Specificity of the Tumor Necrosis Factor-induced Protein 6-mediated Heavy Chain Transfer from Inter-α-trypsin Inhibitor to Hyaluronan

The formation of the hyaluronan-rich cumulus extracellular matrix is crucial for female fertility and accompanied by a transesterification reaction in which the heavy chains (HCs) of inter-α-trypsin inhibitor (IαI)-related proteins are covalently transferred to hyaluronan. Tumor necrosis factor-induced protein-6 (TNFIP6) is essential for this transfer reaction. Female mice deficient in TNFIP6 are infertile due to the lack of a correctly formed cumulus matrix. In this report, we characterize the specificity of TNFIP6-mediated HC transfer from IαI to hyaluronan. Hyaluronan oligosaccharides with eight or more monosaccharide units are potent acceptors in the HC transfer, with longer oligosaccharides being somewhat more efficient. Epimerization of the N-acetyl-glucosamine residues to N-acetyl-galactosamines (i.e. in chondroitin) still allows the HC transfer although at a significantly lower efficiency. Sulfation of the N-acetyl-galactosamines in dermatan-4-sulfate or chondroitin-6-sulfate prevents the HC transfer. Hyaluronan oligosaccharides disperse cumulus cells from expanding cumulus cell-oocyte complexes with the same size specificity as their HC acceptor specificity. This process is accompanied by the loss of hyaluronan-linked HCs from the cumulus matrix and the appearance of oligosaccharide-linked HCs in the culture medium. Chondroitin interferes with the expansion of cumulus cell-oocyte complexes only when added with exogenous TNFIP6 before endogenous hyaluronan synthesis starts, supporting that chondroitin is a weaker HC acceptor than hyaluronan. Our data indicate that TNFIP6-mediated HC transfer to hyaluronan is a prerequisite for the correct cumulus matrix assembly and hyaluronan oligosaccharides and chondroitin interfere with this assembly by capturing the HCs of the IαI-related proteins.

undergoes immense changes, which results in the formation of an expanded extracellular matrix (1). This matrix is involved in the expulsion of the COC from the follicle (2), enhances the pickup and transport of the COC by the oviduct (3,4), and influences sperm penetration and oocyte fertilization (5)(6)(7). The cellular events of matrix formation, which include the synthesis and organization of hyaluronan (HA) around the cumulus cells, have been studied extensively in mouse COCs both in vivo and in vitro (8). After the LH surge, COCs quickly upregulate the expression of hyaluronan synthase 2, the enzyme required to produce hyaluronan (9). Although the mediators that induce hyaluronan synthesis in the COCs in vivo have yet to be determined, prostaglandin E 2 , follicular stimulating hormone and epidermal growth factor are potent initiators of hyaluronan production in vitro (10 -13). Hyaluronan synthesis is also under the control of the oocyte (14,15), and growth differentiation factor-9 and bone morphogenic protein 15 are the most likely oocyte-derived growth factors involved in this regulation (16,17).
Several studies indicate that the organization of hyaluronan into the cumulus extracellular matrix requires the participation of matrix proteins (18 -21). Particularly, inter-␣-trypsin inhibitor (I␣I)-related proteins have been shown to be crucial for cumulus matrix formation both in vitro and in vivo (19,22,23). I␣I and pre-␣-trypsin inhibitor (P␣I) are the two major serum-derived I␣I family proteins synthesized by hepatocytes. I␣I and P␣I are covalent complexes of bikunin (the light chain with trypsin inhibitor activity), and either one or two of three heavy chains (HCs) linked through a chondroitin sulfate chain (24). In mice, these serum proteins diffuse into the preovulatory follicle from the capillaries after the LH surge (25), and their HCs incorporate into the cumulus matrix through covalent linkages to HA (26). The links between HCs and HA are most likely ester bonds formed between the N-acetyl-glucosamine residues of HA and the C-terminal aspartic acid residues of the HCs, as previously described for the HA-HC complexes in the synovial fluid (27). In vitro, COC expansion does not occur in the absence of the I␣I family proteins (i.e. serum) (22,28). Cumulus cells synthesize hyaluronan at a normal rate under these conditions, but release it into the culture medium, rather than form an extracellular matrix (28). The importance of I␣I family proteins in vivo has been established by studies with bikunin-deficient mice (19,23). Bikunin-null mice are unable to synthesize I␣I family members; their females do not correctly form the cumulus extracellular matrix and are infertile.
