Role of carbohydrate in human chorionic gonadotropin. Effect of periodate oxidation and reduction on its in vitro and in vivo biological properties.

Sequential removal of the carbohydrate of human chorionic gonadotropin (hCG) results in the impairment of its biological activity in vitro. The in vivo activity of these derivatives, however, could not be investigated due to their rapid removal from circulation by liver. Therefore, asialo-hCG was subjected to controlled oxidation and reduction in order to render it suitable for in vivo studies. Two periodate oxidizedreduced derivatives, designated as PORA-hCG-I and PORA-hCG-11, were obtained. While PORA-hCG-I lost about 85% of the galactosyl residues, in PORA-hCG-I1 all galactosyl and 40% mannosyl residues were modified. A detailed physico-chemical and immunological characterization of the derivatives established the integrity of the polypeptide chain. The modification did not adversely affect the receptor binding activity of the hormone, rather PORA-hCG-I and -11 both bound to the ovarian receptor with about 2.4 times higher affinity (Ka, 1.8 x 10'' M-') than hCG (Ka, 0.75 X 10'' M-'). Furthermore, the oxidation of the terminal galactosyl residues in asialo-hCG resulted in a drastic decrease of about 290-fold in its binding affinity to the liver receptor compared to asialo-hCG (Ka, 0.19 x 10' "I). As a result, the plasma clearance rate of PORA-hCG-I decreased considerably with concomitant increase in its ovarian uptake to a level comparable to hCG. In vitro response in corpus luteal cells showed that, although PORA-hCG-I and PORA-hCG-I1 stimulated steroidogenesis maximally, they had only 3-5% potency compared to hCG based on the dosage at half-maximal response. On the other hand, in cAMP production, both derivatives stimulated only 20% of the maximal response of hCG at the highest concentration used (1 pg/ ml). In both cAMP production and steroidogenesis, PORA-hCG-I inhibited the action of hCG, inhibition being more pronounced in cAMP production (80%) than in steroidogenesis (20-30%). The properties of PORAhCG-I studied in vivo included the effect on the ovarian ascorbic acid depletion and the trophic effect as indicated by the increase in the mouse uterine weight. This derivative not only showed reduced hormonal activity of 2-5% in these assays, but it also caused a dosedependent inhibition of the hCG activity. These find-

Sequential removal of the carbohydrate of human chorionic gonadotropin (hCG) results in the impairment of its biological activity in vitro. The in vivo activity of these derivatives, however, could not be investigated due to their rapid removal from circulation by liver. Therefore, asialo-hCG was subjected to controlled oxidation and reduction in order to render it suitable for in vivo studies. T w o periodate oxidizedreduced derivatives, designated as PORA-hCG-I and PORA-hCG-11, were obtained. While PORA-hCG-I lost about 85% of the galactosyl residues, in PORA-hCG-I1 all galactosyl and 40% mannosyl residues were modified. A detailed physico-chemical and immunological characterization of the derivatives established the integrity of the polypeptide chain. The modification did not adversely affect the receptor binding activity of the hormone, rather PORA-hCG-I and -11 both bound to the ovarian receptor with about 2.4 times higher affinity (Ka, 1 Furthermore, the oxidation of the terminal galactosyl residues in asialo-hCG resulted in a drastic decrease of about 290-fold in its binding affinity to the liver receptor compared to asialo-hCG (Ka, 0.19 x 1 0 ' ' "I). As a result, the plasma clearance rate of PORA-hCG-I decreased considerably with concomitant increase in its ovarian uptake to a level comparable to hCG. In vitro response in corpus luteal cells showed that, although PORA-hCG-I and PORA-hCG-I1 stimulated steroidogenesis maximally, they had only 3-5% potency compared to hCG based on the dosage at half-maximal response. On the other hand, in cAMP production, both derivatives stimulated only 20% of the maximal response of hCG at the highest concentration used (1 pg/ ml). In both cAMP production and steroidogenesis, PORA-hCG-I inhibited the action of hCG, inhibition being more pronounced in cAMP production (80%) than in steroidogenesis (20-30%). The properties of PORA-hCG-I studied in vivo included the effect on the ovarian ascorbic acid depletion and the trophic effect as indicated by the increase in the mouse uterine weight. This derivative not only showed reduced hormonal activity of 2-5% in these assays, but it also caused a dosedependent inhibition of the hCG activity. These find-* This work was supported by United States Public Health Grant R01-HD-08766 and funds from the World Health Organization H9181337. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  ings strongly suggest that while the carbohydrate is not involved in the hormone-receptor interaction, it is needed for the expression of the hormonal activity.
