate cartilage, and in newborn mouse femoral epiphyseal growth plate. Mice were euthanized in accordance with ethical guidelines approved by the Institutional Animal Care and Use Committee of Thomas Jefferson University.

Articular, end-plate, and epiphyseal cartilage samples were demineralized with EDTA, paraffin-embedded, serially sectioned (5 μm), permeabilized with proteinase K (10 mg/ml), and fixed in 4.0% (weight/volume) paraformaldehyde. Next, sections were treated with antibodies to HIF-2α, LC3, or HIF-1α at a dilution of 1:50 volume/volume. Following treatment with the primary antibody, sections were treated with Alexa Fluor 594–labeled secondary antibodies, counterstained with 4′,6-diamidino-2-phenylindole, and examined by fluorescence microscopy.

Immunohistochemical localization of LC3 in chondrocytes in culture.

LC3 expression in N1511 chondrocytes was assessed by immunohistochemistry. After washing in phosphate buffered saline, cells were fixed for 10 minutes with 4% (v/v) paraformaldehyde (pH 8.0). Following permeabilization with 0.5% (v/v) Triton X-100, antigenic sites were blocked in 10% (v/v) calf serum. The cells were incubated overnight at 4°C with LC3 antibody. Subsequently, the cells were incubated with a fluorescein-labeled secondary antibody. Proteins were examined by confocal microscopy (Fluoview; Olympus, Tokyo, Japan).

Electron microscopic evaluation of autophagy.

Isolated cells were fixed for 2 hours in 2% (v/v) paraformaldehyde and 0.1% (v/v) glutaraldehyde in 0.1M sodium cacodylate, postfixed for 2 hours with 1% (w/v) OsO4, and stained for 1 hour with 2% (w/v) aqueous uranyl acetate. The samples were then washed, dehydrated with graded alcohol, and embedded in Epon–Araldite resin. Ultrathin sections were cut with a Leica ultramicrotome (Leica, Wetzlar, Germany), counterstained with 0.3% (w/v) lead citrate, and examined at 60 kV using a JEOL 1230 electron microscope (JEOL, Palo Alto, CA).

Detection of lysosomal activity.

Lysosomal activity was assessed using the LysoTracker assay. Cells were serum-starved for 4 hours and then incubated for 1 hour at 30°C with LysoTracker Red. Lysosomal activity was assessed by confocal microscopy.

Western blot analysis.

Cells were lysed with M-PER, and 100 μg of protein was then loaded onto 4–20% Tris gradient gels. Following transfer to polyvinylidene difluoride membranes and blocking, blots were treated overnight with the appropriate primary antibody. Membranes were then washed and treated with horseradish peroxidase–labeled secondary antibody. Blots were visualized with Lumigen TMA-6 (Amersham Biosciences, Piscataway, NJ).

Suppression of HIF-2α using small interfering RNA (siRNA) technology.

An siRNA construction kit (pSilencer, Ambion, Austin, TX) was used for silencing of HIF-2α. The following phosphorylated oligonucleotides were used: GATCCGGAGACGGAGGTCTTCTATTTCAAGAGAATAGAAGACCTCCGTCTCCTTTTTTGGAAA (forward) and AGCTTTTCCAAAAAAGGACACGGAGGTCTTCTATTCTCTTGAAATAGAAGACCTCCGTCTCCG (reverse).

Permanent cell lines were generated by clonal selection using 800 μg/ml of hygromycin B. N1511, a cell line with a backbone vector containing scrambled sequences, served as the control (pSHH).

Measurement of chondrocyte ROS levels.

Following culture, cells were treated with dihydroethidium (50 μM) for 30 minutes. The cells were lysed with 0.1% (v/v) Triton X-100 in deionized H2O, and fluorescence was determined at 485 nm and 595 nm.

Suppression of ROS.

