Expressionofthea3/b1isoformof humanNa,K-ATPase in themethylotrophic yeastPichia pastorisCristina Reina1, Gloria Padoani1, Cristina Carotti2, Annamaria Merico2, Grazia Tripodi1, Patrizia Ferrari1

& Laura Popolo2

1Prassis Sigma-Tau Research Institute, Settimo Milanese, Milan; and 2Dipartimento di Scienze Biomolecolari e Biotecnologie, Universita degli Studi di

Milano, Milan, Italy

Correspondence: Laura Popolo, Universita

degli Studi di Milano, Dipartimento di Scienze

Biomolecolari e Biotecnologie, Via Celoria 26,

20133 Milan, Italy. Tel.: 139 02 50314919;

fax: 139 02 50314912; e-mail:

laura.popolo@unimi.it

Received 15 December 2006; revised 25

January 2007; accepted 6 February 2007.

First published online 10 April 2007.

DOI:10.1111/j.1567-1364.2007.00227.x

Editor: Andre Goffeau

Keywords

Pichia pastoris; Na/K pump; Na,K-ATPase a3-

subunit; Na,K-ATPase b1-subunit; protein

expression; protein glycosylation.

Abstract

Na,K-ATPase is a crucial enzyme for ion homeostasis in human tissues. Different

isozymes are produced by assembly of four a- and three b-subunits. The

expression of the a3/b1 isozyme is confined to brain and heart. Its heterologous

production has so far never been attempted in a lower eukaryote. In this work we

explored whether the methylotrophic yeast Pichia pastoris is capable of expressing

the a3/b1 isoform of human Na,K-ATPase. cDNAs encoding the a3 and the b1-

subunits were cloned under the control of the inducible promoter of Pichia pastoris

alcohol oxidase 1. Pichia pastoris could express the single a3- and b1-subunits and

even coexpress them after methanol induction. b1-subunit was produced as a

major 44-kDa glycosylated polypeptide and a3 as a 110-kDa unglycosylated

polypeptide. Expression at the plasma membrane was limited in shaking flask

cultures but by cultivating P. pastoris cells in a fermenter there was a 10-fold

increase of the number of ouabain binding sites per cell. The exported enzyme was

estimated to be about 0.230 mg L�1 at the end of a bioreactor run. Na,K-ATPase

proved active and the dissociation constant of the recombinant enzyme-ouabain

interaction was determined.

Introduction

Na,K-ATPase is an oligomeric membrane protein that

belongs to the P-Type ATPase family of cation transporters.

It plays a crucial role in cellular ion homeostasis and is the

pharmacological receptor for digitalis in man (Muller-

Ehmsen et al., 2002). Na,K-ATPase is composed of stoichio-

metric amounts of two major polypeptides, the a- and

b-subunits. The a-subunit is a multispanning membrane

protein, hydrolyzes ATP, transports the cations and binds

cardiac glycosides with high specificity. The b-subunit is a

type II single-spanning protein necessary for the structural

and functional maturation of the a-subunit, and also

influences the K1 and Na1 activation kinetics of mature

pumps (Blanco & Mercer, 1998; Jorgensen et al., 2003).

Four isoforms for the a-subunit and three for the b-subunit

have been identified and exhibit a tissue-specific distri-

bution and a developmentally regulated pattern of expres-

sion. The a1/b1 complex is found in all tissues and is the

principal isozyme of the kidney. The a2 isoform is expressed

primarily in adipocytes, muscle heart and brain. The a3

isoform is abundant in nervous tissues and present at

lower levels in heart and vessels and the a4 isoform is a

testis-specific isoform. The b2 isoform is mainly expressed

in neural tissue and b3 in testis, retina, liver and lung

(Blanco & Mercer, 1998; Jorgensen et al., 2003). More-

over, the transport properties of Na,K-ATPase can be

modified in a tissue- and isoform-specific way through the

interaction with FXYD proteins (Garty & Karlish, 2006;

Geering, 2006). As a2 and a3 isoforms are usually coex-

pressed in human and other mammalian tissues together

with the a1 isoform, their specific functional and biochem-

ical characterization is hampered by difficulty in purifying

them separately.

The main interest in each isoform function is focused on

heart subunits because Na,K-ATPase is the only known

receptor for cardiac glycosides (ouabain, digoxin and digi-

toxin) and digitalis therapy is widely used in heart failure

treatment (McDonough et al., 1995). Therefore, great efforts

has been made in the last decade to express the individual

functional units of Na,K-ATPase in heterologous systems

with low, or without, background activity (Pedersen et al.,

FEMS Yeast Res 7 (2007) 585–594 c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

1996; Crambert et al., 2000; Muller-Ehmsen et al., 2001;

Blanco, 2005). In general, all these studies reveal new

functional as well as pharmacological characteristics of

Na,K-ATPase isozymes. In particular, the transport charac-

teristics such as turnover, Na1 and K1 affinities and voltage-

dependence, and the pharmacological properties (ouabain

affinity) of human a1, a2 or a3 isoforms together with b1,

b2 or b3-subunits, have been elucidated by their expression

in Xenopus oocytes (Crambert et al., 2000), insect cells

(Blanco et al., 1995) and in part in the yeast Saccharomyces

cerevisiae (Muller-Ehmsen et al., 2001).

Recently, a study on in vitro induction of embryonic stem

cells into neuronal cells indicated that a3 protein is ex-

pressed only when cells have reached a complete morpholo-

gical maturation and proposed a physiological role of a3/b1

isoform. In mature neurons, whereas a1 maintains the basal

ionic gradients, a3 with its lower affinity for Na1 may play a

role in restoring the membrane potential after repeated

action potentials and large influxes of Na1 and its high

affinity for K1 may allow it to function while extracellular

K1 is depleted (Habiba et al., 2000).