Tumor necrosis factor-induced protein 6 (TNFIP6, also known as TSG6) is another matrix protein that is essential for cumulus matrix formation. TNFIP6 is synthesized by both cumulus and granulosa cells during the process of COC expansion (20, 29 -31). The protein exists in two forms, a monomer and a covalent complex with either of the HCs of I␣I, in the extracellular matrix of expanded COCs (30). Recently we and others have demonstrated that the presence of TNFIP6 is crucial for the transfer of HCs from I␣I to HA (21,32). This transesterification reaction appears to be necessary for the formation of the COC extracellular matrix. Female mice lacking Tnfip6 are unable to perform this mandatory reaction and do not assemble the cumulus extracellular matrix leading to infertility (21).
Previous studies have implemented the approach to inhibit cumulus matrix formation by administrating HA oligosaccharides as competitors for endogenous polymeric HA (2,28). However, the mechanism of this inhibition remains largely unexplored. In this study, we report that HA oligosaccharides of defined sizes are potent acceptors in the TNFIP6-mediated HC transfer reaction, with an octamer being the smallest effective size. Our data also indicate that the chemical composition of the acceptor glycosaminoglycan chain greatly influences the TNFIP6-mediated HC transfer reaction. HA oligosaccharides inhibit COC expansion with the same size specificity as the HC transfer, strongly suggesting a direct functional link between the transesterification reaction and the assembly of the cumulus matrix.
Recombinant TNFIP6 -Human recombinant TNFIP6 (94% identical to the mouse protein) was expressed in Schneider 2 cells and purified with the combination of ion exchange and reverse-phase high performance liquid chromatography (HPLC) as described previously (33).
Preparation of HA Oligosaccharides-HA oligosaccharides were isolated and purified from testicular hyaluronidase digests of rooster comb HA as described before (34). Briefly, HA was partially degraded by digestion with bovine testicular hyaluronidase at 37°C for various time periods to achieve the desired sizes of the HA oligosaccharides. The reaction was stopped by boiling the mixture for 20 min. Then, the samples were centrifuged at 10,000 rpm, the supernatant were lyophilized and redissolved in distilled water. HA oligosaccharides were isolated from the digests by anion exchange chromatography on a Dowex 1 ϫ 2 column. The molecular size of each oligomer was determined by electrospray ionization mass spectrometry. The purity and size uniformity of the samples were checked by the HPLC and fluorophore-assisted carbohydrate electrophoresis. The impurities (endotox-ins, DNA, and other proteins) were also routinely checked as described in an earlier publication (34).
TNFIP6-mediated HC Transfer onto Glycosaminoglycans and HA Oligosaccharides-HA, its oligosaccharides, chondroitin, dermatan-4sulfate, chondroitin-6-sulfate, and chondroitin 22-mer (5 g each) were individually incubated with 5 l of mouse serum either in the presence or absence of 250 ng of recombinant TNFIP6 in 50 l of PBS at 37°C for 24 h. An aliquot (10 l) of each reaction was digested with either 100 milliunits of Streptomyces or Streptococcal hyaluronidase (as described under "Results") at 37°C for 1 h. The samples were boiled in reducing Laemmli's loading buffer for 5 min and loaded on 4 -20% SDS-PAGE precast gels. After electrophoresis, the proteins were blotted onto nitrocellulose membranes and probed with anti-I␣I (1:1000). Horseradish peroxidase-labeled anti-rabbit immunoglobulin (1:10000) was used as a secondary antibody. Bands were detected using an enhanced chemiluminescence detection kit.
