- Email: [email protected]
Some species of human leukocyte interferon are glycosylated
ARCHIVES
Vol.
OF BIOCHEMISTRY
232, No. 1, July,
AND BIOPHYSICS pp. 422-426, 1984
COMMUNICATION Some Species of Human Leukocyte JAMES
E. LABDON, AND
Departments
KENNETH SIDNEY
of Biochemistry and *Phy.siologicd Molecular Biology, Roche Research Received
Interferon D. GIBSON,* PESTKA’
Are Glycosylated SHAIU
Chemistry and Pha rmmologg, Center, Nutley, New Jersq
March
SUN,*
Roche
Institute
of
07110
5, 1984
The carbohydrate content of all of the species of human leukocyte interferon (IFNcu) which have been derived from patients with chronic myelogeneous leukemia (CML) and purified to homogeneity has now been determined. Amino sugar content was measured by high-performance liquid chromatography and fluorescamine detection of acid hydrolysates of each sample. Two species showed significant amounts of glucosamine. Most of the purified species of leukocyte interferon from a myeloblast cell line were also tested, and two species were found to contain sugar residues. These forms also differed from the CML interferons in that they revealed the presence of greater amounts of galactosamine. The apparent lack of carbohydrate in some of the higher-molecularweight species of interferon implicated factors other than glycosylation in the molecular weight differences. The results indicate that some species of IFN-(U are glycosylated to various degrees.
The human interferons are proteins which exhibit a variety of effects, including the ability to inhibit virus growth and cell proliferation. Some interferons, such as fibroblast (IFN$)* and immune (IFN-y), are subject to post-translational processing in eucaryotic cells into glycosylated forms. Knight (1) first reported that fibroblast interferon was a glycoprotein by staining the purified form on an SDS-polyaerylamide gel with p-aminosalicylic acid. The binding of IFNfl to columns containing various lectins (2,3) and its specific elution by monosaccharides was consistent with the presence of carbohydrate on this interferon. Friesen et al. (4) directly measured the amino sugar content of purified fibroblast interferon for the first time. Furthermore, the cDNA sequence of a DNA recombinant containing the coding sequence for IFN6 showed a potential asparagine-linked glycosylation
site (5-S). Immune interferon is also believed to be glycosylated on the basis of lectin binding (9) and the presence of two possible glycosylation sites in the cDNA sequence (10,ll). Direct demonstration of carbohydrate on natural immune interferon has recently been reported (12). To date, the presence of carbohydrate moieties has not been shown to be a prerequisite for biological activity. Bose et al (13) have shown that the antiviral activity of a human fibroblast interferon preparation is unchanged after treatment with a mixture of glycosidases, during which up to 89% of the labeled sugars were removed. In a similar experiment, Kelker et al. (14) noticed no change in the antiviral activity of human immune interferon after glycosidase treatment, despite the fact that the enzymes reduced the apparent molecular weight of the protein significantly. Indeed, the recombinant species of the leukocyte, fibroblast, and immune interferons produced in Escherichia coli possess biological activity in spite of a lack of sugar residue in the recombinant products (611, 15, 16). The situation with respect to the glycosylation of the natural leukocyte interferons has been less clear. Unlike fibroblast or immune interferons, the leukocyte
1 To whom correspondence should be addressed. * Abbreviations used: IFN-(3, fibroblast interferon; IFN-7, immune interferon; SDS, sodium dodecyl sulfate; CML, chronic myelogeneous leukemia; IFN-(r, leukocyte interferon; GluNAc, N-acetylglucosamine; GalNAc, N-acetylgalactosamine. 0003-9861/84 Copyright 0 All
rights
$3.00
1984 by Academic Press, Inc. of reproduction in any form reserved.
