Irreversible Heavy Chain Transfer to Hyaluronan Oligosaccharides by Tumor Necrosis Factor-stimulated Gene-6

Background: TSG-6 transfers heavy chains from inter-α-inhibitor to hyaluronan.

Results: Heavy chain transfer to hyaluronan by TSG-6 is reversible for high molecular weight hyaluronan but irreversible for hyaluronan oligosaccharides.

Conclusion: High molecular weight hyaluronan functions as both a heavy chain acceptor and a heavy chain donor, whereas hyaluronan oligosaccharides function only as heavy chain acceptors.

Significance: Hyaluronan oligosaccharides have potential to remove heavy chains from pathological hyaluronan.

Abstract

The covalent transfer of heavy chains (HCs) from inter-α-inhibitor (IαI) to hyaluronan (HA) via the protein product of tumor necrosis factor-stimulated gene-6 (TSG-6) forms the HC-HA complex, a pathological form of HA that promotes the adhesion of leukocytes to HA matrices. The transfer of HCs to high molecular weight (HMW) HA is a reversible event whereby TSG-6 can shuffle HCs from one HA molecule to another. Therefore, HMW HA can serve as both an HC acceptor and an HC donor. In the present study, we show that transfer of HCs to low molecular weight HA oligosaccharides is an irreversible event where subsequent shuffling does not occur, i.e. HA oligosaccharides from 8 to 21 monosaccharide units in length can serve as HC acceptors, but are unable to function as HC donors. We show that the HC-HA complex is present in the synovial fluid of mice subjected to systemic and monoarticular mouse models of rheumatoid arthritis. Furthermore, we demonstrate that HA oligosaccharides can be used, with TSG-6, to irreversibly shuffle HCs from pathological, HMW HC-HA to HA oligosaccharides, thereby restoring HC-HA matrices from the inflamed joint to their normal state, unmodified with HCs. This process was also effective for HC-HA in the synovial fluid of human rheumatoid arthritis patients (in vitro).

Introduction

Under normal conditions, hyaluronan (HA)² is present in nearly every mammalian tissue as a large (typically >1500-kDa) glycosaminoglycan lacking protein modifications of any kind. During many inflammatory events, a unique protein modification is made on HA that significantly promotes the adhesion of leukocytes (Fig. 1A) (1). This protein modification is the covalent transfer of heavy chains (HCs) from inter-α-inhibitor (IαI) to HA to form the HC-HA complex (2, 3). The enzyme that mediates this process is the protein product of tumor necrosis factor-stimulated gene-6 (TSG-6) (4–7). The HC donor IαI is composed of 3 polypeptides: the trypsin inhibitor called “bikunin” (∼16 kDa) and two HCs (∼75–85 kDa each). Each HC is covalently attached to the single chondroitin sulfate (CS) chain of bikunin by an ester linkage between an aspartate in the HC and the 6-OH of N-acetylgalactosamine in CS. TSG-6 transesterifies the HCs to the 6-OH on N-acetylglucosamine in HA. Five different homologous HCs (HC1–5) are present in two gene clusters mapped to different chromosomes (HC1, HC3, and HC4 to chromosome 3, and HC2 and HC5 to chromosome 10) (8). HC4 is unique in that it is not part of IαI, is not transferred to HA, and is secreted as a free HC species (9, 10).

FIGURE 1.

Irreversible heavy chain transfer to hyaluronan oligosaccharides. Panel A, serum-derived IαI is composed of three polypeptides, namely, the protease inhibitor bikunin (blue) and two HCs (green), linked through a single CS chain. During inflammation, as in the synovial fluid of rheumatoid arthritis patients, TSG-6 transfers HCs from IαI onto HMW HA molecules (black) to form the pathological HC-HA complex. Panel B, it is known that TSG-6 reversibly swaps HCs between large HA strands until the HCs are evenly distributed between HA molecules. C, we hypothesized that HC swapping is irreversible when swapped onto much smaller HA oligosaccharides. This irreversible swapping of HCs could be exploited to remove HCs from pathological HC-HA matrices found in RA and other inflammatory diseases, thereby disabling the inflammatory cell retention mechanism and silencing inflammatory signaling cascades at their source.

HA is present at relatively high levels (3–4 mg/ml) in the synovial fluid (SF) of healthy articular joints where it functions as a hydrated, viscoelastic lubricant (11). The HC-HA complex has been detected in the SF of patients with rheumatoid arthritis (RA), and leukocytes have been shown to preferentially bind HC-HA when compared with HA alone (1). This finding is significant because leukocyte counts from inflamed SF are 10–500-fold above normal levels and are important factors in driving irreversible inflammatory changes in RA disease.

HC transfer from IαI to HA is a dynamic process whereby TSG-6 can shuffle HCs between different HA molecules in a reversible manner (12) (Fig. 1B). In this study, we tested the hypothesis that low molecular weight (LMW) HA oligosaccharides are irreversible HC acceptors in contrast to higher molecular weight HA molecules, which function as both HC acceptors and HC donors (Fig. 1C). If true, then HA oligosaccharides could be used as inert HC acceptors to deplete HCs from pathological HC-HA matrices, thereby serving as “bait” to remove the signal for leukocyte adhesion to HA matrices.