Cellular ROS generation was blocked for 1 hour with 2,3-tert-butyl-4-hydroxy-anisole (BHA; Supelco, Bellefonte, PA) at a concentration of 100 μM.

Measurement of chondrocyte catalase and SOD activities.

Catalase activity was measured using a Fluoro Catalase kit (Cell Technology, Mountain View, CA). Samples were treated with 20 μM H2O2 for 1 hour and then treated with the detection reagent. Fluorescence was determined at 540 and 590 nm. SOD activity was measured according to the manufacturer's instructions using a kit from Cell Technology.

Immunoprecipitation of the Bcl-2–beclin 1 complex.

To detect Bcl-2–beclin 1 complex, cells were serum-starved for 4 hours, and whole cell extracts were prepared as described previously (4, 5). Cell lysates (150 μg) were precleared for 1 hour at 4°C using 20 μl of protein A/G–Sepharose beads and 10 μl of rabbit antiserum. The supernatant was then incubated overnight at 4°C with beclin 1 antibody and protein A/G–Sepharose beads. The beads were then washed in lysis buffer. Proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and subjected to Western blot analysis for Bcl-2 and beclin 1.

Immunohistochemical analysis of epiphyseal growth plate from HIF-2α–knockout mice.

Immunohistochemical analyses of HIF-1α, HIF-2α, and LC3 expression were performed on sections of the epiphyseal growth plate obtained from HIF-2α–knockout mice, as described previously (7).

Statistical analysis.

All measurements were performed in triplicate, and the results are presented as the mean ± SD. Differences between groups were analyzed by Student's t-test. P values less than 0.05 were considered significant.


HIF-2 expression in cartilage.

Cartilage tissue sections were immunostained for HIF-2α. As shown in Figure 1A, chondrocytes in articular cartilage from the newborn mouse distal tibia and proximal femur expressed HIF-2α. There was robust expression of HIF-2α, and the highest level of expression was seen in chondrocytes of the superficial zone of the cartilage.

Figure 1.

Expression of hypoxia-inducible factor 2α (HIF-2α) in cartilage. Sections cut through demineralized cartilage were stained with an antibody to HIF-2α, treated with Alexa Fluor 594–labeled secondary antibodies, and nuclei were counterstained blue with 4′,6-diamidino-2-phenylindole. Rabbit IgG was used as a negative control. A, Articular cartilage from the distal tibia and proximal femur of a newborn mouse. Note the intense HIF-2α staining in cells at the articular surface of the joint. B, Articular chondrocytes from the proximal femur of a mature (18-month-old) mouse (1) and from an older (30-month-old) mouse (2). Note the dramatic decrease in HIF-2α expression with age. C, Cartilage from the mineralizing vertebra of a neonate (1) and from the end plate of an older mouse (2). Again, there is a dramatic decrease in HIF-2α expression with age. D, Chondrocytes from articular cartilage obtained from a healthy subject (1) and from a patient with osteoarthritis (OA) (2). HIF-2α expression is significantly reduced in cells of the OA cartilage. E, Prehypertrophic cells of the mouse growth plate. A robust signal is evident in cells of the prehypertrophic zone as compared with cells of the proliferating zone. (Magnification × 200.)

We next examined HIF-2 expression in articular chondrocytes from the proximal femurs of mice of 2 different age groups. While chondrocytes from mature (18-month-old) mice showed abundant HIF-2α expression (Figure 1B, part 1), there was a dramatic reduction in HIF-2α expression in chondrocytes from older (30-month-old) animals (Figure 1B, part 2). When we compared cartilage in mineralizing vertebrae obtained from neonatal mice with end-plate cartilage obtained from older mice (Figure 1C, part 2), a dramatic decrease in HIF-2α expression was seen. Similarly, there appeared to be a reduction in HIF-2α expression in chondrocytes from cartilage obtained from human OA (Figure 1D, part 2) as compared with normal (Figure 1D, part 1) proximal femurs.

