(The FASEB Journal. 2001;15:1865-1876.)
© 2001 FASEB
Glycan-dependent signaling: O-linked N-acetylglucosamine
JOHN A. HANOVER1
LCBB, NIDDK, National Institutes of Health, Bethesda, Maryland 20892, USA
1Correspondence: Chief, LCBB, NIDDK, National Institutes of Health, Bldg. 8, Rm. 402, Bethesda, MD 20892, USA. E-mail: jah{at}helix.nih.gov
 |
ABSTRACT
|
|---|
The addition of O-linked N-acetylglucosamine (O-GlcNAc) to target
proteins may serve as a signaling modification analogous to protein
phosphorylation. Like phosphorylation, O-GlcNAc is a dynamic
modification occurring in the nucleus and cytoplasm. Various analytical
methods have been developed to detect O-GlcNAc and distinguish it from
glycosylation in the endomembrane system. Many target molecules have
been identified; these targets are typically components of
supramolecular complexes such as transcription factors, nuclear pore
proteins, or cytoskeletal components. The enzymes responsible for
O-GlcNAc addition and removal are highly conserved molecules having
molecular features consistent with a signaling role. The O-GlcNAc
transferase and O-GlcNAcase are likely to act in consort with kinases
and phosphatases generating various isoforms of physiological
substrates. These isoforms may differ in such properties as
protein–protein interactions, protein stability, and enzymatic
activity. Since O-GlcNAc plays a critical role in the regulation of
signaling pathways of higher plants, the glycan modification is likely
to perform similar signaling functions in mammalian cells. Glucose and
amino acid metabolism generates hexosamine precursors that may be key
regulators of a nutrient sensing pathway involving O-GlcNAc signaling.
Altered O-linked GlcNAc metabolism may also occur in human diseases
including neurodegenerative disorders, diabetes mellitus and
cancer.—Hanover, J. A. Glycan-dependent signaling: O-linked
N-acetylglucosamine.
Key Words: O-GlcNAc • hexosaminidase C • diabetes mellitus • OGT
|
O-LINKED N-ACETYLGLUCOSAMINE: A BRIEF HISTORY
|
|---|
Although numerous early findings pointed to the existence of short
chain saccharides terminating in N-acetylglucosamine (O-GlcNAc) was
first formally described in 1984 (1)
. This early report
focused on O-GlcNAc as a cell surface modification, but it was soon
recognized that the bulk of the O-GlcNAc residues were intracellular
(2
3
4
5
6)
. One particularly striking finding was that
glycoproteins bearing O-GlcNAc were present in high concentration at
the vertebrate nuclear pore complex (4
5
6
7)
. This was
followed in close order by the observation that components of the
transcription machinery were also modified by O-GlcNAc (8
, 9)
. As the number of proteins shown to contain O-GlcNAc
continued to grow, interest turned to the enzymes mediating glycan
addition and removal. The O-linked GlcNAc transferase was characterized
and partially purified from rat liver and rabbit reticulocyte lysate
(10
, 11)
. Characterization of the hexosaminidase proved
more difficult, and a report describing the activity appeared several
years later (12)
. More recently a protein previously
thought to be a hyaluronidase has been tentatively identified as
O-GlcNAcase (13)
. Analysis of the role of O-GlcNAc in
cellular physiology closely paralleled the advances in enzymology and
substrate characterization. The modification was shown to play a role
in nuclear transport, although a lectin-like interaction was probably
not involved (14
, 15)
. O-GlcNAc addition and removal were
shown to be highly dynamic and changed dramatically during cell
activation (16)
. In addition, transcription and
translation factors were shown to be modified in their activity by
O-GlcNAc modification and, in some cases, their activities were
modified (8
, 17
18
19
20
21
22
23
24
25)
. The O-linked GlcNAc transferase was
molecularly cloned and shown to be a member of the tetratricopeptide
repeat (TPR) family of proteins and similar to the Spindly gene in
Arabidopsis (26
, 27)
. More recently, a detailed
analysis of the O-GlcNAc transferase revealed more of its structure
(28
, 29)
. Certain kinases, including GSK-3 and casein
kinase, were shown to be substrates for the enzyme (29)
.
The gene encoding the transferase has also been analyzed in mice and
humans and appears to be essential (30
; J. A. Hanover
unpublished results). Thus, for O-GlcNAc, it has been an exciting 2
decades from discovery to gene. The reagents are now available for a
more detailed analysis of the functioning of this unique modification.
|
METHODS OF ANALYSIS
|
|---|
Galactosyltransferase
Because O-GlcNAc is a dynamic modification, its detection on
substrate proteins has sometimes proved difficult. The original means
of detection was the use of the catalytic subunit of the enzyme lactose
synthetase, galactosyltransferase, as a topological probe (1
, 31)
. This enzyme transfers galactose to terminal GlcNAc
residues, thus creating a ß 1–4 linkage. By using radiochemically
labeled UDP-galactose as substrate, acceptor molecules could readily be
identified. This technique has its share of limitations, notably, the
cryptic nature of some O-GlcNAc sites. It must also be used in
combination with other techniques to confirm the identity of O-GlcNAc
since the enzyme will modify any terminal GlcNAc, even those present on
membrane and secretory N-linked and O-linked oligosaccharides.