Large scale purification of the recombinant proteins is an

important step in the process of characterization of enzymes

for various purposes, including structural studies. These are

particularly difficult for oligomeric membrane proteins such

as Na,K-ATPase. Pichia pastoris has been developed as host

for the efficient production of foreign proteins. It offers

several potential advantages such as a highly inducible

promoter, a good secretion capability and advanced high-

density fermentation technology (Daly & Hearn, 2005;

Macauley-Patrick et al., 2005). Furthermore, the lack of any

background activity makes it suitable for expressing Na,

K-ATPase isoforms. Recently, expression of Na,K-ATPase

(porcine a1/b1) in the methylotrophic yeast P. pastoris,

allowed the purification of recombinant complexes and

investigation of the mechanism of action of phospholem-

man (Cohen et al., 2005; Lifshitz et al., 2006). The expres-

sion of human a1 and a2 isoforms, together with pig b1, is

currently being investigated. However, the expression of

human a3 and b1 isoforms in P. pastoris is still lacking. Here

we report the expression and a partial biochemical char-

acterization of the a3/b1 isoform of human Na,K-ATPase

in P. pastoris.

Materials and methods

Strains and growth conditions

Table 1 lists P. pastoris His� GS115, the protease-deficient

SMD1168 and derived strains. To screen His1 colonies for

Muts phenotype, minimal dextrose and minimal methanol

plates were used. To induce the expression of recombinant

proteins, His1 Muts cells were inoculated in glycerol-com-

plex medium (MGY) and then shifted to methanol-complex

medium (MMY) as previously described (Carotti et al.,

2004).

Construction of expression vectors

The coding region of b1 Na,K-ATPase (NM001677) was

amplified from human kidney cDNA with FPB1 (50-

CGGAATTCCATGGCCCGCGGGAAAGCC-30) and RPB1

(50-GAAAGATTTGTGCTTGTGA-30) primers and cloned

in pCR-XL-TOPO vector (Invitrogen) generating pCR-XL-

TOPO-b1. The full-length cDNA for the a3-subunit

of human Na,K-ATPase (NM152296), previously cloned

in plasmid pBSK1 (pBSK-a3), was kindly provided by

D. Fornasari (University of Milan). The coding region was

amplified with FPA3 (50-CGGAATTCACCATGGGGGA

CAAGAAAGATGACAAG-30) and RPA3 (50-CGGAATTC

GTGGTGGGGCTGAGGTCAGTAGTA-30) primers and

cloned in pCR-XL-TOPO vector to generate pCR-XL-

TOPO-a3. FPB1, FPA3 and RPA3 oligonucleotides contain

an EcoRI site (underlined). Recombinant plasmids pAO815-

b1 and pAO815-a3 were generated by cloning EcoRI-frag-

ments from pCR-XL-TOPO-b1 or pCR-XL-TOPO-a3 into

the corresponding site of the P. pastoris pAO815 vector

(Invitrogen). pAO815-Ua3, was obtained by cloning the

EcoRI-fragment containing the full-length a3 cDNA, into

pAO815. To construct pAO815-a3/b1 the BglII/BamHI

fragment containing the expression cassette 50AOX1-b1-TT

(AOX1, gene encoding alcohol oxidase 1) was transferred

from pAO815-b1 to the BamHI site of pAO815-a3. The

absence of undesired mutations was confirmed by DNA

sequencing.

Transformation of P. pastoris and expression ofa3- and b1-subunits

BglII-digested plasmids were used to transform P. pastoris

cells (Carotti et al., 2004). To induce the expression, His1

Muts cells were inoculated in MGY at a ratio of 1 : 10

Table 1. Strains of Pichia pastoris used

Strains Relevant genotype Source

GS115 his4 Invitrogen

SMD1168 his4 pep4D Invitrogen

G-vector his4 aox1::pAO815[HIS4] This work

S-vector his4 pep4Daox1::pAO815[HIS4] This work

G-b1 his4 aox1::pAO815[50-AOX1-b1-TT-HIS4] This work

S-b1 his4 pep4Daox1::pAO815[5 0-AOX1-b1-TT-HIS4] This work

S-a3 his4 pep4Daox1::pAO815[5 0-AOX1-a3-TT-HIS4] This work

S-Ua3 his4 pep4Daox1::pAO815[5 0-AOX1-Ua3-TT-HIS4] This work

S-a3/b1 his4 pep4Daox1::pAO815[5 0-AOX1-

a3-TT-50-AOX1-b1-TT-HIS4]

This work

FEMS Yeast Res 7 (2007) 585–594c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

586 C. Reina et al.

between volume of culture and volume of flask and grown

overnight at 30 1C under strong agitation. The cells were

then collected and resuspended in MMY at an initial

OD600 nm = 1. Fresh methanol was added daily to 0.5% (v/v).

Batch fermentation cultivations

Batch cultivations were performed in a Biostat-Q-system (B.

Braun, Germany) with a working volume of 850 mL. An air-

flow of 1.2 L min�1 and a stirrer speed of 700–1200 r.p.m.

maintained a dissolved oxygen concentration above 30% of

air saturation. Temperature was kept at 28 1C and pH was

controlled at 6.0 by automatic addition of 2 M KOH.

Precultures and batch cultivations were done in buffered

mineral glycerol medium BMG (13.4 g L�1 yeast nitrogen

base without amino acids, 0.2 g L�1 biotin, 0.1 M potassium

phosphate pH 6.0 and glycerol at 10 and 4 g L�1, respec-

tively). A preculture was used to inoculate batch cultivations

at an initial OD600 nm of 0.1. After complete exhaustion of

glycerol, methanol induction was started by adding metha-

nol at a final concentration of 0.5% (v/v). Induction was

repeated at intervals of 20 h minimum adding 5 or

2.5 mL L�1 of methanol. When biomass reached a concen-

tration of �14 OD600 nm, half the culture was substituted

with fresh medium. Batch experiments were performed in

duplicate.

Small-scale membrane preparation

The protocol is a modification of a previously described

procedure (Weiss et al., 1998). Cells corresponding to

20 OD600 nm were collected from a 10-mL culture and

resuspended in 500 mL of ice-cold breakage buffer-1

(50 mM sodium phosphate pH 7.4, 1 mM EDTA, 5%

glycerol) freshly supplemented with protease inhibitors

[1 mM phenylmethylsulfonyl fluoride (PMSF) and Comple-

teTM Protease Inhibitor Cocktail (Roche)]. After addition of

an equal volume of cold glass beads, cells were broken by

shaking three times for 45 s in a FastPrep 120 at 4 1C

alternating with 1 min in ice. Unbroken cells and glass beads

were removed by a 5-min centrifugation at 10 000 g at 41C.