To evaluate HIF-2α expression in growth cartilage, sections of mouse neonatal femoral growth plate were immunostained for HIF-2α. While HIF-2α staining was observed in proliferating chondrocytes, the most densely stained cells were localized in the early hypertrophic region (Figure 1E). We and other investigators have demonstrated that cells in this region have elevated alkaline phosphatase activity and express type X collagen (7, 8). HIF-2α expression decreases as chondrocytes become terminally differentiated. Although in most cases, there was very limited staining of the chondrocyte nuclei, this was not unexpected, since there is good evidence to indicate that HIF-2 is preferentially retained or trapped in the cytosol, with a minimum amount of protein present in the nucleus (9). The specificity of HIF-2α staining was determined in sections of growth plate obtained from HIF-2α–knockout mice (results not shown).

HIF-2 regulation of ROS dismutation in chondrocytes.

To examine in greater detail the role of HIF-2 in chondrocyte biology, we evaluated the expression of HIF-2α in N1511 chondrocytes. In culture, these cells expressed aggrecan and type II collagen and generated an abundant extracellular matrix. We silenced HIF-2α gene expression using siRNA technology (Figure 2A) and measured the generation of ROS. Compared with control (pSHH) chondrocytes, the total ROS levels were significantly higher in the silenced chondrocytes (Figure 2B). Under hypoxic conditions, however, while the ROS levels decreased, there was no difference between control and HIF-2α–silenced chondrocytes (Figure 2C).

Figure 2.

Impaired reactive oxygen species (ROS) dismutation and suppressed superoxide dismutase (SOD) and catalase activities upon silencing of hypoxia-inducible factor 2α (HIF-2α). A, Suppression of HIF-2α by use of small interfering RNA technology. Protein was extracted from the chondrocyte cell line, and HIF-2α expression in silenced (SI) cells and in cells transfected with the control construct (pSHH) was measured by Western blotting. There is a marked decrease in HIF-2α protein in the silenced cells. B, ROS levels in maturing chondrocytes. Cells were treated with dihydroethidium, and the change in fluorescence was measured. The increase in dihydroethidium fluorescence in HIF-2α–silenced cells is due to decreased ROS dismutation. C, ROS generation in chondrocytes grown under normoxic (20% O2) and hypoxic (2% O2) conditions. After culture, the cells were treated with dihydroethidium, and the change in fluorescence was measured. The increase in dihydroethidium fluorescence in HIF-2α–silenced cells is due to decreased ROS dismutation. There is no difference in ROS dismutation under hypoxic conditions. D, Catalase activity in chondrocytes. There is a significant decrease in catalase activity in HIF-2α–silenced chondrocytes as compared with control. E, SOD activity in chondrocytes. SOD activity is significantly reduced in HIF-2α–silenced chondrocytes as compared with control. F, Expression of SOD protein in HIF-2α–silenced chondrocytes, as measured by Western blot analysis. Total SOD protein levels are reduced in HIF-2α–silenced chondrocytes compared with control. Lamin A/C was used as a loading control in A and F. Values in B–E are the mean ± SD. # = P < 0.05 versus control.

Since levels of ROS are regulated by the activities of catalase and SOD, we next examined the activity of these dismutating enzymes in chondrocytes. As shown in Figures 2D and E, the activity of both dismutating agents was significantly suppressed in HIF-2α–silenced cells. We also examined the levels of SOD by Western blot analysis. Figure 2F shows that the protein expression level was lowered in silenced chondrocytes.

LC3 expression in cartilage.

