BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 1 024-I 030
THE INFLUENCE OF FLANKING SEQUENCES ON O - G L Y C O S Y L A T I O N
Brian O'Connell, Lawrence A. Tabak*, and Navayanan Ramasubbu 1
Departments of Dental Research and Biochemistry, School of Medicine and Dentistry, University
of Rochester, 601 Elmwood Ave., Box 611, Rochester, NY 14642
1Department of Oral Biology, School of Dental Medicine, State University of New York at
Buffalo, Buffalo, NY 14214
Received September 16, 1991
SUMMARY: The influence of flanking sequences on O-glycosylation of serine and threonine residues was explored by comparison of known acceptor sites. Positions -6, -1 and +3 relative to the site were identified as particularly significant. To test the hypothesis that O-glycosylation could be affected by amino acid sequence, a series of test peptides was made containing substitutions at the sensitive positions. In vitro glycosylation of the peptides confirmed that the acceptor status of threonine was markedly influenced by the residues present at positions -6, -1 and +3. Circular dichroism indicated that peptides which had random structure were glycosylated, except when they contained a charged residue at position -1. ® 1991 Academic Press, Inc.
The addition of O-linked carbohydrates to proteins is common among eukaryotes and is
responsible for a wide range of biological activities (1,2,3). The factors that influence O-
glycosylation of particular serine and threonine residues by GalNAc transferase are not well
defined. The comparison of amino acid sequences flanking GalNAc-modified residues has failed
to uncover a consensus sequence for O-glycosylation (4,5). An alternative strategy has been to
directly assess the ability of various synthetic peptides to be glycosylated in vitro (6,7,8,9).
However, most of the peptides tested in vitro contain a triprolyl motif C-terminal to threonine,
which has been found in only a few glycoproteins (10,11). It has been suggested that
glycosylation by GalNAc transferase is influenced by the local substrate conformation, rather than
a particular amino acid sequence (4,12,13,14). The importance of secondary structure for O-
glycosylation is consistent with the sequence heterogeneity about glycosylated sites and the known
structural information, though there is little direct experimental evidence to support the observation.
In this report, we examine the sequences flanking known glycosylated and non-
glycosylated serine and threonine residues. Comparison of these two data sets indicates that
certain groups of amino acid residues influence the ability of nearby hydroxyamino acids to be
Vol. 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
glycosylated. Synthetic peptides were made to test this hypothesis, and their ability to act as
substrates for GalNAc transferase was assayed by an in vitro glycosylation reaction. CD
spectroscopy of the test peptides suggests that the acceptor status of peptides is related to
secondary structure and amino acid sequence.
MATERIALS AND METHODS
Preparation and analysis of the database of O-glycosylated sites--O-glycosylated protein sequences were collected from a search of the literature and of the GenBank and NBRF databases (15). We only included glycosylated sites where GalNAc was unambiguously assigned to a specific serine or threonine residue either by direct sequencing or compositional analysis. The unmodified serine and threonine residues of the same proteins were used to form a set of non- glycosylated sites. At least six residues of sequence on either side of the sites were required, so that windows of 13 residues were formed. The frequency of each amino acid residue was calculated for each position from -6 to +6 relative to the site. Amino acid residues were divided into 3 groups according to their hydrophobicity (16). The odds ratio was calculated for each amino acid group at each position in the window. Odds ratio is an index of the probability that a pattern of amino acids is related to glycosylation of serine and threonine:
Odds Ratio = (flea of Pattern at ~lvcos sites)x(freo of Dattern not at non-~lvcos sites) (freq of pattern not at glycos sites)x(freq of pattern at non-glycos sites)
Where the term 'pattern' refers to the occurrence of a particular amino acid group at a given position. The data were then filtered so that odds ratios greater than 0.5 and less than 2 were excluded.
In vitro glycosylation assays--Peptides were purchased from Multiple Peptide Systems (San Diego, CA) and purified by reverse phase HPLC. The composition of each peptide was verified by mass spectrometry and amino acid analysis. Prior to use, the peptides were lyophilized twice and dissolved in 50 mM ammonium hydroxide at 20 nmol/~tl.
