| Glycobiology | Pages |
Calf thymus high mobility group proteins are nonenzymatically glycated but not significantly glycosylated
Introduction
Results
Discussion
Materials and methods
Acknowledgments
Abbreviations
References
Calf thymus high mobility group proteins are nonenzymatically glycated but not significantly glycosylated
Over the past decade, there have been many reports suggesting the presence of complex carbohydrates on nuclear and cytoplasmic proteins in mammalian cells. Some of the most often cited of these reports deal with the glycosylation of the high mobility group (HMG) proteins. These are relatively abundant chromosomal proteins that are known to be associated with nucleosomes and actively transcribed regions of chromatin. The original report describing HMG protein glycosylation presented several lines of evidence suggesting that these proteins are glycosylated, including carbohydrate compositional analysis and periodic-acid Schiff staining. We have attempted to repeat these observations with more highly purified protein than was utilized in the original study. Using carbohydrate compositional analysis performed by high pH anion exchange chromatography coupled to pulsed-amperometric detection, we saw no evidence for significant glycosylation of these proteins. In addition, we found no evidence for the presence of O-GlcNAc, a well known form of nuclear glycosylation. The HMG proteins did react with periodate, suggesting the presence of a modification containing cis-diols on the protein. Several tryptic peptides isolated from HMG 14 and 17 which retained the periodate reactivity had in common lysine residues, suggesting a potential modification of the [epsis]-amino groups of lysines such as nonenzymatic glycation. Western blot analysis of the HMG proteins using anti-advanced glycation endproducts (AGE) antibodies confirmed the presence of glycation products on the HMG proteins. Key words: nuclear glycosylation/high mobility group proteins/ periodate oxidation/glycation
Introduction
Numerous reports suggesting the glycosylation of proteins in the nucleus and cytoplasm of mammalian cells have appeared over the past two decades (Hart et al., 1989). The evidence presented in these reports was based on observations made using a wide variety of techniques, including binding of lectins to nuclear proteins, carbohydrate compositional analysis of nuclear proteins, and metabolic labeling of nuclear and cytoplasmic proteins with radioactive sugars (Hart et al., 1989). In spite of the large number of initial reports on the glycosylation of nuclear proteins, very few have been confirmed by more rigorous structural analysis. To date, only three forms of glycosylation have been confirmed and repeatedly shown to exist on nuclear and cytoplasmic proteins in mammalian cells. The best characterized of these is the addition of N-acetylglucosamine to the hydroxyls of serine or threonine residues on proteins (O-GlcNAc; for recent reviews, see Hart et al., 1996; Haltiwanger et al., 1997). Well over a hundred cytoplasmic and nuclear proteins containing this modification have been identified (Hart et al., 1996; Haltiwanger et al., 1997), and enzymes capable of addition and removal of the sugar have been purified and characterized (Haltiwanger et al., 1992; Dong and Hart, 1994; Kreppel et al., 1997; Lubas et al., 1997). The other well characterized forms of cytoplasmic glycosylation in mammalian cells both occur on single proteins: glucose-1-phosphate on phosphoglucomutase (Veyna et al., 1994) and glucose-O-tyrosine on the glycogen primer glycogenin (Smythe et al., 1988). Novel cytoplasmic enzymes responsible for the addition of the sugars to both of these proteins have been identified as well (Pitcher et al., 1988; Srisomsap et al., 1988).
Several other forms of nuclear glycosylation have been reported which have not yet been carefully followed up with more rigorous techniques. One of the most often cited examples of these is the high mobility group (HMG) proteins (Reeves et al., 1981; Reeves and Chang, 1983; Elton and Reeves, 1986). The HMG proteins are among the most abundant non-histone proteins found in the nuclei of eukaryotic cells (Bustin et al., 1990). A protein is defined as an HMG based on extractability from chromatin with 0.35 M NaCl, solubility in 2-5% perchloric acid, and high content of positively charged amino acids (Bustin et al., 1990). More than 10 HMG proteins have been reported in mammalian cells, and they are commonly divided into three groups: the HMG-1/2 family, the HMG-14/17 family, and the HMG-I family. Although a clear function for these proteins is not known, they are known to interact with DNA and are often present at regions of active transcription (Bustin et al., 1990). The HMG proteins have been reported to possess numerous post-translational modifications including phosphorylation (Sun et al., 1980; Ferranti et al., 1992), acetylation (Sterner et al., 1979), ADP-ribosylation (Reeves et al., 1981), methylation (Boffa et al., 1979), and glycosylation (Reeves et al., 1981; Reeves and Chang, 1983; Elton and Reeves, 1986). The reports of glycosylation suggested the presence of fairly complex structures containing N-acetylglucosamine, mannose, galactose, glucose, and fucose. The carbohydrate was proposed to be N-linked to the protein based on resistance of the linkage to dilute alkali (Reeves et al., 1981). These observations raised the possibility of a novel cellular pathway capable of modifying nuclear proteins with N-glycans.
