Báo cáo khoa học: The relative contribution of mannose salvage pathways to glycosylation in PMI-deficient mouse embryonic fibroblast cells pdf

The relative contribution of mannose salvage pathways to
glycosylation in PMI-deficient mouse embryonic fibroblast
cells
Naonobu Fujita
1
, Ayako Tamura
1
, Aya Higashidani
1
, Takashi Tonozuka
1
, Hudson H. Freeze
2
and
Atsushi Nishikawa
1
1 Department of Applied Biological Science, Tokyo University of Agriculture and Technology, Japan
2 Tumor Microenvironment Program, Burnham Institute for Medical Research, La Jolla, CA, USA
Eukaryotic cells contain mannose mainly in the form
of N-linked oligosaccharides and glycophospholipid
anchors [1,2]. The only known pathway providing
GDP-mannose for these molecules requires the follow-
ing conversions: mannose 6-phosphate to mannose
1-phosphate to GDP-mannose and dolichyl-P-mannose
[3,4]. In mammals, mannose 6-phosphate can be
formed in several ways. The first, primary, and by far
the best-known, way is via the following conversions:
Glc to Glc6P to Fru6P to mannose 6-phosphate.
Phosphomannose isomerase (PMI) (which catalyzes
the Fru6P to mannose 6-phosphate conversion) has an
important role in this pathway. The second is by direct

phosphorylation of exogenous mannose that is trans-
ported by a mannose transporter [5]. The third is by
direct phosphorylation of endogenous mannose that is
Keywords
congenital disorders of glycosylation; lipid-
linked oligosaccharide; mannose; N-linked
oligosaccharide; phosphomannose
isomerase
Correspondence
A. Nishikawa, Department of Applied
Biological Science, Tokyo University of
Agriculture and Technology, 3-5-8 Saiwai-
cho, Fuchu-shi, Tokyo 183-8509, Japan
Fax: +81 42 367 5705
Tel: +81 42 367 5905
E-mail:
(Received 22 October 2007, revised 6
December 2007, accepted 17 December
2007)
doi:10.1111/j.1742-4658.2008.06246.x
Mannose for mammalian glycan biosynthesis can be imported directly from
the medium, derived from glucose or salvaged from endogenous or external
glycans. All pathways must generate mannose 6-phosphate, the activated
form of mannose. Imported or salvaged mannose is directly phosphory-
lated by hexokinase, whereas fructose 6-phosphate from glucose is con-
verted to mannose 6-phosphate by phosphomannose isomerase (PMI).
Normally, PMI provides the majority of mannose for glycan synthesis. To
assess the contribution of PMI-independent pathways, we used PMI-null fi-
broblasts to study N-glycosylation of DNase I, a highly sensitive indicator
protein. In PMI-null cells, imported mannose and salvaged mannose make

a significant contribution to N-glycosylation. When these cells were grown
in mannose-free medium along with the mannosidase inhibitor, swainso-
nine, to block the salvage pathways, N-glycosylation of DNase I was
almost completely eliminated. Adding 13 lm mannose to the medium
completely restored normal glycosylation. Treatment with bafilomycin A
1
,
an inhibitor of lysosomal acidification, also markedly reduced N-glycosyla-
tion of DNase I, but in this case only 8 lm mannose was required to
restore full glycosylation, indicating that a nonlysosomal source of man-
nose made a significant contribution. Glycosylation levels were greatly also
reduced in glycoconjugate-free medium, when endosomal membrane traf-
ficking was blocked by expression of a mutant SKD1. From these data, we
conclude that PMI-null cells can salvage mannose from both endogenous
and external glycoconjugates via lysosomal and nonlysosomal degradation
pathways.
Abbreviations
CDG, congenital disorder of glycosylation; GFP, green fluorescent protein; LLO, lipid-linked oligosaccharide; MOI, multiplicity of infection;
PMI, phosphomannose isomerase.
788 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS
salvaged from glycoconjugates that are degraded
within the same cell. An additional minor pathway
may be conversion of GDP-fucose to GDP-mannose
via unstable intermediate GDP-4-keto-6-deoxy-man-
nose [6]. It has previously been assumed that most of
the mannose in macromolecules is derived from glu-
cose [7]. This assumption was based on the universal
distribution of PMI [8], and the fact that PMI is essen-
tial for yeast growth in the absence of mannose [9].
However, exogenous mannose can contribute signifi-

cantly to glycosylation in some cells [10].
The congenital disorders of glycosylation (CDGs)
are metabolic syndromes with a wide symptomatology
and severity, stemming from deficient N-glycosylation
of proteins [11]. CDG type I defects are due to insuffi-
cient synthesis or poor transfer of the lipid-linked
oligosaccharide (LLO) precursor sugar chain to pro-
teins, leaving many glycosylation sites unoccupied.
One of these, CDG-Ib, is due to a deficiency in PMI
that reduces the endogeneous production of man-
nose 6-phosphate [11,12]. As these patients can be
treated with dietary mannose supplements, it is evident
that abnormalities in the mannose metabolic pathways
have serious medical consequences. Therefore, there is
currently great interest in understanding mannose
metabolism, including salvage pathways.
Although the contribution of monosaccharide sal-
vage pathways to glycosylation may be substantial
[13–16], mannose salvage pathways have received rela-
tively little attention and have yet to be systematically
investigated [17]. Sources of salvaged mannose and its
relative contribution to glycoprotein synthesis are
poorly understood. The main difficulty in the study of
mannose salvage pathways is that mannose 6-phos-
phate derived from mannose-salvaging pathways are
indistinguishable from those derived from glucose.
We constructed PMI-knockout mice, and PMI-null
cells were established from embryonic cells [18]. To
study intracellular mannose salvage pathways, we used
an N-linked glycosylation assay system using DNase I

