The phosphatase activity of the isolated H
4
-H
5
loop of Na
+
/K
+
ATPase
resides outside its ATP binding site
Rita Krumscheid
1
,Ru¨ diger Ettrich
2
, Zofie Sovova
´
2
, Kla
´
ra Sus
ˇ
a
´
nkova
´
3
, Zdene
ˇ
kLa
´
nsky

The structural stability of the large cytoplasmic domain (H
4
-
H
5
loop) of mouse a
1
subunit of N a
+
/K
+
ATPase (L354–
I777), the number a nd the location of its binding sites for
2¢-3¢-O-(trinitrophenyl) adenosine 5¢-triphosphate (TNP-
ATP) and p-nitrophenylphosphate (pNPP) were investi-
gated. C- and N-terminal shortening revealed that neither
part of the phosphorylation (P)-domain are necessary for
TNP-ATP binding. There is no indication of a second ATP
site on the P -domain of the isolated loo p, e ven though
others reported previously of its existence by TNP-N
3
ADP
affinity labeling of t he full enzyme. F luorescein isothio-
cyanate (FITC)-anisotropy measurements reveal a consid-
erable stability of the nucleotide (N)-domain suggesting that
it may not undergo a substantial conformational change
upon ATP binding. The FITC modified loop showed only
slightly diminished phosphatase activity, most likely due to a
pNPP site on the N-domain around N398 whose mutation
to D reduced the phosphatase activity by 50%. The amino

ATPase (EC 3.6.3.9) or sodium pump carries
out the coupled extrusion and uptake of Na
+
and K
+
ions
across plasma membranes of mammalian cells. The enzyme
is a heterodimer of a 100 kDa catalytic subunit and a
heavily glycosylated b subunit of about 55 kDa [1–3].
Ouabain, recently recognized as a m ammalian s teroid
hormone [4], uses the sodium pump in the nanomolar
concentration range as a signal transducer [5] but inhibits it
at higher (toxic) concentrations [1–3]. The ion pumping
process is connected to a reaction cycle model with
conformational changes of the catalytic a subunit. Such
changes become visible amongst others in the Na
+
dependent generation of an aspartyl (D369) phosphointer-
mediate with different sensitivities towards the reaction
product ADP or the second transport substrate K
+
.The
observation of high and low affinity ATP sites with
approximate K
d
values of 1 l
M
(E
1
ATP site) and 200 l

reacted with Co(NH
3
)
4
ATP [14,15] or erythrosin isothiocy-
anate [16]. The latter binds to C549 within the nucle otide
(N)-domain of Na
+
/K
+
ATPase [16]. Molecular distance
measurements after specific labeling of the high and low
affinity ATP sites with these fluorescent probes gave
evidence for the existe nce of a n (ab)
2
dimeric structure
[17]. The finding of full-site, half-site and quarter-site
phosphorylation and reactivities, however, l ed Taniguchi
et al. [12] and Froehlich et al.[18]topostulatetheexistence
of a functional (ab)
4
tetrameric structure of N a
+
/K
+
ATPase.
Correspondence to W. Schoner, Institute of Biochemistry and Endo-
crinology, Justus-Liebig-University Giessen, Frankfurter Str. 100,
D-35392 Giessen, Germany. Fax: +49 641 9938179,
Tel.: +49 641 9938170,

+
/K
+
ATPase [20]. The overall structure for the loop
between L354 and L 773 excellently predicted the r eal
structure which was obtained much later by crystallization
and NMR spectroscopy [21,22]. Some deviations from
Hakanson’s crystal structure of a much shorter nucleotide
loop (R378–D586) were noted, however. Hence, some
corrections were recently performed to interpret v ariations
in the location of ATP and 2¢-3¢-O-(trinitrophenyl) adeno-
sine 5¢-triphosphate, trisodium salt (TNP-ATP) binding
within the N-domain (L354–I604) [23]. In silico docking of
ATP as well as NMR studies demonstrated that the H
4
-H
5
loop consisting of the N- and phosphorylation (P)-domains
contains a s ingle ATP site only on the N-domain [20].
Nevertheless, affinity labeling by [
32
P]8-azido-ADP[aP] of
FITC-inactivated and membrane-embedded Na
+
/K
+
ATPase revealed that an amino a cid sequence residing on
the P-domain C-terminally of K736 is involved in ADP
recognition [24,25]. K736 of the a subunit has formerly been
shown to participate in ATP hydrolysis [26,27]. So far it is

