| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
(Received for publication,
January
13, 1995; and in revised form,
March 20, 1995 ) From the
Keratin polypeptides 8 and 18 (K8/18) are intermediate filament
phosphoglycoproteins that are expressed preferentially in glandular
epithelia. We previously showed that K8/18 phosphorylation occurs on
serine residues and that K8/18 glycosylation consists of
O-linked single N-acetylglucosamines
(O-GlcNAc) that are linked to Ser/Thr. Since the function of
these modifications is unknown, we sought as a first step to identify
the precise modification sites and asked if they play a role in keratin
filament assembly. For this, we generated a panel of K18 Ser and Thr
Keratin intermediate filaments (IF),
With regard to IF phosphorylation, there are accumulating
data that this modification plays an important role in filament
reorganization (reviewed in Refs. 23 and 24). In contrast with IF
phosphorylation, IF glycosylation is a more recently recognized
modification that has been well characterized on K13
(10) ,
K8/18
(11) , and neurofilaments
(25) . IF glycosylation
consists of single N-acetylglucosamine (GlcNAc) residues that
are O-linked to serine/threonine (Ser/Thr). This modification
was first described in mouse lymphocytes (26), and, since then has been
shown to be broadly distributed in many nuclear and cytoplasmic
proteins (reviewed in Refs. 27 and 28). The function of this
modification is unknown, but all proteins identified with this
modification share the feature of forming multimeric
complexes
(27, 28) . This suggests that this modification
plays a potential role in protein-protein interaction. The likely
importance of the O-GlcNAc modification is its dynamic
nature
(11, 29) , which together with its simplicity and
potential attachment sites (Ser/Thr) make it similar to
phosphorylation
(27, 28) . Interestingly, the K8 and K18
molecules that are phosphorylated are for the most part not
glycosylated and vice versa, which suggests that each modification may
have different regulatory functions or that one modification may block
or regulate occurrence of the other
(12) . However, in the case
of K8/18, several lines of evidence suggest that the latter possibility
is unlikely
(7) .
In this study, we used a mutational approach
coupled with manual Edman degradation to identify the major
glycosylation sites of human K18 and showed that mutation of these
sites does not appear to play a significant role in filament assembly
in transfected cells. Our study is based on the following earlier data.
(i) K18 glycosylation occurs within the N-terminal 125-amino acid
domain which ends with a biochemically cleavable tryptophan
(Fig. 1)
(12) . (ii) Human K8/18 can be efficiently
expressed in insect Sf9 cells, using baculovirus recombinants, to allow
for detailed biochemical studies. In addition, K18 displays highly
conserved glycosylation in the insect and human cell
systems
(30) . (iii) We have successfully used a similar
mutational approach to identify the major phosphorylation site of human
K18
(7) .
Figure 1:
Amino
acid sequence and tryptic peptides of K18 head and proximal rod
domains. Single-letter abbreviations are used to show the
amino acid sequence of human K18 beginning with the acetylated serine.
Brackets enclose predicted trypsin digestion products. Serine
residues are shown in bold italics and are numbered
consecutively above the brackets, with numbers below indicating the
amino acid position. The beginning of the rod domain and end
of the head domain are also shown. Asterisks highlight the eight threonine residues.
Tryptic glycopeptide mapping was
done using labeled K18 which was isolated from K8/18 immunoprecipitates
using preparative SDS-polyacrylamide gel electrophoresis (SDS-PAGE).