Alkaline ß-elimination
Since O-GlcNAc is in an O-glycosidic linkage to Ser and Thr
residues on proteins, it is susceptible to removal by reduction under
alkaline conditions (1
, 4)
. Although not specific for
O-GlcNAc, this method can be useful in confirming the identity of the
modification after labeling with galactosyltransferase or in tissue
extracts. Some destruction of the GlcNAcitol does occur during this
reaction.
Antibodies to O-GlcNAc containing proteins
Antibodies generated to GlcNAc-modified proteins have been
particularly helpful in identifying this class of proteins
(4
5
6
7)
. In addition, antibodies to streptococcal antigens
have been found to cross-react with O-GlcNAc (32)
.
Antibodies more specific for O-GlcNAc have recently been prepared in
several laboratories; these are likely to help in understanding the
cellular regulation of O-GlcNAc addition. Such antibodies may be used
much the way Ser and Thr phosphorylation specific antibodies have been
used on immunoblots and in combination with immunofluorescence
microscopy.
Mass spectrometry
Mass spectrometry has recently been applied to the detection and
characterization of O-GlcNAc (17
, 24
, 33
34
35
36
37)
. By
examining intact proteins using MALDI, it is possible to look at the
time course and extent of glycosylation of an O-linked GlcNAc
glycoprotein. We have recently used this approach to examine the
glycosylation of recombinant nuclear pore proteins (see
http://www.cipergen.com/F1 case1.html). This method is
both more specific and more quantitative than some of the other
analytical approaches and has been described in greater depth elsewhere
(36)
.
Other analytical approaches
Other approaches such as gel electrophoresis and fluorescent
techniques have been used to detect O-GlcNAc. Perhaps the best
characterized is the FACE (fluorophore-assisted carbohydrate
electrophoresis) procedure in which gel electrophoresis is used to
separate a mixture of fluorescently labeled monosaccharides (see
http://.www.glyko.com/products/face.html). This approach has
recently been applied to O-linked GlcNAc. The difficulty with this
procedure is determining what fraction of the total GlcNAc detected is
derived from the O-linked GlcNAc linkage. This may be facilitated when
recombinant forms of specific O-GlcNAcases become readily available.
|
HOW WIDESPREAD IS O-GlcNAc ADDITION?
|
|---|
Although most of the characterized proteins modified by O-GlcNAc
are from vertebrate sources, it is clear that the modification is much
more widespread. Growing knowledge of genome sequences including
Arabidopsis, other higher plants, Drosophila,
Caenorhabditis, various parasitic protozoa, yeast, and bacteria
have provided some clues. The enzymes of O-GlcNAc metabolism (OGT
and O-GlcNAcase) have been seen in these organisms as expressed
sequence tags, and in some cases the genes have been identified. More
problematic has been the demonstration of the carbohydrate linkage
itself in these diverse organisms. In plants, there is some evidence
that nuclear pore proteins are modified by O-GlcNAc (38
, 39)
, but the modification is capped with addition saccharide
moieties. Enzymes with features suggesting they mediate O-GlcNAc
addition are present in Arabidopsis (40)
. In
protozoa, O-GlcNAc addition has been well documented (41
, 42
, 44)
. There is no clear-cut evidence for O-GlcNAc transferase
activity in budding yeast and detection of the linkage has proved
difficult. In bacteria, the problem has been less well studied. Some
evidence suggests that the enzymes may be present in bacteria, perhaps
being derived by lateral transfer from higher organisms
(45)
. As the role of O-GlcNAc addition comes under closer
scrutiny and the number of completely sequenced genomes increases, we
are likely to learn more about its phylogenetic distribution.
|
CELLULAR PROTEINS MODIFIED BY O-LINKED GLcNAc TRANSFERASE: CLUES TO
FUNCTION?
|
|---|
As shown in Fig. 1
, many proteins are modified by O-GlcNAc. An analysis of the properties
of all of these proteins has suggested some shared features. In
general, these proteins are both phosphorylated and glycosylated. They
also are present in macromolecular complexes whose assembly may be
regulated. A brief description of some of these macromolecular
complexes follows.
View larger version (41K):
[in this window]
[in a new window]
|
Figure 1. Substrates for O-linked GlcNAc transferase (OGT) and similarity to
protein phosphorylation. O-linked GlcNAc has been found to modify
components of supramolecular structures such as the nuclear pore,
transcription complex, and certain cytoskeletal proteins. The figures
on the left depict the supramolecular structures that are targets for
O-GlcNAc addition: a reconstruction of the nuclear pore complex, puffs
on polytene chromosomes, and atomic force micrographs of a
transcription complex and the cytoskeleton, respectively. The figure
below illustrates the dynamic relationship between protein
phosphorylation and O-linked GlcNAc addition. OGT has a substrate
specificity similar to known kinases such as GSK-3 and Casein
kinase.
|
|
Nuclear pore proteins (the nucleoporins)
One of the best studied classes of substrates are the nuclear pore
proteins (4
5
6
7
, 11
, 15
, 27
, 29
, 46
47
48
49
50
51
52
53
54
55
56
57)
. At least six
major vertebrate nuclear pore proteins are modified by O-GlcNAc.
Yet it is still not clear how the modification is involved in the many
functions of the nuclear pore complex. Antibodies or lectins that
interact with glycosylated nucleoporins interfere with nuclear
transport at an energy-dependent step (15
, 46
, 58
59
60
61)
.