422
GLYCOSYLATED
HUMAN
interferons represent a family of proteins with considerable sequence homology among the various species as determined by tryptic mapping (1’7, 18) and amino acid sequencing (19-21). DNA recombinants containing the coding sequences for leukocyte interferons also exhibit such sequence homology (15). However, only one of these clones, IFN-aH, of 12 whose sequence has been determined has two potential asparagine-linked glycosylation sites. Previously, Allen and Fantes (19) detected no carbohydrate in their leukocyte interferon preparations. Rubinstein et al. (17) purified a number of species of leukocyte interferon to homogeneity and found much less than 1 mol of glucosamine or galactosamine per molecule of protein; however, only five of the 11 purified species were tested. Although glycosylation has not been shown to be important in terms of the in vitro biological activity of the interferons, it is important, nevertheless, to examine the role of the sugar residues in these molecules because they may influence their molecular architecture, stability, pharmacodynamics, and antigenicity. In this report, we shall present direct evidence for the first time that certain species of the natural leukocyte interferons are indeed glycosylated. Experimental procedures. Preparation of leukocyte interfercms. Leukocyte interferon species were prepared from CML cells and the KG-l cell line. The interferons were prepared and resolved into the various species as described earlier (17,18). Protein concentrations of each species were determined by fluorescamine analysis (20). Determination of glucosamiw and galuctosamine. Carbohydrate content was estimated by measurement of glucosamine and galactosamine. Up to 113 pmol of each protein was used in the glycosylation determinations. A sample was dissolved in 0.2 ml of 4 M HCl and heated to 100°C for 4 to 8 h. The hydrolysates were dried, redissolved in sodium citrate buffer (0.2 M, pH 2.2), applied to a Dionex DC-IA column, and eluted with the citrate buffer at pH 5.3. The column temperature was maintained at 69”C, and a flow rate of 8.0 ml/h was used. Column effluent was monitored by fluorescamine detection (22). Under these conditions, the retention times for phenylalanine and the amino sugars glucosamine and galactosamine were 28,37, and 41 min, respectively. The quantity of amino sugar was determined by integrating the peaks and comparing areas with known standards. Standards consisted of a sample of ovalbumin, which contained 5 mol glucosamine/mol protein, and the recombinant interferons IFN-(uA, and IFN-@, which were used as negative controls because they have no sugar residues. Phenylalanine liberated by the hydrolysis reaction was used as an internal standard for the quantitation of protein hydrolyzed. The phenylalanine content of most of the sample interferons bad been determined previously t,y amino acid analysis (1’7, 18). The lower
LEUKOCYTE
INTERFERONS
423
limit of detectability of amino sugars by this method was estimated to be 5 pmol (23). Nomenclature. The recommended standard abbreviations are used for the leukocyte, fibroblast, and immune interferons as IFN-(Y, IFN-8, and IFN-7, respectively. However, because the purified species of interferon from CML cells used in these analyses were described previously by similar designations, the designations in the previous reports are used herein to avoid confusion and permit direct reference to these prior communications. Individual purified natural IFN-(Y interferon species will be identified by spelling their Greek letter designations in full to distinguish them from the standard leukocyte, fibroblast, and immune species. For example, alphal, alphaz, betai, bet%, beta,, gammai, gammaa, gammaa, gamma4, gamma,, and delta represent natural leukocyte interferons previously isolated (15, 17). Results and discussion Table I lists the results of the determination of amino sugar content for most of the previously untested natural species. Of the leukocyte interferons derived from CML cells, alphaz, betai, betaz, beta,, and gamma, were previously found to contain no amino sugars (17). None of the KG-l interferons had been previously analyzed for carbohydrate content. The CML-derived species gamma1 appears to have about one residue each of glucosamine and galactosamine per molecule of protein. Gammai has a molecular weight in the range of most of the unglycosylated interferons (17,700), and was therefore not considered a likely candidate for extensive glycosylation. The other CML species containing sugar residues was delta, which contained approximately 5 residues of glucosamine per molecule. Delta is characterized by its unusually high molecular weight compared to the other CML species. On SDS-polyacrylamide gel electrophoresis, it migrated with apparent molecular weight values of 21,000 and 24,000 in two discrete bands (17). It is possible that variable extents of glycosylation are responsible for the lack of molecular weight homogeneity in this species. The high molecular weight may also reflect the presence of neutral sugars which would escape detection by this method. The higher affinity of this species for diol silica HPLC resins (17) may also result from the presence of attached sugar molecules. The increased availability of hydroxyl and amino groups on the glycoprotein would be expected to provide more potential hydrogen bonds with the resin. The delta species is unusual in two other respects. First of all, it does not appear in all preparations of interferon from CML cells. This may be related to the source of the leukocytes. Secondly, delta is the only leukocyte interferon species which is significantly more active on human cells than on bovine cells (approximately eightfold more active). It would be in-
8.0 9.5 9.3 10.3 10.1 7.5 12.6-14.4
4.8 5.7 5.6 5.8 5.1 4.5 6.0”
5.4b 5.4 6.5 6.2
5.3 6.4 6.1 4.5
16,500 16,200 17,700 17,700 21,000 16,500 21-24 K
19 or 17.6 K 17,800 26,200 19,000
18,200 19,500 19,219 20,000
b, bz ba Cl
C2 4 IFN-aA IFN-8
310 225 (W 862 382 220 64 1060 (838) 382 226 544 166 (1333) 377 111 450 340
Phe
,
39 24 93 37 21 9 84-74 (67-58) 37/40 24 32 14 (113) 39 9 45 46
Protein detected (pmol)
interferons
0 0.5 (0) 0.9 0 0 0 4.9-5.6 (5.1-5.8) 2.1-2.3 0 0 0.9 (0.3) 0 . 0 0 0
IFN-cwA, and IFN-8
5.1-5.5 0 0 2.5 (1.7) 0.4 0 0 0
KW
0
0
0
1.2 0
0 cl (0)
Galactosamine/ protein
Molar ratios Glucosamine/ protein
alpha* from CML cells, and recombinant
114 0 0 0 0 (37) 203 0 0 34.9 (196) 16 0 0 0
0 0 (0)
0 11 (0) 88 0 0 0 410 (338) 85 0 0 13 (39) 0 0 0 0
Galactosamine
Glucosamine
Residues detected (pmol)
Note. Values in parentheses are for a second trial. Human leukocyte interferon produced in E. coZi were included as controls. n Average Phe composition for the leukocyte interferons was used. b Phe composition of br was used.
