327
Expression and secretion of a biologically active
glycoprotein hormone, ovine follicle stimulating hormone,
by Pichia pastoris
A E Fidler, S Lun, W Young and K P McNatty
AgResearch Wallaceville, PO Box 40063, Ward Street, Upper Hutt, New Zealand
(Requests for offprints should be addressed to A E Fidler)
ABSTRACT
The methylotrophic yeast, Pichia pastoris, has been
used to co-express recombinant genes formed by
fusion of the mating factor-alpha (MFá) leader and
ovine follicle stimulating hormone (oFSH) á and â
subunit coding sequences. Pichia strains carrying
single copies of the two fusion genes secreted
recombinant oFSH (roFSH) to concentrations of
approximately 51·0 ng/ml and 17·5 ng/ml, measured
by RIA or in vitro bioassay respectively, whereas a
strain with two copies of the á and one copy of the â
subunit fusion genes secreted roFSH to concentrations of 61 ng/ml (RIA) and 22 ng/ml (bioassay).
INTRODUCTION
Follicle stimulating hormone (FSH) secreted by
pituitary gonadotrophs promotes ovarian steroidogenesis and follicular growth (Richards 1994).
Administration of exogenous FSH can be used to
promote the maturation of multiple ovarian follicles
(i.e. superovulation), a procedure that has found
practical application in both human infertility
treatment and animal breeding programmes
(Ben-Cherit et al. 1996, De Koning et al. 1994). At
present, the FSH used for the superovulation of
domestic animals is purified from animal-derived
material; however, the issues of raw material supply
and potential pathogen transmission indicate a
requirement for recombinant FSH (De Koning
et al. 1994).
Structurally, FSH is a heterodimer formed by the
non-covalent association of an á subunit – which is
common to three pituitary glycoprotein hormones,
FSH, luteinizing hormone (LH) and thyroid
stimulating hormone (TSH) (Combarnous 1992,
Nagaya & Jameson 1994) – and a hormone-specific
â subunit that confers receptor binding specificity
(Combarnous 1992). The secondary structures of
Journal of Molecular Endocrinology (1998) 21, 327–336
0952–5041/98/021–327 $08.00/0
It appears that the Pichia-derived roFSH had about
one-third the in vitro bioactivity of native oFSH or,
alternatively, only one-third of the roFSH is
bioactive. Measurements of secreted roFSH á and â
subunit concentrations indicated less than 10% of á
and 25–33% of â subunits were stably dimerized.
The receptor binding properties of the roFSH
resemble those of native oFSH. In summary this
paper reports the production, by P. pastoris, of a
heterodimeric glycoprotein hormone (roFSH) that
has in vitro biological activity.
Journal of Molecular Endocrinology (1998) 21, 327–336
the á and â subunits are constrained by intramolecular disulphide bonds, five in the á subunit and six in the â, and both subunits carry
two N-linked carbohydrate moieties (Combarnous
1992). Variation in the carbohydrate structures
results in both pituitary and serum FSH comprising
a heterogeneous population of molecules varying
in parameters such as receptor-binding affinity and
metabolic clearance rate (Ulloa-Aguirre et al. 1995).
The methylotrophic yeast, Pichia pastoris, has
been developed as an expression system for
high-level production of recombinant proteins
(Buckholz & Gleeson 1991, Cregg et al. 1993).
Pichia offers the features of (i) methanol-induced
expression of heterologous genes integrated into the
genome adjacent to the alcohol oxidase 1 (AOX1)
gene promoter, (ii) growth to high cell density in
inexpensive, chemically defined media, and (iii) the
capacity to carry out post-translational modifications resembling those of mammalian cells (Cregg
et al. 1993). Using either native or heterologous
secretion sequences, recombinant proteins can be
directed into the yeast cell secretory pathway,
wherein disulphide bond formation and glycosylation can occur before secretion into the
? 1998 Society for Endocrinology Printed in Great Britain
328
and others · Recombinant ovine FSH production by yeast
growth media (Cregg et al. 1993). A wide range of
biologically active recombinant proteins have now
been produced in P. pastoris, including secreted
proteins containing intramolecular disulphide bonds
(Clare et al. 1991, Vedvick et al. 1991). Furthermore,
co-expression of two different genes in Pichia has
resulted in the production of biologically active protein heterodimers (Kalandadze et al. 1996). The
success of these experiments indicated that Pichia
pastoris might be a suitable expression system for the
production of biologically active recombinant ovine
FSH (roFSH). This paper reports the co-expression
of ovine FSH á and â subunit cDNA sequences in
P. pastoris to produce biologically active roFSH.
MATERIALS AND METHODS
Choice of yeast expression system
The original cloning procedure was designed with
the objective of co-expressing ovine FSH á and â
subunit cDNA sequences in the budding yeast,
Saccharomyces cerevisiae, using the expression
vector, pYES2 (Invitrogen, San Diego, CA, USA).
To this end, fusion genes were constructed with
the S. cerevisiae mating factor-alpha (MFá) leader
sequence fused in-phase to sequences encoding the
oFSH á and â mature proteins, forming fusion
genes MFáoFSHá and MFáoFSHâ, respectively, as described in detail below. Note that, with
both fusion genes, the FSH subunit mature protein
coding sequences were positioned immediately 3* to
the two codons of the MFá leader sequence
encoding the Lys–Arg dibasic motif recognized
by the kex2 protease (Julius et al. 1984). Attempts
to produce biologically active roFSH using S.
cerevisiae were unsuccessful (A Fidler, unpublished
results), leading to investigation of the yeast, Pichia
pastoris, as an alternative yeast expression system.