[19] in PMI-null cells. Wild-type DNase I has two
potential N-linked glycosylation sites. Although the
N-terminal Asn18–Ala–Thr sequon is fully glycosylat-
ed, the C-terminal Asn106–Asp–Ser sequon differs in
tissues and cultured cells [19]. The occupation of the
C-terminal sequon might reflect the glycosylation capa-
bility of cells. Using PMI-null cells, we can examine the
mannose salvage pathways independently of glucose
interconversion at a physiological concentration of
glucose. In this article, we demonstrate that mannose
salvage pathways make significant contributions to
glycosylation, and that the degradation of glycoconju-
gates occurs mainly at low pH in the lysosomes. We
also show that, when cells were incubated in medium
supplemented with 10% fetal bovine serum, the pre-
dominant source of salvaged mannose was degradation
of endogenous glycoconjugates.
Results
Determination of glycosylation efficiency using
the mutant DNase I expression method
Adenovirus for expression of bovine mutant DNase I
that has only one N-linked glycosylation site was
infected into various cultured cells, and the glycosyla-
tion efficiency of expressed DNase I was determined
[19]. As shown in Fig. 1A, the percentage of glycosy-
lated molecules showed an inherent value depending
on the cells. When another mutant DNase I was
expressed in these cells, e.g. with the Asn106–Ser–Thr
sequon instead of the Asn106–Asp–Ser sequon, the
glycosylation efficiency in these cells was almost

100% (data not shown). We previously reported that
the glycosylation efficiency of the Asn106–Asp–Ser
sequon on DNase I depended on the tissue of
origin [19].
To confirm that the glycosylation percentage of
expressed mutant DNase I reflected the glycosylation
capability of the cells, we next investigated glycosyla-
tion of integrin b
1
. Fibroblasts derived from CDG
patients were infected by adenovirus expressing
mutant DNase I and labeled metabolically with
[
35
S]Met and [
35
S]Cys. After harvesting of the condi-
tioned medium for the DNase I glycosylation assay,
the cells were lysed and integrin b
1
was immunopre-
cipitated. As shown in Fig. 1B, the molecular mass of
the precipitated protein differed according to the type
of CDG. It seems that these differences mainly
depended on the number of N-linked oligosaccha-
rides. It was hard to calculate the exact glycosylation
efficiency of integrin b
1
, however, as integrin b
1

has
more than 10 potential N-glycosylation sites. It is
noteworthy that the percentage of glycosylated
DNase I correlated well with that of the immunopre-
cipitated protein. Thus, the glycosylation efficiency of
this mutant DNase I with the Asn106–Asp–Ser se-
quon seemed to reflect the glycosylation capability of
the cells.
Glycosylation efficiency of mutant DNase I
expressed in CDG-Ib fibroblast cells
We measured the glycosylation efficiency of DNase I
in CDG-Ib fibroblast cells, mouse wild-type fibroblast
cells, and PMI-null embryonic fibroblast cells. CDG-Ib
N. Fujita et al. Intracellular mannose salvage pathways
FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS 789
cells, owing to a substantial deficiency in PMI activity,
do not have enough LLO, which results in unoccupied
N-linked glycosylation sequons. In the absence of
exogenously supplied mannose, GDP-mannose levels
are markedly lower in CDG-Ib cells [20]. In PMI-null
cells, mannose 6-phosphate cannot be supplied from
the glycolytic pathway via PMI, so when PMI-null
cells are cultured in mannose-free medium, mannose
6-phosphate for glycosylation must be salvaged from
elsewhere. To remove the effects of free mannose in
the medium, cells were cultured in MEM supplemented
with 10% dialyzed fetal bovine serum for 24 h, and
glycosylation analyses were then performed. We pre-
dicted that the glycosylation efficiency would be lower
in CDG-Ib cells and decreased markedly in PMI-null

cells. However, CDG-Ib cells maintained sufficient gly-
cosylation ability, and surprisingly, PMI-null cells
showed only a slight reduction in glycosylation effi-
ciency (Fig. 2). From these results, we speculated that
substantial amounts of mannose are supplied by the
salvage pathways.
42 F Ic Ie Ig Ii
1 8 4 2 5 0
70 (%)
72
Huh6 A549
HeLa
COS I
82
62
88 78
(%)
PC12
A

B
1 2 3
82 55 0
]
Fig. 1. Determination of glycosylation efficiency using a mutant
DNase I expression method. Glycosylation efficiencies of mutant
DNase I expressed in various types of cells were determined by
infecting cells with adenovirus carrying mutant DNase I and then
incubated with [
35

S]Met and [
35
S]Cys. (A) The protein was immuno-
precipitated from conditioned medium with rabbit anti-DNase I
serum and protein A–agarose beads, and the eluant was subjected
to SDS ⁄ PAGE. In the first panel, lane 1 is intact immunoprecipitat-
ed protein, and lanes 2 and 3 are endo-b-N-acetylglucosamini-
dase-treated and PNGase F-treated samples, respectively. DNase I
having different types of N-glycan is usually detected as a
broad band [19]. The arrow indicates the location of non-glycosylat-
ed DNase I, and ‘]’ shows the migration of the singly glycosylated
DNase I. The percentage of glycosylated molecules is indicated
under the picture. (B) Upper panel: determination of glycosylation
efficiency of mutant DNase I in several fibroblast cells derived from
CDG type I patients. 42F is a human control fibroblast cell line. The
percentage of glycosylated molecules, quantified by densitometry,
is indicated under the picture. Lower panel: SDS ⁄ PAGE of immuno-
precipitated integrin b
1
derived from several CDG type I cells.
35
S-labeled cells obtained from above experiment were lysed
(100 m
M Tris ⁄ HCl, 150 mM NaCl, 1% NP-40 buffer, pH 8.0, con-
taining Roche protease inhibitor cocktail) and immunoprecipitated
using monoclonal antibody to integrin b
1
and protein A–agarose
beads. The precipitate was subjected to SDS ⁄ PAGE using 7%
polyacrylamide gel.