Since HIF-2 has been shown to regulate ROS levels, and ROS have been shown to induce autophagy, we next examined cartilaginous tissues for the expression of the autophagy marker LC3. Figure 3A shows that LC3 was expressed in chondrocytes from both young and older mice. However, while younger mice exhibited diffuse staining (Figure 3A, part 1), distinct punctate staining was commonly seen in the older mice (Figure 3A, part 2), indicative of autophagy. Figure 3B shows LC3 staining of neonatal and end-plate cartilage from mouse vertebrae. Compared with the end-plate cartilage from older animals, end-plate cartilage cells of the intervertebral disc from young mice displayed diffuse LC3 staining (Figure 3B, part 1). Chondrocytes in the end plate showed distinct puncta characteristic of autophagic cells (Figure 3B, part 2).

Figure 3.

Light chain 3 (LC3) expression in cartilage. Cartilage sections were stained with an antibody to the microtubule-associated protein LC3 and treated with Alexa Fluor 594–labeled secondary antibodies. Rabbit IgG was used as a negative control. A, Articular chondrocytes in the proximal femur of a mature (18-month-old) mouse (1) and from an older (30-month-old) mouse (2). There is diffuse staining for LC3 in chondrocytes from the mature mouse. In contrast, LC3 is present in a distinct punctate distribution in chondrocytes from the older mouse. B, Mineralizing vertebral cartilage from a neonatal mouse (1) and intervertebral end-plate cartilage from an older mouse (2). Cells in the intervertebral disc from the neonatal mouse display diffuse LC3 staining as compared with those in the end-plate cartilage from the older animal. Chondrocytes in the end-plate cartilage from the older animal show distinct puncta, which are characteristic of autophagic cells. C, Articular cartilage obtained from a healthy human subject (1) and from a patient with osteoarthritis (2). Chondrocytes in the OA tissue display numerous autophagic LC3 puncta. In contrast, those in healthy tissue do not show an elevation in punctate LC3. D, Proliferating cells in the growth plate of a wild-type mouse (1) and a hypoxia-inducible factor 2α (HIF-2α)–knockout mouse (2). In the HIF-2α–knockout growth plate, there is LC3 punctate staining (arrows) even in the proliferative zone. In the wild-type growth plate, there is diffuse LC3 staining in the proliferative zone. (Magnification × 200.)

LC3 expression was also evaluated in chondrocytes from human healthy and OA articular cartilage (Figure 3C). Chondrocytes in OA tissue displayed numerous autophagic LC3 puncta (Figure 3C, part 2). In contrast, cells from healthy tissue did not show an elevation in punctate LC3 distribution (Figure 3C, part 1).

Finally, we examined LC3 expression in growth plate cartilage from HIF-2α–knockout mice. Distinct punctate LC3 staining was observed throughout the growth plate, including the proliferating zone cells (Figure 3D, part 2), indicating that suppression of HIF-2α results in the induction of autophagy. In growth plate from wild-type control mice, LC3 staining was seen mainly in the prehypertrophic zone (Figure 3D, part 1)

Induction of autophagy in HIF-2–suppressed chondrocytes.

To determine whether the changes in HIF-2 levels promoted autophagy, cells were probed with an anti-LC3 antibody. As shown in Figure 4A, when HIF-2 was silenced, there was induction of autophagy; thus, there was redistribution of the microtubule-associated protein LC3 to form punctate structures in the HIF-2–silenced cells. Significantly, induction was evident under serum-rich conditions. Under hypoxic conditions, while some LC3 reorganization was seen in control cells, in HIF-2–silenced cells, we again found significant reorganization of LC3 into punctate structures (Figure 4A).

Figure 4.