Cloning of the yeast MFá leader sequence
Nucleotide sequences encoding the yeast MFá
leader sequence were amplified using forward
primer 5*-CCAAGCTTATGAGATTTCCTTCA
ATTTTTA-3* (HindIII site underlined; remaining
primer sequence corresponds to nucleotides 1–22 of
Kurjan & Herskowitz (1982)) and reverse primer
5*-GGGAATTCAGGCCTTTTATCCAAAGAT
ACC-3* (EcoRI and StuI restriction sites underlined; remainder is the complement of nucleotides
255–237 of Kurjan & Herskowitz (1982)). After
HindIII and EcoRI (Boehringer Mannheim,
Mannheim, Germany) digestion, the amplification
products were electrophoresed through 1% (w/v)
low melting point agarose gels (FMC BioProducts,
Journal of Molecular Endocrinology (1998) 21, 327–336
Rockland, ME, USA), in-gel ligated into pUC18
using T4 DNA ligase (Boehringer Mannheim) and
transformed into competent DH5á cells. Plasmids
having inserts were identified by restriction enzyme
digestion and T7 Sequenase catalysed DNA sequencing (Amersham Life Science Ltd, Amersham,
Bucks, UK). The resulting construct was termed
pUC18MFá.
Construction of fusion genes MFáoFSHá and
MFáoFSHâ
First-strand cDNA was synthesized from 5 µg ovine
pituitary total RNA using Superscript reverse transcriptase (Gibco BRL, Bethesda, MA, USA). The
mature protein coding sequences of the oFSH á and
â subunits were amplified by Taq DNA polymerase (Boehringer Mannheim)-catalysed PCR: FSH
á subunit forward primer 5*-TTTCTGATG
GAGAGTTTACAATGCAGGGT-3* (nucleotides
143–172 of Bello et al. (1989)), reverse primer
5* - CCGAATTCAAATATTTAAGATTTGTGA
TAA-3* (EcoRI site underlined; remainder corresponding to nucleotides 439–418 of Bello et al.
(1989)); FSH â subunit forward primer 5*-AG
CTGCGAGCTGACCAACATCACCATCAC3* (corresponding to nucleotides 113–141 of
Mountford et al. (1989)), reverse primer 5*-CC
GAATTCCTCTTTATTCTCTGATGTCACT-3*
(EcoRI site underlined; remaining sequence the
complement of nucleotides 449–430 of Mountford
et al. (1989)). Reaction conditions were: 94 )C for
3 min, 60 )C for 3 min, 72 )C for 5 min, one cycle;
94 )C for 30 s, 62 )C for 1 min, 72 )C for 2 min,
35 cycles; 72 )C for 5 min, one cycle. The
PCR amplification products were incubated with
T4 DNA polymerase (Boehringer Mannheim),
to remove 3*A overhangs, digested with EcoRI,
and ligated with StuI/EcoRI double-digested
pUC18MFá. The sequences of the resulting
HindIII–EcoRI fragments, encoding fusion genes
either MFáoFSHá or MFáoFSHâ, were confirmed
by DNA sequencing (Amersham) as being free of
errors introduced by Taq polymerase.
Subcloning of fusion genes MFáoFSHá and
MFáoFSHâ into pAO815 to form plasmids
pAOMá and pAOMâ
As originally constructed, the MFáoFSHá and
MFáoFSHâ fusion genes were encoded on
HindIII–EcoRI fragments (HindIII at 5* end,
EcoRI at 3* end) to facilitate cloning into the S.
cerevisiae expression vector, pYES2 (Invitrogen).
To allow subcloning into the single EcoRI site
of the P. pastoris expression vector, pAO815
(Invitrogen), EcoRI sites were required at both the
Recombinant ovine FSH production by yeast ·
5* and 3* ends of the fusion gene sequence (Cregg
et al. 1993). Such EcoRI sites were introduced by
co-ligating the HindIII–EcoRI fragments encoding
the MFáoFSHá and MFáoFSHâ fusion genes
into the EcoRI site of pUC18, along with linker
oligomers 5*-AATTCGGTACCA-3* and 5*AGCTTGGTA CCG-3* thereby introducing an
EcoRI site into the 5* end of each fusion gene. The
resulting EcoRI fragments, encoding the two fusion
genes, MFáoFSHá and MFáoFSHâ, were ligated
into EcoRI digested, dephosphorylated pAO815,
forming plasmids pAOMá and pAOMâ respectively. The orientation of the inserts with respect to
the AOX1 promoter was confirmed by restriction
enzyme analysis and DNA sequencing.
and others
grown to high density in rich media with a glycerol
carbon source and then resuspended in media
containing methanol as the sole carbon source
(Barr et al. 1992, Sreekrishna & Kropp 1996). After
2 days of induction, the culture medium was
clarified by centrifugation, a protease inhibitor
cocktail (Complete, Boehringer Mannheim) was
added and the mixture dialysed extensively
against 1#phosphate buffer saline (PBS) at 4 )C.
Samples were stored at "20 )C before analysis by
radioimmunoassay and bioassay.
Radioimmunoassays
The Pichia host strain, GS115 (his4) (Invitrogen),
was transformed with BglII-linearized plasmids
using the lithium chloride method (Ausubel et al.
