(Received for publication, December 10, 1996, and in revised form, January 23, 1997)
From the Department of Biochemistry and Cell Biology, Institute for Cell and Developmental Biology, State University of New York, Stony Brook, New York 11794-5215
O-Linked N-acetylglucosamine (O-GlcNAc) is a ubiquitous and abundant protein modification found on nuclear and cytoplasmic proteins. Several lines of evidence suggest that it is a highly dynamic modification and that the levels of this sugar on proteins may be regulated. Previous workers (Chou, C. F., and Omary, M. B. (1993) J. Biol. Chem. 268, 4465-4472) have shown that mitotic arrest with microtubule-destabilizing agents such as nocodazole causes an increase in the O-GlcNAc levels on keratins in the human colon cancer cell line HT29. We have sought to determine whether this increase in glycosylation is a general (i.e. occurring on many proteins) or a limited (i.e. occurring only on the keratins) process. A general increase would suggest that the microtubule-destabilizing agents were somehow affecting the enzymes responsible for addition and/or removal of O-GlcNAc. Our results suggest that the changes in O-GlcNAc induced by nocodazole are selective for the keratins. The levels of O-GlcNAc on other proteins, including the nuclear pore protein p62 and the transcription factor Sp1, are not significantly affected by this treatment. In agreement with these findings, nocodazole treatment caused no change in the activity of the enzymes responsible for addition or removal of O-GlcNAc as determined by direct in vitro assay. Interestingly, nocodazole treatment did cause a dramatic increase in modification of N-glycans with terminal GlcNAc residues on numerous proteins. Potential mechanisms for this and the change in glycosylation of the keratins are discussed.
O-Linked N-acetylglucosamine (O-GlcNAc)1 is a ubiquitous and abundant form of post-translational modification found on nuclear and cytoplasmic proteins (1-3). It consists of the monosaccharide N-acetylglucosamine attached to the hydroxyl groups of serines and/or threonines. Although over 50 proteins bearing this modification have been identified to date, no example of a specific function for this modification on any of these proteins has yet been demonstrated. Nonetheless, these proteins share some common features that may suggest a role. Significantly, most of these proteins are also known to be phosphoproteins (1-3). Since both phosphate and GlcNAc modify the hydroxyl groups of serines and threonines, it has been proposed that the sugar may function by blocking sites of phosphorylation (4). Several recent reports demonstrating competition between glycosylation and phosphorylation on individual proteins have added support to this concept (5, 6). Modifying a protein with a neutral sugar such as O-GlcNAc could have significantly different effects on the protein compared with phosphorylation. Thus, O-GlcNAc modification of proteins may add an additional level of control to protein phosphorylation by constitutively blocking phosphorylation sites.
In addition to competing with phosphate, several lines of evidence suggest that O-GlcNAc may be a regulated modification much like phosphorylation. For example, in several cases, O-GlcNAc has been shown to turn over more rapidly than the proteins it modifies (7, 8). This suggests that the sugar is dynamically added to and removed from proteins during their lifetimes. Enzymes capable of addition and removal of the sugar from proteins have been purified and characterized (9, 10). They exist as large multisubunit soluble proteins in the cytoplasm of cells. Thus, a system capable of regulated addition and removal of the sugar exists. In addition to simply turning over more rapidly than the protein, changes in the levels of glycosylation induced by various stimuli have been observed in two separate systems. Kearse and Hart (11) showed that mitogenic activation of either isolated T lymphocytes or T-cell hybridomas resulted in rapid (<1 h) changes in the level of glycosylation of several nuclear and cytoplasmic proteins. Subsequently, Chou and Omary (12, 13) demonstrated that mitotic arrest of a human colon cancer cell line (HT29) with okadaic acid or microtubule-destabilizing agents (such as nocodazole) resulted in an increase in glycosylation of keratins 8 and 18. In addition, Chou and Omary (13) observed increased glycosylation of proteins in a crude nuclear extract as well as on several known plasma membrane proteins. The increase in terminal GlcNAc residues of the plasma membrane proteins could be mimicked by brefeldin A, a Golgi disrupter, suggesting that these changes were on extracellular glycans, not O-GlcNAc. The brefeldin A treatment had no effect on the glycosylation of keratins 8 and 18. Nonetheless, the increase in terminal GlcNAc residues on proteins from the nuclear extracts of these cells suggested that nocodazole may cause a general increase in the levels of O-GlcNAc on proteins in these cells.
