(Received for publication, April 22, 1996, and in revised form, September 9, 1996)
From the Laboratoire de Chimie-Physique des Macromolécules aux Interfaces, CP 206/2, Université Libre de Bruxelles, Campus Plaine, B-1050 Brussels, Belgium
Membrane topology of the
H+,K+-ATPase has been studied after proteolytic
degradation of the protein by proteinase K. Proteinase K had access to
either the cytoplasmic part of the protein or to both sides of the
membrane. Fourier transform infrared attenuated total reflection
spectroscopy indicated that membrane-associated domain of the protein
represented about 55% of the native protein, meanwhile the cytoplasmic
part represented only 27% of the protein. The secondary structure of
the ATPase and of its membrane-associated domains was investigated by
infrared spectroscopy. The secondary structure of the
membrane-associated structures and of the entire protein was quite
similar (-helices, 35%;
-sheets, 35%; turns, 20%; random,
15%). These data were in agreement with 10
-helical transmembrane
segments but suggested a participation of
-sheet structures in the
membrane-associated part of the protein. Polarized infrared
spectroscopy indicated that the
-helices were oriented nearly
perpendicular to the membrane plane. No preferential orientation could
be attributed to the
-sheets. Monitoring the amide
hydrogen/deuterium exchange kinetics demonstrated that the membrane
associated part of the ATPase molecule is characterized by a relatively
high accessibility to the solvent, quite different from that observed
for bacteriorhodopsin membrane segments.
The gastric H+,K+-ATPase is an ,
heterodimer (1, 2) that belongs to the P-type ATPase family. These
ATPases, which include the Na+,K+-ATPase, the
Ca2+-ATPases and the yeast and Neurospora plasma
membrane H+-ATPases, are thought to share a common
mechanism and a common structure. Both N- and C-terminals have been
found cytoplasmic (3), thus the catalytic subunit must have an even
number of transmembrane segments. According to the hydropathy plots of
their catalytic subunits, the P-type ATPases are thought to adopt a similar topological insertion in the membrane. The first four transmembrane segments located in the N-terminal third of the protein
are clearly defined on the hydropathy plots. Their existence was
demonstrated by proteolysis and in vitro expression
experiments (4, 5). The other predicted transmembrane segments, located in the C-terminal third of the protein, are more ambiguous. Predictions and experiments suggested from four to six transmembrane segments in this region of the protein (4, 6, 7).
It must be pointed out that until now all the transmembrane segments of
the ATPase are predicted to be -helices, according to the hydropathy
plots designed for localizing potential transmembrane
- helices on
the sequence of membrane proteins. This idea was further strengthened
by the recent knowledge of the three-dimensional structure of
bacteriorhodopsin (8), of a photosynthetic complex (9), and of a
bacterial cytochrome c oxidase (10). Bacterial porins (11)
whose membrane segments adopt a
-sheet structure were considered as
an exception. However, recently a 9-Å resolution structure of the
nicotinic acetylcholine receptor obtained by high resolution electron
microscopy (12) indicated that three out of the four predicted
transmembrane
-helices do not adopt the expected helical
conformation. The secondary structure of the membrane part of the
nicotinic receptor was analyzed by Fourier transform infrared
spectroscopy after removal of the receptor's extramembrane moieties by
enzymatic proteolysis (13).
-Sheet structures were clearly present
in the membrane part of the protein.
In the present paper, we investigated by a similar approach the
structure of the membrane part of the
H+,K+-ATPase inserted in tubulovesicles
isolated from stomach parietal cells. After proteolysis of either the
cytoplasmic part or of both the vesicle interior and the cytoplasmic
parts of the protein, the analysis of the secondary structure of the
remaining membrane-associated part of the ATPase by Fourier transform
infrared spectroscopy showed an amount of -helical structure
compatible with 10
-helical transmembrane segments. These segments
were oriented perpendicularly with respect to the membrane plane
according to their infrared linear dichroism. Our results also
suggested the presence of a large amount of
-sheet structures in the
membrane-associated segments of the tubulovesicle proteins.
Proteinase K, ATP, and nigericin were purchased from Sigma. Deuterium oxide was from Janssen Chimica (Geel, Belgium). All other reagents were of the highest purity grade commercially available.
Preparation of the TubulovesiclesTubulovesicles were
isolated from hog gastric fundus by differential centrifugation as
described (14). They were further purified from the microsomal
fractions by centrifugation on a sucrose discontinuous density gradient
at 100,000 × g overnight. The material collected at
the 8-30% sucrose interface is referred to as the tubulovesicles. A
single band at 95 kDa corresponding to the -subunit of the
H+,K+-ATPase was visible after
SDS-PAGE1 analysis of a 6-µg sample on a
mini-protean II (Bio-Rad).