1994). His+/MutS transformants were isolated by
selection for His+ auxotrophy and screening for
slow growth on media containing methanol as the
sole carbon source (MutS) (Sreekrishna & Kropp
1996). The presence of oFSH á and â subunit
sequences in His+/MutS transformants was confirmed by PCR from genomic DNA templates
(Linder et al. 1996).
Immunoreactive FSH dimer and oFSHá were
measured using RIA kits supplied by the NIH
National Hormone and Pituitary Program (NIH,
Bethesda, MD, USA). Samples were assayed for
FSH dimer using United States Department of
Agriculture (USDA)-oFSH–19-selective immunoaffinity purified (SIAFP)-I-2 for iodination,
USDA-oFSH–19-SIAFP-RP-2 (94#NIH-oFSHS1; biological potency=2351 IU/mg) for reference
preparation and National Institute of Diabetes and
Digestive and Kidney Diseases-anti oFSH-1 antiserum. For the assay of oFSH á subunit,
WRR-1-Alpha was used for iodination and reference preparation, and the antiserum was National
Institute of Arthritis, Metabolism and Digestive
Diseases-anti-oLH Alpha-1. For the assay of oFSH
â subunit, oFSHâ (Bioscan Continental Inc.,
Quebec, Canada) was used for iodination and
reference preparation, and the antiserum was
anti-porcine FSHâ (Biogenesis Ltd, Poole, Dorset,
UK). Standards were iodinated using the Chloramine T method and purified by gel chromatography (FSH á and â subunits) or ion exchange
chromatography (FSH dimer) (Moore et al. 1997).
Assays were performed by incubating 100 µl sample
or standard, 100 µl primary antibody and 50 µl
tracer overnight at room temperature. Separation of
free from bound hormone was by the second
antibody method, using sheep anti-rabbit IgG,
followed by 1 ml 6% polyethylene glycol 8000
(Carbowax Union Carbide Co., Danbury, CT,
USA). Sensitivities of the assays for FSHá, FSHâ
and FSH dimer (90% of zero binding) were 0·1, 0·5
and 0·2 ng/ml respectively. Intra- and inter-assay
coefficients of variation for the above assays were 9%
and 18·8%, 10·9% and 15·4%, and 7·9% and 10·8%
respectively.
Induction of recombinant protein production
Radioreceptor assay (RRA)
Pichia culturing and induction followed ‘shake tube’
and ‘shake flask’ procedures in which cultures were
Clarified supernatants
‘shake flask’ cultures
Construction of plasmids pAOMáMâ and
pAO(Má)2Mâ
Plasmid pAOMâ was double-digested with restriction enzymes BglII and BamHI and a
1·9-kb fragment, encoding the AOX1 promoter,
MFáoFSHâ fusion gene and 3*AOX1 transcription
termination sequences, was ligated into BamHIdigested, dephosphorylated pAOMá, producing
plasmid pAOMáMâ. The relative orientation of the
two fusion genes in pAOMáMâ was determined by
both restriction enzyme analysis and DNA sequencing. Plasmid pAOMáMâ was digested with BglII
and BamHI and an approximately 3·8-kb fragment, encoding the two fusion genes, MFáoFSHá
and MFáoFSHâ, with their transcription control sequences, was ligated into BamHI-digested,
dephosphorylated pAOMá, producing plasmid
pAO(Má)2Mâ. Insert orientation was determined
by restriction enzyme analysis and DNA
sequencing.
GS115 transformation and screening of
transformants
from methanol-induced
of strains GSpAO815,
Journal of Molecular Endocrinology (1998) 21, 327–336
329
330
and others · Recombinant ovine FSH production by yeast
GSpAOMáMâ and GSpAO(Má)2Mâ were dialysed extensively against 10 mM MgCl2, 50 mM
Tris–HCl pH 7·3 before being concentrated tenfold
by ultrafiltration (Amicon, Beverly, MA, USA).
FSH concentrations measured by RRA were
quantified using membrane fractions from bovine
testes as previously reported (Moore et al. 1997)
using USDA-125I oFSH–19-SIAFP-I-2 as tracer
and USDA-oFSH–19-SIAFP-RP-2 for reference
preparation. Samples from methanol-induced
Pichia cultures were assayed at supernatant protein concentrations of less than 30 µg per tube
(300 µg/ml) to avoid the detection of non-specific
binding interference when the supernatants were
assayed at higher concentrations. Total protein
concentrations were determined by the bicinchoninic acid method (Pierce, Rockford, IL, USA).
In vitro FSH bioassay
The FSH bioassay, using a Chinese Hamster Ovary
(CHO) cell line expressing the human FSH
receptor, was essentially as described previously
(Albanese et al. 1994). However, the assay endpoint was cAMP production, as measured using
RIA rather than luciferase reporter gene expression.
Native oFSH (NIH-FSH-RP2) (NIH) was used as
the standard.