Autophagic response of hypoxia-inducible factor 2α (HIF-2α)–suppressed cells. A, Light chain 3 (LC3) expression in HIF-2α–suppressed cells cultured under normoxic (20% O2) or hypoxic (2% O2) conditions, as determined by immunofluorescence analysis. Under normoxic conditions, LC3 is expressed in speckled structures (autophagosomes) in HIF-1–silenced (SI) cells, even under serum-replete conditions. In contrast, LC3 fluorescence is diffuse in control (pSHH) cells. Under hypoxic conditions, both HIF-2α–silenced cells and control cells show a speckled distribution of LC3 (arrows). B, Association of Bcl-2 and beclin 1 in HIF-2α–suppressed cells, as determined by immunoprecipitation and Western blot analyses. Control and HIF-2α–suppressed cells were grown under serum-rich conditions, lysed, immunoprecipitated with anti–beclin 1, and subjected to Western blotting. Note the association of Bcl-2 with beclin 1 in control cells. This association is blocked in HIF-2α–silenced cells, even under serum-rich conditions. C, Analysis of total Bcl-2 expression in whole cell lysates, as determined by Western blotting. There is no significant difference in total Bcl-2 expression between HIF-2α–silenced and control cells. D, Lysosomal activity of HIF-2α–suppressed chondrocytes. Cells were incubated with LysoTracker Red (70 nM) for 1 hour at 37°C in serum-replete media and examined by confocal microscopy. Note the near-complete suppression of lysosomal activity in control cells. In contrast, there is extensive lysosomal fluorescence in HIF-2α–silenced cells. E, Increased expression of the LC3-II isoform upon treatment with bafilomycin A1 (BAF), as determined by Western blotting of whole cell lysates. Note the presence of LC3-II in HIF-2α–silenced cells grown in serum and the accumulation of this isoform in bafilomycin A1–treated cells. F and G, Transmission electron microscopy of control and HIF-2α–silenced chondrocytes. In F, images on the right are high-magnification views of the boxed areas of the images on the left. Note the presence of autophagosomes and isolation membranes in HIF-2α–silenced cells cultured under serum-rich conditions (arrows). In G, note the presence of a mitochondrion surrounded by a double-layered membrane (cluster of 3 arrows) and the presence of isolation membranes (single arrows) in HIF-2α–silenced cells viewed at still-higher magnification. (Magnification × 200 in A, D, and F, left; × 8,000 in F, right; and × 10,000 in G.)

The association of the autophagic protein beclin 1 with Bcl-2 was also evaluated in chondrocytes under serum-replete conditions. We noted that in control cells, Bcl-2 coimmunoprecipitated with beclin 1. In contrast, immunoprecipitates of silenced cells indicate a reduced affinity of Bcl-2 for beclin 1 (Figure 4B). However, there was no difference in total Bcl-2 levels between the 2 cell lines (Figure 4C).

We also used LysoTracker Red to probe for autophagic activity. The increase in LysoTracker fluorescence (Figure 4D), which resulted from an elevation in lysosomal activity, confirmed that the silenced cells in serum-replete medium were autophagic. Similarly, bafilomycin A1, an agent that inhibits the fusion of autophagosomes with lysosomes, caused an accumulation of the LC3-II isoform (Figure 4E).

We confirmed the presence of autophagosomes by transmission electron microscopy. Figure 4F shows that autophagosomes were formed in HIF-2–silenced cells under serum-rich conditions. Double-membrane–lined vesicles as well as autophagic isolation membranes were visible in these silenced cells. The presence of mitochondria in the vesicles (Figure 4G) suggested that HIF-2 silencing results in the onset of mitophagy in these cells.

Suppression of autophagy by ROS inhibition.

Since silencing of HIF-2 resulted in the loss of dismutating activity, it was necessary to determine if the autophagic activity was due to elevated generation of ROS. To test this, we inhibited ROS generation in these cells using BHA. Figure 5A shows that while HIF-2–silenced cells displayed a punctate pattern of LC3 distribution, this pattern was absent from cells treated with BHA. Thus, ROS inhibition results in the diffuse redistribution of LC3. Similarly, while distinct LC3-I and LC3-II isoforms were seen in HIF-2–silenced cells, ROS suppression resulted in the disappearance of the LC3-II isoform (Figure 5B).

Figure 5.