Each 50 ~tl assay contained: 125 mM Tris, pH 7.1, 0.5% Triton X-100, 10 mM MnC12, 10 nCi UDP-[14C]GalNAc (55 mCi/mmol) and 100 nmol test peptide. The source of GalNAc transferase was 0.7-1 ktl human colostrum. The colostrum was collected two days before parturition and delipidated by repeated centrifugation at 16000xg. The protein concentration of the colostrum was 24 mg/ml (17). The reactions were incubated at 37 ° for 2 h after which they were immediately frozen at -20oc.
Glycosylated peptides were separated from unincorporated UDP-[14C] GalNAc on a column (6.6 x 930 mm) of BioGel P2, using 0.1% TFA and a flow rate of 1.5 mls/min. Fractions were collected at 1 min intervals and cpm determined by liquid scintillation spectrophotometry. The UDP-[14C]GalNAc incorporated in the peptides was calculated as a percentage of the total counts added.
To preclude the possibility that impurities in the peptides could interfere with the incorporation of the UDP-[14C]GalNAc, peptides #1, 2, 3 and the serine variant of #1 were assayed in the presence of 240 ~tM UDP-GalNAc. The addition of excess unlabeled UDP-GalNAc did not affect the acceptor status of the peptides.
Circular dichroism--All spectra were recorded on a Jasco 600A spectropolarimeter interfaced to an IBM PS30 computer at room temperature (20 ° C) with a 2 s time constant and a scan rate of 10 nm/min. The spectra were obtained over the 185-250 nm range using 0.1 cm cylindrical cuvettes. The instrument was calibrated using 0.06% (w/v) of aqueous ammonium (+)-10- camphorsulfonate.
1025
Vol . 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH C O M M U N I C A T I O N S
R E S U L T S A N D D I S C U S S I O N
The database we developed consisted of 16 glycoproteins, having a total of 83 glycosylated
sites (11, 18-32). For comparison, the 662 non-glycosylated serine and threonine residues of the
proteins in the database were used. The data for serine and threonine are combined, since no
difference was seen between them.
The frequency of amino acid residues differs around glycosylated and non-glycosylated
sites (Fig. 1). Around glycosylated sites, proline is especially enriched at positions -3 (22%), -1
(28%) and +3 (35%)--it is greater than 10% at all other positions. Alanine is particularly abundant
from -4 to +2 and, along with glycine, is frequently found at -6. Glycosylated serine and
threonine residues occur frequently in proximity to glycosylated sites, showing the tendency for O-
linked sugars to be added in clusters. Since clustered glycosylated sites represent the majority of
our database (71 out of 83), the analysis will be biased towards this pattern, though it is possible
that there are different requirements for single glycosylation sites.
Odds ratios were calculated to relate the probability of glycosylation to the frequency of
amino acids flanking a site (Table 1). Residues were divided into groups in order to determine if a
set o f residues is associated with O-glycosylation. Initially, residues were grouped according to
hydrophobicity (16), surface accessibility (33), radius of gyration (34) and ability to form ~l-turns
(35). Grouping the residues on the basis of hydrophobicity yielded the greatest contrast between
glycosylated and non-glycosylated sites: Group 1 - Ile, Val, Leu, Phe, Cys, Met, Ala, Group 2 -
Gly \ ~ \ A N N N \ \ N \ N " X ~ , , , ~ \ \ \ . \~ .~ 'x A N N \ ~ ' \ N \
set ,'\~\\'\\\k,\~x\\~'\ " 54 ~'\\'\'\\,\\\'\~,,\\\'\~ Trp Tyr
His Glu ~ \ \~N; \ \ \ ,N 'q , "
,~'Q .'k"< .\N\\ \ \\\ Ash ,.\\ Ly~ k ~
Ser ° Thr °
45 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Position relative to S s r / T h r Position relative to S e r / T h r
Fie. 1. Frequency distribution of amino acids flanking glycosylated and non-glycosylated serine and threonine residues. The frequency of each amino acid residue is expressed in the plot as a percentage for positions -6 to +6, relative to a hydroxyamino acid at position 0. The shading at each location indicates the percentage frequency of a residue. Amino acids occurring at less than 5% are unshaded. Ser* and Thr* are the glycosylated form of these residues. "Blank" refers to the absence of any amino acid, i.e. the site occurs near the extremity of a protein.
