Proc. Natl. Acad. Sci. USAVol. 92, pp. 11120-11124, November 1995Biophysics

Molecular mechanism of protein-retinal couplingin bacteriorhodopsin

(energy transduction/proton pump/membrane protein/spectroscopy/retinal isomerization)

JOHN K. DELANEY*, ULRIKE SCHWEIGERt, AND SRIRAM SUBRAMANIAM*t*Department of Biological Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD 21205; and tDepartment of Biochemistry, Max PlanckInstitute for Biochemistry, Martinsried, Germany

Communicated by Walther Stoeckenius, University of California, Santa Cruz, CA, August 10, 1995

ABSTRACT Bacteriorhodopsin is a membrane proteinthat functions as a light-driven proton pump. Each cycle ofproton transport is initiated by the light-induced isomeriza-tion of retinal from the all-trans to 13-cis configuration and iscompleted by the protein-driven reisomerization of retinal- tothe all-trans configuration. Previous studies have shown thatreplacement of Leu-93, a residue in close proximity to the13-methyl group of retinal, by alanine, resulted in a 250-foldincrease in the time required to complete each photocycle.Here, we show that the kinetic defect in the photocycle of theLeu-93 -> Ala mutant occurs at a stage after the completionof proton transport and can be overcome in the presence ofstrong background illumination. Time-resolved retinal-extraction experiments demonstrate the continued presence ofa 13-cis intermediate in the photocycle of the Leu-93 -* Alamutant well after the completion of proton release and uptake.These results indicate that retinal reisomerization is kineti-cally the rate-limiting step in the photocycle of this mutantand that the slow thermal reisomerization can be bypassed bythe absorption of a second photon. The effects observed for theLeu-93 -> Ala mutant are not observed upon replacement ofany other residue in van der Waals contact with retinal orupon replacement of Leu-93 by valine. We conclude that thecontact between Leu-93 and the 13-methyl group of retinalplays a key role in controlling the rate of protein conforma-tional changes associated with retinal reisomerization andreturn of the protein to the initial state.

Isomerization of retinal by light is the first step in the trans-duction of light energy by proteins in the rhodopsin family. Inbacterial rhodopsins such as bacteriorhodopsin and halorho-dopsin, retinal isomerization is coupled to the vectorial trans-port of protons and chloride ions, respectively, across the cellmembrane, whereas in the visual rhodopsins, retinal isomer-ization is coupled to the activation of guanine nucleotide-binding proteins (G-proteins) present in photoreceptor cells(1, 2). In the bacterial rhodopsins, each cycle of light energytransduction is completed by protein-driven reisomerization ofretinal to the starting configuration. In contrast, in the visualrhodopsins, the return of retinal to the initial state is accom-plished either by absorption of a second photon or by exchangeof the isomerized retinal with 11-cis-retinal (3, 4). The molec-ular mechanism by which energy stored in the polypeptidechain is used to drive a change in retinal geometry is offundamental interest in understanding light energy transduc-tion by the bacterial rhodopsins.

In bacteriorhodopsin, light-induced retinal isomerizationfrom the all-trans to the 13-cis configuration triggers a seriesof changes in protein conformation that result in three impor-tant stages in the photocycle: I, the release of a proton into theextracellular medium; II, the uptake of a proton from the

cytoplasmic medium; and III, the thermal reisomerization ofretinal to the starting all-trans configuration. These confor-mational changes are reflected as changes in the absorptionspectrum, as evidenced by the sequential formation and decayof the optical intermediates J, K, L, M, N, and 0 (5, 6).Spectroscopic studies of wild-type bacteriorhodopsin haveshown that the release of a proton (T- 50 ,tsec) into theextracellular medium coincides with the formation of the Mintermediate (7). Proton uptake and retinal reisomerizationare thought to occur during the N-to-O transition (8). At 250C,the cycle is complete in about 10 msec. Detailed spectroscopicand biochemical analyses of many site-specific mutants (9, 10)have established that Asp-85, Arg-82, and Asp-212 form acluster of interacting residues in the extracellular half ofbacteriorhodopsin that play a key role in the proton-releasepathway (i.e., stage I of the photocycle) and that Asp-96,Thr-46, and Arg-227 form a cluster of residues in the cyto-plasmic half that are involved in proton uptake (stage II). Incontrast, little is known about the molecular mechanism ofretinal reisomerization (stage III), which is thought to involvereisomerization of retinal from the 13-cis to a twisted all-transstate (11), followed by conversion to the starting all-transconfiguration.From an analysis of the properties of mutants carrying