Because of the biological significance of the putative N-glycosylation of nuclear proteins, we have reexamined the reported glycosylation of the HMG proteins. Here we report that we could detect little or no glycosylation of HMGs 1, 2, 14, or 17 from calf thymus. In contrast, we did observe the presence of a significant amount of periodate-reactive material on the HMG proteins. Several lines of evidence suggest that the basis of the periodate reactivity is the nonenzymatic modification of [epsis]-amino groups of lysine residues with sugars through the process known as glycation (Bucala and Cerami, 1992; Brownlee, 1995; Drickamer, 1996).
Results
Purification of the HMG proteins
The original reports describing glycosylation of the HMG proteins utilized two different techniques for extraction of these proteins from calf thymus (0.35 M NaCl and 5% perchloric acid; Reeves et al., 1981; Reeves and Chang, 1983; Elton and Reeves, 1986). We utilized both procedures and found that the O.35 M NaCl extracts had higher levels of HMG 14. Thus, HMG proteins partially purified from the 0.35 M NaCl extracts of calf thymus nuclei were used for subsequent experiments. The HMG proteins were further purified by reverse phase HPLC on a C4-silica column (Figure 1A) as described by Elton and Reeves (Elton and Reeves, 1985a,b). Aliquots of fractions from the HPLC were analyzed by acid-urea gel electrophoresis (Figure 1B). The HMG proteins have characteristic mobilities on acid-urea gels which makes them easy to identify in these systems (Elton and Reeves, 1986). HMGs 14 and 17 (peaks labeled 'II" and 'I", respectively, in Figure 1A; see Figure 1B, lanes 20-27, for identification of the proteins migrating at those positions) were completely resolved from one another using this technique. The level of purity obtained utilizing the reverse phase HPLC appeared to be much better than that of the CM-Sephadex purified protein utilized in the original reports (Elton and Reeves, 1985a,b). The identity of HMG 14 and 17 was confirmed by Western blot analysis with HMG 14/17 specific antibodies (data not shown) and by gas phase sequencing of tryptic peptides generated from each protein (see below). HMGs 1 and 2 (eluting from 40-47 min) were not well resolved from each other or from histone H1 (Figure 1B, lanes 40-47). Thus, in all subsequent experiments, HMGs 1 and 2 were either analyzed together (with histone H1) or in the partially purified HMG fraction of the 0.35 M NaCl extract (see 'Before" lane, Figure 1B).
Figure
HMG proteins are not modified with O-GlcNAc
Since the best characterized form of glycosylation on nuclear and cytosolic proteins is O-GlcNAc (Hart et al., 1996; Haltiwanger et al., 1997), we decided to examine the HMG proteins for this modification. Although O-GlcNAc had not yet been discovered when the glycosylation of the HMG proteins was originally reported, the proteins were reported to be modified with GlcNAc (Reeves et al., 1981). We performed galactosyltransferase labeling on both the partially purified and HPLC purified samples of HMG proteins (Figure 2). Numerous proteins in the partially purified HMG preparation labeled with [3H]galactose, indicating the presence of terminal GlcNAc moieties on several minor, contaminating proteins (Figure 2A). Interestingly, no radioactivity was associated with the HMG proteins, even though they were clearly the most abundant proteins in the extract. The lack of radioactivity associated with the HMG proteins indicates a complete lack of terminal GlcNAc moieties on these proteins. This data also demonstrates that partially purified fractions of the HMG proteins are heavily contaminated with numerous glycoproteins. In contrast, no [3H]galactose was incorporated into HPLC purified HMG 14 or 17 (Figure 2B), indicating that the majority of the contaminating glycoproteins shown in Figure 2A were successfully purified away using this technique. Some [3H]galactose was incorporated into galactosyltransferase itself and a contaminating protein in the HMG 14 sample (see asterisk in Figure 2B), indicating that the enzyme was active during this radiolabeling. These results suggest that the HMG proteins are not modified with O-GlcNAc or any other structure containing a terminal, nonreducing GlcNAc moiety.
