The Accumulation of Intracellular ITEGE and DIPEN Neoepitopes in Bovine Articular Chondrocytes Is Mediated by CD44 Internalization of Hyaluronan

Abstract

Objective

A dramatic loss of aggrecan proteoglycan from cartilage is associated with osteoarthritis. The fate of residual G1 domains of aggrecan is unknown, but inefficient turnover of these domains may impede subsequent repair and retention of newly synthesized aggrecan. Thus, the objective of this study was to determine whether ITEGE- and DIPEN-containing G1 domains, generated in situ, are internalized by articular chondrocytes, and whether these events are dependent on hyaluronan (HA) and its receptor, CD44.

Methods

ITEGE and DIPEN neoepitopes were detected by immunofluorescence staining of bovine articular cartilage chondrocytes treated with or without interleukin-1α (IL-1α). Additionally, purified ITEGE- or DIPEN-containing G1 domains were aggregated with HA and then added to articular chondrocytes, articular chondrocytes transfected with CD44Δ67, or COS-7 cells transfected with or without full-length CD44. Internalized epitopes were distinguished by their resistance to extensive trypsinization of the cell surface.

Results

Both ITEGE and DIPEN were visualized within the extracellular cell-associated matrix of chondrocytes as well as within intracellular vesicles. Following trypsinization, the intracellular accumulation of both epitopes was clearly visible. IL-1 treatment increased extracellular as well as intracellular ITEGE epitope accumulation. Once internalized, the ITEGE neoepitope became localized within the nucleus and displayed little colocalization with HA, DIPEN, or other G1 domain epitopes. The internalization of both ITEGE and DIPEN G1 domains was dependent on the presence of HA and CD44.

Conclusion

One important mechanism for the elimination of residual G1 domains following extracellular degradation of aggrecan is CD44-mediated co-internalization with HA.

One of the early events associated with osteoarthritis is the loss of the proteoglycan, aggrecan, from the extracellular matrix of cartilage (1). This loss continues to occur even as the resident chondrocytes mount a reparative response that includes enhanced aggrecan biosynthesis. The current paradigm suggests that aggrecan turnover is initiated within the extracellular environment, due to the activity of neutral-pH–optimum endoproteinases, primarily aggrecanases, and to a lesser extent, matrix metalloproteinases (MMPs). The aggrecanases, including ADAMTS-4 (2) and ADAMTS-5 (3), are thought to be the key mediators of early aggrecan loss and cartilage damage, and ADAMTS-5 is known to be the major aggrecanase in the mouse (4,5). MMP cleavage of aggrecan correlates with late-stage cartilage damage in mouse models of arthritis (6–8), and it may also be involved in the baseline turnover of aggrecan in vitro (9) and in vivo (10). The products of in vivo proteolysis at both the MMP and the aggrecanase sites are present in humans (11–14) and in mice with experimental arthritis (7,15–17). Following an initial cleavage in the aggrecan interglobular domain, the C-terminal chondroitin sulfate–rich portion of aggrecan is lost from the cartilage by diffusion. Newly generated N-terminal neoepitopes can now be detected within synovial fluid and provide a diagnostic measure of proteoglycan loss from the tissue (14).

The fate of the original N-terminal domain of aggrecan (termed the G1 domain), presumably still bound to hyaluronan (HA) and stabilized by link protein, is less clear. However, 2 studies have shed light on this aspect of aggrecan turnover. Fosang et al (9) demonstrated, using confocal microscopy, that the G1 domain derived from aggrecanase-mediated cleavage of aggrecan (containing a new C-terminal neoepitope termed ITEGE) was localized intracellularly, inside chondrocytes present within sections of intact porcine articular cartilage. After their study, we demonstrated that a commercially purified and biotinylated aggrecan G1 domain–link protein complex (isolated following trypsin digest of aggrecan) could be co-internalized with HA via a CD44-mediated endocytosis event (18). Together, these results suggest that the final pathway for complete catabolism of aggrecan includes receptor-mediated co-internalization of HA, the G1 domain of aggrecan, and other HA-bound proteins such as link protein. It was also determined in our study that full-size, intact aggrecan monomer, bound to HA and retained at the cell surface of chondrocytes, cannot be internalized; neither the HA nor the aggrecan is internalized (18). This implies that the internalization event requires an initial cleavage of aggrecan.

The question remains whether the same CD44-mediated endocytosis pathway is also utilized by chondrocytes for the turnover of aggrecan G1 fragments generated by aggrecanase or MMPs in situ. G1 fragments generated by ADAMTS-4 or ADAMTS-5 exhibit a unique carboxy-terminal sequence, ITEGE, recognized by an anti-ITEGE neoepitope antibody (9,19). Cleavage of aggrecan by MMPs, such as MMP-13, yields a G1 with a different carboxy-terminal sequence, DIPEN, that is recognized by an anti-DIPEN neoepitope antibody (9,19). Therefore, these G1 domains can be readily distinguished.

We have shown that the HA receptor, CD44, mediates the binding and endocytosis of HA in articular chondrocytes (20–22). The internalized HA is degraded to small fragments within low-pH intracellular organelles. Internalized fluorescein-conjugated HA colocalizes with a red fluorescent probe (LysoTracker Red; Molecular Probes–Invitrogen, Carlsbad, CA) that identifies low-pH intracellular organelles (18,23). Several approaches have been used to document the role of chondrocyte CD44 in the endocytosis of HA. First, the intracellular accumulation of HA by chondrocytes is enhanced following the exposure of cells to the catabolic cytokine interleukin-1α (IL-1α), a condition that dramatically stimulates the expression of chondrocyte CD44 (22,24). As a more direct, gain-of-function approach, COS-7 cells (a CD44-negative cell type) gain the capacity to bind and internalize HA following transient transfection with full-length recombinant human CD44 (25). This capacity of COS-7 cells to bind and internalize HA can be blocked by cotransfection of COS-7 cells with full-length CD44 together with a carboxy-terminal truncation mutant of CD44 (CD44Δ67). The expression of CD44Δ67 results in cells that can neither bind nor internalize extracellular HA (18,25).