replacements of residues that make contact with retinal,Subramaniam et al. (12) and Greenhalgh et al. (13) reportedthat replacement of Leu-93 by alanine or Val-49 by alanineincreased the time required to complete each photocycle by-250-fold and -10-fold, respectively. This increase is due tothe accumulation of an equilibrium mixture of intermediatesoptically resembling the N and 0 intermediates of the wild-type photocycle, populated during stage III of the photocycle.A surprising feature of these mutants was that despite theslower cycling time, the steady-state vectorial proton-pumpingefficiency in reconstituted vesicles under saturating yellowlight was comparable to that of wild-type bacteriorhodopsin(12), leading to two hypotheses: (i) the interactions of thenonpolar side chains of Leu-93 and Val-49 with retinal areimportant for either proton uptake or reisomerization or both,and (ii) absorption of a photon by the long-lived intermediatesgreatly accelerates completion of the photocycle, thus ac-counting for the normal steady-state transport activity. Here,we test these hypotheses and present evidence that the van derWaals contact between Leu-93 and retinal plays a key role inthe rapid thermal reisomerization of retinal observed in thephotocycle of bacteriorhodopsin.

MATERIALS AND METHODSConstruction and Purification of Mutants. Construction of

halobacterial strains expressing the Leu-93 -> Ala mutant andpurification of purple membranes containing the mutant bac-teriorhodopsins were carried out as described (14, 15). Purified

tTo whom reprint requests should be addressed.

11120

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

May

30,

202

1

Proc. Natl. Acad. Sci. USA 92 (1995) 11121

membranes (at an optical density of 0.5 at 540 nm) weresuspended in 10 mM sodium phosphate buffer (pH 7.0)containing 0.15 M KCI.

Transient Absorption Spectroscopy. A 10-Asec flash at 550nM (-0.1 mJ/cm2 per flash) from a xenon lamp was used toinitiate the photocycle, and a weak probe beam was used tofollow the subsequent transient absorbance changes. Theprobe beam was filtered by a 0.25-m monochromator (270M,Spex), and the transient photocurrent was amplified anddigitized by a 12-bit analog/digital board (DAS 1401,Keithley). The source of the continuous background illumi-nation for experiments presented in Fig. 2 was a 150-W slideprojector, whose light was filtered with an orange glass cut onfilter (A > 560 nm). The irradiance at the sample (peak value-20 mW/cm2) was varied by using neutral density filters. In atypical experiment, 64 transient absorption traces were aver-aged from a sample volume of 1 ml (optical density 0.5 at540 nm). The rate of repopulation of the initial state wasfollowed at 483 nm, the decay of the M intermediate wasfollowed at 405 nm, and the formation and decay of the 0intermediate were followed at 640 nm for wild-type bacterio-rhodopsin (Amax, at 570 nm) and at 625 nm or 605 nm for theLeu-93 -- Ala mutant (Am.x at 540 nm). Since the N and 0intermediates are in equilibrium (12), the kinetics measuredfor the formation and decay of the 0 intermediate reflect thekinetics of formation and decay of the equilibrium mixture ofN and 0 intermediates.Proton Kinetics Measurements. Kinetics of proton release

and uptake was measured in 0.15 M KCl as described (7) bytaking the difference between absorbance transients recordedon samples with and without pyranine (8-hydroxyl-1,3,6-pyrene trisulfonate; Sigma) at a concentration of -35 ,uM.

Retinal Extraction. Membranes (in 20mM phosphate bufferat pH 6.0; optical density at 540 nm 0.2) were illuminatedfor 4 sec at -5°C with saturating yellow light (A > 475 nm).Transient absorbance measurements under identical condi-tions showed that only the N and 0 intermediates were presentunder these conditions. Conversion to the initial bacteriorho-dopsin state was complete (>95%) within 40 sec. Retinalextraction was carried out as described in ref. 16. At differenttimes after illumination, the conversion of the N and 0intermediates to the initial state was quenched (protein de-naturation time 1-2 sec) by the addition of 700 ,lI of2-propanol to 300 ,ul of membranes. Retinal was extracted inton-hexane, and the composition of retinal isomers was deter-mined by HPLC.