expressed in Escherichia coli retains both TNP-ATP binding
[23,34–36] and phosphatase activity [36], the possibility
arose to localize both activities within this loop and to study
their properties. This paper localizes by truncation, single
site mutation and in silico docking experiments, the position
of the binding site for ATP at the front side of the N-domain
between I390 and L576 and reveals the separate existence
of a p-nitrophenylphosphatase at the rear site o f the
N-domain. Our studies gave no indication of a s econd
ATP binding site in the self-forming conformer of the
isolated H
4
-H
5
loop.
Experimental procedures
All chemicals were of the highest purity available and were
obtained from Applichem (Darmstadt, Germany), Bio-Rad
(Munich, Germany), B oehringer-Mannheim ( Mannheim,
Germany),E.Merck(Darmstadt,Germany),Sigma-
Aldrich (Taufkirchen, Germany) Molecular Probes
(Eugene, OR, USA) or Carl Roth (Karlsruhe, Germany).
Pfu-polymerase was from Stratagene (La Jolla, CA, USA)
and the restriction endonucleases BamHI and EcoRI were
from Promega ( Mannheim, Germany). The pGEX-2T
expression vector was from Amersham Biosciences (Frei-
burg, Germany). DNA miniprep and DNA gel extraction
kits were from peqLab (Erlangen, Germany) and Qiagen
(Hilden, Germany). Supercompetent E. coli XL1 b lue cells
were bought from Stratagene. BL21DE3 cells were a

4
-H
5
loop–glutathione S-transferase (GST) fusion protein
and its truncation products. K
+
activated p-nitrophenyl-
phosphatase as a partial activity of Na
+
/K
+
ATPase was
measured as described previously [17]. S tatistical analysis of
the comparison of the phosphatase activities in truncated
H
4
-H
5
loop was carried out with Student’s t-test.
Construction and purification of H
4
-H
5
loop–GST fusion
proteins
The part of the DNA sequence of the a subunit of mouse
brain Na
+
/K
+

TAGGTCATCATGGTC-3¢; antisense with stop
3924 R. Krumscheid et al.(Eur. J. Biochem. 271) Ó FEBS 2004
codon: 5¢-CTCCTGTGACCATGATGACCTAAATCCC
AGC-3¢; I390–S601 sense with BglII site: 5¢-GC GT
AGA
TCTATCCATGAAGCTGACACCACAG-3¢;antisense
with EcoRI restriction site: 5¢-AT
GAATTCGCGCTGCG
GCATTTGCCCACAGC-3¢; L354–P588* sense with stop
codon: 5¢-ATTGACCCTCCT
TGAGCTGCTGTCCCCG
ATGCTGTG-3¢; L354–L576* sense with stop codon:
5¢-CCCGTGGATAACCTC
TGATTCGTGGGTCTTAT
CTCC-3¢; L354–L541* sense with stop codon: 5¢-GGCC
TTGGA
TAGCGTGTGCTAGGTTTCTGCCACCTC-3¢;
L354–L527* sense with stop c odon: 5¢-C CCCTGGACGA
AGAGCTG
TAAGACGCCTTTCAGAATGCC-3¢;the
* means that antisense primers of the C-terminally shor-
tened constructs were usually complementary. The primer
sequence of the N398D construct was GCTGACACCA
CAGAG
GATCAGAGTGGGGTCTCC and that of the
D369A construct C CACCATCTGCTCC
GCCAAGACT
GGAACTCTGAC. The underlined nucleotides encode the
mutated amino acid.
Expression and purification of the GST fusion proteins

and 0.8 m
M
Na
4
EDTA, pH 7.4, at 37 °Cwithincreasing
concentrations of pNPP (0–2 m
M
) i n a total volume of
1 mL. The r eaction was stopped after 24–48 h by
addition of 3
M
NaOH. Proteins were sedimented and
the absorption of the supernatant was monitored at
405 nm. Background (hydrolysis of pNPP under the
same conditions in the absence of protein) was
substracted. The velocity of substrate cleavage was
calculated assuming a molar absorption coefficient of
18 500 LÆmol
)1
Æcm
)1
. D ata were fitted to the Michaelis–
Menten equation.
Test for protein tyrosine phosphatase activity
The assay was performed using the EnzCheck Phosphate
Assay Kit (Molecular Probes, Eugene, O R, USA). T he
GST fusion proteins L354–I777 and L354–I604 were
tested for their ability to release phosphate from
O-phospho-
L

and 545 nm, respectively, after 3 min of incubation at 37 °C
in the dark and gentle stirring. TNP-ATP binding to the
protein was detected as an increase of fluorescence intensity
in the p resence of p rotein compared to the fluorescence
intensity in its absence [23,42].
Determination of eosin binding to the fusion proteins
Interaction of eosin Y with GST fusion proteins was studied
in similarity to Skou & Esman [43] in 20 m
M
Tris/HCl,
pH 7.8, at 37 °C. Excitation (480–530 nm with k
Emm
¼
538 nm) and emission (530–580 nm with k
Exc
¼ 518 nm)
spectra in the presence and absence of 1 l
M
or 10 l
M
(L354–P588)–GST fusion protein were recorded on a
Hitachi F-3000 Fluorescence Spectrophotometer with
5 nm bandpass. Steady-state fluorescence studies were
performed with the (L354–I777)–GST fusion protein in
thesamebufferat37°C on a PerkinElmer LS50B
Luminescence Spectrometer exciting the probe at 518 n m
and recording the emitted fluorescence at 530 nm (5 nm
band passes each) and using an emission filter of 530 nm.
The following ligands were tested with respect to their
influence on the steady-state fluorescence of 100 n