9.6 12.5 10.0 9.0
10.2619.5 9.6 17.0 11.8
Phe residues/ molecule
Molecular weight
Mel% Phe
IFN-(Y species
I
GLYCOSYLATIONOF INTERFERONS
TABLE
G-
z
g
8
F-
GLYCOSYLATED
HUMAN
teresting to determine if these properties of delta result from the presence of sugar residues. Of the KG-l-derived interferons tested, only bi and c1 were found to contain amino sugar residues. Unlike their CML counterparts, these interferons possessed larger amounts of galactosamine in addition to glucosamine. The c1 species appears to contain about two molecules of galactosamine and possibly one of glucosamine per molecule of protein. The most heavily glycosylated of all leukocyte interferon species tested was bi, which has about two residues of glucosamine and five of galactosamine per molecule of protein. However, the bi species migrated as two bands on SDS-polyacrylamide gel electrophoresis, corresponding to molecular weight values of 19,090 and 17,699. It is likely that the lower-molecular-weight species represents contaminating be (18), which was found to contain no amino sugars in this study. Therefore, the values in Table I probably reflect smaller than actual amounts for the sugar content of bl. An unexpected result was the lack of detectable amino sugar content in most of the higher-molecularweight species, particularly in betas (17), gammrq, and ba. The amount of protein hydrolyzed was well in excess of that required to reliably detect 1 mol amino sugar/m01 protein. It seems more likely that the apparent high molecular weights of these species on SDS-polyacrylamide gel electrophoresis reflect the anomalous migration of these species, rather than the presence of carbohydrate, although neutral sugars may also play a role. Due to a limited amount of material available, only about 8 to 9 pmol of species gammas and di were analyzed. However, one residue of amino sugar per molecule of protein would have been detected were it present. Since no significant trace of either sugar was observed on the chromatogram for either sample, it seems unlikely that these species are glycosylated. Insufficient material precluded testing of KG-l species dp. The “a” species derived from KG-1 cells was not tested either because these forms were not resolved and were present in only very small amounts (18). The presence of glucosamine in some of the species of the leukocyte interferons is consistent with several of the known polysaccharide structures of glycoproteins. These structures have been reviewed in detail by Wash and Bahl(24). The delta species could contain several asparagine-linked, high mannose-type structures containing only two residues of N-acetylglucosamine (GluNAe) each, although the presence of mannose cannot be confirmed by this method of analysis. Alternatively, its high glucosamine content could be explained by the presence of a single, complextype structure containing two to four extra GluNAc residues in the peripheral portion of the chain. The absence of galactosamine in the delta species would seem to rule out the presence of 0-glycosidic linkages to serine and threonine in these structures. The sit-
LEUKOCYTE
INTERFERONS
425
uation is quite different with respect to gammai and the interferons derived from the KG-l cell line. The presence of galactosamine in species gammai, bi, and ci would indicate an unusual structure or combination of structures for the carbohydrate chains in these molecules. To date, the only reported asparaginelinked polysaccharide structures containing N-acetylgalactosamine (GalNAc) on proteins have been for the pituitary glycoprotein hormones (24). In these structures, the GalNAc residues are located in the peripheral region of the carbohydrate chain. A notable exception exists for the a subunits of the human pituitary hormones leutropin, follotropin, and thyrotropin, in which a GalNAc residue is located in the core (25). The presence of galactosamine in gamma1 and two of the KG-l species may be significant in terms of the hormone-like properties of the interferons. The unusually high ratio of galactosamine to glucosamine in bl, however, may be due to the presence of additional serineor threonine-linked structures containing predominantly GalNAc residues. Examples of such structures are antifreeze glycoprotein, human IgA, cartilage keratan sulfate, and epiglycanin from TA-3 cells (24). Other proteins containing this type of structure but with GalNAc, N-acetylneuraminic acid, or neutral sugars in the periphery include the @ subunit of human chorionic gonadotropin, bovine fetuin, human erythrocyte membrane protein, and some canine and porcine submaxillary glycoproteins (24). Simple O-linked structures to serine or threonine would also explain the composition of species gammai and ci; how ‘er, the possibility of novel carbohydrate structures in these interferon species cannot be dismissed either. It is also possible that the presence of relatively large amounts of galactosamine in the KG1 interferon species represents an abnormality associated with the transformed nature of this tumor cell line. Further information must await a more detailed examination of the structures of these molecules. The results show that only two leukocyte interferon species derived from CML cells are glycosylated. These interferons, gammai and delta, both contained glucosamine, whereas gammai was found to possess one residue of galactosamine as well. The glycosylated species derived from KG-1 cells differed in that they contained greater amounts of galactosamine in addition to glucosamine. Variations in the distribution of the glycosylated species, particularly delta, in the interferon preparations may account for variability in the extents of glycosylation of crude leukocyte interferons. In our laboratory, delta was present in only some of the preparations, yet it was the most extensively glycosylated interferon from CML cells. In addition, the results indicate that the presence of carbohydrate alone cannot adequately explain why some of the interferon species (beta,, gammq, and hs) migrate with apparent molecular weights signif-
426
LABDON
icantly higher than expected. Furthermore, the amount of amino sugar detected in some species (e.g., delta) cannot entirely account for its high molecular weight. It remains to be seen to what extent longer polypeptide chains, neutral sugars, or other residues play a part in the molecular weight or structure of these species.