Human chorionic gonadotropin is a glycoprotein hormone in which the carbohydrate constitutes about 30% of the molecule. Because of the recent advances in our understanding of the detailed structures of the protein and carbohydrate parts of hCG' (1-4) and the vital role it plays in maintaining pregnancy in humans, it has been the target for the development of newer means of fertility regulation. It was reported earlier that sequential removal of monosaccharides from hCG by exoglycosidases resulted in a considerable impairment of biological activity in vitro as judged by steroidogenesis and cAMP production in granulosa and leydig cells (5-8). Furthermore, all the carbohydrate-modified derivatives were able to bind to ovarian and testicular receptors with equal or greater affinity than the native hCG. However, in vivo evaluation of the biological properties of these derivatives could not be undertaken because of their rapid clearance from circulation by the carbohydrate specific receptors on the liver cell surface (9-12). This paper deals with further modification of the nonreducing terminal galactosyl residues in asialo-hCG by controlled periodate oxidation and reduction with the objective of prolonging its plasma half-life. The in uitro and in vivo agonistic and antagonistic properties of the resulting PORA-hCG have been investigated and are the subject of the present. A preliminary report of this work has been made earlier (13).

RESULTS
Preparation and Characterization of PORA-hCG-I and PORA-hCG-11-For the preparation of PORA-hCG-I and 11, a highly purified hCG with specific activity of 12,000 IU/mg was used as the starting material (22). Sialic acid was quantitatively removed with neuraminidase treatment. The terminal galactosyl residues thus exposed were se1:ctively oxidized with periodate oxidation under controlled conditions. The optimal experimental conditions for oxidation with respect to time, temperature, and concentration of sodium metaperiodate were established using asialo-fetuin. As is clear from Fig. 1, in 3.5 h 1.8 mol of periodate was consumed per mol of galactose compared to a theoretical value of 2 mol for complete oxidation. Thus, 85% galactosyl residues were destroyed under these conditions without any significant effect on any other carbohydrate. It is worth pointing out that for NaBH., reduction, the pH should be strictly controlled and kept below 7.0. Even a brief exposure of the derivative to alkaline conditions may lead to reduction of the disulfide bonds and thereby inactivation of the hormone. The PORA-hCG-I thus obtained was further treated with periodate for an additional period of 4 h to oxidize completely the galactosyl residues. The second preparation was designated as PORA-hCG-11. The recovery of the derivatives was about 8045%.
