Covalent linkage between proteins of the inter-alpha-inhibitor family and hyaluronic acid is mediated by a factor produced by granulosa cells.

The direct interaction of hyaluronic acid (HA) and proteins of the inter-α-inhibitor family plays a critical role in organization and stabilization of the expanding cumulus extracellular matrix (cECM) following an ovulatory stimulus. Despite similarities in the morphology of cumulus oocyte complexes (COCs) expanding in vivo and in vitro, we find that the cECM of COCs which expand within intact follicles are more elastic and resistant to shear stress than the cECM of those stabilized in vitro. Western blot analysis shows that only the heavy chains of inter-α-inhibitor are incorporated into the cECM and appears to be covalently linked to HA after stabilization in vivo while intact inter-α-inhibitor is bound to the HA-enriched cECM by a non-covalent mechanism in in vitro stabilized COCs. However, purified pre-α-inhibitor and HA can form covalent linkage in the presence of granulosa cells or with granulosa cell-conditioned medium. In addition, COCs resistance to shear stress is also enhanced by coincubation with granulosa cells. Upon formation of the apparent covalent linkage between heavy chains and HA in culture medium, the light chain (bikunin) is concomitantly released into the medium as a complex with chondroitin sulfate moieties of inter-α-inhibitor supporting the possibility that HA may replace the chondroitin sulfate linkage to the heavy chains. We speculate that a factor(s) secreted by granulosa cells within the follicle may catalyze a transesterification reaction resulting in an exchange of chondroitin sulfate with HA at the heavy chain/chondroitin sulfate junction followed by release of chondroitin sulfate-bikunin into the follicular fluid. It is also possible that the consequent further stabilization of the cECM through the covalent interaction of HA and heavy chains of inter-α-inhibitor may play an important role in the process of ovulation.

The direct interaction of hyaluronic acid (HA) and proteins of the inter-␣-inhibitor family plays a critical role in organization and stabilization of the expanding cumulus extracellular matrix (cECM) following an ovulatory stimulus. Despite similarities in the morphology of cumulus oocyte complexes (COCs) expanding in vivo and in vitro, we find that the cECM of COCs which expand within intact follicles are more elastic and resistant to shear stress than the cECM of those stabilized in vitro. Western blot analysis shows that only the heavy chains of inter-␣-inhibitor are incorporated into the cECM and appears to be covalently linked to HA after stabilization in vivo while intact inter-␣-inhibitor is bound to the HA-enriched cECM by a non-covalent mechanism in in vitro stabilized COCs. However, purified pre-␣-inhibitor and HA can form covalent linkage in the presence of granulosa cells or with granulosa cellconditioned medium. In addition, COCs resistance to shear stress is also enhanced by coincubation with granulosa cells. Upon formation of the apparent covalent linkage between heavy chains and HA in culture medium, the light chain (bikunin) is concomitantly released into the medium as a complex with chondroitin sulfate moieties of inter-␣-inhibitor supporting the possibility that HA may replace the chondroitin sulfate linkage to the heavy chains. We speculate that a factor(s) secreted by granulosa cells within the follicle may catalyze a transesterification reaction resulting in an exchange of chondroitin sulfate with HA at the heavy chain/chondroitin sulfate junction followed by release of chondroitin sulfate-bikunin into the follicular fluid. It is also possible that the consequent further stabilization of the cECM through the covalent interaction of HA and heavy chains of inter-␣-inhibitor may play an important role in the process of ovulation.
In most mammalian species (including mouse, rat, and human), cumulus-oocyte complexes (COCs) 1 of pre-ovulatory follicles undergo a dramatic change following an ovulatory stim-ulus. The tightly packed cumulus cells first disaggregate and then synthesize and secrete large amounts of hyaluronic acid (HA) into their extracellular matrices (ECMs). The ECM, cumulus cells, and oocyte are thus integrally bound within an expanded mucoid complex which is about 20 to 40 times larger (volume) dependent upon the species (1). This process of cumulus expansion is required for ovulation and may also facilitate the process of fertilization (2)(3)(4).