WT
A
B
PMI-null
Baf. A1
SW
CDG-Ib
W
T PMI-n
u
ll
Glycosylation efficiency (%)
C
D
G
-I
b

Control
0
20
40
60
80
100
Control
SW
Baf. A1
]
]

]
Fig. 2. Effect of swainsonine and bafilomycin A
1
on glycosylation.
CDG-Ib fibroblast cells, mouse wild-type embryonic fibroblast cells
and PMI-null fibroblast cells were incubated for 24 h in MEM sup-
plemented with 10% dialyzed fetal bovine serum with or without
10 l
M swainsonine (SW) or 100 nM bafilomycin A
1
(Baf.A1) prior to
metabolic labeling. The percentages of glycosylated molecules
were analyzed as described in Experimental procedures. (A)
SDS ⁄ PAGE analysis representative of three experiments with simi-
lar results. The arrow indicates the location of nonglycosylated
DNase I. (B) The bands in (A) were quantified by densitometry, and
percentages of the glycosylated molecules were calculated. Values
shown are means of five independent experiments.
Intracellular mannose salvage pathways N. Fujita et al.
790 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS
To investigate the relative contribution of lysosomal
a-mannosidases to mannose salvage pathways in PMI-
null cells, they were treated with either swainsonine or
bafilomycin A
1
, and glycosylation analysis was per-
formed. Swainsonine is an indolizidine alkaloid that
acts as a reversible inhibitor of lysosomal a-mannosi-
dase and of the Golgi complex a-mannosidase II [21].
Bafilomycin A

1
is a highly specific inhibitor of vacuo-
lar-type H
+
-ATPase [22], and inhibits acidification
and degradation in lysosomes of cultured cells [23].
Treatment with these drugs did not affect the percent-
age of glycosylated molecules in normal and CDG-Ib
cells, whereas the values in PMI-null cells were reduced
almost to zero. These results indicate that, in PMI-null
cells, almost all mannose 6-phosphate for glycosylation
is supplied by salvage pathways that involve lyso-
somal a-mannosidases.
Supplemental mannose corrected glycosylation
deficiency
If swainsonine and bafilomycin A
1
block the mannose
salvage pathways, then supplemental mannose should
be sufficient to correct the glycosylation deficiency in
swainsonine-treated or bafilomycin A
1
-treated PMI-
null cells. We thus performed mannose titration
experiments in swainsonine-treated and bafilomycin
A
1
-treated PMI-null cells. Cells were treated with
10 lm swainsonine or 100 nm bafilomycin A
1

along
with the indicated concentration of mannose for 12 h
prior to metabolic labeling. The percentages of
glycosylated molecules were then measured (Fig. 3). As
expected, the glycosylation efficiencies increased in
swainsonine-treated and bafilomycin A
1
-treated PMI-
null cells with increasing mannose concentrations in
the medium. About 13 lm and 8 lm mannose were
sufficient to fully restore glycosylation in swainsonine-
treated and bafilomycin A
1
-treated PMI-null cells,
respectively. From these results, we conclude that the
reduction in glycosylation efficiency in swainso-
nine-treated and bafilomycin A
1
-treated PMI-null cells
is caused by blockade of the mannose salvage path-
ways and not of the glycoprotein maturation steps.
Overexpression of SKD1
E235Q
strongly inhibited
core glycosylation in PMI-null cells
In mannose salvage pathways, important sources of
mannose are probably glycoconjugates, which are
transported to lysosomes by a membrane trafficking
process. To test the involvement of membrane traffick-
ing in mannose salvage pathways, we examined the

effect of dominant-negative SKD1 mutants on the
mannose supply in PMI-null cells. SKD1 is a member
of the ATPase family associated with cellular activities,
and for which the yeast homolog Vps4p has been
shown to be involved in endosomal ⁄ vacuolar mem-
brane transport [24]. Expression of a mutant SKD1
molecule, named SKD1
E235Q
, that lacks ATPase activ-
ity in mammalian cells exerted dominant-negative
effects on various membrane-transport processes that
involve endosomes [25]. As shown in Fig. 4A, when an
adenovirus delivery system was used, almost 100% of
the cell population overexpressed green fluorescent
protein (GFP)–SKD1
E235Q
[26]. When the glycosyla-
tion assay was performed in GFP–SKD1
E235Q
-overex-
pressing cells, the percentage of glycosylated molecules
in wild-type cells was unchanged, but the percentage in
PMI-null
0 2 . 5 5 7 . 5 1 0 12.5 15
2 0 (µ
M)
WT
Man conc.
Baf.A 1
0 1 2. 5 5 7.5

1 0 (µ
M)
PMI-null
WT
Man conc.
SW
A

B
0
20
40
60
80
100
0 5 10 15 20
Mannose conc. (µ
M)
Glycosylation efficiency (%)
PMI-null (SW)
WT (SW)
PMI-null (Baf.A1)
WT (Baf.A1)
Fig. 3. Mannose titration in swainsonine-treated or bafilomycin A
1
-
treated PMI-null cells. PMI-null cells were cultured with 10 l
M
swainsonine or 100 nM bafilomycin A
1

for 12 h prior to labeling.
The indicated concentrations of mannose (0–20 l
M) were also
added to the media simultaneously. (A) Glycosylation analysis. The
data shown are from a single experiment that is representative of
three replicates. The arrow indicates the location of nonglycosylat-
ed DNase I. (B) The graph shows the mannose titration curve in
wild-type (WT) and PMI-null cells.
N. Fujita et al. Intracellular mannose salvage pathways
FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS 791
PMI-null cells was markedly reduced (Fig. 4B). These
results demonstrate that most salvaged mannose comes
from material transported to lysosomes via endosomal
trafficking pathways.
The contribution of serum glycoproteins and
endogenous glycoconjugates to mannose salvage
pathways
When PMI-null cells are incubated in media supple-
mented with 10% fetal bovine serum, the glycoconju-
gates used in mannose salvage pathways could be
derived from either endogenous or exogenous mole-
cules. To identify the relative contributions of exoge-
nous and endogenous glycoconjugates, we performed
the glycosylation assay under serum-free conditions.
To remove glycoconjugates from the culture medium,
10% fetal bovine serum was replaced with 1% BSA or
2% TCH, which is a serum replacement product and
contains extremely low concentrations of glycoprotein.
Cells were incubated for 12 h prior to metabolic label-
ing in MEM supplemented with 1% BSA or 2% TCH,

