Effects of Glycosylation on the Secretion and Enzyme Activity of Mucor Rennin, an Aspartic Proteinase of MucorpusiZZus, Recombinant Yeast*

The Mucor rennin gene encoding a prepro form of the fungal aspartic proteinase from Mucorpusillus was expressed under the control of the yeast GAL7 promoter in Saccharomyces cerevisiae. The mature M. pusillus rennin secreted efficiently by yeast was a highly glycosylated protein. Analysis by a combination of site-directed mutagenesis of each of the three pos- sible glycosylation sites and treatment of the secreted M. pusillus rennins with

The aspartic proteinases produced extracellularly by the two closely related species of zygomycete fungi, Mucorpusillus (1) and Mucor miehei (a), possess relatively high milk-clotting activity along with low proteolytic activity. These are called Mucor rennins and are widely used as milk coagulants in industrial cheese production. We reported previously the cloning and sequencing of the structural gene for M. pusillus rennin (3). The gene encodes a prepro enzyme composed of 361 amino acids of the mature M. pusillus rennin and an additional NH*-terminal sequence of 66 amino acid residues. When the gene was expressed in Saccharomyces cerevkiae cells under the control of the yeast GAL7 promoter, a highly glycosylated form of the mature M. pusillus rennin which was excreted into the medium efficiently (4). We also confirmed that the primary secretion product of the enzyme was a pro-M. pusillus rennin possessing a pro-sequence of 44 amino acids; the pro-sequence was removed by autocatalytic processing at acidic pH (5).
The three possible N-glycosylation sites, i.e. Asn7', Asn"", and Asn'=. Analysis of the mature M. pusillus rennin secreted from the recombinant yeast indicated that the enzyme was highly glycosylated mainly with mannose moieties (about 37 residues/mol) (4). In contrast, the commercial preparation of M. pusillus rennin produced by M. pusillus contains only a few glycosidic moieties in its molecule (6). Comparison of the catalytic properties of both the preparations has revealed that the ratio of milk-clotting activity to proteolytic activity of the recombinant yeast M. pusillus rennin is distinctly lower than that of the commercial M. pu.sil1u.s rennin.
The present study was undertaken to determine the exact glycosylation sites of yeast M. pusillus rennin together with the effect of glycosylation on the catalytic properties of the enzyme. We also found that mutant M. pusillus rennins lacking the glycosylation sites obtained by means of sitedirected mutagenesis were secreted in decreased levels from the yeast cells.    (Fig. 2b), suggesting that the commercial M. (14). Two chromogenic oligopeptides I and II (Table II)  Sites-In order to determine the glycosylated sites, we con- (Fig. 1) had a larger molecular mass (46 kDa) than that of the structed mutant M. pusillus rennins in which each of the 3 commercial M. pusillus rennin (40 kDa) produced by the asparagine residues described above was exchanged to glutaindustrial strain of M. pusillus (4). This seemed to be due to mine by site-directed mutagenesis.

Strains
the different extents of asparagine-linked glycosylation in S. cerevisiae transformants carrying the mutated M. pusillus these two M. pusillus rennins because endo H treatment of rennin genes were cultured in YPGal medium for 4 days, and the culture supernatants were subjected to SDS-polyacrylamide gel electrophoresis.
As shown in Fig. 3, both the Gin'" and GlnlRX M. pusillus rennins in the culture supernatant had a reduced molecular mass of about 43 kDa whereas the molecular mass of Gin"" M. pusillus rennin was the same as that of nonmutated M. pusillus rennin (46 kDa) (Fig. 3). This is consistent with the observations on the time course of endo H digestion of yeast M. pusillus rennin. When both Asn'" and Asn'** were changed to glutamine, the mass of the M. pudus rennin with the double mutation was the same as that of the commercial M. pusillus rennin. All these data clearly showed that the number of glycosylation sites of yeast M. pusillus rennin was two and that they were AsnTg and Asn"".
Effect of Glycosylation on Secretion of M. pusillus Rennin-All the lanes of SDS-polyacrylamide gel shown in Fig. 3 contained M. pusillus rennin samples prepared from equal volumes (100 ~1) of the culture supernatants.