Gel electrophoresis was carried out on a 7.5% gel according to Laemmli (15) or on a 16% Tricine gel according to Schagger and von Jagow (16). Typically, 6-9 µg of protein were applied per lane on a 7-cm gel in the case of SDS-PAGE and 40-50 µg on a 10-cm 16% gel (+2 cm of spacer gel) in the case of SDS-Tricine gel. Bands were revealed by Coomassie Blue coloration. The samples were not boiled before running. In the case of SDS-Tricine gel, all the steps (polymerization and running) were performed at 4-5 °C.
ATPase ActivityATPase activity of the tubulovesicles was determined in a medium containing 40 mM HEPES, 2 mM ATP, 2 mM MgCl2 at pH 7.2 in the presence or in the absence of 20 mM KCl. The vesicles were incubated at 37 °C for 15 min in this medium. The reaction was stopped by addition of 7% SDS (final concentration 1.75%). Inorganic phosphate formation was assayed according to Stanton (17) excepted that the coloration was revealed with ascorbate. Sucrose was 8% (w/v) in order to keep iso-osmotic conditions and nigericin was 14 µM.
Protein AssayProteins were assayed using the BCA kit from Pierce or by a modified Lowry method (18). Bovine serum albumin was used as standard in both cases.
Lipid AssayLipids were assayed with the enzymatic test for the choline reaction group from Boehringer Mannheim, assuming that lipids carrying a choline function represent 50% of the total of the lipids present in the tubulovesicles (19).
Proteolytic DigestionWe used two conditions for the proteolytic degradation of the water-exposed domains of the H+,K+-ATPase.
In the first case, vesicles containing the H+,K+-ATPase were diluted to a concentration of 2.5 mg/ml in the conservation buffer (8% sucrose, 50 mM HEPES, pH 7.2 with Tris) in order to keep iso-osmotic conditions. A freshly prepared solution of proteinase K (Sigma) in the same conservation buffer was added at a protease/protein ratio of 1/4 (w/w).
In the second case, tubulovesicles were first diluted and washed by centrifugation in a 1 mM HEPES, pH 7.2 (with Tris) buffer and then resuspended in the same buffer at a final concentration of 2.5 mg/ml. A freshly prepared solution of proteinase K in the 1 mM HEPES buffer was added at a protease/protein ratio of 1/2 (w/w). In this case, the same amount of protease was added after 3 and 7 h.
In both cases, samples were incubated at 37 °C and aliquots were
removed after 0 min (in this case the inhibitor phenylmethylsulfonyl fluoride is added just before the protease), 5 min, 30 min, 1 h,
4 h, 7 h, 24 h, 48 h, and 72 h. A fresh
solution of phenylmethylsulfonyl fluoride (100 mM in EtOH)
was added to each aliquot at a final concentration of 6 mM
to stop the proteolysis. NaCl was added to the digested vesicles to 0.5 M final concentration. The sample was then shaken on ice
for a few minutes, and the samples were centrifuged at 45,000 × g for 45 min. The resulting pellets were twice resuspended
in 1 mM HEPES, pH 7.2, and recentrifuged at 45,000 × g for 45 min. The final pellets were resuspended in 100 µl
of 1 mM HEPES, pH 7.2, and stored at 20 °C.
The -subunit of the
H+,K+-ATPase was deglycosylated using the
N-glycosidase F (Boehringer Mannheim) in denaturing
conditions. To 100 µl of tubulovesicles (1 mg/ml) was added 3 µl of
10% (w/v) SDS and 7 µl of 10%
-mercaptoethanol (v/v). The sample
was boiled for 1 min. Then, 21 µl of
n-octyl-
-D-glucopyranoside 10% (w/v), 154 µl of phosphate-buffered saline-EDTA buffer (137 mM NaCl, 2.7 mM KCl, 1.47 mM
KH2PO4, 8.1 mM
Na2HPO4, and 20 mM EDTA) and 15 µl of N-glycosidase F (250 units/ml) were added, before
incubation at 37 °C during 24 h.
The protein was isolated on a 14% SDS-polyacrylamide gel electrophoresis as described by Laemmli (15). The protein band was electroblotted onto polyvinylidene difluoride membranes by the method of Matsudaira (20), and the blot was stained with Coomassie Blue. Alkylation of cysteine in proteins with acrylamide was carried out in the reaction cartridge of the sequencer as described by Brune (21). Amino acid microsequence analysis was performed by automated Edman degradation of 1-10 pmol of peptides on a Beckman LF3400 protein sequencer (Beckman Instruments, Inc.). The samples was sequenced using standard Beckman sequencer procedure 4. All sequencing reagents were from Beckman Instruments, Inc.