and the glycosylation assay was then performed. As
shown in Fig. 5A, although glycoconjugates were
almost completely removed from the culture medium,
the percentage of glycosylated molecules in PMI-null
cells was only slightly reduced. We obtained the same
result when preincubation in serum-free medium was
extended to 24 h (data not shown). This indicates that,
when the culture medium does not contain fetal bovine
serum, substantial amounts of mannose can be derived
from endogenous rather than exogenous glycoconju-
gates.
If endogenous glycoconjugates are the main source
of mannose, the percentage of glycosylated molecules
should be reduced with time when PMI-null cells are
cultured in mannose-free medium,. Prior to metabolic
labeling, PMI-null cells were incubated with MEM
supplemented with 10% dialyzed fetal bovine serum
for 0, 12, 24, 36 or 48 h (Fig. 5B), and glycosylation
assays were then performed. As shown in Fig. 5C,
increasing the preincubation time reduced the percentage
0
20
40
60
80
100
120
W
T PMI n
u

ll
MOI = 0
MOI = 1000
MOI = 2000
SKD1
E
23
5
Q

MOI = 1000
SKD1
E
23
5
Q

MOI = 2000
MOI = 0
DIC
Fluorescence
A
B
Glycosylation efficiency (%)
Fig. 4. Overexpression of SKD1
E235Q
inhib-
ited mannose salvage pathways. To gener-
ate the overexpression of SKD1
E235Q

, cells
were coinfected with AxCALSKD
EQ
at an
MOI of 0, 1000, or 2000, and AxCANCre at
an MOI of 150. (A) Twenty-four hours post-
infection, the cells were washed twice with
NaCl ⁄ P
i
and fixed with 4% paraformalde-
hyde for 15 min. Fluorescent images were
obtained with a Zeiss Axiocam controlled by
AXIOVISON software. (B) Prior to metabolic
labeling, the cells were incubated with
MEM supplemented with 10% dialyzed
fetal bovine serum, containing AxCALSKD
EQ
and AxCANCre at the indicated MOI. The
glycosylation analysis was performed as
described in Experimental procedures. The
results are presented as percentage of
untreated cells, which was defined as
100%.
Intracellular mannose salvage pathways N. Fujita et al.
792 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS
MEM+ 20% FBS
MEM + 10% dialyzed FBS
012243648
60
Pre-incubation time (h)

0
20
40
60
80
100
120
A
B
E
C
D
Control
(10 % dialyzed FBS)
1 % BSA 2 % TCH
Glycosylation efficiency (%)
Glycosylation efficiency (%)
0
0 1020304050
20
40
60
80
100
Pre-incubation time (h)
Fluorescence
Retention time (min)
04
812
16

0 h
24 h
48 h
G0 1 2 3
Glycoprotein Total protein
G. R. 1 0.94 0.64 (Gl
y
co
p
rotein/Total
p
rotein)
F. I. 100 87 56 % 100 93 87 %
M0
24 48 h
M0
24 48 h
82 k
42 k
180 k
Fig. 5. The contribution of serum glycoproteins in culture medium to N-glycosylation. (A) To remove glycoproteins from the culture medium,
10% dialyzed fetal bovine serum was replaced with 1% BSA or 2% TCH. PMI-null cells were incubated for 12 h in MEM supplemented with
1% BSA or 2% TCH prior to metabolic labeling. The percentages of glycosylated molecules were analyzed as described in Experimental proce-
dures. The number of glycosylated molecules is presented as percentage of the value in 10% dialyzed fetal bovine serum. (B) Scheme of the
time course for an experiment. After 0, 12, 24, 36 or 48 h of incubation with MEM supplemented with 10% dialyzed fetal bovine serum (indi-
cated with dotted line), the glycosylation analyses were performed. (C) Time course of the reduction in glycosylation efficiency of DNase I in
PMI-null cells. The medium was replaced every 4 h with fresh MEM supplemented with 10% dialyzed fetal bovine serum. Values represent
mean ± SD of three independent experiments. (D) LLO patterns of cells harvested after 0, 24 and 48 h of incubation with MEM supplemented
with 10% dialyzed fetal bovine serum. The cells were collected, and LLOs were labeled with 2-aminopyridine and analyzed using HPLC accord-
ing to the method described in Experimental procedures. Ten per cent volume of the product of each sample was injected into the column. The

arrows indicate the elution positions of standard pyridylaminated oligosaccharides, G0 is Man
9
GlcNAc
2
-PA, and G3 is Glc
3
Man
9
GlcNAc
2
-PA
(2-aminopyrimidine). (E) Glycosylation analysis of total cell protein. About 30% of each sample was harvested after 0, 24 and 48 h of incubation
with 10% dialyzed fetal bovine serum, and subjected to SDS ⁄ PAGE using 12% polyacrylamide gel. The gel was first stained with Pro-Q Emer-
ald 300 for glycoprotein determination, and next with SYPRO Ruby for total protein determination, according to the manufacturer’s instructions.
Total fluorescence intensity (FI) in each lane was calculated; the amount after 0 h of incubation with 10% dialyzed fetal bovine serum was
100%. The total FI of glycoprotein ⁄ total protein is shown as glycosylation ratio (GR). The standard protein (M) molecular masses are indicated.
N. Fujita et al. Intracellular mannose salvage pathways
FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS 793
of glycosylated molecules in PMI-null cells. In
contrast, the percentage of glycosylated molecules in
wild-type cells was unchanged (data not shown). As
the amount of LLO also decreased, depending on the
increase in preincubation time (Fig. 5D), this seems to
be attributable to the lack of mannose salvage and an
inability to derive sufficient mannose from other
sources. We then determined the variation of amount
of glycoprotein in cells. As shown in Fig. 5E, although
the total amount of protein in cells decreased to 87%
after 48 h of incubation, the amount of glycoprotein
decreased more. The results of culturing for 48 h

showed that the percentage of glycosylated protein in
cells decreased to 64% and the total amount of glyco-
protein dropped to nearly half the initial amount. On
the other hand, the amount of glycoprotein in the
medium did not differ substantially from the beginning
to the end of the experiment, because the medium was
changed every 4 h. From these findings, we conclude
that PMI-null cells incubated in medium supplemented
with 10% fetal bovine serum can salvage mannose
from endogenous glycoconjugates, but that the amount
is insufficient to maintain normal protein synthesis
(growth) or normal levels of LLO.
Discussion
In this study, the mannose salvage pathways have been
systematically investigated in PMI-null cells. The four
main findings are as follows: (a) the contribution of
mannose salvage pathways to glycosylation is quite
substantial; (b) glycoconjugate degradation mainly
occurs in lysosomes under low-pH conditions; (c) gly-
coconjugates are transported to lysosomes via endo-
somal trafficking pathways; and (d) when cells are
incubated in medium supplemented with 10% fetal
bovine serum, mannose can be derived from endoge-
nous glycoconjugates.
The unexpected finding that normal and CDG-Ib
cells showed nearly equal levels of glycosylation
(Fig. 2) led us to investigate the mannose salvage path-
way in mammalian cells. PMI-null cells cannot gener-
ate mannose from glucose, and they are forced to rely
on mannose salvage pathways. This allowed us exam-