EXPERIMENTAL PROCEDURES

Reagents

IαI/pre-IαI sources were either mouse serum (S7273; Sigma-Aldrich) or human serum as noted (donor 736; Equitech-Bio Inc., Kerrville, TX). They were not purified from other serum components with the exception that the human serum was dialyzed into a phosphate-buffered saline (PBS) solution containing 1 mm MgCl₂ using a 20-kDa molecular mass cut-off membrane. We dialyzed the serum to rule out the possibility that endogenous LMW HC acceptors were present in serum and responsible for the HC gel shift when serum was incubated with TSG-6 alone. Because dialysis had no effect on this phenomenon (data not shown), the mouse serum was not dialyzed. The mouse serum was also used in the presence of PBS and 1 mm MgCl₂ because the ability of TSG-6 to transfer HCs is dependent upon divalent cations (7). TSG-6 incubated with human serum dialyzed into PBS containing no divalent cations was unable to transfer HCs to HA, whereas nondialyzed human serum and serum dialyzed into PBS containing 1 mm MgCl₂ showed equal ability to promote HC transfer to HA (data not shown). The rest of the experiments in this study used human serum dialyzed into PBS containing 1 mm MgCl₂ defining the amount and identity of the divalent cation present. Because the mouse serum was not dialyzed, other cations were present. In addition, PBS and an additional 1 mm MgCl₂ were also added to the samples containing mouse serum.

Recombinant human TSG-6 (2104-TS; R&D Systems, Minneapolis, MN) was resuspended at 0.005 mg/ml. HA oligosaccharides 4, 6, 8, and 10 monosaccharide units in length and monodisperse 1,000-kDa HA (¯M_w/¯M_n = 1.007), where ¯M_w and ¯M_n indicate weight average molecular weight and weight average molecular number, were purchased from Hyalose, L.L.C. (Oklahoma City, OK). HA of 12, 14, 18, and 21 monosaccharide units in length was kindly provided by Hyalose. Antibodies against HC1 (SC-33944 anti-mouse; SC-21968 anti-human), HC2 (SC-21978 anti-mouse; SC-21975 anti-human), or HC3 (SC-21979 anti-mouse and human) (Santa Cruz Biotechnology) were employed. Streptomyces hyaluronidase (100740-1, Seikagaku, East Falmouth, MA) was used at 8 milliunits/μl of reaction volume.

Heavy Chain Transfer and Swapping Reactions

The reaction volumes were 25 μl of PBS containing 1 mm MgCl₂, 5% serum supplemented with 1.25 μg HA, and/or 0.005 μg of TSG-6. TSG-6 was always added last and marked the beginning of the incubation period (time 0). EDTA was used to stop the reaction by adding 0.5 μl of a 0.5 m (pH 8.0) solution. In HC swapping studies, HCs were first transferred to HA molecules of a specific size, and after a specified time, equal mass amounts of HA molecules of a different size were added to the same reaction mixture for a specified time.

Fluorophore-assisted Carbohydrate Electrophoresis

Fluorophore-assisted carbohydrate electrophoresis (FACE) was used to assess the purity of the HA oligosaccharides used in the size-range study. This method has been previously described (13).

Models of Experimental Rheumatoid Arthritis in Mice

The systemic proteoglycan-induced arthritis (PGIA) and the monoarticular antigen-induced arthritis (AIA) models of rheumatoid arthritis in BALB/c mice have been previously described (14–16). Synovial fluids were obtained from several affected knee and ankle joints. The samples were harvested 7–10 days after the onset of arthritis using a 25-gauge needle and by lavaging the joints with an equal or double volume of saline containing 0.5 unit of heparin, which was then pooled with the original synovial fluid extract. Typically, the extracts were 4–6 μl from ankle and knee joints of mice with PGIA and 8–10 μl from the affected knee joints of mice with AIA. The extracts and lavages from several knees (n = 3–4) or ankles (n = 3–4) were pooled. For AIA, an extra 10-fold saline lavage (by synovial fluid volume) was used. When loading the TGX electrophoresis gel (described later), equal volumes of synovial fluid were added to each lane, taking into consideration the -fold dilution of the synovial fluid by the saline lavages. Synovial fluid samples (2.5 μl) were incubated with 2 units of Streptomyces hyaluronidase in a final reaction volume of 25 μl in PBS for 18 h at 37 °C before loading onto the gel. Loading 2.5 μl of 1:2 diluted synovial fluid, pretreated with hyaluronidase, per lane gave a strong heavy chain signal by Western blot with anti-HC antibodies.

Irreversible Transfer of Heavy Chains to Hyaluronan Oligosaccharides in Human Synovial Fluid

Synovial fluid was collected from the knee joint of RA patients during a routine arthrocentesis at their clinic visit under approved Institutional Review Board (IRB) protocols with samples de-identified. The analysis was the same as described for the mouse synovial fluid.

Western Blot Analysis

Samples were electrophoresed on 4–15% Mini-Protean TGX gels (Bio-Rad) and blotted using the Bio-Rad nitrocellulose and Trans-Blot Turbo system. Samples of 25 μl with 1.25 μl of serum gave a strong HC signal on the blots with antibodies used in this study. The molecular weight standard was purchased from LI-COR (928-40000). The blots were blocked for 1 h with LI-COR blocking buffer (927-40000; LI-COR) and then probed with antibodies against HC1, HC2, or HC3 (dilution 1:1000) in the blocking buffer with 0.1% Tween 20 for 1 h. The blots were washed 5× in PBS with 0.1% Tween 20 and probed with an IRDYE secondary antibody (LI-COR; 926-32214) at 1:15,000 dilution in blocking buffer with 0.1% Tween 20 and 0.01% lauryl sulfate for 45 min. The blots were washed as before and imaged on an Odyssey infrared imaging system (LI-COR).