Figure
Carbohydrate compositional analysis shows no significant level of sugar on the HMG proteins
The most dramatic result of the previous reports suggesting glycosylation of the HMG proteins was the carbohydrate compositional analysis showing the presence of GlcNAc, galactose, mannose, glucose, and fucose on these proteins (Reeves et al., 1981). A later report suggested that several of the HMG proteins are modified with up to 5-8% carbohydrate by weight (Elton and Reeves, 1986). Using the highly purified HMG preparations shown above, we utilized a highly sensitive carbohydrate compositional analysis technique to re-examine the glycosylation of these proteins. As shown in Table 1, very little carbohydrate was detected on any of the HMG proteins examined. The reported values for each of the monosaccharides (the average of three separate determinations) was within the noise level of the technique. The highest level of any individual sugar was that of glucose, which was present at similar levels even in the control hydrolysis due to a contaminant present in the water. The level of glucose in each determination was highly variable, as seen in the high standard deviations reported for glucose in each sample. Thus, no significant conclusions can be drawn from the glucose data. Ovalbumin is a glycoprotein modified by approximately 5% carbohydrate by weight. Using the same procedures on this standard glycoprotein, we were able to obtain the expected values for the levels of GlcNAc and mannose (Kakehi and Honda, 1993). This demonstrates that the techniques used here can easily detect the glycosylation of a protein modified by as much as 5-8% carbohydrate by weight. In contrast, we could not detect significant glycosylation of HMG 1, 2, 14, or 17.
In addition to showing that the HMG proteins bear carbohydrate based on compositional analysis, several other techniques were used in previous reports to examine the glycosylation of the HMG proteins. For instance, the HMG proteins were shown to react with the fucose-specific lectin Ulex europeus (Reeves et al., 1981). We have shown that this lectin does interact with purified HMG 14, although we could not inhibit the interaction with 100 mM fucose, indicating that the binding is nonspecific (data not shown). We could not detect any interaction of the lectin with HMG 1, 2, or 17. Previous reports had also suggested that HMG 1, 2, 14, and 17 could be metabolically labeled in Murine Friend erythroleukemia cells with [3H]-fucose, -mannose, -glucosamine, and -galactose (Reeves et al., 1981; Elton and Reeves, 1986). We attempted to reproduce this metabolic labeling under identical conditions, but we could detect no significant radiolabeling of the HMG proteins with any of these sugars (data not shown).
The HMG proteins react with periodate
Another method used previously to detect carbohydrate on the HMG proteins was periodic acid-Schiff staining (Reeves et al., 1981). Periodate reacts with cis-diols, and carbohydrates are the major type of cis-diol containing species in biological systems. We have used a method for detecting periodate-oxidized sugars which is more sensitive than the original periodic acid-Schiff stain (Angel and Nilsson, 1990). In this method, biotin-hydrazide is used to react with the aldehyde groups generated by the periodate oxidation. Using this method, reactivity of positive controls such as fetuin were totally dependent on the presence of periodate (Figure 3A). Using partially purified HMG proteins, we could detect strong reactivity of HMGs 1 and 2 in the presence of periodate and reduced reactivity in its absence (Figure 3A). The reason for the reactivity in the absence of periodate was not clear. HMGs 14 and 17 in the same samples reacted very weakly. Analysis of HPLC purified HMGs 14 and 17 showed more clearly that reactivity with the biotin-hydrazide was dependent on the presence of periodate (Figure 3B). These results suggested that the HMG proteins are modified with cis-diol containing groups.