ine the mannose salvage pathways at physiological
concentrations of glucose.
We consider that the glycosylation analysis method
using DNase I works well because N-glycosylation
may not be crucial for the correct folding of DNase I.
In the pulse-chase experiment, the percentage of gly-
cosylated molecules of DNase I was fairly constant
over the time course of the experiments (data not
shown). We also observed that the percentage of
glycosylated molecules was not affected by treatment
with the proteasome inhibitor MG132 [27] (data not
shown). These results, however, indicate that DNase I
is a glycoprotein, so the glycosylation analysis method
using DNase I may be unaffected by endoplasmic
reticulum quality control mechanisms.
One of the most important findings of this study is
that glycosylation deficiency in swainsonine-treated
PMI-null cells was completely restored by only 13 lm
supplemental mannose (Fig. 3). This concentration is
equivalent to about 25% of the normal blood level of
mannose in humans [28,29]. It has been reported that,
under physiological concentrations of glucose and
mannose, human fibroblasts can derive about 70% of
the mannose in N-linked chains from mannose, and
about 10% of the transported mannose is used for gly-
cosylation, whereas the remainder is isomerized to
Fru6P [10]. Therefore, we consider the results of the
mannose titration to be reasonable.
We then considered the relative contributions of
diet, salvaging and glucose interconversion to glyco-

protein synthesis (Fig. 6). The mannose titration
results and those of a previous report [10] indicate that
a low concentration of mannose is sufficient for proper
glycosylation in fibroblast cells. Furthermore, we also
observed that a substantial amount of mannose can be
Glucose
Mannose
GDP-
-6-P
-6-P
-1-P
Fru-6-P
Lysosome
PMI
PMI
Glycolysis/
TCA cycle
LLO synthesis/
Glycosylation/
Processing
Glycoprotein
degradation
ER
Glucose transporter Mannose transporter
M
M
M
M
M
G

G
G
(A)
(B)
(D)
Oligosaccharide
degradation
(C)
Oligosaccharide
processing
Golgi
M
(C)
Endocytosis
(D1) (D2)
Fig. 6. Metabolic pathways of mannose 6-phosphate in mammalian
cells. Mannose 6-phosphate for N-linked glycosylation may be
obtained via the following routes. (A) From glucose via a pathway
involving PMI. (B) Plasma mannose may be transported inside the
cell. (C) It may be generated in the cytosol, and transported
through the endoplasmic reticulum (ER) and Golgi apparatus via a
degradation and processing pathway. (D) It may be salvaged from
lysosomal degradation of oligosaccharide (D1) and glycoconjugates
via membrane-traffic-dependent salvage pathways (D2).
Intracellular mannose salvage pathways N. Fujita et al.
794 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS
supplied by the salvage pathways in PMI-null cells.
The relative contributions made by exogenous sugar,
salvage pathways and interconversions probably vary
with cell type and amount of glycoprotein synthesized;

therefore, further research is required to clarify these
points.
Although this study was focused on the lysosomal
mannose salvage pathway, which is dependent on
membrane trafficking, there are also two membrane
traffic-independent mannose salvage pathways [30]. The
first involves the free mannose that is generated during
glycoprotein maturation steps. As mammalian cells
generally contain higher proportions of complex oligo-
saccharides than unprocessed high-mannose-type
chains, it is clear that most of the nine mannose residues
initially incorporated into the LLO precursor will be lost
as free mannose. The second involves the free oligo-
saccharides that are generated in the cytosol, and are
products of LLO breakdown, glycopeptides and incor-
rectly folded glycoproteins generated during quality
control screening of the biosynthesis of glycoproteins in
the endoplasmic reticulum [31]. Indeed, the glycosyla-
tion efficiency in PMI-null cells was reduced to almost
zero both by swainsonine and bafilomycin A
1
treatment,
but there were considerable differences in the mannose
titration curves (Fig. 3B). This distinction may be due
to differences in the mechanism of action of swainsonine
and bafilomycin A
1
. Swainsonine inhibits both lyso-
somal and nonlysosomal mannosidases, whereas bafilo-
mycin A

1
blocks only lysosomal mannosidases.
Therefore, some mannose used in glycosylation can
come from membrane traffic-independent salvage path-
ways. Studies are underway to clearly define the relative
contributions of these mannose salvage pathways.
Experimental procedures
Materials
The ViraPower Adenoviral Gateway Expression kit, Lipofec-
tamine 2000, Opti-MEM, LR Clonase Enzyme mix and the
Pro-Q Emerald 300 Glycoprotein Gel Stain with SYPRO
Ruby protein gel stain kit were all purchased from Invitrogen
Life Technologies (Carlsbad, CA, USA). Swainsonine and
antibody to integrin b
1
(SG ⁄ 19) were from Seikagaku Cor-
poration (Tokyo, Japan), bafilomycin A
1
was from Wako
Pure Chemicals (Tokyo, Japan), endo-b-N-acetylglucosa-
minidase and PNGase F were from New England Biolabs
(Ipswich, MA, USA), Pro-mix [
35
S]Met ⁄ Cys labeling mixture
was from GE Healthcare Bioscience (Little Chalfont, UK),
the serum replacement medium TCH was from CELOX
Labolatories, Inc. (St Paul, MN, USA), the C
18
SepPak
column and ENVI-CARB solid-phase extraction tube