Effect of hypoxia-inducible factor 2α (HIF-2α) silencing on the generation of reactive oxygen species (ROS), the induction of autophagy, the activity of mammalian target of rapamycin (mTOR), and the expression of Akt-1 and Bcl-xL. Cellular ROS generation was blocked with 2,3-tert-butyl-4-hydroxy-anisole (BHA) at a concentration of 100 μM for 1 hour. Cells were silenced (SI) for the expression of HIF-2α or were transfected with the control construct (pSHH). A and B, Expression of light chain 3 (LC3), as determined by immunofluorescence (A) or Western blot (B) analysis. The punctate pattern of LC3 expression in HIF-2α–silenced cells disappeared upon treatment with BHA (A) (magnification × 200). Note the presence of the LC3-II isoform in HIF-2α–silenced cells, which also disappeared upon treatment with BHA (B). C, Expression of mTOR, S6 kinase (S6K), Akt-1, and Bcl-xL in total protein isolated from cells cultured under serum-replete conditions, as determined by Western blotting. Note the reduced activity of mTOR and its target gene S6K, as well as the reduced levels of phospho-mTOR and phospho-S6K. Note also the reduced expression of total Akt-1, phospho–Akt-1, and Bcl-xL. Tubulin was included as a loading control.

Expression of mTOR, Akt-1, and Bcl-xL in HIF-2–silenced cells.

Since HIF-2 silencing resulted in the induction of autophagy, we next examined the expression and activity of the autophagic repressor mTOR. As shown in Figure 5C, while there was a minimal decrease in mTOR expression, there was a significant reduction in activated mTOR. Thus, while control cells showed robust expression of phospho-mTOR, HIF-2–silenced cells showed minimal expression of the activated form of mTOR. In addition, there was a reduction of the mTOR target phospho-S6K. We also examined the expression of another mTOR-activated protein, the survival pathway kinase Akt. Figure 5C shows that not only was Akt activation (phospho–Akt-1) suppressed in HIF-2–silenced cells, but overall expression was also reduced. We next examined the expression of the antiapoptotic protein Bcl-xL. While total Bcl-2 expression was not affected in HIF-2–silenced cells (Figure 4C), the expression of Bcl-xL was significantly reduced (Figure 5C).

HIF-1 expression in HIF-2–silenced cells and cartilage.

To explore the possible interaction between HIF-1 and HIF-2, we examined HIF-1 expression in the HIF-2–silenced cells. As shown in Figure 6A, there was robust expression of HIF-1α in the HIF-2–silenced cells.

Figure 6.

Increased expression of hypoxia-inducible factor 1α (HIF-1α) in HIF-2α–suppressed cells and in cells from aging and osteoarthritic (OA) cartilage. A, HIF-1α expression in HIF-2α–silenced (SI) and control (pSHH) cells, as determined by Western blotting. Note the robust expression of HIF-1α in HIF-2α–silenced cells. Tubulin was included as a loading control. B, HIF-1α expression in cells of the articular cartilage from the proximal and distal femur obtained from a mature (18-month-old) mouse (1) and an older (30-month-old) mouse (2). Note the low levels of HIF-1α staining in superficial cells of the articular cartilage in the mature mouse. In the older mouse, there is increased staining of the superficial chondrocytes. C, Mineralizing vertebral cartilage from a neonatal mouse (1) and intervertebral end-plate cartilage from an older mouse (2). Cells in the neonatal mouse cartilage show low levels of HIF-1α expression, whereas cells in the end-plate cartilage from the older mouse show significantly higher levels of HIF-1α. D, Chondrocytes in articular cartilage obtained from a healthy human subject (1) and from a patient with OA (2). OA chondrocytes display high levels of HIF-1α expression as compared with those in healthy tissue. E, HIF-1α expression in chondrocytes of the proliferative zone of a growth plate from a wild-type mouse (1) and an HIF-2α–knockout mouse (2). Note the low level of expression in proliferative-zone cells in the growth plate of the wild-type mouse. In contrast, the proliferative-zone cells in the growth plate of the HIF-2α–knockout mouse show an increase in HIF-1α expression. (Magnification in B–E × 200.)