Some evidence suggests that O-GlcNAc may represent an alternative
nuclear transport signal (62)
. Nuclear transport and
nuclear pore assembly can occur when O-GlcNAc is capped with a
galactose residue (14
, 54)
. Yet cytoplasmic expression of
a galactosyltransferase appears to be toxic to cells (63)
.
Nuclear pore proteins are both glycosylated and phosphorylated
(51
, 54
, 64
65
66
67)
. Phosphorylation of the nucleoporins is
regulated during the cell cycle whereas glycosylation appears stable
throughout the cell cycle (54)
. Although the precise role
of the O-GlcNAc modification in nuclear import/export is poorly
understood, the finding that so many nucleoporins are modified in this
manner remains provocative.
Transcription machinery (Pol II and transcription factors)
RNA polymerase II and many of its associated transcription factors
are modified by O-GlcNAc (8
, 17
18
19
, 22
23
24
25
, 33
, 68
69
70
71
72
73)
.
Much work has focused on the CTD domain of RNA polymerase, where
phosphorylation and glycosylation appear to be reciprocal or mutually
exclusive (74)
. The precise role of the modification in
the functioning of the polymerase has remained elusive, however.
Evidence suggests that O-GlcNAc may modulate the turnover and
protein–protein interactions of transcription factors such as SP1 and
the estrogen receptor (8
, 22
, 23
, 25
, 68
, 69
, 72
, 75)
.
Like the work on the nuclear transport machinery, efforts to understand
the role of glycosylation in the normal function or regulation of the
transcription machinery have proved difficult. This is due to the
tremendous complexity of that machinery itself, the absence of good in
vitro systems, and the finding that so many components are modified by
O-GlcNAc.
O-GlcNAc-modified proteins and cellular architecture
Many cytosolic proteins have been identified as substrates for
O-GlcNAc addition, as Fig. 1
demonstrates. Some common themes emerge
upon examination of these cytoskeletal substrates. Many of the known
regulatory proteins that interact with actin and tubulin contain
O-GlcNAc (66
, 76
77
78)
. Proteins thought to bridge the
cytoskeleton to the plasma membrane such as band 4.1, vinculin, talin,
and the synapsins are also modified (2
, 79
80
81
82)
. Some of
the cytokeratins appear to be dynamically modified by O-GlcNAc and
phosphate, with some indications of changes during the cell cycle
(83
84
85)
. The presence of distinct isoforms of these
various proteins is consistent with the notion that O-linked GlcNAc
addition may a final step in an incompletely defined signal
transduction pathway.
|
THE O-LINKED GlcNAc TRANSFERASE: A CONSERVED TPR PROTEIN
|
|---|
Efforts to understand the dynamic addition of GlcNAc to a large
number of substrates has led to the purification and cloning of the
enzyme(s) mediating this glycan addition.
Purification, characterization, and molecular identification of
O-GlcNAc transferase
ß-N-acetylglucosaminyltransferase (OGT) activity has been
detected in a wide range of tissues including rat liver, rabbit blood,
and an enriched reticulocyte fraction (10
, 11
, 55)
. The
strategies used to identify the activity relied heavily on the
identification of protein substrates such as the nuclear pore proteins,
which provided peptide substrates or could be produced in recombinant
form (10
, 11
, 49
, 55
, 67)
. Subsequent purification of the
enzyme from these sources revealed a catalytic component of 
100 kDa,
and microsequencing of the protein led rapidly to the molecular cloning
of the transferase from various sources (26
, 27)
.
O-linked GlcNAc transferase: molecular genetics
Transcripts encoding OGT
O-linked GlcNAc transferase cDNAs and the genes encoding those
transcripts have been identified in humans, rats, and
Caenorhabditis elegans (26
, 27)
. In humans,
four predominant transcripts encoding OGT ranging from 10 to 3 kb
transcripts were found in most tissues. Transcripts were particularly
enriched in the pancreas (27)
. Even more striking was the
high concentration of OGT transcripts in the ß cells of the islets of
Langerhans within the pancreas (72
, 86)
(Fig. 2
). These observations were particularly intriguing in light of our
hypothesis that OGT plays a role in glucose sensing. By regulating the
secretion of insulin, the pancreatic ß cell is the key regulator of
glucose homeostasis. Other tissues such as skeletal muscle, brain, fat,
and cardiac muscle also had relatively high mRNA levels. Smaller
amounts of the transcript are present in the kidney (26
, 27)
. The current databases suggest that many alternatively
spliced forms of OGT exist in certain tissues.
View larger version (77K):
[in this window]
[in a new window]
|
Figure 2. In situ localization of OGT transcripts in the pancreas to the ß
cells of the islet of Langerhans. Double-label immunofluorescent
detection of mRNA transcripts within the mouse pancreas with anti DIG
antibody (red) and either insulin (at left) or glucagon (at right,
green) FITC labeled antibody). Colocalization of the labels in the ß
cells suggests enrichment of the OGT transcripts in this cell type.
Note the lack of colocalization in the  cells marked by
anti-glucagon antibody.
|
|
The OGT gene
The gene encoding OGT is present as a single copy on the same
region of the mouse and human X chromosome (Fig. 3
; see also ref 30
). Positional cloning efforts have yet to
positively identify any human diseases that precisely map to this
region, although the OGT locus is a candidate gene for X-linked
Parkinson dystonia (see ‘Neurodegenerative diseases: a role for
O-linked GlcNAc?’ below). On ablation in embryos, the gene has been
shown to be essential for stem cell viability (J. A.