RESULTS
Plasmid pAO815-based constructions and
transformation into Pichia host strain GS115
Four plasmid constructs were produced: (i)
pAOMá, encoding the MFáoFSHá fusion gene
alone, (ii) pAOMâ, encoding the MFáoFSHâ
fusion gene alone, (iii) pAOMáMâ, encoding both
the MFáoFSHá and MFáoFSHâ fusion genes in
tandem, and (iv) pAO(Má)2Mâ, encoding two
copies of the MFáoFSHá gene and a single copy of
the MFáoFSHâ gene (Fig. 1A). In those plasmids
with multiple fusion genes, each gene is transcribed
from its own AOX1 promoter and is separated from
adjacent genes by the 3*AOX1 transcription
termination sequence. As it has been found that the
level of recombinant protein production is frequently positively correlated with the number of
copies of the corresponding gene (Sreekrishna &
Kropp 1996), plasmid pAO(Má)2Mâ was constructed to mimic more closely the stoichiometry of
the á/â subunit dimerization reaction in gonadotrophs, in which the á subunit is present in excess of
the â subunit. Plasmids were linearized by BglII
digestion and transformed into the strain GS115
(his4) (Invitrogen) and MutS/His+ transformant
Journal of Molecular Endocrinology (1998) 21, 327–336
strains identified. For each plasmid construct, six to
ten MutS/His+ transformants were screened for
secreted recombinant protein production, and
transformed strains that consistently produced the
greatest amount of protein were selected for further
characterization. The presence of oFSHá or oFSHâ
cDNA sequences, or both, in the genomes of the
transformed strains was confirmed by PCR using
primers specific for the oFSH á and â subunit
sequences (Fig. 1B,C).
Measurement of the concentrations of
secreted recombinant oFSH á and â subunits
in the growth media of GS115 transformants
RIA was used to determine the concentrations of
secreted oFSH á and â subunits present in the
growth media of methanol-induced ‘shake tube’
cultures of GS115 transformants (Fig. 2). No
immunoreactivity to either the anti-oFSHá or the
anti-oFSHâ antisera was detected in the media of
either untransformed GS115 (GS0) or GS115 transformed with the vector pAO815 (GSpAO815). In a
representative triplicate induction, the growth media
of strain GSpAOMá contained a mean&... of
752&27 ng/ml immunoreactive oFSHá and no
detectable oFSHâ, whereas the reverse was found
with strain GSpAOMâ, with which immunoreactive
oFSHâ was detected in the supernatant at a
mean&... concentration of 209&15 ng/ml, with
no detectable oFSHá. These findings are consistent
with the presence and absence of corresponding
genes as determined by PCR (Fig. 1B,C) and confirm that there is no significant cross reaction
between the two recombinant á and â subunits in the
RIA. Co-expression of the á and â fusion genes, in
strain GSpAOMáMâ, resulted in the secretion of
recombinant oFSH á and â subunits into the growth
media to mean&... concentrations of 547&6
and 196&9 ng/ml respectively. The growth media
of strain GSpAO(Má)2Mâ contained á subunit at
a mean&... concentration of 1386&174 ng/ml
and oFSHâ subunit at a mean&... concentration
of 192&18 ng/ml.
RIA and bioassay measurements of the
concentration of secreted recombinant oFSH
dimer in the growth media of GS115
transformants
As FSH is biologically active as an á/â-subunit
heterodimer, both RIA and an in vitro bioassay were
used to determine the concentration of roFSH
dimer present in the growth media of the
transformed Pichia strains. Neither FSH immunoreactivity nor bioactivity was detected either in
Recombinant ovine FSH production by yeast ·
1. (A) Arrangement of fusion genes
MFáoFSHá and MFáoFSHâ in plasmids pAOMá,
pAOMâ, pAOMáMâ and pAO(Má)2Mâ. The cloning
strategy is outlined in Materials and Methods. Plasmid
pAOMá encodes the fusion gene MFáoFSHá, which
consists of the S. cerevisiae MFá leader sequence fused
in phase to the ovine FSHá mature protein coding
region, flanked by the AOX1 promoter (5*AOX1) and
transcription termination (TT) sequences. Plasmid
pAOMâ resembles pAOMá, except that it encodes the
fusion gene MFáoFSHâ. Plasmid pAOMáMâ encodes
both fusion genes MFáoFSHá and MFáoFSHâ, each
flanked by 5*AOX1 and the TT sequences and
orientated with respect to each other as shown. Plasmid
pAO(Má)2Mâ encodes two copies of the MFáoFSHá
fusion gene and a single copy of the MFáoFSHâ gene,
with each fusion gene being flanked by the AOX1
transcription control sequences. All pAO815-derived
plasmid constructs were linearized by BglII digestion
before being transformed into Pichia strain GS115.
Transformants in which the linearized plasmids had
incorporated into the AOX1 gene by homologous
recombination were identified by sequentially selecting
for the His+ phenotype and then screening for the MutS
phenotype. The resulting His+/MutS strains were
designated GSpAOMá, GSpAOMâ, GSpAOMáMâ
and GSpAO(Má)2Mâ. (B,C) Detection of oFSH á and
â subunit cDNA sequences in the genomes of
transformed Pichia strains. Genomic DNAs from strains
GS115 (host strain untransformed) (lane 1), GSpAO815
(host strain transformed with plasmid pAO815) (lane 2),
GSpAOMá (lane 3), GSpAOMâ (lane 4),
GSpAOMáMâ (lane 5) and GSpAO(Má)2Mâ (lane 6)
were used as templates for PCR with primers specific to
the ovine FSHá (B) and ovine FSHâ (C) cDNA
sequences. The PCR primers were those used to amplify
the oFSH á and â subunit coding regions for
construction of the fusion genes and amplified products
of 296 bp (FSHá) and 336 bp (FSHâ). The
amplification products were electrophoresed through 2%
(w/v) agarose gels, ethidium bromide-stained and
photographed under u.v. illumination.
and others
the growth media of control strains, GS0 and
GSpAO815, or the media of strains transformed
with only one of the two fusion genes, strains
GSpAOMá and GSpAOMâ (Fig. 3). In contrast,
the growth media of Pichia strains carrying at least
one copy of the á and â subunit fusion genes
(i.e. strains GSpAOMáMâ and GSpAO(Má)2Mâ)
secreted roFSH dimer into their growth media:
GSpAOMáMâ (RIA mean&... concentration
50·9&1·4 ng/ml; bioassay mean&... concentration 17·5+2·7 ng/ml); GSpAO(Má)2Mâ (RIA
mean&... concentration 61·2&5·6 ng/ml; bioassay mean&... concentration 21·8&2·4 ng/ml)
(Fig. 3).
