6604
Biochemistry 1996, 35, 6604-6611
Proton Transport by Halorhodopsin†
György Váró,‡,§ Leonid S. Brown,‡ Richard Needleman,| and Janos K. Lanyi*,‡
Department of Physiology and Biophysics, UniVersity of California, IrVine, California 92717, and Department of Biochemistry,
Wayne State UniVersity School of Medicine, Detroit, Michigan 48201
ReceiVed January 17, 1996; ReVised Manuscript ReceiVed March 25, 1996X
ABSTRACT:
In halorhodopsin from Natronobacterium pharaonis, a light-driven chloride pump, the chloride
binding site also binds azide. When azide is bound at this location the retinal Schiff base transiently
deprotonates after photoexcitation with light >530 nm, like in the light-driven proton pump bacteriorhodopsin. As in the photocycle of bacteriorhodopsin, pyranine detects the release of protons to the bulk.
The subsequent reprotonation of the Schiff base is also dependent on azide, but with different kinetics
that suggest a shuttling of protons from the surface as described earlier for halorhodopsin from
Halobacterium salinarium. This azide-dependent, bacteriorhodopsin-like photocycle results in active
electrogenic proton transport in the cytoplasmic to extracellular direction, detected in cell envelope vesicle
suspensions both with a potential-sensitive electrode and by measuring light-dependent pH change. We
conclude that in halorhodopsin an azide bound to the extracellular side of the Schiff base, and another
azide shuttling between the Schiff base and the cytoplasmic surface, fulfill the functions of Asp-85 and
Asp-96, respectively, in bacteriorhodopsin. Thus, although halorhodopsin is normally a chloride ion pump,
it evidently contains all structural requirements, except an internal proton acceptor and a donor, of a
proton pump. This observation complements our earlier finding that when a chloride binding site was
created in bacteriorhodopsin through replacement of Asp-85 with a threonine, that protein became a chloride
ion pump.
The bacteriorhodopsins and the halorhodopsins are small
integral membrane proteins in various species of halobacteria
that, upon illumination, transport protons out of the cells and
chloride ions into the cells, respectively (Lanyi, 1990, 1993;
Oesterhelt et al., 1992; Rothschild, 1992; Ebrey, 1993). Since
the amino acid sequence (Blanck & Oesterhelt, 1987; Lanyi
et al., 1990; Otomo et al., 1992; Soppa et al., 1993) and the
tertiary structure (Henderson et al., 1990; Havelka et al.,
1995) of these proteins are similar, and the ion transport is
triggered in both cases by photoisomerization of the all-transretinal to 13-cis-retinal, a shared transport mechanism for
protons and chloride has been long regarded as likely
(Oesterhelt & Tittor, 1989; Keszthelyi et al., 1990; Oesterhelt
et al., 1992).
Under special conditions halorhodopsin from Halobacterium salinarium was reported to transport protons. This
occurred through a two-photon reaction in which the unprotonated retinal Schiff base was first accumulated by
sustained illumination with green light and then photoexcited
with blue light (Bamberg et al., 1993). The proton transport
was from the extracellular to the cytoplasmic surface, i.e.,
in the direction opposite from bacteriorhodopsin. It was
proposed that the protons were translocated because (a)
proton conductive pathways exist between the Schiff base
and the two membrane surfaces and (b) the accessibility of
† This work was funded partly by grants from the National Institutes
of Health (GM 29498 to J.K.L.), the Department of Energy (DEFG0386ER13525 to J.K.L. and DEFG02-92ER20089 to R.N.), the National
Science Foundation (MCB-9202209 to R.N.), and the U.S. Army
Research Office (DAAL03-92-G-0406 to R.N.).
* Author to whom correspondence should be addressed.
‡
University of California.
§ Permanent address: Institute of Biophysics, Biological Research
Center of the Hungarian Academy of Sciences, H6701 Szeged,
Hungary.
| Wayne State University.
X
Abstract published in AdVance ACS Abstracts, May 1, 1996.
S0006-2960(96)00115-8 CCC: $12.00
the Schiff base to the surfaces changes upon the sequential
photoisomerizations. In the 13-cis state produced by green
light, the access had to be to the cytoplasmic side where the
Schiff base proton was released, and in the all-trans state
produced upon blue-light illumination of the 13-cis state the
access would be changed to the extracellular side, where a
proton was taken up. The transport is very similar to proton
transport by bacteriorhodopsin in which Asp-85 is replaced
with an asparagine. This recombinant protein does not
transport in the usual bacteriorhodopsin mode, i.e., in a
single-photon reaction of the protonated Schiff base, because
the extracellular proton acceptor is absent (Subramaniam et
al., 1990; Kataoka et al., 1994). However, it transports
protons in a two-photon reaction (green plus blue light) and
in the extracellular to cytoplasmic direction as halorhodopsin
(Tittor et al., 1994).
The analogy of halorhodopsin and bacteriorhodopsin with
respect to chloride transport appears to be somewhat more
direct. According to recent evidence (Sasaki et al., 1995),
chloride transport replaces proton transport when Asp-85 of
bacteriorhodopsin is exchanged for a threonine, the residue
found in the halorhodopsins at this location. Chloridedependent spectral shifts and the photocycles of this mutant
bacteriorhodopsin and halorhodopsin from Natronobacterium
pharaonis (Scharf & Engelhard, 1994; Váró et al., 1995a)
suggested that chloride binds initially near both Thr-85 and
the Schiff base. The similarities of these single-photon
photocycles and the way their steps were affected by
changing the chloride concentration, as well as the direction
of the translocation, suggested that the mechanism of the
chloride transport in D85T recombinant bacteriorhodopsin
was analogous to chloride transport by halorhodopsin. Thus,
in this case it was the single amino acid replacement, rather
than a difference of the nature of the photoreaction, that
changed the ion specificity of the transport.