1026
Vol. 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
Table 1. Relationship of amino acid type and position to glycosylation of serine and threonine. Amino acids are grouped according to their hydrophobicity (16); Group I - Ile, Val, Leu, Phe, Cys, Met, Ala; Group 2 - Gly, Thr, Thr*, Ser, Ser*, Trp, Tyr, Pro; Group 3 - His, Glu, Gin, Asp, Asn, Lys, Arg. Odds ratios >1 denote patterns that are associated more with glycosylated sites than non-glycosylated sites. Odds ratios <1 suggest patterns that are more associated with non-glycosylated sites.
AMINO ACID GROUP AND POSITION RELATIVE TO SER/]'HR
FREQUENCY AT FREQUENCY ODDS GLYCOSYLATED AT NON- RATIO
S I T E S GLYCOSYLATED SrI'ES
(n=83) t ~ )
-6-5-4-3-2-1 * 1 2 3 4 5 6
1
1
17 246 0.436
15 230 0.414
1 16 231 0.446
41 211 2.087
45 219 2.395
49 218 2.935
2 44 217 2.314
2 56 207 4.559
13 209 0.403
7 174 0.258
3 8 204 0.239
However, this observation may be due to the fact that certain residues were fortuitously grouped
together, rather than suggesting that glycosylation depends on the hydrophobicity of residues, per
se. For example, glycosylated serines and threonines were included with the unmodified residues,
even though they are several-fold less hydrophobic (36). Since we do not know the order in
which sites are glycosylated, we cannot separate the modified from unmodified residues. Also, if
we had used the hydrophobicity scale of Black and Mould, the odds ratios would have been more
extreme as alanine would appear in the same group as proline (36).
In general, group 1 and group 3 residues were associated with non-glycosylated sites,
while group 2 residues correlated with glycosylation (Table 1). Position +3 had the highest and
the lowest odds ratios caused by group 2 and group 3 residues, respectively. The same
relationship was seen at positions -1 and -6, though to a lesser degree. Since there was clear
evidence of a positive and negative relationship between the type of amino acid at positions -6, -1
and +3 and the glycosylation status, it was decided to study these positions experimentally.
The importance of amino acid residues at positions -6, -1 and +3 was tested by
synthesizing a series of peptides that contained a single threonine (Table 2). The parent peptide is
based on a sequence from human Von Willebrand factor and has a sequence typical of glycosylated
sites (Table 2, #1) (18). Single substitutions were made in the peptide at positions -6, -1 and +3
relative to threonine so that amino acids of each group were represented.
The ability of each peptide to serve as an acceptor for O-linked GalNAc is listed in Table 2.
In general, the test peptides behaved as predicted from the odds ratio analysis. While peptide #1
1027
Vol . 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
Table 2. The test peptides were used in in vitro glycosylation assays to determine their effectiveness as substrates for O-glycosylation. Alterations in the parent peptide sequence are boxed and all peptides were acetylated at the N-terminus. Values given are the mean of duplicate assays. The background level of glycosylation (no test peptide added) was 1.9%, and this was subtracted from the value obtained for each test peptide.
proved to be an effective substrate for O-glycosylation, the substitution of proline (group 2) at
position +3 by isoleucine (group 1), arginine (group 3) or glutamic acid (group 3) eliminated its
ability to act as a substrate (peptides #8, #9 and #10). At position -1, the substitution of valine
(group 1) by proline (group 2) gave a peptide that was still a substrate for O-glycosylation (peptide
#5). Substitution by glutamic acid or arginine at -1 effectively abolished the ability of the peptide
to O-glycosylate (peptides #6 and #7). When proline at -6 was replaced by isoleucine, the peptide
did not glycosylate (peptide #2). Unexpectedly, alteration of the -6 position to arginine or glutamic
acid did not diminish the peptide's ability as an acceptor (peptides #3 and #4). It is possible that
residues at the N-terminus of a small peptide do not exert the same influence in the context of a
larger protein. Peptide #11--with glutamic acid at the three sensitive positions--was not a good
acceptor. Though we did not observe a difference between the known flanking sequences of
glycosylated serine and threonine residues, a peptide similar to #1, but containing serine instead of
threonine, failed to be glycosylated. This is consistent with previous observations and suggests
that there may be a separate GalNAc transferase for serine which is unstable or that there are
specific flanking sequence requirements for serine which were not met (9). Though a wide range
of sequences are compatible with the O-glycosylation of hydroxyamino acids, the process is
sensitive to single amino acid changes in the substrate.
