Efficient uptake of recombinant α-galactosidase A produced with a
gene-manipulated yeast by Fabry mice kidneys
Takahiro Tsukimura, 1 Ikuo Kawashima, 2,3 Tadayasu Togawa,1 Takashi Kodama,1
Toshihiro Suzuki,1 Toru Watanabe,4 Yasunori Chiba,2,4 Yoshifumi Jigami,4 Tomoko
Fukushige,5 Takuro Kanekura, 5 and Hitoshi Sakuraba1,2,*
From the Departments of 1Analytical Biochemistry and 2Clinical Genetics, Meiji
Pharmaceutical University, Tokyo, Japan, 3Department of Molecular Medical Research,
The Tokyo Metropolitan Institute of Medical Science, Tokyo Metropolitan
Organization for Medical Research, Tokyo, Japan, 4Research Center for Medical
Glycoscience, National Institute of Advanced Industrial Science and Technology,
Tsukuba, Japan, and 5Department of Dermatology, Kagoshima University Graduate
School of Medical and Dental Sciences, Kagoshima, Japan.
※Correspondence: Hitoshi Sakuraba
Department of Analytical Biochemistry, Meiji Pharmaceutical University,
2-522-1 Noshio, Kiyose, Tokyo 204-8588, Japan.
Tel: +81-42-495-8923; Fax: +81-42-495-8923;
Running headline: Uptake of α-galactosidase A by Fabry mice kidneys
Keywords: Fabry disease, α-Galactosidase A, Yeast, Mannose 6-phosphate, Enzyme
To economically produce recombinant human α-galactosidase A (GLA) with a cell
culture system that does not require bovine serum, we chose methylotrophic yeast cells
with the OCH1 gene deleted and overexpressing the MNN4 gene as a host cell line. The
enzyme (yr-hGLA) produced with the gene-manipulated yeast cells has almost the same
enzymological parameters as those of the recombinant human GLA produced with
cultured human fibroblasts (agalsidase alfa), which is currently used for enzyme
replacement therapy for Fabry disease. However, the basic structures of their sugar
chains are quite different. yr-hGLA has a high content of phosphorylated N-glycans and
is well incorporated into the kidneys, the main target organ in Fabry disease, where it
cleaves the accumulated glycosphingolipids. A glycoprotein production system
involving this gene-manipulated yeast cell line will be useful for the development of a
new enzyme replacement therapy for Fabry disease.
Fabry disease (MIM 301500) is an X-linked inborn error of metabolism with a high
incidence of 1 in 1,250-4,000 male live births (1-3). In Fabry disease, a deficiency of
α-galactosidase A (GLA, MIM 300644, EC 220.127.116.11) activity causes the systemic
accumulation of glycosphingolipids including globotriaosylceramide (Gb3) and
globotriaosylsphingosine (lyso-Gb3), and patients with the disease exhibit progressive
renal, cardiac and vascular disorders (4-7).
So far, two different recombinant human GLAs, agalsidase alfa (Replagal®; Shire
HGT, Cambridge, MA) (8), produced with cultured human fibroblasts, and agalsidase
beta (Fabrazyme®; Genzyme, Cambridge, MA) (9,10), generated by Chinese hamster
ovary (CHO) cells, have been developed, and they are used for enzyme replacement
therapy (ERT) for Fabry disease. These recombinant GLAs are glycoproteins having
both complex type and high-mannose type sugar chains produced by cultured
mammalian cells, and it is thought that they are incorporated into cells mainly via
mannose 6-phosphate (M6P) receptors in many organs (11) except for the liver, in
which their uptake by hepatocytes and Kupffer cells occurs mainly through
asialoglycoprotein receptors and mannose ones, respectively (12). The incorporated
GLAs cleave the glycosphingolipids deposited in Fabry organs, and there have been
many clinical reports describing the efficacy of these recombinant enzymes (13-16).
However, there are some disadvantages of producing recombinant enzymes using
mammalian cells. As mammalian cell cultures usually require bovine serum or a
serum-free synthetic medium, the production of recombinant enzymes is very expensive
and/or careful monitoring for contamination by microbes is required to exclude
infectious diseases such as bovine spongiform encephalopathy (BSE). Furthermore, the
efficacy of uptake of the currently available recombinant GLAs by the kidneys is not so
high (17,18). As renal insufficiency is the most significant disorder determining the
prognosis of Fabry disease, the economical production of a safe recombinant enzyme
that can be highly incorporated into the kidneys is urgently needed.