RESULTS

Kinetics of HC Transfer from HC-Bikunin to Hyaluronan

It has been previously reported that the enzyme TSG-6 binds HC-bikunin molecules prior to forming a HC-TSG-6 intermediate (7). Subsequent noncovalent binding by this HC-TSG-6 intermediate to HA via the TSG-6 link module allows transfer of HCs to HA. We were interested in the effect of HA size on the kinetics of HC transfer by TSG-6 from bikunin to HA. Fig. 2 shows the kinetics of HC transfer by recombinant TSG-6 from human serum HC-bikunin to high molecular weight (HMW) 1,000-kDa HA (HA1000K, ∼5,300 monosaccharides in length, panel A) or to LMW ∼1.5-kDa HA oligosaccharide (HA8, 8 monosaccharides in length, panel B) or in the absence of HA (panel C). Human serum HC-bikunin is primarily in the form of IαI (two HCs per bikunin). In Fig. 2, all reactions were incubated for 24 h at 37 °C in PBS containing 1 mm Mg²⁺ with addition of EDTA at the times indicated (0–24 h, lanes 3–10) to stop the reaction (10 mm final concentration). In lane 2, serum alone (no TSG-6 or HA) incubated at 37 °C for 24 h was run as control. Blots were probed with a mixture of antibodies against human HC1- and HC2-specific peptides. Fig. 2 shows TSG-6-mediated transfer of HCs from HC-bikunin to HMW HA (Fig. 2A, HA1000k; HC disappears from the blot as the HMW prevents entry into the stacking gel or from transfer during blotting; hyaluronidase control was done but not shown) and to LMW HA oligosaccharide (Fig. 2B, HA8). Transfer was complete within 2–4 h with no detectable formation of a pre-IαI (one HC per bikunin) intermediate. Interestingly, in the absence of any HA (Fig. 2C), after 1 h of incubation, TSG-6 initially shuffled HCs on bikunin forming a mixture of bikunin molecules with one (pre-IαI), two (IαI), and three HCs per bikunin molecule followed by the slow release of free HCs by 4 h ending in release of all HCs as free HCs by 24 h. Once released or “chucked” as free HCs, these HCs could not be subsequently transferred by TSG-6 to HA of any size (Fig. 3).

FIGURE 2.

Kinetics of human serum HC transfer. Panels A–C, Western blots showing the kinetics of HC transfer by TSG-6 from human serum HC-bikunin (IαI) to HA1000K (molecular mass ∼1000 kDa, or ∼5,300 monosaccharides long) (A), HA8 (8 monosaccharides long) (B), or serum incubated with TSG-6 alone (no HA) (C). These blots were simultaneously probed with antibodies against human HC1 and HC2. All reactions were incubated for 24 h at 37 °C. Lane 2 shows untreated serum alone. In lane 3, EDTA was added to the reactants before TSG-6 was added, showing that EDTA stops the reaction. In lanes 4–10, EDTA was added 0.5, 1, 2, 4, 6, 8, and 24 h (respectively) after the reaction was started by the addition of TSG-6 (respectively). Molecular mass standards are shown in red on the left of the gels. Note that in panel A, the HC-HA complex does not enter the gel and thus vanishes from the blot. After 2–4 h, the reactions with HA were complete.

FIGURE 3.

Free HCs chucked by TSG-6 cannot be transferred to hyaluronan. Western blot simultaneously probed with antibodies against HC1 and HC2. Lanes 2 and 6 show human serum alone, incubated at 37 °C for 48 h. In lane 3, human serum was incubated with TSG-6 alone for 24 h, at which point EDTA was added and the reaction mixture was incubated for an additional 24 h. In lane 7, human serum was incubated with TSG-6 alone for 48 h, at which point EDTA was added. In lanes 4 and 5, human serum was incubated with TSG-6 alone for 24 h, and then an aliquot of HA14 or HA1000K (respectively) was added to the reaction mixture in the presence of EDTA for an additional 24 h. In lanes 8 and 9, human serum was incubated with TSG-6 alone for 24 h, and then an aliquot of HA14 or HA1000K (respectively) was added to the reaction mixture for an additional 24 h, after which EDTA was added.

Fig. 4 shows the kinetics of HC transfer by TSG-6 from mouse serum HC-bikunin, which is primarily in the form of pre-IαI (one HC) with lower amounts of IαI (two HCs). The predominance of pre-IαI in mouse serum and full-length IαI in human serum was a common trend found in other isolates (data not shown). In Fig. 4, blots were probed with a mixture of antibodies against mouse HC1- and HC2-specific peptides. Similar to human serum IαI, HC transfer from mouse serum IαI to HMW HA1000k (Fig. 4A) and LMW HA8 oligosaccharide (Fig. 4B) was completed very quickly (within 1 h), However transfer from mouse pre-IαI was relatively slow, requiring 24 h. Although release of free HCs from mouse IαI in the absence of HA (Fig. 4C) was similar to that of human IαI, release of free HCs from mouse pre-IαI was slower and incomplete by 24 h.

FIGURE 4.

Kinetics of mouse serum HC transfer. Panels A–C, Western blots showing the kinetics of HC transfer by TSG-6 from mouse serum HC-bikunin (pre-IαI and IαI) to HA1000K (A), HA8 (B), or serum incubated with TSG-6 alone (no HA) (C). The rest of the general description is the same as for Fig. 2, except the reactions with HA were complete in 1 h.