subtracted from all further raw data as a background.
Volume corrections were applied and background values of
TNP-ATP fluorescence in the absence of GST f usion
proteins were subtracted. Fluorescence intensity was nor-
malized so that a fluorescence of 1 l
M
TNP-ATP (i.e. in the
absence of protein) w as equal to unity. The dependence of
fluorescence intensity on the concentration of TNP-ATP
was fi tted to Eqn (1) [44], describing a model w ith one
binding site per protein molecule:
F ¼½Pþ
1
2
ðc À 1Þ
½Pþ½EþK
d
À
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð½Pþ½EþK
d
Þ
2
À 4½P½E
q

ð1Þ
or to Eqn (2) [44], describing a model with n identical,
noninteracting, noncooperative binding sites per protein
molecule:

loop [41]. A ll parameters except K
d
were kept
constant during the fitting procedure. Data are presented as
mean ± SEM from the indicated number of independent
measurements.
Modification of the ATP binding site by FITC labeling
for phosphatase studies
Na
+
/K
+
ATPase from pig kidney (6 U; 300 lg; 2 l
M
)was
incubated in a t otal volume of 1 mL for 30 min in the dark
at room temperature in a solution of 40 m
M
Tris/HCl, pH
9, and 10 l
M
FITC. Excess fluorophore was removed by
sedimentation of the protein a t 100 000 g in an ultracentri-
fuge. Modification of 5 l
M
of the (L354–I604)–GST fusion
protein proceeded for 2 h in 40 m
M
Tris/HCl,pH9,inthe
presence of 10 l

Tris/HCl,
pH 9, was labeled for 30 min with 30 l
M
FITC in the dark
at room temperature. Residual free FITC was removed by
dialysis over night against a large excess of 50 m
M
Tris/HCl,
150 m
M
NaCl, 2.5 m
M
CaCl
2
, pH 7.8. The GST tag was split
off by 10 U of human thrombin per mg of GST fusion
protein for 1 h at room temperature with gentle shaking. The
GST protein was removed by incubation of the mixture with
1 mL of pre-equilibrated glutathione Sepharose (see above).
This procedure was repeated once more. Finally, thrombin
and buffer c omponents w ere r emoved by size exclusion
chromatography on a 3 mL Sephadex G-25 column pre-
equilibrated with 20 m
M
Tris/HCl, pH 7.8. The concentra-
tion of the FITC labeled loop was 145 lgÆmL
)1
(2.88 l
M
),

1
subunit
Na
+
/K
+
ATPase (R378–D586) was generated by analogy
to the crystal structure of the corresponding sequence of
porcine a
2
sodium pump [21,47]. The latter, recently
published structure lacks three parts of 6, 10 and 6 amino
acid residues that exist in the mouse brain a
1
subunit.
Hence, the three-dimensional structure of these three
peptides was additionally modeled according to the proce-
dure published previously for the H
4
-H
5
loop of pig kidney
Na
+
/K
+
ATPase [20]. The primary structure of the mouse
brain Na
+
/K

/K
+
ATPase
[20,23]. Several dynamics runs were set up for a canonical
ensemble. One dynamics run was a single i nterval of
120 ps at 3 00 K, and 343 K, respectively, with a femto-
second time step result being r ecorded every 25 fs. The
shake technique was applied to all bonds. Force field
parameters were the same as for the minimization. FITC
was connected to K501 via a covalent bond using the
BUILDER
module included in
INSIGHT II
and its position in
the binding site was optimized.
Results
Effects of truncation of the cytoplasmic H
4
-H
5
loop of the
a subunit of Na
+
/K
+
ATPase on TNP-ATP binding and
p
-nitrophenylphosphatase activity
Molecular modeling of the H
4

I604
L354
L527
ATP
S477
F475
Q482
E446
L354
E505
K501
K480
L773
Fig. 1. C-Terminal truncation of the H
4
-H
5
loop leads to loss of TNP-ATP binding due to unfolding of the N-domain as revealed by molecular
modelling. Molecular modeling was performed as described previously [20]. (A) The complete H
4
-H
5
loop starting at L354 and ending at L773
contains the nucleotide (N)-domain interac ting with TNP-ATP and th e phosphorylation (P)-d omain (D369 shown in blue). The amino acid
sequence ALLK known to interact with a nkyrin [60] is colored in green. (B) The size of the isolated H
4
-H
5
loops shortened by the C-terminal part of
the P-domain to L354–I604 is without effect on TNP-ATP binding. (C) The stability of the N-domain and its ability to bind TNP-ATP with high