Figure Table
GalN
GlcN
Gal
Glc
Man
HMG 14
ND
ND
0.0170 (0.0170)
0.120 (0.170)
0.0900 (0.0700)
HMG 17
ND
ND
ND
0.0950 (0.0440)
0.0450 (0.0250)
HMG 1/2
ND
0.0439 (0.0123)
0.0268 (0.0174)
0.900 (0.584)
0.0506 (0.0145)
Ovalbumin
ND
8.12 (0.898)
0.450 (0.243)
0.653 (0.183)
6.33 (0.717)
Blanka
ND
ND
ND
0.560 (0.300)
ND
Since the compositional analysis shown above suggested the absence of most carbohydrates on these proteins, we examined other possible explanations for the periodate reactivity. The HMG proteins have been reported to be ADP-ribosylated (Giri et al., 1978; Reeves et al., 1981), and ribose would give a positive reaction with the periodate but would be destroyed in the conditions used in the carbohydrate compositional analysis. Thus, we examined the possibility that the periodate reactivity on HMG 14 and 17 was due to ADP-ribosylation. Treatment of proteins with neutral 0.5 M hydroxylamine removes ADP-ribose from three of the amino acids known to be modified in proteins (arginine, glutamic acid, aspartic acid; Payne et al., 1985). Hydroxylamine treatment of HMG 14 and 17 had no effect on the periodate-dependent labeling with biotin-hydrazide (Figure 3B), suggesting that the periodate sensitivity is not due to the presence of ADP-ribosylation. ADP-ribosylation has also been reported to occur on cysteine residues (Jacobson et al., 1990), but the HMG proteins do not contain any cysteine residues.
Isolation and characterization of peptides containing periodate-reactive material
In order to learn more about the nature of the periodate-reactive material on HMG 14 and/or 17, we decided to attempt to localize the modification along the polypeptide backbone. Tryptic peptides of purified HMG 14 and 17 were generated and separated by reverse-phase HPLC. (Figure 4A shows the data for HMG 17. HMG 14 gave comparable results.) Aliquots from HPLC fractions were assayed for the presence of periodate-reactive material using the 'dot-blot" assay described in Materials and methods (Figure 4B). Surprisingly, many of the peptides demonstrated periodate-sensitive reactivity with biotin-hydrazide. Two of these periodate-positive peptides ('17" and '18", Figure 4A,B) were further purified and identified by gas phase sequence analysis. Fraction 17 (originally eluting at 17 min) corresponded to a partial tryptic peptide from HMG 17: 16VKDEPQRR23. Fraction 18 (originally eluting at 18 min) corresponded to a second partial tryptic peptide: 47KGEKVPK53. A periodate-positive peptide was also isolated from HMG 14 and sequenced using similar techniques (23LSAKPAPAKVETK35; data not shown).
Figure
Since so many tryptic peptides reacted with periodate in the dot blot assay, we were concerned that the reactivity may be due to an unusual reaction with the unmodified peptides themselves. To test this idea, we had the peptides corresponding to Fractions 17 and 18 chemically synthesized. The peptides were separated by reverse phase HPLC as in Figure 4 and subjected to the same dot-blot assay to test for the presence of periodate reactive material. In neither case was there any reactivity in this assay, even though significantly more material was analyzed than in Figure 4 (data not shown). Thus, we concluded that the peptides isolated from HMG 14 and 17 must be modified with a periodate-reactive substance.
The HMG proteins are glycated
All of the periodate-sensitive peptides which we sequenced contained lysine, glutamic acid, and proline residues suggesting that a modification of one of these amino acids may be responsible for the periodate reactivity. Glycation is a modification of the [epsis]-amino group of lysines which would also be periodate reactive. It results from the nonenzymatic reaction of reducing sugars (such as glucose or glucose 6-phosphate) with primary amino groups and is known to occur on a number of proteins (Brownlee, 1995). Glycated proteins have been shown to interact tightly with immobilized boronic acid affinity columns and specifically elute with 0.2 M sorbitol (Takahashi et al., 1995). We examined the ability of the HMG proteins to bind to and elute from an immobilized boronic acid column (Figure 5). Although the majority of HMGs 1 and 2 flowed through the column, a small amount bound and specifically eluted with 0.2 M sorbitol. Based on densitometric scans of the stained gel, approximately 0.1% of the total HMG 1/2 eluted from the column with sorbitol. These results strongly suggest that a small portion of these proteins are glycated. To rule out the possibility that the fraction of HMGs 1 and 2 which eluted with sorbitol were nonspecifically bound to the column, the material which flowed through the column was re-applied and eluted. No HMG 1 or 2 could be detected in the sorbitol eluate from the reapplied sample, suggesting that the interaction with the column was specific (data not shown). In contrast to HMGs 1 and 2, the majority of HMGs 14 and 17 bound to the column. Unfortunately, neither HMG 14 nor 17 was eluted with sorbitol, or even after extensive washing with acidic buffers. Thus, for reasons which are not clear, it appears that HMGs 14 and 17 interact very tightly with the boronic acid column. Therefore, although conclusions about HMGs 14 and 17 cannot be made, these results strongly suggest that HMGs 1 and 2 are substoichiometrically glycated.