Effect of HC Isotype on Kinetics of HC Transfer from HC-Bikunin to Hyaluronan

We were interested in the role that HC isotype might have in the differences in kinetics seen between HC transfer by TSG-6 from human and mouse HC-bikunin (Figs. 2 and 4). Fig. 5 shows the kinetics of HC transfer by recombinant TSG-6 from human (panels A and B) and mouse (panel C) serum HC-bikunin to LMW HA8 oligosaccharides, but probed separately with HC isotype-specific antibodies. As seen in Fig. 5, A and B, human HC-bikunin contains both HC1 and HC2. The human HC1-bikunin was in the form of IαI with no detectable pre-IαI (panel A, lane 1). The human HC2-bikunin was predominantly in the form of IαI, but with detectable amounts of pre-IαI (panel B, lane 1). The relative staining of human HC1 was significantly stronger than staining for human HC2 (compare band intensities between panel A, lane 1 and panel B, lane 1a). To allow visualization of the bands in panel B, the gain was increased for lanes 1–10.

FIGURE 5.

Kinetics of transfer to hyaluronan of different HC isotypes. Panels A–C, Western blots, probed with HC isotype-specific antibodies, show the kinetics of HC1 (A) or HC2 (B and C) transfer from human (A and B) or mouse (C) serum-derived IαI and pre-IαI to HA8 (8 monosaccharides long) via the addition of recombinant TSG-6. Lane 1 shows serum alone. In lanes 2–9, EDTA was added 0, 0.5, 1, 2, 4, 6, 8, and 24 h, respectively, after the addition of TSG-6 to the solution of HA8 and serum to stop the reaction. The gain was increased for the image of lanes 1–10 in panel B in comparison with panels A and C to allow better visualization of all bands. Therefore, in panel B, the image for lane 1a was exposed similar to the images in panels A and C for comparison.

In Fig. 5A, transfer of HC1 from human IαI to HA8 was complete within 2–4 h with no detectable pre-IαI intermediate similar to the results in Fig. 2B. However, although transfer of the first HC2 from human IαI was completed within 2–4 h, transfer of the second HC2 was slower, generating a pre-IαI intermediate (Fig. 5B). By 24 h, there was still HC2 that had not been transferred from the pre-IαI intermediate to the HA8 oligosaccharide. Thus, the results from probing the blot in Fig. 2 with both HC1 and HC2 antibodies are representative of those obtained with the HC1 antibody alone in Fig. 5 as the antibody against HC1 gave a much stronger signal than the antibody against HC2 in human serum.

Mouse HC-bikunin was positive for HC2 (Fig. 5C), although completely negative when probed with a similar antibody against HC1 (not shown). The mouse HC2-bikunin was predominantly in the form of pre-IαI, with lower amounts of IαI (panel C, lane 1). However, although transfer of the first HC2 from mouse IαI was complete within 1 h, transfer of the second HC2 was slower, generating a pre-IαI intermediate (Fig. 5C). By 24 h, there was still HC2 that had not been transferred from the pre-IαI intermediate to HA8 oligosaccharide. This was similar to the transfer results for human HC2 in panel B. As no HC1 was detected by the HC1-specific antibody in mouse serum, the kinetics of transfer to HA8 seen in Fig. 5C are the same as in Fig. 4B.

Irreversible Transfer of HCs to Hyaluronan Oligosaccharides

It has previously been reported that TSG-6-mediated HC transfer from HC-bikunin to HA is followed by reversible swapping of HCs from site to site on the same and different HA molecules (12). We were interested in the effect that HA size would have on this swapping by TSG-6. In Fig. 6, blots show HC swapping from HMW HA (HA1000K) to HA14 (HA oligosaccharide containing 14 monosaccharides) via recombinant TSG-6. In panel A, the HCs of IαI from human serum (lane 1) were transferred to HA14 (lane 2) or to HA1000K (lane 3) for 24 h at 37 °C. Note that most of the HCs transferred to HA1000K and thus did not enter the gel because the HC-HA complex was too large, thus giving the appearance that they disappeared. Also, note that small portions of HCs were chucked as free HCs, either from IαI (as shown in Fig. 2C) or from HC-HA1000K (as will be shown in Fig. 8). In lane 4, HCs were first transferred to HA1000K for 24 h, after which the reaction mixture was spiked with HA14 to allow swapping of HCs from HA1000K to HA14 for 72 h at 37 °C. In lane 5, the reverse sequence was applied (i.e. transfer of HCs to HA14 for the first 24 h followed by incubation for 72 h with HA1000K). Although the HCs were able to swap from HA1000K to HA14 (lane 4), the reverse of this did not occur (lane 5). Thus, unlike for HMW HA, transfer of HCs to LMW HA14 oligosaccharide is an irreversible event. This effect is further illustrated in lane 6 in which both HA1000K and HA14 were applied simultaneously for 72 h. Thus, although HA1000K is both an HC acceptor and an HC donor, HA14 is exclusively an irreversible HC acceptor. Similar results were obtained with mouse serum (Fig. 6B).

FIGURE 6.