We have previously identified a serum factor (proteins of the inter-␣-inhibitor family), critical in organizing and stabilizing the expanding cumulus matrix (5). This protein factor appears to be excluded from follicular fluid until the ovulatory gonadotropin surge and then quickly diffuses into the follicular fluid where it becomes integrated within the cumulus ECM (5,6). Two major forms of this factor, pre-␣-inhibitor (P␣I) and inter-␣-inhibitor (I␣I), exist in mammalian species including mouse, bovine, and human (7,8). They each include a common light chain (about 40 kDa) which has two domains of the Kunitz-type trypsin inhibitor and so this protein is termed bikunin. P␣I is composed of bikunin and a single heavy chain connected by chondroitin sulfate (9 -12). I␣I consists of bikunin and two heavy chains also joined by chondroitin sulfate. According to a model proposed by Enghild et al. (10,11), a single chondroitin sulfate chain extends from a glycosylation site at Ser-10 of the bikunin subunit to link with the C-terminal Asp residue of each heavy chain via an ester bond to form a novel carbohydrate linkage. The three different heavy chains are highly homologous and, in fact, the specific heavy chain combinations identified in different species may differ from one another. For example, P␣I of human and mouse is composed of heavy chain 3 (HC3) and the light chain, while bovine P␣I consists of heavy 2 (HC2) and the light chain (13).
Both P␣I and I␣I are almost identical in their ability to stabilize the expanding cumulus ECM in vitro (14) where a direct interaction between proteins of the I␣I family and HA seems to play a critical role in preventing the release of HA into the culture medium. As demonstrated in an earlier in vitro study, this initial interaction appears to be a non-covalent charge-mediated interaction (14). Although COCs which expand in vitro are morphologically indistinguishable from those expanding in vivo, ovulated COCs appear to be more elastic and more resistant to mechanical shear force. It has been reported that proteins of the I␣I family could form covalent interactions with HA in various systems including follicular fluid (15)(16)(17)(18), however, the degree of native protein that forms covalent linkage with HA appears to be very low and the mechanism of the covalent interaction has not been clarified. Nonetheless, it is possible that a covalent interaction between I␣I and HA could result in this observed increased stability of the cumulus ECM.
In this study, we show that the majority of I␣I within the ovulated cumulus ECM is covalently linked with HA and that this covalent interaction can be partially achieved in vitro by incubating purified P␣I and HA with granulosa cells. Like ovulated COCs which expand within the intact follicle, COCs stabilized in medium containing granulosa cells or granulosa cell-conditioned medium, possess greater resistance to shear forces than those stabilized in medium lacking granulosa cells or granulosa cell-conditioned medium. This increased stability may be required for maintenance of integrity of the cumulus mass during extrusion of the COC through the rupture site within the follicular wall.

Methods
Preparation of COCs-Mice were injected with 5 IU of pregnant mare's serum gonadotropin and sacrificed 48 h later. Ovaries were placed in MEM with penicillin-G (100 units/ml) and streptomycin (50 g/ml). COCs (about 50 -80 COCs per animal) for in vitro expansion assays were isolated and incubated in medium containing MEM, 2.5 mM glucosamine, porcine follicle-stimulating hormone (2 g/ml), and other factors (FBS, purified bovine or mouse P␣I as specified in each experiment) at 37°C and 5% CO 2 for 16 h as described previously (5). In vivo stabilized ovulated COCs were collected about 12 h after an injection of an ovulatory dose of hCG (5 IU) in animals primed 48 h earlier with pregnant mare's serum gonadotropin (5 IU).