FTIR SpectroscopyAttenuated total reflection infrared
(ATR-FTIR) spectra were recorded on a Perkin-Elmer infrared
spectrophotometer 1720X equipped with a liquid nitrogen-cooled
mercury-cadmium telluride detector. The internal reflection element was
a germanium ATR plate (50 × 20 × 2 mm) with an aperture
angle of 45° yielding 25 internal reflections. 128 scans were
averaged for each spectrum. Spectra were recorded at a nominal
resolution of 4 cm1. The spectrophotometer was
continuously purged with air dried on a silica gel column (5 × 130 cm) at a flow rate of 7 liters/min. Every 4 scans, reference
spectra of a clean germanium plate were automatically recorded by a
sample shuttle accessory. At the end of the scans, the spectra were
transferred from the memory of the spectrophotometer to a computer for
subsequent treatments.
Thin films were obtained as described by Fringeli and Günthard (22) by slowly evaporating 40-100 µl of the cleaved and noncleaved tubulovesicles under a N2 stream on one side of a germanium plate. This results in the formation of oriented multilayers at the surface of the plate. The ATR plate was then sealed in a universal sample holder and rehydrated by flushing D2O-saturated N2 at room temperature.
Lipid/ATPase RatioThe lipid/protein (w/w) ratio is related
to the (C=O)lipid/
(N-H)protein
absorption bands ratio in the FTIR spectra. The ATPase/lipid ratio
after hydrolysis was estimated from the ratio of the protein absorption
peak (amide I (1600-1700 cm
1)) with respect to the lipid
(C=O) absorption peak (1700-1800 cm
1) as established by
Goormaghtigh et al. (23).
Fourier
self-deconvolution was applied to narrow the different components of
the amide I region, which is the most sensitive to the secondary
structure of proteins. The self-deconvolution was carried out using a
Lorentzian line shape for the deconvolution and a Gaussian line shape
for the apodization. To quantify the area of the different components
of amide I revealed by the self-deconvolution, a least squares
iterative curve fitting was performed to fit Lorentzian line shapes to
the spectrum between 1700 and 1600 cm1. Prior to curve
fitting, a straight base line passing through the ordinates at 1700 cm
1 and 1600 cm
1 was subtracted. The
spectrum arising from the lipid part of the system was found to be
completely flat between 1700 cm
1 and 1600 cm
1 and was therefore not subtracted. To avoid
introducing artifacts due to the self-deconvolution procedure, the
fitting was performed on the nondeconvoluted spectrum. The proportion
of a particular structure was computed to be the sum of the area of all
the fitted Lorentzian bands having their maximum in the frequency
region where that structure occurs divided by the total area of amide I. In these conditions, the difference with the x-ray determination amounted to 8% (standard deviation) for a large set of proteins tested
(23).
The determination
of molecular orientations by infrared ATR spectroscopy was performed as
described by Goormaghtigh and Ruysschaert (24). When orientation was to
be evaluated, additional spectra were recorded with perpendicular
(0°) and parallel (90°) polarized incident light. The dichroism
spectrum was computed by subtracting the 0° polarized spectrum from
the 90° polarized spectrum taking into account the difference in the
relative power of the evanescent field for each polarization as
described before (25). A larger absorbance at 90° (upward deviation
on the dichroism spectrum) indicates a dipole oriented close to a
normal to the ATR plate. Conversely, a larger absorbance at 0°
(downward deviation on the dichroism spectrum) indicates a dipole
oriented closer to the plane of the ATR plate. The dichroic ratio R is
defined as the ratio of the amide I area recorded for the parallel
polarization (A90°) and perpendicular
polarization (A0°). If only the -helix
component is ordered while the other structures have no particular
orientation in the membrane,
![]() |
(Eq. 1) |
![]() |
(Eq. 2) |
![]() |
(Eq. 3) |
![]() |
(Eq. 4) |
![]() |
(Eq. 5) |
Films containing 70-100 µg
of protein were prepared on a germanium plate as described above.
Nitrogen gas was saturated with D2O (by bubbling in a
series of five vials containing D2O) at a flow rate of 51 ml/min (controlled by a Brooks flow meter). Bubbling was started at
least 1 h before starting the experiment. At zero time, the tubing
was connected to the cavity of the sealed chamber surrounding the film.
For each kinetic time point, 12 spectra are recorded and averaged at a
resolution of 4 cm1. Two kinetics were recorded
simultaneously as described before (25, 27). The areas of amides I, II,
and II
were obtained by integration between 1702 and 1596, 1596 and
1502, and 1492 and 1412 cm
1, respectively. For each
spectrum, the area of amide II and amide II
was divided by the area of
amide I. This permitted us to take into account small but significant
variations of the overall spectral intensity due in part to the
presence of D2O, which, by virtue of its presence, induces
the swelling of the sample layer and therefore increases the average
distance between the protein sample and the germanium crystal surface.