We have sought to determine whether the increase in glycosylation of
the keratins in HT29 cells induced by nocodazole is a selective or a
general event. If there is an induction of an
O-GlcNAc-transferase or an inhibition of an
O-GlcNAcase by a stimulus such as nocodazole, one would
expect to see changes in the level of glycosylation of several
proteins. This appears to be what was observed by Kearse and Hart (11)
in mitogenically activated T lymphocytes. We have examined the
glycosylation of whole cell extracts as well as several specific
proteins (keratins, a nuclear pore protein, and the transcription factor Sp1) in nocodazole-treated HT29 cells. Our data suggest that the
change in glycosylation of the keratins is a selective event. We saw no
major change in O-GlcNAc levels on the other proteins we
examined. In agreement with Chou and Omary (12, 13), we saw a
generalized increase in terminal GlcNAc residues on numerous proteins
upon nocodazole treatment. We have demonstrated that this increase is
largely due to an accumulation of incompletely processed
N-glycans in nocodazole-treated cells.
Nocodazole, bovine milk galactosyltransferase,
protein G-agarose, and ovalbumin were from Sigma. Galactosyltransferase
was autogalactosylated prior to use as described (14). Peptide
N-glycosidase F (PNGase F) was purified from culture
filtrate of Flavobacterium menigisepticum as described (15).
Anti-keratin monoclonal antibody (L2A1) ascites fluid was generously
provided by Dr. Bishr Omary (Stanford University) (16). Anti-nuclear
pore monoclonal antibody (414) ascites fluid was obtained from Babco
(Berkeley, CA), and an isotype-matched control IgG (ascites fluid) was
obtained from Sigma (mineral oil plasmacytoma 21). Rabbit anti-human
Sp1 polyclonal antibody (PEP2) and competing peptide were obtained from
Santa Cruz Biotechnology (Santa Cruz, CA). Horseradish
peroxidase-conjugated goat anti-mouse IgG was obtained from Jackson
ImmunoResearch Laboratories, Inc. (West Grove, PA). The human colon
cancer cell line HT29 was obtained from the American Type Culture
Collection (Rockville, MD). Dulbecco's modified Eagle's medium, fetal
bovine serum, penicillin/streptomycin, and trypsin/EDTA were obtained
from Life Technologies, Inc. UDP-[6-3H]galactose (50-60
Ci/mmol), UDP-[6-3H]GlcNAc (25.8 Ci/mmol), and
EN3HANCE were from DuPont NEN. The disaccharide standards
(galactose-1,3-N-acetylgalactosamine, galactose-
1,3-N-acetylglucosamine, and
galactose-
1,4-N-acetylglucosamine) were obtained from
Sigma. The alditol derivatives of these standards were prepared by
reduction with sodium borohydride as described (17). The peptide
YSDSPSTST was synthesized at the Center for the Analysis and Synthesis
of Macromolecules, State University of New York at Stony Brook. All
electrophoresis reagents were from Bio-Rad. Enhanced chemiluminescence
reagents were from Amersham Corp. All other reagents were of the
highest quality available.
HT29 cells were
maintained in a humidified incubator with 5% CO2 in
Dulbecco's modified Eagle's medium containing 10% fetal bovine serum
and penicillin/streptomycin (complete Dulbecco's modified Eagle's
medium). Nocodazole treatments were performed by replacing the media on
subconfluent plates (50-75% confluent) with fresh complete media with
or without 0.5 µg/ml nocodazole. At appropriate time points, the
cells were washed once with 10 mM Tris, pH 7.5, 0.15 M NaCl (Tris-buffered saline) and removed from the plates
with trypsin/EDTA. The cells were then washed three times with cold
Tris-buffered saline, counted, and frozen at 80 °C until use.