Hanover, unpublished results) and embryonic development
(30)
. Conditional knockout experiments involving ablation
of the gene in the various target tissues are under way. Much remains
to be learned about the processing and expression of the OGT gene.
View larger version (26K):
[in this window]
[in a new window]
|
Figure 3. Location of the OGT gene on mouse and human chromosome X. The gene
encoding OGT spans greater than 50 kb of mouse and human chromosome X.
This localization has been determined by fluorescence in situ
hybridization (left for mouse) and by radiation hybrid analysis and
direct sequencing for the human gene encoding. The gene is essential
for both embryonic stem cell viability and early embryonic
development.
|
|
Domain structure and post-translational modifications
The coding sequence of OGT suggested a tripartite structure
consisting of a repetitive domain, a linker and targeting domain, and a
catalytic domain (26
, 27)
(Fig. 4
). This overall structure has been largely confirmed as a result of
site-directed mutagenesis and expression in Escherichia coli
and baculovirus (28
, 29)
. The structures shown in Fig. 4
are proteins (whose X-ray structure has recently been solved) that
share the conserved domains of OGT (see below).
View larger version (22K):
[in this window]
[in a new window]
|
Figure 4. Domain structure of O-linked GlcNAc transferase. OGT has an
amino-terminal repeat domain containing tetratricopeptide repeats
(TPR), a central linker and targeting domain, and a carboxyl-terminal
catalytic domain. A molecular model of the structure of the 9–12 TPR
repeats was derived from the crystal structure of the PP1 protein
(87)
. The structure shown below at right is from the
coordinates of N-acetylglucosaminyltransferase I, which represents a
new protein superfamily. OGT is likely to adopt a similar structure
based on sequence similarity with transferases such as
N-Acetylglucosaminyltransferase I.
|
|
Tetratricopeptide repeats
The TPR repeats are present in 9–12 copies in the known O-linked
GlcNAc transferases (26
, 27
, 29
, 45)
(Fig. 4)
. The TPR
repeat is found in many proteins where it is thought to mediate
protein–protein interactions. The TPR repeats of OGT may play a role
in substrate recognition and trimerization of the enzyme (28
, 29
, 45)
. In general, these repeats act in groups of three, thus
forming a groove that will accommodate peptides (45)
. The
structure of a TPR repeat protein was solved recently, leading to the
proposal that repeats may form a superhelix with a diameter of roughly
50 Å (87)
. Other studies suggest that certain TPR
proteins may serve to bridge other proteins such as those of the heat
shock proteins in the multichaperone machine (88)
. The
precise role of the TPR domain of OGT is currently under active
investigation.
Linker/targeting domain
Following the TPR repeats in the central region of C.
elegans, OGT is a sequence that appears to be a canonic nuclear
targeting signal (Fig. 4)
. This motif is interrupted in human and rat
OGT, and precise mutagenesis studies have not yet been performed to
demonstrate whether this domain is important for nuclear targeting of
OGT. Overexpression of the enzyme in C. elegans confirmed
that it enters the nucleus when expressed in the nematode
(27)
. Clearly, some of the enzyme is present in the
nucleus and many substrates are nuclear. It is also clear that a pool
of OGT exists in the cytoplasm and on the surface of certain
cytoplasmic organelles. The regulation of the intracellular trafficking
of the enzyme is under intense investigation.
The catalytic domain
We have suggested that the catalytic domain of OGT shares some
features with known glycosyltransferases and certain lectins
(45)
. This similarity is not striking at the primary
sequence level, but some motifs present in OGT are also present in
these other enzymes (Fig. 4)
. The regions of similarity are present in
two subdomains designated catalytic domains 1 and 2. Recently, the
crystal structure of an O-linked GlcNAc has been solved
(89)
, and the critical residues interacting with sugar
nucleotide appear to be conserved between these enzymes and OGT (such
as the ‘... DXD... ’ motif). Mutational analysis
suggests that any substantial deletion in this region of the molecule
dramatically inhibits the enzymatic activity of OGT (28
, 29
, 45)
(Fig. 4)
.
Post-translational modifications
Although it is the product of a unique single gene, OGT appears to
be subject to proteolysis and post-translational modifications. A 78
kDa form of the enzyme has been detected (10)
, which may
represent that proteolytic product. In addition, OGT is modified
post-translationally in some intriguing ways (26
27
28
29)
.
These include autoglycosylation with O-GlcNAc and tyrosine
phosphorylation. The effects of these modifications on enzyme activity
or substrate recognition have not been examined in detail.
|
O-GlcNAcase (HEXOSAMINIDASE C)
|
|---|
O GlcNAcase: purification and identification
Some studies have been directed at purifying and identifying the
enzyme (or enzymes) normally responsible for O-GlcNAc removal
(12
, 13
, 86)
. This has proved difficult since
hexosaminidases exist in the lysosomes of cells that are released upon
subcellular fractionation. One candidate enzyme that removed O-GlcNAc
was purified from the cytosolic fraction of rat spleen. This
preparation was shown to consist of a 51 and 54 kDa subunit
(12)
. More recently, candidates for the O-GlcNAcase have
been examined in a number of tissues, including the brain (13
, 86
, 90)
. This latter activity has been attributed to a
previously identified protein originally thought to be a hyaluronidase
associated with meningioma (13
, 91)
. This information
suggests that O-GlcNAcase is a 103 kDa protein with an acidic
isoelectric point (13
, 91)
. Analysis of the protein
structure shows no obvious protein motifs. The only similarity
occurring with a probability greater than chance is the BglG
anti-terminator motif from E. coli. This RNA binding protein
prevents transcription termination of the bgl operon that is regulated
by ß-glucosides. The significance of this similarity is not yet
known. The O-GlcNAcase also shows limited homology (near its amino
terminus) to Clostridium perfringens Mu toxin and to a
lacto-N-biosidase precursor from Streptomyces. This sequence
similarity may reflect the position of the catalytically important
residues in the molecule. Another aspect of the regulation of the
putative O-GlcNAcase is the existence of at least two
isoforms in the brain that are produced by alternative splicing.