The relationship between the secondary structure of our test peptides and their ability to O-
glycosylate was explored by CD spectroscopy. When spectra were acquired in methanol, we
observed a close correlation between peptide structure and acceptor status (Fig. 2). The pepfides
that were good acceptors all had spectra characteristic of random coil and turn structure (peptides
#1, #3, #4, #5). Eckhardt et al. have suggested that CD spectra with "random coil" characteristics
1028
Vol. 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
1.0
..-. o
-0.25 I I I I I f
200 210 220 230 240 250 Wavelength (nm)
Fig. 2. Circular dichroism spectra of peptides #2, 3, 6 and 10 are shown to represent the results found for the 11 test peptides (see text). Spectra were obtained for the test peptides in methanol at a concentration of approximately 100 ~tg/ml. Peptide spectra did not differ from one another in water, trifluoroethanol or 125 mM Tris, pH 7.1.
actually represent 13-turn and aperiodic structure with some antiparalle1 [3-sheet (14). Conversdy,
most of the peptides that failed to glycosylate appeared to adopt a more ordered structure; the CD
spectra of peptides #2 and #8 exhibited significant extended conformation, whereas the peptides
#9, #10 and #11 showed an increase in helicity. Peptides #6 and #7 also had spectra that are
composed of random coil and turn structure but did not accept GalNAc. We suggest that this
inhibition of glycosylation is due to the local effect of the charged residues at position -1.
Our data suggest that the size or charge of residues near a site may have effects on O-
glycosylation that supersede an otherwise favorable structural environment. NMR studies of
glycosylated peptides have indicated that the orientation of GalNAc to the peptide backbone is
dependent on the size of the group N-terminal to the hydroxyamino acid (37). The effect of
charged residues on N-glycosylation has been reported (38). Peptides containing aspartic acid at
position +1 do not serve as substrates for N-glycosylation, even when they have the "correct"
conformation. Similarly, an acidic residue at position +1 prevents O-mannosylation ofpeptides in
vitro (39). It appears that the substrate specificity of GalNAc transferase differs from that of O-
GlcNAc transferase whose sites of glycosylation have an adjacent acidic residue (40).
Given the experimental evidence that O-glycosylation is influenced by the flanking amino
acid sequence and secondary structure, it will be necessary to evaluate the importance of other
positions by systematic mutagenesis.
Acknowledgments: This work was supported in part by National Institutes of Health Grants DE-
08511, DE-08108 (L.A.T.) and DE-00159. B.O.C. is supported by Dentist-Scientist Award K16
1029
Vol. 180, No. 2, 1991 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
DE-00159; these studies constitute work toward fulfillment of the Ph.D. degree. The authors
would like to thank R. Raubertas for his advice on statistical aspects of this work. We
acknowledge M.J. Levine for use of the CD facility which is supported in part by DE-08240. We
thank B. VanWuyckhuyse for his technical assistance and P. Noonan for preparing the
manuscript.
REFERENCES
1. Tabak, L.A., Levine, M.J., Mandel, I.D., and EUison, S.A. (1982) J. Oral PathoL 11, 1-17. 2. Jentoft, N. (1990) Trends Biochem. Sci. 15, 291-294. 3. Gallagher, J.T. and Corfield, A.P. (1978) Trends Biochem. Sci. 3, 38-41. 4. Aubert, J.P., Biserte, G., and Loucheux-Lefebvre, M.H. (1976) Arch. Biochem.
Biophys. 175, 410-418. 5. Wilson, I.B.H., Gavel, Y., and yon Heijne, G. (1991) Biochem. J. 275, 529-534. 6. Hagopian, A., Westall, F.C., Whitehead, J.S., and Eylar, E.H. (1971) J. Biol. Chem.
246, 2519-2523. 7. Young, J.D., Tsuchiya, D., Sandlin, D.E., and Hotroyde, M.J. (1979) Biochemistry 18,
4444-4448. 8. Briand, J.P., Andrews, S.P., Jr., Cahill, E., Conway, N.A., and Young, J.D. (1981) J.