Mg
2+
activated phosphatase
K
m
(l
M
) V
max
(nmolÆh
)1
Æmg
)1
)
L354–L777 324 3.55 ± 0.35 0.76 ± 0.05 10.35 ± 0.73
L354–I604 251 3.30 ± 0.06 0.83 ± 0.07 11.28 ± 0.72
I390–S601 212 3.50 ± 0.07 0.57 ± 0.08 5.22 ± 1.08
L354–P588 235 2.95 ± 0.05 0.88 ± 0.03 11.36 ± 0.48
L354–L576 222 3.60 ± 0.07 0.83 ± 0.08 11.42 ± 2.42
L354–L541 188 4.73 ± 0.19 0.88 ± 0.22 9.18 ± 1.00
L354–L527 174 10.05 ± 0.95 1.10 ± 0.18 8.75 ± 1.30
N398D (L354–I604) 251 3.30 ± 0.2 0.93 ± 0.10 5.65 ± 0.53
D369A (L354–I604) 251 3.50 ± 0.50 1.17 ± 0.15 8.96 ± 0.24
Ó FEBS 2004 Mapping of ATP and pNPP binding sites of the Na pump (Eur. J. Biochem. 271) 3927
of the fluorescence enhancement to a single site nucleotide
binding model (Fig. 2), gave no indication for a second site.
TNP-ATP b inding was suppressed b y the presence of ATP
and ADP but not by AMP, as previously reported [34,35]
(data not shown).
A further means to s earch for a second ATP and

for TNP-ATP for
the shortest construct, L354–L527 (200%, K
d
¼ 10 .05 l
M
;
Table 1 ). We should add that shorter constructs could not
be purified because they showed an increasing tendency to
precipitate in solution. This shortest construct L354–L527
still contained all amino acids known to be necessary to bind
ATP (Fig. 1C) [46].
Amino terminal shortening of the loop protein was also
tested. We prepared the construct I390–S601 which lacks
the phosphorylation site at D369. This construct showed a
single TNP-ATP binding site as well. The protein had the
same TNP-ATP binding properties as the longest protein
L354–I777 with K
d
¼ 3.5 0 l
M
(Table 1). Interestingly, its
Mg
2+
activated phosphatase activity was more than 50%
reduced as compared to the corresponding construct L354–
I604 containing the N-terminal part o f the P-domain.
Studies on the structural stability of the N-domain by
FITC anisotropy decay and eosin fluorescence
Overall Na
+

Additionally, to have a closer look to the rigidity of the
FITC-labeled L354–P588 H
4
-H
5
loop, the lifetime of the
excited state and the anisotropy decay of the labeled protein
were determined using a phase domain fluorometer with
modulation f requencies f rom 10 MHz to 200 MHz. We
observed a two-component fluorescence intensity decay
with the major lifetime component s
1
¼ 3.5 ns (f
1
¼ 0.77)
and the minor component s
2
¼ 1.7 ns (f
2
¼ 0.23). The
average lifetime of the excite d state was determined as s ¼
3.1 ns.
The anisotropy decay of FITC-labeled L354–P588 H
4
-H
5
loop was determined in L-format over the range of
modulation frequencies from 10 M Hz to 200 MHz. A
two-component decay with a longer component of q
1

+
/K
+
ATPase [43,53,54]. The largest construct, the
(L354–I777)–GST fusion protein, and the C-terminally
shortened construct L354–P588 were used for a comparat-
ive study. I n contrast to results reported for the membrane-
embedded Na
+
/K
+
ATPase [43,54], we observed neither a
change of the excitation nor of the emission fluorescence
spectra in the presence of any of these H
4
-H
5
loop–GST
Fig. 2. Binding of TNP-ATP to the (L354–I777)–GST fusion protein,
and fit of the data to equations for 1 or 2 TNP-ATP binding sites. The
(L354–I777)–GST fusion protein (1.6 l
M
) was titrated with TNP-ATP
in 50 m
M
Tris/ H Cl , pH 7 .5 at 3 7 °C. Excitation and emission wave-
lengths were 462 nm and 527 nm, respectively. Regression analysis
according to the Eqns (1) and (2) (solid line: Eqn (1); broken line:
Eqn (2); also Table 1) demonstrates that the e quation describing the
properties of a single TNP-ATP binding site gives the best fit.

Characterization of the three-dimensional structure of
the complete and truncated H
4
-H
5
loops by molecular
modeling
The interpretation o f the above r eported data on the
truncation of the H
4
-H
5
loop is considerably facilitated by
the availability of a molecular model [20]. Molecular
modeling of the truncated H
4
-H
5
loop revealed t hat a big
part of the N-domain can be removed without any loss in
TNP-ATP binding properties (compare Table 1 with
Fig. 1A–C). The increase i n K
d
value o f the TNP-A TP
protein complex by the extreme C-terminal shortening to
the L354–L527 construct could be described by dynamic
and energy minimization runs to be the result o f an
increased mobility of parts of the loop structure (Fig. 1C).
The location of FITC within the ATP site obtained either
by in silico docking to the previously described full H