© 1996 American Chemical Society
Proton Transport by Halorhodopsin
The observation of chloride transport by bacteriorhodopsin
suggested that under suitable conditions halorhodopsin might
transport protons in the manner of wild-type bacteriorhodopsin. Demonstration of this would complete the analogy of
the two proteins. Introducing aspartate residues into halorhodopsin from H. salinarium, at locations where they should
function like Asp-85 and Asp-96 in bacteriorhodopsin, did
not convert it into a proton pump [mentioned in Havelka et
al. (1995)]. Since the pKa of the Schiff base of halorhodopsin
becomes considerably lower after photoisomerization of the
retinal to 13-cis (Lanyi, 1986), as in bacteriorhodopsin
(Govindjee et al., 1994; Brown & Lanyi, 1996), this negative
finding means that in the halorhodopsin mutant the engineered aspartate was not a suitable proton acceptor. The
reason for this is not clear, since in the absence of transducing
protein, sensory rhodopsin I, the third retinal protein of
halobacteria which contains an aspartic acid at the position
of Asp-85 in bacteriorhodopsin, is a proton pump in both a
single-photon cycle (Olson & Spudich, 1993; Bogomolni et
al., 1994) and a two-photon cycle (Haupts et al., 1995). We
have therefore examined another alternative. Azide catalyzes
the light-dependent deprotonation and reprotonation of the
Schiff base in halorhodopsin (Hegemann et al., 1985; Lanyi,
1986; Scharf & Engelhard, 1994), and the reprotonation of
the Schiff base in the photocycle of the D96N mutant of
bacteriorhodopsin where this process is otherwise very slow
(Tittor et al., 1989; Otto et al., 1989; Cao et al., 1991). The
proton transport reactions in bacteriorhodopsin and halorhodopsin with unprotonated Schiff base were enhanced by
azide also (Bamberg et al., 1993; Tittor et al., 1994; Kataoka
et al., 1994; Dickopf et al., 1995). The dependence of
protonation and deprotonation of the Schiff base of halorhodopsin on the pKas of various weak acids suggested that
the mechanism for the catalysis was a shuttling of protons
between the Schiff base and the cytoplasmic surface (Lanyi,
1986). The azide therefore did not traverse the entire
membrane and could not facilitate net proton translocation.
Thus, under some conditions azide can act as both a proton
acceptor and a donor to the Schiff base. Unlike in the case
of halorhodopsin in H. salinarium, a blue shift of the
absorption maximum upon adding chloride to the halorhodopsin to N. pharaonis indicated that this protein contains
a high-affinity binding site for chloride near the Schiff base.
We had argued from indirect evidence that this site is to the
extracellular side of the Schiff base (Váró et al., 1995a,b).
Site-specific mutagenesis of halorhodopsin from H. salinarium identified an arginine near this location as essential
for chloride binding (Rüdiger et al., 1995). Could azide
bound to this site function in N. pharaonis halorhodopsin
like Asp-85 in bacteriorhodopsin? In the present report we
demonstrate that azide bound to the chloride binding site
functions as an acceptor of the Schiff base proton during
the photocycle and that, after deprotonation of the Schiff
base, protons are released to the extracellular surface.
Another molecule of azide then shuttles protons from the
cytoplasmic surface to the unprotonated Schiff base, as in
H. salinarium halorhodopsin. The result is active electrogenic proton transport in the cytoplasmic to extracellular
direction. Thus, azide converts N. pharaonis halorhodopsin
into a light-driven proton pump analogous, with respect to
both the direction and the internal mechanism of the proton
translocation, to wild-type bacteriorhodopsin.
Biochemistry, Vol. 35, No. 21, 1996 6605
MATERIALS AND METHODS
Membrane fragments enriched for halorhodopsin were
prepared from H. salinarium containing a plasmid vector with
the structural genes for either H. salinarium or N. pharaonis
halorhodopsin and the bop promoter by the method described
before (Váró et al., 1995a). The T126D mutant was
constructed by conventional methods, using this vector. All
samples contained Na2SO4, with or without NaCl, so as to
keep the [Na+] at 2 M. Stationary spectra were measured
with a Shimadzu 1601 spectrophotometer connected to a
desktop computer.
Absorption changes were followed after photoexcitation
with a Nd-YAG laser pulse (532 nm, 7 ns), as in numerous
earlier publications of ours [e.g., Váró et al. (1995c)].
Transient pH changes during the photocycle were followed
with the indicator dye pyranine1 (Grzesiek & Dencher, 1986).
The temperature was regulated at 20 °C throughout.
Cell envelope vesicles were prepared by the method
described before (Lanyi & MacDonald, 1979). Transport
was measured at about pH 6 in 1.5 M Na2SO4, with or
without NaCl or K2SO4, so as to keep the [Na+] plus [K+]
at 3 M. In transport assays, transmembrane electrical
potential was measured with a TPP+ electrode (Shinbo et
al., 1978) and proton extrusion or uptake with a pH electrode
as in Váró et al. (1995a).