High Performance Liquid Chromatography Coupled ELISA for COCs-Ovulated COCs were washed 3 times in phosphate-buffered saline and then transferred to 500 l of 6 M guanidine HCl with 8% lauryl sulfobetaine or to 100 l of phosphate-buffered saline with 2 units of Streptomyces hyaluronidase for 3 h at 37°C and then transferred to 400 l of 6 M guanidine HCl with 8% lauryl sulfobetaine. About 100 l of each of these samples were fractionated using a gel-filtration column (TSK-G-4000, Bio-Rad) on a Waters HPLC unit eluted with phosphate-buffered saline at a 6 ml/min flow rate. Fractions were collected from 10 to 34 min. 100 l of each collected fraction was placed in a microwell plate (Dynatech Labs, Alexandria, VA) overnight at 4°C. The plates were then washed and blocked with 1% dry milk in 10 mM Tris-HCl buffer (pH 8.0) with 0.05% of Tween 20 (TBT). Following incubation with anti-human I␣I (1:1000 dilution with TBT) for 1 h at room temperature and three consecutive washes with TBT, the plates were then incubated with goat anti-rabbit IgG conjugated to horseradish peroxidase (1:1000 dilution, Bio-Rad). Following three washes with TBT, the plates were developed according to the manufacturer's instructions. The relative absorbance at 450 nm was determined with a Bio-Tek EL 309 Autoreader (Bio-Tek, Winnoski, VT).
Cumulus ECM Shear Resistance Assay-Pasteur pipettes were flamed to an average size of 180 m (inside diameter; about half the size of fully expanded and stabilized COCs). Individual COCs were sucked fully into the pipette and gently blown out (defined as one cycle). This process was repeated until the outer half of the cumulus mass had been stripped away and the remaining complex had been reduced to the size of the bore of the pipette. Although this parameter is judged subjectively, the release of the outer layer of the expanded cumulus mass is an all or nothing event which permits an accurate quantitation of the stability of expanded cECM. The resistance to shear stress (shear resistance index) is defined as the number of cycles necessary to strip off the outer half of the mass of cumulus cells.
Purification of Mouse and Bovine P␣I-Both mouse and bovine P␣I were purified from mouse serum or FBS through four consecutive steps including ammonium sulfate precipitation, HPLC using gel filtration, DEAE, and gel filtration as described previously (5,14). The purity of P␣I/I␣I used in this experiment is about 99% free of other proteins judged by Coomassie Blue staining following sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE). The cross-contamination of P␣I with I␣I, however, is about 5%.
Generation of Bikunin Site Specific Antiserum-Peptide corresponding to human bikunin sequence 116 -130 (QGNGNKFYSEKECRE) was synthesized and conjugated to keyhole limpet hemocyanin in the Department of Biochemistry, University of Kentucky (Lexington, KT). New Zealand White rabbits were immunized with the conjugated peptide and antiserum was collected as described previously (19). The specificity of the antiserum was characterized by Western blot as shown in Fig. 4B.
Western Blot Analysis-Protein samples from FBS, mouse serum, ovulated COCs, and various in vitro stabilized COCs or media (see figure legends for details of individual sample preparation) were heated to 100°C in SDS-PAGE sample buffer (62.5 mM Tris-HCl at pH 6.8 and containing 2% SDS, 10% glycerol, and 5% 2-mercaptoethanol) for 90 s. Prior to treatment in this sample buffer, most samples in MEM were divided equally into two parts and one part was treated with Streptomyces hyaluronidase for 3 h at 37°C. The Streptomyces hyaluronidase (Sigma, H1136) is specific for hyaluronic acid and does not have chondroitinase activity (20). As shown in this study, this enzyme will not dissociate the heavy chains and light chain of native P␣I/I␣I. The reduced samples were then resolved on polyacrylamide gel (Bio-Rad pre-casted 4 -15%, 4 -20% gradient, or homemade 7 or 10% as specified in figure legends) and then transferred to nitrocellulose paper. Following incubation with rabbit anti-human I␣I IgG (Dako) (1:1000) or antibikunin site-specific anti-serum (1:500) and alkaline phosphatase-conjugated goat anti-rabbit IgG (1:1000), the blots were developed using substrates according to the manufacturer's instructions (Bio-Rad).