The triplicate induction results indicated that
duplication of the á subunit fusion gene, strain
GSpAO(Má)2Mâ, resulted in an increase in
production of roFSH dimer when compared with
strain GSpAOMáMâ, which carries a single copy of
the á subunit fusion gene. To confirm this, ten
‘shake-tube’ cultures of strains GSpAOMáMâ and
GSpAO(Má)2Mâ were methanol-induced and the
concentration of roFSH in the growth medium
measured by both RIA and bioassay. The results
showed a greater mean concentration of roFSH in
the growth media of strain GSpAO(Má)2Mâ (RIA
mean&... concentration 63·1&2·8 ng/ml; bioassay mean&... concentration 22·6&1·8 ng/
ml) compared with strain GSpAOMáMâ (RIA
mean&... concentration 45·0&1·6 ng/ml; bioassay mean&... concentration 15·3&1·2 ng/
ml), with the differences from both assays being
statistically significant at the 1% level.
FSH receptor binding characteristics of the
secreted recombinant oFSH
RRAs were used to compare the receptor binding
characteristics of the recombinant oFSH with that of
native oFSH. Radiolabelled native oFSH was competitively displaced from the bovine FSH receptor
by increasing amounts of unlabelled native oFSH
or increasing amounts of dialysed and concentrated
culture media from methanol-induced cultures
of strains GSpAOMáMâ and GSpAO(Má)2Mâ
(Fig. 4). The sigmoidal displacement curves have
been transformed into linear plots using the logit–log
transformation. The regression lines for hormone
displacement by the culture media of strains
GSpAOMáMâ and GSpAO(Má)2Mâ parallelled
each other, with slopes of "0·876 and "0·811
respectively, and that of native oFSH standard, slope
"0·811, indicating that the binding properties of
the roFSH from the two strains resemble each
other and that of the native oFSH. Control RRAs
using the dialysed culture media from induced
Journal of Molecular Endocrinology (1998) 21, 327–336
331
332
and others · Recombinant ovine FSH production by yeast
2. RIA measurements of the concentration of FSH á and â subunits secreted
into the growth media of transformed P. pastoris strains. Pichia strains GS0,
GSpAO815, GSpAOMá, GSpAOMâ, GSpAOMáMâ and GSpAO(Má)2Mâ were
grown for 2 days before heterologous gene expression was methanol-induced for a
further 2 days. The cells were pelleted by centrifugation and the clarified supernatant
dialysed extensively against 1#PBS before the FSH á (/) and â (.) subunit
concentrations (ng/ml) were assayed by RIA. Values shown are means of three
inductions, with ... bars indicated.
cultures of negative control strain GSpAO815 indicated that non-specific interference with native
oFSH binding occurred at culture media protein
concentrations greater than 300 µg/ml (30 ng/tube)
(Fig. 4). Using ED50 values from the RRA, the
concentrations of roFSH in the tenfold concentrated culture media were calculated as 230 ng/ml
for strain GSpAOMáMâ and 342 ng/ml for
GSpAO(Má)2Mâ.
DISCUSSION
Biologically active recombinant FSH has been
produced previously using transformed mammalian
and insect cell lines and transgenic mice (Keene
et al. 1989, Greenberg et al. 1991, Mountford et al.
1994, Arey et al. 1997, Hakola et al. 1997). The
CHO cell-derived recombinant human FSH has
been used successfully in infertility treatment
(Devroey et al. 1993). Furthermore, it was recently
reported that Pichia-expressed bovine FSH â
subunit could be combined with native bovine
á subunit to produce biologically active FSH
(Samaddar et al. 1997). Extending these approaches,
we have now reported that co-expression of
Journal of Molecular Endocrinology (1998) 21, 327–336
recombinant genes encoding the yeast MFá leader
sequence fused to the ovine FSH á and â subunit
mature protein coding regions results in the
secretion of biologically active recombinant ovine
FSH (roFSH).
It is of interest to note that the concentration of
roFSH as measured by in vitro bioassay was only
about one-third of that determined by RIA,
indicating either that the roFSH has one-third
the biopotency of native oFSH or, alternatively,
that only about one-third of the immunoreactive
roFSH is biologically active. The difference in
the measurements of the two assays cannot be
attributed to non-specific factors within the growth
media interfering with the bioassay, as native oFSH
standards diluted in strain GSpAO815 growth
media were accurately measured using the bioassay
(data not shown). It should be borne in mind that
the two assays used (RIA and bioassay) differ with
respect to the end-points of measurements, with the
RIA requiring interaction of the roFSH with a
polyclonal antisera, whereas the bioassay requires
roFSH binding to a G-protein-coupled membrane
receptor, followed by cAMP synthesis (Albanese
et al. 1994). The RRA measurements gave roFSH
concentration estimates between those of the RIA
Recombinant ovine FSH production by yeast ·
and others
3. RIA and bioassay measurements of the concentration of recombinant oFSH
heterodimer secreted into the growth media of transformed P. pastoris strains. Clarified
culture supernatants from methanol-induced strains GS0, GSpAO815, GSpAOMá,
GSpAOMâ, GSpAOMáMâ and GSpAO(Má)2Mâ were dialysed against 1#PBS and
assayed for roFSH heterodimer by RIA (/) or in vitro bioassay (.). RIA values are the
means of three inductions, the bioassay values the means of six inductions. ... bars
are shown.
and the bioassay, suggesting the existence of three
classes of secreted roFSH: (i) receptor-binding,
biologically active, (ii) receptor-binding, biologically inactive, and (iii) non-receptor-binding.