ET
11.
12. 13.
REFERENCES 14. 1. KNIGHT, E., JR. (1976) Proc. Natl. Acad. Sci USA 73,520-523. 2. TAN, Y. H., BARAKAT, F., BERTHOLD, W., SMITHJOHANNSEN, H., AND TAN, C. (1979) J. Bill Chew. 254, 8067-8073. 3. DAVEY, M. W., HUANG, J. W., SULKOWSKI, E., AND CARTER, W. A. (1974) J. Bid Chem. 249,63546355. 4. FRIESEN, H.-J., STEIN, S., EVINGER, M., FAMILLETTI, P. C., MOSCHERA, J., MEIENHOFER, J., SHIVELY, J., AND PESTKA, S. (1981) Arch. Biochem, Bie phys. 206, 432-450. 5. TANIGUCHI, T., OHNO, S., FUJI-KURIYAMA, Y., AND MURAMATSU, M. (1980) Gene 10.11-15. 6. DERYNCK, R., CONTENT, J., DE CLERCQ, E., VOLCKAERT, G., TAVERNIER, J., DEVOS, R., AND FIERS, W. (1980) Nature (London) 285,542-547. 7. HOUGHTON, M., STEWART, A. G., DOEL, S. M., EMTAGE, J. S., EATON, M. A. W., SM~~H, J. C., PATEL, T. P., LEWIS, H. M., PORTER, A. G., BIRCH, J. R., CARTWRIGHT, T., AND CAREY, N. H. (1980) Nucl. Acio!s Res. 8, 1913-1931. 8. GOEDDEL, D. V., SHEPARD, H. M., YELVERTON, E., LEUNG, D., CREA, R., SLOMA, A., AND PESTKA, S. (1980) Nucl. Acids Res. 8, 4057-4074. 9. JANKOWSKI, W. J., DAVEY, M. W., O’MALLEY, J. A., SULKOWSKI, E., AND CARTER, W. A. (1975) J. Viral. 16, 1124-1130. 10. GRAY, P. W., LEUNG, D. W., PENNICA, D., YELVERTON, D., NAJARIAN, R., SIMONSEN, C. C., DERYNCK, T., SHERWOOD, P. J., WALLACE, D. M.,
15. 16.
17.
18. 19. 20.
21.
22.
23.
24. 25.
AL. BERGER, S. L., LEVINSON, A. D., AND GOEDDEL, D. V. (1982) Nature @vm!un) 295, 503-508. DEVOS, R., CHEROUTRE, H., TAYA, Y., DEGRAVE, W., VAN HEU~ERS~~N, H., AND FIERS, W. (1982) NULL Acids Res. 10, 24874501. BRAUDE, I. In Proceedings of the FDA Workshop on Interferon, in press. BOSE, S., GURARI-ROTMAN, D., RUEGG, U. T., CORLEY, L., AND ANFINSEN, C. (1976) J. Biol. Chem. 251, 1659-1662. KELKER, H. C., YIP, Y. K., ANDERSON, P., AND VILCEK, J. (1983) J. Bid Chem 258.8010-8013. PESTKA, S. (1983) Arch Biochem. Biophys. 221, l37. PESTKA, S., KELDER, B., FAMILLE~I, P. C., MosCHERA, J. A., CROWL, R., AND KEMPNER, E. S. (1983) J. Biol. Chem 258, 9706-9709. RUBINSTEIN, M., LEVY, W. P., MOSHERA, J. A., LAI, C.-Y., HERSHBERG, R. D., BARTLETT, R. T., AND PESTKA, S. (1981) Arch. Bi&ewz. Biuphys. 210, 307-31s. HOBBS, D. S., AND PESTKA, S. (1982) J. Bid Chew. 257, 4071-4076. ALLEN, G., AND FANTES, K. H. (1980) Nature (Lendon) 287,408-411. ZOON, K. C. (1981) in Methods in Enzymology (Pestka, S., ed.), Vol. 78, pp. 457-464, Academic Press, New York. LEVY, W. P., RUBINSTEIN, M., SHIVELY, J., DEL VALLE, U., LAI, C.-Y., MOSCHERA, J., BRINK, L., GERBER, L., STEIN, S., AND PESTKA, S. (1981) Proc. Natl. Acad Sci. USA 78, 6186-6190. STEIN, S., AND MOSCHERA, J. (1981) in Methods in Enzymology (S. Pestka, ed.), Vol. 79, pp. 716, Academic Press, New York. KILPATRICK, D. L., GIBSON, K. D., AND JONES, B. N. (1983) Arch. Biochem. Biophys. 224,402404. WAGH, V. W., AND BAHL, 0. M. (1981) CRC &it. Rev. Biochem. 10, 307-377. HARA, K., RATHNAM, P., AND SAXENA, B. B. (1977) J. Biol. Chem 253, 1582-1591.