The K18 bands were visualized by brief Coomassie Blue staining,
followed by electroelution and then acetone precipitation. Keratin
precipitates were digested with trypsin (tosylphenylalanyl chloromethyl
ketone-treated) (375 µg/ml in 80 µl of 50 mM
NH
Figure 2:
Characterization of the glycosylation of
K18 serine mutants expressed in Sf9 cells. K8/18 immunoprecipitates
were obtained from Sf9 insect cells infected with K8/18 recombinant
baculovirus. The immunoprecipitates were labeled with
UDP-[
Table I
In order to assign the three major tryptic glycopeptides shown in
Fig. 2A to specific K18 tryptic peptides, we generated a
series of K18 constructs that have single or multiple mutations at
serine or threonine residues. These mutants were then expressed as
recombinant baculoviruses in Sf9 cells followed by immunoprecipitation
of K8/18, labeling of the K18 glycosylation sites, and then tryptic
peptide mapping to determine which mutation results in the absence of
labeling of the specific peptides. As shown in Table I, mutation
of all the potential threonine glycosylation sites did not affect the
labeling of any of the glycopeptides. Hence, all the threonine mutant
constructs showed K18 tryptic glycopeptide maps that appeared identical
(not shown) with that of the wild type K18 map shown in
Fig. 2A. In contrast, mutation of some of the serine
residues did abolish the glycosylation of specific tryptic peptides
(Table I). For example, constructs that contained mutations of
Ser-12,13, Ser-12-15, and Ser-4,5,9,12-15 showed lack of
labeling of peptide 3 as represented by the map of Ser-12,13 shown in
Fig. 2B. This suggests that glycosylation of peptide 3
occurs on Ser-12 and/or -13. Similarly, constructs that contained
mutations of Ser-7-9 and Ser-7,8,11,14 showed lack of labeling of
peptides 1 and 2 as shown in the maps of Fig. 2D and
Fig. 2C, respectively. This suggests that glycosylation
of peptides 1 and 2 occurs on Ser-7 and/or -8. Confirmation of the
assignment of the peptide numbers shown in Fig. 2,
B-D, was obtained by performing mixture maps with the
wild type construct (not shown). The four minor spots seen in
Fig. 2
, C and D, are background spots that are
also seen in A and B of Fig. 2. Of note, the
construct containing mutations of Ser-7-9 resulted in the
disappearance of spots 1 and 2 (Fig. 2D) even though the
mutations are contained within the same predicted tryptic peptide (see
Fig. 1
). This may be consistent with incomplete trypsin digestion
at Arg-26 or the presence of two forms of the Ser-7-11-containing
peptide (e.g. phosphorylated or other post-translationally
modified forms), or that mutations of Ser-7-9 indirectly alter
the glycosylation at distant sites of other tryptic peptides (see
below).
Figure 3:
Analysis of K18 constructs containing
single or multiple mutations at potential glycosylation sites. The
numbers represent mutated serines shown as consecutive numbers
in Fig. 1. A, K8/18 were immunoprecipitated from Sf9 cells
infected with recombinant wild type K8/18 or WT K8/mutant K18.
Immunoprecipitates were labeled and then analyzed by SDS-PAGE as
described under ``Materials and Methods.'' B,
individual labeled K18 bands were isolated and analyzed by tryptic
peptide mapping. x = origin where samples were spotted
on the cellulose plates. Circled spots represent minor labeled
peptides that are inconsistently seen.
Although the major tryptic glycopeptides of K18 expressed in insect
cells or as found in wild type human colonic tissue cultured cells are
shared, it is possible that variability may occur in the glycosylation
sites between the insect and human systems if more than one serine is
present within a given glycopeptide. We addressed this possibility
using manual Edman degradation of
[
Figure 4:
Identification of the glycosylated
residues of human K18 tryptic glycopeptides using manual Edman
degradation. K8/18 immunoprecipitates from HT29 cells were labeled with
UDP-[
Manual
Edman degradation of the identical radiolabeled peptides isolated from
Sf9 cells gave a pattern very similar to that in Fig. 4except
that cycle 4 of peptides 1 and 2 was 1.6 times higher in counts than
cycle 3 (e.g. 728 versus 467 cpm for peptide 1 of K18
from Sf9 cells, not shown), indicating that the specific activity of
Ser-30 glycosylation is more than Ser-29 in Sf9 cells, while the
opposite is true in HT29 cells. This supports the finding shown in
Fig. 3A, where panel c (i.e. mutation
at Ser-30 of K18) demonstrates slightly lower incorporation of
[
Figure 5:
We also asked if
obliteration of the glycosylation of K18 interferes with the ability of
K8/18 to form filaments after transfection into mammalian cells. As
shown in Fig. 6, transfection of WT K18 or a panel of the
glycosylation K18 mutants with WT K8 into NIH-3T3 or BHK cells resulted
in similar-appearing filaments.