RESULTS
Binding of Chloride and Azide by Halorhodopsin. The
binding of chloride to halorhodopsin from N. pharaonis can
be detected through a blue shift of the absorption maximum
(Scharf & Engelhard, 1994; Váró et al., 1995a). Figure 1A
shows difference spectra that reveal a simple two-state
equilibrium, dependent on the binding of chloride. Essentially the same difference spectra, with an isosbestic point
at 602 nm, were obtained with azide (not shown), suggesting
that azide binds to the same site as chloride. In Figure 1B
the amplitudes of the difference spectra are plotted as
functions of chloride or azide concentrations. The apparent
dissociation constant for the chloride complex is 1.7 mM
and for the azide complex 10 mM, consistent with earlier
values (Scharf & Engelhard, 1994; Váró et al., 1995a). As
shown below, proton concentration does not change the
calculated dissociation constants of either chloride or azide.
Since the pKa of azide is 4.7 and therefore above pH 6 [N3-]
= [azide]total and [HN3] varies linearly with [H+], this pH
independence implicates the anionic form of azide in the
binding and the spectral shift.
If the anion binding is near the retinal Schiff base, as
suggested by the blue shift of the absorption band, it can be
expected to elevate the pKa of the protonated Schiff base
through coulombic effects and/or possibly through hydrogen
bonding. Spectroscopic titrations of the protonation state
of the Schiff base, utilizing the shift of the absorption
maximum from about 580 to 410 nm upon deprotonation,
were carried out in Na2SO4 and in various mixtures of sulfate
and NaCl or sodium azide. A series of such pH-dependent
different spectra, but in sulfate alone, is given in Figure 2A.
1 Abbreviations: TPP+, tetraphenylphosphonium ion; CCCP, carbonyl cyanide m-chlorophenylhydrazone; CAPS, 3-[cyclohexylamino]1-propanesulfonic acid; Bis-tris propane, 1,3-bis[[tris(hydroxymethyl)methyl]amino]propane; pyranine, 8-hydroxy-1,3,6-pyrenetrisulfonate.
6606 Biochemistry, Vol. 35, No. 21, 1996
Váró et al.
FIGURE 1: Binding of chloride and azide to halorhodopsin. (A)
Spectroscopic titration of membranes with NaCl in Na2SO4.
Chloride plus sulfate minus sulfate difference spectra 1-6 with
chloride concentrations of 1, 3, 10, 30, 100, and 300 mM. Total
[Na+] was kept at 2 M, buffered at pH 7.0 with 50 mM phosphate.
(B) Titration curves based on the blue shift of the absorption
maximum in panel A, upon adding chloride (O, dashed line) or
azide (b, solid line), and the amplitude of light-induced absorption
change at 410 nm (0) as in Figure 3A.
FIGURE 2: Effects of chloride and azide on the pKa of the protonated
Schiff base. (A) Deprotonation of the Schiff base of N. pharaonis
halorhodopsin in 1 M Na2SO4 upon raising the pH. Higher pH
minus pH 6.0 difference spectra 1-5 are given for pH 7.5, 8.0,
8.5, 9.0, and 9.5. Buffer was 25 mM CAPS plus 25 mM Bis-tris
propane. (B) pKa of the Schiff base, calculated from spectra such
as in panel A, as a function of chloride (O) or azide (b) in N.
pharaonis halorhodopsin, or chloride (0) in H. salinarium halorhodopsin.
The pKa values obtained for halorhodopsin from N. pharaonis
from these titrations are shown in Figure 2B as functions of
the chloride or azide concentration (open and closed circles).
The limiting pKa in the absence of chloride or azide is 7.98.0. In the presence of added anion the apparent pKa is
significantly raised, and to diffferent extents with chloride
and azide. The kinetic model that uniquely describes the
data is
obvious clues. The results show further that at the physiological chloride concentration of several molar the pKa of
the Schiff base in the N. pharaonis protein is raised to above
11.5, i.e., far above that of the H. salinarium protein. This
is in keeping with the fact that the pH of the natural
environment of N. pharaonis cells is above 10 (Tindall et
al., 1980), while H. salinarium grows near neutral pH.
Below pH about 7.5 no significant deprotonation of the
Schiff base occurred in the unphotolyzed proteins, in either
sodium sulfate alone or with added chloride. While sustained
illumination accumulated a state with unprotonated Schiff
base that reprotonated on the tens of minutes time scale as
described before (Lanyi & Schobert, 1983; Hegemann et al.,
1985), up to pH 7.5 its amount in the H. salinarium protein
was no more than about 10% of the total, and less in the N.
pharaonis protein (not shown). All the experiments described below were with halorhodopsin from N. pharaonis,
and the decrease of the amount of protein with protonated
Schiff base was not significant under any of the conditions
used.
Deprotonation of the Schiff Base upon Flash Illumination.
Absorption changes at 410 and 590 nm after pulse photoexcitation of halorhodopsin in Na2SO4 revealed that virtually
no deprotonation of the Schiff base (absorption increase at
410 nm) occurred in the absence of azide (Figure 3A). With
azide concentrations between 1 and 100 mM, however, such
a state, analogous to the M photointermediate of bacteriorhodopsin, did arise, and in significant amounts relative to
the depletion signal at 590 nm (Scharf & Engelhard, 1994;
Figure 3). At 100 mM azide the ratio of depletion at 590
nm to absorption rise at 410 nm was 1.5, a value not very
different from 1.7, the ratio calculated for 100% conversion
to the deprotonated Schiff base in the bacteriorhodopsin
photocycle (Zimányi & Lanyi, 1993). Thus, the transient
deprotonation of the Schiff base in halorhodopsin under these
-CdN + H+ S -CdNH+ + A- S -CdNH+ Awhere -CdN and -CdNH+ represent the unprotonated and
protonated retinal Schiff base, respectively, and A- is the
anion (chloride, or azide at pH well above its pKa). Thus,
chloride or azide binding depends on the positive charge of
the protonated Schiff base. The dissociation constant in the
best fit of this model (lines in Figure 2B) is 0.5 mM for
chloride and 8 mM for azide, consistent with the values from
the spectral shifts of the chromophore.