Figure
Several laboratories have recently generated antibodies which react with glycation products on proteins (Makita et al., 1992; Takahashi et al., 1995). We have examined the HMG proteins in a 0.35 M NaCl extract using one of these antibodies (Makita et al., 1992) by Western blot analysis. Interestingly, all of the HMG proteins specifically reacted with the anti-glycation antibody (Figure 6A). Incubation of the antibody with excess glycated BSA substantially reduced its reactivity with both the HMG proteins and the glycated BSA positive control, indicating that the reactivity of the HMG proteins with the antibody was due to glycation (Figure 6B). The specificity of the antibody was demonstrated by lack of reactivity with any of the molecular weight standards (Figure 6A). In addition, an equivalent amount of an isotype matched control monoclonal antibody did not react with the HMG proteins (data not shown). These data strongly suggest that HMGs 1, 2, 14, and 17 are all glycated.
Figure
Discussion
In this report we have carefully analyzed several of the high mobility group proteins (HMG 1, 2, 14, and 17) for the presence of carbohydrate. In contrast to what was originally reported by Reeves et al. (Reeves et al., 1981), we have not detected significant glycosylation of these proteins. Using a combination of techniques including highly sensitive carbohydrate compositional analysis, lectin blotting and metabolic labeling with radioactive sugars, we can find no significant levels of glycosylation on any of the HMG proteins. In contrast, HMG 1, 2, 14, and 17 all react with periodate, putatively a carbohydrate specific reagent. The periodate reactivity appears to be the result of glycation of [epsis]-amino groups on several lysines of the HMG proteins.
Glycation results from the nonenzymatic reaction of reducing sugars (such as glucose or glucose 6-phosphate) with free amino groups, usually the [epsis]-amino groups of lysine residues (Bucala and Cerami, 1992; Brownlee, 1995; Drickamer, 1996). The reaction proceeds through a reversible Schiff-base to stable, Amadori rearrangement products and finally to a heterogeneous group of protein modifications termed advanced glycation endproducts (AGEs; Bucala and Cerami, 1992; Brownlee, 1995; Drickamer, 1996). These modifications typically occur on long-lived proteins (e.g., collagens) and are more pronounced under conditions of increased glucose concentrations, such as in diabetes mellitus (Bucala and Cerami, 1992; Brownlee, 1995). Most of the work thus far on glycation has dealt with extracellular proteins, although several recent reports have demonstrated that histones are glycated (Sakurai and Tsuchiya, 1988; Gugliucci and Bendayan, 1995). The fact that the HMG proteins are glycated is consistent with these findings. Both the histones and the HMG proteins are long-lived and rich in lysines (HMG 14 is 21% lysine, HMG 17 is 25% lysine). Since both the histones and HMG proteins interact with DNA through lysine residues, it is likely that significant glycation of these proteins could affect their proper function. Thus, increased glycation could result in alterations in chromatin structure leading to changes in gene expression. Such a mechanism has been proposed to play a role in some of the pathologies associated with diabetes (Brownlee, 1995; Gugliucci and Bendayan, 1995).
The glycation of the HMG proteins explains all of our observations dealing with the periodate-sensitivity of these proteins. It provides a periodate-sensitive modification which occurs at substoichiometric levels at several sites along the polypeptide backbone. This explains why we observed numerous periodate-positive peptides after trypsin digestion (Figure 4). All of the peptides identified contained lysine residues, providing potential sites for glycation. It also explains why a small fraction of the HMG 1 and 2 specifically eluted from the boronic acid column with sorbitol (Figure 5).