4
-H
5
loop down to amino acid
number 600 revealed no change in phosphatase activity
(Table 1). This may mean that the phosphatase is located on
a part of the N-domain. Because a transfer of the phosphate
group of pNPP to the protein has been reported for the
membrane-embedded enzyme [11,33], a possible participa-
tion of the phosphorylation s ite D369 in the isolated H
4
-H
5
loop’s p-nitrophenylphosphatase activity was tested. Muta-
tion of D369 to A had no significant effect on the V
max
of
substrate hydrolysis (Fig. 4). We therefore conclude that the
isolated loop does not form a phosphointermediate during
catalysis. Interestingly, when the amino terminal part of the
P-domain was deleted, the resulting (I390–S601)–GST
fusion protein showed significantly lower V
max
and K
m
values for pNPP (Table 1, F ig. 4). Because the Mg
2+
dependent phosphatase of the H
4
-H

I604)–GST fusion proteins.
Surprisingly, FITC-labeling o f the loop protein i n the
H
4
-H
5
loop had only a small effect on Mg
2+
dependent
phosphatase activity (Table 2). The observed effect is in
the same range as for the K
+
activated, Mg
2+
depend-
ent phosphatase activity in the membrane-embedded full
enzyme (Table 2). Hence, it seemed possible that a pNPP
binding site might exist separately from the ATP site.
Tran & Farley had reported that N398 is labeled by
radioactive 4-azido-2-nitrophenylphosphate and that this
labeling l eads to an inactivation o f Na
+
/K
+
ATPase
[51]. pNPP docking experiments to the H
4
-H
5
loop

(Fig. 5D), respectively. In the Ca
2+
ATPase derived
model [20], a hydrophobic environment was formed by
M384, L414, W 411, which may stabilize the substrate’s
phenyl ring from both sides (Fig. 4, Table 1). The NO
2
group of pNPP seems to interact w ith N 398, while the
phosphate group may be stabilized by interaction with
S400 and S408 (Fig. 6B). Docking to a model based on
Hakansson’s crystal structure [21,23] missing the N-
terminal part of the P-domain (R378–D586), however,
showed a d ifferent pNPP binding site 16 A
˚
away f ro m
N398 with an estimated 10% lower interaction energy as
compared to the full loop (Fig. 5B,D). I n this case, the
NO
2
group seems to point to the direction of S408, while
the phosphate group may lie between S401 and Q389.
The only hydrophobic interaction of the phenyl ring i n
this model is achieved by H517.
In the Ca
2+
ATPase-derived model (Fig. 5A), the phos-
phate group of pNPP is 3.2 nm from the phosphorylation
Fig. 4. Truncation of the residual P-domain, mutation of the phosphorylation site D369, and N398 as part of the putative phosphatase site. Effects on
V
max

4
-H
5
loop–GST fusion protein. Labeling of pig
kidney Na
+
/K
+
ATPase as well as the purified L354–I604 loop pro-
tein was performed with 10 l
M
FITC at pH 9 fo r 30 min and the
labeled protein was handled as describedinExperimental procedures.
Phosphatase activity of the L354–I604loopproteinwastestedat5 m
M
pNPP. NA, not applicable.
Molar binding
ratio of FITC
Phosphatase
activity
Na
+
/K
+
ATPase
(control)
NA 1.54 lmolÆmg
)1
Æmin
)1

Even millimolar concentrations of ATP did not inhibit
the phosphatase activity (data not shown), indicating that
the pNPP site is unable to bind ATP and that within the
isolated H
4
-H
5
loop protein, binding of ATP to the
nucleotide binding site does not lead to a conformational
change of the N-domain, or to an alteration of the pNPP
site.
Fig. 5. p-Nitrophenylphosphate can be docked to the ATP site as well as to a phosphatase site at the rear surface of the N-domain (overview). (A)
Docking of pNPP to the ATP binding site of the model containing N- and P-domains [20]. The amino acids interacting with the adenine ring of
ATP may also interact weekly with p-nitrophenylphosphate (pNPP). (B) pNPP docked to a surface around N398 at the rear site of the N-dom ain
(pNPP site) of the same model. (C) Docking of pNPP to the ATP binding site of the model containing the N-domain only, modeled according to
Hakansson’s crystal structure [21,23]. Comparison of (A) and (C) show little structural difference in the environment of the docked ligand. The final
docking energies w ithout solvation and desolvation effects were estimated as )6.7 kcalÆmol
)1
for (A) and )6.4 kcalÆmol
)1
for structure (C),
respectively. (D) Docking attempts of pNPP to the rear side of the model of the isolated N-domain [23] revealed that this structure does not contain
a binding site aroun d N398 as in (B) (full loop). The in teraction of the substr ate with the N-terminally shortened loop s tructure is lower as compared
to the E1-Ca
2+
ATPase derived structure of the full H
4
-H
5
protein [20] [) 6.8 kcalÆmol