Irreversible transfer of human and mouse serum-derived HCs to a hyaluronan oligosaccharide. Panels A (human) and B (mouse) show Western blots of HC swapping from HMW HA1000K (high molecular weight HA with a molecular mass of ∼1000 kDa) to LMW HA14 oligosaccharide (14 monosaccharides long) via recombinant TSG-6. These blots were simultaneously probed with antibodies against HC1 and HC2. In panel A, the HCs of IαI from human serum (lane 1) completely transfer to HA14 (lane 2) and to HA1000K (lane 3) in 24 h. Note that HCs transferred to HA1000K do not enter the gel because the HC-HA complex is too large, thus giving the appearance that this species disappeared. On the other hand, the gel does not resolve HC from HC-HA14 complexes due to the small molecular weight differential. In lane 4, HCs were first transferred to HA1000K for 24 h, after which the reaction mixture was spiked with HA14 to allow swapping of HCs from HA1000K to HA14 for 72 h. In lane 5, the reverse sequence was applied (i.e. transfer of HCs to HA14 for 24 h followed by incubation for 72 h with HA1000K). In lane 6, both HA1000K and HA14 were applied simultaneously for 72 h. Similar results were obtained with mouse serum (panel B). The description for panel B is the same as for panel A with the exception that mouse serum was used as the HC-bikunin source. In summary, HA14 can only serve as an HC acceptor, not as a donor, whereas HMW HA can serve both roles.

FIGURE 8.

Release of free HCs from HMW HC-HA. The Western blot depicts HC chucking from HMW HC-HA. HCs were transferred to HA1000K by the addition of serum, and recombinant TSG-6 and EDTA were added 4, 24, 48, 72, and 96 h later (lanes 2–6, respectively) to inhibit TSG-6 activity. In lane 7, after the formation of HC-HA1000K for 24 h, an HA oligosaccharide 21 monosaccharides in length was added to demonstrate HC swapping from HC-HA1000K to HA21. Lane 8 shows HC transfer to HA21 (alone) in 4 h. Lane 9 shows HC transfer to HA21 (alone) for 96 h. All reactants were incubated at 37 °C for the entire 96-h incubation.

The kinetics of TSG-6-mediated swapping of human HCs between HMW HA (HA1000K) and LMW HA (HA8) are shown in Fig. 7, A and B. In panel A, all HCs were transferred to HA1000K from human serum IαI via TSG-6 for 24 h, after which HA8 was added and the reaction was stopped at different time points by the addition of EDTA. All samples were kept at 37 °C for the entire 96-h period to control for the stability of the reactants. Panel B is similar to panel A except that all of the HCs were first transferred from human serum IαI to HA8 for 24 h, after which HA1000K was added and the reaction was stopped by the addition of EDTA at the times indicated. In panel A, lane 9 shows the complete transfer of HCs from HC-bikunin in serum to HA8 as a positive, end point control (i.e. no HA1000K was used in lane 9). Panel A shows that swapping of human HCs from HMW HA to LMW HA oligosaccharide is relatively slow, taking 72 h to near completion. This was in contrast to human HC transfer from bikunin to either HMW or LMW HA, which was completed in 2–4 h (Fig. 2). In panel B, lane 9 shows the complete transfer of HCs from HC-bikunin in serum to HA1000K as a positive, end point control (i.e. no HA8 used in lane 9). Panel B shows no swapping of HCs from the LMW HA to the HMW HA during the 72-h incubation as the LMW HA8 oligosaccharide, although able to accept HCs, is unable to serve as an HC donor. Similar results were seen for the TSG-6-mediated swapping of mouse HCs (Fig. 7, C and D). Thus, unlike HC transfer from bikunin to HA, there does not appear to be species, or HC isotype, differences in the kinetics of HC swapping.

FIGURE 7.

Kinetics of irreversible transfer of human and mouse serum-derived HCs to a hyaluronan oligosaccharide. In panel A, all HCs were transferred to HA1000K from human serum IαI via TSG-6 for 24 h, after which HA8 was added and the reaction was stopped at 0, 2, 4, 6, 8, 24, 28, and 72 h (lanes 1–8) by the addition of EDTA. In lane 9, HA8, TSG-6, and serum (no HA1000K) were incubated for 24 h as an end point control. Panel B is the same as panel A except that all of the HCs were transferred from human serum IαI to HA8 for 24 h, after which HA1000K was added and the reaction was stopped as before. In lane 9, HA1000K, TSG-6, and serum (no HA8) were incubated for 24 h as an end point control. The descriptions for panels C and D are the same as for panels A and B, respectively, with the exception that mouse serum was used as the HC-bikunin source.

Because TSG-6 released (or chucked) free HCs from the CS chain of HC-bikunin in the absence of HA as an HC acceptor (Figs. 2C and 4C), we investigated the possibility that TSG-6 might also release HCs from HMW HC-HA. As shown in Fig. 8, TSG-6 transferred all of the HCs from serum-derived HC-bikunin to HA1000K (compare lanes 1 and 2) by 4 h. At later time points (i.e. 24, 48, 72, and 96 h after the reaction had started), TSG-6 released free HCs from HA1000K, albeit to a much less, and slower, extent than the release of HCs from the CS chain of HC-bikunin. The release of free HCs was not likely the result of general hydrolysis of the HCs from HA over the 96-h reaction period because hydrolysis was not observed in lanes 1 or 2, although the samples were incubated for the entire 96-h period. Swapping of HCs from HC-HA1000K to HA21 induced a gel shift to slightly higher than the free HC, and much stronger in intensity, demonstrating that although TSG-6 can release free HCs from HMW HC-HA, the number of free HCs released were minor in comparison with the number of HCs transferred to the HA21. Furthermore, the near absence of the faster migrating free HC band in samples incubated with HA21 (lanes 7–9) suggests that TSG-6 cannot release free HCs from irreversible HC acceptors such as HA21 (as shown in Fig. 9).