Both variants have the same amino terminus. This observation, coupled
with the sequence similarities mentioned above, allows us to
tentatively assign a region corresponding to 63–283 in the amino
terminus as the catalytic domain of O-GlcNAcase (Fig. 5
).
A previous analysis of the tissue distribution of this transcript
suggests that it is evolutionarily highly conserved and enriched in the
brain, skeletal muscle, and pancreas (91)
. This is similar
to the known distribution of OGT transcripts (26
, 27)
. The
gene is present on human chromosome 10 (91)
.
Inhibitors of O-GlcNAcase
The best-characterized inhibitor of the O-GlcNAcase is
O-(2-acetamido-2-deoxy D-glucopyranosylidene)amino-N-phenylcarbamate,
also called PUGNAc (65)
. This is a nontoxic compound that
leads to the accumulation of O-GlcNAc-modified proteins in treated
cells. This increase can be as much as twofold in certain cell types.
Another, less potent inhibitor of the O-GlcNAcase is streptozotocin.
This O-GlcNAc analog is an irreversible inhibitor of the enzyme and can
lead to a dramatic accumulation of O-GlcNAc in certain target tissues
such as the pancreas (72
, 86
, 90
, 92
, 93)
. Streptozotocin
has long been thought to act by inducing fragmentation of DNA in the
ß cells of the islet. Some evidence points to the inhibition of
O-GlcNAcase as another potential mechanism for selective ß cell
destruction in animals treated with streptozotocin (72
, 86
, 92
, 93)
. However, this hypothesis has recently been challenged
(90)
. Future work on these inhibitors may provide
additional insight into to the role of O-GlcNAcase in O-GlcNAc
metabolism.
|
THE HEXOSAMINE BIOSYNTHETIC PATHWAY: GLUCOSE TO UDP-HEXOSAMINES
|
|---|
The pathway through which UDP-GlcNAc and UDP-GalNAc are
synthesized from glucose is often termed the hexosamine biosynthetic
pathway. Only a small percentage of the glucose entering cells is
routed to this pathway, yet flux through the pathway is thought to be
regulated largely by the levels of glucose and the rate-limiting enzyme
GFAT (glutamine:fructose-6-phosphate amidotransferase). An intriguing
aspect of hexosamine biosynthesis is its apparent link to cellular
glucose sensing pathways, particularly that of insulin
(94
95
96
97)
. Overexpression of GFAT has been found to
recapitulate many of the physiological features of insulin resistance
in vertebrates such as glucose transporter translocation and
transcriptional effects (98
99
100
101
102
103
104)
. Figure 6
illustrates the suspected link between the hexosamine biosynthetic
pathway and the glycosylation of substrates by OGT. In this model,
glucose entering cells is routed via the hexosamine biosynthetic
pathway into the production of UDP-GlcNAc. This elevated UDP-GlcNAc has
two effects on the pathway: 1) UDP-GlcNAc feedback inhibits
GFAT, thus blunting synthesis, and 2) it accelerates the
glycosylation of proteins by OGT. The range of concentrations over
which OGT acts matches the cytosolic concentrations of sugar
nucleotides (in the micromolar range) in this model. One problem with
this type of regulation is that the true cytoplasmic concentration of
UDP-GlcNAc has not been determined. The bulk of the sugar nucleotide is
transported into the endomembrane system, where it is involved in
membrane and secretory glycoprotein biosynthesis (86
, 105)
. Another complication is the inhibition of GFAT by excess
nucleotide that would severely limit the range over which UDP-GlcNAc
concentrations might be expected to accumulate. Despite these
complications, a large body of evidence argues that this pathway plays
some role in signal transduction (20
, 86
, 92
, 93
, 98
, 100
101
102
103
104
105
106
107
108
109)
. The extent to which these biological phenomena can be
explained by addition of O-GlcNAc to target substrates remains to be
determined.
View larger version (13K):
[in this window]
[in a new window]
|
Figure 6. The hexosamine biosynthetic pathway and O-GlcNAc-dependent signaling.