Figure 6:
Immunofluorescence of mammalian cells
transfected with WT K8 and WT K18 or K18 glycosylation mutants. NIH-3T3
or BHK cells grown on coverslips were transiently co-transfected with
WT K8/18 or WT K8 and the indicated K18 single or multiple serine
glycosylation mutants. After 3 days, cells were fixed and analyzed by
immunofluorescence as described under ``Materials and
Methods.'' The numbers represent the mutated serines
shown as consecutive numbers in Fig. 1.
A somewhat unexpected observation in this study is
that mutation of Ser-29,30 which reside in the same tryptic peptide
abolished two radiolabeled spots. Although mutation within one peptide
may indirectly affect glycosylation at a distant site, Edman
degradation release of radioactivity during the third and fourth cycles
for both spots suggests that the two spots correspond to the same
peptide. Additional support for this is based on the sequence of human
K18 (Refs. 34 and 35 and Fig. 1) which indicates that Ser-29 and
Ser-30 are the only Ser-Ser amino acids in K18 that are located at the
third and fourth position of a predicted tryptic peptide. Furthermore,
the third and fourth amino acids of the predicted tryptic peptides of
human K18 lack Ser-Thr or Thr-Ser, and the only Thr-Thr is within the
first predicted tryptic peptide of K18 containing amino acids 1-5
(Fig. 1). However, as shown by our mutational analysis, threonine
glycosylation does not contribute to the three radiolabeled
glycopeptide spots (Table I), and a double mutation of the two
threonines in the first tryptic peptide (SFTTR) does not alter K18
glycosylation (not shown). Therefore, the most likely interpretation of
our results, although not directly proven, is that spots 1 and 2 (
Fig. 2
and Fig. 3) correspond to differently modified forms
of the same peptide. The precise modification that gives rise to two
different forms of the same glycopeptide remains to be determined, but
phosphorylation is a likely candidate. This is based on phosphorylation
of the peptide containing Ser-29,30 (not shown and Ref. 7) and the
slight difference of migration of spots 1 and 2 by electrophoresis
(horizontal direction of the peptide maps shown in Fig. 2and
Fig. 3
) which is consistent with a charge difference.
Previous
indirect biochemical evidence we obtained, based on the generation of
Table II
The sequence
ISVSR is conserved in mouse and human K18 and the sequence PVSSAASV in
human K18 is nearly identical with the corresponding sequence PASSAASV
in mouse K18
(34, 35) . This suggests that mouse K18
glycosylation is likely to be identical with its human counterpart. The
only identified glycosylation sites of intermediate filament proteins
are those shown in Table II. However, glycosylation is likely to
be a modification that is found in a broad range of IF proteins. For
example, the O-GlcNAc type of glycosylation has been
characterized in K13
(10) , and glucosamine was shown to be
incorporated into a number of epidermal keratins
(8, 9) .
In addition, there is some evidence for a porcine lamin A-like protein
that binds to concanavalin A
(41) , although direct evidence for
lamin A glycosylation and its nature are not known. Interestingly, the
sequence
Pro-X
The function of single
O-GlcNAcs remains speculative although several lines of
evidence indicate that it is likely to be important (reviewed in Refs.