The cell-free extracts were centrifuged at 100 000 g for

30 min at 4 1C. The supernatant (S100) and the membrane

(P100) fractions were obtained. The membrane fraction was

resuspended in 80 mL membrane suspension buffer (50 mM

Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA) supplemen-

ted with protease inhibitors.

Large-scale membrane preparation and ATPasepurification procedure

From an 850-mL culture in a fermenting vessel, cells

corresponding to 5000 OD600 nm were collected by centrifu-

gation. Cells were broken using a Bead beater at 4 1C in

15 mL of breakage buffer-2 (1.4 M sorbitol, 10 mM MOPS/

Tris-HCl pH 7.2, 1 mM EDTA and supplemented with

protease inhibitors) after addition of an equivalent volume

of cold glass beads. Membranes were essentially prepared as

already described (Strugatsky et al., 2003; Cohen et al.,

2005). The membrane fraction was resuspended in 4 mL of

a buffer containing 10 mM MOPS/Tris-HCl pH 7.2, 25%

glycerol, 1 mM PMSF and stored at � 80 1C until use. The

purification procedure consisted of incubating the mem-

branes (2 mg mL�1) with 0.3–0.6 mg mL�1 SDS for 30 min at

room temperature under continuous stirring in a solution

containing 3 mM Na2ATP, 25 mM imidazole/HCl, 1 mM

Na2EDTA and protease inhibitors (Pedersen et al., 1996).

The SDS-treated membranes were loaded on a step gradient

made up of three successive layers of sucrose: 3.6 mL at

29.4%, 2.2 mL at 15%, 1 mL at 10% and centrifuged in a Ty-

65 Beckman fixed-angle rotor at 231 000 g for 90 min. After

centrifugation, the pellet was resuspended in 25 mM imida-

zole, 1 mM EDTA, pH 7.5 and stored at � 80 1C.

Electrophoresis and immunoblottingprocedures

Protein concentration was determined using a D-C Protein

Assay (Bio-Rad) after membrane solubilization in SDS-

minus buffer (0.0625 M Tris-HCl pH 6.8, 2.3% SDS) for

20 min at 37 1C. After supplementing the samples with

glycerol, b-mercaptoethanol and bromophenol blue, pro-

teins were analyzed by sodium dodecyl sulfate polyacryla-

mide-gel electrophoresis (SDS-PAGE). Total protein extracts

were prepared according to Yaffe & Schatz (1984). As a

control, extracts from dog brain stem, dog renal medulla

and rat brain were loaded. SDS-PAGE and immunoblotting

were carried out as previously described (Gatti et al., 1994).

Rabbit anti-GERK epitope of b1-subunit serum (diluted

1 : 2000) and rabbit anti-KETYY epitope of a-subunits were

kindly donated by S.J.D. Karlish (Weizmann Institute). A

mouse anticanine a3-subunit monoclonal antibody (BIO-

MOL Research Laboratories Inc., Plymouth Meeting, PA)

was used at a concentration of 1 mg mL�1. Peroxidase-con-

jugated affinity-purified F(ab’)2 fragment donkey antirabbit

or antimouse IgG (Jackson ImmunoResearch Laboratories,

West Grove, PA) was used at a 1 : 10 000 dilution. Bound

antibodies were revealed with ECL Western blotting detec-

tion reagents (Amersham Pharmacia Biotech, UK).

Endo-b-N -acetylglucosaminidase H (EndoH)treatment

Membrane proteins (40 or 80 mg) were added with an

equivalent volume of de-glycosylation buffer [200 mM Na

acetate pH 5.5, 0.4% SDS and 2% (v/v) b-mercaptoethanol]

and incubated at 37 1C for 20 min. After halving the sample,

FEMS Yeast Res 7 (2007) 585–594 c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

587a3/b1 isoform of Na,K-ATPase in P. pastoris

aliquots were incubated for 18 h at 37 1C with or without

50 mU of EndoH (Roche).

Equilibrium binding of [3H]ouabain to intactcells

Equilibrium binding to whole cells was performed using

[3H]ouabain (Pedersen et al., 1996). In a saturation experi-

ment, aliquots containing 109 cells were resuspended in

1 mL of 1 M sorbitol, 3 mM MgSO4, 1 mM NaTris2VO4,

1 mM EDTA, 10 mM Tris-HCl pH 7.4 and incubated in the

presence of [3H]ouabain (13 Ci mmol�1, Amersham) at

concentrations ranging from 4� 10�10 to 5� 10�7 M. After

90 min at 37 1C with vigorous shaking, the bound and

unbound ouabain was separated by centrifugation at 1000 g

for 5 min at 4 1C. Samples were then washed twice with 1 mL

of ice-cold water and the amount of bound [3H]ouabain was

determined by liquid-scintillation counting. The nonspeci-

fic binding was evaluated in the presence of an excess of cold

ouabain (10�3 M) and also using cells transformed with the

empty vector. All the measurements were done in triplicate.

The dissociation constant (Kd) of the ouabain-enzyme

complex and the total number of possible binding sites

(Bmax) were calculated by plotting the data according to the

Scatchard method. Competition displacement studies were

performed using a constant concentration of [3H]ouabain

(4� 10�10, 8� 10�8, 10�9, 5� 10�9, 10�8 or 5� 10�7 M) and

variable concentrations of unlabeled ouabain (from 10�9 to

10�6 M) and provided another estimation of the enzyme

affinity for the ligand (Erdmann & Schoner, 1974). The IC50

(concentration of the inhibitor corresponding to 50% of

displacement) was calculated from the displacement curves

by the use of the KALEIDA-GRAPH program (SYNERGY Software

ver 3.6). The Bmax was also determined in separate experi-

ments using a saturating dose of [3H]ouabain (5� 10�7 M)

under the above-mentioned conditions.