One intriguing hypothesis is that the pathway by which glucose is
converted to UDP-hexosamine is part of a signaling cascade. A fraction
of the glucose entering cells is converted to UDP-GlcNAc through a
series of interconversions involving glutamine and catalyzed in part by
the enzyme GFAT (glutamine:fructose-6-phosphate amidotransferase). The
levels of UDP-GlcNAc would then reflect the flux through the pathway
and could serve as a glucose sensor or integrator. UDP-GlcNAc levels
would fluctuate in response to glucose and glutamine levels, leading to
differential activity of OGT within a physiologically range dictated by
the transferase’s Km with respect to the sugar
nucleotide (right panel). Increased OGT activity would lead to enhanced
O-GlcNAc levels on physiologically relevant substrates. The
physiological manifestations of elevation of UDP-GlcNAc levels are
summarized in the text and include the phenomenon known as ‘insulin
resistance’.
|
|
|
EVIDENCE FOR AN O-GlcNAc-DEPENDENT SIGNALING PATHWAY
|
|---|
In addition to the anecdotal evidence implicating hexosamine
biosynthesis in insulin resistance, other features of the modification
suggest a role in signal transduction. First, there is the obvious
analogy with phosphorylation. Like phosphorylation, O-GlcNAc addition
is a dynamic modification (16
, 19
, 67
, 83
, 110
, 111)
. Most
of the proteins bearing O-GlcNAc are also phosphorylated such that
distinct isoforms bearing different constellations of modifications
have been documented extensively (17
18
19
, 21
, 23
, 54
, 67
68
69
, 76
, 79
, 112
113
114
115
116
117
118)
. The substrate specificity of OGT with respect to
peptide recognition is similar to certain kinases such as GSK-3. Taken
together, these observations suggest that O-GlcNAc addition may be an
integral component of certain signaling cascades, much like
phosphorylation. This hypothesis is further strengthened by certain
experiments that appear to link the glycosylation of certain
transcription factors with an alteration in transcription (8
, 17
18
19
20
, 25
, 68
, 70
, 71
, 119)
. At least one form of bacterial
toxin has striking similarity to OGT and may represent an O-linked
GlcNAc transferase (120
, 121)
. The Clostridium
novyi toxin, like many bacterial toxins, interferes with a
normal cellular mechanism leading to selective cell toxicity in this
modification of the Rho proteins (120
, 121)
.
The most direct evidence that O-GlcNAc addition plays a role in
signaling comes from studies with the higher plant
Arabidopsis. One of the key regulators of plant growth and
development is gibberellin, a hormone that promotes cell elongation and
modulates germination and flowering. A mutation termed Spindly (Spy) is
a recessive mutation leading to a constitutive activation of the
gibberellin signaling pathway (40
, 122)
. Thus, the genetic
evidence suggests that SPY encodes a negative regulator of GA signal
transduction acting early in the pathway (123)
. The
Spindly protein was shown to have a great deal of similarity to OGT
(26
, 27)
. Although it has not yet been demonstrated to
mediate O-Linked GlcNAc transfer, the genetic and sequence data
implicating OGT in this signal transduction cascade are compelling.
|
GLYCAN-DEPENDENT SIGNALING AND HUMAN DISEASES
|
|---|
Neurodegenerative diseases: a role for O-linked GlcNAc?
The enzymes of O-GlcNAc metabolism are enriched in the brain and
O-GlcNAc is abundant in that tissue (26
, 27
, 91)
. Many
proteins important for neuronal function are modified by O-GlcNAc
including neurofilaments, clathrin assembly proteins, and the
ß-amyloid precursor protein (76
, 78
, 79
, 110
, 115
, 116
, 118
, 124
, 125)
. In Alzheimer brains, Tau is hyperphosphorylated and
AP-3 is underglycosylated (76
, 115
, 116
, 124)
. Since
glucose metabolism is essential for neural function, this tissue may be
particularly susceptible to lowered glycosylation by the O-GlcNAc
pathway. This could lead to the altered post-translational
modifications observed in Alzheimer’s and could be part of the
pathological manifestation of the disease. The distribution of OGT
transcripts in the brain is certainly not uniform (S. M. Sato,
J. A. Hanover, and Z. Lai, unpublished results), making certain
neurons more susceptible to such damage. Another intriguing finding is
the observation that X-linked Parkinson dystonia maps to the location
of the X chromosome corresponding roughly to OGT (J. A.
Hanover, and S. Yu, unpublished results; ref 30
). This
disorder, also termed Lubing, has been extensively studied recently and
is one of the hereditary movement disorders. It will be of interest to
determine whether the pathology correlates with altered O-GlcNAc
metabolism (126
127
128
129)
.
Diabetes mellitus and O-linked GlcNAc
As described above, several lines of indirect evidence points to a
role for O-linked GlcNAc in mammalian insulin resistance. There is also
evidence that O-linked GlcNAc may be part of the glucose sensing
mechanism in the ß cells of the pancreas. Transcripts encoding OGT
are enriched in the ß cells (86
, 92)
. In addition,
glucosamine can blunt insulin secretion under some circumstances
(130)
. Streptozotocin, a ß cell toxin, has recently been
shown to inhibit O-GlcNAcase; the resulting elevation of O-GlcNAc
levels has been suggested to be part of the mechanism of selective ß
cell destruction (86
, 92)
. The hyperglycemia resulting
from NIDDM could also lead to a chronic elevation of GlcNAc levels and
result in ß cell loss. The progressive loss of insulin secreting
cells is well documented in NIDDM. Additional studies are required to
determine the extent to which O-GlcNAc metabolism is involved in ß
cell function and in insulin action in peripheral tissues.
Cancer, tumor suppression, and O-linked GlcNAc
The striking changes in cell morphology, gene expression, and
energy utilization observed in transformed cells may be associated with
changes in O-GlcNAc metabolism. Proteins known to be oncogenes, such as
c-myc and the SV40 T antigen, are O-GlcNAc modified
(17
, 18
, 131)
. The c-myc transcription factor
is involved in transcriptional regulation of some glucose-responsive
genes. In the case of the myc oncogene, the glycosylation and
phosphorylation occur at residue Thr 58 in the transactivation domain,
a mutational hotspot for Burkitt’s lymphoma. The problem in
interpreting much of the mutational analysis performed on oncogenes
such as myc is that any given site can bear both phosphate and GlcNAc.