were from Waters (Milford, MA, USA) and Supelco (Belle-
fonte, PA, USA) respectively, and fetal bovine serum was
obtained from ICN Biomedicals (Costa Mesa, CA, USA).
Other media and reagents were from Sigma (St Louis, MO,
USA).
Plasmid and adenovirus preparation
Previously, we described an N-glycosylation efficiency assay
[19]. In this work, we used 1-delta wild-type bovine
DNase I, in which Asn18 was exchanged for Glu, with only
one glycosylation site, Asn106–Asp–Ser. To obtain higher
reproducibility, an adenovirus expression system of mutant
DNase I was constructed as follows. Adenoviruses bearing
mutant bovine DNase I was prepared using the ViraPower
Adenovirus Expression System according to the manufac-
turer’s instructions. Briefly, the mutant bovine DNase I
cDNAs were subcloned into the pENTER 1A vector. After
purification of the plasmids, the cDNA inserts were trans-
ferred to the pAd ⁄ CMV ⁄ V5-DEST vector by means of the
Gateway system using LR clonase. The plasmids were puri-
fied and digested with the PacI restriction enzyme. One
microgram of linearized plasmid was diluted with 200 lL
of Opti-MEM, and then mixed with 200 lL of lipofecta-
mine solution, dissolving 4 lL of Lipofectamine 2000 in
Opti-MEM, and transfected into subconfluent 293A cells in
1 mL of Opti-MEM in six-well plates. The 293A cells were
then cultured for 8 days in DMEM supplemented with
10% fetal bovine serum. The medium was replaced every
2 days. When most cells became detached from the plates,
the cells and culture medium were harvested together,
freeze–thawed three times, and centrifuged (1500 g for

20 min) to obtain the adenovirus-enriched supernatants.
Aliquots of the supernatants were then added to fresh
293A cells and cultured for 3 days to amplify adenoviruses.
Viral titers were determined by tissue culture infective dose
50 (TICD
50
) methods with 293A cells.
Cell culture and infection
PMI-deficient mouse embryonic fibroblast cells (PMI-null
cells) [18] and wild-type mouse embryonic fibroblast cells
(WT cells) were grown in DMEM containing 20% fetal
bovine serum. CDG-Ib patient cells were grown in
a-MEM containing 10% fetal bovine serum. All media
were supplemented with 10 UÆmL
)1
penicillin and
100 mgÆmL
)1
streptomycin. The cells were infected with
adenoviruses bearing mutant bovine DNase I as follows.
One day before infection, approximately 2 · 10
5
cells were
plated into six-well plates and incubated at 37 °C for 16 h
in a CO
2
incubator. The medium was replaced with 1 mL
of culture medium containing adenoviruses bearing mutant
bovine DNase I at a multiplicity of infection (MOI) of
300. After incubation for 24 h, the medium containing

N. Fujita et al. Intracellular mannose salvage pathways
FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS 795
adenoviruses was replaced with 1 mL of MEM supple-
mented with 10% dialyzed fetal bovine serum. Following
an additional incubation for 12 h, the medium was again
replaced with 1 mL of MEM supplemented with 10% dia-
lyzed fetal bovine serum. After incubation for another
12 h, the cells were washed twice with NaCl ⁄ P
i
and incu-
bated for 4 h at 37 °C with 0.7 mL of MEM without
Met ⁄ Cys but supplemented with 10% dialyzed fetal bovine
serum and 2.65 MBq of [
35
S]Met ⁄ Cys labeling mixture.
The culture medium was then harvested.
Immunoprecipitation and glycosylation analysis
The extent of glycosylation of mutant DNase I was mea-
sured as previously described [19]. Briefly, the harvested cell
culture medium was incubated with 1 lL of anti-DNase I
serum overnight. Then, 20 lL of a 50% protein A–agarose
bead suspension was added. After 1 h of additional rota-
tion, the beads were washed three times. The immunopre-
cipitated protein was then eluted from the beads in 2X
SDS ⁄ PAGE sample buffer by boiling for 5 min. The eluted
samples were separated by SDS ⁄ PAGE using 12% poly-
acrylamide gel, and the intensities of the bands correspond-
ing to DNase I were quantified using a BAS1000 bioimage
analyzer (Fuji Film Co., Tokyo, Japan).
Glycosidase digestion

Immunoprecipitated DNase I was released from the beads
by boiling for 10 min in 0.5% SDS ⁄ 1.0% 2-mercaptoetha-
nol solution. After centrifugation (15 000 g for 1 min), the
concentrated reaction mixtures were adjusted to contain
50 mm sodium phosphate (pH 7.5) and 1.0% NP-40 for
PNGase F digestion, or 50 mm sodium citrate (pH 5.5) for
endo-b-N-acetylglucosaminidase digestion, according to the
manufacturer’s instructions. PNGase F (2 units) or endo-
b-N-acetylglucosaminidase (2 units) was added, and the
reactions were incubated at 37 °C for 30 min. Following
incubation, the samples were subjected to SDS ⁄ PAGE.
Swainsonine and bafilomycin A
1
treatment
Cells were treated with 10 lm swainsonine, 100 nm bafilo-
mycin A
1
or 20 lm mannose for 24 h prior to metabolic
labeling in 1 mL of MEM supplemented with 10% dialyzed
fetal bovine serum. Metabolic labeling was then performed
in the presence of 10 lm swainsonine, 100 nm bafilo-
mycin A
1
,or20lm mannose, respectively. Glycosylation
analysis was performed as described above.
Mannose titration
Prior to metabolic labeling, cells were incubated with 10 lm
swainsonine or 100 nm bafilomycin A
1
and different

concentrations of mannose (0–20 lm) for 12 h. Metabolic
labeling was also performed in the presence of 10 lm
swainsonine or 100 nm bafilomycin A
1
and the indicated
concentrations of mannose. The glycosylation analysis was
then performed as described above.
GFP–SKD1
E235Q
overexpression
A Cre ⁄ loxP inducible system was utilized to express
SKD1
E235Q
[26], because constitutive expression of
SKD1
E235Q
is toxic for 293A cells, in which recombinant
adenoviruses are grown. Twenty-four hours after infection
with adenoviruses bearing mutant DNase I at an MOI of
300, cells were washed and coinfected with adenoviruses
bearing SKD1
E235Q
(AxCALSKD
EQ
) at an MOI of 0, 500,
1000, or 2000, and Cre recombinase (AxCANCre) at an
MOI of 150. After incubation for 12 h, the medium con-
taining adenoviruses was replaced with 1 mL of MEM
containing 10% dialyzed fetal bovine serum. Following an
additional incubation for 12 h, the glycosylation analysis