In order to assure that the chemical treatment used in the preparation of hCG derivative produced only the desired modification of the carbohydrate, a detailed physico-chemical characterization of PORA-hCG-I and its subunits was undertaken. The separation of PORA-hCG-I subunits, after dissociation in 8 M urea, was achieved by gel filtration on Sephadex G-75 (Fig. 2). The two subunits of PORA-hCG-I, a and P, eluted as sharp peaks with elution coefficients (&) of 0.41 and 0.22 compared with 0.35 and 0.14, respectively, for native hCG subunits. Tables I and I1 show the carbohydrate and amino acid compositions of the derivatives, respectively. While 85% of the exposed galactosyl residues were destroyed in PORA-hCG-I, almost all galactose was oxidized in PORA-hCG-11. In addition, the latter also lost 40% of mannose. The bulk of 15% galactosyl residues still remaining in PORA-hCG-I were found to be in the P-subunit. These were further located in the COOH-terminal chymotryptic peptide (Table   I),  (residues 109-145) which contains all O-glycosidically-linked sugar chains (2,4). The periodate treatment did not affect the hexosamine content in both preparations. The amino acid analyses of both PORA-hCG preparations did not show any significant change from that of hCG. It should be noted that cystinyl and methioninyl residues which are quite susceptible to oxidation did not undergo any loss, suggesting that the protein backbone was essentially unaffected during oxidation. The NH2-terminal amino acids of a and P subunits of PORA-hCG-I were found by dansyl method to be alanine -. E - and serine, respectively. In addition, a small amount of valine was observed in PORA-hCG-I-a, consistent with the microheterogeneity in the NH2-terminal amino acid sequence of hCG-a as reported earlier (1). Both preparations of PORA-hCG were homogeneous when examined by polyacrylamide gel electrophoresis in acidic and alkaline buffer systems (Fig. 3). At pH 8.3, while hCG with highly negative charge due to 14-16 sialic acid residues (3,4) was able to penetrate the gel (RF, 0.21), A-hCG and both preparations of PORA-hCG did not migrate appreciably. Conversely, at pH 4.5, while hCG failed to enter the gels, its derivatives migrated with identical RF values of about 0.26. When the above derivatives were examined by SDS-polyacrylamide gel electrophoresis under completely denaturing condition, their a and P-subunits moved with identical RF values of 0.50 and 0.42, respectively. Under similar conditions, the

Carbohydrate-modified Analogs of
hCG 12627 a-and P-subunits of hCG moved slower with R.F values of 0.45 and 0.35, respectively. Fig. 4 depicts the absorption spectra from 240 to 340 nm of hCG, PORA-hCG-I, and PORA-hCG-11. The absorption spectra of the derivatives closely resemble with that of hCG. HCG, PORA-hCG-I, and PORA-hCG-I1 have absorption maxima at 278 nm with molar extinction coefficients of 19,880, 17,950, and 17,840, respectively. These studies suggest the absence of any detectable denaturation in hCG on oxidation. This is further supported by the identical immunological activity of hCG and its derivatives (91-102%) as determined by radioimmunoassay using "'I-hCG and anti-hCG antiserum (Fig. 5). Furthermore, hCG and the derivatives yielded parallel dose response curves, again confirming the integrity of the polypeptide chains.

Binding of hCG and Its Derivatives to Ovarian Receptor-
In order to assess further the effect of modification of carbohydrates of A-hCG on binding to the ovarian receptor, radioligand receptor assay was performed with crude ovarian plasma membranes (Fig. 6 ) . The equilibrium binding constants (KO) listed in Table I11 were determined from the Scatchard plots of the competitive binding curves shown in Fig. 6. Desialylation of hCG resulted in about 2.6-fold increase in its binding affinity (K,,, 1.99 X 10"' "I). The K , value for native hCG was found to be 0.75 X 10" M". These results are in close agreement with those reported by Moyle et al. (8), and Tsuruhara et al. (34) for rat testicular and ovarian mem-   6. In vitro binding of hCG and derivatives to rat ovarian homogenates. 100,000 cpm of '251-hCG was incubated with ovarian homogenate in the presence of different concentrations of serially diluted hCG or derivatives. In the absence of hCG, 30,000 cpm were bound and assumed as 100% binding. Specific binding was calculated by subtracting the nonspecific binding, i.e. in the presence of 10 pg hCG, which was 2,500 cpm. The per cent binding, displayed as logit brane receptor, respectively. Further oxidation of the terminal galactosyl residues resulted in a slight decrease in the binding affinity as compared to A-hCG, but it was still 2.4 times higher than hCG. Similar results were obtained when testicular plasma membranes were used in the assay. The above data, thus, suggests that higher ovarian uptake of PORA-hCG-I in vivo was probably due to its higher affinity for the receptor (see below).