Thus, the mutational analysis must be interpreted with caution. Like
the oncogenes, tumor suppressor genes may also be modified by O-GlcNAc.
Evidence suggests that p53, the most extensively studied tumor
suppressor, bears O-GlcNAc (132)
. The role of this
modification in the many cellular functions of p53 such as tumor
suppression, apoptosis, and gene regulation is presently unknown.
Tumor cells are known to have dramatically altered glucose metabolism.
Transformed cells switch from oxidative to glycolytic metabolism, where
energy requirements of the rapidly dividing cells are provided by
glycolysis-derived ATP. This phenomena, termed the ‘Crabtree effect’,
has puzzled researchers for years, and many explanations have been
offered for the metabolic transition. Given the central role of glucose
in regulating O-GlcNAc metabolism in cells, future research no doubt
will focus on the significance of O-GlcNAc modification in the Crabtree
effect.
|
SUMMARY AND FUTURE PROSPECTS
|
|---|
A model for O-linked GlcNAc-dependent cell signaling
The molecular characterization of the enzymes involved O-GlcNAc
metabolism, and the growing list of known substrates have led to a
simple model for how O-GlcNAc may function to modulate various cellular
functions (Fig. 7
). The O-GlcNAc transferase and O-GlcNAcase are likely to act in consort
with kinases to generate various isoforms of physiological substrates.
The TPR repeats of OGT, in particular, may be involved in substrate
recognition and perhaps bridging of substrates in larger assemblies.
The catalytic domain of OGT is directly responsive to the levels of
UDP-GlcNAc and perhaps is regulated by autoglycosylation or
phosphorylation of the transferase. The subcellular localization of OGT
may also be modulated; this relocalization may play a key role in OGT
function. The O-GlcNAcase may be subject to additional forms of
regulation including post-translational modifications or changes in
subcellular localization. The result of these complex
interactions would be modulation of such parameters as
protein–protein interactions, protein stability, and perhaps enzymatic
activity of the substrates themselves. Thus, the system would provide a
means for integrating information derived from the levels of key
components of the nutrient sensing hexosamine pathway such as glucose
and glutamine. This model suggests that O-GlcNAc, like phosphorylation,
performs a variety of functions depending on the substrate recognized
and modified.
View larger version (48K):
[in this window]
[in a new window]
|
Figure 7. Summary of the central role of O-GlcNAc metabolism in cell signaling.
This model for how O-GlcNAc may modulate various cellular functions
reflects many of the current findings. The levels of UDP-GlcNAc appear
to be responsive to nutrient concentrations within cells that include
glucose, essential amino acids (glutamine in particular), and fatty
acids. Each has been shown to elevate UDP-GlcNAc levels. Other forms of
regulation may result form the localization or modifications of the
enzymes of O-GlcNAc metabolism, as shown. The enzymes OGT and GlcNAcase
oppose each other but are independently regulated. There is clearly a
dynamic interplay between the action of cellular kinases, phosphatases,
and the enzymes of O-GlcNAc metabolism. Certain kinases are known to be
substrates for OGT; OGT may be modified by specific kinases. This
dynamic interplay creates tremendous potential for intracellular
regulation. Among those cellular functions modulated by O-GlcNAc
metabolism are nuclear transport, transcription, translation, cell
signaling, apoptosis, and cell shape. Each function is described in
greater detail in the text. Current evidence suggests that defects in
O-GlcNAc metabolism are associated with human diseases such as
neurodegeneration, diabetes mellitus, and cancer.
|
|
Future prospects
It is now clear that O-GlcNAc is involved in key cellular events
such as transcription, translation, nuclear transport, and cell
signaling. Dysregulation of O-GlcNAc metabolism is likely to be
involved in certain human diseases. Given these possibilities,
analytical approaches to accelerate research in this area will be
increasingly important. This brief review was written in the hope that
young investigators may find some aspect of glycan-dependent signaling
to be of interest and begin to explore some of these
possibilities.