The secreted amounts of Gin" and GlnlXR M. pusillus rennins were apparently reduced. Secretion of the double mutant Gln'g/Gln'~ M. pusillus rennin was reduced to about %O that of the nonmutated M. pusillus rennin. In order to determine whether the lack of glycosylation in the mutated M. pusillus rennins caused a decrease in intracellular to extracellular transport, we examined the amounts of M. pusillus rennin in both the culture supernatants and the cells. In this experiment, the yeast transformants were induced in YPGal medium for 2 days because prolonged cultivation for 4 days caused extensive decrease in the intracellular M. pusillus rennins, probably due to proteolytic activity of the host cells. As shown in Fig. 4, most of the extracellular proteins cross-reactive with the anti-M. pusillus rennin antibody was present in two forms. The yeast transformants were cultured in YPGal medium at 30 "C for 4 days, and each 100 ~1 of supernatant was subjected to SDS-polyacrylamide gel electrophoresis. Lane A4, molecular weight standards; lane C, purified commercial M. pusillus rennin (1.4 pg). Each 25 ~1 of culture was centrifuged at 6000 x g for 5 min, and all the proteins in the supernatants and the cells were subjected to SDS-polyacrylamide gel electrophoresis. Asterisked bands indicate pro-M. pusillus rennins with different extents of glycosylation. Lane Y, purified yeast M. pusillus rennin (300 ng). rennin to mature M. pusillus rennin occurred in the culture broth (5), we concluded that the upper band (asterisked in Fig. 4) was pro-M. pusillus rennin, and the lower band was mature M. pusillus rennin. When the cultivation was prolonged for 2 more days, the pro-M. pusillus rennins (upper band) were converted to the mature M. pusillus rennin (lower band), and the pattern of SDS-polyacrylamide gel electrophoresis was identical to that shown in Fig. 3. The Gln'"/Gln"" M. pusillus rennin gave a single band of pro-M. pu.sil1u.s rennin due to a very small extent of self-processing at a low concentration of the protein (5). Prolonged cultivation resulted in yielding a single band of mature M. pusillus rennin that had the same mass as that shown in Fig. 3. The mutations of Gin'", GlnlR8, and Gln'g/Gln'm caused a marked decrease in the extracellular M. pusillus rennins whereas Gin".' M. pusillus rennin was excreted as efficiently as the nonmutated M. pusillus rennin. In contrast, distinct intracellular accumulation of the immunoreactive proteins was observed only with the mutations at the actual glycosylation sites. A major band detected in the cells of each Gln5', Gin",', and Gin'= transformant showed a larger molecular mass than that of the nonglycosylated pro-M. pusillus rennin, suggesting that they were pro-M. pusillw rennins glycosylated at the residual asparagine residues in the secretory apparatus. Several degradation products were also detected with Glnlm and Gln'S/Gln'm.
All these data clearly indicated that glycosylation at the two sites of M. pusillus rennin was required for efficient secretion from yeast cells.
Effect of Glycosylation on Enzyme Actiuity-The catalytic properties of the endo H-treated yeast M. pusillus rennin and the mutated M. pusillus rennins were compared with those of the nonmutated yeast M. pusillus rennin and the commercial M. pusillus rennin. The results are shown in Table III.
The nonmutated yeast M. pusi1lu.s rennin showed distinctly lower milk-clotting activity with relatively higher proteolytic activity than those of the commercial M. pusillus rennin. As a consequence, the clotting activity/proteolytic activity ratio of the yeast M. pusillus rennin was only one-fourth that of the commercial M. pusillus rennin. The endo H treatment of the yeast M. pusillus rennin improved the ratio to a value very similar to that of the commercial M. pusillus rennin mainly because of an increase of clotting activity. Similar changes were observed with the mutated M. pusillus rennins lacking one or both of the N-glycosylation sites. As the number of glycosylated residues was reduced, their clotting activity increased, and their proteolytic activity decreased. GlnT9 and Gin'= had almost the same clotting activity, and Gln7g/Gln'R8 showed the most improved clotting activity/ proteolytic activity ratio, being almost equal to those of the commercial M. pu.sil1u.v rennin and endo H-treated yeast M. pusillus rennin but slightly different from that of the commercial M. pusillus rennin.
We analyzed further the kinetic properties of these mutants using synthetic oligopeptides I and II (Table IV). Although the K,,, and k,., values of the yeast M. pusillus rennin for both the substrates were slightly different from those of the commercial nonglycosylated M. pusillus rennin, they still did not coincide after endo H treatment or site-directed mutagenesis.