was performed as described above.
Fluorescence microscopy
PMI-null cells were grown in eight-well chamber slides
(Nalge Nunc International, Rochester, NY, USA) and stimu-
lated to overexpress SKD1
E235Q
by coinfection with
AxCALSKD
EQ
at an MOI of 0, 1000, or 2000, and AxCAN-
Cre at an MOI of 150. After 24 h of incubation, the cells
were washed twice with NaCl ⁄ P
i
and fixed with 4% parafor-
maldehyde in NaCl ⁄ P
i
for 15 min. The cells were then
analyzed by fluorescence microscopy using a Zeiss Axiovert
200 (Carl Zeiss Inc., Thornwood, NY, USA).
Measurement condition in serum-free media
TCH is a serum replacement product containing
0.6 mgÆmL
)1
protein and no sugars. Cells were preincubated
for 12 h at 37 °C in 1 mL of MEM supplemented with 1%
BSA or 2% TCH. Metabolic labeling was then performed in
MEM supplemented with 1% BSA or 2% TCH, respectively.
Prolonged incubation with mannose-free media
After incubation for 0, 12, 24, 36 or 48 h in MEM supple-
mented with 10% dialyzed fetal bovine serum, the glycosyl-

ation efficiency of DNase I and total cell protein were
analyzed. In this experiment, medium was replaced every
4 h with fresh MEM supplemented with 10% dialyzed fetal
bovine serum. For measurement of glycosylation of total
cell protein, collected cells were lysed by 8 m urea contain-
ing 2% Chaps, and then each lysate was subjected to
SDS ⁄ PAGE using 12% polyacrylamide gel. First, glycopro-
teins were stained with Pro-Q Emerald 300, and second,
Intracellular mannose salvage pathways N. Fujita et al.
796 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS
total proteins were stained with the SYPRO Ruby protein
gel stain kit according to the manufacturer’s instructions.
Using a Lumino LAS-3000 imaging analyzer and multi
gauge v2.1 software (Fuji Film), the stained proteins were
imaged.
Analysis of LLOs
Approximately 1.0 · 10
7
harvested cells were suspended
in methanol and dried under N
2
. Afterwards, LLOs
were extracted as previously described [32]. Briefly, LLOs
were extracted in chloroform ⁄ methanol ⁄ water (CMW;
10 : 10 : 3), and the materials in the CMW extract were
treated with weak acid to generate soluble oligosaccha-
rides. The hydrolysates were then loaded onto a C
18
SepPak column directly connected to a 3 mL ENVI-CARB
solid-phase extraction tube to remove residual salt and lip-

ids, as previously described [33]. After loading of the sam-
ple, the columns were washed with 9 mL of 2%
acetonitrile ⁄ 0.1 m ammonium acetate in H
2
O. For elution
of the oligosaccharides, the C
18
SepPak column was
removed and the oligosaccharides were eluted from the
ENVI-CARB tube with 6 mL of H
2
O ⁄ acetonitrile (3 : 1,
v ⁄ v). Then, the dried samples were labeled with 2-amino-
pyridine for HPLC analysis. Pyridylamination was per-
formed as described previously [34]. Pyridylaminated
oligosaccharides were further purified with a Cellulose
Cartridge Glycan preparation kit (Takara Bio Inc., Shiga,
Japan), and separated by HPLC using an Asahipak NH
2
P-
50 4D column (150 · 4.6 mm; Shodex, Showa Denko KK,
Tokyo, Japan). Solvent A was 97% acetonitrile adjusted to
pH 7.0 with 0.3% acetic acid. Solvent B was 10% acetoni-
trile adjusted to pH 7.0 with 0.3% acetic acid. Gradient
conditions were a linear gradient of 70% solvent A and
30% solvent B to 40% solvent A and 60% solvent B over
20 min at a flow rate of 0.8 mLÆmin
)1
, followed by 5 min
of 40% solvent A and 60% solvent B to 100% solvent B

at the same flow rate. Elution was monitored by fluores-
cence (excitation wavelength, 320 nm; emission wavelength,
400 nm). Each peak was identified by comparison with a
standard pyridylaminated oligosaccharide elution time.
Acknowledgements
We express our gratitude to Dr Tamotsu Yoshimori
for providing adenoviruses bearing Cre-recombinase
and GFP–SKD1
E235
, to Dr Sumihiro Hase for pro-
viding standard pyridylaminated oligosaccharides, and
to Mr Kazuyuki Iimura for technical assistance. To
Dr Nobuhiro Takahashi for permission to use his
laboratory for fluorescence microscopy. This work
was supported by CREST of JST, the National Insti-
tutes of Health grant R01 DK55695, and the Rocket
Williams Fund (HHF).
References
1 Kornfeld R & Kornfeld S (1985) Assembly of aspara-
gine-linked oligosaccharides. Annu Rev Biochem 54,
631–664.
2 Menon AK (1994) Structural analysis of glycosylphos-
phatidylinositol anchors. Methods Enzymol 230, 418–
442.
3 Varki A (1991) Radioactive tracer techniques in the
sequencing of glycoprotein oligosaccharides. FASEB J
5, 226–235.
4 Varki A (1994) Metabolic radiolabeling of glycoconju-
gates. Methods Enzymol 230, 16–32.
5 Panneerselvam K & Freeze HH (1996) Mannose enters