Upon transfection of normal bovine articular chondrocytes with CD44Δ67, the transfected chondrocytes also lose their capacity to bind and internalize HA (18,25) and their capacity to mediate CD44-mediated signaling events (26). These results demonstrate that CD44Δ67 functions as a dominant-negative mutant receptor for CD44, interfering with all CD44-related cell functions. Therefore, the use of CD44Δ67 as the dominant negative provides a useful loss-of-function approach for study of the role of CD44 in receptor-mediated endocytosis events. All of these approaches were used in this study to determine whether the aggrecan G1 domain fragments generated in situ, by either aggrecanases or MMPs, are co-internalized with HA via a CD44-mediated mechanism.

MATERIALS AND METHODS

Cell culture

Chondrocytes were isolated from full-thickness slices of the articular surface of the metacarpophalangeal joints of young adult (18–24-month-old) steers (27). Cartilage slices were digested in 0.2% Pronase (Calbiochem, San Diego, CA) in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Herndon, VA) plus 10% fetal bovine serum (FBS; Hyclone, Logan, UT) for 1.5 hours at 37°C, followed by an overnight digestion with 0.025% collagenase P (Calbiochem) in DMEM containing 10% FBS. Isolated chondrocytes were cultured in alginate beads (1.2% Keltone LV in 150 mM NaCl; Kelco, Chicago, IL), as previously described (27). The beads were maintained in DMEM/Ham's F-12 medium plus 5% FBS and 25 μg/ml ascorbate and 50 μg/ml gentamicin (Gibco BRL, Grand Island, NY) for 5 days to recover. The mammalian COS-7 cell line (simian virus 40–transformed African Green monkey kidney cells; American Type Culture Collection, Rockville, MD) was cultured in DMEM plus 10% FBS as monolayer cultures (25).

Detection of cell surface and intracellular aggrecan G1 domain neoepitopes

Adult bovine articular chondrocytes were cultured in alginate beads for 5 days, as described previously (18). The chondrocytes were released from the alginate beads with 55 mM sodium citrate in 150 mM NaCl, washed, pelleted, and resuspended in DMEM containing 10% FBS. In some experiments the chondrocytes were treated with 10 ng/ml IL-1α (Genzyme, Cambridge, MA) in DMEM containing 10% FBS for 48 hours prior to release from the beads. In other experiments, the chondrocytes were incubated in the continuous presence of 2 units/ml Streptomyces hyaluronidase (Sigma, St. Louis, MO) in DMEM containing either 10% FBS or a 1:100 dilution of anti-CD44 monoclonal antibody IM7.8.1 (PharMingen BD Biosciences, Bedford, MA) for 48 hours prior to release from the beads. One aliquot of released chondrocytes was immediately fixed with 1% paraformaldehyde and then quenched with 0.2M glycine in phosphate buffered saline (PBS). A second aliquot was trypsinized (0.25%; Gibco BRL) for 30 minutes at 37°C to remove all cell surface–bound protein.

Using flow cytometry, we documented that these trypsinization conditions removed all CD44-immunoreactive epitope from the chondrocyte cell surface, resulting in mean channel fluorescence equivalent to negative controls (phycoerythrin–streptavidin alone) (28) (data not shown). The cells were then fixed with 1% paraformaldehyde, quenched with 0.2M glycine in PBS, and then permeabilized using 0.1% Triton X-100 for 15 minutes at room temperature. After washing in PBS, the chondrocyte aliquots were incubated overnight at 4°C with rabbit anti-G1 (Antibody A; 1:1,000), rabbit anti-DIPEN³⁴¹ (Antibody B; 5 μg/ml), mouse monoclonal anti-³⁴²FFGVG (Antibody C), or rabbit anti-ITEGE³⁷³ (Antibody D; 0.05 μg/ml) as described (9,29) (Figure 1). Polyclonal anti–G1 domain antibody (Antibody A) was a gift from Prof. T. Hardingham (University of Manchester, Manchester, UK).

Figure 1.

Aggrecan N-terminal fragments and antibody recognition sites. Depicted is a schematic representation of the single-letter amino acid code sequence for the interglobular domain of aggrecan, the predominant cleavage sites within it adjacent to the G1 domain, and the resultant peptide fragments. The antibodies used in this study were the anti–G1 domain (A), anti–G1-DIPEN neoepitope (B), anti–FFGVG neoepitope (C), and anti–G1-ITEGE neoepitope (D). Aggrecan G1 domains generated by the action of matrix metalloproteinases (MMPs) such as stromelysin or MMP-13 would be recognized by antibodies A and B, whereas G1 domains generated by the action of aggrecanases would be recognized by antibodies A and D. Cleavage by both enzymes would liberate a peptide recognized by antibodies C and D.

Cells were then washed and incubated with either rhodamine red X–conjugated AffiniPure Fab fragment goat anti-rabbit IgG antibody (1:1,000; Jackson ImmunoResearch, West Grove, PA), fluorescein isothiocyanate (FITC)–conjugated goat anti-rabbit IgG antibody (1:50; Jackson ImmunoResearch), or Cy3-conjugated AffiniPure goat anti-mouse IgG (1:6,000; Jackson ImmunoResearch) for 1 hour at 4°C, followed by the addition of the nuclear stain, 4′,6-diamidino-2-phenylindole (DAPI; Molecular Probes). Chondrocytes were visualized using a Nikon Eclipse E600 microscope equipped with Y-Fl Epi-fluorescence (Nikon, Melville, NY), a 60 × 1.4 n.a. oil-immersion objective, and rhodamine red (red), FITC (green), and DAPI (blue) filters. Images were captured digitally in real time using a Spot-RT camera (Diagnostic Instruments, Sterling Heights, MI) and processed using MetaView imaging software (Universal Imaging, West Chester, PA). To document the pH of intracellular organelles, live chondrocytes were incubated with 50 nM LysoTracker Red for 30 minutes prior to trypsinization and fixation.