Since the ATR spectrum intensity depends on this distance (28), this
resulted in a loss of a few percent of the band intensity for all
measured bands. Undeuterated spectra were recorded before the kinetic
experiment as explained above, and 100% deuterated samples values were
extrapolated by assuming a value of zero for the amide II surface. The
area of amide II and II
(reported to the area of amide I) was finally expressed between 0 and 100% for each kinetic time point.
For the analysis of the deconvoluted spectra during the time course of
exchange, each spectrum was self-deconvoluted with a
28-cm1 FWHH Lorentzian lineshape and apodized with a
Gaussian lineshape with a FWHH of 14 cm
1, yielding to a
resolution enhancement factor of 2 (29).
In a
first approach, the integrity and the orientation of the vesicles was
assessed from ATPase activity measurements. ATPase activity (mean
values ± S.D. from four experiments) expressed in
µmol·h1·mg
1 protein at 37 °C is
36 ± 7 in isotonic conditions; 111 ± 30 after addition of
14 µM nigericin and 122 ± 20 in hypotonic
conditions (see "Experimental Procedures"). With these figures of
activity, it was possible to evaluate the proportion of sealed vesicles oriented with the cytoplasmic side outside (30). The increased activity
upon nigericin addition indicated that at least 68% of the ATPase
molecules were pumping into the interior of a sealed vesicle. This is a
minimum estimate arrived at by assuming that in the absence of
nigericin, 32% (= 36/111 × 100) of the ATPase molecules was fully
active, and 68% was completely inhibited by the electrochemical
gradient. In fact, it is more likely that more ATPase molecules pumped
protons into sealed vesicles but that the ion leak through the membrane
was largely responsible for the residual ATPase activity. The fact that
rupture of the vesicles by hypotonic conditions did not reveal
significantly more activity than collapsing the electrochemical
gradient by nigericin indicated that the ATP binding site of the ATPase
molecules was homogeneously oriented toward the outside of the
vesicles. The 10% activity increase recorded in hypotonic conditions
indicated that no more than 10% of the vesicles was oriented with the
ATP binding site located inside. That the hypotonic conditions opened the vesicles is demonstrated below.
In a second approach, we took advantage of the asymmetric localization
of the -subunit of the ATPase. While most of the
-subunit mass is
located on the cytoplasmic side (7, 31), the largest part of the
-subunit is located inside (~35 amino acids outside for ~228
amino acids inside) and spans the lipid bilayer only once.
Accessibility of the
-subunit to proteolysis was therefore a good
test for defining the orientation of the ATPase molecule. Visualization
of native
-subunit on SDS-PAGE was difficult, since it migrates as a
broad band between 60 and 90 kDa. When deglycosylated (see
"Experimental Procedures"), the
-subunit appeared as a well defined band around 31 kDa (32, 33). Proteolytic degradation of the
-subunit was therefore followed by SDS-PAGE after deglycosylation of
the
-subunit by N-glycosidase F in denaturing conditions
(see "Experimental Procedures"). After proteolysis in iso-osmotic
conditions, the
-subunit appeared at 30.5 kDa on the gel just
below the intact
-subunit (31 kDa), with the same intensity (data
not shown). Sequencing of the band at 30.5 kDa showed that the
N-terminal part of the proteolyzed
-subunit in iso-osmotic
conditions begins at the 31st amino acid of the
-subunit
(G31RTLSRWVWISLYYV). This result indicated the
accessibility of the sole cytoplasmic side of the ATPase to proteases
even though the cytoplasmic part of the subunit represents only about
35 amino acids. After proteolysis in hypotonic conditions, the band of the
-subunit completely disappeared (data not shown), indicating accessibility of both sides (cytoplasmic and vesicle interior) of the
ATPase to proteases.
In order to isolate the membrane-embedded domain of the gastric H+,K+-ATPase, gastric tubulovesicles were treated with proteinase K. In a first set of experiments, we treated sealed tubulovesicles with proteinase K in iso-osmotic conditions (see "Experimental Procedures"). With this procedure, we potentially removed the cytoplasmic domain of the H+,K+-ATPase, leaving intact the membranous and vesicle interior parts of the protein. In a second set of experiments, the vesicles were opened by an osmotic shock before proteinase K was added. In this case, the protease had access to both sides of the membrane and was able to hydrolyze the cytoplasmic and extracytoplasmic parts of the ATPase molecules.