Other than the periodate-reactivity, our results on the glycosylation of the HMG proteins are in conflict with the previous reports. There are many potential reasons for this conflict. The most likely one is that the proteins analyzed here were more highly purified than those analyzed previously. The reverse-phase HPLC purification of HMG 14 and 17 used here has been demonstrated to be superior to the CM-cellulose chromatography used in previous studies (Elton and Reeves, 1985a,b). The differences between highly purified HMG proteins and partially purified protein was made very clear by the galactosyltransferase analysis shown in Figure 2. The partially purified preparations, in which the HMG proteins were the major Coomassie staining bands, contained numerous proteins with terminal GlcNAc residues as evidenced by the galactosyltransferase labeling. In contrast, the HPLC purified HMG proteins had none of these contaminants.
In addition to the differences in purity, some of the techniques used by Reeves et al. (1981) were problematic. The original reports on glycosylation of these proteins suggested that the HMG proteins were modified with as much as 5-8% carbohydrate by weight (Elton and Reeves, 1986). These results were based on the anthrone reaction, a colorimetric assay for the presence of sugars (Reeves et al., 1981; Elton and Reeves, 1986). In fact, using several different colorimetric sugar assays, Reeves et al. (1981) showed that HMG 14 and 17 both appeared to be more heavily glycosylated than standard glycoproteins such as ovalbumin. Colorimetric assays for sugars are relatively insensitive and can give false positive reactions with high concentrations of proteins (Chaplin, 1986). We have quantified the level of sugar on these proteins using a highly sensitive and widely used method (Hardy and Townsend, 1994). As a positive control, we analyzed ovalbumin using this technique and found the expected large molar ratios of mannose and GlcNAc per mole of protein. Ovalbumin was selected because it was also the positive control used by Reeves et al. (1981) in their sugar analyses. In contrast, we found essentially background levels of carbohydrate on all of the HMG proteins examined. Based on these results, we conclude that the HMG proteins are not glycosylated and that the previous results were compromised by impurities in the samples and unreliable techniques.
Several other laboratories have also examined posttranslational modifications on HMG proteins in the past few years and failed to find significant glycosylation of these proteins. Ferranti et al. (1992) performed mass spectral analysis of the HMG Y and the only posttranslational modification found was phosphorylation. Chao et al. (1994) reported, using similar techniques to those used here, that they could not detect significant glycosylation on HMG 1 from chicken erythrocytes or calf thymus. Finally, Boumba et al. (1993) analyzed HMG 17 from porcine thymus by mass spectrometry and found it to be modified only with phosphate. In combination with our results, these studies all suggest that the high mobility group proteins are not glycosylated.
In order for a protein to be identified as a nuclear or cytoplasmic glycoprotein, it needs to meet two criteria. First, the presence of covalently associated carbohydrate by direct structural analysis needs to be demonstrated. Second, the protein must be demonstrated by rigorous techniques to be present in the nuclear or cytoplasmic spaces of the cell. O-GlcNAc modified proteins (Hart et al., 1996; Haltiwanger et al., 1997), phosphoglucomutase (Srisomsap et al., 1988; Veyna et al., 1994) and glycogenin (Pitcher et al., 1988; Smythe et al., 1988) all fulfill these criteria. Numerous other proteins have been reported to be nuclear or cytoplasmic glycoproteins even though both of these criteria have not been met. The HMG proteins are a prime example. They are well known to be nuclear proteins, but no direct structural analysis of their putative carbohydrates had been performed. Since the original reports of the glycosylation of HMG proteins were published (Reeves et al., 1981), they have been cited numerous times as an example of nuclear proteins modified with complex carbohydrate structures (Hart et al., 1989; Codogno et al., 1992; Felin et al., 1997). Our results (and those of the above mentioned reports) shed serious doubt on the conclusions that the HMG proteins are glycosylated. Thus, we believe that many of the reports of complex carbohydrates (e.g., classical N- or O-glycans) on nuclear or cytoplasmic proteins need to be seriously questioned and reexamined. A careful combination of structural analysis and subcellular localization must be performed on any potential candidate before it can be concluded that these types of modifications occur in nuclear and/or cytoplasmic compartments.