The possible lack of biological activity of two-thirds
of the roFSH may result from the recombinant
molecules adopting tertiary structures inappropriate for receptor binding, activation, or both.
In particular, N-linked carbohydrate moieties
significantly influence the biological activity of
gonadotropins (Thotakura & Blithe 1995,
Ulloa-Aguirre et al. 1995, Arey et al. 1997), as they
are required for activation of intracellular signalling
pathways in addition to being determinants of
in vivo bioactivity through their influence on the
metabolic clearance rate (MCR) of the hormones
(Ulloa-Aguirre et al. 1995, Arey et al. 1997). As the
glycosylation moieties synthesized by P. pastoris
differ from those of mammalian cells, being of the
high-mannose type characteristic of yeast (Grinna &
Tschopp 1989), an important future goal will be to
determine the glycosylation patterns and MCR of
Pichia-derived roFSH. Comparison of the mean
concentrations of the secreted roFSH á and â
subunits with those of the roFSH heterodimer
indicates that only approximately 9% and 4% of the
á subunits, 26 and 32% of the â subunits, secreted
by strains GSpAOMáMâ and GSpAO(Má)2Mâ
respectively, are present in the culture media as
dimers. Thus it would appear that dimerization of
the roFSH á and â subunits occurs inefficiently
or the roFSH heterodimers formed are unstable,
dissociating into their subunits after secretion. At
the molecular level, there are a number of
explanations for the low level of stable roFSH dimer
formation: the 85-amino acid MFá leader sequence
may sterically hinder dimerization (Kurjan &
Herskowitz 1982, Julius et al. 1984), N- and
O-linked carbohydrate moieties may block dimerization, as has been reported for the native á subunit
(Blithe 1990, Thotakura & Blithe 1995), or a â
subunit intramolecular disulphide bond that forms a
‘seat-belt’ structure believed to be important in
stabilizing gonadotrophin dimers (Lapthorn et al.
1994) may not be formed efficiently in Pichia
cells.
Comparison of the á and â subunit concentrations
in the culture media of GS115 transformants in
which the fusion genes were expressed alone,
GSpAOMá and GSpAOMâ, with those of strain
Journal of Molecular Endocrinology (1998) 21, 327–336
333
334
and others · Recombinant ovine FSH production by yeast
4. Comparative displacement of 125I-labelled
oFSH from bovine FSH receptors by native oFSH and
the dialysed culture media of transformed Pichia strains
GSpAO815, GSpAOMáMâ, and GSpAO(Má)2Mâ.
Radio-labelled (125I) oFSH (USDA-oFSH–19SIAFP-I-2) was specifically displaced from bovine FSH
receptors by increasing concentrations of oFSH
standards (USDA-oFSH–19-SIAFP-RP-2) (-) and
tenfold concentrated dialysed culture media from
methanol-induced cultures of strains GSpAOMáMâ
(5) and GSpAO(Má)2Mâ (1). At supernatant
concentrations of protein greater than 30 µg/tube,
non-specific displacement of the 125I-oFSH was
observed using the culture media from negative control
strain GSpAO815 (,). Concentrations of native oFSH
standards (-) are Ln (ng oFSH/tube) while
concentrations of culture media (GSpAO815 (,),
GSpAOMáMâ (5), GSpAO(Má)2Mâ (1)) are Ln (ng
total protein/tube). Logit B/Bo, loge (b/100"b) where b
is the proportion of tracer bound expressed as a
percentage of that in the zero standard. Values shown
are the means of duplicate assays.
GSpAOMáMâ indicates that â subunit production
is unaffected by co-expression of the á subunit gene,
whereas the secreted á subunit concentrations are
significantly reduced when co-expressed with the
â subunit gene (P<0·001). In contrast, á subunit production is increased by duplicating the
MFáoFSHá gene, the mean concentration of á
subunit secreted by strain GSpAO(Má)2Mâ
(1386 ng/ml) being approximately double that
produced by strain GSpAOMá (752 ng/ml) (a
difference that is statistically significant at the
5% level). Increased á subunit secretion with
doubling of gene copy number is more dramatic
when comparing strains GSpAOMáMâ and
GSpAO(Má)2Mâ. The more than twofold differJournal of Molecular Endocrinology (1998) 21, 327–336
ence in á subunit secretion between these strains
may indicate that increased á subunit production
overcomes the apparent inhibitory effect of â
subunit co-expression. Although the fraction of á
subunit secreted by strain GSpAO(Má)2Mâ that is
dimerized with â subunit is low at 4% (compared
with 9% with strain GSpAOMáMâ), duplicating
the MFáoFSHá gene did result in a 40–50%
increase in roFSH dimer production as measured
by RIA, bioassay and RRA.
As noted above, at 20–60 µg/l, the amounts of
roFSH secreted by the transformed Pichia strains
are modest when compared with the mg/l to g/l
values reported for some Pichia-produced heterologous proteins (Sreekrishna & Kropp 1996).