27 and 28). (i) It occurs on molecules that clearly have an important
biologic function (Table II), and to that end it may regulate the
function of these proteins. For example, wheat germ agglutinin inhibits
the transcriptional but not DNA binding capacity of Sp1 presumably by
binding the O-GlcNAcs of Sp1 (47). (ii) It is a dynamic
modification that has a higher turnover rate than the turnover of the
K8/18 proteins
(11) , and its levels can change upon cell
activation as shown for mouse T cells treated with concanavalin
A
(29) . (iii) It is a simple modification that may easily come
on/off similar to phosphorylation. In fact, most if not all proteins
with this modification are also phosphorylated. Examples include
K8/18
(12) , eukaryotic peptide chain initiation factor
2-associated p67 polypeptide
(48) , RNA polymerase
II
(49) , clathrin assembly protein AP-3
(50) , and
neurofilaments
(25, 51) . Interestingly, in the cases
that have been studied, it appears that for the most part molecules
that are phosphorylated are not glycosylated and vice versa (12, 48,
49; but see Ref. 50 for exception). This indicates that one
modification may block the other and/or each modification plays
different regulatory functions. For K8/18, it appears that
glycosylation and phosphorylation are not related in several systems
tested. For example, inhibition of basal phosphorylation of K8/18 in
human HT29 cells using staurosporine
(12) , or increased
phosphorylation of K8/18 during the S or G
With regard to K8/18 glycosylation, several potential functions may
be considered. These include a role in protein-protein interaction at
the keratin-keratin level and/or at an associated protein-keratin
level. Such roles are attractive possibilities for the
O-GlcNAc modification in general since most identified
proteins with this modification form multimeric polypeptide
complexes
(27, 28) . The transfection experiments using
wild type or K18 glycosylation mutants did not show an obvious role for
K18 glycosylation in filament assembly (Fig. 6). We cannot
exclude the possibility that K18 glycosylation plays a role in filament
organization not measured by our immunofluorescence assembly assay. The
alternative, that of an associated protein-keratin interaction role, is
attractive based on the location of the identified modifications in
NF-L, M
(25) , and K18 (this report). Hence, glycosylation of K18
and NF-L appear to be restricted to the head domain, and glycosylation
of NF-M appears to be restricted to the head and tail domains. These
are divergent but shared IF protein domains where most of the
structural heterogeneity and presumed tissue specific functions
reside
(15, 16, 17) . Availability of the K18
glycosylation mutants should allow us to begin testing this
possibility.
Two additional functions of K8/18 glycosylation that
may be considered are subcellular localization and degradation
protection roles. For example, prolonged heat stress
(52) or
mitotic arrest
(12) , which ultimately result in cell death, were
associated with increased K8/18 glycosylation and phosphorylation at a
stage prior to eminent cell death. Although it is not possible to
separate glycosylation from phosphorylation effects in these two
systems, glycosylation could protect from protein degradation (reviewed
in Ref. 53) during impending cell death and basal states. Finally,
glycosylated keratin molecules, which appear to be distinct from the
phosphorylated species, may be localized in different subcellular
compartments. For example, site-specific antiphospho glial fibrillary
acidic protein showed that specific sites were phosphorylated within
the spatial region proximal to the cleavage furrow during
cytokinesis
(54) . Although the function(s) of IF protein
glycosylation and the single O-GlcNAc type of modification are
unclear, the abundance of K8/18 coupled with identification of their
glycosylation sites should make addressing functional aspects of this
modification more amenable to study.
Volume 270,
Number 20,
Issue of May 19, 1995 pp. 11820-11827
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
Ala mutants at potential glycosylation sites followed by
expression in a baculovirus-insect cell system. We identified the major
glycosylation sites of K18 by comparing the tryptic
H-glycopeptide pattern of the panel of mutant and wild type
K18 expressed in the insect cells with the glycopeptides of K18 in
human colonic cells. The identified sites occur on three serines in the
head domain of K18. The precise modified residues in human cells were
verified using Edman degradation and confirmed further by the lack of
glycosylation of a K18 construct that was mutated at the molecularly
identified sites then transfected into NIH-3T3 cells. Partial or total
K18 glycosylation mutants transfected into mammalian cells manifested
nondistinguishable filament assembly to cells transfected with wild
type K8/18. Our results show that K18 glycosylation sites share some
features with other already identified O-GlcNAc sites and may
together help predict glycosylation sites of other intermediate
filament proteins.
()
which are preferentially expressed in epithelial
cells
(1, 2) , undergo several post-translational
modifications including phosphorylation (3-7) and
glycosylation
(8, 9, 10, 11, 12, 13) .
Cells in different epithelia express varying complements of keratins
that form obligate heteropolymers in a cell-specific manner. For
example, glandular type ``simple'' epithelia typically
express keratin polypeptides 8 and 18 (K8/18)
(1) . In general,
the function of IF proteins is poorly understood, although at least one
function for epidermal keratins is to provide cells with structural
stability (reviewed in Refs. 14-17). The most striking evidence
for this is the identification of several inherited blistering skin
diseases that are caused by mutations in epidermal keratins (reviewed
in Refs. 18-20). In the case of K8/18, there is no clear function
or human disease association, but gene disruption of mouse K8 resulted
in fetal death with extensive liver hemorrhage
(21) or in
colorectal hyperplasia and rectal prolapse
(22) depending on the
genetic background of the mice. If one accepts the hypothesis that
post-translational modifications of IF proteins are likely to regulate
their function(s), then characterization of these modifications should
provide a handle for studying the function and regulation of these
proteins.