Previously, we produced a recombinant human GLA using budding yeast
Saccharomyces cerevisiae, and revealed that the enzyme cleaved the Gb3 accumulated
in cultured Fabry fibroblasts and organs from Fabry mice. However, the productivity of
the enzyme with the Saccharomyces cerevisiae strain was very low (0.1 mg per 1 liter
In this study, we used a methylotrophic yeast, Ogataea minuta, in which the OCH1
gene is disrupted (21) and the MNN4 gene is overexpressed (22,23), as host cells to
produce a recombinant GLA having mammalian-like, phosphorylated N-glycans, and
examined its effects on the organs of Fabry mice.
MATERIALS AND METHODS
Plasmid Construction and Transformation of a Methylotrophic Yeast
To prepare a human GLA expression vector for a methylotrophic yeast, O. minuta, an
open reading frame encoding human GLA was amplified using primers Xb-GLA-F
(5’-CCCTCTAGAAAAATGCAGCTGAGGAACC-3’) and Ba-GLA-R (5’-
GGGGGGATCCTTAAAGTAAGTCTT-3’), with pCXN2Gal (24) including the human
GLA cDNA sequence as a template. The PCR fragment obtained was digested with
XbaI and BamHI, and the resultant DNA fragment was inserted between the XbaI and
BamHI sites of the pOMEU1 vector (16). On the other hand, plasmid
pOMEG-ScMNN4 (25) was used for overproduction of ScMnn4p in O. minuta. These
vectors were cut with NotI and used for transformation of the O. minuta TK-3-A strain
(∆och1) (21), which was kindly provided by Kyowa Hakko Kirin (Tokyo, Japan).
Transformation of O. minuta was performed as described previously (21,25). The
correct integration was confirmed by PCR.
Expression and Purification of a Yeast Recombinant Human GLA (yr-hGLA)
The transformed O. minuta was precultured in 100 ml of YPAD broth (2% peptone,
1% yeast extract, 2% glucose, and 0.2 mg/ml adenine) and then transferred to 6 liters of
BMGY broth (6% peptone, 1% yeast extract, 1.34% yeast nitrogen base without amino
acids, 1% glycerol, and 0.1 M potassium phosphate, pH 6.0) in a jar fermentor. When
the glycerol had been completely consumed, methanol was added as a carbon source
and inducer. Methanol induction was performed at 28°C and continued until the GLA
activity in the culture broth reached saturation. The temperature and dissolved oxygen
concentration were monitored and controlled by a computer during fermentation. After
induction, the supernatant of the cultured medium was concentrated by ultrafiltration
(Microza UF; Asahi Kasei Chemicals Co., Tokyo, Japan) and used as the crude enzyme.
Purification of GLA was performed at 4°C, all of the column materials being
purchased from GE Healthcare Bio-Sciences (Piscataway, NJ). The crude enzyme was
precipitated by 55% ammonium sulfate saturation and the precipitate was dissolved in
25 mM 2-(N-morpholino)ethanesulfonic acid (MES) buffer, pH 6.0. A sample was
dialyzed against the same buffer and then applied to a HiLoad 16/10 Q Sepharose HP
column equilibrated with 25 mM MES buffer, pH 6.0. After the column had been
washed, the GLA protein was eluted with a linear gradient of NaCl, from 0 to 1 M, in
the same buffer. The fractions containing the enzyme activity were recovered, and then
ammonium sulfate was added to the solution to a final concentration of 0.3 M. The
resultant solution was applied to a HiLoad 16/10 Phenyl Sepharose HP column
equilibrated with 25 mM MES buffer, pH 6.0, containing 0.3 M ammonium sulfate.
After the column had been washed, the GLA protein was eluted with 25 mM MES
buffer, pH 6.0. The fractions containing the enzyme activity were pooled and
concentrated by ultrafiltration. This solution was applied to a HiLoad 16/60 Superdex
200 pg column using 0.2 M NaCl in the same buffer. The fractions containing the
enzyme activity were pooled. Treatment of the fractions with a culture supernatant of
the Cellulomonas sp. SO-5 strain for exposure of M6P residues on glycans was
performed as described previously (19,25). After the treatment, a sample was further
purified with a Q Sepharose HP column followed by a Superdex one under the same
conditions as those given above, and the active fractions were collected.