In other experiments, fixed chondrocytes incubated overnight with primary antibodies were then incubated for 1 hour at room temperature with FITC-conjugated goat anti-rabbit IgG antibody (1:50; Jackson ImmunoResearch). Cell surface immunofluorescence was then quantified using a FACScan cytometer (Becton Dickinson, Palo Alto, CA) with CellQuest software. For 10,000 cells, log fluorescence channel versus cells per channel was plotted.

In some experiments, individual chondrocyte nuclei were isolated prior to incubation with primary G1 antibodies. Briefly, following release of the cells from alginate beads, the washed chondrocytes were allowed to swell for 1 hour, with gentle shaking, in hypotonic buffer (0.01M NaPO₄ [pH 7.3]) on ice (30). The cells were sheared via 40 strokes of a tight-fitting Dounce homogenizer, followed by the addition of 2 volumes 0.5M sucrose in 0.02M Tris (pH 7.3). The lysed cells were then centrifuged at 4,000g for 20 minutes at 4°C to pellet the nuclear fraction (the supernatant was discarded). The cell nuclei were suspended in PBS, fixed with 1% paraformaldehyde, and then incubated with various primary antibodies, as described above.

Generation of G1-ITEGE and G1-DIPEN in explant culture for use as exogenous HA ligands

Slices of pig articular cartilage were cultured for 3 days in 10% fetal calf serum, then switched to serum-free medium with 10 ng/ml IL-1α for a further 4 days. At the end of the culture period the tissue was extracted with buffer containing 4M guanidinium hydrochloride (GuHCl), 10 mM EDTA, and 50 mM sodium acetate (pH 5.8), and analyzed for G1 species by Western blotting with anti-G1 antibody, anti-DIPEN³⁴¹, or anti-ITEGE³⁷³ antibodies (19). All the G1 in the tissue extracts contained the ITEGE³⁷³ neoepitope; no G1-DIPEN fragments were detected. G1-ITEGE was purified by HA–Sepharose affinity chromatography. The dialyzed extracts were applied to an HA–Sepharose column equilibrated in PBS, and the G1-NITEGE fragments were eluted with buffered 4M GuHCl as described previously for G1–G2 (19,31). Purified G1-ITEGE (1.776 mg) was digested with 100 μg/ml trypsin-activated MMP-13 for 22 hours at 37°C to generate G1-DIPEN. Recombinant human proMMP-13 (32) was a gift from Dr. V. Knäuper and Prof. G. Murphy (University of East Anglia, Norwich, UK). G1-DIPEN was then purified by elution from HA–Sepharose, dialyzed, and lyophilized. The concentration of the G1-ITEGE preparation was estimated by absorbance at 278 nm. The purity of the G1-DIPEN and the G1-ITEGE fragments was assessed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and Western blotting with the polyclonal G1 domain antibody.

Transient transfections

Adult bovine articular cartilage chondrocytes, cultured in alginate beads for 5 days, were released from alginate with 55 mM sodium citrate in 150 mM NaCl, and treated with 0.25% trypsin in DMEM for 10 minutes at 37°C. The chondrocytes were then plated into 100-mm tissue culture dishes in DMEM plus 10% FBS at high density. COS-7 cells were plated at a similar density. The following day the chondrocytes were transfected with 10 μg human pCD44HΔ67 (pTracer-SV40, green fluorescent protein [GFP] coexpression vector; Invitrogen) in serum-free, antibiotic-free DMEM, in the presence of Lipofectamine 2000 reagent (Gibco BRL) as previously described (25). It should be noted that the GFP is driven by a separate cytomegalovirus promoter on the SV40-pTracer plasmid. The COS-7 cells were transfected in a similar manner with 10 μg human pCD44Hwt (pTracer-SV40 with GFP). Following 5 hours of incubation at 37°C in a humidified CO₂ incubator, FBS was added to the transfected chondrocytes or COS-7 cells to a concentration of 10%. Medium was changed the following day, and binding and uptake experiments were performed 48 hours posttransfection.

Binding and internalization of G1-decorated HA by chondrocytes and COS-7 cells

To prepare each ml of G1-decorated HA, 10 μg G1 domain was combined with 50 μg of high molecular weight HA (Genzyme) and incubated for 3 hours with rocking at room temperature. Following aggregate formation the decorated HA was brought to 70% ethanol containing 1.3% (weight/volume) potassium acetate and precipitated overnight by incubation at –20°C. Following collection by centrifugation, the G1-decorated HA pellets were resuspended in PBS at a final concentration of 25 μg/ml.

The G1 domains used included biotinylated HA binding protein (HABP; Seikagaku America, Falmouth, MA), as described and characterized previously (18), affinity-purified G1-ITEGE, and affinity-purified G1-DIPEN. The G1-decorated HA was added to the chondrocytes or COS-7 cells for 1 hour at 4°C to optimize the visualization of the cell surface–bound probe (18,25). For internalization studies, chondrocytes or COS-7 cells were incubated with the G1-decorated HA for 3 hours at 37°C and then treated with 0.25% trypsin for 30 minutes at 37°C. The cells were washed, fixed in 1% paraformaldehyde, quenched with 0.2M glycine in PBS, and then permeabilized with 0.1% Triton-X for 15 minutes at room temperature. The fixed chondrocytes or COS-7 cells were then incubated with primary and secondary antibodies as described above.