All
galactosyltransferase labelings were performed as modifications of
described procedures (14). Cell pellets prepared as described above
were extracted with hot 1% SDS (0.5 ml/107 cells) for 5 min at 100 °C. Chromatin was sheared by repeated passage of the
extracts through a 25-gauge needle. Insoluble material was removed by
centrifugation at 10,000 × g for 10 min at 4 °C. Equivalent protein aliquots (100 µg) were precipitated with 8 volumes
of acetone for >4 h at 20 °C. Protein pellets were collected by
centrifugation and solubilized in 50 µl of 1% SDS by heating to
100 °C for 5 min. After cooling, the samples were diluted to 0.5 ml
with galactosyltransferase labeling buffer (final concentrations: 50 mM Hepes, pH 7.4, 5 mM MnCl2, 100 mM galactose) containing 2% (w/v) Triton X-100. At this
concentration of detergents (0.1% SDS and 2% Triton X-100),
galactosyltransferase is fully active (data not shown). The labeling
was conducted under saturating conditions for galactosyltransferase
(41.3 milliunits) and UDP-[6-3H]galactose (2 µCi) at
37 °C for 1 h. Unincorporated radiolabel was separated from the
labeled proteins by gel filtration chromatography (Sephadex G-50) as
described (14). The protein fraction was concentrated by lyophilization
and acetone precipitation prior to analysis by SDS-PAGE and
fluorography.
For galactosyltransferase labeling of keratins, cells were extracted essentially as described by Chou et al. (7). Briefly, cell pellets prepared as described above were extracted for 30 min on ice with 10 mM sodium phosphate, pH 7.4, 0.15 M NaCl (phosphate-buffered saline) containing 0.5% (w/v) Nonidet P-40 and protease inhibitor cocktails 1 and 2 (Pic 1 and Pic 2) (0.5 ml/107 cells) (14). Insoluble material was removed by centrifugation at 10,000 × g for 10 min at 4 °C. Equivalent amounts of protein (100 µg) of the extracts were labeled with galactosyltransferase and UDP-[6-3H]galactose as described above. The labeling was stopped by the addition of 25 mM EDTA, and the keratins were immunoprecipitated from the labeling mixture using monoclonal antibody L2A1 and protein G-agarose as described (18).
Galactosyltransferase Labeling of Nuclear Pore Proteins and Sp1For galactosyltransferase labeling of nuclear pore proteins and Sp1, cells were extracted with RIPA buffer (50 mM Tris, pH 8.0, 0.15 M NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40) containing Pic 1 and Pic 2 for 30 min on ice. Aggregates were broken by passage of the lysates through a 25-gauge needle several times. Insoluble material was removed by centrifugation at 10,000 × g for 10 min at 4 °C. All immunoprecipitations were performed on extract containing 1 mg of protein. Nuclear pore proteins were immunoprecipitated with monoclonal antibody 414, and Sp1 was immunoprecipitated with polyclonal antibody PEP2 essentially as described (18). A control for the monoclonal antibody 414 immunoprecipitations was performed with isotype-matched mouse IgG ascites fluid. The Sp1 immunoprecipitations were done with and without the control peptide as described by the manufacturer (Santa Cruz Biotechnology). Immune complexes were collected with protein G-agarose and washed five times with RIPA buffer. Galactosyltransferase labeling was performed on the immune complexes bound to the protein G-agarose beads in 0.5 ml with galactosyltransferase labeling buffer containing 2% Triton X-100, 41.3 milliunits of galactosyltransferase, and 2 µCi of UDP-[6-3H]galactose for 1 h at 37 °C. The labeled immune complexes were then washed twice with RIPA buffer and analyzed by SDS-PAGE and fluorography.
Carbohydrate AnalysisPNGase F digestions of
galactosyltransferase-labeled protein fractions were performed as
described (19). The released glycans were separated from the protein
fraction by gel filtration on Sephadex G-50 (1 × 50 cm) in 50 mM ammonium formate, 0.1% SDS, 0.02% sodium azide.
Quantitation was performed by measuring the radioactivity of aliquots
from fractions collected. The protein fractions were lyophilized and
acetone-precipitated. Alkali-induced -elimination was performed on
the precipitated, PNGase F-digested protein fractions as described
(19). Released saccharides were separated from the protein fraction by
gel filtration on Sephadex G-50 (1 × 50 cm) as described above.