FIGURE 9.

HC transfer to a series of purified hyaluronan oligosaccharides. The HA oligosaccharides used in this study were analyzed by FACE to determine their monodispersity (panel A). The order of HA oligosaccharides is 8, 10, 12, 14, 18, and 21 monosaccharides (lanes 1–6 respectively). The same amount of oligosaccharide by mass was loaded in each lane, but as there is one mole of AMAC per mole of oligosaccharide, the signal decreases with increasing oligomer size. In panels B and C, these HA oligosaccharides, as well as HA4 and HA6 (not shown on the FACE gel), were incubated with human serum (as the IαI source) and TSG-6 for 4 h, and then HA1000K was added to an aliquot of these samples (shown as plus signs in lanes 3, 5, 7, and 9) for an additional 24 h. Thus, every sample was incubated for a total of 28 h, after which the reaction was stopped with EDTA. These samples were analyzed by Western blot, probing the blots simultaneously with antibodies against HC1 and HC2. Lane 1 shows human serum alone for this incubation period. HA4, HA6, HA8, and HA10 are shown in lanes 2–9 of panel A, and HA12, HA14, HA18, and HA21 are shown in lanes 2–9 of panel B. AMAC, 2-aminoacridone.

Range of Hyaluronan Oligosaccharides that Serve as Irreversible HC Acceptors

We were interested in defining the range of HA oligosaccharide sizes that serve as irreversible HC acceptors. HA oligosaccharides that were 4 (HA4), 6 (HA6), 8 (HA8), 10 (HA10), 12 (HA12), 14 (HA14), 18 (HA18), and 21 (HA21) monosaccharide units in length were tested. The purity of these oligosaccharides was confirmed by FACE analysis (Fig. 9A). In Fig. 9, B and C, recombinant TSG-6 was used to covalently transfer HCs from human serum-derived IαI (lane 1) to HA oligosaccharides of different sizes for 4 h at 37 °C to form HC-HA complexes with these oligosaccharides (−, lanes 2, 4, 6, and 8). After separation by SDS-PAGE and Western blotting, blots were probed with a mixture of antibodies against human HC1 and HC2. After this initial 4 h, samples were incubated an additional 24 h at 37 °C with (+, lanes 3, 5, 7, and 9) or without (−, lanes 2, 4, 6, and 8) HA1000K. The results show that once HCs were transferred to HA oligosaccharides containing 8, 10, 12, 14, 18, and 21 monosaccharides, TSG-6 could not remove the HCs from these oligosaccharides for transfer to the HA1000k (note that HC-HA1000K is too large to enter the gel or to be transferred to the blot). In contrast, HA1000K accepted HC transfer in the presence of HA4 and HA6 (Fig. 9B, lanes 3 and 5), indicating that they are too small to be HC acceptors. Thus, HA oligosaccharides with sizes 8–21 monosaccharides in length serve as irreversible HC acceptors via TSG-6 from either HC-bikunin or HMW HC-HA, but are unable to serve as HC donors.

In Figs. 2 and 4, we showed that in the absence of HA, TSG-6 slowly released or chucked HCs from bikunin. This is seen in Fig. 9B for the HA4 (lane 2) and HA6 (lane 4) oligosaccharides, which are too small to accept TSG-6-mediated HC transfer from bikunin (5). The small amounts of free HC seen in lanes 3 and 5 upon the addition of HA1000k are a result of HCs being chucked by TSG-6 during the first 4 h of incubation in the absence of HA1000K followed by transfer of the remaining HCs from bikunin to HA1000K during the remaining 24 h of incubation. The inability of HA6 to function as an HC acceptor was also confirmed in a separate study in which HA6 and HA18 were incubated with TSG-6 and HC-bikunin at 4-fold different ratios from each other, after which HCs were only found to transfer to HA18 (Fig. 10).

FIGURE 10.

Hyaluronan hexamers (six monosaccharide units) cannot function as an HC acceptor. A Western blot simultaneously probed with antibodies against HC1 and HC2 is shown. Lane 2 shows human serum alone, incubated for 4 h at 37 °C (all the other treatments were for the same time and temperature). Lane 3 shows human serum incubated with TSG-6 alone for the same time and temperature. In lane 4, HA6 was added to a reaction mixture of human serum and TSG-6. In lane 5, HA18 was added to a reaction mixture of human serum and TSG-6. In lanes 6–8, various ratios of HA6 and HA18 were co-incubated as noted. HA6 does not accept HCs.

Furthermore, we tested the hypothesis that HC swapping from HA1000K to HA8 might show differential kinetics than swapping from HA1000K to HA21, as might be the case if cooperativity might be observed in transfer to the smaller HA oligosaccharide (HA8). We found that the kinetics of HC swapping from HA1000K to HA8 and HA21 were identical (Fig. 11); thus, no cooperativity was evident.

FIGURE 11.

Comparison of the kinetics of HC swapping from HC-HA1000K to HA8 and HA21. HCs from serum-derived IαI were first transferred to HA1000K, in the presence of recombinant TSG-6, for 24 h, and portions of this reaction were spiked with an aliquot of HA8 or HA21. These reactions were stopped by the addition of EDTA 2, 4, 6, 8, 24, 48, and 72 h after the addition of the HA oligosaccharides (lanes 2–8 respectively). Controls included samples in which the HCs were transferred only to HA1000K (C1 for both panels A and B) or transferred only to HA8 (C2; lane 9, panel A) or only to HA21 (C2; lane 9, panel B).