Kinetics of the Proteolytic DegradationThe percentage of
proteolysis as a function of digestion time was monitored by following
the evolution of the vesicle lipid:protein ratio after removal of
digestion products. Lipid:protein ratios were calculated from the FTIR
spectra of the re-isolated tubulovesicles (Fig.
1A) (see "Experimental Procedures"). The
relative amount of protein remaining associated to the vesicles was
reported as a function of the incubation time for the two proteolysis
conditions (Fig. 2). The proteolysis kinetic was
analyzed as the sum of two components (see Fig. 2). The fast component
could represent the easily accessible part of the protein, and the slow
one could represent less accessible segments (presumably because of
their association with the membrane). For the sealed vesicles, the fast component represented 27% of the protein. This component rose up to
45% for the opened vesicles. The additional 18% could account for the
extracytoplasmic parts of the ATPase made accessible to the protease
after opening the vesicles. The kinetics constants (k1 and k2) (see Fig. 2)
are similar for both conditions of proteolysis. Since
k1 = 12 h1, the rapid components
(45% of the protein in hypotonic digestion conditions) were virtually
totally hydrolized (99.999%) after 4 h. On the other hand,
k2 = 6.10
3 h
1,
indicating that less than 3% of the slow component was hydrolyzed after the same period of time.
Membrane-associated Peptides
On SDS-PAGE (Fig.
3), after proteolysis in hypotonic conditions, we
observed the disappearance of the 95-kDa -subunit band of the
H+,K+-ATPase. This disappearance was
accompanied with the appearance of several bands with molecular weight
below 11 kDa (see Fig. 3, lane 1) which were not present in
the intact tubulovesicles (Fig. 3, lane 2). Four bands were
resolved by SDS-PAGE and migrated at 10,800, 8900, 7800, and 6600 Da,
respectively. This peptide pattern was already stabilized after 4 h of incubation for both conditions of proteolysis and was not modified
for longer digestion times (up to 48 h). The peptide pattern was
identical for both proteolysis conditions.
Structure of the ATPases in the Tubulovesicles: Secondary Structure of the ATPases and of Its Membrane-associated Fraction
Attenuated
total reflection IR spectroscopy (ATR-IR) was used to determine the
structure of the ATPase and of its isolated membrane-associated part.
Fig. 1A reports the shape of amide I in the course of the
proteolysis in isotonic conditions (from 5 min to 24 h). Amide I
bands were relatively broad with a maximum at 1655 cm1
and a shoulder near 1635 cm
1 for all digestion times.
Fig. 1B reports the spectra obtained by subtraction of the
spectrum of the native ATPase (i.e. the 0-min spectrum in
A) from every spectrum reported in A. The spectra in B are therefore the spectra of the cleaved parts of the
ATPase molecules and are representative of the structure of the removed peptides as they were in the native protein. The maximum near 1655 cm
1 and the shoulder near 1640 cm
1 suggest
the presence of both
-helix and
structure in the peptides accessible to the protease. Assigning the different amide I components revealed by Fourier self-deconvolution to different secondary structures is best substantiated by monitoring their shift upon hydrogen/deuterium exchange. Figs. 6 and 7 report, respectively, the
spectra of the intact ATPase (Fig. 6) and of the membrane-associated ATPase fragments isolated after incubation in the presence of proteinase K for 48 h in hypotonic conditions (Fig. 7) for
deuteration times from 0 to 512 min. In B of Figs. 6 and 7,
spectra have been self-deconvoluted with a 28-cm
1 FWHH
(full width at half-height) Lorentzian line shape and apodized with a
14-cm
1 FWHH, Gaussian line shape as described elsewhere
(24). The lipid
(C=O) band was located at 1744 cm
1
before and after protease treatment (Figs. 6 and 7). For the native
ATPase (Fig. 6B), the amide I region presented two main features located at 1654 and 1636 cm
1. We assign them,
respectively, to the
-helix and
-sheet structure. The exchange of
the amide hydrogen by deuterium (52% of exchange after 60 min, see
below) did not alter the frequency of the two main features resolved
here (the frequency shift upon deuteration reaches large values only
when deuteration is close to completion). This confirmed the assignment
of the 1654 and 1636 cm
1 bands to
-helix and
-sheet
stable secondary structures. This observation is important, since
several structures (loops, various turns, extended chains; see
Goormaghtigh et al. (26) for a more exhaustive compilation)
may absorb in the same frequency range but are expected to exchange
much faster because of weaker internal hydrogen bonds. The random
coil-associated band was probably too broad to be resolved in the
self-deconvolution conditions used here (25, 34) and was therefore
hidden in the base line underlying the resolved features described
above. Two similar peaks, located at 1653 and 1633 cm
1,
were found in the membrane part of the ATPase (Fig. 7B) and were assigned, respectively, to
-helix and
-sheet structures. The
additional small 1695-cm
1 component resolved here was
tentatively assigned to the high frequency component of the
-sheet
structure. These results suggest a significant amount of
-sheet
structure in the membrane-associated domain of the ATPase.