Preparation of Chondroitin Sulfate Radiolabeled I␣I and Immunopreciptation-To radiolabel the chondroitin sulfate component of I␣I, a mouse hepatoma cell line was generated from an SV40 large T antigen transgenic mouse provided by Dr. J. S. Butel, Baylor College of Medicine. The cells were maintained and propagated in MEM supplemented with 10% FBS at 37°C under 5% CO 2 . After achieving confluence in a 75-cm 2 flask, the cells were washed 3 times with phosphate-buffered saline and then incubated with 5 ml of labeling medium (sulfate free MEM containing 2 mCi of carrier-free [ 35 S]sulfuric acid, 0.5% FBS) overnight. The medium was centrifuged (1000 ϫ g) for 10 min and passed through a 0.2-m filter (Millipore) to remove cell debris. The unincorporated radioisotope was removed using ultrafiltration with a molecular mass cut-off of 100 kDa (Centricone-100, Amicon) and the medium concentrated to a final volume of 0.5 ml. About 20 l of this sulfate-labeled protein mixture was then added to 80 l of MEM containing granulosa cells under various conditions specified in the figure legends and incubated overnight at 37°C and 5% CO 2 . These cellmedium mixtures were then centrifuged (at 1000 ϫ g for 10 min) to remove cell debris and 2 l of rabbit anti-human I␣I was added and the mixture then incubated at room temperature for 2 h following addition of 20 l of protein A-agarose suspension. The resulting mixture was incubated for another 2 h with gentle agitation and washed 4 times in Tris-HCl buffer (pH 8.0 with 1% bovine serum albumin and 0.05% Tween 20). 20 l of sample buffer (62.5 mM Tris-HCl, pH 6.8, with 2% SDS, 10% glycerol, and 5% 2-mercaptoethanol) was added and this mixture was heated to 100°C for 90 s. After centrifugation (16,000 ϫ g for 5 min), the supernatant was resolved on a 4 -15% gradient SDS-PAGE. The gels were then soaked with EN 3 HANCE (DuPont NEN) and fluorographed as recommended by the manufactures manual.

RESULTS AND DISCUSSION
In vitro and in vivo expanded COCs exhibit marked differences in resistance to shear forces. In vivo stabilized, ovulated COCs exhibit shear resistant indices greater than 60 in every ovulated COC tested (n ϭ 12). In fact, trituration of ovulated COCs was arbitrarily terminated at the 60th cycle since every ovulated COC tested in this manner was still intact. In sharp contrast, in vitro stabilized COCs (stabilized in the absence of granulosa cells) exhibit a shear resistance index of 8 Ϯ 2 (n ϭ 26), while those stabilized in vitro in the presence of granulosa cells exhibit a shear resistance index of 20 Ϯ 4 (n ϭ 23). While the number of cumulus cells in either conditions were not assessed, incubating COCs with granulosa cell-conditioned medium also resulted in an enhancement of the shear resistance index 18 Ϯ 4 (n ϭ 11). This suggests a different cECM stabilization mechanism in vivo and in cultured COCs, despite the similarity in morphology of stabilized COCs of both preparations. The difference may be in part because of the participation of granulosa cells.
Since the molecular mass of ovarian HA is Ͼ2000 kDa (21), gel filtration HPLC should be able to distinguish native P␣I/I␣I from the protein-HA complex. Thus, to assess the interaction of HA and proteins of the I␣I family within the ECM of in vivo ovulated COCs, the ovulated COCs were subjected to 6.0 M guanidine HCl and 8% lauryl sulfobetaine (a reagent which would dissociate most non-covalent interactions (16)), prior to subjecting the sample to gel filtration HPLC. The results are summarized in Fig. 1. Fig. 1, A and B, are controls of native bovine P␣I and P␣I treated with chondroitinase ABC where the bimodal peak in panel B represents the heavy chain and the light chain fraction of the native protein as verified by Western blot (not shown). However, following guanidine HCl-lauryl sulfobetaine treatment, the majority of the I␣I immunopositive fraction was still present in the void volume (Fig. 1C). When COCs were treated with Streptomyces hyaluronidase, the immunopositive fraction shifted to a position as a broader peak that corresponds roughly to the naked heavy chain (panel D). Moreover, these results implied that most of the inter-␣-inhibitor immunopositive material in the in vivo stabilized cumulus ECM was covalently associated with HA.