HC-HA in Synovial Fluid of Mice with Experimental Rheumatoid Arthritis

The formation of HC-HA complexes has been previously reported in the synovial fluid of patients with RA (2). However, at present, nothing is known about the formation of HC-HA complexes in two widely used experimental mouse models of rheumatoid arthritis: systemic PGIA and monoarticular AIA. To determine whether these mouse models of RA mimic the human disease, we collected SF from the joints of mice in which PGIA and AIA had been induced as described previously (14–16). In Fig. 12, knee and ankle SF from mice subjected to PGIA (panel A) or knee SF from mice subjected to AIA (panel B) were probed with a mixture of antibodies against mouse HC1 and HC2. Aliquots of the SF from these mice were treated either with (+) or without (−) Streptomyces hyaluronidase for 24 h at 37 °C to release the HCs from the HC-HA complex, which is too large to enter the gel. Panel A shows an aliquot of SF pooled from several knees (lanes 1 and 2) and an aliquot of SF pooled from two sets of ankle joints (lanes 3–6) of mice subjected to PGIA. Panel B shows SF from three different sets of pooled knee SF of mice subjected to AIA. Without hyaluronidase treatment (lanes 1, 3, and 5), only a pre-IαI band is seen in the mice with PGIA, whereas pre-IαI and IαI bands are seen in the mice with AIA as the HCs attached to the HMW SF HA (HMW HC-HA) are too large to enter the gel. It is likely that at least a portion of the pre-IαI found in the PGIA fractions and of the pre-IαI and IαI found in the AIA fractions was from small amounts of serum that may have leaked into the SF during the isolation procedure (which we observed happened to a greater extent with the AIA isolate). In lanes 2, 4, and 6, hyaluronidase-released HCs are readily detected in SF from PGIA and AIA mice, similar to results previously reported in human RA SF (2).

FIGURE 12.

The pathological HC-hyaluronan complex is present in the synovial fluid of mice subjected to two different models of arthritis. These Western blots analyze the knee/ankle synovial fluid from mice subjected to systemic PGIA (panel A) or monoarticular AIA (panel B). These blots were simultaneously probed with antibodies against HC1 and HC2. Aliquots of the synovial fluid from these mice were either untreated (minus) or treated (plus) with Streptomyces hyaluronidase to release the HCs from the HC-HA complex. Panel A shows synovial fluid from pooled knee samples (lanes 1 and 2) and two different pooled fractions of synovial fluid derived from ankle joints (lanes 3–6) of mice subjected to PGIA. Panel B shows synovial fluid from three different fractions of knee synovial fluid of mice subjected to AIA.

Detectable levels of HC2 and HC3 were observed in both mouse models of RA (Fig. 13, E and F), but the antibody did not detect HC1 (panel D). Furthermore, HC1 was detectable in the SF of two of three human patients with RA (panel A), whereas HC2 was only detected in one of the three patients (panel B), and the antibody against HC3 did not detect any HCs in any of the patients (panel C).

FIGURE 13.

Heavy chain isotypes in arthritic human and mouse synovial fluid. An aliquot (2 μl) of synovial fluid from the knee joints of three different human rheumatoid arthritis patients (SF1–3) (panels A–C) and the synovial fluid from the knee and/or ankle joints of mice subjected to the PGIA and AIA models (panels D–F) was treated with Streptomyces hyaluronidase to release free HCs from HC-HA and analyzed by Western blot, probing three individual blots with antibodies against HC1 (panels A and D), HC2 (panels B and E), or HC3 (panels C and F). An aliquot (1.25 μl) of human serum (lane 2, panels A–C) or mouse serum (S in lane 4, panels D and F) was used as reference control (but accidentally left out in the fourth lane of panel E). The blots in panels A–C were scanned at the same time under the same settings and are shown as raw images. Similarly, the blots in panels A–C were scanned at the same time under the same settings and are shown as raw images.

Irreversible HC Transfer from HC-HA in Pathologic Synovial Fluid

We examined whether LMW HA oligosaccharides, which serve as irreversible HC acceptors, but not HC donors, could remove HCs from HC-HA complexes in human and mouse RA SF. In Fig. 14, human (panel A) and mouse (panel B) SFs were treated with (lane 2) and without (lane 1) hyaluronidase as in Fig. 12, but for 72 h at 37 °C. In lane 4, in the presence of exogenous recombinant TSG-6, HCs from HMW HC-HA were irreversibly transferred to HA14 oligosaccharides. In lane 3, in the absence of exogenous TSG-6, HC transfer to HA14 occurred as a result of endogenous SF TSG-6 activity.

FIGURE 14.

Irreversible transfer of HCs from synovial fluid hyaluronan to a hyaluronan oligosaccharide. Panels A and B, Western blots showing evidence for irreversible HC transfer from HMW HC-HA to HA14 oligosaccharides in synovial fluid derived from the knee of a human patient with rheumatoid arthritis (panel A) or the knees of mice subjected to the systemic PGIA model of rheumatoid arthritis (panel B). These blots were simultaneously probed with antibodies against HC1 and HC2. In lanes 1–2, synovial fluid was treated without or with Streptomyces hyaluronidase (Hyase) for 72 h. In lane 3, HA14 was added to test for endogenous TSG-6 activity by its ability to transfer HCs from HC-HA in the synovial fluid to HA14. In lane 4, exogenous recombinant TSG-6 was added to further promote HC swapping from HMW HC-HA to HA14. Similar results were obtained with synovial fluid derived from the ankles of mice subjected to the PGIA model (not shown).