the N-domain is a lso evident fr om the high steady-state
fluorescence anisotropy o f r ¼ 0.25 for the FITC-labeled
H
4
-H
5
loop and from its long anisotropy decay of q
1
of
11.3 ns favoring the view that the whole loop tumbles in
solution. Additional support for this conclusion comes also
from the fact that the loop does not, in contrast to the
membrane-embedded Na
+
/K
+
ATPase, respond to eosin
Y by fluorescence changes upon addition of ATP, Na
+
or
Mg
2+
[43]. Thus, the N-domain of the isolated H
4
-H
5
loop
is unable to twist down to the P-domain [3]. Such a
conformational change is needed in the membrane embed-
ded Na

formed from ATP in both enzymes [12,31]. Kinetic experi-
Fig. 6. A closer look at the ATP and pNPP sites on the N-domain of the H
4
-H
5
loop of Na
+
/K
+
ATPase. The models u sed for (A–D ) are the same a s
in Fig. 5. (A) and (C) This comparison of pNPP binding to the ATP sites of both models shows that in the N- and P-domains con taining model [20]
(A) a hydrophobic interaction of the pNPP’s phenyl resid ue with F475 and F548 exists. The nitro-group of the substrate interacts with K501 and
Q482 in both models. (B) In the model containing N- and P-domains [20], recognition of p NPP by the pNPP site at t he rear su rface of the N -domain
is due the formation of an hydrogen bond of the NO
2
-group of pNPP with the amino group of N398. The hydrophobicity of the binding pocket is
achieved by W411 and L414. The phosphate group seems to form a hydrogen bridge with the OH-group of S408. (D) In the structure based on the
crystal of the a
2
structure of the N-domain [21,23], the N398 does not interact with the substrate. pNPP may bind 16 A
˚
away from this amino acid
residue and with a lower interaction energy as compared to the model in (B).
3932 R. Krumscheid et al.(Eur. J. Biochem. 271) Ó FEBS 2004
ments in the membrane-embedded enzyme could not decide
whether Na
+
/K
+
ATPase and phosphatase sites overlap

bend together when the H
4
-H
5
loop is disconnected from
the transmembrane helices. Removal of the C-terminal
sequence of the P-domain is without effect on the
p-nitrophenylphosphatase activity (Table 1). Although
pNPP may interact with the ad enosine subsite of the ATP
site (Figs 5A,C and 6A,C), modification of this site by FITC
(Fig. 3) had only a minor effect on the pNPP activity
(Table 2). Additionally, and contrary to the expectations
from the literature reporting an inhibition of K
+
activated
phosphatase by ATP in t he full en zyme [31,60], even
millimolar concentrations of ATP did not affect the
phosphatase (data not shown). All these findings support
the conclusion that in the i solated H
4
-H
5
loop, pNPP is not
hydrolyzed via the ATP site although it may be able to bind
there (Figs 5A and 6A).
Structural models describing the three-dimensional fold-
ing of the a subunit f orming the H
4
-H
5

formation of the structural backbone of the ATP binding
pocket. The adjacent amino acid residue C549, was shown
to be labeled by erythrosin isothiocyanate in the membrane-
embedded Na
+
/K
+
ATPase after blocking of the E
1
ATP
binding site with FITC [16]. Modification of this site with
the sulfhydryl- reactive 8-thiocyano-ATP forming a mixed
disulfide bridge may inactivate Na
+
/K
+
ATPase [63]. In
the H
4
-H
5
loop model of Na
+
/K
+
ATPase [20] obtained
analogously to E1-Ca
2+
ATPase [19], F548 is part of the
ATP binding site and C549 is accessible by induced fit [7];

amino acids interacting with pNPP (see Fig. 6B); italic, amino acids not conserved in the compared sequences; dashes, amino acids not present in
the sequence.
Ó FEBS 2004 Mapping of ATP and pNPP binding sites of the Na pump (Eur. J. Biochem. 271) 3933
to a site i n close vicinity to N398, a residue that has been
affinity-labeled by 4-azido-2-nitrophenylphosphate in the
membrane-embedded N a
+
/K
+
ATPase [51]. The model
showing exclusively the isolated N-domain [21,23] (Figs 5D
and 6D), refused to dock pNPP close to N398. The new
position found for pNPP binding is 16 A
˚
away from N398
and shows a lower interaction energy than to the site in the
other model. Apparently, the model respecting N- and
C-terminal peptide extensions of the N-domain [20] des-
cribes more adequately the experimental findings (Figs 5B
and 6B): mutation of N398 to aspartate and truncation of
the P-domain’s N-terminal part, caused a drop of the
phosphatase activity (Fig. 4). The latter finding points to a
stabilizing effect o f the N-terminal sequence in forming the
pNPP site in the neighborhood of N398. The hydrolysis of
pNPP does not need the phosphorylation site D369 (Fig. 4).
The finding of a phosphatase site around N398 outside of
the ATP binding site, which is involved in the overall
Na
+
/K

phosphatase activity in H
+
/K
+
ATPase [31] (Table 3).
Tyrosine phosphorylation can generally be achieved from
pNPP [49], but phosphotyrosine is not hydrolyzed by the
isolated H
4
-H
5
loop. Because Mg
2+
is needed for pNPP
hydrolyis by the isolated H
4
-H
5
loop, the pNPP site close to
N398 seems to recognize this divalent cation. It is an open
question w hether it participates in the recently reported
MgATP dependent interaction of isolated H
4
-H
5
loops [55].
In conclusion, our data show that the isolated H
4
-H
5