A quantitative analysis of the amide I band of deuterated spectra by
Fourier self-deconvolution and least squares curve fitting (23) allowed
us to quantify this secondary structures (Fig. 4).
Results are reported in Table I.
|
Spectra of native and digested tubulovesicles were
recorded using polarized light at 90 and 0° (see Ref. 24 for the
geometry of the experiment). The dichroic spectra were obtained by
subtracting spectra recorded with 0° polarized light from spectra
recorded with 90° polarized light, for native tubulovesicles (Fig.
5) and for the vesicles digested in the two conditions
used (data not shown). In all cases, a positive deviation of the
dichroic spectra was present in the 1656-1660-cm1
spectral domain assigned to the
-helical structure. As the amide I
(C=O) dipole lies approximately parallel to the helix axis, this
deviation suggested a perpendicular orientation of the helix axes with
respect to the membrane plane for the majority of the
-helices. In
the
-strand region of the spectra (1637-1613 cm
1), no
deviation could be observed. Integration of the amide I area between
1706 and 1597 cm
1 yielded a dichroic ratio of 1.53 for
the native ATPase, 1.63 for the ATPase digested in isotonic conditions,
and 1.66 for the ATPase digested in hypotonic conditions with a value
of Riso (see "Experimental Procedures") of
1.45. Taking into account the
-helical content reported in Table I,
the corresponding R
values are, respectively,
1.59, 1.77, and 1.82. Since it is not legitimate to assume an uniaxial
distribution of the helix axis orientations in the entire protein, it
was not possible to translate the value of R
= 1.59 into a mean angular orientation. On the other hand, the values
of 1.77 and 1.82 correspond to a near-perpendicular orientation of the
helix axes with respect to the membrane plane, taking into account a
27° angle between the transition dipole and the helix axis (see Refs.
24 and 27 for a thorough discussion on the method and on the 27°
value). The latter values of R
implied little
orientational disorder in the helical component.
Amide Hydrogen/Deuterium Exchange Kinetics
At constant pH and
temperature, the rate of amide hydrogen exchange by deuterium is
related to the stability of the secondary structures and to the solvent
accessibility to the NH amide groups. Amide hydrogen exchange was
followed by monitoring the amide II absorption peak (maximum at 1544 cm1) decrease or amide II
absorption peak (maximum at
1450 cm
1) increase as a function of the time of exposure
to D2O-saturated N2 flow (from 15 s to
8 h; for details, see "Experimental Procedures"). This is
illustrated in Figs. 6 and 7, where a series of spectra recorded as a
function of the deuteration time for the ATPase (Fig. 6)
and for the ATPase digested in hypotonic conditions (Fig. 7) are reported. The evolution of the area of the amide
II for each sample was then computed between 0 and 100% of deuteration as explained under "Experimental Procedures." The
hydrogen/deuterium exchange was faster for the digested ATPase than for
the native one (Fig. 8). After 1-h deuteration, 67% of
the peptide N-H was exchanged in the protease-treated in hypotonic
conditions sample and 61% in isotonic conditions of digestion, whereas
for the intact ATPase the exchange was only 52% after 1 h.
The aim of this work was to test experimentally the models proposed to describe the mode of insertion of the gastric H+,K+-ATPase into the membrane. The approach chosen was to use proteases to attempt to cleave the segments of ATPase protruding from the membrane while preserving the integrity of the membrane domain and to investigate the structure of this membrane domain by ATR-FTIR spectroscopy.
Critical assumptions in the present work are that 1) proteolytic cleavage is efficient outside the membrane but does not reach the membrane part of the protein, and 2) the structure of the membrane part of the protein is not affected by removal of the extramembrane loops. These two assumptions have been discussed previously in detail and experimentally verified in similar experimental conditions on a membrane protein of know structure: bacteriorhodopsin (25).
The choice of proteinase K has been suggested by the fact that it is
one of the most active unspecific protease and because it can be
efficiently separated from lipid vesicles after proteolysis as
described before (25). The fact that proteinase K completely cleaves
the parts of the protein protruding from the vesicle is crucial for the
conclusions of this paper. Proteinase K has been shown to be one of the
most effective proteases to remove the extramembrane part of proteins
(e.g. Refs. 13, 25, and 35, 36, 37, 38, 39). Furthermore, as some
specificity of proteinase K for selected amino acids or secondary
structures could eventually hinder the proteolytic action, the
association of different proteases is likely to overcome this barrier.