DISCUSSION

For simplicity of discussion, we define HC transfer as the TSG-6 catalyzed covalent transfer of HCs from the bikunin CS chain to HA. This is in contrast to HC shuffling and/or swapping of HCs between different HA molecules or the TSG-6-mediated release of free HCs, which we refer to as chucking.

Kinetics of HC Transfer Are HA Size-independent

In this study, we investigated the hypothesis that the kinetics of HC transfer from IαI to HA might be different when TSG-6 transferred HCs to large HA molecules (HA1000K) when compared with HC transfer to small HA molecules (HA8, which is 8 monosaccharides long). Our results show that the kinetics of HC transfer by TSG-6 from HC-bikunin to HA are independent of HA size (compare the blot in Fig. 2A with the blot in Fig. 2B, and compare the blot in Fig. 4A with the blot in Fig. 4B). Transfer of HCs from both human and mouse IαI to HA1000K and HA8 was complete within 2–4 h with almost no detectable formation of a pre-IαI intermediate. In contrast, HC transfer from mouse pre-IαI was relatively slow, requiring a minimum of 24 h.

We also investigated the hypothesis that the kinetics of HC transfer from IαI to HA might be different for the different HC isotypes. The data showed that when two HCs are present on the CS chain of bikunin, the transfer of the first HC to HA was independent of species or HC isotype. However, when a single HC is on bikunin (i.e. in the form of pre-IαI), this HC is quickly transferred if it is HC1, but transferred slowly if it is HC2, independent of species.

We observed that in the absence of HA, TSG-6 shuffled the HCs on the CS chain of bikunin, forming an equilibrium mixture of bikunin molecules with 1–3 HCs per bikunin that was noticeable in the first hour of incubation. After 4 h of HC shuffling, we observed the TSG-6-mediated chucking of free HCs, which was completed within 24 h. This phenomenon cannot be explained as the simple hydrolysis of an unstable HC bond with the CS chain of bikunin because IαI showed no signs of cleavage in the absence of TSG-6 or when TSG-6 was applied in the presence of EDTA. Once chucked, these free HCs could not be transferred by TSG-6 to HA of any size (Fig. 3). Because it is hard to imagine an in vivo scenario in which TSG-6 and IαI would interact in the absence of HA, we largely view HC shuffling on IαI and the chucking of free HCs as an in vitro artifact. This conclusion is also likely because HC transfer is faster than HC chucking.

Kinetics of HC Shuffling Are HA Size-dependent

Reversible HC shuffling between different HA molecules has been previously demonstrated (12). This shuffling presumably involves a process whereby TSG-6 binds to HC-HA through its link module near the HC attachment site on HA followed by the formation of a HC-TSG-6 intermediate (7) and the subsequent transfer of that HC to another site on the same, or different, HA molecule. Throughout this study, we have shown that the shuffling of HCs among HA molecules was unidirectional when the HCs were transferred to HA oligosaccharides. Thus, although the initial kinetics of HC transfer were HA size-independent, HC shuffling was dependent on HA size. Our interpretation is that HA oligosaccharides are too small to allow rebinding of TSG-6 and/or removal of HC from HA oligosaccharides due to the presence of the bulky HC adduct. Although we have shown that HC transfer to HA21 is an irreversible event, the upper HA size limit of irreversibility remains to be established. We suspect that the upper size limit may not be precise (i.e. ±1–2% of the molecular weight around a defined average chain length), but rather a more disperse range of HA sizes (i.e. > ±1–2%). Unlike HC transfer from bikunin to HA, we did not observe species, or HC isotype, differences in the kinetics of HC shuffling.

Shuffling of HCs from Pathological HC-HA to HA Oligosaccharides

The formation of pathological HC-HA complexes has been previously reported in the SF of patients with RA (1). Such complexes enhance the CD44-mediated adhesion of leukocytes to HA matrices (1). We reported in this study that the HC-HA complex is also present in the SF of two experimental mouse models of RA: PGIA and antigen-induced arthritis AIA (14–16). These two experimental models of RA fundamentally differ in that PGIA is a systemic model of RA similar to that found in humans, whereas AIA is a monoarticular model of arthritis in which the disease is confined to the joint that is injected with antigen after immunization of the animal with the same antigen. Interestingly, our results show that similar to human RA patients, the animals with both PGIA and AIA forms of the disease accumulate significant amounts of HC-HA complex in the SF of their affected synovial joints (Fig. 12).

From the SF of both human patients and mouse models of RA, we demonstrated that HA oligosaccharides can be used to “cleanse” pathological HC-HA complexes of HCs during an HC shuffling process mediated by endogenous and exogenous TSG-6 activity. Thus, by removing HCs from HC-HA, in RA, we have disrupted the HC-HA complex, which is known to significantly promote leukocyte adhesion to HA matrices in SF. Although there is clearly therapeutic potential for HA oligosaccharides in this pathology, important obstacles remain, such as the potential proinflammatory effects of small HA molecules (17), their rapid clearance in the bloodstream (18), and the relatively slow kinetics of HC swapping in our in vitro tests. Thus, it is probably more feasible to consider the use of irreversible HC transfer to HA oligosaccharides in cooperation with existing anti-inflammatory strategies.