Biochemical Analyses of yr-hGLA
To determine the purity and molecular mass of yr-hGLA, sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by staining with
Coomassie brilliant blue R was performed using yr-hGLA treated or not treated with
glycopeptidase F (PNGase F). Deglycosylation of the enzyme was performed with an
enzymatic deglycosylation kit (Prozyme, San Lendodro, CA). Briefly, the enzyme
proteins were electrophoresed on a Tris-glycine polyacrylamide gel and then stained
with Coomassie brilliant blue R. Immunoblotting analysis with anti-GLA antibodies
was performed as described previously (24). As a control, agalsidase alfa was used.
GLA activity was fluorometrically measured with 4-methylumbelliferyl
(MU)-α-D-galactopyranoside (Calbiochem, LaJolla, CA) as a substrate in the presence
of N-acetyl-D-galactosamine (Sigma-Aldrich, St. Louis, MO), a specific inhibitor of
N-acetylgalactosaminidase, in 0.1 M citrate-phosphate buffer, pH 4.6, as described
previously (26). The fluorescence of 4-methylumbelliferone released on the enzyme
reaction was measured with a Wallac 1420 ARVO MX Multilabel Counter
(Perkin-Elmen, Waltham, MA) at excitation and emission wavelengths of 355 nm and
460 nm, respectively. Protein determination was performed with a Micro BCA Protein
Assay Kit (Thermo Scientific, Rockford, IL), using bovine serum albumin as a standard.
A kinetic experiment involving yr-hGLA and various concentrations of the substrate
was performed to determine the Michaelis constant (Km) value.
Pyridylamine (PA)-labeled N-glycans derived from yr-hGLA were separated by
high-performance liquid chromatography (HPLC), and the percentage of
phosphorylated N-glycans was calculated based on the peak areas, as described
previously (25,27). Briefly, after PNGase F digestion, the yr-hGLA N-glycans were
PA-labeled and fractionated on a normal-phase column, Shodex Asahipak NH2P-50 4E
(Showa Denko, Tokyo, Japan), to roughly separate phosphorylated and
non-phosphorylated glycans. Then, each fraction was applied to a TSK-gel Amide-80
column (Tosoh, Tokyo, Japan), and the amount of each fraction was calculated as the
percentage of the total peak area.
The percentage of phosphorylated N-glycans was also confirmed by bacterial alkaline
phosphatase digestion. PA-labeled N-glycans derived from 10 μg of yr-hGLA and
agalsidase alfa were treated with 2 units of alkaline phosphatase from E. coli C75
(TAKARA BIO., Shiga, Japan) in 50 mM Tris-HCl, pH 9.5, at 37°C for 15 h. After
boiling, the samples were filtered and analyzed on a COSMOSIL 5C18-AR-II column
(4.6 x 250 mm, Nacalai Tesque, Kyoto, Japan). Isocratic separation was performed with
10 mM sodium phosphate, pH 3.8, containing 0.075% 1-butanol for 30 min at a flow
rate of 1 ml/min at 55˚C. Samples before alkaline phosphatase treatment were used as
controls. The fractions of which the retention times shifted on the treatment were judged
to be phosphorylated N-glycans and the amount of phosphorylated N-glycans was
calculated as the percentage of the total peak area. Three independent experiments
involving three independent preparations were performed and average values with ±
standard deviation (SD) are calculated.
Examination of the Effects of yr-hGLA on Organs of Fabry Mice
Mice were bred and maintained at the Animal Center of our university. Both male and
female Fabry mice (GLA knock-out mice denoted by A.B. Kulkarni, T. Ohshima, et al.)
(28,29) and wild-type C57BL/6 mice (14 months old) were used in this experiment,
which was approved by the Animal Ethics Committee of our university.
To determine the biodistribution of the enzyme, a single dose, 1.0 mg/kg body weight,
of yr-hGLA was injected into a tail vein of three Fabry mice. As a control, agalsidase
alfa was used. The mice were sacrificed 1 h after administration of the enzymes, and
then the GLA activity in the kidneys, heart, and liver was measured. The mice were
perfused with phosphate-buffered saline (PBS), pH 7.4, before removal of the organs.
To examine cleavage of the glycosphingolipids accumulated in the organs, two groups
of Fabry mice, each consisting of three mice, were injected with yr-hGLA or agalsidase
alfa (1.0 mg/kg body weight) separately every day for four days, and then sacrificed 24
h after the last injection.
For determination of the Gb3 levels, tissues including the kidneys, heart, and liver
were analyzed by means of high performance thin-layer
chromatography-immunostaining with an anti-Gb3 monoclonal antibody (30), followed
by densitometry, as described previously (31).