We next examined HIF-1 expression in various cartilaginous tissues. While moderate levels of HIF-1 were seen in the superficial chondrocytes of the articular surface of the tibia and femur (Figure 6B, part 1), there was a significant increase in HIF-1 expression in cartilage from older mice (Figure 6B, part 2). Similarly, while cells of the intervertebral cartilage from young mice displayed low levels of HIF-1 (Figure 6C, part 1), in the end plate of mature mice, there was elevated HIF-1 expression (Figure 6C, part 2). Furthermore, there were high levels of HIF-1 expression in OA chondrocytes (Figure 6D, part 2) as compared with normal articular chondrocytes (Figure 6D, part 1).

We next examined the expression of HIF-1 in the growth plate. Previous studies have demonstrated high levels of HIF-1 expression in hypertrophic chondrocytes and moderate-to-low expression in proliferating cells. However, the pattern of HIF-1 expression in the growth plate of an HIF-2–knockout mouse was found to be different. Thus, high levels of HIF-1 expression were seen in cells of the proliferative zone (Figure 6E, part 2), unlike in the wild-type mouse, in which very low levels of HIF-1 expression were seen in cells of the proliferative zone (Figure 6E, part 1).


A number of recent studies have shown that local oxygen tension provides a microenvironmental signal that influences skeletal cell function (10, 11). While it is clear that the oxemic response is transduced by HIF-1, there is some evidence to indicate that HIF-2 expression may be a requirement for cartilage function. For example, a recent study of HIF-2α–knockout mice indicates that deletion of this gene (EPAS–/–) resulted in small animals, suggesting an abnormality of chondrocyte function (7). Herein, we showed that the HIF-2 homolog was expressed by chondrocytes of a number of different cartilage types. If HIF-2α has a protective function and serves to promote chondrocyte survival and function, then loss or deletion of this homolog is likely to be associated with development of disease or possibly with aging. Lending support to this notion, we noted that while HIF-2 was expressed abundantly by cells in articular cartilage and in the cartilage of mineralizing vertebrae, expression was reduced in cartilage from older animals and was low in mature end-plate cartilage. Similarly, there was a significant drop in expression in OA chondrocytes compared with healthy cells. Based on these findings, it is plausible to consider that this HIF homolog, together with HIF-1, may play a role in maintaining chondrocyte functional activity and tissue health.

We also examined the expression of HIF-2α in the epiphyseal growth plate and in the N1511 chondrocytic cell line. This HIF homolog was robustly expressed in vivo; in hypoxic culture, there was expression of this transcription factor. When we evaluated ROS generation in HIF-2α–silenced cells, we noted a small, but significant (20–30%), increase in total ROS values as compared with the pSHH controls. Thus, we confirmed earlier findings that indicated that HIF-2 regulates dismutation of ROS. Next, we evaluated the activity of catalase and SOD in HIF-2–silenced cells. These 2 genes encode proteins that serve to decrease levels of intracellular ROS. Predictably, when HIF-2α was suppressed, there was a marked decrease in the activities of both enzymes, and the SOD protein levels were also low.

Since these 2 dismutating enzymes are HIF-2 target genes, it is likely that their expression serves to minimize ROS generation in the chondrocytes. Minimization is of considerable importance, since ROS accumulation can influence structural as well as signaling activities of a number of cell types. The recent observation by Scherz-Shouval et al (12) that H2O2 modifies a conserved cysteine residue in the autophagy protein ATG4 couples HIF-2α activity with the induction of autophagy. We explored this possible relationship by evaluating autophagy and HIF-2 expression in a number of cartilage types. We noted that low levels of HIF-2 or suppression of HIF-2 correlated with a robust autophagic response in OA tissue as well as in aging cartilage; in growth-plate cartilage from HIF-2α–knockout mice, there was elevated autophagy. From this perspective, HIF-1 and HIF-2α have separate and opposing effects on the induction of autophagy.