Figure 2B contains this kind of data also for halorhodopsin
from H. salinarium, in the presence of chloride. Although
the detergent-solubilized protein behaved differently with
respect to pH and chloride (Schobert & Lanyi, 1986), the
pKa of the Schiff base in the membrane-bound protein from
this species varies with chloride concentration similarly to
the N. pharaonis protein. It appears that while the absorption
maximum of the H. salinarium protein shifts toward the red
rather than toward the blue upon chloride binding (Schobert
et al., 1983; Steiner & Oesterhelt, 1983; Steiner et al., 1984),
suggesting that the binding site is different from that in the
N. pharaonis protein, the binding of the chloride is near
enough to the Schiff base to depend on its protonation state.
The nature of the difference in the binding sites in the
halorhodopsins from the two species is not yet clear, and
the residues in the vicinity of the Schiff base provide no
Proton Transport by Halorhodopsin
Biochemistry, Vol. 35, No. 21, 1996 6607
FIGURE 4: Flash-induced deprotonation of the Schiff base as a
function of pH. Experiments were performed as in Figure 3A, but
at 10 mM azide and the pH indicated.
FIGURE 3: Flash-induced transient absorption changes in the
presence of azide. (A) The appearance of deprotonated Schiff base
is detected at 410 nm in 1 M Na2SO4, pH 7.0, with azide
concentrations 0, 1, 3, 10, 30, and 100 mM, as indicated. (B) The
corresponding absorption changes at 590 nm, at the azide concentrations indicated in panel A in the direction of increasing negative
amplitudes.
conditions is extensive. Analysis of the traces in Figure 3A
indicated that the time constant for the rise of this M-like
state was nearly independent of azide but its amplitude was
strongly azide-dependent. Figure 1B includes also the
amplitudes of the absorption increase at 410 nm as functions
of azide concentration. They nearly coincide with the
amplitudes of the spectral shifts. The correlation between
the appearance of the M-like state in the photocycle and the
spectral shifts in the unphotolyzed protein suggests that the
deprotonation of the Schiff base depends on the prior binding
of the anionic azide near the Schiff base. It seems likely,
therefore, although not proven by these results, that the azide
bound at this location is the acceptor of the Schiff base
proton.
Unlike the rate of the rise, the rate of the decay of the
absorption at 410 nm is strongly dependent on the concentration of azide (Figure 3A). This dependence is linear with
azide concentration up to at least 100 mM. The catalysis of
the reprotonation of the Schiff base by azide is therefore
not related to the binding of azide detected by spectral shifts
that saturates with a binding constant of 10 mM (Figure 1B).
It must be based on a different mechanism. Figure 4 shows
the absorption changes at 410 nm at different pH values
between 5.5 and 7.0. Again, the rise and the decay behave
differently. The deprotonation of the Schiff base is pHindependent, while its reprotonation is strongly accelerated
when the pH is lowered.
The azide dependence of the M-like state in these
experiments implicates binding of azide at the anion site that
also can be occupied by chloride. Figure 5A shows that, as
expected, 10 mM azide is ineffective in the presence of 30
mM chloride. Increasing the concentration of azide to 300
mM partly overcomes this inhibition (Figure 5B), suggesting
that azide and chloride compete for the same site but
deprotonation of the Schiff base will occur only when azide
is the bound anion.
FIGURE 5: Competition of azide and chloride for binding. Flashinduced deprotonation of the Schiff base was followed as in Figure
3A, at different chloride concentrations with 10 mM azide (A) or
300 mM azide (B).
Transient Proton Release in the Presence of Azide. The
observed deprotonation of the Schiff base suggests that, as
in bacteriorhodopsin, under these conditions a proton might
be released in the halorhodopsin photocycle also. Figure 6
shows the absorption increase at 410 nm that originates from
deprotonation of the Schiff base, and the net absorption
change at 457 nm (with pyranine minus without pyranine)
that detects pH change in the bulk (Grzesiek & Dencher,
1986). The latter indicates that protons are transiently
released from the protein after deprotonation of the Schiff
base. The observed proton release is delayed relative to the
release to the surface, probably because buffering groups at
the surface delay the transit of the protons to the bulk, as
demonstrated for bacteriorhodopsin (Heberle & Dencher,
1992). The amplitude of the pH decrease relative to the
amplitude of the absorption rise at 410 nm is about half that
in bacteriorhodopsin, probably for the reason that, unlike
purple membranes, the membranes used here contained
significant amounts of proteins other than halorhodopsin and
these would contribute buffering. Another alternative would
be that not all of the protons from the Schiff base are released
to the surface. Because of its lesser amplitude, and because
6608 Biochemistry, Vol. 35, No. 21, 1996
FIGURE 6: Transient proton release and uptake after photoexcitation
in the presence of 10 mM azide. The trace at 410 nm shows
deprotonation of the Schiff base as in Figure 3A. The trace labeled
as pyranine reflects pH change in the bulk. Controls (not shown)
include wild-type bacteriorhodopsin, which revealed no significant
buffering by azide at the concentration used, and a halorhodopsin
sample without azide that revealed no pyranine signal.
the quantum yield of the photoisomerization of halorhodopsin
is about half that of bacteriorhodopsin (Lanyi, 1984; Oesterhelt et al., 1985), the pyranine signal was more difficult to
detect than in the other protein. Inasmuch as the signal/
noise ratio permits estimation of the time constant of the
subsequent proton uptake in Figure 6, it correlates with the
absorption decrease at 410 nm. No pH change was observed
in the absence of azide (not shown).