Enzyme-linked Immunosorbent Assay for the TNFIP6-mediated HC Transfer-Maxisorp plates were coated with 50 g (or as the experimental procedures required) of high molecular weight hyaluronan in 50 l of PBS at room temperature overnight. The wells were blocked with 150 l of 2.5% skim milk in PBS at room temperature for 2 h. Mouse serum, human recombinant TNFIP6, and glycosaminoglycans were added in a total volume of 50 l in PBS as the experimental protocols required. The optimal amount of mouse serum that produced low nonspecific binding was found to be 0.1 l and was used in all the reactions. HC transfer reactions were carried out at 37°C for 24 h. In the time course experiments, the reaction was stopped at the given time points by the addition of 50 l of 100 mM EDTA, since previous studies suggested the requirement of divalent cations (32,35). After completion of the reaction, the plates were washed three times with PBS containing 0.05% Tween 20 for 10 min, and then incubated with 100 l of anti-I␣I antibody (1:1000) in PBS/0.05% Tween 20 at 37°C for 1 h. The plates were washed three times as above, and then 100 l of HRPO-labeled goat anti-rabbit immunoglobulin (1:5000) was added in PBS/0.05% Tween 20. The plates were incubated at 37°C for 1 h and then washed three times with PBS/0.05% Tween 20 and three times with detergentfree PBS. The color reaction was developed by adding 100 l of 1 mg/ml orthophenylenediamine and 0.1 l/ml hydrogen peroxide in 0.1 M citrate buffer, pH 5.0. Color development was stopped by the addition of 50 l of 4 M sulfuric acid, and the absorbance values were read at 490 nm using a Spectramax 250 ELISA plate reader (Molecular Devices Corp, Sunnyvale, CA).
Isolation and Culture of COCs-Sexually immature (21 days old) Swiss CD1 female mice were primed by intraperitoneal injection of 5 international units of PMSG, in 0.1 ml PBS. After 46 -48 h, the animals were sacrificed, and their ovaries were dissected out in MEM containing 25 mM HEPES, 0.1% bovine serum albumin and 50 ng/ml gentamycin. Compact COCs were collected by puncturing large follicles, and washed in culture medium containing MEM, 3 mM glutamine, 0.3 mM sodium pyruvate, and 50 ng/ml gentamycin. COCs were cultured in 300 l of culture medium containing 1 mM dbcAMP and mouse serum (generally 5% if not indicated otherwise) at 37°C, for 16 h. HA oligosaccharides, chondroitin, and chondroitin 22-mer were added at concentrations indicated in the Result section. Certain cultures were also supplemented with 1 g/ml recombinant TNFIP6. COC expansion was assessed with an Olympus inverted microscope equipped with a CCD camera.
Characterization of Medium and Matrix Fractions of Cultured COCs-COCs with their culture medium were transferred to a microcentrifuge tube and centrifuged at 300 ϫ g, 4°C for 3 min. The culture medium was carefully removed, and the COCs were washed three times with PBS and centrifuged as above. After resuspension in PBS, the complexes were digested with Streptomyces hyaluronidase (10 milliunits/100 COC) at 37°C for 1 h to release matrix molecules bound to hyaluronan. After digestion, the cells were pelleted at 300 ϫ g, 4°C for 5 min, and the supernatants were aspirated and called as matrix fraction. Aliquots of medium were similarly digested with Streptomyces hyaluronidase at 37°C for 1 h. The samples were analyzed by Western blot using an anti-I␣I antibody as described above.