We thank Dr K.O. Ha
˚
kansson for providing the atomic coordinates of
the Na
+
/K
+
ATPase N-domain crystal structure.
This work was supported by the German and C zech Governments
through TSR-088-97, and CZE 00/33, by the Ministry of Education,
Youth and Sports of the Czech Republic (LN 00A141), by grants no.
204/01/0254, 204/01/1001, 206/03/D082, 309/02/1479 and MSMT
113100003 and the research projects no. AVOZ11922 of the Grant
Agency of the Czech Republic and the Deutsche Forschungsge me-
inschaft, Bonn Scho 139/21-2+3. The contributions of Drs T. Obs
ˇ
il
and M. Kubala to this work are gratefully acknowledged.
References
1. K aplan, J.H. (2002) Biochemistry of Na,K-ATPase. Ann. Rev.
Biochem. 71, 511–535.
2. Lingrel, J.B. & Kuntzweiler, T. (1994) Na
+
,K
+
-ATPase. J. Biol.
Chem. 269, 19659–19662.
3. J orgensen, P.L., Hakansson, K.O. & Karlish, S.J.D. (2003)
Structure and mechanism of Na,K-ATPase: Functional Sites and
Their Interactions. Ann. Rev. Physiol. 65, 817–849.

+
-ATPase.
Analysis of structural requirements of fluorescent ATP derivatives
with a Koshland-Nemethy- Filmer model of two interacting ATP
sites. J. Biol. C hem. 274, 1971–1978.
9. Bader, H. & Sen, A.K. (1966) K
+
-Dependent acyl phosphatase as
part of the (Na
+
+K
+
)-dependent ATPase o f cell membranes.
Biochim. Biophys. Acta 118, 116–123.
10. Bond, H., Bader, H. & Post, R.L. (1971) Acetyl phosphate as a
substitute for ATP in (Na
+
+K
+
)-ATPase. Biochim. Biophys.
Acta 241, 57–67.
11. Intur risi, C.E. & Titus, E. (1970) Ouabain-dependent incorpora-
tion of
32
Pfromp-nitroph enylphosphate into a mic rosomal
phosphatase. Mol. Pharmacol. 6, 99–107.
12. Taniguchi, K., Kaya, S., Abe, K. & Mardh, S. (2001) The oligo-
mericnatureofNa
+
/K

+
-transporting ATPase lowers
the conformational flexibility of i ts E
1
ATP site. Eur. J. Biochem.
253, 245–250.
16. Linnertz, H., Kost, H., Obsil, T., Kotyk, A., Amler, E. & Schoner,
W. (1998) Erythrosin 5¢-isothiocyanate labels Cys549 as part of the
low-affinity ATP binding site of Na
+
/K
+
-ATPase. FEBS Lett.
441, 103–105.
17. Linnertz,H.,Urbanova,P.,Obsil,T.,Herman,P.,Amler,E.&
Schoner, W. (1998) Molecular distance measurements reveal an
(ab)
2
dimeric structure of Na
+
/K
+
-ATPase. High affinity ATP
binding site and K
+
-activated phosph atase reside on different
alpha-subunits. J. Biol. Chem. 273, 28813–28821.
18. Froehlic h, J.P., Bamberg, E., Kane, D.J., Mahaney, J.E. & Albers,
R.W. (2000) Contribution of quarternary protein interactions to
the mechanism of energy transduc tion in Na

1182.
22. Hilge, M., Siegal, G., Vuister, G.W., Gu
¨
ntert, P., Gloor, S.M. &
Abrahams, J.P. (2003) ATP- indu ced conformational changes of
the nucleotide-binding domain of Na,K-ATPAse. Nature Struct.
Biol. 10, 468–474.
23. La
´
nsky´ , Z., Kubala, M., Ettrich, R., Kuty´ ,M.,Pla
´
s
ˇ
ek, J., Tei-
singer, J., Schoner, W. & Amler, E. (2004) The hydrogen bond
between Arg
423
and Glu
472
and another key r esidues A sp
443
,Ser
477
and Pro
489
are responsible for forming and a different positioning
of TNP-ATP and ATP within the nucleotide binding site of Na
+
/
K