Materials and methods
Materials
Bovine milk galactosyltransferase, ovalbumin and glycated BSA were from Sigma. Galactosyltransferase was autogalactosylated prior to use as described (Roquemore et al., 1994). Calf thymus was obtained from Pel Freeze Biologicals, Inc. Rabbit anti-HMG 14/17 was generously provided by Dr. Michael Bustin (NIH, Bethesda, MD). The glycation-specific monoclonal antibody (Makita et al., 1992) and an isotype matched control antibody were generously provided by Dr. Richard Bucula (Picower Institute, Manhasset, NY). Horseradish peroxidase-conjugated goat anti-rabbit IgG was from Promega. UDP-[6-3H]-galactose (50-60 Ci/mmol) and En3Hance were from Dupont-NEN. The periodate labeling of proteins was performed using the Glycotrak Kit from Oxford Glycosystems and used as described by the manufacturer. The immobilized boronic acid column and trifluoroacetic acid (TFA) were from Pierce. All water used for HPLC was purified on a Milli Q system (Millipore). Peptides were synthesized by Quality Controlled Biochemicals (Hopkinton, MA). Peptide sequencing was performed at the Center for the Analysis of Macromolecules at SUNY-Stony Brook. All electrophoresis reagents were from Bio-Rad. Enhanced chemiluminescence reagents were from Amersham, Inc. All other reagents were of the highest quality available.
Purification of the HMG proteins
The HMG proteins were purified from calf thymus essentially as described (Elton and Reeves, 1985a,b). Two different extraction protocols were utilized. One involved extraction of calf thymus nuclei with 0.35 M NaCl (salt extraction) and the other extraction of the HMG proteins directly from the intact tissue with 5% perchloric acid (PCA extraction). The HMG proteins were further purified by reverse-phase HPLC on a C4-column (4.6 × 250 mm, 300 Å pore, 5 µm particles, Vydac) as described previously (Elton and Reeves, 1985a,b). The partially purified proteins from the salt extracted nuclei (1 mg) were chromatographed on a C4-column using a 0-60% acetonitrile linear gradient in 0.1% TFA over 70 min at 1 ml/min. Protein was followed by absorbance at 214 nm. Fractions (1 min) were collected for further analyses.
Carbohydrate compositional analysis
Carbohydrate compositional analysis was performed as described previously (Hardy and Townsend, 1994). HPLC purified HMG proteins (40-50 µg) were hydrolyzed in 2 M TFA for 4 h at 100°C. Ovalbumin (2 nmol) was also hydrolyzed as a positive control. Water blank samples were subjected to hydrolysis to attempt to control for the level of glucose found in the hydrolysates. Samples were then concentrated by evaporation in a Speed Vac (Savant) and resuspended in water. Samples were analyzed by high pH anion exchange chromatography with pulsed-amperometric detection (HPAEC-PAD) (Dionex, Inc.) as described previously (Hardy and Townsend, 1994). All separations were performed on a Carbopac PA-1 column (Dionex, Inc.). To each sample 500 pmol of 2-deoxyglucose was added (posthydrolysis) as an internal standard. The limit of sensitivity of our system is approximately 20 pmol for monosaccharides. Thus, with the amounts of proteins being analyzed, ratios of sugar to protein of less than 0.01 mol/mol were considered below the detection limit (Not Detected).
Periodate labeling of carbohydrates
Proteins were reacted first with periodate and then with biotin-hydrazide in solution, prior to electrophoresis, using the Glycotrak Kit from Oxford Glycosystems (as described by the manufacturer). A no periodate control was performed on each sample to detect proteins which react directly with the biotin-hydrazide. After electrophoresis on 12% SDS-PAGE, the proteins were transferred to nitrocellulose and probed with strepavidin-alkaline phosphatase to detect the biotinylated proteins. The alkaline phosphatase conjugate was detected using bromochloroindolyl phosphate/nitro blue tetrazolium detection as described (Harlow and Lane, 1988). To rule out modification of the proteins by ADP-ribosylation, the biotinylated proteins were treated with 0.5 M hydroxylamine in 0.1 M Tris-HCl, pH 7.0 (Payne et al., 1985) for 1 h at room temperature prior to electrophoresis.
Isolation of tryptic peptides from HMG proteins
HPLC purified HMG 14 and 17 were subjected to digestion with TPCK-trypsin (Sigma, 1:25 trypsin to protein) for 24 h at 37°C in sterile 0.2 M ammonium bicarbonate. TPCK-trypsin was added again and incubated for another 24 h. The resulting peptides were separated by reverse-phase HPLC on a C18-silica column (4.6 × 30 cm, 300 Å pore, 5 µm particles, Rainin Dynamax). A linear 0-60% acetonitrile gradient in 0.1% TFA over 60 min was used at a flow rate of 0.5 ml/min. Peptides were detected by absorbance at 214 nm, and 1 min fractions were collected.