RESULTS

Generation of endogenous G1-ITEGE by bovine articular chondrocytes

Bovine articular chondrocytes were released from alginate beads after 5 days in culture, fixed, and then incubated with anti-ITEGE antibody (Antibody D, Figure 1). Endogenous ITEGE epitope could be clearly visualized on the cell surface (as well as intracellular sites) by immunofluorescence using FITC-conjugated secondary antibodies (Figure 2A). Only faint background fluorescence was observed under conditions in which the primary antibody was absent (results not shown). When the chondrocytes were preincubated for 24 hours in the presence of 10 ng/ml IL-1α, there was a significant increase in the level of total detectable ITEGE epitope, as quantified by flow cytometry (Figure 2B) (control chondrocyte mean channel fluorescence 15.9; mean channel fluorescence in the presence of IL-1 44.3). This 2.8-fold increase could also be visualized by comparing low-power images of chondrocytes incubated without or with IL-1α (Figures 2F and G).

Figure 2.

Generation of endogenous G1-ITEGE by bovine articular chondrocytes. Bovine articular chondrocytes released from alginate beads were fixed and then incubated with anti-ITEGE antibody (Antibody D, Figure 1). A, Total endogenous G1-ITEGE. B, Expression of total G1-ITEGE expressed by chondrocytes in the absence (no interleukin-1 [IL-1]) or presence (+ IL-1) of pretreatment exposure to IL-1α was quantified by flow cytometry. C, Endogenous ITEGE epitope subsequent to trypsinization, fixation, and permeabilization; both antibodies were visualized using fluorescein isothiocyanate–conjugated secondary antibodies. Only faint background fluorescence was observed under conditions in which the primary antibody was absent (results not shown). Intracellular ITEGE neoepitope expressed by chondrocytes treated without (D) or with (E) IL-1α, as detected using rhodamine X–conjugated Fab fragment as a secondary antibody. F and G, Total G1-ITEGE neoepitope expressed by chondrocytes treated without (F) or with (G) IL-1α was also viewed at lower power by immunofluorescence microscopy for comparison; F and G have been digitally lighted by a matched, equivalent amount to enhance the weaker fluorescence observed with low-power objectives. The nuclei are identified by blue fluorescence due to 4′-6-diamidino-2-phenylindole staining.

To determine whether the ITEGE-containing G1 domains were internalized, the chondrocytes were treated for 30 minutes with trypsin following release from the beads, fixed, permeabilized, and then incubated with anti-ITEGE antibody, followed by either FITC-conjugated secondary antibody (Figure 2C) or rhodamine X–conjugated Fab fragment secondary antibody (Figures 2D and E). Intracellular ITEGE, localized within spherical vesicles, was clearly detectable using either green or red fluorescence. Following incubation in the presence of IL-1α, a substantial increase in intracellular ITEGE epitope (Figure 2E) was also observed in comparison with untreated control chondrocytes (Figure 2D).

Endogenous G1-DIPEN is also generated by articular chondrocytes

Bovine articular chondrocytes were again released from alginate beads after 5 days in culture, fixed, and then incubated with anti-DIPEN antibody (Antibody B, Figure 1). Endogenous DIPEN epitope could be clearly visualized on the cell surface by immunofluorescence, using either FITC-conjugated secondary antibody (Figure 3A) or rhodamine X–conjugated Fab fragment secondary antibody (Figure 3C). To visualize intracellular DIPEN, the chondrocytes were treated for 30 minutes with trypsin following release from the beads, fixed, permeabilized, and then incubated with anti-DIPEN antibody followed by either FITC-conjugated secondary antibody (Figure 3B) or rhodamine X–conjugated Fab fragment secondary antibody (Figure 3D). Intracellular DIPEN was clearly detectable using either green or red fluorescence. Although the intracellular DIPEN was also localized in spherical vesicles (e.g., Figure 3D), few, if any, appeared localized within or in proximity to the cell nucleus.

Figure 3.

Generation of endogenous G1-DIPEN by bovine articular chondrocytes. Bovine articular chondrocytes released from alginate beads were fixed and then incubated with anti-DIPEN antibody (Antibody B, Figure 1). A, Total endogenous G1-DIPEN detected using fluorescein isothiocyanate (FITC)–conjugated secondary antibody. B, Intracellularly localized endogenous G1-DIPEN neoepitope detected using FITC-conjugated secondary antibody, subsequent to trypsinization, fixation, and permeabilization. C, Total endogenous G1-DIPEN detected using rhodamine X–conjugated Fab secondary antibody. D, Intracellularly localized endogenous G1-DIPEN neoepitope detected using rhodamine X–conjugated secondary antibody, subsequent to trypsinization, fixation, and permeabilization. The nuclei are identified by blue fluorescence due to 4′-6-diamidino-2-phenylindole staining.

Characteristics of the endogenous intracellular ITEGE neoepitope generated by bovine articular chondrocytes

Bovine articular chondrocytes were released from alginate beads after 5 days in culture that included treatment with 10 ng/ml IL-1α on days 3–5. In 1 experiment, the chondrocytes were incubated with Lyso-Tracker Red for 5 minutes prior to trypsinization. The cells were then washed, fixed, permeabilized, and incubated with anti-ITEGE antibody. Two representative images in Figures 4A and B show that the intracellular ITEGE epitope (green fluorescence) appeared in small vesicles within the cytoplasm and nucleus, but did not colocalize with the LysoTracker Red probe (red fluorescence), demarking low-pH organelles such as late endosomes and lysosomes. In our previous study (18), internalized fluorescein-conjugated HA and LysoTracker Red colocalization was clearly evident as the digitally overlaid images depicted the blend of red and green fluorescence as yellow fluorescence. Few of the vesicles seen in Figures 4A and B appeared yellow, however, and there were no red fluorescent vesicles within the nucleus. These results suggest that internalized ITEGE is not in the same intracellular compartment as internalized HA.