The released saccharide fractions were desalted by passage over Dowex
50-X8 (H+ form) and Dowex 1-X8 (formate form). The desalted
saccharides were subjected to size fractionation on a Superdex peptide
fast protein liquid chromatography column (Pharmacia Biotech Inc.). The
column was chromatographed at 0.5 ml/min in Milli Q water (Millipore
Corp.) and was calibrated with partially hydrolyzed dextran standards
detected using an in-line Rainin refractive index monitor (20). The
pooled saccharide fraction from the Superdex column was then separated
by high pressure anion-exchange chromatography analysis on a CarboPac
MA-1 column (4 × 25 mm; Dionex Corp.) with disaccharide standards
as described (8).
Cytosolic fractions for enzyme assays were made by
sonicating cell pellets in 20 mM Tris, pH 7.5, 0.25 M sucrose containing Pic 1 and Pic 2 (2 × 107 cells/ml). Sonication was for 10 s on ice with a
Branson Model 185E probe sonicator set on level 3. Sonicates were
centrifuged at 10,000 × g for 10 min at 4 °C. The
supernatants were then centrifuged at 100,000 × g for
60 min at 4 °C. The supernatants (cytosol) were desalted on 1-ml
Sephadex G-50 columns as described (9) prior to assay.
O-GlcNAc-transferase assays were performed essentially as
described (9) using the SP-Sephadex assays and YSDSPSTST as the
acceptor peptide, although the final volume of the assay was reduced to
25 µl. The O-GlcNAcase activity was assayed as described
(10) using p-nitrophenyl-
-N-acetylglucosamine
as substrate. O-GlcNAc
ase assays were performed at pH 6.5 and in the presence of 50 mM GalNAc to inhibit lysosomal
hexosaminidases. Similar O-GlcNAc
ase assays were also
performed with [3H]GlcNAc-labeled YSDSPSTST as described
(10). Essentially identical results were obtained.
UDP-HexNAc levels were determined by high pressure liquid chromatography analysis of perchloric acid extracts of cell pellets as described (21). Since UDP-GlcNAc and UDP-GalNAc do not separate using this technique, data are presented as total UDP-HexNAc levels. We have analyzed the ratio of UDP-GlcNAc to UDP-GalNAc by performing carbohydrate analysis after acid hydrolysis (22) of the UDP-HexNAc fraction. The ratio of GlcNAc to GalNAc in the pool is ~2:1, consistent with previously published results (21).
Other MethodsAll gels were 10% SDS-polyacrylamide gels as described by Laemmli (23). Western blots were performed as described (18). Detection on Western blots was by ECL as described by the manufacturer (Amersham Corp.). Protein was estimated using the BCA assay (Pierce).
Chou and Omary (12) demonstrated that nocodazole
treatment of the human colon cancer cell line HT29 resulted in a
significant increase in the glycosylation of keratins 8 and 18. We
sought to extend this observation to determine whether the increase in glycosylation was a general phenomenon, or if the change was specific for the keratins. To demonstrate that nocodazole treatment of the HT29
cells does indeed cause a change in the level of O-GlcNAc on
the keratins, a time course of nocodazole treatment was performed (Fig.
1). Cells were lysed at the indicated times, and
portions of each lysate were labeled with galactosyltransferase under
conditions where all accessible terminal GlcNAc residues are modified
with [3H]galactose (see "Experimental Procedures").
Keratins 8 and 18 were then isolated by immunoprecipitation and
examined by SDS-PAGE and fluorography. Glycosylation of keratin 18 was
barely detectable in the untreated cells, but it increased markedly
between 8 and 24 h of nocodazole treatment (Fig. 1A),
coincident with mitotic arrest (12). Under these same conditions, we
saw very little glycosylation of keratin 8. Chou and Omary (12) also
reported significantly less glycosylation of keratin 8 than keratin 18. During the time course, there was essentially no change in the level of
the keratins themselves (Fig. 1B), indicating that there is
a true increase in glycosylation of keratin 18 and not just an increase
in the level of the protein. We also saw an increase in the labeling of
a protein that coprecipitates with the keratins. This protein is most
likely an N-glycosylated protein called KAP85 (13). Thus, as
described by Chou and Omary (12), nocodazole induces a dramatic
increase in the level of glycosylation of keratin 18.