For measurement of lyso-Gb3 in mouse tissues, lyso-Gb3 was extracted and then
derivatized with o-phthalaldehyde (OPA) reagent. The OPA-derivatized lyso-Gb3 was
separated by HPLC and then quantitated by fluorescence detection. A calibration curve
for lyso-Gb3 was prepared by the addition of authentic lyso-Gb3 to a normal tissue
For morphological analysis, mouse tissues were analyzed electron microscopically, as
described previously (20).
Data are expressed as means ± SD. We performed statistical analyses with Student’s
t-test. Values were considered statistically significant at p < 0.05.
Production and Purification of yr-hGLA
To produce a recombinant human GLA having mammalian-like,
phosphomannosylated sugar chains, we expressed human GLA in a gene-manipulated O.
minuta strain with the OCH1 gene, which encodes α-1, 6-mannosyltransferase
catalyzing the initial reaction of the outer sugar chain synthesis specific to yeast, deleted
and overexpressing the S. cerevisiae MNN4 gene, which encodes a positive regulator of
mannosylphosphate transferase. The productivity of the yr-hGLA was estimated to be
12 mg per 1 liter culture.
Then, the enzyme was purified from the culture medium by means of column
chromatography. The purification degree was 5,230-fold, and the recovery of the
enzyme was 35%.
Characterization of yr-hGLA
yr-hGLA was detected as two or more broad bands on SDS-PAGE, but it only gave
one band after deglycosylation. This suggested that yr-hGLA has heterogeneous sugar
chains (Figure 1).
The enzymological parameters of yr-hGLA were determined using
MU-α-D-galactopyranoside, an artificial substrate for GLA. The specific activity of
yr-hGLA was 2.0 mmol/h/mg protein and its Km value was 3.0 mM, these values being
almost the same as those of agalsidase alfa (specific activity, 2.1 mmol/h/mg protein;
and Km, 2.8 mM).
The monosaccharide composition of yr-hGLA was determined, and then compared
with that of agalsidase alfa, which had been reported previously (32). yr-hGLA has
mannose residues, but does not contain any fucose, galactose, or sialic acid ones (data
not shown). To determine the content of phosphorylated N-glycans, N-glycans derived
from yr-hGLA and agalsidase alfa were PA-labeled and then separated by HPLC. For
agalsidase alfa, 15.3 ± 2.6 % of the total N-glycans was found to be phosphorylated; on
the other hand, for yr-hGLA, the phosphorylated N-glycan content was 28.7 ± 2.7 %.
Effects of yr-hGLA in Fabry Mice
We injected a single dose of yr-hGLA into Fabry mice, and 1 h after the administration,
the GLA activity in the kidneys, heart, and liver was measured. As a control, agalsidase
alfa was used. The results are shown in Figure 2. An apparent increase in enzyme
activity was observed in the organs of Fabry mice after the yr-hGLA administration.
The increase in enzyme activity on the administration of yr-hGLA was greater in the
kidneys compared with in the case of agalsidase alfa, although it was apparently lower
in the liver.
The effects of yr-hGLA and agalsidase alfa on the degradation of Gb3 and lyso-Gb3
that had been accumulated in organs were examined separately after repeated
administration. The results are summarized in Figure 3. Both the enzymes decreased the
Gb3 and lyso-Gb3 accumulated in the kidneys, heart, and liver. There were no
differences in the decreases in the glycosphingolipid levels between yr-hGLA and
agalsidase alfa under the experimental conditions. The lyso-Gb3 level in the kidneys
decreased to approximately 30 % of that in untreated Fabry mice, but the Gb3 level
remained at about 80 % of that in untreated Fabry mice.
Morphological analysis revealed that many lamellar inclusion bodies resulted from the
accumulation of glycolipids in the kidneys, hearts, and livers of Fabry mice, and that the
number of the inclusion bodies markedly decreased after the repeated administration of
yr-hGLA, as in the case of agalsidase alfa (Figure 4).
ERT with recombinant human GLAs produced by mammalian cells has been reported
to have a lot of beneficial effects on Fabry patients; alleviation of neuropathic pain,
improvement of sweat function, regression of hypertrophic cardiomyopathy, and also
stabilization of kidney function, although it is not effective in the advanced stages of the
disease (8-10,13-16). However, there are various disadvantages of the production of
enzymes using mammalian cells, i.e., the high cost, the complicated procedure and
difficulty in performing a large scale-up, and the risk of infections through cultivation in
the presence of bovine serum. Although there is no experimental evidence that the BSE
infectious agent is present in bovine serum, both the Center for Biologics Evaluation
and Research of the US Food and Drug Administration, and the European Medicines
Agency have issued guidelines for the controlled use of materials of animal origin to
reduce the risk of BSE infection (33,34). To overcome these problems, we chose O.
minuta cells as candidate host cells for the production of a recombinant GLA for ERT,
because a methylotrophic yeast is suitable for large scale and economic production of
glycoproteins, and does not require bovine serum for culture.