To further examine the relationship between HIF-2 and autophagy, we probed for LC3 protein expression in the suppressed chondrocytes. There was redistribution of the microtubule-associated protein to form punctate structures that are characteristic of the autophagic state. Since LC3 redistribution was evident under serum-rich conditions, we inferred that autophagy is directly controlled by HIF-2 and is not an artifact induced by the culture system. We also used LysoTracker Red to probe cells for autophagic activity. The increase in LysoTracker fluorescence as a result of an elevation in lysosomal activity confirmed that the silenced cells in serum-replete medium displayed the functional characteristics of an autophagic chondrocyte. In addition, ultramicroscopic analysis indicated the presence of double-membrane vacuoles and isolation membranes, which are morphologic characteristics of an autophagic cell. Based on these observations, we concluded that HIF-2 is a potent regulator of autophagy in maturing chondrocytes. Thus, in contrast to HIF-1, which serves to promote autophagy (4, 13), HIF-2 regulates the extent of the autophagic response and can be viewed as acting as a brake for the accelerator function of HIF-1.

Our earlier studies (4), together with those discussed above, suggest that both HIF-1 and HIF-2 regulate chondrocyte maturation and the development of autophagy. Moreover, since the impact of each of these transcription factors is maximal at different stages of chondrocyte maturation and disease (HIF-2α is maximal in prehypertrophic cells, whereas HIF-1α is maximal in cells of the hypertrophic zone of the growth plate), the following question was raised: Does HIF-2 expression regulate HIF-1 activity? While the details of this relationship are fragmentary, one possible indication may come from the contrasting suppressive effects of HIF-1 and HIF-2 on each other (14). Thus, suppression or loss of HIF-2 results in higher expression of HIF-1, a potent inducer of autophagy (4, 13). In young healthy chondrocytes, HIF-2–dependent expression of SOD and catalase would serve to down-regulate ROS generation. Furthermore, HIF-1 is expressed at a minimal level. With aging and the onset of disease, the subsequent lowered expression of HIF-2 would lead to a loss of dismutating activity, an increase in ROS generation, and the promotion of HIF-1 expression. In addition, there was a decrease in mTOR activity and a reduction in the expression of both Akt-1 and the antiapoptotic protein Bcl-xL. These changes would cause an increase in chondrocyte autophagy, as was evident in cells from OA and aging cartilage.

Finally, from the perspective of the oxemic microenvironment, there were no significant differences in total ROS levels between hypoxic control cells and HIF-2–silenced cells. However, under normoxic conditions, there was a significant increase in ROS levels in HIF-2–silenced cells. These observations may explain the refractory nature of hypoxic chondrocytes to apoptogen challenge (8, 15). From a physiologic perspective, the hypoxic environment would allow the cells to complete their developmental and maturation processes in the presence of potential apoptogens generated by matrix degradation (16). We hypothesize that at the chondro-osseous junction, the increase in oxygen tension as a result of vascular invasion would lead to an increase in ROS generation in a manner similar to that reported for ischemia-reperfusion injury. In addition to the elevation in ROS generation, the decrease in HIF-2 levels and the stabilization of HIF-1 would further elevate the autophagic activity of chondrocytes, resulting in sensitization to apoptogen challenges and final removal by apoptosis.


Dr. Srinivas had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Bohensky, Terkhorn, Adams, Shapiro, Srinivas.

Acquisition of data. Bohensky, Terkhorn, Freeman, Adams, Garcia, Srinivas.

Analysis and interpretation of data. Bohensky, Terkhorn, Freeman, Adams, Shapiro, Srinivas.

Manuscript preparation. Bohensky, Terkhorn, Freeman, Shapiro, Srinivas.

Statistical analysis. Terkhorn, Srinivas.