RESULTSKinetics of Proton Uptake in the Leu-93 -* Ala Mutant. To

determine whether proton release or uptake (stages I and II)is a rate-limiting step in the photocycle of the Leu-93 -+ Alamutant, we measured the kinetics of light-driven protonationchanges in parallel with the formation and decay of the M, N,and 0 intermediates (Fig. 1). Proton release and formation ofthe M intermediate occurred in - 1 msec. The kinetics ofproton uptake (v 6 msec) was similar to both the rate ofdecay of the M intermediate (T 4 msec) and the formationof the 0 intermediate (T 5.5 msec) as observed for wild-typebacteriorhodopsin (7, 17). However, the lifetime of the 0intermediate in the Leu-93 -- Ala mutant was increased by-250-fold (T 1.8 sec) relative to wild-type bacteriorhodopsin(v 7 msec). [This lifetime for the 0 intermediate in theLeu-93 -- Ala mutant, which was obtained by using a weaknarrow-band (3 nm) probe beam, is about 1.5 times longer thanthe corresponding lifetime reported in ref. 12, where themeasurements were carried out in detergent micelles with acontinuous white-light probe beam.] The time required forcompletion of the photocycle is therefore "100 times longerthan the time required for proton uptake, implying that the

6

cf,0

x

2

4

2

0

ALeu 93- Ala

wild-type

0.01 0.1 1 10

Time (sec)

I

e

H uptake

0 10 20 30 40

Time (msec)FIG. 1. Measurements of photocycle kinetics at pH 6.3 in wild-type

bacteriorhodopsin and the Leu-93 -* Ala mutant at 22'C. (A) Tran-sient changes in absorbance showing the kinetics of decay of the 0intermediate in the photocycle of the Leu-93 -> Ala mutant (measuredat 625 nm) and in wild-type bacteriorhodopsin (measured at 640 nm).(B) Transient absorption traces showing the decay of the M interme-diate (magnified 4-fold, measured at 405 nm) and the rise of the 0intermediate following flash excitation of the Leu-93 -> Ala mutant.The negative transient absorbance at 405 nm for times in excess of -30msec is due to complete decay of the M intermediate and due to thepresence of N and 0 intermediates, which are expected to have lowerextinction coefficients than bacteriorhodopsin at this wavelength. Therecovery of this negative absorbance to zero occurs on the same timescale observed for the decay of the 0 intermediate and matches therecovery of the initial bacteriorhodopsin state. (C) Kinetics of light-driven proton uptake in the Leu-93 -- Ala mutant. The maximal pHdrift during the course of the entire experiment was Ala mutant is greatly delayed at

stage III of the photocycle.Effect of Background Light on the Photocycling Rate in the

Leu-93 -- Ala Mutant. To test whether light absorption by the

long-lived intermediate(s) could accelerate the photocycle ofthe Leu-93 -> Ala mutant, we measured the photocycling time

in the presence of increasing amounts of constant backgroundlight (Fig. 2). Under these conditions, the photocycling timeconstant in the Leu-93 -> Ala mutant decreased progressivelyfrom 1.8 sec to 88 msec with increasing orange (A > 560 nm)background illumination, approaching the photocycling time

2.0

._~ ~~~~~~

1.5CO)

0 10 20 30SLight Intensity (%)

0.5

0 25 50 75 100

Light Intensity (%)

FIG. 2. Acceleration of photocycling time in the Leu-93 -* Alamutant in the presence of background illumination at pH 7.0. Theabsorbance changes at 483 nm were fit to a single rising exponential.The time constants of repopulation of the initial state in the presenceof orange background light (A > 560 nm) (-) are shown as a functionof light intensity, which was varied by using neutral density filters. Thecycling time was reduced from 1.8 sec, when no background illumi-nation was present, to 88 msec at the highest illumination levels used.(Similar results were obtained at pH 6.0.) (Inset) The same data arepresented in a plot of the rate of ground-state repopulation vs.background illumination, which shows that the rate extrapolates atzero light intensity to the experimentally observed value (-) in theabsence of background illumination.

0

0.5

0.0

-0.5

Biophysics: Delaney et aL

Dow

nloa

ded

by g

uest

on

May

30,

202

1

Proc. Natl. Acad. Sci. USA 92 (1995)

observed for wild-type bacteriorhodopsin (Fig. 1). At lowerbackground light levels, the photocycling rate varied linearlywith background light intensity (Fig. 2 Inset) and, whenextrapolated to zero background light intensity, matched therate experimentally obtained without background illumina-tion. These results demonstrate that one or both of thelong-lived intermediates can be photochemically converted tothe initial bacteriorhodopsin state, thus bypassing the slowthermal recovery. Since the only likely effect of the back-ground light is to reisomerize retinal, we conclude that retinalreisomerization is the rate-limiting step in the photocycle ofthe Leu-93 -> Ala mutant. Hence, under background illumi-

nation, a rapid wild-type-like photocycling time is achieved inthe Leu-93 -- Ala mutant using two photons. The first photoninitiates proton transport by isomerizing retinal from theall-trans to the 13-cis configuration. Upon completion of stepsin proton translocation, the second photon resets retinal andthe protein to the initial bacteriorhodopsin state.