Transport ActiVity in the Presence of Azide. Cell envelope
vesicles prepared from halobacteria (Lanyi & MacDonald,
1979) can be used to assay light-driven ion transport activity,
through measurement of either membrane potential with a
TPP+ electrode (Shinbo et al., 1978) or the appearance or
disappearance of protons in the medium with a glass
electrode (Lanyi & Oesterhelt, 1982). The former is a
sensitive method for detecting electrogenic transport but does
not identify the transported ion, while the latter can distinguish between active proton transport (short-circuited by
protonophores but enhanced by abolishing the transmembrane electrical potential) and the transport of other species
when the latter is electrogenic and therefore results in passive
proton movement (enhanced by protonophores but eliminated
by abolishing the transmembrane electrical potential).
Illumination of envelope vesicles containing N. pharaonis
halorhodopsin in the presence of chloride produced a large
interior negative electrical potential from the inward chloride
transport and an accompanying passive proton uptake (Duschl et al., 1990; Váró et al., 1995a). In Na2SO4 a residual
small but reproducible pH change upon illumination had
suggested that, although much less effectively, SO42- or
HSO4- is also transported. Figure 7 shows the amplitudes
of transmembrane electrical potential measured with a TPP+
electrode under these conditions, but as functions of azide
concentration. The uptake of TPP+ indicated that in sulfate
a small negative interior electrical potential develops,
consistent with the passive proton influx reported earlier.
With added azide the amplitude of this electrical potential
increased, suggesting that additional electrogenic transport
occurs. However, above 2 mM azide the amplitude declined
again. The reason for this is evident from the observed effect
of azide on the electrical potential from chloride transport
(Figure 7). Significant competition of azide for the chloride
Váró et al.
FIGURE 7: Azide-dependent electrogenic transport upon illumination
of halorhodopsin. Cell envelope vesicles containing N. pharaonis
halorhodopsin in 1.5 M Na2SO4 were illuminated with continuous
yellow light (>530 nm). Light-dependent transmembrane electrical
potential was measured with a TPP+ (tetraphenylphosphonium)
electrode in the absence of chloride (O) or its presence in 1.35 M
Na2SO4 plus 0.3 M NaCl (b). In the measured range the electrode
response is roughly linear with the transmembrane electrical
potential in the vesicles. Dashed line is the calculated effect of azide
without its protonophoric activity, with the response from sulfate
subtracted.
binding site in the protein would not occur at the chloride
concentration of 300 mM used. However, as is well-known,
azide will abolish transmembrane electrical potential as a
protonophore and/or membrane-permeant anion because both
its protonated and anionic forms will cross the membranes.
The data in sulfate plus azide were corrected for this effect,
using a simple pump and leak model in which the leak is
linearly dependent on the amount of accumulated charges
in the vesicles and defined by the parameters of the inhibition
in the control curve in Figure 7. This calculation produced
the curve shown with a dotted line. If it represents the
electrical potential that would be created without the protonophoric effect of azide, the azide-dependent electrogenic
transport is comparable in extent to the chloride transport.
The apparent binding constant for azide is then 9 mM. This
value is consistent with the dissociation constant for the
binding of azide, from both spectral shift and the deprotonation of the Schiff base (Figure 1B). These results therefore
correlate the electrogenic transport with the binding of the
azide and the azide-dependent deprotonation of the Schiff
base in the photocycle, but they do not identify the
transported ion.
Measurements of illumination-dependent pH change in
suspensions of these vesicles in 1.5 M sodium sulfate, under
the conditions used in Figure 7, demonstrated that, depending
on the vesicle preparation, 7-20 protons/halorhodopsin were
taken up (not shown). This is expected to occur in response
to the interior negative electrical potential that develops from
sulfate transport. Added at 5 µM, the protonophore CCCP
somewhat increased the amplitude of this passive proton
influx and caused it to be more rapid, also as expected. After
addition of 2 mM azide, however, the proton uptake changed
to proton extrusion (not shown), in spite of the fact that under
Proton Transport by Halorhodopsin
Biochemistry, Vol. 35, No. 21, 1996 6609
FIGURE 9: Photocycle of T126D halorhodopsin in 1 M Na2SO4.
Flash-induced transient absorption changes are shown at 640, 410,
and 520 nm. Buffer was 50 mM Bis-tris propane at pH 7.0. Solid
lines, no azide; dashed lines, 100 mM azide.
FIGURE 8: Azide-dependent active transport of protons upon
illumination of halorhodopsin. Suspensions of cell envelope vesicles
containing N. pharaonis halorhodopsin were assayed in 1.3 M Na2SO4 plus 0.2 M K2SO4 for light-dependent pH changes. The
indicated additions were cumulative.
these conditions the interior negative electrical potential
increases (Figure 7). Whether this extrusion represents active
proton transport, as its direction suggests, was better studied
with the transmembrane potential clamped at zero by
equilibrating the vesicles with 1.3 M Na2SO4 plus 0.2 M
K2SO4 and adding the K+ ionophore valinomycin. Under
these conditions any possibility for passive proton transport
was removed. Indeed, as shown in Figure 8, the lightdependent passive proton influx was entirely eliminated upon
addition of 10 µM valinomycin. After addition of 2 mM
azide to this mixture, illumination caused rapid and extensive
proton extrusion that was sensitive to the protonophore CCCP
(Figure 8). This result demonstrates unequivocally that the
azide-dependent interior negative electrical potential in Figure
7 originates from active, outward-directed proton transport.