DISCUSSION
The present work has revealed that the 2 asparagine residues, Asn7" and Asn'%, among the three possible N-linked glycosylation sites in the M. pusillus rennin polypeptide are actually glycosylated during the secretory processes whereas the Asn"" residue is not glycosylated in yeast cells. Although x-ray crystallographic analysis of M. pusillus rennin has not  Aspa and ASPIC' are the catalytic residues of this enzyme. The figure was drawn according to Subramanian et al. (16). yet been done, the highly conserved tertiary structure among several aspartic proteinases allows speculation on the possible locations of these residues in the M. pudlus rennin molecule. The side chain of Asnu3 seems to exist in the exposed region of the molecule, as depicted in the structure of endothiapepsin (Fig. 5) (16). The specificity of glycosylation may therefore not be due to the accessibility of the asparagine residues to glycosyl transferase. Another possible explanation is the relative rate of glycosylation of asparagine residues in Asn-X-Thr and Asn-X-Ser. Kaplan et al. (17) reported that the threonine-containing sequence was glycosylated more rapidly than the serine-containing one both in vivo and in vitro (17). Consistent with this, the Asn79 and Asn's8 residues of M. pusillus rennin are in the Asn-X-Thr sequence whereas AsnIl is in the Asn-X-Ser sequence. Recently, the same group discovered a glycosylation site binding protein, a component of the oligosaccharide transferase (18). The glycosylation site binding protein might play an important role in sorting the glycosylation of M. pusillus rennin.
Removal of the glycosylation sites of M. pusillus rennin resulted in a significant decrease in the amount of secreted M. pusillus rennin along with accumulation of M. pusi11u.s rennin in the yeast cells. A relationship between N-linked glycosylation and protein secretion by yeast has also been reported for human tissue-type plasminogen activator expressed in S. cereuisiae (19). In Chinese hamster ovary cells, a mutant tissue-type plasminogen activator lacking possible N-linked glycosylation sites remained in the cell as a stable complex with a protein GRP78, which regulates the transport of secretory proteins (20). It is possible that a similar protein is associated with the accumulation of the mutant M. pusillus rennins in S. cereuisiae cells (21,22).
The clotting activity of yeast M. pusillus rennin was distinctly low but recovered after endo H treatment. Increase in the relative milk-clotting activity was also observed with the mutated M. pusillus rennins lacking 1 or both of the 2 glycosylated asparagine residues. These results indicate that glycosylation of M. pusillus rennin causes distinct modulation of its enzymatic activity, probably due to a change in the specificity for scissile peptide bonds in the high molecular weight protein substrates. A similar result was obtained when the tissue-type plasminogen activator secreted from 5'. cereuisiae was treated with endo H (19). However, only a slight change in the kinetic parameters was observed with the synthetic substrates examined. Similar results were obtained upon mutagenesis of the glycosylation site of another aspartic proteinase, renin (23). The difference in the lengths of substrates probably accounts for this inconsistency. For example, various aspartic proteinases show different pH optima for acid-denatured hemoglobin but almost identical optima for shorter peptide substrates (14,24). A longer synthetic peptide with an appropriate sequence will be required for further analysis of the effect of glycosylation on catalytic activity.
One of the glycosylation sites, Asn79, is located in the "flexible flap region" that partially covers the substrate binding cleft. Several residues on the flap may interact with a bound substrate. One of the flap residues in chymosin, Tyr77, corresponding to Tyr8* in M. pusillus rennin, seems to be involved in both substrate binding and catalytic activity (14). It seems possible that Asn7'-linked glycosylation causes perturbation of the flap structure, thus influencing the enzyme activity. Another glycosylation site, AsnIB, is located at the junction of the NH*-terminal and the COOH-terminal domains of the enzyme. We may assume that Asnlss-linked glycosylation influences the global tertiary structure of the enzyme. However, such a possible effect should be examined carefully by several means such as NMR spectroscopy.
The clotting and proteolytic activities of the endo H-treated yeast M. pusillus rennin and the mutant M. pusillus rennin (Gln7g/Gln*88) lacking the N-glycosylation sites were not completely identical to those of commercial M. pusillus rennin. The reason for this is still unknown, but the difference of the host is one possible explanation; the process of protein folding or some other modification might differ slightly between S. cerevisiae and M. pusillus.