Peptide Dot-Blot assay
Aliquots (5%) of the fractions from the HPLC were reacted with periodate and biotin-hydrazide as described above. The reactions were then spotted onto nitrocellulose membrane using a Dot-Blot apparatus (Bio-Rad). The biotinylated peptides were detected with horseradish peroxidase-conjugated streptavidin and ECL. Peptides which gave periodate-positive reactivity were subjected to a second round of HPLC as described above. Those peptides which gave clean, single peaks were sent for gas-phase sequence analysis.
Lectin blot
Samples containing either HMG 14 or 17 were electrophoresed on a 12% SDS-PAGE gel and transferred to nitrocellulose. The membrane was blocked with 1% BSA in TBS (10 mM Tris-HCl, pH 7.5, 0.15 M NaCl), for 1.5 h. After blocking, the membrane was incubated with horseradish peroxidase-conjugated Ulex europeus lectin (EY labs) at 0.002 mg/ml for 2 h. The membrane was then washed three times with TBS (10 mM Tris-HCl, pH 7.5, 0.15 M NaCl) for 15 min with rocking, and the peroxidase was detected using ECL. Specificity controls were performed by the addition of 100 mM fucose and glucose to the lectin blot during incubation with the membranes.
Immobilized boronic acid chromatography
Samples were chromatographed on the immobilized boronic acid column by the manufacturers instructions. Briefly, partially purified HMG proteins (100 µg) were loaded onto a 2 ml boronic acid column equilibrated in 0.25 M ammonium acetate, pH 8.5 (column buffer). The flow through was collected as a single fraction, and the column was then washed with column buffer until the absorbance at 230 nm was the same as buffer alone (~30 ml). Glycated proteins were specifically eluted with 10 ml of 0.2 M sorbitol in column buffer. Fractions (1 ml) were collected for the wash and eluate. The material in the Flow Through fraction was reapplied to the same column (after washing with 20 volumes of column buffer to remove the sorbitol). The column was then washed and eluted as before. Aliquots (10%) of fractions were analyzed by SDS-PAGE followed by staining with Coomassie blue. Estimates of the fraction of protein which bound to the column were obtained by densitometric scanning of the stained gel on a Bio-Rad Model GS-670 Imaging Densitometer.
Other methods
SDS-PAGE was performed as described by Laemmli (Laemmli, 1970). Fluorography was performed with En3Hance by the manufacturers instructions. Western blot analysis was performed as described previously (Harlow and Lane, 1988). All electrophoretic transfers were performed with a Millipore semidry transfer unit by the manufacturers instructions. Electrophoresis on acid-urea gels (Elton and Reeves, 1986), galactosyltransferase labelings (Roquemore et al., 1994) and protein assays (Schaffner and Weissmann, 1973) were performed as described previously.
Acknowledgments
We thank Dr. Deborah Brown, Mr. Scott Busby, Ms. Kathleen Grove, Mr. Daniel Moloney, Mr. Glenn Philipsberg and Mr. Richard Scartozzi for critical reading of the manuscript and helpful discussions. R.S.H. is the recipient of an American Cancer Society junior faculty research award. This work was supported by NIH Grant GM 48666 to R.S.H.
Abbreviations
O-GlcNAc, O-linked N-acetylglucosamine; HMG, high mobility group; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; HPLC, high-pressure liquid chromatography; AGE, advanced glycation endproducts.
References
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 27 Jan 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. J. Thornalley Glycation, receptor-mediated cell activation and vascular complications of diabetes Diabetes and Vascular Disease Research, May 1, 2004; 1(1): 21 - 22. [PDF]
|
|
|
|
 |
|
 C. M. West, H. van der Wel, and E. A. Gaucher Complex glycosylation of Skp1 in Dictyostelium: implications for the modification of other eukaryotic cytoplasmic and nuclear proteins Glycobiology, February 1, 2002; 12(2): 17R - 27R. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Teng-umnuay, H. R. Morris, A. Dell, M. Panico, T. Paxton, and C. M. West The Cytoplasmic F-box Binding Protein SKP1 Contains a Novel Pentasaccharide Linked to Hydroxyproline in Dictyostelium J. Biol. Chem., July 17, 1998; 273(29): 18242 - 18249. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||