Figure 4.

Characteristics of endogenous intracellular ITEGE neoepitope generated by bovine articular chondrocytes. Bovine articular chondrocytes were released from alginate beads following a 48-hour pretreatment with 10 ng/ml interleukin-1α. A and B, Overlay images of intracellular endogenous G1-ITEGE (green fluorescence) and Lyso-Tracker Red probe demarking low-pH organelles (red fluorescence). C and D, Intracellular endogenous G1-ITEGE (C) did not colocalize with internalized hyaluronan binding probe (D), detected by streptavidin Texas Red. E, Two-color overlay of the 4′-6-diamidino-2-phenylindole-stained images in C and D. F and G, Isolated chondrocyte nuclei were fixed and incubated with anti-G1 antibody (F) and anti-ITEGE antibody (G); both were detected with fluorescein isothiocyanate (FITC)–conjugated secondary antibody. H, Intracellular endogenous G1 domain was observed in cytoplasm-localized vesicles (FITC-conjugated secondary antibody). I, Intracellular, endogenous FFGVG peptide neoepitopes (Antibody C, Figure 1) were also observed in cytoplasm-localized vesicles (detected with rhodamine X–conjugated Fab secondary antibody).

In a second experiment, chondrocytes were released from beads and incubated for an additional 24 hours with exogenous HA decorated with biotinylated HABP-G1 domain. The cells were then trypsinized and intracellular ITEGE neoepitope was detected. Intracellular ITEGE was again depicted by green fluorescence (Figure 4C). As in previous views, the neoepitope was localized predominantly in the central region of the cell (a region that includes the nucleus, as seen in Figure 4E). In contrast, internalized HA decorated with HABP (red fluorescence) was localized within larger vesicles exclusively outside of the central region (Figure 4D). Thus, intracellular ITEGE does not colocalize with newly internalized exogenous HA decorated with HABP-G1 domain. Figure 4E shows a 3-color overlay of Figures 4C and D, as well as blue DAPI staining for the nucleus; there was little colocalization of vesicles containing green and red fluorescent components, indicating that ITEGE neoepitope does not colocalize with HABP-G1 domain.

In a third experiment, chondrocytes were released from beads, the cells swelled in hypotonic buffer, lysed, and the individual nuclei isolated. These nuclei were then fixed and incubated with anti-G1 antibody (Antibody A, Figure 1) or anti-ITEGE antibody. Figure 4F shows that there was no intact G1 domain epitope present in the isolated nuclei, whereas the ITEGE neoepitope could be clearly observed in these cell-free nuclear preparations (Figure 4G). Often the ITEGE associated with the nucleus was present within small vesicles, as shown on the right side of Figure 4G; however, in other instances, larger vesicles were observed seemingly attached to the outer edge of the nuclei (left side of Figure 4G). These results suggest that the ITEGE epitope that is associated with the nucleus no longer contains an intact G1 domain. One possibility is that the ITEGE-containing G1 domain has undergone additional proteolytic processing prior to its targeting to the nucleus.

Another explanation for these results is that the aggrecanase-generated G1 fragments are subjected to further enzymatic processing such as a second cleavage by MMP, and it is this ITEGE-containing peptide that is selectively transported to the nucleus. If the additional proteolytic processing involved cleavage by MMP, a new amino-terminal neoepitope would be generated. As illustrated in Figure 1, this peptide would exhibit an amino-terminal sequence of FFGVG (recognized by Antibody C), yet still include the carboxy-terminal ITEGE (recognized by Antibody D). To determine whether this peptide was present intracellularly, chondrocytes were released from beads, trypsinized, fixed, permeabilized, and incubated with either anti-G1 antibody (Antibody A, Figure 1) or anti-FFGVG antibody (Antibody C, Figure 1). As shown in Figure 4H, the G1 domain itself can be visualized within trypsinized and permeabilized chondrocytes. Thus, G1 domains are internalized by chondrocytes but do not appear to be delivered to the nucleus.

Similarly, the FFGVG neoepitope is also generated in these cultures, and can be detected (Figure 4I). However, the FFGVG peptides are also localized in vesicles outside the nucleus. At this point it is not known whether the FFGVG was generated extracellularly and then internalized, or whether it was generated by intracellular processing. Nonetheless, it appears that only the ITEGE neoepitope becomes localized within the nucleus of chondrocytes, unlike HA or other related G1 domain epitopes. Additional proteolytic processing of G1-ITEGE may be occurring; however, it is likely that cleavage occurs at sites other than the VDIPEN-FFGVG site.

Intracellular endogenous ITEGE neoepitope originates from neoepitope bound at the cell surface

Forty-eight hours prior to release from alginate beads, the bovine articular chondrocytes were incubated with a dilute solution of Streptomyces hyaluronidase to remove HA-bound proteins at the cell surface. Upon release, the cells were either fixed and stained immediately with the anti-ITEGE antibody (Figure 5A), or trypsinized, fixed, permeabilized, and then incubated with the anti-ITEGE antibody (Figure 5B). No ITEGE epitope was detected at the cell surface nor within the cytoplasm or nucleus of the cells (green fluorescence), compared with untreated control chondrocytes, which showed abundant cellular staining (Figure 5C). As another approach, 48 hours prior to release from alginate beads, the chondrocytes were incubated with the anti-CD44 antibody IM7.8.1. Previous results from our laboratory have shown that this antibody functions as a partial blocking antibody for HA binding in bovine chondrocyte CD44 (33). When the cells were released and incubated with anti-ITEGE antibody, no ITEGE epitope was detectable at the cell surface of the chondrocytes (i.e., no green fluorescence was observed; Figure 5D). Following trypsinization (Figure 5E), a barely detectable level of nuclear intracellular ITEGE neoepitope was observed.