TNFIP6-mediated HCs Transfer from I␣I and P␣I to HA
Oligosaccharides-We and others (21,32) have previously shown that TNFIP6 is essential for the transfer of HCs from I␣I and P␣I to high molecular weight HA. In order to investigate the size specificity of the TNFIP6-mediated HC transfer to HA, small oligosaccharides of defined sizes were incubated with mouse serum (as the source of I␣I and P␣I) in the presence of recombinant TNFIP6 at 37°C for 24 h. Western blot analysis with anti-I␣I antiserum indicated that in this cell free biochemical reaction, it was possible to transfer HCs from I␣I and P␣I not only to high molecular weight HA but also to HA oligosaccharides ( Fig. 1). The transfer of the HCs were apparent from the disappearance or the reduction of the I␣I and P␣I bands and from the appearance of new bands either at the level of HCs (ϳ85 kDa, in the case of oligosaccharides) or at the wells (in the case of high molecular weight HA). As we described previously, incubation of high molecular weight HA and serum without TNFIP6 did not transfer HCs to HA (Fig. 1A, lane 2). The addition of recombinant TNFIP6 caused the whole I␣I band and most of the P␣I band to disappear and a high molecular weight HA-HC complex to appear (Fig. 1A, lane 1). HA oligosaccharides with 8 or more monosaccharide units were potent acceptors in the HC transfer reaction (Fig. 1B, lanes [5][6][7][8]. In these reactions, the size of the I␣I-reactive bands at the HC level increased with the size of the HA oligosaccharide, indicating that the HCs were actually transferred onto the oligosaccharides. In the case of the HA4 and HA6 oligosaccharides only minor bands were detected at the HC level (Fig. 1B,  lanes 3 and 4), and the intensity of these bands did not significantly differ from the HC band observed in the absence of the oligosaccharides (Fig. 1B, lane 2). However, the intensity of the I␣I band in the HA6 reaction was somewhat decreased, suggesting that either minor amounts of HC were transferred to this oligosaccharides, or spontaneous HC release from I␣I occurred. It is also important to note that the I␣I band disappeared completely in the case of all oligosaccharides except for HA6 and HA4 but the P␣I band was still present at a low intensity in all the cases. This observation suggested that the HCs transfer from the I␣I was probably a more favorable reaction than that from the P␣I.
In order to quantitatively determine the differences in the HC acceptor abilities of the HA oligosaccharides, we established an ELISA system in which the HC transfer reaction was completed on surface immobilized HA (Fig. 2). As expected the reaction was dependent on the concentration of the HA coating solution ( Fig. 2A), and the amounts of TNFIP6 (Fig. 2B) and serum (not shown) added. The transfer reaction showed a saturation curve when 50 l of 1 mg/ml HA coating solution, 25 ng of TNFIP6 and 0.1 l of serum were used, and could be accelerated by the addition of magnesium ions (Fig. 2C). Under these conditions, we used different amounts of high molecular weight HA and HA oligosaccharides to compete with the surface immobilized hyaluronan for the HC transfer reaction (Fig.  2D). The competitor amounts that caused 50% inhibition were compared assess the HC acceptor ability of these competitors. As expected from our Western-blot results HA8 or larger oligosaccharides in the range of 1-400 ng effectively inhibited the HC transfer to the immobilized HA (Fig. 2D). In the same range, neither HA6 (Fig. 2D) nor HA4 (not shown) were able to compete. However, a drastically larger amount of HA6 (10 g) could cause partial inhibition (see below). Generally, the larger oligosaccharide resulted in better inhibition, with HA8, HA10, HA12, and HA14 being ϳ6.4, 4.9, 3, and 2 times weaker competitors, respectively, than high molecular weight HA (inhibition curves for HA10 and HA12 are not shown).
Glycosaminoglycan Specificity of the TNFIP6-mediated HC Transfer-With the purpose of exploring the structural specificity of the HC transfer, we used different glycosaminoglycans, either sulfated or unsulfated in our biochemical transfer studies and checked their ability to accept HCs from the I␣I and P␣I in the TNFIP6-mediated HC transfer reaction. Sulfated glycosaminoglycans, such as dermatan-4-sulfate and chondroitin-6sulfate could not serve as acceptors. Neither I␣I nor P␣I could be eliminated with these glycosaminoglycans, and no new bands were apparent in Western blot analysis (Fig. 3A, lanes 5 and 7) compared with the control reaction without glycosaminoglycans (Fig. 3A, lane 9). Chondroitin, an unsulfated glycosaminoglycan, was acceptor for the TNFIP6-mediated HC transfer as indicated by the disappearance of the I␣I band and the reduction of the P␣I band in Western blot analysis (Fig. 3, A, lane 3, and B, lane 4). In addition, new bands appeared on the gel in the range between the HC and P␣I. The size and polydispersity of these new I␣I-positive bands were consistent with chondroitin chains bearing one HC. The digestion of the chondroitin reaction product with Streptococcal hyaluronidase (an enzyme that is able to degrade chondroitin but not chondroitin sulfates) resulted in a strong band at the HC level and the disappearance of the multiple bands in the range between the HC and P␣I (Fig. 3, A, lane 4, and B, lane 3). The HC transfer reaction to chondroitin took place only in the presence of TNFIP6 (Fig. 3B, lanes 2 and 4), indicating the essential role of this protein. Furthermore, a monodisperse chondroitin oligomer with 22 monosaccharide units was also able to serve as an acceptor in the TNFIP6-mediated transfer reaction (Fig.  3C). These data indicated that beside the N-acetyl-glucosamine-containing HA, unsulfated N-actyl-galactosamine-containing chondroitin was also able to participate in the transesterification reaction as an acceptor.