ATPase activity assay

The ATPase assay was performed both on crude membranes

from a large-scale preparation and on the purified enzyme

(see ‘Large-scale membrane preparation and ATPase purifi-

cation procedure’): aliquots containing 0.3–5 mg of protein

were incubated for 10 min at 37 1C with 140 mM NaCl,

3 mM MgCl2, 3 mM EGTA, 67 mM Hepes, 1 mM ATP,

10 mM KCl (using g[32P]ATP as a tracer), 1.2 mM EDTA

pH 7.4 in the presence or absence of 7.5 mM ouabain. The

reaction was stopped with ice-cold 10% perchloric acid and

the tubes were centrifuged at 2000 g for 5 min. The radio-

active excess ATP was adsorbed by adding 500 mL of charcoal

for 15 min at room temperature and the released

[g-32P]phosphate was counted in 300mL of supernatant

from each sample after centrifugation at 2000 g for 30 min.

The Na1,K1-ATPase activity was calculated as the difference

in c.p.m. for tubes with or without 7.5 mM ouabain.

Results

Expression of the single b1- and a3-subunits ofNa,K-ATPase in P. pastoris

The human b1 and a3 cDNAs were cloned in the P. pastoris

pAO815 vector that harbors an expression cassette consti-

tuted by the 50-flanking region of AOX1, an EcoRI cloning

site, the 30-transcription termination of AOX1, the HIS4

marker and an extended 30 flanking region of the AOX1 gene

(Fig. 1). The insertion of cDNA places the coding sequence

under the control of the AOX1 promoter and makes the

expression inducible by methanol. The long 50- and 30-

regions of AOX1 promote the homologous recombination

of the BglII/BglII restriction fragment into the AOX1

chromosomal locus. Integration into the AOX1 locus causes

replacement of the AOX1 gene with the exogenous fragment

and determines the acquisition of His1 and Muts pheno-

types. The pAO815-b1 and pAO815-a3 plasmids and the

vector were used to transform the wild-type GS115 and

the protease-deficient SMD1168 strains. About 10% of the

selected His1 transformants were also Muts. His1 Muts cells,

pregrown in MGY, were routinely shifted at time zero to

methanol-containing medium (MMY).

Figure 2a shows representative growth kinetics. Cells grew

exponentially for about 24 h after the shift, then growth

progressively decreased and ceased between 48 and 72 h after

the shift. Similar kinetics was obtained for the GS115-

derived strains, GS-vector and G-b1, and for the protease-

deficient SMD1168-derived strains, S-vector, S-b1 and S-a3

strains, although GS115-derived strains reached a higher

density. A 44-kDa polypeptide was detected by immunoblot

using anti-b1 serum in total protein extracts from G-b1 and

S-b1 strains at 48 and 72 h after induction (Fig. 2b) and in

membrane fractions starting 24 h from induction (data not

shown). This band was identified as a b1-subunit because it

was absent in extracts from GS-vector and S-vector (V in

Fig. 2) and from recombinant clones at time zero after

induction. An additional closely migrating band, indicated

by an asterisk in Fig. 2b, is a cross-reactive that antiserum

exhibited in initial uses.

To determine whether b1 was a membrane protein,

membrane (P100) and soluble (S100) fractions were ana-

lyzed. The immunoblot in Fig. 2c shows that b1 was totally

recovered in the membrane fraction as a major band of

44 kDa. As the expected molecular mass of b1-subunit is

about 35 kDa and three potential N-glycosylation sites are

present in the ectodomain, the 44-kDa polypeptide could be

glycosylated. After treatment with EndoH, which removes

the N-linked chains, the 44-kDa polypeptide shifted to a

FEMS Yeast Res 7 (2007) 585–594c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

588 C. Reina et al.

band of about 35 kDa (see also below), suggesting that a

glycan moiety of about 9 kDa is linked to the polypeptide

(Fig. 2d). Thus, the b1-subunit is less glycosylated in

P. pastoris than in mammalian cells, where its molecular

mass is about 55 kDa (Fig. 2d, Control). The presence of

short N-linked chains is consistent with a less frequent

hyperglycosylation in this microorganism than in S. cerevi-

siae [for a review see Daly & Hearn (2005)] and with

previously reported data on porcine b1-subunit expressed

in P. pastoris (Cohen et al., 2005). b1 appeared to be more

susceptible to in vitro degradation in the GS115-derived

fractions, therefore all the following studies were performed

using SMD1168.

Pichia pastoris cells were also able to efficiently express the

human a3-subunit. Plasmid pAO815-a3 was used to trans-

form SMD1168 cells. Growth kinetics in methanol of the S-

a3 recombinant strain was unchanged with respect to S-b1

(Fig. 2a). A polypeptide of about 110 kDa was detected in

the membrane fraction at 24 h after induction and slightly

decreased at 48 h. This band was recognized by a mono-

clonal anti-a3-subunit antibody (Fig. 2e) as well as by anti-

a-subunit KETYY antiserum (data not shown), indicating

that it corresponds to the a3-subunit. This molecular mass

is consistent with the predicted value of 111 734.5. Interest-

ingly, the a3-polypeptide was expressed only when the

AOX1 50-flanking region was directly fused to the a3 cDNA

coding region. Recombinant clones isolated by transforma-

tion with pAO815-Ua3 (Fig. 1) expressed a high amount of

a3 mRNA but the a3-subunit was undetectable (data not

shown). Thus, the presence of the G1C-rich 50UTR (un-

translated region) of a3 mRNA was detrimental to produc-

tion of the protein, as observed for other UTRs (Sreekrishna

et al., 1997) and also for the same human 50UTR of a3

cDNA in Xenopus oocytes (Crambert et al., 2000).