Vol.
OF BIOCHEMISTRY
232, No. 1, July,
AND BIOPHYSICS pp. 422-426, 1984
COMMUNICATION Some Species of Human Leukocyte JAMES
E. LABDON, AND
Departments
KENNETH SIDNEY
of Biochemistry and *Phy.siologicd Molecular Biology, Roche Research Received
Interferon D. GIBSON,* PESTKA’
Are Glycosylated SHAIU
Chemistry and Pha rmmologg, Center, Nutley, New Jersq
March
SUN,*
Roche
Institute
of
07110
5, 1984
The carbohydrate content of all of the species of human leukocyte interferon (IFNcu) which have been derived from patients with chronic myelogeneous leukemia (CML) and purified to homogeneity has now been determined. Amino sugar content was measured by high-performance liquid chromatography and fluorescamine detection of acid hydrolysates of each sample. Two species showed significant amounts of glucosamine. Most of the purified species of leukocyte interferon from a myeloblast cell line were also tested, and two species were found to contain sugar residues. These forms also differed from the CML interferons in that they revealed the presence of greater amounts of galactosamine. The apparent lack of carbohydrate in some of the higher-molecularweight species of interferon implicated factors other than glycosylation in the molecular weight differences. The results indicate that some species of IFN-(U are glycosylated to various degrees.
The human interferons are proteins which exhibit a variety of effects, including the ability to inhibit virus growth and cell proliferation. Some interferons, such as fibroblast (IFN$)* and immune (IFN-y), are subject to post-translational processing in eucaryotic cells into glycosylated forms. Knight (1) first reported that fibroblast interferon was a glycoprotein by staining the purified form on an SDS-polyaerylamide gel with p-aminosalicylic acid. The binding of IFNfl to columns containing various lectins (2,3) and its specific elution by monosaccharides was consistent with the presence of carbohydrate on this interferon. Friesen et al. (4) directly measured the amino sugar content of purified fibroblast interferon for the first time. Furthermore, the cDNA sequence of a DNA recombinant containing the coding sequence for IFN6 showed a potential asparagine-linked glycosylation
site (5-S). Immune interferon is also believed to be glycosylated on the basis of lectin binding (9) and the presence of two possible glycosylation sites in the cDNA sequence (10,ll). Direct demonstration of carbohydrate on natural immune interferon has recently been reported (12). To date, the presence of carbohydrate moieties has not been shown to be a prerequisite for biological activity. Bose et al (13) have shown that the antiviral activity of a human fibroblast interferon preparation is unchanged after treatment with a mixture of glycosidases, during which up to 89% of the labeled sugars were removed. In a similar experiment, Kelker et al. (14) noticed no change in the antiviral activity of human immune interferon after glycosidase treatment, despite the fact that the enzymes reduced the apparent molecular weight of the protein significantly. Indeed, the recombinant species of the leukocyte, fibroblast, and immune interferons produced in Escherichia coli possess biological activity in spite of a lack of sugar residue in the recombinant products (611, 15, 16). The situation with respect to the glycosylation of the natural leukocyte interferons has been less clear. Unlike fibroblast or immune interferons, the leukocyte
1 To whom correspondence should be addressed. * Abbreviations used: IFN-(3, fibroblast interferon; IFN-7, immune interferon; SDS, sodium dodecyl sulfate; CML, chronic myelogeneous leukemia; IFN-(r, leukocyte interferon; GluNAc, N-acetylglucosamine; GalNAc, N-acetylgalactosamine. 0003-9861/84 Copyright 0 All
rights
$3.00
1984 by Academic Press, Inc. of reproduction in any form reserved.