Cells and Reagents
The cell lines used were:
HT29 (human colon), NIH-3T3 (mouse fibroblast), and BHK-21 (hamster
kidney). They were obtained from the American Type Culture Collection
(Rockville, MD) and cultured as recommended by the supplier. Insect Sf9
cells were from Pharmingen (San Diego, CA). Monoclonal antibody L2A1,
which recognizes human K18, was used for immunopurification of
K8/18
(11) . Other reagents used were: uridine diphosphate
(UDP)-[4,5-H]galactose (36.7 Ci/mmol) and
ENHANCE spray (DuPont NEN), galactosyltransferase (Sigma),
and trypsin (Worthington Biochemical Corp.). Construction of Mutants and Baculovirus
Recombinants
The cDNA for K8 and K18 were subcloned into the
pBluescript SK
plasmid
(7) . Site-directed
mutagenesis to generate serine and threonine alanine (Ala)
mutants were carried out using a Transformer
kit
(Clontech). For mutations at adjacent serines or threonines, a single
primer was typically used, whereas multiple primers were used for
mutations at nonadjacent residues. All mutations were confirmed by
sequencing the mutagenized regions. Mutants were subcloned into the
pVL1392 vector for expression in Sf9 cells
(30) or downstream of
the hCMV promoter in the pMRB101 mammalian expression
vector
(31) . Baculovirus recombinants were generated using the
BaculoGold Transfection kit (Pharmingen) exactly as described (30). Immunofluorescence Microscopy
Transfected
mammalian cells were grown on coverslips using 6-well plates. After 3
days of transient transfection, cells were fixed for 3 min in -20
°C methanol and then washed. Cells were then incubated with
monoclonal antibody L2A1 (30 min), washed, and then incubated with
Texas Red-conjugated goat anti-mouse antibody (30 min). Color slides
were taken using Ektochrome Kodak Elite 400 film from which black and
white pictures were generated. Immunoprecipitation and Tryptic Peptide
Mapping
Immunoprecipitation was carried out using Sf9 cells
infected with the recombinant baculovirus constructs (4 days) or
NIH-3T3 cells transfected with the mammalian expression vector
constructs (3 days). Cells were solubilized with 2% Empigen BB in
phosphate-buffered saline containing 5 mM EDTA, 0.1
mM phenylmethylsulfonyl fluoride, 25 µg/ml aprotinin, 10
µM leupeptin, 10 µM pepstatin (45 min, 4
°C)
(32) . The nonsolubilized residual was pelleted, followed
by incubation of the solubilized material with monoclonal antibody L2A1
conjugated to agarose (1-2 h), and then washing. Galactosylation
of the K8/18 immunoprecipitates was carried out using
UDP-[H]galactose and galactosyltransferase
exactly as described
(12) . 
HCO) for 16-24 h, followed by
lyophilization, then analysis in the horizontal dimension by
electrophoresis (pH 1.9 buffer) and in the vertical dimension by
chromatography using cellulose glass plates (10,000-15,000 cpm
loaded per plate) as described
(33) . After being allowed to
air-dry, plates were evenly sprayed with ENHANCE and then
exposed for 7-14 days. Manual Edman Degradation
Individual K18
glycopeptides, isolated from HT29 or Sf9 cells, were water-extracted
from their corresponding spots on the cellulose plates (after
fluorography and using the developed films as templates) as
described
(33) . Prior to extraction, the plates were placed in a
fume hood to allow evaporation of the ENHANCE material. The
extracted peptides were lyophilized and then subjected to manual Edman
degradation exactly as described
(25) . RESULTS
Localization of the Major Tryptic Glycopeptides of
K18
Our previous results showed that human K18 glycosylation in
HT29 colonic tumor cells occurs within the N-terminal 125-amino acid
domain that is generated after tryptophan cleavage using HCl/dimethyl
sulfoxide/HBr (Fig. 1) as described previously
(12) . This
domain (Fig. 1) contains 20 and 8 of the potential 37 serine and
30 threonine glycosylation sites of K18, respectively, based on the
known primary structure of human K18
(34, 35) . Since
initial attempts at obtaining unambiguous sequences after isolating
radiolabeled glycopeptides followed by microsequencing were
unsuccessful in our hands (not shown), we elected to use a molecular
mutational approach to localize the tryptic peptides where K18
glycosylation occurs. To do so, we used a baculovirus expression system
in insect Sf9 cells to generate large quantities of K18 proteins that
were mutated at potential Ser/Thr K18 glycosylation sites. The basis
for this approach is the conserved glycosylation when comparing human
K18 isolated from HT29 cells or expressed in insect cells
(30) .