|
REFERENCES
|
|---|
-
Torres, C. R., Hart, G. W. (1984) Topography and polypeptide distribution of terminal N-acetylglucosamine residues on the surfaces of intact lymphocytes. Evidence for O-linked GlcNAc. J. Biol. Chem 259,3308-3317[Abstract/Free Full Text]
-
Holt, G. D., Haltiwanger, R. S., Torres, C. R., Hart, G. W. (1987) Erythrocytes contain cytoplasmic glycoproteins. O-linked GlcNAc on band. 4.1 J. Biol. Chem. 262,14847-14850[Abstract/Free Full Text]
-
Holt, G. D., Hart, G. W. (1986) The subcellular distribution of terminal N-acetylglucosamine moieties Localization of a novel protein-saccharide linkage, O-linked GlcNAc. J. Biol. Chem 261,8049-8057[Abstract/Free Full Text]
-
Hanover, J. A., Cohen, C. K., Willingham, M. C., Park, M. K. (1987) O-linked N-acetylglucosamine is attached to proteins of the nuclear pore Evidence for cytoplasmic and nucleoplasmic glycoproteins. J. Biol. Chem 262,9887-9894[Abstract/Free Full Text]
-
Holt, G. D., Snow, C. M., Senior, A., Haltiwanger, R. S., Gerace, L., Hart, G. W. (1987) Nuclear pore complex glycoproteins contain cytoplasmically disposed O-linked N-acetylglucosamine. J. Cell Biol. 104,1157-1164[Abstract/Free Full Text]
-
Snow, C. M., Senior, A., Gerace, L. (1987) Monoclonal antibodies identify a group of nuclear pore complex glycoproteins. J. Cell Biol. 104,1143-1156[Abstract/Free Full Text]
-
Park, M. K., D’Onofrio, M., Willingham, M. C., Hanover, J. A. (1987) A monoclonal antibody against a family of nuclear pore proteins (nucleoporins): O-linked N-acetylglucosamine is part of the immunodeterminant. Proc. Natl. Acad. Sci. USA 84,6462-6466[Abstract/Free Full Text]
-
Jackson, S. P., Tjian, R. (1988) O-glycosylation of eukaryotic transcription factors: implications for mechanisms of transcriptional regulation. Cell 55,125-133[Medline]
-
Kelly, W. G., Hart, G. W. (1989) Glycosylation of chromosomal proteins: localization of O-linked N-acetylglucosamine in Drosophila chromatin [published erratum: Cell (1989) Vol. 58, p. 419]. Cell 57,243-251[Medline]
-
Haltiwanger, R. S., Holt, G. D., Hart, G. W. (1990) Enzymatic addition of O-GlcNAc to nuclear and cytoplasmic proteins. Identification of a uridine diphospho-N-acetylglucosamine:peptide beta-N-acetylglucosaminyltransferase. J. Biol. Chem 265,2563-2568[Abstract/Free Full Text]
-
Starr, C. M., Hanover, J. A. (1990) Glycosylation of nuclear pore protein p62. Reticulocyte lysate catalyzes O-linked N-acetylglucosamine addition in vitro. J. Biol. Chem 265,6868-6873[Abstract/Free Full Text]
-
Dong, D. L., Hart, G. W. (1994) Purification and characterization of an O-GlcNAc selective N-acetyl-beta-D-glucosaminidase from rat spleen cytosol. J. Biol. Chem. 269,19321-19330[Abstract/Free Full Text]
-
Gao, Y., Wells, L., Comer, F. I., Parker, G. J., Hart, G. W. (2001) Dynamic O-glycosylation of nuclear and cytosolic proteins: cloning and characterization of a neutral, cytosolic ß-N-acetylglucosaminidase from human brain. J. Biol. Chem. 276,9838-9845[Abstract/Free Full Text]
-
Miller, M. W., Hanover, J. A. (1994) Functional nuclear pores reconstituted with beta 1–4 galactose-modified O-linked N-acetylglucosamine glycoproteins. J. Biol. Chem. 269,9289-9297[Abstract/Free Full Text]
-
Wolff, B., Willingham, M. C., Hanover, J. A. (1988) Nuclear protein import: specificity for transport across the nuclear pore. Exp. Cell Res. 178,318-334[Medline]
-
Kearse, K. P., Hart, G. W. (1991) Lymphocyte activation induces rapid changes in nuclear and cytoplasmic glycoproteins. Proc. Natl. Acad. Sci. USA 88,1701-1705[Abstract/Free Full Text]
-
Chou, T. Y., Hart, G. W., Dang, C. V. (1995) c-Myc is glycosylated at threonine 58, a known phosphorylation site and a mutational hot spot in lymphomas. J. Biol. Chem. 270,18961-18965[Abstract/Free Full Text]
-
Chou, T. Y., Dang, C. V., Hart, G. W. (1995) Glycosylation of the c-Myc transactivation domain. Proc. Natl. Acad. Sci. USA 92,4417-4421[Abstract/Free Full Text]
-
Comer, F. I., Hart, G. W. (1999) O-GlcNAc and the control of gene expression. Biochim. Biophys. Acta 1473,161-171[Medline]
-
Daniels, M. C., Kansal, P., Smith, T. M., Paterson, A. J., Kudlow, J. E., McClain, D. A. (1993) Glucose regulation of transforming growth factor-alpha expression is mediated by products of the hexosamine biosynthesis pathway. Mol. Endocrinol. 7,1041-1048[Abstract]
-
Datta, B., Ray, M. K., Chakrabarti, D., Wylie, D. E., Gupta, N. K. (1989) Glycosylation of eukaryotic peptide chain initiation factor 2 (eIF-2)-associated 67-kDa polypeptide (p67) and its possible role in the inhibition of eIF-2 kinase-catalyzed phosphorylation of the eIF-2 alpha-subunit. J. Biol. Chem. 264,20620-20624[Abstract/Free Full Text]
-
Han, I., Roos, M. D., Kudlow, J. E. (1998) Interaction of the transcription factor Sp1 with the nuclear pore protein p62 requires the C-terminal domain of p62. J. Cell. Biochem. 68,50-61[Medline]
-
Jiang, M. S., Hart, G. W. (1997) A subpopulation of estrogen receptors are modified by O-linked N-acetylglucosamine. J. Biol. Chem. 272,2421-2428[Abstract/Free Full Text]
-
Reason, A. J., Morris, H. R., Panico, M., Marais, R., Treisman, R. H., Haltiwanger, R. S., Hart, G. W., Kelly, W. G., Dell, A. (1992) Locali
TOP
ABSTRACT
O-LINKED N-ACETYLGLUCOSAMINE: A...