Incubation of the ATPase with trypsin or subtilisin before the
incubation with proteinase K did not improve the extent of digestion
(data not shown). Therefore, we consider the possibility of
extramembrane domain(s) resistant to the protease as unlikely. The
second critical assumption is that the isolated transmembrane segments
keep their native structure. From thermodynamic consideration, it seems
likely that the structure of the membrane embedded segments is
essentially dependent on the membrane environment and that changes in
the extramembrane region would have little influence on it. The only
well documented membrane protein behavior upon proteolysis is that of
bacteriorhodopsin, which confirms this hypothesis. Experiments carried
out previously under similar experimental conditions indicated that the
strong amide I polarization which demonstrates a transmembrane
orientation of the helices is identical in both the intact
bacteriorhodopsin molecule and in its membrane domain isolated after
proteinase K treatment (25). These data bring additional indication of the integrity of isolated membrane domain structures. These conclusions are in line with cleavage experiments performed by Draheim et al. (40) which showed that the secondary structure and the
orientation of the isolated membrane domain of bacteriorhodopsin do not
change after proteolysis. Moreover, the integrity and stability of
membrane fragments of bacteriorhodopsin after proteolysis or after
separated reconstitution is well documented (41, 42, 43). In the present work, the similar polarization of the
-helical component of amide I
in the intact protein and in its membrane part indicated the stability
of the orientation of this membrane protein structure upon removal of
the connecting segments. A similar stability of the isolated
transmembrane segments was also demonstrated on the Neurospora
crassa plasma membrane H+-ATPase (25).
The amount of protein removed by the proteolytic treatment when either
only the outside of the vesicle or both sides of the vesicle were
accessible to the proteases can be discussed with respect to the
current model of the ATPase. The comparison of the model with the
experimental data supposes that the orientation of the vesicles is
characterized. ATPase activity measurements indicated that more than
68% of the ATPase molecules was oriented with their cytoplasmic side
facing the outside of sealed vesicles. Proteolytic cleavage confirmed
that more than 95% of the ATPase molecules had this orientation. In
isotonic conditions, proteinase K removed only the first 30 amino acids
of the -subunit of the ATPase. These 30 amino acids correspond to
the N-terminal and sole cytoplasmic part of the subunit. Whereas after
opening of the vesicles in hypotonic conditions cytoplasmic and
extracytoplasmic sides of the subunit became accessible to the
protease. Our experimental results indicated that the sole
membrane-associated domain of the ATPase represented about 55% of the
protein (i.e. 728 amino acids, see Table I), i.e.
more than expected from the hydropathy plots (about 15-20%) (7, 31).
Only a few works addressed the question of the amount of protein
present in the lipid bilayer. Sequencing of the isolated
membrane-associated peptides could demonstrate the presence of some
peptide sequences in the membrane, although the estimation of their
length rely on the apparent mass deduced from SDS-PAGE. The presence of
other peptides which either do not migrate as well resolved bands or do
not stain is likely (44). An additional hypothesis to explain the
discrepancy between prediction and experimental data is that only
transmembrane
-helix secondary structures have been considered so
far. It has also been shown for the
Na+,K+-ATPase that models based on the
hydrophobicity of the amino acid sequence underestimate there protein
insertion into the bilayers, in comparison with models based on
chemical labeling experiments (45). Accessibility increases upon
opening the vesicles in hypotonic conditions precisely accounted for
the extracytoplasmic side of the
-subunit, suggesting little
degradation of the
-subunit from the extracytoplasmic side of the
membrane. Extracytoplasmic parts of the
-subunit of other P-type
ATPases have been described as small: the Ca2+-ATPase
extracytoplasmic domain is small according to crystal analysis by x-ray
diffraction (46, 47, 48), and the extracytoplasmic loops of the
H+-ATPase from N. crassa (49) and of the
Na+,K+-ATPase
-subunit (50) are highly
resistant to tryptic digestion.
Quantitative evaluation of the structure of the ATPase before
incubation with protease (Table I) is in good agreement with secondary
structure determined for other P-type ATPases (25, 51, 52). Analysis of
the sole membrane-associated part of the protein indicated the presence
of 34% -sheet (248 amino acid residues). Particularly convincing as
to the presence of
-sheet structures in the membrane-associated
proteolytic peptides of the ATPase were the IR spectra reported in Fig.