In this study, we used an OCH1-disrupted strain to produce a recombinant GLA
having mammalian-like N-glycans. As humans do not have any β-linked mannose, α
-1,3-linked mannose, or mannose residue with phosphodiester linkage (Manα1-PO4-),
they could exhibit strong immunogenicity in humans (35-37). However, glycoproteins
produced by O. minuta do not contain any β-mannose or α-1,3-mannose residues
(21), and mannose attached to phosphate residues would be removed by the α
-mannosidase treatment in the procedure for exposure of M6P residues on N-glycans.
There have been reports that there is no difference in antibody production on ERT
between agalsidase alfa produced by human fibroblasts and agalsidase beta produced by
CHO cells when they are administrated at the same dose (38,39). Considering this, it is
expected that the mammalian-like sugar chains of yr-hGLA produced by the
gene-manipulated O. minuta do not exhibit high antigenicity in humans. We obtained a
productivity level of over 10 mg/ l, which will be increased by further improvement and
scaling up of the culture in the future.
We examined the enzymological characteristics of the recombinant human GLA
produced by the gene-manipulated O. minuta cells. There is no large difference in
enzyme activity or substrate affinity between yr-hGLA and agalsidase alfa. However,
results of monosaccharide composition analysis suggested that yr-hGLA has
high-mannose type sugar chains but not complex type ones, and that its content of M6P
residues is very high. These features will be advantageous for incorporation through
M6P receptors into the cells of many organs including the kidneys.
Lee et al. (40) examined the biodistribution of agalsidase alfa and agalsidase beta
injected into tail veins of Fabry mice. The results suggested that the enzymes are
predominantly incorporated into the liver. Recombinant lysosomal enzymes produced
by mammalian cells usually contain both complex type and high-mannose type sugar
chains, and most of the infused enzymes are rapidly incorporated into hepatocytes and
Kupffer cells via asialoglycoprotein and mannose receptors, respectively. So, effective
delivery of lysosomal enzymes to various organs except for the liver through M6P
receptors should be compromised on rapid clearance of administered enzymes by
asialoglycoprotein and mannose receptors in the liver. Sly et al. (41) reported that
genetical elimination of mannose receptors slowed the plasma clearance of recombinant
β-glucuronidase, resulting in enhanced degradation of the accumulated substrates in the
kidneys. They also reported that saturation of the mannose receptor clearance system
with high doses of enzyme improved the targeting to M6P receptor-containing tissues
including the kidneys. Thus, inhibition of uptake of lysosomal enzymes by the liver
should facilitate their incorporation into the kidneys via M6P receptors. Our present
animal study also revealed that the administration of agalsidase alfa greatly increased
the GLA activity in the livers of Fabry mice. On the other hand, the increase in enzyme
activity in the liver was moderate when yr-hGLA, which does not have complex type
sugar chains, was injected. As to the kidneys, yr-hGLA exhibited markedly increased
enzyme activity compared with in the case of agalsidase alfa. As there should be no
difference in protein sequence between yr-hGLA and agalsidase alfa, the difference in
their biodistributions after injection should be due to a difference in their sugar chains.
The mechanism underlying the high incorporation of yr-hGLA into the kidneys is
unknown. But low clearance of the enzyme by the liver and/or a high content of
phosphorylated N-glycans of yr-hGLA may play a role in the mechanism. Recently,
Prabakaran et al. revealed that at least three endocytic receptors including the M6P one,
megalin, and sortilin were related to GLA delivery in human podocytes (Prabakaran, T.,
et al. Receptor-mediated uptake of α-galactosidase A in human podocytes in Fabry
disease. The 11th European Round Table on Fabry Disease, October 15, 2010, Istanbul).
Considering the results, many receptors other than the M6P one are involved in the
uptake of GLA in the kidneys, and they may play an important role in the mechanism.
Takamatsu et al. (42) reported that a recombinant fibroblast growth factor produced by
gene-manipulated yeast cells was predominantly incorporated into the kidneys after
intravenous injection into mice. Glycoproteins with a specific sugar chain structure
produced by gene-manipulated yeast cells may be applicable to kidney-targeting