As described previously
(30) , immunoprecipitation of K18 from
HT29 cells or Sf9 cells, followed by labeling of accessible terminal
GlcNAcs, showed nearly identical tryptic glycopeptides. The tryptic
H-glycopeptide profile of K18 (human or after expression in
Sf9 cells) is exemplified by the profile shown in
Fig. 2A, which shows that K18 contains three major
tryptic glycopeptides. The other unnumbered spots in
Fig. 2A are sometimes seen and correspond to labeled
galactosyltransferase enzyme contamination (not shown and Ref. 30).
H]galactose using galactosyltransferase.
Radiolabeled K18 was eluted from the gel, digested with trypsin, and
then analyzed in the horizontal dimension by electrophoresis and in the
vertical dimension by chromatography as described under
``Materials and Methods.'' A representative group of K18
serine constructs from Table I is shown with numbers corresponding to mutated serines indicated as consecutive numbers
in Fig. 1. Numbered spots correspond to major labeled K18
glycopeptides. Note the absence of peptide 3 in B and peptides
1 and 2 in C and D.
Identification of the Major Glycosylation Sites of Human
K18
The results shown in Table Iand Fig. 2indicated
that one or both serines in the Ser-12,13- and the Ser-7,8-containing
peptides are glycosylated. In order to identify the precise
modification residues, we generated individual and multiple serine
mutants of K18 that cover these potential glycosylation sites. As shown
in Fig. 3A, expression of the mutants in Sf9 cells using
recombinant baculoviruses resulted in significant protein levels as
noted after immunoprecipitation and Coomassie staining of the SDS-PAGE
separated products. Analysis of the glycosylation of the individual
serine K18 mutants: Ser-7, Ser-8, Ser-12, and Ser-13 using
one-dimensional SDS-PAGE was unrevealing except that some decrease in
the glycosylation of the Ser-8 as compared with the wild type (WT) K18
construct was observed (Fig. 3A, lanes
a-e). However, analysis of the tryptic glycopeptide maps of
the individual serine mutants showed that mutation of Ser-13 (i.e. Ser-48 of K18, Fig. 1) abolished glycopeptide 3
(Fig. 3B, compare panels a and b).
Furthermore, although individual Ser-7 and Ser-8 constructs showed
tryptic glycopeptide maps that were identical with wild type K18 (not
shown), the double mutant construct Ser-7,8 (Fig. 3A,
lane f) showed the absence of glycopeptides 1 and 2
(Fig. 3B, panel c) indicating that both serines
are also glycosylated in Sf9 cells (i.e. Ser-29 and -30 of
K18, Fig. 1). Verification that Ser-29,30,48 of K18 are the major
glycosylation sites in Sf9 cells was obtained by analyzing the
glycosylation of the construct Ser-7,8,13 which showed minimal
glycosylation (Fig. 3A, lane g) and lack of
labeling of peptides 1-3 (Fig. 3B, panel
d). Of note, the construct containing Ser-9 mutation had a tryptic
glycopeptide pattern identical with wild type K18 (not shown).
H]galactose-labeled individual glycopeptides of
K18 isolated from human or insect cells. As shown in Fig. 1, the
tryptic peptides containing Ser-29,30 and Ser-48 have five and two
serines, respectively, as potential glycosylation sites in human HT29
cells. Manual Edman degradation of peptides 1 and 2, individually
isolated from HT29 cells, showed that most of the released
[H]Gal-GlcNAc-amino acid label was in cycles 3
and 4 which corresponds to Ser-29 and Ser-30 (Fig. 4, A and B). Similar processing of peptide 3, also isolated
from HT29 cells, showed that most of the released label was in cycle 4
which corresponds to Ser-48 of K18 (Fig. 4C).