mammalian cells using a specific transporter that is
insensitive to glucose. J Biol Chem 271, 9417–9421.
6 Aebi M & Hennet T (2001) Congenital disorders of gly-
cosylation: genetic model systems lead the way. Trends
Cell Biol 11, 136–141.
7 Yurchenco PD, Ceccarini C & Atkinson PH (1978)
Labeling complex carbohydrates of animal cells with
monosaccharides. Methods Enzymol 50, 175–204.
8 Gracy RW & Noltmann EA (1968) Studies on phos-
phomannose isomerase. J Biol Chem 243, 3161–3168.
9 Payton MA, Rheinnecker M, Klig LS, De Tiani M &
Bowden E (1991) A novel Saccharomyces cerevisiae
secretory mutant possesses a thermolabile phosphoman-
nose isomerase. J Bacteriol 173, 2006–2010.
10 Panneerselvam K, Etchison JR & Freeze HH (1997)
Human fibroblasts prefer mannose over glucose as a
source of mannose for N- glycosylation. J Biol Chem
272, 23123–23129.
11 Marquardt T & Denecke T (2003) Congenital disorders
of glycosylation: review of their molecular bases, clinical
presentations and specific therapies. Eur J Pediatr 162,
359–379.
12 Freeze HH & Westphal V (2001) Balancing N-linked
glycosylation to avoid disease. Biochimie 83, 791–799.
13 Rome LH & Hill DF (1986) Lysosomal degradation of
glycoproteins and glycosaminoglycans. Efflux and recy-
cling of sulphate and N-acetylhexosamines. Biochem J
235, 707–713.
14 Trujillo JL & Gan JC (1973) Glycoprotein biosynthesis:
VI. Regulation of uridine diphosphate N-acetyl-D-glu-

cosamine metabolism in bovine thyroid gland slices.
Biochim Biophys Acta 304, 32–41.
15 Shetlar MR, Capps JC & Hern DL (1964) Incorpora-
tion of radioactive glucosamine into the serum proteins
of intact rats and rabbits. Biochim Biophys Acta 83, 93–
101.
16 Aronson NN & Docherty PA (1983) Degradation of
[6-3H]- and [1-14C]-glucosamine-labeled asialo-alpha
1-acid glycoprotein by the perfused rat liver. J Biol
Chem 258, 4266–4271.
N. Fujita et al. Intracellular mannose salvage pathways
FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS 797
17 Freeze HH (1999) Monosaccharide metabolism. In
Essentials of Glycobiology (Varki A, Cummings R, Esko
J, Freeze HH, Hart G & Marth J, eds), pp. 69–84. Cold
Spring Harbor Press, New York, NY.
18 DeRossi C, Bode L, Eklund EA, Ahang F, Davis JA,
Westphal V, Wang L, Borowsky AD & Freeze HH
(2006) Ablation of mouse phosphomannose isomerase
(Mpi) causes mannose 6-phosphate accumulation, toxic-
ity, and embryonic lethality. J Biol Chem 281, 5916–
5927.
19 Nishikawa A & Mizuno S (2001) The efficiency of
N-linked glycosylation of bovine DNase I depends on
the Asn-Xaa-Ser ⁄ Thr sequence and the tissue of origin.
Biochem J 355, 245–248.
20 Rush JS, Panneerselvam K, Waechter CJ & Freeze HH
(2000) Mannose supplementation corrects GDP-man-
nose deficiency in cultured fibroblasts from some
patients with congenital disorders of glycosylation

(CDG). Glycobiology 10, 829–835.
21 Elbein AD (1987) Inhibitors of the biosynthesis and
processing of N-linked oligosaccharide chains. Annu
Rev Biochem 56, 497–534.
22 Bowman EJ, Siebers A & Altendorf K (1988) Bafil-
omycins: a class of inhibitors of membrane ATPases
from microorganisms, animal cells, and plant cells. Proc
Natl Acad Sci USA 85, 7972–7976.
23 Yoshimori T, Yamamoto A, Moriyama Y, Futai M &
Tashiro Y (1991) Bafilomycin A1, a specific inhibitor of
vacuolar-type H(+)-ATPase, inhibits acidification and
protein degradation in lysosomes of cultured cells.
J Biol Chem 266, 17707–17712.
24 Babst M, Sato TK, Banta LM & Emr SD (1997)
Endosomal transport function in yeast requires a novel
AAA-type ATPase, Vps4p. EMBO J 16, 1820–1831.
25 Yoshimori T, Yamagata F, Yamamoto A, Mizushima
N, Kabeya Y, Nara A, Miwako I, Ohashi M, Ohsumi
M & Ohsumi Y (2000) The mouse SKD1, a homologue
of yeast Vps4p, is required for normal endosomal traf-
ficking and morphology in mammalian cells. Mol Biol
Cell 11, 747–763.
26 Nara A, Mizushima N, Yamamoto A, Kabeya Y,
Ohsumi Y & Yoshimori T (2002) SKD1 AAA ATPase-
dependent endosomal transport is involved in autolyso-
some formation. Cell Struct Funct 27, 29–37.
27 Lee DH & Goldberg AL (1998) Proteasome inhibitors:
valuable new tools for cell biologists. Trends Cell Biol
8, 397–403.
28 Pitkanen E & Kanninen T (1994) Determination of

mannose and fructose in human plasma using deute-
rium labelling and gas chromatography ⁄ mass spectrom-
etry. Biol Mass Spectrom 23, 590–595.
29 Etchison JR & Freeze HH (1997) Enzymatic assay of
D-mannose in serum. Clin Chem 43, 533–538.
30 Winchester B (2005) Lysosomal metabolism of glyco-
proteins. Glycobiology 15, 1R–15R.
31 Spiro RG (2004) Role of N-linked polymannose oligo-
saccharides in targeting glycoproteins for endoplasmic
reticulum-associated degradation. Cell Mol Life Sci 61,
1025–1041.
32 Gao N & Lehrman MA (2002) Analyses of dolichol
pyrophosphate-linked oligosaccharides in cell cultures
and tissues by fluorophore-assisted carbohydrate elec-
trophoresis. Glycobiology
12, 353–360.
33 Grubenmann CE, Frank CG, Hu
¨
lsmeier AJ, Schollen
E, Matthijs G, Mayatepek E, Berger EG, Aebi M &
Hennet T (2004) Deficiency of the first mannosylation
step in the N-glycosylation pathway causes congenital
disorder of glycosylation type Ik. Hum Mol Genet 13,
535–542.
34 Yanagida K, Natsuska S & Hase S (1999) A pyridy-
lamination method aimed at automatic oligosaccharide
analysis of N-linked sugar chains. Anal Biochem 274,
229–234.
Intracellular mannose salvage pathways N. Fujita et al.
798 FEBS Journal 275 (2008) 788–798 ª 2008 The Authors Journal compilation ª 2008 FEBS