10. Smyth, D.S. and Utsumi, S. (1967) Nature 216, 332-335. 11. Oppenheim, F.G., Offner, G.D., and Troxler, R.F. (1985) J. Biol. Chem. 260,10671-10679. 12. Hill, H.D., Jr., Schwyzer, M., Steinman, H.M., and Hill, R.L. (1977) J. Biol. Chem.
252, 3799-3804. 13. Shogren, R., Gerken, T.A., and Jentofl, N. (1989) Biochemistry 28, 5525-5536. 14. Eckhardt, A.E., Timpte, C.S., Abemethy, J.L., Toumadje, A., Johnson, W.C., Jr., and
Hill, R.L. (1987) J. Biol. Chem. 262, 11339-11344. 15. Devereux, J., Haeberli, P., and Smithies, O. (1984) Nuc. Acids Res. 12, 387-395. 16. Kyte, J. and Doolitfle, R.F. (1982) J. Mol. Biol. 157, 105-132. 17. Lowry, O.H., Rosebrough, N.J., Farr, A.L., and Randall, R.J. (1951) J. Biol. Chem.
193, 265-275. 18. Titani, K., Kumar, S., Takio, K., Ericsson, L.H., Wade, R.D., Ashida, K., Walsh,
K.A., Chopek, M.W., Sadler, J.E., and Fujikawa, K. (1986) Biochemistry 25, 3171-3184. 19. Tornita, M. and Marchesi, V.T. (1975) Proc. Nat. Acad. Sci. 72, 2964-2968. 20. Putnam, F.W., Liu, Y.-S. V., and Low, T.L.K. (1979) J. Biol. Chem. 254, 2865-2874. 21. Kessler, M.J., Mise, T., Ghai, R.D., and Bahl, O.P. (1979) J. Biol. Chem. 254, 7909-7914. 22. Robinson, E.A. and Appella, E. (1979) J. Biol. Chem. 254, 11418-11430. 23. Honma, K., Tomita, M., and Hamada, A. (1980) J. Biochem., 88, 1679-1691. 24. Klapper, D.G., Goodfriend, L., and Capra, J.D. (1980) Biochemistry 19, 5729-5734. 25. Seidah, N.G. and Chr6tien, M. (1981) Proc. Nat. Acad. Sci. 78, 4236-4240. 26. Putnam, F.W., Takahashi, N., Tetaert, D., Debuire, B., and Lin, L.-C. (1981) Proc. Nat.
Acad. Sci. 78, 6168-6172. 27. Murayama, J.-I., Tomita, M., and Hamada, A. (1982) J. Membr. Biol. 64, 205-215. 28. Murayama, J.-I., Yamashita, T., Tomita, M., and Hamada, A. (1983) Biochim. Biophys.
Acta 742, 477-483. 29. Robb, R.J., Kumy, R.M., Panico, M., Morris, H.R., and Chowdhry, V. (1984) Proc.
Nat. Acad. Sci. 81, 6486-6490. 30. Lai, P.-H., Everett, R., Wang, F.-F., Arakawa, T., and Goldwasser, E. (1986) J. Biol.
Chem. 261, 3116-3121. 31. Wernette-Hammond, M.E., Lauer, S.J., Corsini, A., Walker, D., Taylor, J.M., and Rail,
S.C., Jr., (1989) J. Biol. Chem. 264, 9094-9101. 32. Yet, M.-G., Chin, C.C.O., and Wold, F. (1990) FASEB J. 4, Abstr. 3542, A2306. 33, Chothia, C. (1976) J. Mol. Biol. 105, 1-14. 34. Levitt, M. (1976) J. Mol. Biol. 104, 59-107. 35. Chou, P.Y. and Fasman, G.D. (1978) Ann. Rev. Biochem. 47, 251-276. 36. Black, S.D. and Mould, D.R. (1991) Analyt. Biochem. 193, 72-82. 37. Maeji, N.J., Inoue, Y., and Chfij6, R. (1987) Biopolymers 26, 1753-1767. 38. Imperiali, B. and Shannon, K.L. (1991) Biochemistry 30, 4374-4380. 39. Strahl-Bolsinger, S. and Tanner, W. (1991) Eur. J. Biochem. 196, 185-190. 40. Haltiwanger, R.S., Holt, G.D., and Hart, G.W. (1990) J. Biol. Chem. 265, 2563-2568.