This conclusion was strengthened by Western blot analysis of the ovulated COCs using antibody against native I␣I (Dako) as shown in Fig. 2. Lane 2 illustrates the relative positions of native mouse I␣I and P␣I on SDS-PAGE. These bands migrate at about 220 and 130 kDa, respectively. Lane 3 shows that this antibody recognizes both the heavy and light chains of I␣I on SDS-PAGE followed by Western blot. These bands migrate at about 100 and 50 kDa, respectively. Both P␣I and I␣I could be dissociated from in vitro stabilized COCs by treatment of the sample with SDS and 2-mercaptoethanol (Fig. 2, lane 4). Prior treatment of the same sample of in vitro stabilized COCs with Streptomyces hyaluronidase, however, did not release any more detectable native I␣I, P␣I, or heavy chain (Fig. 2, lane 5), indicating that the interaction of I␣I/P␣I with HA without granulosa cells is non-covalent in nature. In contrast, major I␣I components of in vivo stabilized COCs could not enter the gel upon treatment of the sample with SDS and 2-mercaptoethanol (lane 6) without prior hyaluronidase treatment. The Coomassie Blue staining of the transferred gel shows similar transfer efficiency in both lanes 6 and 7 (not shown). After Streptomyces hyaluronidase treatment prior to SDS and 2-mercaptoethanol of the in vivo stabilized COCs, however, prominent immunostaining was visible with a major component of I␣I at about 100 kDa (probably the heavy chain) and a minor component migrating at about 200 kDa (probably a double heavy chain; see Fig. 3, below). There is, however, a very small amount of immunopositive material corresponding to the heavy chains and the native P␣I in the sample that is not treated with hyaluronidase (lane 6, compare to lane 7). It may be that a small amount of heavy chain spontaneously falls off during extraction. If all of the immunopositive material shown in lane 7, following hyaluronidase treatment, is covalently linked with HA, the conversion from native protein to the covalently bound form is almost complete. Such a high degree of covalent linkage between heavy chains of the I␣I family and HA in an extracellular matrix is unprecedented.
The time course of incorporation of I␣I heavy chains into the expanding HA enriched cumulus ECM in vivo is shown in Fig.  3. During the time frame from 3 to 12 h after the hCG injection, only trace amounts of intact I␣I/P␣I become incorporated into the cECM as shown by Western blot (lanes 2-4) without prior treatment with hyaluronidase. However, a large amount of P␣I/I␣I heavy chains were incorporated into the matrix by about 6 h after injection of hCG revealed by treating the sample with Streptomyces hyaluronidase (lanes 5-7). There is no detectable incorporation of native protein or heavy chains of P␣I/I␣I during the first hour following hCG, regardless of whether the samples are treated with hyaluronidase or not (not shown). In addition, a band was again observed at about 200 kDa, which was suspected to be comprised of double heavy chains, possibly derived from the I␣I that forms a covalent linkage with a single HA molecule in such a way that it is protected from the action of hyaluronidase. Indeed, further treatment of the hyaluronidase-treated sample with NaOH (0.1 M for 10 min at room temperature followed by neutralization with 0.1 M HCl; a method previously shown to dissociate the heavy chains from O-linked carbohydrates as well as the ester bond that links the heavy chain of I␣I with chondroitin sulfate (9, 18)), the 200-kDa band disappeared and only the single . Moreover, direct treatment of ovulated COCs with NaOH (without pretreatment with hyaluronidase, lane 15) also coverts this 200-kDa band to a sharp 100-kDa band that corresponds to the position of the heavy chain of I␣I. These results taken together support the possibility that a small fraction of the two heavy chains of I␣I form a covalent linkage with the same HA molecule in close proximity that is resistant to Streptomyces hyaluronidase or chondroitinase but sensitive to NaOH treatment.