Nocodazole Induces a Significant Increase in Terminal GlcNAc Residues on Numerous Proteins in HT29 Cells
To determine whether
the effect of nocodazole on the keratins was a general or a selective
event, we examined the glycosylation of proteins from crude lysates of
HT29 cells. Lysates of HT29 cells were galactosyltransferase-labeled
exactly as described above except that no immunoprecipitation was
performed. A dramatic increase in the labeling of numerous protein
species occurred between the 8- and 24-h time points (Fig.
2A), coincident with the increase in the
glycosylation of keratin 18 shown in Fig. 1. During this same period,
no significant changes were seen in the levels of the proteins
themselves (Fig. 2B). These results demonstrate that mitotic
arrest with nocodazole induces a large increase in terminal GlcNAc
residues on numerous proteins from HT29 cells. Interestingly, the
labeling of at least one protein decreased upon nocodazole treatment
(Fig. 2A, asterisk). We believe that this
decrease is due to a decrease in the solubility of this protein under
the extraction conditions used (phosphate-buffered saline and 0.5%
Nonidet P-40). The decrease is no longer observed when cell extracts
are made with either RIPA buffer or 1% SDS (see Fig. 4). For this
reason, subsequent lysates were all made with either RIPA buffer or 1%
SDS.
Increase in Terminal GlcNAc Residues Is Due to N-Glycans and Not O-GlcNAc
Chou and Omary (13) had demonstrated that nocodazole
could induce an increase in the level of terminal GlcNAc residues on cytokeratins, proteins in nuclear extracts, and extracellular proteins.
Since the changes in GlcNAc on extracellular proteins were likely to
result from changes in N-glycans, we sought to determine
whether the increase in terminal GlcNAc residues shown in Fig. 2 was
due to changes in O-linked and/or N-linked
GlcNAc. We used the enzyme PNGase F to differentiate between GlcNAc
linked to proteins through N-glycans or other means. In
untreated cells, essentially all of the radiolabel was resistant to
PNGase F digestion (Fig. 3A, compare 0 h PNGase F with 0 h + PNGase F), implying that the majority of
the GlcNAc residues are O-linked. In contrast, the
nocodazole-treated samples showed significant sensitivity to the PNGase
F digestion (Fig. 3A, compare 32 h
PNGase F with 32 h + PNGase F). In fact, the increase in galactose incorporation caused by the nocodazole treatment could be completely accounted for by
the increase in PNGase F-sensitive (i.e. N-linked) material. Thus, nocodazole appeared to induce an increase in terminal GlcNAc residues on N-glycans in these cells, but it caused no
significant changes in the PNGase F-resistant (i.e.
O-linked) material.
To confirm that the labeled sugars remaining on the proteins after
PNGase F digestion were in the form of O-GlcNAc, the PNGase F-resistant material from both the control and nocodazole-treated samples was subjected to alkali-induced -elimination. This caused the majority (>90%) of the radioactivity to be released from the protein fraction (Fig. 3B), indicating that the PNGase
F-resistant radioactivity was O-linked to the protein. To
determine the structure of the material released by the
-elimination, the samples were subjected to size (Fig.
3C) and high pressure anion-exchange chromatography (Fig.
3D) analyses. These analyses demonstrated that the released sugars were essentially all in the form of
[3H]galactose-
1,4-N-acetylglucosaminitol,
the expected product of galactose-labeled O-GlcNAc. Thus,
essentially all of the PNGase F-resistant [3H]galactose
was linked to O-GlcNAc.