Coexpression of b1- and a3-subunits and surfacedetection of Na,K-ATPase

Figure 3 shows the coexpression of b1- and a3-subunits in

cells transformed with pAO815-. The growth rate of S-a3/b1

is similar to that of S-vector, S-a3 and S-b1 strains, indicat-

ing that coexpression of the two subunits does not affect

growth (Fig. 3a). Figure 3b shows the immunoblot of

equivalent amounts of membrane proteins prepared at

different time intervals after methanol induction. In fact,

24 h after methanol induction, the a3-polypeptide was

detected as a single 110-kDa band. After that time, the level

of the protein decreased, which is consistent with the fact

that protein synthesis and secretion are tightly coupled to

growth in yeast. A densitometric analysis indicated that the

protein level slowly declined as growth ceased (half-time

�35 h). In the same immunoblot the b1-subunit was also

detected (Fig. 3b, lower panel). Three bands of apparent

molecular mass of 44, 41 and 35 kDa were observed. Their

levels decreased as growth ceased. The 44-kDa band comi-

grated with the 44-kDa band detected in cells expressing

only b1 (data not shown). After treatment of the membrane

fraction with EndoH, the 44- and 41-kDa polypeptides were

no longer detected and the intensity of the 35-kDa band

increased (Fig. 3c, lower panel). This indicates that the

44 kDa form is the fully glycosylated form of b1 and the 41-

and 35-kDa polypeptides are the underglycosylated and

unglycosylated forms, respectively. The 41-kDa form might

represent an intermediate precursor of the maturation

Fig. 1. Scheme of the expression plasmids used. 50AOX1, 50 flanking region of AOX1; 30 AOX1, 30 flanking region of AOX1 containing the

transcription termination sequence (TT); EcoRI, cloning site. pAO815-Ua3 contained 133 nt of the 50-UTR of human a3 cDNA.

FEMS Yeast Res 7 (2007) 585–594 c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

589a3/b1 isoform of Na,K-ATPase in P. pastoris

pathway of the b1-subunit. The endoplasmic reticulum

(ER) to Golgi transit is likely to be a limiting step because

coexpression/assembly of b1 with a3 occurs in the ER and

precedes the transport to the plasma membrane. The

electrophoretic mobility of the a3 polypeptide was unaf-

fected by EndoH, indicating that this protein is not glycosy-

lated (Fig. 3c, upper panel).

The presence of human Na,K-ATPase at the level of the

plasma membrane was tested by measuring the ability of

intact P. pastoris cells to bind radioactive ouabain, a specific

ligand and inhibitor of Na,K-ATPase. Both saturation and

competition experiments were performed. The binding of

[3H]ouabain was measured in a saturation experiment using

different concentrations of radioligand from 4� 10�10 up to

5� 10�7 M. The curve revealed the presence of specific and

saturable binding sites for ouabain at the yeast cell surface

(data not shown). The dissociation constant (Kd) of the

exported enzyme was calculated from the linearization of

the data of binding according to the Scatchard analysis

(Fig. 4a). From the slope of the straight line (� 1/Kd), the

dissociation constant was determined to be 6.29� 10�8 M.

On the basis of the total number of possible binding sites

(Bmax) obtained from the intercept of the plot on the

abscissa, a total of 180 binding sites per cell was determined.

Control cells transformed with the empty vector exhibited

no binding (data not shown). Ligand displacement curves

were obtained using fixed concentrations of radioactive

ouabain and increasing concentrations of the same cold

ligand. Similar curves were obtained. The curve resulting

from different displacement experiments is shown in Fig. 4b.

The mean of the values of IC50 is 3.29� 2.7� 10�8 M

(mean� SD, n = 7). As previously reported (Erdmann &

Fig. 2. Expression of the single human b1- and

a3-subunits in Pichia pastoris. (a) Growth ki-

netics of representative clones of GS-vector (m),

GS-b1 (n) and S-vector (’), S-b1 (�) and S-a3

(^) strains inoculated at time zero in MMY in

shaking flasks at 30 1C. (b) Immunoblot with

anti-b1 serum of total protein extracts prepared

from culture of S-b1 and G-b1 cells grown as

shown in (a) at the indicated time intervals.

Asterisk (�) indicates a cross-reactive band. (c)

Human b1-subunit is present in the membrane

fraction. A homogenate of Pichia pastoris cells

collected at 48 h was subjected to a centrifuga-

tion fractionation at 100 000 g for 30 min at

4 1C and S100 and P100 fractions were ana-

lyzed by immunoblotting with anti-b1 serum.

T, total homogenate (�150mg total proteins);

S, S100 (120mg of proteins); P, P100 (30 mg).

(d) Human b1 isoform is N-glycosylated in Pichia

pastoris. Immunoblot analysis with anti-b1 anti-

serum of the P100 fraction after EndoH treat-

ment. Forty micrograms of membrane protein

was incubated in the absence (� ) or presence

(1) of EndoH. Control, rat brain extract. (e) a3-

subunit is present in the membrane fraction.

Immunoblot of S100 (S) and P100 (P) fractions,

120 and 30 mg respectively of proteins, were

analyzed by immunoblot with anti-a3 Mab.

Control, 2 mg of protein extract from dog brain

stem.

FEMS Yeast Res 7 (2007) 585–594c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

590 C. Reina et al.

Schoner, 1974), the IC50 represents an accurate estimation of

the Kd of the ouabain-Na,K-ATPase interaction provided

that the equilibrium conditions used in the competition

experiments are the same of those used in the saturation

experiments. Consistently, the mean value of IC50

(3.29� 10�8 M) is very close to the value obtained from the

Scatchard transformation of the saturation curve.

The Bmax was also determined using a saturating dose of

radioactive ouabain (5� 10�7 M). A value of 234� 135 sites

per cell (mean� SD, n = 5) was obtained.

Properties of a3/b1 isoforms of Na,K-ATPaseexpressed in batch fermentation

To increase the expression of a3/b1 Na,K-ATPase we grew

P. pastoris cells in a fermenter. The best condition was a two-

phase growth/induction. After the first growth phase in

glycerol-containing medium, methanol was added (0.5%

final concentration) at the time indicated by the arrow to

induce the expression of the proteins (Fig. 5a). Muts

phenotype determines a very slow growth rate on methanol.