However, the amounts are comparable to the values
of 300–400 µg/l previously reported for recombinant
heterodimer secretion by Pichia (Kalandadze et al.
1996). Although the low level of roFSH production
may, in part, reflect the low efficiency of roFSH
dimer formation, concentrations of secreted á and â
subunit themselves are at the lower end of the range
of reported recombinant protein production by
Pichia (Sreekrishna & Kropp 1996). roFSH subunit
synthesis is probably influenced by the conditions of
yeast growth and induction, and by the sequences of
the recombinant genes. The growth and induction
procedures used in this work were standard ‘shake
tube’ or ‘shake flask’ procedures, both of which
produced similar concentrations of secreted recombinant proteins. However, these growth and
induction conditions are by no means optimal, and
the use of bioreactors frequently results in a
substantial increase in heterologous protein production (Sreekrishna & Kropp 1996). Both the
nucleotide composition and codon usage of heterologous genes influence their level of expression in
Pichia (Sreekrishna & Kropp 1996). Although the
A+T content of the FSH á and â subunit coding
sequences, 54% and 44% respectively, is within the
30–55% range considered compatible with high level
expression, both the á and â subunit coding
sequences include one or more codons believed to
be unfavourable for translation in Pichia: in the
á subunit, GGG encoding Gly34, and in the â
subunit, GGG (Gly99), ATA (Ile20), CGC (Arg88)
and CGA (Arg96) (Bello et al. 1989, Mountford
et al. 1989, Sreekrishna & Kropp 1996). Furthermore, it is possible that heterodimer formation and
stability could be enhanced using fusion gene
constructs encoding secretory signals other than the
MFá leader sequence. In addition, heterodimer
formation and stability might be enhanced by
cloning of the â and á subunit coding regions in
tandem, to produce a biologically active single
polypeptide, as has been reported for human FSH
Recombinant ovine FSH production by yeast ·
expressed by mammalian cells (Sugahara et al.
1996a,b).
In summary, the key finding of this paper is that
biologically active recombinant ovine follicle stimulating hormone has been produced by Pichia
pastoris. To our knowledge, this is the first report of
the production of a biologically active glycoprotein
hormone using this particular yeast. Further studies
will be needed to evaluate whether yields of P.
pastoris-derived roFSH can be increased, and
whether the roFSH is biologically active in vivo.
ACKNOWLEDGEMENTS
Our thanks go to Dr Chip Albanese for the
generously providing the transformed CHO cellline used in the FSH bioassay, to Drs David Tisdall
and Lloyd Moore for helpful comments on this
manuscript, and to Alan Barkus and Ruth Kiel for
assistance with figure preparation. This work was
financially supported by the New Zealand Foundation for Research, Science and Technology
(FoRST).
REFERENCES
Albanese C, Christin-Maitre S, Sluss PM, Crowley WF &
Jameson JL 1994 Development of a bioassay for FSH using
a recombinant human FSH receptor and a cAMP responsive
luciferase reporter gene. Molecular and Cellular
Endocrinology 101 211–219.
Arey BJ, Stevis PE, Deecher DC, Shen ES, Frail DE,
Negro-Vilar A & Lopez FJ 1997 Induction of promiscuous
G protein coupling of the follicle-stimulating hormone
(FSH) receptor: a novel mechanism for transducing
pleiotropic actions of FSH isoforms. Molecular Endocrinology
11 517–526.
Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG,
Smith JA & Struhl K (Eds) 1994 Saccharomyces cerevisiae.
In Current Protocols in Molecular Biology, vol 2, ch 13,
pp 13·7·1–13·7·2. New York: John Wiley & Sons, Inc.
Barr KA, Hopkins SA & Sreekrishna K 1992 Protocol for
efficient secretion of HSA developed from Pichia pastoris.
Pharmaceutical Engineering 12 48–51.
Bello PA, Mountford PS, Brandon MR & Adams TE 1989
Cloning and DNA sequence analysis of the cDNA for the
common á-subunit of the ovine pituitary glycoprotein
hormones. Nucleic Acids Research 17 10494.
Ben-Cherit A, Gotlieb L, Wong PY & Casper RF 1996
Ovarian response to recombinant human follicle-stimulating
hormone in luteinizing hormone-depleted women:
examination of the two cell, two gonadotropin theory.
Fertility and Sterility 65 711–717.
Blithe DL 1990 N-linked oligosaccharides on free á interfere
with its ability to combine with human chorionic
gonadotropin-â subunit. Journal of Biological Chemistry 265
21951–21956.
Buckholz RG & Gleeson MAG 1991 Yeast systems for the
commercial production of heterologous proteins.
Biotechnology 9 1067–1072.
and others
Clare JJ, Romanos MA, Rayment FB, Rowedder JE, Smith
MA, Payne MM, Sreekrishna K & Henwood CA 1991
Production of mouse epidermal growth factor in yeast:
high-level secretion using Pichia pastoris strains containing
multiple gene copies. Gene 105 205–212.
Combarnous Y 1992 Molecular basis of the specificity of
binding of glycoprotein hormones to their receptors.
Endocrine Reviews 13 670–691.
Cregg JM, Vedvick TS & Raschke WC 1993 Recent advances
in the expression of foreign genes in Pichia pastoris.
Biotechnology 11 905–910.
De Koning WJ, Walsh GA, Wrynn AS & Headon DR
1994 Recombinant reproduction. Biotechnology 12
988–992.