To examine whether we could see changes in the level of O-GlcNAc on proteins in response to nocodazole treatment, we examined the pattern of O-GlcNAc-modified proteins from control and nocodazole-treated cells by SDS-PAGE and fluorography after PNGase F treatment (Fig. 4A). As shown above, a large increase in PNGase F-sensitive terminal GlcNAc residues was induced by the nocodazole treatment. In contrast, the pattern of the PNGase F-resistant species (which represent the O-GlcNAc-labeled species as demonstrated above) did not change dramatically upon nocodazole treatment. Some minor changes could be seen on the gel, but the overall pattern was unchanged. Likewise, the overall total protein pattern was largely unchanged by nocodazole treatment (Fig. 4B). Thus, it appears that nocodazole does not induce major changes in the levels of glycosylation on the more abundant O-GlcNAc-bearing proteins (i.e. those that can be detected in a crude cell lysate) in HT29 cells. We can conclude from these data that global changes in the pattern of O-GlcNAc-modified proteins like those observed by Kearse and Hart (11) in mitogenically activated T-cells are not induced by nocodazole treatment.
Nocodazole Treatment Has Little or No Effect on the Glycosylation of the Nuclear Pore Protein p62 or the Transcription Factor Sp1To more carefully examine changes in the glycosylation of
specific O-GlcNAc-modified proteins other than the keratins,
we performed galactosyltransferase labelings of the nuclear pore protein p62 and the transcription factor Sp1 from control and nocodazole-treated cells. Both of these proteins are known to be
modified with O-GlcNAc (24-26). The nuclear pore proteins
are among the most abundant and heavily O-GlcNAc-modified
proteins in mammalian cells (24). Thus, they serve as a good marker of the level of O-GlcNAc modification in a cell. Monoclonal
antibody 414, which reacts primarily with p62 and more weakly with
other nuclear pore proteins (25), was used to immunoprecipitate the protein from control and nocodazole-treated HT29 cells. The
immunoprecipitates were then labeled with galactosyltransferase to
detect O-GlcNAc. As can be seen in Fig.
5A, p62 labeled quite well using this
technique. The presence or absence of nocodazole had very little effect
on the level of its glycosylation. The levels of the p62 protein were
monitored by Western blot analysis with monoclonal antibody 414 (Fig.
5B). If anything, this analysis showed that nocodazole induces a slight increase in the amount of the p62 protein. Thus, there
does not appear to be an increase in the sugar/protein ratio on p62. We
also examined the glycosylation of the transcription factor Sp1 using
the same procedure. Sp1 exists as two major molecular mass species in
cells (95 and 105 kDa), both of which have been shown to be
glycosylated (26). Both species were present in untreated HT29 cells
(Fig. 5D), and both appeared to be glycosylated (Fig.
5C). Nocodazole treatment caused an accumulation of the 105-kDa form (Fig. 5, C and D), but the overall
level of glycosylation on the protein did not increase. Taken together,
these data suggest that the changes seen in glycosylation of the
keratins are selective in that similar changes do not occur on all
O-GlcNAc-modified proteins in the cell.
Nocodazole Treatment Has No Effect on the O-GlcNAc Glycosylation Machinery
In addition to examining the level of
O-GlcNAc on proteins in nocodazole-treated cells, we assayed
for changes in the activity levels of the enzymes responsible for the
addition (O-GlcNAc-transferase) or removal
(O-GlcNAcase) of O-GlcNAc. Changes in the
activity of these enzymes would be indicative of a general effect of
nocodazole on glycosylation. We have modified the standard assay for
both of these enzymes so that they can be detected in crude extracts of
~107 cells (see "Experimental Procedures"). Using
these modified assays on extracts of HT29 cells treated for increasing
times with nocodazole, we saw no significant changes in activity that
correlate with the increase in glycosylation seen on the keratins (Fig.