Eighty-nine hours after the beginning of the experiment,

cells were diluted by addition of fresh medium. This practice

ensured a highly aerobic environment and prompt avail-

ability of salts and vitamins, leading to a new active growth

phase during the next 5 days. The number of ouabain-

binding sites per cell (Bmax) was determined in the second

induction phase by measuring the binding of ouabain to

intact cells at a saturating dose of the ligand (5� 10�7 M).

There was a rapid increase of this value and the number of

binding sites per cell was 1414� 34 (mean� SD, n = 3) at

186 h (Fig. 5a, triangles). Protein production was also

monitored. Figure 5b shows that a3 and b1 were undetect-

able at the moment of methanol addition (18 h, lane 1),

became detectable at 42, 66 and 88 h (lanes 2–4) and

increased at 120, 137 h (lanes 5 and 6) and again at 161 and

186 h after the beginning of the experiment (lanes 7 and 8).

At 186 h, the a3 level reached c. 3% of membrane protein

(0.4% of total protein) as determined by densitometric

analysis of the Coomassie-stained 110-kDa band, using b-

galactosidase as a reference. The glycosylated nature of b1

caused poor staining and hampered quantification. In con-

clusion, the batch growth in the fermenter allowed more

protein to be exported to the plasma membrane.

The functionality and binding property of the protein

were examined. Saturation experiments were performed

using cells collected at 186 h by batch fermentation. The Kd

was the same obtained using cells grown in shaking flasks

(see above), indicating that, independent of the modes of

growth, the binding property of the surface-available en-

zymes is unchanged. Total ATPase activity was determined

using a 32P-ATP hydrolysis assay in the raw membrane

fraction from a large-scale preparation at the 186 h time

point. Specific activities of 65.94� 21 (mean� SD, n = 9)

and 22.95� 10.8 (mean� SD, n = 6) mmol of Pi h�1 mg�1

were obtained for the P. pastoris recombinant and control

Fig. 3. Coexpression of a3/b1 isoforms of hu-

man Na,K-ATPase. (a) Growth kinetics of a

culture of S-a3/b1 strain in MMY in a shaking

flask at 30 1C. Arrows indicate membrane pre-

paration time. (b) Kinetics of the coexpression:

immunoblot analysis of 30mg of membrane

fractions (P100) prepared at 0, 24, 48, 72 and

96 h after methanol induction from S-a3/b1

strain. V indicates the P100 fraction from S-

pAO815 (empty vector) at 48 h after induction.

C, Control: extract from dog brain stem. (c)

Glycosylation profile of b1 and a3 isoforms

coexpressed in Pichia pastoris. About 20mg of a

P100 fraction, prepared from S-a3/b1 cells at

48 h after induction, was incubated in the ab-

sence (lane 1) or presence (lane 2) of EndoH. The

upper part of the blot was immunodecorated

with anti-a3 Mab, and the lower part with anti-

b1 antibodies. The 44-, 41- and 35-kDa forms of

the b1 subunit are given in brackets. Lane 3,

2 mg of protein extract from dog renal medulla;

lane 4, 6 mg of protein extract from dog brain

stem.

FEMS Yeast Res 7 (2007) 585–594 c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

591a3/b1 isoform of Na,K-ATPase in P. pastoris

strains, respectively. Thus, the recombinant strain exhibited

approximately three times more ATPase activity than the

control. To measure the fraction inhibited by ouabain,

membranes were treated with SDS, in a concentration range

known to preferentially inhibit yeast plasma membrane H1-

ATPase and also demask Na,K-ATPase present in closed

vesicles [see Pedersen et al. (1996) and ‘Large-scale mem-

brane preparation and ATPase purification procedure’]. The

specific activity was 4.8� 2.86 (mean� SD, n = 5) mmol

mg�1 h�1, corresponding to c. 15% of total ATPase activity

measured under these conditions [32.82� 7.5 (mean� SD,

n = 5) mmol of Pi h�1 mg�1.

From the number of ouabain binding sites per cell and the

total amount of a3-subunit produced we calculated that a

fraction of about 15% of the total a3 protein is present at the

cell surface. This value and the low activity measured suggest

that a significant fraction of inactive or partially active pools

of Na,K-ATPase is recovered from intracellular membranes.

Thus, the a3/b1 isoform of human Na,K-pump is not

exported efficiently in P. pastoris either for inefficient folding

and assembly of the subunits or for a disparity of production

of the two subunits, as previously reported for a3/b1

expressed in other heterologous systems (Blanco et al.,

1995).

Discussion

The a3/b1 isoform of human Na,K-ATPase, a notably

difficult enzyme to express in a lower eukaryote, was

produced in P. pastoris and partially characterized. The

recombinant a3-subunit is a 110-kDa polypeptide that is

not glycosylated as it is in its natural form (Crambert et al.,

2000). The recombinant b1-subunit is a 44-kDa glycopro-

tein that is less glycosylated than in a natural environment.

Fig. 4. Radioligand binding studies of a3/b1 isoform heterologously

produced in Pichia pastoris. Binding studies were performed on intact

cells grown in shaking flasks at 30 1C. (a) Scatchard transformation of the

binding data obtained from a saturation curve of [3H]ouabain binding

performed as described in ‘Materials and methods’. R2 = 0.9598. A value

of Kd = 6.29�10�8 M was calculated from the slope of the line (� 1/Kd).

From the intercept on the x-axis (0.295), a value of Bmax/cell = 180 was

obtained. (b) Ligand-displacement curve. Each point is the mean of three

measurements� SE. An IC50 of 3.29� 2.7� 10�8 M (mean� SD, n = 7)

was calculated by the KALEIDA-GRAPH program from seven independent

ligand displacement curves. R = 0.997.