HA Oligosaccharides Suppress Cumulus Cell-Oocyte Complex Expansion by Inhibiting HC Transfer to Endogenous HA-The inhibitory effect of HA oligosaccharides on COC expansion has been described previously (2,28). However, the mechanism of this inhibition has not yet been determined. In order to investigate HA oligosaccharide specificity of the suppression of COC expansion, we used highly purified HA oligosaccharides of defined sizes (34) to competitively inhibit COC expansion in vitro. Freshly isolated mouse COCs were cultured in the pres- ence of 5% mouse serum, 1 mM dbcAMP and 1 mg/ml HA oligosaccharides of different sizes (HA4-HA14) for 16 h. COCs treated with HA4 or HA6 oligosaccharides showed normal expansion (Fig. 4, C and D). Inhibition of expansion could be first observed with HA8 oligosaccharides and was more prominent with larger oligosaccharides. COCs treated with either HA8 or HA10 oligosaccharides showed similar morphology. Most of the cumulus cells were settled down to the dish, but some of them were still maintained in a loose three-dimensional structure (Fig. 4, E and F). HA12 and HA14 oligosaccharides caused the complete abolishment of the three-dimensional structure of COCs (Fig. 4, G and H). The degree of inhibition was dependent on the concentration of the serum as well as on the concentration of the oligosaccharides. When serum concentration was decreased to 1% (and thus the amounts of I␣I and P␣I were limited), HA8 was able to achieve the same degree of inhibition as HA12 or HA14 in 5% serum (Fig. 5). On the other hand, decreasing the concentration of HA12 caused less prominent suppression of COC expansion, with ϳ250 g/ml HA12 in 1% serum causing approximately the same morphology as 1 mg/ml HA8 oligosaccharide in 5% serum (Fig. 6). Therefore, depend-ing on its concentration and the amount of serum used, HA8 could be just as effective inhibitor of COC matrix expansion as HA12. Since the TNFIP6-mediated HC transfer and the inhibition of COC expansion demonstrated the same HA oligosaccharide specificity (i.e. HA8 or higher), we investigated the medium of the oligosaccharide-treated cultures (pictured on Fig. 4) for the presence of HCs. The media of COCs treated with HA4 or HA6 did not contain oligosaccharide-linked HCs (Fig. 7, lanes 2 and  3). However, these HCs were easily detectable in the culture media treated with HA8 or larger oligosaccharides (Fig. 7,  lanes 4 -7). In order to investigate the compartmentalization of HCs between the cumulus matrix and the culture medium in HA oligosaccharide-treated cultures, we performed large scale cultures in the absence or presence of either HA6, HA8, or HA12 oligosaccharides. The matrix fraction (COCs digested with Streptomyces hyaluronidase) contained hyaluronanlinked HCs only in the untreated and HA6-treated cultures (Fig. 8, lanes 7 and 8), but not in the HA8-and HA12-treated cultures. The absence of the hyaluronan-linked HCs in the HA8-and HA12-treated COCs (Fig. 8, lanes 9 and 10) coincided with the lack of the stable COC matrix in these cultures. In agreement with this observation, HA oligosaccharide-linked HCs were detected in the culture medium of the HA8 and HA12-treated COCs, but not in the untreated and the HA6-treated ones (Fig. 8, lanes 3-6). These data suggested that HA8 or larger oligosaccharides successfully competed for the HC transfer reaction with endogenous polymeric HA synthesized by the cumulus cells, and this competition is the most likely reason for interfering with correct COC expansion.