The apparent covalent interaction observed in vivo within the ovulated cECM can be partially reproduced in vitro in the presence of granulosa cells or granulosa cell-conditioned medium. This system consisted of HA, purified I␣I or P␣I, and granulosa cells in MEM or granulosa cell-conditioned medium. As shown in Fig. 4A, only when granulosa cells or granulosa cell-conditioned medium are added into the reaction mixture, will the system generate the free light chain of P␣I and the heavy chain of P␣I which is released upon treating the sample with Streptomyces hyaluronidase (Fig. 4A, lanes 6 -9). Heattreated granulosa cell-conditioned medium is unable to facilitate the covalent binding of HA with heavy chain (Fig. 4A,  lanes 4 and 5). The same experiments illustrated in lanes 6 -9 of Fig. 4A were repeated and illustrated in lanes 2-5 of Fig. 4B but with a higher concentration of purified P␣I and HA. As stated above, Western blot of medium extracts showed two bands corresponding to P␣I and bikunin (Fig. 4B, lanes 2 and  4). Treatment of the samples with Streptomyces hyaluronidase prior to SDS and 2-mercaptoethanol treatment, however, revealed a prominent band corresponding to the heavy chain position (Fig. 4B, lanes 3 and 5). Purified bovine P␣I displayed the same pattern of interaction with HA when incubated with granulosa cells or granulosa cell-conditioned medium (not shown). The identity of the 50-kDa band as free bikunin (light chain) of P␣I was strengthened by using bikunin site-specific antiserum in the Western blot. In this experiment, a sample of purified P␣I was electrophoresed before (Fig. 4B, lanes 7 and 9) or after treatment with chondroitinase ABC to dissociate the light and heavy chains (Fig. 4B, lanes 8 and 10). The commercial anti-human I␣I rabbit IgG recognizes native protein (lane 7) and the dissociated heavy chain and light chain (lane 8). In contrast, the bikunin site-specific antiserum only recognizes native P␣I (lane 9) and the light chain (lane 10). Lane 12 is a Western blot illustrating an experiment in which the antibikunin site-specific antiserum was used to detect the presence of bikunin-positive epitopes after a sample of P␣I was incubated with granulosa cells and HA and then subsequently treated with Streptomyces hyaluronidase. As expected, the Western blot only showed the native protein band and the light chain. No heavy chain was detected. It should noted that the migration patterns for I␣I/P␣I and bikunin are somewhat different in different Western blots in this figure because different percentages of polyacrylamide were utilized as specified in the figure legend.
After treatment of native P␣I with chondroitinase ABC, the bikunin (light chain) fraction always appeared as two closely migrating bands on Western blot (e.g. Fig. 2 and Fig. 4B). This pattern may reflect the heterogeneity in the length of chon- droitin sulfate linkage in the native protein which has been shown to vary from 16 to 21 polydisaccharide units (11,12). Alternatively, it is possible that the unit of polysaccharide that connects the heavy chain and light chain may not be exclusively composed by chondroitin sulfate such that alternative cutting sites of chondroitinase may exist on the polysaccharide. This later possibility is consistent with the present study showing that upon forming the covalent linkage with HA, the released light chain-chondroitin sulfate complex migrates as a single band in Western blots (Fig. 4).
The amount of heavy chain binding covalently in our in vitro system is also dependent upon the concentration of exogenous HA (Fig. 5). The amount of native protein apparently binding covalently with HA progressively increases with increasing exogenous HA as judged by the progressive loss of native protein bands (lanes 3-6). In contrast, incubation of purified P␣I and HA with granulosa cells alone (Fig. 5, lane 2) or with exogenous HA alone in MEM (Fig. 5, lanes 7-9), does not generate any detectable heavy chain covalently bound to HA and the intensity of the native protein band is unaltered.
The physiological significance of the covalent interaction of the heavy chains of P␣I/I␣I with HA is not yet clear. However, further stabilization of the cumulus ECM seems to be achieved by this interaction as quantified indirectly by the shear-resistance assay in vitro. It should be pointed out that the efficiency of P␣I/I␣I incorporation into the cECM in vitro is much lower than that in vivo (Fig. 2, compare lanes 5 and 7). Since this low level of P␣I/I␣I incorporation occurs even at high serum concentration (10%), we are currently unable to adequately assess the status of the P␣I-HA complex in COCs stabilized in vitro when coincubated with granulosa cells. It is unlikely that the moderate enhancement of shear resistance of COCs by coincubating with granulosa cells results from recruitment of granulosa cells into the expanding COCs because the addition of granulosa cell-conditioned medium results in almost the same degree of enhancement. However, Salustri et al. (22) have shown that ovulated COCs have about 3 times more cumulus cells than those compact COCs expanded in vitro and that the origin of those extra cumulus cells occurs by recruitment of mural granulosa cells. Thus, the high shear resistance of in vivo stabilized COCs may occur as a consequence of recruitment of granulosa cells by the inner layer of the cumulus. Since granulosa cells possess the ability to catalyze the covalent interaction between heavy chains of P␣I/I␣I and HA, this may also lead to the packing and incorporation of large amounts of heavy chain into the expanding cECM. We speculate that the additional stabilization possibly achieved through the covalent interaction of heavy chains of P␣I/I␣I and HA, may provide elasticity required by the complex to maintain its integrity and to protect the oocyte during its extrusion from the ruptured follicle.