Fig. 5. Expression of a3/b1 isoforms of human Na,K-ATPase in a

fermenter. (a) Growth kinetics of the S-a3/b1 strain during batch

fermentation cultivation: ’, OD600 nm values; m, number of ouabain

binding sites per cell. Standard deviations were o 5% of the mean value

and are not reported in the graph. At time zero, cells were inoculated in

an 850-mL fermenter. The black arrow indicates moment of first

methanol addition. After 89 h, cells were diluted (second arrow on the

x-axis), as described in ‘Materials and methods’. Empty arrows indicate

the time points for membrane preparation. (b) Immunoblot analysis of

membrane fractions (P100) from cells collected at the time indicated by

the empty arrows in (a) and corresponding to the lanes: lane 1, 18 h

(methanol addition); lane 2, 42 h; lane 3, 66 h; lane 4, 88 h; lane 5, 120 h;

lane 6, 137 h; lane 7, 161 h; lane 8, 186 h. For each time, 5 mg of

membrane proteins were loaded. After blotting, the filter was cut into

two parts: the upper part was decorated with anti-a3 Mab and the lower

part with anti-b1 serum. Lane C, extract from dog brain stem.

FEMS Yeast Res 7 (2007) 585–594c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

592 C. Reina et al.

Both a3- and b1-subunits can also be expressed separately,

whereas in Xenopus and mammalian cells, the a-subunit

expressed in the absence of the b-subunit undergoes degra-

dation as b-subunit in these systems is crucial for protein

stabilization and transport to the plasma membrane (Beguin

et al., 2000).

Cells expressing the a3/b1 complex showed a number of

ouabain binding sites per cell in a shaking flask, generally in

agreement with the number obtained for other isoforms in

yeast S. cerevisiae (Horowitz & Farley, 1988; Horowitz et al.,

1990) but lower than that reported in S. cerevisiae and

P. pastoris for isoforms other than a3/b1 (Pedersen et al.,

1996; Crambert et al., 2000; Strugatsky et al., 2003). How-

ever, P. pastoris expressed about 10 times more ouabain

binding sites per cell in a batch fermenter, confirming

the advantages this microorganism offers under appropriate

fermentation conditions. The dissociation constant of the

enzyme-ouabain interaction calculated from the displace-

ment curves was 3.29� 2.7� 10�8 M and is close to the

value of Kd obtained from the Scatchard analysis

(6.29� 10�8 M). This value is in agreement with those

reported for human Na,K-ATPase a/b1 isoforms expressed

in the yeast S. cerevisiae (Muller-Ehmsen et al., 2001),

human tissues, insect cells and cultured cells expressing

different combinations of Na,K-ATPase isoforms (Blanco

et al., 1995; Wang et al., 2001). However, this value is about

10 times lower than that reported for the human a3/b1

isoform expressed in Xenopus oocytes (Beguin et al., 2000).

Ligand affinities measured for other membrane proteins,

e.g. mouse 5-hydroxytryptamine or human b2-adrenergic

receptors produced in P. pastoris, were likewise lower than

the ones the receptors expressed in their natural environ-

ment (Weiss et al., 1998). It was proposed that ergosterol,

which replaces cholesterol in yeast membranes, can affect

membrane fluidity or, in a direct way, the ligand-binding

properties of the heterologously produced membrane pro-

teins. In this regard, recent evidence indicates that lipids play

an essential role, particularly in preserving Na,K-ATPase

activity during purification, and lipid-protein interactions

are probably integral structural components of various

membrane proteins (Lee, 2002; Cohen et al., 2005). Another

possible explanation for the observed discrepancies could

be that the reported values were measured in enriched

membrane preparations, whereas we used intact yeast cells

that might leak K1 ions known to affect the ouabain binding

(Crambert et al., 2000).

The recombinant a3/b1 isoform of Na,K-ATPase de-

scribed here was active. The recombinant clone produced

three times more ATPase activity than the control strain.

Moreover, Na,K-ATPase activity was measured using a

treatment with low concentrations of SDS, which is known

to inhibit H1-ATPase, an abundant proton pump of the

yeast plasma membrane which interferes with the assay of

ATPase activity assay. From the number of ouabain sites per

cell obtained through batch fermentation, a yield of about

0.23 mg of surface enzymes per liter of culture can be

calculated at an OD600 nm of about 14. Further optimization

of production could be achieved by testing different growth-

induction strategies or growth temperatures and by exploit-

ing the great versatility of P. pastoris to grow at high density

in fermenters. In addition, the combination with protein-

fusion strategies (Yokoyama, 2003), as recently shown for

the porcine a1/b1 isozyme (Strugatsky et al., 2003) could

also facilitate the purification of the human a3/b1 isoform

for further biochemical and structural characterizations and

comparison with other isoforms.

Acknowledgements

The work in the Popolo lab was partially supported by

Fondo Interno Ricerca Scientifica e Tecnologica 2004–2005

and 2005–2006 to L.P. C.C. was recipient of a fellowship

from Prassis Sigma-Tau. The authors wish to thank Dr

Concetta Compagno for advice in the fermentation experi-

ments and helpful discussions and Marlene Deutsch for

English revision.

Statement

Human a1 and a2 isoforms, together with pig His10b1, have

recently been expressed. Results communicated by

Y. Lifshitz, H. Garty, and S.J.D. Karlish at the 11th Interna-

tional ATPase Conference, September 6–11, 2005, Marine

Biological Laboratory, Woods Hole, MA.

References

Beguin P, Hasler U, Staub O & Geering K (2000) Endoplasmic

reticulum quality control of oligomeric membrane proteins:

topogenic determinants involved in the degradation of the

unassembled Na,K-ATPase alpha subunit and in its

stabilization by beta subunit assembly. Mol Biol Cell 11:

1657–1672.

Blanco G (2005) The NA/K-ATPase and its isozymes: what we

have learned using the baculovirus expression system. Front

Biosci 10: 2397–2411.

Blanco G & Mercer RW (1998) Isozymes of the Na-K-ATPase:

heterogeneity in structure, diversity in function. Am J Physiol

275: F633–F650.

Blanco G, Sanchez G & Mercer RW (1995) Comparison of the

enzymatic properties of the Na,K-ATPase alpha 3 beta 1 and

alpha 3 beta 2 isozymes. Biochemistry 34: 9897–9903.

Carotti C, Ragni E, Palomares O, Fontaine T, Tedeschi G,

Rodriguez R, Latge JP, Vai M & Popolo L (2004)

Characterization of recombinant forms of the yeast Gas1

protein and identification of residues essential for

glucanosyltransferase activity and folding. Eur J Biochem 271:

3635–3645.