Devroey P, Mannaerts B, Smitz J, Coelingh Bennink H &
Steirteghem A 1993 First established pregnancy and birth
after ovarian stimulation with recombinant human follicle
stimulating hormone (Org 32489). Human Reproduction 8
863–865.
Greenberg NM, Anderson JW, Hsueh AJW, Nishimori K,
Reeves JJ, DeAvila DM, Ward DN & Rosen JM 1991
Expression of biologically active heterodimeric bovine
follicle-stimulating hormone in milk of transgenic mice.
Proceedings of the National Academy of Sciences of the USA
88 8327–8331.
Grinna LS & Tschopp JF 1989 Size distribution and
general structural features of N-linked oligosaccharides
from the methylotrophic yeast, Pichia pastoris. Yeast 5
107–115.
Hakola K, Van der Boogaart P, Mulders J, de Leeuw R,
Schoonen W, Van Heyst J, Swolfs A, Van Casteren J,
Huhtaniemi I & Kloosterboer H 1997 Recombinant rat
follicle-stimulating hormone; production of Chinese hamster
ovary cells, purification and functional characterization.
Molecular and Cellular Endocrinology 127 59–69.
Julius D, Schekman R & Thorner J 1984 Glycosylation and
processing of prepro-á-factor through the yeast secretory
pathway. Cell 36 309–318.
Kalandadze A, Galleno M, Foncerrada L, Strominger JL &
Wucherpfennig KW 1996 Expression of recombinant
HLA-DR2 molecules. Replacement of the hydrophobic
transmembrane region by a leucine zipper dimerization motif
allows the assembly and secretion of soluble DR alpha beta
heterodimers. Journal of Biological Chemistry 271
20156–20162.
Keene JL, Matzuk MM, Otani T, Fauser BCJM, Galway AB,
Hsueh AJW & Boime I 1989 Expression of biologically
active human follitropin in Chinese hamster ovary cells.
Journal of Biological Chemistry 264 4769–4775.
Kurjan J & Herskowitz I 1982 Structure of a yeast pheromone
gene (MFá): a putative á-factor precursor contains four
tandem copies of mature á-factor. Cell 30 933–943.
Lapthorn AJ, Harris DC, Littlejohn A, Lustbader JW,
Canfield RE, Machin KJ, Morgan FJ & Isaacs NW 1994
Crystal structure of human chorionic gonadotropin. Nature
369 455–461.
Linder S, Schliwa M & Kube-Granderath E 1996 Direct
PCR screening of Pichia pastoris clones. BioTechniques 20
980–982.
Moore LG, Ng-Chie W, Lun S, Lawrence SB, Young W &
McNatty KP 1997 Follicle-stimulating hormone in the
brushtail possum (Trichosurus vulpecula): purification,
characterization, and radioimmunoassay. General and
Comparative Endocrinology 106 30–38.
Mountford PS, Bello PA, Brandon MR & Adams TE 1989
Cloning and DNA sequence analysis of the cDNA for the
precursor of ovine follicle stimulating hormone â-subunit.
Nucleic Acids Research 17 6391.
Journal of Molecular Endocrinology (1998) 21, 327–336
335
336
and others · Recombinant ovine FSH production by yeast
Mountford PS, Brandon MR & Adams TE 1994 Expression
and characterization of biologically active ovine FSH from
mammalian cell lines. Journal of Molecular Endocrinology 12
71–83.
Nagaya T & Jameson JL 1994 Structural features of the
glycoprotein hormone genes and their encoded proteins. In
The Pituitary Gland, pp 63–89. Ed H Imura. New York:
Raven Press.
Richards JS 1994 Hormonal control of gene expression in the
ovary. Endocrine Reviews 15 725–751.
Samaddar M, Catterall JF & Dighe RR 1997 Expression of
biologically active beta subunit of bovine follicle-stimulating
hormone in the methylotrophic yeast Pichia pastoris. Protein
Expression and Purification 10 345–355.
Sreekrishna K & Kropp KE 1996 Pichia pastoris. In
Nonconventional Yeasts in Biotechnology. A Handbook, pp
203–253. Ed K Wolf. Berlin: Springer Verlag.
Sugahara T, Grootenhuis PDJ, Sato A, Kudo M, BenMenahem D, Pixley MR, Hsueh AJW & Boime I 1996a
Expression of biologically active fusion genes encoding the
common á subunit and either the CGâ or FSHâ subunits:
Journal of Molecular Endocrinology (1998) 21, 327–336
role of a linker sequence. Molecular and Cellular
Endocrinology 125 71–77.
Sugahara T, Sato A, Kudo M, Ben-Menahem D, Pixley MR,
Hsueh AJW & Boime I 1996b Expression of biologically
active fusion genes encoding the common alpha subunit and
the follicle-stimulating hormone beta subunit. Role of a
linker sequence. Journal of Biological Chemistry 271
10445–10448.
Thotakura NR & Blithe DL 1995 Glycoprotein hormones:
glycobiology of gonadotrophins, thyrotrophin and free á
subunit. Glycobiology 5 3–10.
Ulloa-Aguirre A, Midgley AR Jr, Beitins IZ & Padmanabhan
V 1995 Follicle-stimulating isohormones: characterization
and physiological relevance. Endocrine Reviews 16
765–787.
Vedvick T, Buckholz RG, Engel M, Urcan M, Kinney J,
Provow S, Siegel RS & Thill GP 1991 High-level secretion
of biologically active aprotinin from the yeast Pichia pastoris.
Journal of Industrial Microbiology 7 197–202.
5 June 1998