7B, where one of the two main resolved features in amide I
was located at 1633 cm
1, a frequency assigned to the
-sheet structure and which is never overlapped by helical
contribution (26, 53). Extended polypeptide chains have been found at
similar frequencies in some protein (34), but this hypothesis is
unlikely here since only
-sheet structures can form the internal
hydrogen bonding required in the membrane environment. Furthermore, if
not buried in the membrane environment, such an extended polypeptide
chain should be easily accessible to the proteases and available for
fast H/D exchange, at odds with the experimental results. The absence
of a distinct dichroism from the
-sheet structure implies that these
-sheets are either disordered or are oblique, in agreement with the
lack of distinct infrared linear dichroism observed for porine (54). The presence of
-sheet structures in the membrane-associated domain
of the ATPase implies that the current models for these proteins need
to be revised. In the transmembrane structure of the porin monomer, 16 segments of 11 ± 5 residues length form a
-pleated sheet
structure and are connected by turns (55), but segments as short as 5 residues in length are sufficient to span the hydrophobic core of the
membrane (56). It is premature to describe how
-sheet structures
could be associated with the membrane in the case of the ATPase, but
the presence of a rather large number of short stretches spanning the
membrane would explain the relatively large number of amino acids
involved in turn structures. The 248 amino acid residues involved in
-helices are compatible with the presence of about 10 transmembrane
helices in the ATPase, in good agreement with the predictions based on
the protein sequence. The analysis of the dichroic spectra reveals an
orientation of the
-helices in agreement with a transmembrane
orientation.
Little data exist on the amide hydrogen/deuterium exchange rate in
membrane proteins (for a review, see Ref. 57). For bacteriorhodopsin, the exchange reaches a plateau when 30% of the amide protons has been
exchanged (25, 58, 59). This value corresponds roughly to the fraction
of the protein protruding from both sides of the membrane and suggests
that the transmembrane helices are exchanged at an extremely slow rate.
Similarly, the single transmembrane -helical segment of glycophorin
is virtually not exchangeable (38, 60). On the other hand, the human
erythrocyte glucose transporter exchanges about 80% of its amide
protons by deuterium within the first hour (61), even though structure
predictions suggest the presence of 12
-helices accounting for 50%
of the protein in the membrane. It was concluded that an access to the aqueous medium exists in the transmembrane part of the protein. It must
be mentioned that the 12-
-helix model has been recently challenged
(62). Similarly, the channel-forming peptide of colicin E1, which has
been shown to be predominantly
-helical, also has very fast exchange
(80% exchange in 5 min (63)). The results obtained here for the
H+,K+-ATPase show that for the
membrane-associated part of the ATPase, 67% of exchange is observed
after 1 h (corresponding to about 488 amino acids out of 728),
indicating that the membrane region is much less tightly protected
against hydrogen/deuterium exchange than the membrane region of
bacteriorhodopsin. These data suggest that building a model of the
transmembrane region of the ATPase based on the structure and on the
packing of the transmembrane segments of bacteriorhodopsin is not
legitimate, as already evident from secondary structure analysis.
In a recent paper, Corbalan-Garcia et al. (64) applied a
proteolytic approach to the sarcoplasmic reticulum
Ca2+-ATPase. Digestion by trypsin (30 min) and proteinase K
(30 min) on the cytoplasmic side of the membrane resulted in the
removal of 63% of the protein in agreement with the prediction model
derived value of 66% (65). Fourier self-deconvolution and partial
least squares analysis of amide I demonstrated the absence of -sheet structure in the proteinaceous material associated with the membrane after proteolytic digestion. Experiments carried out in our laboratory (not shown) confirm the Ca2+-ATPase results obtained by
Corbalan-Garcia et al. (64) and suggest that gastric ATPase
and N. crassa H+-ATPase (25) show a larger
membrane-associated protease-resistant fraction than the
Ca2+-ATPase. Whether a serious structural difference exists
between H+,K+- and H+-ATPase on one
hand and Ca2+-ATPase on the other hand is now to be
confirmed by other means. The analysis of the primary structure of the
H+,K+-ATPase by the algorithm of Schirmer and
Cowan (66), designed to reveal porin-like
-sheet transmembrane
segments, indicated the presence of 12 such segments in the catalytic
subunit of the H+,K+-ATPase, but only 7 in the
Ca2+-ATPase (not shown).
In conclusion, our results showed the presence in the
membrane-associated domain of the gastric
H+,K+-ATPase of 248 amino acid residues
involved in -helical structures, in agreement with the presence of
10 transmembrane helices. Linear dichroism data confirmed their
transmembrane orientation. In addition, 248 amino acid residues were
also involved in
-sheet structures associated with the membrane, 131 were involved in turns, and 102 in random coil structure. The
hydrogen/deuterium exchange rate of the membrane part of the protein
suggested an organization of these different secondary structures in
such a way that the solvent had access to a large fraction of the
protein, perhaps through a membrane hole or pore delimited by the
protein.