Stimulation of CAMP Production and Steroidogenesis in Corpora LutcBal
Cells-The dose response curves of cAMP production a.nd steroidogenesis obtained with varying amounts of hormone is shown in Fig. 7. Although both preparations of PORA-hCG were able to stimulate steroidogenesis maximally, they had only 3-5% hCG activity calculated on the basis of ED5(, (see "Experimental Procedures"). Half-maximal stimulation of steroidogenesis was obtained at a concentration of about 1 ng/ml of PORA-hCG compared with a value of 0.03 ng/ml obtained with the native hormone. In case of cAMP production, both preparations of PORA-hCG stimulated only 20% response of that of hCG at the highest concentration used ( 1 pg/ml). A-hCG produced about 50% of the maximal response of hCG a t 1 pg/ml. Thus, PORA-hCG acted as a partial of hCG agonist of hCG. Consequently, the ability of PORA-hCG-I to inhibit hCG action in the two assays was determined. As shown in Fig. 8, equal or higher amount of PORA-hCG-I was able to neutralize about 80% of the activity of 10 ng and 100 ng of hCG in CAMP assay. In steroidogenesis, PORA-hCG-I neutralized the activity of hCG only partially (20-30%). Similar results were obtained previously with the glycosidasetreated hCG derivatives in Leydig cells (8).
It is clear from the data that the two preparations of PORA-hCG were indistinguishable in their physico-chemical, immunological, and in vitro biological properties. Therefore, further in vitro as well as in vivo studies were conducted with only PORA-hCG-I.
Binding of hCG a n d Its Derivatives to Liver Membrane Receptors-In order to assess the effect of oxidation of terminal galactosyl residues of A-hCG on its binding to the liver Gal/Glu receptor, competitive binding assay was performed using Iz5I A-hCG and crude rat liver membranes. As may be seen from Fig. 9, the oxidation of terminal galactosyl residues led to a greatly diminished ability of A-hCG to bind to the liver membrane receptor. While 20 ng of A-hCG was able to cause 50% inhibition of the "' 1 A-hCG binding, EDSO for PORA-hCG-I was 12,000 ng. At the highest level tested (25 pg), hCG produced only 20% inhibition. Scatchard analyses of the above data showed that, while A-hCG had very high affinity for Gal/Glu receptor & 0.19 X 10"' M"), PORA-hCG-I possessed 290-fold less affinity, with K,, of 6.54 X 10" M " (Table 111). The association constant K , for hCG was 0.93 X 10" M-I. Thus, the increased plasma survival time of PORA-hCG-I was probably due to its decreased binding to liver Gal/ Glu receptor (see below).
Plasma Clearance of hCG and Its Analogs-The plasma survival curves of '251-labeled hCG, A-hCG, and PORA-hCG-I are shown in Fig. 10. As expected, the desialylation of hCG led to a dramatic increase in its plasma clearance rate with

FIG. 9. Competition of binding of '"I-labeled A-hCG and unlabeled hormones to the liver-homogenate.
About 250 pg of crude plasma membrane protein from liver (2,000 X g pellet of liver homogenate) was incubated with 5 ng of "'I-labeled A-hCG (100,000 cpm) in 50 mM Tris-HCI buffer, pH 7.2, containing 0.2 mM CaCL in a total volume of 1.2 ml. In the absence of unlabeled A-hCG, 28,000 cpm of '"'I-labeled A-hCG were bound, which represents 100% binding on the ordinate. Nonspecific binding was determined in the presence of 10 pg of asialo-fetuin and was always less than 5% of the total cpm added.  tlr2 of less than 0.5 min as a result of the exposed galactosyl residues. The plasma half-life of the oxidated derivative was substantially increased (t1,2, 15 min), though still less than that of native hormone (tlrS, 48 min). During the duration of these experiments (30 min), no phosphotungstic acid-soluble counts were detectable in the plasma, indicating that little or no degradation of hCG and its analogs had occurred.