6, A and B). In addition, since
UDP-GlcNAc is a substrate for O-GlcNAc-transferase and could
be limiting, we examined the levels of UDP-GlcNAc over the course of
nocodazole treatment (Fig. 6C). The technique used for
analysis of UDP-GlcNAc was not capable of resolving UDP-GlcNAc from
UDP-GalNAc (21). Thus, the data are shown as the total UDP-HexNAc
levels. The ratio of UDP-GlcNAc to UDP-GalNAc in a cell is determined
by an epimerase capable of interconverting the two, and a relatively
constant ratio of ~2:1 (UDP-GlcNAc/UDP-GalNAc) is found in most cells
(see "Experimental Procedures"). A dramatic increase in UDP-HexNAc
levels was observed during the latter time points of the nocodazole
treatment, but these changes occurred much later than the observed
increase in O-GlcNAc levels on the keratins. Thus, we could
detect no gross changes in the glycosylation machinery responsible for
the addition and/or removal of O-GlcNAc from proteins. These
data are consistent with the observations discussed above and suggest
that the change in glycosylation of the keratins induced by nocodazole
is a selective phenomenon.
The modulation of post-translational modifications such as phosphate groups on proteins is well documented. The level of phosphate on proteins is controlled by the relative activity of the enzymes responsible for addition (kinases) and removal (phosphatases) of the phosphate. One implication of such systems is that stimuli that activate or inhibit these enzymes can cause changes in modification of numerous target proteins. For example, stimulation of cAMP-dependent protein kinase with cAMP results in phosphorylation of numerous target proteins. Thus, if nocodazole were a modulator of a system controlling O-GlcNAc levels, one would predict that the levels of glycosylation on numerous proteins would change. Since nocodazole is a microtubule-destabilizing agent that arrests the cell at the G2/M interface of the cell cycle, such an effect could be associated with cell cycle-induced changes. The goal of this report was to test this prediction and to determine whether such a system is operating in HT29 cells with nocodazole as a stimulator.
In this report, we have shown that nocodazole causes a marked increase
in the level of O-GlcNAc on keratin 18, as was originally reported by Chou and Omary (12). In contrast, analysis of
O-GlcNAc levels on other proteins in the nocodazole-treated
cells showed few, if any, changes. Consistent with this observation is
the fact that we saw no changes in the glycosylation machinery
(O-GlcNAc-transferase, O-GlcNAcase, and
UDP-GlcNAc levels) induced by nocodazole that would explain the
increase in glycosylation of keratin 18. Thus, nocodazole appears to
induce an increase in glycosylation of keratin 18 selectively. This
implies that the glycosylation machinery is not activated by nocodazole
treatment and that the increase in glycosylation of keratin 18 must
result from a change in the keratins themselves.
Since nocodazole causes a reorganization of the cytoskeletal components, including the intermediate filaments (12), the simplest explanation for the change in glycosylation of the keratins is that it is related to this reorganization. Omary's group (12, 13) has done extensive work showing that agents that cause reorganization of intermediate filaments in HT29 cells result in dramatic increases in both the glycosylation and phosphorylation of the keratins. In addition to the changes induced by inhibitors that arrest the cell cycle at the G2/M interface (okadaic acid and microtubule-destabilizing agents such nocodazole and colcemid) (12, 13), Omary and co-workers (27) have shown that heat shock causes a rapid increase in the glycosylation and phosphorylation of keratins 8 and 18. These results suggest that the changes in phosphorylation and/or glycosylation are related to the reorganization. Mutation of a major phosphorylation site in keratin 18 does not alter intermediate filament assembly, but it does interfere with filament reorganization of keratins 8 and 18 induced by G2/M arrest in transfected cells (28). Thus, phosphorylation appears to play an active role in filament reorganization in HT29 cells. A similar role for the glycosylation of the keratins is not yet clear. Mutation of the major glycosylation sites of keratin 18 had no effects on filament assembly in transfected cells (29), although the effects of the mutation on reorganization induced by G2/M arrest have not been examined. Recent studies have demonstrated that mutation of a conserved arginine in keratin 18 is sufficient to induce a significant reorganization of intermediate filament structures. The glycosylation and phosphorylation of this mutated keratin 18 are significantly increased compared with the wild-type protein (30). The increase in glycosylation observed on this keratin 18 bearing a point mutation strongly suggests that the change in glycosylation is a result of intermediate filament reorganization, and not a cause. Such a conclusion is consistent with our findings that the increase in glycosylation caused by nocodazole is apparently selective for the keratins. Further work will be needed to conclusively determine the role glycosylation plays in intermediate filament reorganization. Interestingly, not all treatments that cause reorganization of intermediate filaments have the same effects. Rotavirus infection of HT29 cells, which causes intermediate filament reorganization, results in an increase in phosphorylation of keratins 8 and 18, but no change in their glycosylation (27). Thus, the signals that result in changes in glycosylation or phosphorylation of the keratins appear to be distinct.