Chondroitin Interferes with COC Expansion Only in the Presence of Exogenous TNFIP6 -Since our earlier experiment (Fig.  3) demonstrated that HCs can be successfully transferred to chondroitin (and its oligomer) in the cell-free system, we attempted to inhibit COC expansion with this glycosaminoglycan. Interestingly, neither chondroitin nor its 22-mer was able to interfere with COC expansion at a concentration (1 mg/ml) where HA oligosaccharides were fully efficient (Fig. 9, B and  C). Similarly, dermatan-4-sulfate and chondroitin-6-sulfate were also unable to suppress COC expansion (data not shown). The inability of chondroitin to interfere with COC expansion raised the possibility that this glycosaminoglycan is a poor competitor of HA in the transesterification reaction. Since HA synthesis starts at ϳ3-4 h after the initiation of matrix expan-sion in culture (36), we attempted to inhibit COC expansion by adding exogenous recombinant TNFIP6 together with chondroitin or its oligomer at the start of the culture. Under these conditions chondroitin was expected to have a 3-4 h head start (in the absence of endogenous HA synthesis) to complete the HC transfer reaction and deplete I␣I and P␣I from the culture medium. Indeed, chondroitin and its oligomer successfully inhibited COC expansion when they were added with exogenous recombinant TNFIP6 (Fig. 9, E and F). On the other hand HA6 oligosaccharide, dermatan-4-sulfate and chondroitin-6-sulfate were ineffective in suppressing COC expansion even in the presence of recombinant TNFIP6 (data not shown). Westernblot analysis of the medium and matrix fractions of chondroitin-treated COCs supported our hypothesis regarding chondroitin as a weak competitor in the transesterification reaction. When chondroitin was added alone to the culture, HC incorporation into the matrix was inhibited in a certain extent, but still was sufficient to stabilize the matrix (Fig. 10, lane 1). However, when chondroitin was added with recombinant TN-FIP6 to the cultures, HC transfer to polymeric HA was completely inhibited (Fig. 10, lane 2), suggesting that chondroitin sufficiently reduced the amount of I␣I and P␣I by the time endogenous HA started to synthesize. The culture medium of chondroitin-treated COCs (with or without exogenous TNFIP6) contained the same chondroitin-linked polydisperse bands as was observed in the cell-free HC transfer reaction (Fig. 10,  lanes 3 and 4).
In order to further assess the effectiveness of chondroitin in the HC transfer reaction, we tested its competitive efficiency against immobilized HA in ELISA. Both chondroitin and its 22-mer oligosaccharide were ϳ20 times weaker competitors in this assay than high molecular weight HA (Fig. 11). Dermatan-4-sulfate did not inhibit the HC transfer reaction even at 10 g, while chondroitin-6-sulfate caused ϳ13% inhibition at that level (data not shown). Interestingly, when HA6 was tested as a control at high concentrations, it caused ϳ16 and ϳ30% inhibition at 2.5 and 10 g, respectively (Fig. 11). The extrapolation of this curve suggested that HA6 was ϳ3,000 times weaker competitor than high molecular weight HA. Since we never observed HC transfer to HA6 oligosaccharide in Western blot analysis (see Figs. 1, 7, and 8), the competition of HA6 in ELISA at high concentrations was not likely the result of the transfer of HA to this oligosaccharide. Rather, HA6 at these concentrations probably just decreased the interaction of TN-FIP6 with polymeric HA and thus slowed down the HC transfer to the immobilized hyaluronan. DISCUSSION One of the major differences between HA and other glycosaminoglycans is that HA is not synthesized on a protein core. However, studies in the last two decades have demonstrated that under special circumstances HA is capable of establishing covalent linkage with a certain group of proteins. These HAprotein complexes have been detected in serum-containing cultures of fibroblasts and smooth muscle cells (37,38), in synovial fluids and sera of arthritic patients (27,39,40) and in the extracellular matrix of cumulus cells during ovulation (26). The protein components of this complex derive from serum (and such they were first named as serum-derived hyaluronan-associated proteins or SHAPs, Ref. 35) and have been identified as the heavy chains (HCs) of I␣I-related serum proteins (27). The linkage between HA and the HCs is an ester bond that forms between C-6 hydroxyl group of a Nacetyl-glucosamine residue of HA and the peptide carboxyl group of an aspartate residue of the HCs (27). The HCs of the I␣I family members are transferred to HA by a transesterification reaction, although the mechanism of this reaction has not been completely understood. Recently, we and others (21,32) have demonstrated that TNFIP6 facilitates the transfer of HCs onto HA. In this report, we have used HA oligosaccharides and other glycosaminoglycans to further characterize the specificity of the TNFIP6-mediated HC transfer to HA, and to provide a direct link between this transesterification reaction and the assembly of the cumulus extracellular matrix.