Ovarian Uptake of hCG and Its Derivatives in Immature Superovulated Rats-Although the mature female rats provided a good system to examine the plasma clearance of hCG and its derivatives, it was a poor model for the comparison of their ovarian uptake because of the low levels of uptake, ie. less than 1% of the injected dose. Moreover, the corpora lutea of mature rats contain few receptors and remain nonfunctional unless activated by mating or other stimuli (35). Therefore, to study the effect of oxidation of A-hCG on ovarian uptake, immature, superovulated rats were used. The results of these studies are shown in Fig. 11. While A-hCG did not accumulate in the ovaries to a significant extent, PORA-hCG-I was taken up by the ovaries at a level comparable to hCG.
Biopotency of PORA-hCG-I-Since PORA-hCG-I binds to the ovarian or testicular receptors in vivo or in vitro equally or better than hCG, it was possible to assay its biological activity in uiuo. Two bioassays, ovarian ascorbic acid depletion and mouse uterine weight were performed to determine the specific activity of PORA-hCG-I, the former measures hCG/ LH activity and the latter total gonadotropic activity. The

TABLE V
Ability of PORA-hCG-Z to neutralize the biological actiuity of hCG in ovarian ascorbic acid depletion assay SDecific activitv of hCG 12.000 IU/me (see leeend to Table IV). results of these bioassays are shown in Fig. 12 and Table IV.
While a mean specific activity of 499 IU/mg (369-629, 95% confidence limit) was obtained with OAAD assay, a lower value of 181 IU/mg (146-216,95% confidence limit) was estimated by MUW. The average specific activity obtained from the two assay systems was 357 f 68.9 IU/mg (222-492, 95% confidence limit). Thus, oxidation of galactosyl residues re- sulted in a derivative which had only 2 4 % agonist activity as compared with hCG.
Ability of PORA-hCG-I to Neutralize the hCG Activity in Vivo-Since PORA-hCG-I binds to the ovarian receptor in vitro as well as in vivo, it was examined for its ability to neutralize the effect of hCG in the above two bioassay systems. Tables V and VI show the results of these studies. As may be seen from Table V in OAAD assay. Similarly, in the MUW assay (Table V), it is obvious that PORA-hCG-I significantly reduced the effect of the hCG to about 40% (II:IV, p < 0.001; VI:VIII, p < 0.001).

DISCUSSION
The role of carbohydrate in the function of glycoproteins has attracted a great deal of attention in recent years. Since the classical findings of Ashwell and Morel1 of the presence of galactose/glucose receptor on hepatocytes (36), several other receptors specific for sugars such as mannose, fucose and Nacetylglucosamine, and mannose-6-phosphate have been found in reticuloendothelial system and skin fibroblast, respectively (11,12,(37)(38)(39). Although the physiological significance of the receptors in the metabolism of glycoproteins remains to be clarified, these studies do establish the function of carbohydrates in cellular recognition. As far as the role of carbohydrate in the mechanism of action of glycoproteins of biological significance is concerned, only a limited number of such studies have been reported thus far (5)(6)(7)(8). We first initiated studies on the role of carbohydrate in the biological and immunological function of hCG in vitro (8). The sequential removal of sugar residues from hCG resulted not only in the loss of hormonal activity but also imparted antagonistic properties to the hormone. However, these in vitro studies could not be extended to in vivo situation in the whole organism because of the removal of the derivatives from plasma by the specific receptors for sugars on the liver cell surface. This prompted the present investigation, in which further modification of A-hCG was necessary in order to prolong its plasma half-life by inhibiting the binding to the liver glycoprotein receptors without affecting its ability to bind to the ovarian receptor. This was achieved by the controlled periodate oxidation and reduction of the terminal galactosyl residues in A-hCG.