Inward transport of azide, while consistent with the increased
electrical potential, would have produced increased passive
proton uptake and not when the electrical potential was
clamped at zero. Outward transport of azide is inconsistent
with the sign of the electrical potential. It would have caused
the observed proton extrusion, but not with the electrical
potential kept at zero. In neither case would CCCP abolish
the pH change.
Properties of T126D Halorhodopsin. The threonine
residue in halorhodopsin from N. pharaonis at the position
of Asp-85 in bacteriorhodopsin (Thr-126) was replaced with
an aspartate. Such a recombinant protein, constructed from
halorhodopsin in H. salinarium and also containing an
aspartate equivalent to Asp-96 in bacteriorhodopsin, did not
transport protons [mentioned in Havelka et al. (1995)]. The
absorption maximum of the T126D mutant was blue shifted
by 40 nm relative to wild type in 1.5 M sulfate. Presumably,
the anionic aspartate near the Schiff base shifts the absorption
band in the same manner as chloride or azide bound to the
wild-type protein. Lowering the pH, between 7 and 2, did
not result in a red shift, as in bacteriorhodopsin, that would
indicate that the aspartate becomes protonated (not shown).
Possibly, the pKa of Asp-126 is lower than the pKa of its
equivalent in bacteriorhodopsin. Interestingly, however, the
mutant protein was found to be stable at acid pH, unlike the
wild-type halorhodopsin that irreversibly denatures below
pH 5. In experiments similar to those in Figure 1, chloride
(up to 2 M) or azide (up to 100 mM) did not shift the
absorption maximum of the mutant, suggesting that the anion
binding site is abolished by the threonine to aspartate
replacement.
Significantly, as evident from Figure 9, no deprotonation
of the Schiff base occurs in the photocycle of the T126D
protein. The absorption changes are essentially the same as
in the wild-type protein in sulfate (Váró et al., 1995a).
Unlike the wild-type protein, however, the absorption
changes in the photocycle of the mutant did not change in 2
M NaCl (not shown). Furthermore, lack of absorption
change at 410 nm indicated that no deprotonation of the
Schiff base occurred in the presence of 100 mM azide, in
either sulfate (Figure 9) or chloride (not shown). Evidently,
the engineered aspartate prevents the binding of azide, but a
low pKa, or an unfavorable geometry relative to the Schiff
base, does not allow it to act as proton acceptor. Lack of
pH change upon illumination of cell envelope vesicles
containing T126D halorhodopsin in the presence of K+ and
valinomycin (not shown) confirmed that the threonine to
aspartate mutation does not confer proton transport activity
on this protein.
DISCUSSION
We had found earlier that the ionic specificity of bacteriorhodopsin changed from protons to chloride when the
proton acceptor residue Asp-85 was replaced with a threonine
(Sasaki et al., 1995). In this mutant, chloride was transported
in the extracellular to cytoplasmic direction, as in halorhodopsin. Thus, we concluded that bacteriorhodopsin
contains all structural properties necessary for chloride
transport, except an anion binding site on the extracellular
side of the Schiff base. Does halorhodopsin contain, in the
same way, all structural attributes that are necessary to
transport protons in the cytoplasmic to the extracellular
direction, except for a proton acceptor? We report results
demonstrating that when an appropriate proton donor (and
acceptor) is provided to the Schiff base, halorhodopsin will
indeed transport protons in the way bacteriorhodopsin does.
In the presence of the weak acid azide, the photocycle of
the N. pharaonis halorhodopsin is changed so as to include
the deprotonation and the subsequent reprotonation of the
6610 Biochemistry, Vol. 35, No. 21, 1996
retinal Schiff base (Scharf & Engelhard, 1994; Figures 3-5)
and the release and the subsequent uptake of protons (Figure
6). These proton exchange reactions must occur in a
vectorial fashion because their consequence is the active
electrogenic transport of protons from the cytoplasmic to the
extracellular direction (Figures 7 and 8). The observations
related to the binding of azide and the kinetics of the
protonation reactions in the photocycle suggest a mechanism
for the transport. Azide appears to bind to the chloride
binding site of this protein that is near the retinal Schiff base,
and the azide dependence of the light-induced deprotonation
of the Schiff base correlates with the dissociation constant
of this binding (Figure 1B). The location of this anion
binding site is not firmly identified, but consideration of the
residues that are in the vicinity of the Schiff base (Blanck
& Oesterhelt, 1987; Lanyi et al., 1990) and the phenotype
of a mutant with replaced arginine in this region (Rüdiger
et al., 1995) suggest that it is to the extracellular side of the
Schiff base. Lack of chloride and azide binding in the
T126D mutant are consistent with the extracellular location
of the site. Such a location is suggested further by the
chloride-dependent spectroscopic changes in the photocycles
of this halorhodopsin (Váró et al., 1995b) and D85T
bacteriorhodopsin (Sasaki et al., 1995) in the context of the
extracellular to cytoplasmic direction of the chloride transport, and the observation that the chloride-binding affinity
of D85S is greatly different from that of D85T bacteriorhodopsin.2 We had considered three possibilities for how
the azide bound to this site could facilitate deprotonation of
the Schiff base. First, the azide anion could be translocated
like chloride from the extracellular to the cytoplasmic surface
but transiently protonated at the same time from the Schiff
base. The subsequent reprotonation of the Schiff base from
the cytoplasmic side would result in the net translocation of
N3-. This alternative is ruled out by the observations in
Figure 8 that demonstrate active transport of protons rather
than of azide. Reprotonation of the Schiff base from the
extracellular rather than the cytoplasmic side would result
in the net translocation of HN3. This is ruled out by the
fact that the transport is electrogenic (Figure 7). Second,
the bound azide could facilitate deprotonation of the Schiff
base but with the proton passing through a hydrogen-bonded
chain to the cytoplasmic direction, as suggested for bacteriorhodopsin (Le Coutre et al., 1995). This would produce
the observed transient proton release to the bulk (Figure 6)
but would not result in the negative inside electrical potential
(Figure 7) or proton transport in the cytoplasmic to extracellular direction (Figure 8). In any case, the azide binding