The most dramatic change in glycosylation induced by nocodazole is the increase in terminal GlcNAc residues on N-glycans. Other workers have shown that short-term treatments with nocodazole (e.g. 1-3 h) have no effect on the movement or glycosylation of proteins through the Golgi apparatus (31). Microtubules are believed to be involved in retrograde transport of proteins from the Golgi apparatus to the endoplasmic reticulum (32), and nocodazole has been used to inhibit this transport. The effects of nocodazole on retrograde transport are also seen after fairly short treatments (<3 h). We see no effect on glycosylation until after 8 h (see Fig. 2A). Thus, the effects we see are apparently not related to the short-term effects of microtubule disassembly, but are more likely related to arrest of the cells at the G2/M interface. Most vesicular traffic in cells, including endoplasmic reticulum to Golgi apparatus and intra-Golgi traffic, has been shown to be interrupted during mitosis (33). The Golgi apparatus fragments, assisting in even distribution of the Golgi apparatus between the two daughter cells. Thus, the accumulation of the terminal GlcNAc residues on N-glycans that we see is most likely a result of the mitosis-induced interruption of medial to trans-Golgi transport. Since galactosyltransferase is a trans-Golgi enzyme (34), lack of medial to trans-Golgi movement would result in an accumulation of unmodified GlcNAc residues. The accumulation we see may suggest that there is a slight time delay between interruption of intra-Golgi transport (medial to trans) and that of endoplasmic reticulum to Golgi transport. The mechanism of Golgi fragmentation, as well as the rationale for why transport stops at mitosis, is not well understood (33). Several workers have suggested that Golgi vesiculation may result from the normal process of vesicular budding from Golgi stacks coupled with an inhibition of vesicle fusion. Thus, vesicles form but are unable to fuse with the next stack. The fact that terminal GlcNAc moieties on N-glycans increase under these conditions may serve as a useful assay to follow this process and allow for a better understanding of the underlying mechanisms.
The results presented in this paper suggest several important factors that must be kept in mind when analyzing changes in the levels of O-GlcNAc on proteins. Galactosyltransferase is still the best method available for analyzing O-GlcNAc levels, and PNGase F digests must be performed on each labeling to assess the contribution of N-glycans to the final pattern. If we had analyzed only the galactosyltransferase-labeled samples from the untreated cells with PNGase F, we would have concluded that >90% of the galactose incorporated was on O-GlcNAc and that PNGase F controls were unnecessary in this cell type. In contrast, the nocodazole treatment caused a dramatic change in galactose labeling of N-glycans. In addition, this study underscores the importance of following protein levels as well as sugar levels before conclusions about changes in glycosylation of any protein can be made. Initial studies with Nonidet P-40 lysates of HT29 cells showed an apparent decrease in glycosylation of some proteins (see Fig. 2). We believe that the decrease has to do with changes in the solubility of proteins induced by the nocodazole treatment of the cells, and not with true changes in the level of glycosylation. Once the change in the levels of the proteins was abrogated by a more efficient extraction technique (e.g. RIPA buffer or 1% SDS), the changes in the glycosylation level disappeared. Thus, in future studies on the levels of O-GlcNAc on proteins, it will be essential to control for labeling of N-glycans with PNGase F and to closely follow the level of the proteins being examined as well as the level of the sugar on them.
We thank Scott Busby, Kathleen Grove, Sean
Li, Lillian Medina, Daniel Moloney, Dr. Deborah Brown, and Dr. James
Trimmer for critical reading of this manuscript and helpful
discussions. We also thank Mona Sheth, Scott Busby, and Patrick Reilly
for assistance in modifying the assays used for both
O-GlcNAc-transferase and O-GlcNAcase. Peptide
synthesis was performed at the Center for Analysis and Synthesis of
Macromolecules and the Center for Biotechnology, State University of
New York at Stony Brook.