The current study also supports the likelihood that only the heavy chain of I␣I and P␣I forms a covalent linkage with HA while bikunin is concomitantly released into the culture medium when they are incubated with granulosa cells or in granulosa cell-conditioned medium. It is plausible to speculate that this process may be catalyzed or assisted by a factor(s) secreted by granulosa cells in response to an ovulatory stimulus. We have postulated that this conversion may involve an enzyme (esterase) synthesized by granulosa cells which catalyzes a transesterification between HA and chondroitin sulfate at the junction between the carboxyl end of the heavy chain of I␣I and chondroitin sulfate where chondroitin sulfate serves as the linker between the heavy chain and light chain (23). Indeed, Huang et al. (16) originally proposed that an enzyme(s) in serum could catalyze the exchange of HA with the chondroitin sulfate moiety of I␣I based upon the observation that incuba- All the samples consisted of 20 g/ml purified mouse P␣I in MEM medium with or without 6 ϫ 10 6 /ml granulosa cells. After incubation at 37°C and 5% CO 2 overnight, the mixtures were centrifuged at 200 ϫ g to remove granulosa cells. The SDS-PAGE sample buffer was then added to the supernatant followed by Western blot using antibody against human I␣I. Lane 1, high molecular weight standards. Lanes 2-6, P␣I incubated with granulosa cells and increasing amounts of HA (0, 0.1, 0.5, 1.0, and 2.0 mg/ml). Lanes 7-9, P␣I incubated without granulosa cells and with increasing amounts of HA (0.5, 1.0, and 2.0 mg/ml). Note the progressive increase in the apparent HC-HA smear above the native P␣I band and the concomitant progressive increase of the light chain at about 50 kDa. tion of serum with HA generates a heavy chain that is covalently linked with HA. In our system, however, overnight incubation of either purified I␣I/P␣I or serum (both mouse and bovine) with various amounts of HA alone could not generate any detectable heavy chain-HA complex or free bikunin. While the proportion of heavy chains covalently binding with HA after incubation in serum was not reported (16), conversion from a charge-mediated interaction between P␣I/I␣I and HA to a covalent binding of the heavy chain and HA in in vivo matured COCs is virtually complete. Serum may, however, contain a very low level of the hypothetical enzyme capable of catalyzing this charge mediated to covalent conversion. It will be interesting to determine whether or not long-term incubation of purified I␣I and HA can spontaneously generate a low level of heavy chain-HA complexes.
More recently, Zhao et al. (24) have found that I␣I heavy chain-HA complexes isolated from pathological synovial fluid form an ester bond with the C-terminal Asp residues of I␣I heavy chain. This finding and the present study are consistent with the transesterification model of covalent binding of the heavy chain and HA involving exchange of HA and the chondroitin sulfate component of I␣I as depicted in Fig. 6. This model predicts that formation of the covalent linkage with the I␣I heavy chain results in release of the chondroitin sulfate moiety of the native molecule along with bikunin. This was partially confirmed by radiolabeling the chondroitin sulfate moiety of the native I␣I and chasing this component into the "freed" bikunin fraction after formation of the covalent interaction between heavy chains and HA in vitro which occurs in the presence of granulosa cell-conditioned medium (Fig. 7). Equally supportive of the authenticity of this postulated interaction, however, is the finding that the heavy chain is absent (Fig. 7, lane 3) following treatment of the medium with hyaluronidase. This result indicates that the release of chondroitin sulfate is at least a necessary step in the formation of the HA-heavy chain complex. Further experiments are necessary to determine whether heavy chains of I␣I can independently form an equivalent covalent linkage with HA.