site in N. pharaonis halorhodopsin must be different from
the proposed site in bacteriorhodopsin (Le Coutre et al.,
1995) because unlike in halorhodopsin (Figure 1A) no
spectral shifts can be seen in bacteriorhodopsin when azide
(or chloride) is added (not shown). Third, the azide anion
could be the acceptor for the Schiff base proton in the
photocycle, analogous to Asp-85 in bacteriorhodopsin. In
this alternative, azide itself cannot be transported because
once protonated it is no longer an anion. The pH changes
observed with pyranine argue that after protonation the azide
either promptly releases the proton to the extracellular surface
or the HN3 is lost from the binding site and dissociates to
H+ and N3- in the bulk. This would be unlike Asp-85 in
2
L. S. Brown, R. Needleman, and J. K. Lanyi, unpublished results.
Váró et al.
bacteriorhodopsin, which retains the proton until the end of
the photocycle but causes instead proton release from Glu204 (Brown et al., 1995; Richter et al., 1996). On the other
hand, halorhodopsin does not contain a protonatable residue
at the location equivalent to Glu-204. The third alternative
explains all of the results.
The subsequent reprotonation of the Schiff base in the
photocycle is also facilitated by azide, but similarly to how
azide functions in H. salinarium halorhodopsin (Lanyi, 1986)
and in the D85N/D96N bacteriorhodopsin mutant (Kataoka
et al., 1994) that do not contain suitable extracellular binding
sites, i.e., merely by shuttling protons between the Schiff
base and the cytoplasmic surface. Shuttling is suggested by
the observation of nonsaturating kinetics (at least up to 100
mM), inconsistent with the binding constant of 10 mM for
the azide (Figure 3A). This distinguishes it from the
deprotonation process. Reprotonation by a shuttling reaction
would be either by uptake of HN3 from the bulk or by H+
uptake followed by conduction of the proton from the
membrane surface to the Schiff base via N3- bound with
low affinity. The pH dependence of the reprotonation
(Figure 4) rules out the alternative in which the proton donor
is the bound HN3 produced during the deprotonation of the
Schiff base. The finding of active electrogenic proton
transport in the cytoplasmic to the extracellular direction
requires unequivocally that if azide is the proton acceptor
to the Schiff base from the extracellular side, as strongly
suggested by the results, then it must be from the cytoplasmic
side that it acts as proton donor. Thus, it appears that the
azide molecules at these two locations are functionally
equivalent to Asp-85 and Asp-96, respectively, of bacteriorhodopsin.
The results in this report, demonstrating active proton
transport by halorhodopsin, complement the earlier finding
of chloride transport by bacteriorhodopsin (Sasaki et al.,
1995). In both cases a binding site for the unphysiologically
transported ion was changed or added, but the photoreactions
and conditions under which the transport of the physiologically transported ion would occur were unaltered. We
conclude therefore that the ion specificities of bacteriorhodopsin and halorhodopsin reside in a single binding site
rather than in the structure of the ion-conductiVe pathways.
Likewise, the mechanism of the reorientation switch that
confers directionality to the translocation must be common
to the two kinds of transport and not dependent on the kind
of ion transported. It must be inherent in the common
structure of these proteins.
REFERENCES
Bamberg, E., Tittor, J., & Oesterhelt, D. (1993) Proc. Natl. Acad.
Sci. U.S.A. 90, 639-643.
Blanck, A., & Oesterhelt, D. (1987) EMBO J. 6, 265-273.
Bogomolni, R. A., Stoeckenius, W., Szundi, I., Perozo, E., Olson,
K. D., & Spudich, J. L. (1994) Proc. Natl. Acad. Sci. U.S.A. 91,
10188-10192.
Brown, L. S., & Lanyi, J. K. (1996) Proc. Natl. Acad. Sci. U.S.A.
93, 1731-1734.
Brown, L. S., Sasaki, J., Kandori, H., Maeda, A., Needleman, R.,
& Lanyi, J. K. (1995) J. Biol. Chem. 270, 27122-27126.
Cao, Y., Váró, G., Chang, M., Ni, B., Needleman, R., & Lanyi, J.
K. (1991) Biochemistry 30, 10972-10979.
Dickopf, S., Alexiev, U., Krebs, M. P., Otto, H., Mollaaghababa,
R., Khorana, H. G., & Heyn, M. P. (1995) Proc. Natl. Acad.
Sci. U.S.A. 92, 11519-11523.
Proton Transport by Halorhodopsin
Duschl, A., Lanyi, J. K., & Zimányi, L. (1990) J. Biol. Chem. 265,
1261-1267.
Ebrey, T. G (1993) in Thermodynamics of membranes, receptors
and channels, (Jackson, M., Ed.) pp 353-387, CRC Press, Boca
Raton, FL.
Govindjee, R., Balashov, S., Ebrey, T., Oesterhelt, D., Steinberg,
G., & Sheves, M. (1994) J. Biol. Chem. 269, 14353-14354.