27. Ovchinnikov, Y.A., Dzhandzhugazyan, K.N., Lutsenko, S.V.,
Mustayev, A.A. & Modyanov, N.N. (1987) Affinity modification
of E
1
-form of Na
+
,K
+
-ATPase revealed Asp-710 in the catalytic
site. FEBS Lett. 217, 111–116.
28. Toy oshima, C. & Nomura, H. (2002) Structural changes in the
calcium pump accompanying the dissociation of calcium. Nature
418, 605–611.
29. X u, C. (2002) A structural model f or the catalytic c ycle of
Ca
2+
-ATPase. J. Mol. Biol. 316, 201–211.
30. D avis, R.L. & Robinson, J.D. (1988) Substrate sites of (Na
+
+
K
+
)-ATPase: pertinence of the adenine and fluorescein binding
sites. Biochim. Biophys. Acta 953, 26–36.
31. Schuurmans Stekhoven, F. & Bonting, S.L. (1981) Transport
Adensone Triphosphatses: Properties and Function. Physiol. Rev.
61, 1–76.
32. Schoner, W., Tho
¨
nges,D.,Hamer,E.,Antolovic,R.,Buxbaum,

cytoplasmic loop of Na,K-A TPase overexpressed
in Escherichia coli retains its ability to bind ATP. FEBS Lett. 426,
297–300.
36. Tran, C.M. & Farley, R.A. (1999) Catalytic actvivity of an isolated
domain of Na,K-ATPase expressed in Escherichia coli. Biophys. J.
77, 258–266.
37. J orgensen, P.L. (1974) Purification and characterization of (Na
+
+K
+
)-ATPase. 3. Purification from the outer medulla of mam-
malian kidney after selective removal of mem brane compone nts
by sodium dodecyl sulphate. Biochim. Biophys. Acta 356, 36–52.
38. Schone r, W., Ilberg, C.V., Kramer, R. & Seubert, W. (1967) On
the m echanism of Na
+
-andK
+
-stimulated hydrolysis of
adenosine triphosphate. 1. Purification and properties of a Na
+
-
and K
+
-activated ATPase from ox brain. Eur. J. Biochem. 1,334–
343.
39. Lowr y, O.H., Rosebrough, N.J., F arr, A.L. & Randall, R.J.
(1951) Protein measurement with the Folin ph enol reagent. J. Biol.
Chem. 293, 235–275.
40. Bradfor d, M.M. (1976) A rapid and sensitive method for the

5
loop of Na/K-ATPase. Eur. Biophys. J. 32, 363–369.
45. Sali, A. & Blundell, T.L. (1993) Comparative protein modeling by
satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815.
46. Laskowski, R.A., McArthur, M.W., Moss, D.S. & Thornton,
J.M. (1993) PROCHECK: a program to check the stereochemical
quality of protein structures. J. Appl. Crystallogr. 26, 283–291.
47. Haue, L., Pedersen, P.A., Jorgensen, P.L. & Hakansson, K.O.
(2003) Cloning, expression, purification and crystallization of the
N-domain from the a
2
subunit of the membrane-spanning Na,K-
ATPase protein. Acta Cryst. D59, 1259–1261.
48. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin , F. &
Higgins, D.G. (1997) The Clustal–Windows interface: Flexible
strategies for multiple sequence alignment aid ed by quality ana-
lysis tools. Nucleic Acids Res. 25, 4876–4882.
49. Wang,S.,Taberno,L.,Zhang,M.,Harms,E.,VanEtten,R.L.&
Staufacher, C .V. (2000) Crystal structure of a l ow-molecular
weight protein tyrosine phosphatase from Saccharomyce s cervisiae
and its complex with the substrate p-nitrophenyl phosphate. Bio-
chemistry 39, 1903–1914.
50. Morris , G.M., Goodsell, D., Huey, R. & Olson, A.J. (1996) Dis-
tributed automated docking of flexible ligands to proteins: Parallel
applications of AutoDock 2.4. J. Comp. Aided Mol. Design 10,
293–304.
51. Tran, C.M. & Farley, R.A. (1996) Photoaffinity labeling o f the
active site of the Na
+
/K

-ATPase.
Biochem. Biophys. Res. Commun. 306, 416–420.
57. H uang, W. & Askari, A. (1975) (Na
+
+K
+
)-activated adeno-
sinetriphosphatase: Fluorimetric Determination of th e associated
K
+
-dependent 3-O-methylfluorescein phosphatase and its use for
the assay of enzyme samples with low activities. Anal. Biochem. 66,
265–271.
58. Cav ieres, J.D., Lilley, K.S. & Ward, D.G. (2000) Dissecting and
mapping two nucleotide binding sites in Na,K-ATPase. In Na/K-
ATPase and Related ATPases (Taniguchi, K. & Kaya, S., eds), pp.
115–122. Elsevier Science, Amsterdam.
59. Linnertz, H., Miksik, I., Kvasnicka,P.,Bertoli,E.,Mazzanti,L.,
Schoner, W. & Amler, E. (1998) Binding of pyrene isothiocyanate
to the E
1
ATPsitemakestheH
4
-H
5
cytoplasmic loop of Na
+
/K
+
-