Figure 5.

Requirement of hyaluronan and CD44 for the retention and internalization of endogenous G1-ITEGE. Bovine articular chondrocytes were released from alginate beads following a 48-hour pretreatment with a dilute solution of Streptomyces hyaluronidase (A and B), no pretreatment (C), a 48-hour preincubation with anti-CD44 antibody IM7.8.1 (D and E), or preincubation with 10 ng/ml interleukin-1α (IL-1α) to enhance the generation of endogenous G1-ITEGE (F–H). Upon release, the cells were either fixed and stained immediately with the anti-ITEGE antibody (A, C, and D) or trypsinized, fixed, permeabilized, and then incubated with the anti-ITEGE antibody (B, E–H). In A–E, G1-ITEGE was detected using fluorescein isothiocyanate–conjugated secondary antibody. F shows a 2-color overlay of IL-1α–pretreated chondrocytes transfected with pCD44Δ67, trypsinized, fixed, permeabilized, and then incubated with the anti-ITEGE antibody followed by detection with a rhodamine X–conjugated Fab secondary antibody. The nuclei are identified by blue fluorescence due to 4′-6-diamidino-2-phenylindole staining. G shows nuclei in the red fluorescence channel alone, used to document the complete lack of red fluorescence (representing G1-ITEGE) of 2 cells at the right of the field. H shows a 3-color overlay that includes the green fluorescence channel; successfully transfected cells (pCD44Δ67-expressing cells) in this panel are identified by the intense expression of green fluorescent protein.

In a third experiment, 48 hours prior to release from alginate beads, the chondrocytes were preincubated with 10 ng/ml IL-1α to enhance generation of G1-ITEGE. The cells were released from the beads, placed into monolayer culture, and then transfected with pCD44Δ67 (subcloned into a GFP-positive pTracer expression plasmid). We have shown previously that expression of CD44Δ67, a truncated CD44 mutant, acts as a dominant-negative receptor, blocking all HA binding functions dependent on full-length native CD44 (25). Forty-eight hours posttransfection the cells were trypsinized, fixed, permeabilized, and incubated with anti-ITEGE antibody (red fluorescence). Because the level of cell transfection never reaches 100%, successfully transfected chondrocytes (GFP-positive) can be viewed within the same field as nontransfected cells (GFP-negative).

As shown in the 2-color overlay (Figure 5F), a cluster of chondrocytes exhibited positive expression of the ITEGE neoepitope, with the majority of the epitope (shown in red fluorescence) colocalized with the cell nucleus (blue fluorescence). However, another group of cells to the right in the field of view did not exhibit intracellular ITEGE (Figure 5G). These same cells that did not display the ITEGE neoepitope were the cells transfected by CD44Δ67, as noted by their intense expression of GFP (Figure 5H). All 3 of these approaches suggest that blocking HA binding to CD44 at the cell surface of chondrocytes, by the use of either hyaluronidase, anti-CD44 antibody, or a dominant-negative CD44, blocks the binding and subsequent internalization of G1-ITEGE.

Binding and internalization of exogenous, purified G1-ITEGE depends on expression of CD44

Aggrecan G1-ITEGE was isolated and purified. The purified protein was then incubated with high molecular mass HA, allowed to aggregate, and then repurified by ethanol precipitation. The HA, decorated with G1-ITEGE, was incubated with COS-7 cells or bovine articular chondrocytes for 1 hour at 4°C to visualize cell surface–bound G1, or for 3 hours at 37°C followed by trypsinization to visualize internalized G1. The ITEGE epitope was then detected using an anti-ITEGE antibody. In the first experiment, COS-7 cells (which are naturally CD44 negative) were transfected with full-length human pCD44. As shown in Figure 6A, GFP-positive COS-7 cells (inset) gained the capacity to bind ITEGE-containing G1 domains onto the cell surface (red fluorescence). To visualize internalized G1-ITEGE, the COS-7 cells were trypsinized, fixed, and permeabilized and then incubated with anti-ITEGE (Figure 6B, red fluorescence). Internalized G1-ITEGE was detected (see 2-color images in Figure 6). A proportion of the internalized G1-ITEGE appeared to localize with the nucleus.

Figure 6.

Binding and internalization of hyaluronan (HA)–G1-ITEGE complexes by transfected COS-7 cells and bovine articular chondrocytes. COS-7 cells were transfected with full-length human pCD44, and bovine articular chondrocytes were transfected with pCD44Δ67. A–F, Forty-eight hours posttransfection, the cells were incubated with HA–G1-ITEGE complexes for 1 hour at 4°C (A and D) to visualize cell surface–bound G1-ITEGE, or for 3 hours at 37°C (B, C, E, and F) followed by trypsinization, fixation, and permeabilization to visualize internalized G1-ITEGE. For experiments with chondrocytes, the cells were preincubated with Streptomyces hyaluronidase and rinsed prior to the addition of the HA–G1-ITEGE complexes. G1-ITEGE epitope was then detected using anti-ITEGE antibody, followed by rhodamine X–conjugated Fab secondary antibody. A, Total G1-ITEGE present on green fluorescent protein (GFP)–positive COS-7 cells (inset shows 3-color overlay). B, Intracellular G1-ITEGE present on GFP-positive COS-7 cells (3-color overlay shown by upper cells in the panel). C, Intracellular G1-ITEGE present on a representative GFP-negative COS-7 cell (3-color overlay shown by upper cell in the panel). D, Total G1-ITEGE present on GFP-negative chondrocytes. E, Intracellular G1-ITEGE present on GFP-negative cells. F, Intracellular G1-ITEGE present on a representative GFP-positive chondrocyte (3-color overlay).