H]galactose followed by isolation of K18.
After tryptic peptide mapping, individual labeled peptides were eluted
from the cellulose plates and subjected to manual Edman degradation as
described under ``Materials and Methods.'' The counts
released from each reaction cycle are represented by the black
bars. Peptide numbers correspond to the major glycopeptides shown
in Fig. 2. The serines marked by arrows represent the major
glycosylation sites on individual peptides based on the
counts/min/cycle obtained.
H]galactose than panel b (i.e. mutation at Ser-29 of K18). Mutation of K18 Glycosylation Sites Abolishes
Glycosylation of the Protein Expressed in Mammalian Cells but Does Not
Interfere with Filament Assembly
We compared the glycosylation
of WT K8/18 versus WT K8/glycosylation mutant K18 after
expression in NIH-3T3 cells, in order to further confirm the K18
glycosylation sites identified using the mutational Sf9 cell expression
and biochemical approaches. Transient expression of a K18 Ser-29,30,48
Ala mutant or WT K18, with their obligate heteropolymer WT K8,
in NIH-3T3 cells showed that the mutant construct was not glycosylated
(Fig. 5, compare lanes 3 and 4) although
protein levels of the mutant and wild type K18 were equal
(Fig. 5, lanes 1 and 2).
H-galactosylation of K8/18 in
NIH-3T3 cells expressing WT K8/18 or WT K8 with the K18 glycosylation
mutant. NIH-3T3 cells were transiently transfected with wild type K8/18
(lanes 1 and 3) or wild type K8 and a K18
Ser-29,30,48 Ala mutant (gly-) (lanes 2 and 4). After 3 days, K8/18 were immunoprecipitated,
galactosylated with UDP-[H]galactose, and then
analyzed by SDS-PAGE.
DISCUSSION
Identification of the Major Human K18 Glycosylation
Sites
This study reports the identification of the major sites
of the single O-GlcNAc type of glycosylation for human keratin
18. Several lines of evidence support the assignment of the identified
sites. First, a systematic mutational approach coupled with expression
in insect cells identified the major tryptic glycopeptides that are
shared between K18 isolated from human cells or expressed in insect
cells and indicated that Ser-29,30,48 of K18 are glycosylated when K18
is expressed in insect cells. Second, manual Edman degradation of the
K18 H-labeled glycopeptides isolated from human or insect
cells confirmed the glycosylation positions that were identified using
the mutational approach. Third, expression of the
glycosylation-negative K18 mutant and lack of its glycosylation in
NIH-3T3 cells provided additional evidence for the monosaccharide
attachment sites. 
-aminobutyrate after 
-elimination and then acid hydrolysis of
purified K8/18, suggested that threonine is a likely glycosylation site
of K8/18
(12) . However, the results obtained here unambiguously
identified three serine but no threonine K18 major glycosylation sites.
Our results here do not exclude the possibility that other
glycosylation sites of K18 (serine or threonine) that are not
accessible to labeling in the presence of
UDP-[H]galactose and galactosyltransferase may be
present. How Conserved Are the Human K18 Glycosylation
Sites?
Although a consensus sequence does not exist for the
O-GlcNAc type of cytoplasmic and nuclear protein
glycosylation, the K18 glycosylation sites
(
PVSSAASVY, and 
ISVSRSTS) show some
similarity to already identified sites in other proteins
(Table II). For example, most, but not all, identified sites have
a nearby proline and/or valine. The glycosylated Ser-48 of K18 has an
adjacent valine but no nearby proline. The only other identified
O-GlcNAc site that does not have an adjacent proline is
ANQLTNDY in talin
(36) (Table II). It is not clear yet if
one or more GlcNAc transferases is involved in this type of
glycosylation and if distinct transferases to serine and threonine
residues are present
(37) . To date, only one GlcNAc transferase
has been characterized
(38) . Another common feature of the
identified O-GlcNAc sites is a high density of Ser/Thr
residues in proximity to the modified residue. These features of the
O-GlcNAc type of glycosylation are similar to the general
features of the O-linked mucin type of
glycosylation
(37, 39, 40) .