Our findings further strengthen the view that TNFIP6 is essential for the transesterification reaction, i.e. in the absence of this protein HC transfer cannot be detected. This observation is somewhat contradictory to a previous study that achieved HC (SHAP) transfer by incubating HA with serum only (37). This contradiction, however, could be explained by the amounts and origin of the sera used. Particularly, potential circulating levels of TNFIP6 could be varying among different species or even among individuals of the same species. Our results also indicate that TNFIP6-mediated HC transfer is enhanced by magnesium ions. This finding is in agreement with previous observations that EDTA is able to inhibit the transfer reaction. However, in contrary to a previous report (32), we were unable to enhance the reaction with extra amounts of calcium ions (data not shown), indicating that either calcium ion is not necessary for the reaction or the level of this ion in the serum (even at 1/500 dilution) is still at saturation for the reaction. Our results also determine the structural requirements necessary for the successful completion of the TNFIP6-mediated HC transfer. HA8 or larger oligosaccharides are potent acceptors in the transesterifiation reaction, an observation consistent with the HA oligosaccharide specificity of the inhibition of the SHAP transfer to high molecular weight HA (37). Our data indicate that longer oligosaccharides are somewhat more effective in inhibiting the HC transfer to high molecular weight HA, but the difference between HA8 and high molecular weight HA is still within one order of magnitude. Since TNFIP6 is a HA-binding protein, it is reasonable to speculate that there could be some correlation between the HC transferring and HA binding specificities of this protein. Binding of hyaluronan to cells expressing a chimeric membrane-bound form of TNFIP6 can be inhibited with HA6 oligosaccharide just as efficiently as with HA8 (41). Recent isothermal calorimetric measurements, on the other hand, show that TNFIP6 binds to HA6 with an ϳ15 times lower affinity than to HA8 (42). However, it is unlikely that the difference observed in the HC acceptor abilities of HA6 and HA8 is due to the different TNFIP6 binding affinities of these oligosaccharides, since our HC transfer experiments (Fig. 1) contained HA6 oligosaccharide in a ϳ600fold molar excess over TNFIP6. Rather, it seems more likely that the difference in the HC-transferring abilities of these oligosaccharides is due to the way these oligomers fit into the HA binding groove of TNFIP6. In this regard, the conformation of the particular N-acetyl-glucosamine residue that acts as an acceptor for the HC transfer is likely to be critical for this reaction to occur. For example, recent nuclear magnetic resonance studies on the Link module of TNFIP6 in the presence of HA oligosaccharides clearly indicate that the reducing termini of bound HA6 and HA8 oligomers lie in different positions relative to the protein (42). Further structural studies should elaborate on HA conformation necessary for the HC transfer to take place.
Of the other glycosaminoglycans we used, only chondroitin (an unsulfated epimer of hyaluronan) was able to serve as a substrate in the HC transfer reaction. However, the efficiency of chondroitin to compete with hyaluronan for the HCs is significantly lower than that of the HA oligosaccharides. This observation indicates that the epimerization of C-4 hydroxyl group (i.e. the change to N-acetyl-galactosamine) dramatically reduces the HC accepting ability of the glycosaminoglycan. This reduced ability is likely the consequence of the weaker binding of chondroitin to TNFIP6, due to the structural change from N-acetyl-glucosamine to N-acetyl-galactosamine. Although the physiological role of the HC transfer to chondroitin is not understood, HCs covalently linked to large proteoglycans (presumably through chondroitin chains) have been detected in follicular fluid (43). Sulfated glycosaminoglycans, such as chondroitin-6-sulfate or dermatan-4-sulfate are not acceptors for the HCs. The bulky and negatively charged sulfate groups most likely prevent the fit of these glycosaminoglycans into the HA binding groove of TNFIP6, and therefore are not compatible with the HC transfer.