Binding and internalization of G1-ITEGE was observed only in successfully transfected (GFP-positive) COS-7 cells (see upper images in Figure 6B). G1 domains were not detected on the surface of nontransfected COS-7 cells (results not shown), and were not detected intracellularly in nontransfected COS-7 cells (Figure 6C). Thus, CD44 expression in these nonchondrogenic cells was required for internalization and G1-ITEGE retention at the cell surface.

To determine the requirement for CD44 in chondrocyte internalization of G1-ITEGE, a loss-of-function approach was used. Again, purified HA decorated with G1-ITEGE was added to the cells. However, for these experiments, the chondrocytes were first preincubated with Streptomyces hyaluronidase, washed, and then incubated with a complex of HA–G1-ITEGE. As shown in Figure 6D, HA decorated with G1-ITEGE displayed prominent binding at the cell surface of the bovine articular chondrocytes (red fluorescence). Following trysinization, fixation, and permeabilization, the HA decorated with G1-ITEGE domains (Figure 6E) could be clearly visualized intracellularly and localized predominantly in association with the nucleus. Bovine articular chondrocytes transfected with the CD44 dominant-negative receptor CD44Δ67 (GFP-positive chondrocyte shown in Figure 6F) displayed no capacity to bind or internalize HA–G1-ITEGE complexes. These results further support the observation that CD44 is required for the retention and internalization of G1-ITEGE. Experiments involving the addition of HA decorated with purified G1-DIPEN to COS-7 cells and bovine articular chondrocytes showed that CD44 was also required for internalization of preformed HA–G1-DIPEN complexes (results not shown).

DISCUSSION

In this study, the catabolic fate of residual aggrecan G1 domains was explored using adult articular chondrocytes released from alginate beads after 5 days in culture. The 5-day bead culture model allows chondrocytes to recover from the stress of enzymatic isolation from the cartilage, and the 3-dimensional environment serves to promote the maintenance of the chondrocyte phenotype (34,35). During this recovery period, the proteoglycan-rich cell-associated matrix is reestablished and cell surface proteins, such as CD44, are resynthesized. More relevant to this study, however, is that after 5–7 days of culturing in alginate beads the chondrocytes approach steady-state conditions of aggrecan metabolism, conditions in which the turnover of aggrecan begins to balance biosynthesis (34). In this study, we demonstrate that cell-based catabolism of aggrecan does occur after 5 days of culture, resulting in the expression of both G1-DIPEN and G1-ITEGE fragments retained within the cell-associated matrix of chondrocytes in alginate beads (Figures 2A, 2F, 2G, 3A, and 3C). This implies that, under these culture conditions, the chondrocytes have not only synthesized sufficient aggrecan to reestablish a cell-associated matrix (28), but have also begun maintenance turnover of this proteoglycan by ADAMTS proteinases, as well as by MMPs.

Reagents such as IL-1α or retinoic acid are often used to enhance aggrecan turnover by chondrocytes (9,24,36). As shown in Figures 2B and 2D–G, endogenous G1-ITEGE was substantially up-regulated by IL-1α, with only a minimal change in the accumulation of G1-DIPEN. This differential sensitivity to IL-1α is consistent with expression of G1-ITEGE and G1-DIPEN observed in porcine cartilage explant cultures (9). The IL-1–induced increase in G1-ITEGE supports the concept that the generation of these aggrecan fragments is due to a cell-derived inducible proteinase activity. It is also interesting to note that during 48-hour incubation with IL-1α, aggrecan catabolites, including G1-ITEGE, accumulate even though synthesis of aggrecan has been significantly inhibited, as has been shown in numerous studies (22,24,37).

Another exciting observation from our study was that a proportion of the G1-ITEGE and G1-DIPEN present at the surface of chondrocytes was internalized by these cells. Intracellular localization of G1 domains was defined morphologically as neoepitope concentrated within small, round vesicular structures, and functionally as neoepitope resistant to extensive trypsinization of the cell surface. The endocytosis of G1 domains also appears linked to the larger overall mechanism of aggrecan catabolism. The IL-1α–induced increases in cell surface ITEGE neoepitope (e.g., Figures 2F and G) were paralleled by an increased accumulation of intracellular ITEGE neoepitope (Figures 2D and E). Thus, IL-1 increases not only ADAMTS expression and activity (4,38–44), but also the relevant receptors, adapter proteins, and cytoskeletal components necessary for the internalization of the residual aggrecan G1 domains. We have shown previously that IL-1 treatment of chondrocytes results in a 6–8-fold increase in CD44 protein and a 3-fold increase in accumulation of internalized HA (22).

The retention of G1 domains at the chondrocyte plasma membrane and the internalization of G1 domains is dependent on the presence of HA and CD44. There is no retention or internalization of G1-ITEGE in chondrocytes exposed to dilute Streptomyces hyaluronidase (Figures 5A and B), antibodies to CD44 (Figures 5D and E), or chondrocytes expressing a CD44 dominant-negative receptor (Figures 5F–H). These treatments interfere with the extracellular retention of HA on chondrocytes, thus implying that all intracellularly localized G1 domains were derived from uptake at the cell surface. In other words, intracellular G1-ITEGE or G1-DIPEN neoepitopes are not the result of intracellular processing of newly synthesized aggrecan, as occurs with misfolded aggrecan in nanomelia (45). That the intracellular G1-ITEGE or G1-DIPEN fragments are derived from an extracellular pool is further substantiated by the binding and endocytosis of exogenous purified HA–G1-ITEGE and HA–G1-DIPEN complexes by chondrocytes and CD44-transfected COS-7 cells (Figure 6).