FEMS Yeast Res 7 (2007) 585–594 c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

593a3/b1 isoform of Na,K-ATPase in P. pastoris

Cohen E, Goldshleger R, Shainskaya A, Tal DM, Ebel C, le Maire M

& Karlish SJ (2005) Purification of Na1,K1-ATPase expressed

in Pichia pastoris reveals an essential role of phospholipid-

protein interactions. J Biol Chem 280: 16610–16618.

Crambert G, Hasler U, Beggah AT, Yu C, Modyanov NN,

Horisberger JD, Lelievre L & Geering K (2000) Transport and

pharmacological properties of nine different human Na,

K-ATPase isozymes. J Biol Chem 275: 1976–1986.

Daly R & Hearn MT (2005) Expression of heterologous proteins

in Pichia pastoris: a useful experimental tool in protein

engineering and production. J Mol Recognit 18: 119–138.

Erdmann E & Schoner W (1974) Ouabain-receptor interactions

in (Na1 plus K1)-ATPase preparations. IV. The molecular

structure of different cardioactive steroids and other

substances and their affinity to the glycoside receptor. Naunyn

Schmiedebergs Arch Pharmacol 283: 335–356.

Garty H & Karlish SJ (2006) Role of FXYD proteins in ion

transport. Annu Rev Physiol 68: 431–459.

Gatti E, Popolo L, Vai M, Rota N & Alberghina L (1994) O-linked

oligosaccharides in yeast glycosyl phosphatidylinositol-

anchored protein gp115 are clustered in a serine-rich region

not essential for its function. J Biol Chem 269: 19695–19700.

Geering K (2006) FXYD proteins: new regulators of Na-K-

ATPase. Am J Physiol Renal Physiol 290: F241–F250.

Habiba A, Blanco G & Mercer RW (2000) Expression, activity and

distribution of Na,K-ATPase subunits during in vitro neuronal

induction. Brain Res 875: 1–13.

Horowitz B & Farley RA (1988) Development of a heterologous

gene expression system for the Na,K-ATPase subunits in the

yeast Saccharomyces cerevisiae. Prog Clin Biol Res 268B: 85–90.

Horowitz B, Eakle KA, Scheiner-Bobis G, Randolph GR, Chen

CY, Hitzeman RA & Farley RA (1990) Synthesis and assembly

of functional mammalian Na,K-ATPase in yeast. J Biol Chem

265: 4189–4192.

Jorgensen PL, Hakansson KO & Karlish SJ (2003) Structure and

mechanism of Na,K-ATPase: functional sites and their

interactions. Annu Rev Physiol 65: 817–849.

Lee AG (2002) Ca21-ATPase structure in the E1 and E2

conformations: mechanism, helix–helix and helix–lipid

interactions. Biochim Biophys Acta 1565: 246–266.

Lifshitz Y, Lindzen M, Garty H & Karlish SJ (2006) Functional

interactions of phospholemman (PLM) (FXYD1) with

Na1,K1-ATPase. Purification of alpha1/beta1/PLM complexes

expressed in Pichia pastoris. J Biol Chem 281: 15790–15799.

Macauley-Patrick S, Fazenda ML, McNeil B & Harvey LM (2005)

Heterologous protein production using the Pichia pastoris

expression system. Yeast 22: 249–270.

McDonough AA, Wang J & Farley RA (1995) Significance of

sodium pump isoforms in digitalis therapy. J Mol Cell Cardiol

27: 1001–1009.

Muller-Ehmsen J, Juvvadi P, Thompson CB, Tumyan L, Croyle M,

Lingrel JB, Schwinger RH, McDonough AA & Farley RA

(2001) Ouabain and substrate affinities of human Na(1)-

K(1)-ATPase alpha(1)beta(1), alpha(2)beta(1), and

alpha(3)beta(1) when expressed separately in yeast cells.

Am J Physiol Cell Physiol 281: C1355–C1364.

Muller-Ehmsen J, McDonough AA, Farley RA & Schwinger RH

(2002) Sodium pump isoform expression in heart failure:

implication for treatment. Basic Res Cardiol 97 (Suppl 1):

I25–I30.

Pedersen PA, Rasmussen JH & Joorgensen PL (1996) Expression

in high yield of pig alpha 1 beta 1 Na,K-ATPase and inactive

mutants D369N and D807N in Saccharomyces cerevisiae.

J Biol Chem 271: 2514–2522.

Sreekrishna K, Brankamp RG, Kropp KE, Blankenship DT, Tsay

JT, Smith PL, Wierschke JD, Subramaniam A & Birkenberger

LA (1997) Strategies for optimal synthesis and secretion of

heterologous proteins in the methylotrophic yeast Pichia

pastoris. Gene 190: 55–62.

Strugatsky D, Goldshleger R, Bibi E & Karlish SJ (2003)

Expression of Na,K-ATPase in P. pastoris: Fe21-catalyzed

cleavage of the recombinant enzyme. Ann NY Acad Sci 986:

247–248.

Wang J, Velotta JB, McDonough AA & Farley RA (2001) All

human Na(1)-K(1)-ATPase alpha-subunit isoforms have a

similar affinity for cardiac glycosides. Am J Physiol Cell Physiol

281: C1336–C1343.

Weiss HM, Haase W, Michel H & Reilander H (1998) Comparative

biochemical and pharmacological characterization of the mouse

5HTSA 5-hydroxytryptamine receptor and the human beta2-

adrenergic receptor produced in the methylatrophic yeast.

Pichia pastoris. Biochem J 330: 1137–1147.

Yaffe MP & Schatz G (1984) Two nuclear mutations that block

mitochondrial protein import in yeast. Proc Natl Acad Sci USA

81: 4819–4823.

Yokoyama S (2003) Protein expression systems for

structural genomics and proteomics. Curr Opin Chem Biol 7:

39–43.

FEMS Yeast Res 7 (2007) 585–594c� 2007 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. All rights reserved

594 C. Reina et al.