422
GLYCOSYLATED
HUMAN
interferons represent a family of proteins with considerable sequence homology among the various species as determined by tryptic mapping (1’7, 18) and amino acid sequencing (19-21). DNA recombinants containing the coding sequences for leukocyte interferons also exhibit such sequence homology (15). However, only one of these clones, IFN-aH, of 12 whose sequence has been determined has two potential asparagine-linked glycosylation sites. Previously, Allen and Fantes (19) detected no carbohydrate in their leukocyte interferon preparations. Rubinstein et al. (17) purified a number of species of leukocyte interferon to homogeneity and found much less than 1 mol of glucosamine or galactosamine per molecule of protein; however, only five of the 11 purified species were tested. Although glycosylation has not been shown to be important in terms of the in vitro biological activity of the interferons, it is important, nevertheless, to examine the role of the sugar residues in these molecules because they may influence their molecular architecture, stability, pharmacodynamics, and antigenicity. In this report, we shall present direct evidence for the first time that certain species of the natural leukocyte interferons are indeed glycosylated. Experimental procedures. Preparation of leukocyte interfercms. Leukocyte interferon species were prepared from CML cells and the KG-l cell line. The interferons were prepared and resolved into the various species as described earlier (17,18). Protein concentrations of each species were determined by fluorescamine analysis (20). Determination of glucosamiw and galuctosamine. Carbohydrate content was estimated by measurement of glucosamine and galactosamine. Up to 113 pmol of each protein was used in the glycosylation determinations. A sample was dissolved in 0.2 ml of 4 M HCl and heated to 100°C for 4 to 8 h. The hydrolysates were dried, redissolved in sodium citrate buffer (0.2 M, pH 2.2), applied to a Dionex DC-IA column, and eluted with the citrate buffer at pH 5.3. The column temperature was maintained at 69”C, and a flow rate of 8.0 ml/h was used. Column effluent was monitored by fluorescamine detection (22). Under these conditions, the retention times for phenylalanine and the amino sugars glucosamine and galactosamine were 28,37, and 41 min, respectively. The quantity of amino sugar was determined by integrating the peaks and comparing areas with known standards. Standards consisted of a sample of ovalbumin, which contained 5 mol glucosamine/mol protein, and the recombinant interferons IFN-(uA, and IFN-@, which were used as negative controls because they have no sugar residues. Phenylalanine liberated by the hydrolysis reaction was used as an internal standard for the quantitation of protein hydrolyzed. The phenylalanine content of most of the sample interferons bad been determined previously t,y amino acid analysis (1’7, 18). The lower
LEUKOCYTE
INTERFERONS
423
limit of detectability of amino sugars by this method was estimated to be 5 pmol (23). Nomenclature. The recommended standard abbreviations are used for the leukocyte, fibroblast, and immune interferons as IFN-(Y, IFN-8, and IFN-7, respectively. However, because the purified species of interferon from CML cells used in these analyses were described previously by similar designations, the designations in the previous reports are used herein to avoid confusion and permit direct reference to these prior communications. Individual purified natural IFN-(Y interferon species will be identified by spelling their Greek letter designations in full to distinguish them from the standard leukocyte, fibroblast, and immune species. For example, alphal, alphaz, betai, bet%, beta,, gammai, gammaa, gammaa, gamma4, gamma,, and delta represent natural leukocyte interferons previously isolated (15, 17). Results and discussion Table I lists the results of the determination of amino sugar content for most of the previously untested natural species. Of the leukocyte interferons derived from CML cells, alphaz, betai, betaz, beta,, and gamma, were previously found to contain no amino sugars (17). None of the KG-l interferons had been previously analyzed for carbohydrate content. The CML-derived species gamma1 appears to have about one residue each of glucosamine and galactosamine per molecule of protein. Gammai has a molecular weight in the range of most of the unglycosylated interferons (17,700), and was therefore not considered a likely candidate for extensive glycosylation. The other CML species containing sugar residues was delta, which contained approximately 5 residues of glucosamine per molecule. Delta is characterized by its unusually high molecular weight compared to the other CML species. On SDS-polyacrylamide gel electrophoresis, it migrated with apparent molecular weight values of 21,000 and 24,000 in two discrete bands (17). It is possible that variable extents of glycosylation are responsible for the lack of molecular weight homogeneity in this species. The high molecular weight may also reflect the presence of neutral sugars which would escape detection by this method. The higher affinity of this species for diol silica HPLC resins (17) may also result from the presence of attached sugar molecules. The increased availability of hydroxyl and amino groups on the glycoprotein would be expected to provide more potential hydrogen bonds with the resin. The delta species is unusual in two other respects. First of all, it does not appear in all preparations of interferon from CML cells. This may be related to the source of the leukocytes. Secondly, delta is the only leukocyte interferon species which is significantly more active on human cells than on bovine cells (approximately eightfold more active). It would be in-
8.0 9.5 9.3 10.3 10.1 7.5 12.6-14.4
4.8 5.7 5.6 5.8 5.1 4.5 6.0”
5.4b 5.4 6.5 6.2
5.3 6.4 6.1 4.5
16,500 16,200 17,700 17,700 21,000 16,500 21-24 K
19 or 17.6 K 17,800 26,200 19,000
18,200 19,500 19,219 20,000
b, bz ba Cl
C2 4 IFN-aA IFN-8
310 225 (W 862 382 220 64 1060 (838) 382 226 544 166 (1333) 377 111 450 340
Phe
,
39 24 93 37 21 9 84-74 (67-58) 37/40 24 32 14 (113) 39 9 45 46
Protein detected (pmol)
interferons
0 0.5 (0) 0.9 0 0 0 4.9-5.6 (5.1-5.8) 2.1-2.3 0 0 0.9 (0.3) 0 . 0 0 0
IFN-cwA, and IFN-8
5.1-5.5 0 0 2.5 (1.7) 0.4 0 0 0
KW
0
0
0
1.2 0
0 cl (0)
Galactosamine/ protein
Molar ratios Glucosamine/ protein
alpha* from CML cells, and recombinant
114 0 0 0 0 (37) 203 0 0 34.9 (196) 16 0 0 0
0 0 (0)
0 11 (0) 88 0 0 0 410 (338) 85 0 0 13 (39) 0 0 0 0
Galactosamine
Glucosamine
Residues detected (pmol)
Note. Values in parentheses are for a second trial. Human leukocyte interferon produced in E. coZi were included as controls. n Average Phe composition for the leukocyte interferons was used. b Phe composition of br was used.