The HA oligosaccharide specificity of the TNFIP6-mediated HC transfer has important implications for the assembly of the cumulus matrix. HA oligosaccharides inhibit COC expansion with the same size specificity (HA8 or larger) as their HC transfer specificity. Moreover, this inhibition is accompanied by the appearance of oligosaccharide-linked HCs in the culture medium. The extent of inhibition of COC expansion appears to be more prominent with the increasing size of the oligosaccharides (at the same concentrations), resembling the same size effect on the HC transfer in the cell-free reaction. However, the inhibition of COC expansion is clearly dependent on the oligosaccharide and serum (i.e. I␣I and P␣I) concentrations, and even HA8 can achieve full inhibition under appropriately chosen conditions. The weaker HC acceptor ability of chondroitin is also demonstrated by its weaker ability to inhibit COC expansion. Chondroitin inhibits cumulus matrix formation only when it is allowed to deplete I␣I and P␣I before HA synthesis starts by the cumulus cells. All these data strongly suggest that HA oligosaccharides and chondroitin interfere with COC expansion by preventing the HC transfer to endogenously synthesized high molecular weight HA, and this transfer reaction is crucial for cumulus matrix assembly. Although HA oligosaccharides and chondroitin may also interfere with the binding of HA to surface receptors on the cumulus cells, the contribution of these interactions to the stability of the cumulus matrix may not be significant. Female mice deficient in CD44, the major cell surface HA-binding protein (44 -46), do not appear to have the same reproductive problems than those deficient in either TNFIP6 or bikunin (i.e. I␣I and P␣I) (19,21,23). Although the participation of other cell surface receptors cannot be excluded, the mechanism by which chondroitin inhibits COC expansion precludes the participation of these receptors. Chondroitin has been reported to inhibit HA-binding to cell surface receptors (47,48). However, if this were the case in the COC cultures, this inhibition should have been observed in the absence of exogenous TNFIP6. The fact, that chondroitin inhibits COC expansion only in the presence of exogenous TNFIP6 (i.e. when I␣I and P␣I are depleted before endogenous HA synthesis starts), indicates that the TNFIP6-mediated HC transfer to HA, rather than HA receptor binding, is crucial for the cumulus matrix assembly.
The findings of this report also provide further insights on the mechanism by which the cumulus matrix is stabilized. It has been hypothesized that cross-linking of hyaluronan chains are necessary for successful cumulus matrix assembly (19,21,26,49). This cross-linking has been proposed to occur through either the HA-linked HCs or HC-TNFIP6 complexes with the direct participation of the HA-binding domain of TNFIP6. These hypotheses are further strengthened by a recent observation that a blocking antibody against the HA-binding domain of TNFIP6 prevents the correct assembly of the cumulus matrix in vitro (50). However, our current results suggest that the mechanism of the cross-linking is probably more complex. First, if HA-linked HCs were alone to mediate cross-linking of HA chains, soluble HA would not be expected to inhibit HC transfer to immobilized HA in our ELISA system (Fig. 2D). Rather, the soluble HA chains would have stacked on each other and on the immobilized HA (all cross-linked by the HCs) possibly bearing even more HCs. The fact, that soluble HA inhibits the HC transfer to immobilized HA, implies that HA-linked HCs alone may not be sufficient for cross-linking HA. Second, the oligosaccharide specificity of the inhibition of cumulus matrix assembly suggests that the HA-binding domain of TNFIP6 may not be directly involved in the cross-linking of the HA chains. As we discussed above, TNFIP6 binding to hyaluronan can be effectively inhibited by HA6 oligosaccharide (41), while cumulus matrix formation is inhibited by HA8 but not HA6 oligosaccharides. Therefore, the HA binding domain may not contribute directly to the cross-linking (but appears to do so indirectly through the HC transfer). However, cooperativity between two TNFIP6 molecules in the matrix cannot be excluded and could result in a HA binding specificity of larger than HA6. Alternatively, TNFIP6 and/or the HCs may interact with additional matrix components to stabilize the cumulus matrix. Recently, pentraxin-3-null female mice have been shown to demonstrate very similar reproductive deficiencies (including impaired cumulus matrix formation) (20) 2 as TN-FIP6 and bikunin-null mice. Thus, pentraxin-3 could poten-tially be the additional candidate for the matrix stabilizing role.