Grzesiek, S., & Dencher, N. A. (1986) FEBS Lett. 208, 337-342.
Haupts, U., Haupts, C., & Oesterhelt, D. (1995) Proc. Natl. Acad.
Sci. U.S.A. 92, 3834-3838.
Havelka, W. A., Henderson, R., & Oesterhelt, D. (1995) J. Mol.
Biol. 247, 726-738.
Heberle, J., & Dencher, N. A. (1992) Proc. Natl. Acad. Sci. U.S.A.
89, 5996-6000.
Hegemann, P., Oesterhelt, D., & Steiner, M. (1985) EMBO J. 4,
2347-2350.
Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann,
E., & Downing, K. H. (1990) J. Mol. Biol. 213, 899-929.
Kataoka, M., Kamikubo, H., Tokunaga, F., Brown, L. S., Yamazaki,
Y., Maeda, A., Sheves, M., Needleman, R., & Lanyi, J. K. (1994)
J. Mol. Biol. 243, 621-638.
Keszthelyi, L., Száraz, S., Dér, A., & Stoeckenius, W. (1990)
Biochim. Biophys. Acta 1018, 260-262.
Lanyi, J. K. (1984) Biochem. Biophys. Res. Commun. 122, 91-96.
Lanyi, J. K. (1986) Biochemistry 25, 6706-6711.
Lanyi, J. K. (1990) Physiol. ReV. 70, 319-330.
Lanyi, J. K. (1993) Biochim. Biophys. Acta 1183, 241-261.
Lanyi, J. K., & MacDonald, R. E. (1979) Methods. Enzymol. 56,
398-407.
Lanyi, J. K., & Oesterhelt, D. (1982) J. Biol. Chem. 257, 26742677.
Lanyi, J. K., & Schobert, B. (1983) Biochemistry 22, 2763-2769.
Lanyi, J. K., Duschl, A., Hatfield, G. W., May, K. M., & Oesterhelt,
D. (1990) J. Biol. Chem. 265, 1253-1260.
Le Coutre, J., Tittor, J., Oesterhelt, D., & Gerwert, K. (1995) Proc.
Natl. Acad. Sci. U.S.A. 92, 4962-4966.
Oesterhelt, D., & Tittor, J. (1989) Trends Biochem. Sci. 14, 5761.
Oesterhelt, D., Hegemann, P., & Tittor, J. (1985) EMBO J. 4, 23512356.
Oesterhelt, D., Tittor, J., & Bamberg, E. (1992) J. Bioenerg.
Biomembr. 24, 181-191.
Biochemistry, Vol. 35, No. 21, 1996 6611
Olson, K. D., & Spudich, J. L. (1993) Biophys. J. 65, 2578-2585.
Otomo, J., Tomioka, H., & Sasabe, H. (1992) Biochim. Biophys.
Acta 1112, 7-13.
Otto, H., Marti, T., Holz, M., Mogi, T., Lindau, M., Khorana, H.
G., & Heyn, M. P. (1989) Proc. Natl. Acad. Sci. U.S.A. 86,
9228-9232.
Richter, H.-T., Brown, L. S., Needleman, R., & Lanyi, J. K. (1996)
Biochemistry 35, 4054-4062.
Rothschild, K. J. (1992) J. Bioenerg. Biomembr. 24, 147-167.
Rüdiger, M., Haupts, U., Gerwert, K., & Oesterhelt, D. (1995)
EMBO J. 14, 1599-1606.
Sasaki, J., Brown, L. S., Chon, Y.-S., Kandori, H., Maeda, A.,
Needleman, R., & Lanyi, J. K. (1995) Science 269, 73-75.
Scharf, B., & Engelhard, M. (1994) Biochemistry 33, 6387-6393.
Schobert, B., & Lanyi, J. K. (1986) Biochemistry 25, 4163-4167.
Schobert, B., Lanyi, J. K., & Cragoe, E. J. J. (1983) J. Biol. Chem.
258, 15158-15164.
Shinbo, T., Kamo, N., Kurihara, K., & Kobatake, Y. (1978) Arch.
Biochem. Biophys. 187, 414-422.
Soppa, J., Duschl, J., & Oesterhelt, D. (1993) J. Bacteriol. 175,
2720-2726.
Steiner, M., & Oesterhelt, D. (1983) EMBO J. 2, 1379-1385.
Steiner, M., Oesterhelt, D., Ariki, M., & Lanyi, J. K. (1984) J.
Biol. Chem. 259, 2179-2184.
Subramaniam, S., Marti, T., & Khorana, H. G. (1990) Proc. Natl.
Acad. Sci. U.S.A. 87, 1013-1017.
Tindall, B. J., Mills, A. A., & Grant, W. D. (1980) J. Gen.
Microbiol. 116, 257-260.
Tittor, J., Soell, C., Oesterhelt, D., Butt, H.-J., & Bamberg, E. (1989)
EMBO J. 8, 3477-3482.
Tittor, J., Schweiger, U., Oesterhelt, D., & Bamberg, E. (1994)
Biophys. J. 67, 1682-1690.
Váró, G., Brown, L. S., Sasaki, J., Maeda, A., Needleman, R., &
Lanyi, J. K. (1995a) Biochemistry 34, 14490-14499.
Váró, G., Needleman, R., & Lanyi, J. K. (1995b) Biochemistry 34,
14500-14507.
Váró, G., Zimányi, L., Fan, X., Sun, L., Needleman, R., & Lanyi,
J. K. (1995c) Biophys. J. 68, 2062-2072.
Zimányi, L., & Lanyi, J. K. (1993) Biophys. J. 64, 240-251.
BI9601159