9.6 12.5 10.0 9.0
10.2619.5 9.6 17.0 11.8
Phe residues/ molecule
Molecular weight
Mel% Phe
IFN-(Y species
I
GLYCOSYLATIONOF INTERFERONS
TABLE
G-
z
g
8
F-
GLYCOSYLATED
HUMAN
teresting to determine if these properties of delta result from the presence of sugar residues. Of the KG-l-derived interferons tested, only bi and c1 were found to contain amino sugar residues. Unlike their CML counterparts, these interferons possessed larger amounts of galactosamine in addition to glucosamine. The c1 species appears to contain about two molecules of galactosamine and possibly one of glucosamine per molecule of protein. The most heavily glycosylated of all leukocyte interferon species tested was bi, which has about two residues of glucosamine and five of galactosamine per molecule of protein. However, the bi species migrated as two bands on SDS-polyacrylamide gel electrophoresis, corresponding to molecular weight values of 19,090 and 17,699. It is likely that the lower-molecular-weight species represents contaminating be (18), which was found to contain no amino sugars in this study. Therefore, the values in Table I probably reflect smaller than actual amounts for the sugar content of bl. An unexpected result was the lack of detectable amino sugar content in most of the higher-molecularweight species, particularly in betas (17), gammrq, and ba. The amount of protein hydrolyzed was well in excess of that required to reliably detect 1 mol amino sugar/m01 protein. It seems more likely that the apparent high molecular weights of these species on SDS-polyacrylamide gel electrophoresis reflect the anomalous migration of these species, rather than the presence of carbohydrate, although neutral sugars may also play a role. Due to a limited amount of material available, only about 8 to 9 pmol of species gammas and di were analyzed. However, one residue of amino sugar per molecule of protein would have been detected were it present. Since no significant trace of either sugar was observed on the chromatogram for either sample, it seems unlikely that these species are glycosylated. Insufficient material precluded testing of KG-l species dp. The “a” species derived from KG-1 cells was not tested either because these forms were not resolved and were present in only very small amounts (18). The presence of glucosamine in some of the species of the leukocyte interferons is consistent with several of the known polysaccharide structures of glycoproteins. These structures have been reviewed in detail by Wash and Bahl(24). The delta species could contain several asparagine-linked, high mannose-type structures containing only two residues of N-acetylglucosamine (GluNAe) each, although the presence of mannose cannot be confirmed by this method of analysis. Alternatively, its high glucosamine content could be explained by the presence of a single, complextype structure containing two to four extra GluNAc residues in the peripheral portion of the chain. The absence of galactosamine in the delta species would seem to rule out the presence of 0-glycosidic linkages to serine and threonine in these structures. The sit-
LEUKOCYTE
INTERFERONS
425
uation is quite different with respect to gammai and the interferons derived from the KG-l cell line. The presence of galactosamine in species gammai, bi, and ci would indicate an unusual structure or combination of structures for the carbohydrate chains in these molecules. To date, the only reported asparaginelinked polysaccharide structures containing N-acetylgalactosamine (GalNAc) on proteins have been for the pituitary glycoprotein hormones (24). In these structures, the GalNAc residues are located in the peripheral region of the carbohydrate chain. A notable exception exists for the a subunits of the human pituitary hormones leutropin, follotropin, and thyrotropin, in which a GalNAc residue is located in the core (25). The presence of galactosamine in gamma1 and two of the KG-l species may be significant in terms of the hormone-like properties of the interferons. The unusually high ratio of galactosamine to glucosamine in bl, however, may be due to the presence of additional serineor threonine-linked structures containing predominantly GalNAc residues. Examples of such structures are antifreeze glycoprotein, human IgA, cartilage keratan sulfate, and epiglycanin from TA-3 cells (24). Other proteins containing this type of structure but with GalNAc, N-acetylneuraminic acid, or neutral sugars in the periphery include the @ subunit of human chorionic gonadotropin, bovine fetuin, human erythrocyte membrane protein, and some canine and porcine submaxillary glycoproteins (24). Simple O-linked structures to serine or threonine would also explain the composition of species gammai and ci; how ‘er, the possibility of novel carbohydrate structures in these interferon species cannot be dismissed either. It is also possible that the presence of relatively large amounts of galactosamine in the KG1 interferon species represents an abnormality associated with the transformed nature of this tumor cell line. Further information must await a more detailed examination of the structures of these molecules. The results show that only two leukocyte interferon species derived from CML cells are glycosylated. These interferons, gammai and delta, both contained glucosamine, whereas gammai was found to possess one residue of galactosamine as well. The glycosylated species derived from KG-1 cells differed in that they contained greater amounts of galactosamine in addition to glucosamine. Variations in the distribution of the glycosylated species, particularly delta, in the interferon preparations may account for variability in the extents of glycosylation of crude leukocyte interferons. In our laboratory, delta was present in only some of the preparations, yet it was the most extensively glycosylated interferon from CML cells. In addition, the results indicate that the presence of carbohydrate alone cannot adequately explain why some of the interferon species (beta,, gammq, and hs) migrate with apparent molecular weights signif-
426
LABDON
icantly higher than expected. Furthermore, the amount of amino sugar detected in some species (e.g., delta) cannot entirely account for its high molecular weight. It remains to be seen to what extent longer polypeptide chains, neutral sugars, or other residues play a part in the molecular weight or structure of these species.