-X
-X (where one X
= a hydrophobic
residue, and two Xcorrespond to
Ser/Thr residues or Ser/Thr and any other amino acid) is found in the
head domain of a number of intermediate filament proteins. This
includes PLSS in human desmin
(42) , PLSP in human
lamins
(43) , and PAST in keratin 16
(44) . Since the
sequence
Pro-X-X-X is a glycosylation motif in neurofilament L and M and in K18
(Table II), it is possible that similar motifs in other
intermediate filament proteins are also glycosylated. Potential Functions to Consider for K18
Glycosylation
The few available functional data regarding K8/18
glycosylation are in general exclusionary rather than supportive of
several potential tested functional roles. For example, glycosylation
does not appear to play a significant role in keratin solubility since
the soluble and insoluble K8/18 pools had similar specific activities
of glycosylation and similar tryptic glycopeptide maps
(45) . In
addition, although G/M arrest of HT29 cells using
anti-microtubule agents or okadaic acid resulted in a marked increase
in K8/18 glycosylation, analysis of G
/G, S
phase, and G/M phase synchronized cells did not reveal any
changes in K8/18 glycosylation
(46) . /M phases of the
cell cycle
(46) are not associated with altered K8/18
glycosylation. Furthermore, mutation of the major phosphorylation site
of human K18 (Ser-52) does not affect K18 glycosylation
(7) .
)
We are very grateful to Kris Morrow and Sally
Morefield for preparing the figures, to Dr. Dennis Dong for advice
regarding manual Edman degradation of H-peptides, and to
Linda P. Jacob and Theresa L. Hooper for preparing the manuscript.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Slawson, T. Lakshmanan, S. Knapp, and G. W. Hart A Mitotic GlcNAcylation/Phosphorylation Signaling Complex Alters the Posttranslational State of the Cytoskeletal Protein Vimentin Mol. Biol. Cell, October 1, 2008; 19(10): 4130 - 4140. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Hatsell, L. Medina, J. Merola, R. Haltiwanger, and P. Cowin Plakoglobin Is O-Glycosylated Close to the N-terminal Destruction Box J. Biol. Chem., September 26, 2003; 278(39): 37745 - 37752. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Akimoto, H. Kawakami, K. Yamamoto, E. Munetomo, T. Hida, and H. Hirano Elevated Expression of O-GlcNAc-Modified Proteins and O-GlcNAc Transferase in Corneas of Diabetic Goto-Kakizaki Rats Invest. Ophthalmol. Vis. Sci., September 1, 2003; 44(9): 3802 - 3809. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N.-O. Ku, J. Liao, and M. B. Omary Apoptosis Generates Stable Fragments of Human Type I Keratins J. Biol. Chem., December 26, 1997; 272(52): 33197 - 33203. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. S. Haltiwanger and G. A. Philipsberg Mitotic Arrest with Nocodazole Induces Selective Changes in the Level of O-Linked N-Acetylglucosamine and Accumulation of Incompletely Processed N-Glycans on Proteins from HT29 Cells J. Biol. Chem., March 28, 1997; 272(13): 8752 - 8758. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N.-O. Ku and M. B. Omary Phosphorylation of Human Keratin 8in Vivo at Conserved Head Domain Serine 23and at Epidermal Growth Factor-stimulated Tail Domain Serine 431 J. Biol. Chem., March 14, 1997; 272(11): 7556 - 7564. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Goletz, F. Hanisch, and U Karsten Novel alphaGalNAc containing glycans on cytokeratins are recognized invitro by galectins with type II carbohydrate recognition domains J. Cell Sci., January 7, 1997; 110(14): 1585 - 1596. [Abstract] [PDF]
|
|
|
|
|
|
D. L.-Y. Dong, Z.-S. Xu, G. W. Hart, and D. W. Cleveland Cytoplasmic O-GlcNAc Modification of the Head Domain and the KSP Repeat Motif of the Neurofilament Protein Neurofilament-H J. Biol. Chem., August 23, 1996; 271(34): 20845 - 20852. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Eto, S. S. Yoon, B. P. Bode, S. Kamidono, K. Makino, H. Saya, H. Nakamura, and K. K. Tanabe Mapping and Regulation of the Tumor-associated Epitope Recognized by Monoclonal Antibody RS-11 J. Biol. Chem., August 25, 2000; 275(35): 27075 - 27083. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ABSTRACT