These experiments also demonstrated the requirement of CD44 expression for the retention and endocytosis of bound G1 domains. Only COS-7 cells transfected with full-length CD44 exhibited the capacity to bind and internalize HA–G1-ITEGE complexes. Conversely, chondrocytes expressing the CD44 dominant-negative mutant lost the capacity to bind or internalize purified HA–G1-ITEGE complexes. Our previous studies on the endocytosis of HA by CD44 showed that the internalization does not represent fluid-phase pinocytosis, that the initial internalized HA is of high molecular mass (21), and that there do appear to be steric size limitations to this process because HA decorated with intact aggrecan monomers is not internalized (18). These results imply that only the pool of aggrecan G1 domains that remains bound to HA and, in turn, bound to CD44 has the potential to undergo endocytosis and presumably terminal catabolism. Free G1 domains, or HA-bound G1 domains entrapped within extracellular matrix, would presumably exhibit a substantially longer turnover half-life.

These results differ from those of Fujimoto et al, who reported the direct binding of aggrecan chondroitin sulfate chains to CD44, a binding that was not susceptible to Streptomyces hyaluronidase (46). In our study, the retention of aggrecan G1 domains was sensitive to Streptomyces hyaluronidase, and furthermore, these domains do not contain chondroitin sulfate chains. Nonetheless, it is possible that some CD44 receptors bind G1 domains via HA, while others interact with intact aggrecan monomer via their chondroitin sulfate chains. However, in our previous studies, the retention of aggrecan to CD44 has always displayed susceptibility to Streptomyces hyaluronidase, small HA oligosaccharides, or non-sulfated chondroitin (18,33,47,48), but not chondroitin sulfate (49).

One of the intriguing observations of this study is the nuclear localization of the ITEGE neoepitope compared with other intracellular G1 domain epitopes and HA itself. While HA and HABP colocalize with the low-pH organelle marker LysoTracker Red (18,23), the ITEGE neoepitope exhibited no such colocalization within a low-pH organelle (Figures 4A and B), or with internalized HABP (Figures 4C–E). In repeated experiments, the ITEGE neoepitope appeared in cytoplasmic vesicles, but also uniquely concentrated in the nucleus of the chondrocytes (Figures 2, 4, 5, and 6). From 2-dimensional images, it is impossible to conclude that the neoepitope is within the nucleus, rather than above or below it. Therefore, to verify this association, individual chondrocyte nuclei were isolated. Again, the ITEGE neoepitope was colocalized within individual nuclei (Figure 4G).

Another interesting observation was that the size of the ITEGE-positive vesicles associated with the nuclei was typically smaller than that of the vesicles containing G1-DIPEN, HABP, or HA. However, larger ITEGE-positive vesicles were sometimes observed on the periphery of the nucleus (left side of Figure 4G). Similar, larger and more peripheral ITEGE-positive nuclear inclusions were seen in the COS-7 cells and chondrocytes that had internalized purified, exogenous, HA–G1-ITEGE complexes. This may imply a transition from larger vesicles bound at the nuclear membrane to the ITEGE neoepitope distributed throughout the nuclear matrix.

While the ITEGE neoepitope remained in association with isolated nuclei (Figure 4G), no G1 domain epitope was present in the nuclei (Figure 4F). Since the G1 epitope is present within the cytoplasmic organelles (Figure 4H), the results suggest that some form of enzymatic processing occurred, releasing a peptide containing the carboxy-terminal ITEGE neoepitope, a peptide that subsequently becomes targeted to the nucleus. The nature of this targeting or the exact processing that occurs is unknown. Several investigators have described the association of HA with the nucleus (50,51). However in this case, the ITEGE neoepitope–containing peptide is being delivered to the nucleus without an intact HA-binding amino-terminal domain (Figures 4F and H). The other G1 domains, G1-DIPEN (Figure 3) or HABP (Figure 4D), presumably bound to HA at the time of endocytosis, were not associated with chondrocyte nuclei, as is the ITEGE-containing peptide. Thus, it is unlikely that HA is being used as a transport vehicle to target the ITEGE peptide to the nucleus.

Furthermore, we have not observed nuclear-localized fluorescein–HA in our previous studies of chondrocytes (18); this may be due to the short time frame of the studies (3 hours) or because the chondrocytes are not highly proliferative. Again, nuclear ITEGE accumulation was blocked when extracellular binding of G1-ITEGE to chondrocytes (Figures 4 and 6) or to COS-7 cells was blocked. Thus, the ITEGE neoepitope must be bound to HA (i.e., part of a functional G1 domain) to be internalized initially. Then, following a subsequent intracellular processing event, a proportion of the ITEGE-containing peptide is targeted to the nucleus. The function, if any, of this nuclear localization is unknown. The intracellular processing events are also unknown, except that the event is not due to the action of an MMP that would liberate an FFGVG and ITEGE-containing peptide.

We speculate that the new amino-terminal generated on the ITEGE-containing peptide likely represents the sequence that is recognized for nuclear targeting. Otherwise, G1-ITEGE as well as the FFGVG-ITEGE peptides would have also been delivered to the nucleus. Thus, identification of the new N-terminal of this ITEGE-containing peptide and of its potential chaperone recognition protein represent exciting future aims. It is interesting to speculate that this nuclear-localized ITEGE peptide might represent an end-product, feedback inhibition signal factor that functions to down-regulate genes involved in aggrecan catabolism. Whether this supposition can be borne out by further research remains to be seen.