ET
11.
12. 13.
REFERENCES 14. 1. KNIGHT, E., JR. (1976) Proc. Natl. Acad. Sci USA 73,520-523. 2. TAN, Y. H., BARAKAT, F., BERTHOLD, W., SMITHJOHANNSEN, H., AND TAN, C. (1979) J. Bill Chew. 254, 8067-8073. 3. DAVEY, M. W., HUANG, J. W., SULKOWSKI, E., AND CARTER, W. A. (1974) J. Bid Chem. 249,63546355. 4. FRIESEN, H.-J., STEIN, S., EVINGER, M., FAMILLETTI, P. C., MOSCHERA, J., MEIENHOFER, J., SHIVELY, J., AND PESTKA, S. (1981) Arch. Biochem, Bie phys. 206, 432-450. 5. TANIGUCHI, T., OHNO, S., FUJI-KURIYAMA, Y., AND MURAMATSU, M. (1980) Gene 10.11-15. 6. DERYNCK, R., CONTENT, J., DE CLERCQ, E., VOLCKAERT, G., TAVERNIER, J., DEVOS, R., AND FIERS, W. (1980) Nature (London) 285,542-547. 7. HOUGHTON, M., STEWART, A. G., DOEL, S. M., EMTAGE, J. S., EATON, M. A. W., SM~~H, J. C., PATEL, T. P., LEWIS, H. M., PORTER, A. G., BIRCH, J. R., CARTWRIGHT, T., AND CAREY, N. H. (1980) Nucl. Acio!s Res. 8, 1913-1931. 8. GOEDDEL, D. V., SHEPARD, H. M., YELVERTON, E., LEUNG, D., CREA, R., SLOMA, A., AND PESTKA, S. (1980) Nucl. Acids Res. 8, 4057-4074. 9. JANKOWSKI, W. J., DAVEY, M. W., O’MALLEY, J. A., SULKOWSKI, E., AND CARTER, W. A. (1975) J. Viral. 16, 1124-1130. 10. GRAY, P. W., LEUNG, D. W., PENNICA, D., YELVERTON, D., NAJARIAN, R., SIMONSEN, C. C., DERYNCK, T., SHERWOOD, P. J., WALLACE, D. M.,
15. 16.
17.
18. 19. 20.
21.
22.
23.
24. 25.
AL. BERGER, S. L., LEVINSON, A. D., AND GOEDDEL, D. V. (1982) Nature @vm!un) 295, 503-508. DEVOS, R., CHEROUTRE, H., TAYA, Y., DEGRAVE, W., VAN HEU~ERS~~N, H., AND FIERS, W. (1982) NULL Acids Res. 10, 24874501. BRAUDE, I. In Proceedings of the FDA Workshop on Interferon, in press. BOSE, S., GURARI-ROTMAN, D., RUEGG, U. T., CORLEY, L., AND ANFINSEN, C. (1976) J. Biol. Chem. 251, 1659-1662. KELKER, H. C., YIP, Y. K., ANDERSON, P., AND VILCEK, J. (1983) J. Bid Chem 258.8010-8013. PESTKA, S. (1983) Arch Biochem. Biophys. 221, l37. PESTKA, S., KELDER, B., FAMILLE~I, P. C., MosCHERA, J. A., CROWL, R., AND KEMPNER, E. S. (1983) J. Biol. Chem 258, 9706-9709. RUBINSTEIN, M., LEVY, W. P., MOSHERA, J. A., LAI, C.-Y., HERSHBERG, R. D., BARTLETT, R. T., AND PESTKA, S. (1981) Arch. Bi&ewz. Biuphys. 210, 307-31s. HOBBS, D. S., AND PESTKA, S. (1982) J. Bid Chew. 257, 4071-4076. ALLEN, G., AND FANTES, K. H. (1980) Nature (Lendon) 287,408-411. ZOON, K. C. (1981) in Methods in Enzymology (Pestka, S., ed.), Vol. 78, pp. 457-464, Academic Press, New York. LEVY, W. P., RUBINSTEIN, M., SHIVELY, J., DEL VALLE, U., LAI, C.-Y., MOSCHERA, J., BRINK, L., GERBER, L., STEIN, S., AND PESTKA, S. (1981) Proc. Natl. Acad Sci. USA 78, 6186-6190. STEIN, S., AND MOSCHERA, J. (1981) in Methods in Enzymology (S. Pestka, ed.), Vol. 79, pp. 716, Academic Press, New York. KILPATRICK, D. L., GIBSON, K. D., AND JONES, B. N. (1983) Arch. Biochem. Biophys. 224,402404. WAGH, V. W., AND BAHL, 0. M. (1981) CRC &it. Rev. Biochem. 10, 307-377. HARA, K., RATHNAM, P., AND SAXENA, B. B. (1977) J. Biol. Chem 253, 1582-1591.