In order to ensure that the polypeptide chain was unaffected by the modification, and that any change in the biological properties of the hormone was due to the carbohydrate modification, a detailed physicochemical and immunological characterization of the derivative was undertaken. The data on the amino acid composition, NH,-terminal amino acid analysis, polyacrylamide gel electro-phoresis, and gel filtration chromatography failed to show any detectable alteration in the polypeptide backbone. Also, there seems to be no apparent denaturation of the derivative as indicated by the UV absorption spectrum (Fig. 4) and its immunological activity determined by radioimmunoassay (Fig. 5). The retention of the receptor binding activity as measured by radioreceptor-assay further substantiates the above conclusion. In fact, PORA-hCG has 2.4 times higher affinity for the receptor than hCG (Fig. 66, Table 111).
Prior to evaluating the efficacy of PORA-hCG in uivo, initial studies were carried out in vitro. The data on the cellular response of PORA-hCG show that the modification markedly impaired the ability of the hormone to stimulate cAMP production and steroidogenesis without affecting the receptor binding activity. While the derivative stimulated maximal progesterone production at higher dosage (about 40 times), it could produce only 20% of the maximal level of cAMP stimulated by hCG. These results agree with those of previously reported by Moyle et al. (8) in Leydig cells. While the carbohydrate modification resulted in the loss of the binding of the derivative to the liver receptor ( Fig. 9), the binding to the target tissue was maintained or even enhanced. In agreement with the in vitro activity of PORA-hCG, the in vivo bioassays showed that it had 2-5% biopotency of hCG; 5% as measured by OAAD assay, a value significantly higher than that of 2% determined by MUW assay. Furthermore, PORA-hCG was found to be more effective in inhibiting the activity of hCG in OAAD than MUW assay. No definitive explanation can be given for the above discrepancy except that the time course of experiment varied considerably in the two assays. While OAAD assay required 4 h, MUW assay was performed over a period of 4 days. The plasma half-life of PORA-hCG-I (15 min) was less than that of hCG (48 min) because of its increased removal by the kidney (22). However, it had higher binding affinity (Kc,, 1.78 X 10"' M") for the receptor than hCG (& 0.75 X 10"' M").
The present data strongly indicate that the carbohydrate plays a crucial role in the function of the hormone, although the precise molecular site of its action has not been established. Since the derivative does bind to the receptor, the lesion must lie in the post receptor binding event(s). Based upon the currently available model of polypeptide hormone action, several possibilities can be considered for the molecular site of action of the Carbohydrate. It is conceivable that the carbohydrate modification may prevent the aggregation of hormone-receptor complex essential for the expression of hormonal response (40,41). This may occur due to a change in the fluidity of the hormone-receptor complex in the membrane. A recent report by Amsterdam et al. (40) suggests that small clusters of hCG/LH receptors in granulosa cells may play a role in the cellular responsiveness. Similarly, another report by Schechter et al. (41) shows that while a CNBr cleaved analog of epidermal growth factor is devoid of biological activity, it retains sufficient receptor binding activity, but cannot form cell surface clusters in fibroblast. Addition of bivalent, but not monovalent, antibody of epidermal growth factor restores the biological response as well as the patch formation. It is also possible that the modification may affect the conformation of the hormone-receptor complex and thereby its fluidity in the membrane and interaction with the regulatory subunit of adenylate cyclase. Whether the carbohydrate binding component is an independent entity or a part of the adenylate cyclase system is not known. It is conceivable, however, that one of the subunits of the guanine nucleotide binding protein (G) may be involved in the interaction with the carbohydrate. The G protein from rat liver and turkey erythrocyte membrane has been found to be oligomeric in nature (42,43). If this is the case, the presence of such a 17. Reisfeld, R. A., Lewis, U. J., and Williams, D. E. (1962) Nature subunit may represent an additional step in the regulation of ( in A-hCG by periodate oxidation and reduction. The iesults presented here further substantiate our previous observation that carbohydrate is not required for hormone-receptor interaction, but is important in the expression of the hormonal activity in the target tissue. The carbohydrate-modified derivatives of hCG can serve as probes in the investigation of the mechanism of hormone action. Last, but not least, such derivatives have also potential in fertility control.