Extraction of Retinal Following Illumination of the Leu-93> Ala Mutant. To directly establish that retinal reisomeriza-

tion is slowed down in the Leu-93 -- Ala mutant, we carried

out time-resolved chemical extraction of retinal at differenttimes after light excitation of the membranes at -5°C (Fig. 3).The concentration of the 13-cis isomer increases to -65%following illumination and then decreases slowly to the initialvalue of '20%. The conkentration of the all-trans isomerdecreases initially to -35% following illumination and in-creases slowly to the initial value of '80%. From thesemeasurements, the time constant for the thermal interconver-sion of 13-cis and all-trans isomers is '20-25 sec, although thetime required for protein denaturation and the temperaturefluctuations during the experiment make this only an approx-imate estimate. Transient absorption spectroscopic measure-ments carried out under the same conditions showed that thedecay of the N and 0 intermediates to bacteriorhodopsinoccurred on a similar time scale (T 16 sec at 5°C from Fig.4), implying that the conversion of the 13-cis form to theall-trans form occurs with the conversion of the long-lived Nand 0 intermediates to bacteriorhodopsin. These experimentsdemonstrate the continued presence of a 13-cis intermediatein the photocycle well after the completion of proton uptake.Temperature Dependence of the Amplitude and Kinetics ofIntermediate in the Leu-93 -- Ala Mutant. To further

characterize the nature of the kinetic defect in the Leu-93Ala mutant, we determined the temperature dependence of

l r

800

E0

0cn=ta:

60

40

20

1500 50 100

Time (sec)

FIG. 3. Time-dependent changes in the concentration of 13-cis (0)and all-trans (0) retinal isomers at different times after illuminationof the Leu-93 -> Ala mutant at pH 6.0. The ratio of the all-trans and13-cis isomers is about 80%:20% in both the dark- and light-adaptedstates of this mutant (12). The concentration of the 13-cis isomerincreased by -45% following illumination and subsequently decayedto the initial value. The concentration of the all-trans isomer decreasedby -45% after illumination and returned to the initial value. No otherisomers besides the all-trans and 13-cis forms were detected.

_ .

0

x 7.5

7.0a)70~0

* 6.5E

6.0

0

C.)6)

0

0

C

-1

-2

-33i.3 3.4 3.5 3.6

1/T x 1000 (1/K)

FIG. 4. Temperature dependence of the amplitude (A) and timeconstant (B) of decay of the absorbance at 605 nm in the photocycleof the Leu-93 -- Ala mutant at pH 7.0 between 25.5°C and 5°C. Thedecays were fit to single exponentials to determine the respective timeconstants. The slope of the line in B leads to an activation energy of-77 kJ/mol for the transition from the 0 intermediate to the initialbacteriorhodopsin state. (Similar results were obtained at pH 6.0.)

the amplitude and lifetime of the 0 intermediate between 25°Cand 5°C. The amplitude of the 0 intermediate in the Leu-93-- Ala mutant increased slightly with a decrease in tempera-ture (Fig. 4A), in marked contrast to the decrease observed inthe case of wild-type bacteriorhodopsin (18). An Arrheniusplot of the rate of decay of the 0 intermediate showed that theactivation energy for the transition from the 0 intermediate tothe initial bacteriorhodopsin state was -77 kJ/mol (Fig. 4B),which is '18 kJ/mol higher than that observed for wild-typebacteriorhodopsin (19). The linearity of the Arrhenius plotalso demonstrates that no change in heat capacity occursduring this transition. These measurements show that thelong-lived 0 intermediate observed in the photocycle of theLeu-93 -+ Ala mutant is different from the twisted, all-trans 0

intermediate observed in wild-type bacteriorhodopsin (11).

DISCUSSIONThe above experiments demonstrate that the Leu-93 -* Ala

mutation delays the photocycle at a stage following initialretinal isomerization, proton release, and proton uptake butbefore retinal reisomerization. The effect of the Leu-93 -> Ala

mutation is reversed by light-induced retinal reisomerization.Mutagenesis studies have shown that replacement of any otherresidue in contact with retinal, including those in the generalvicinity of the C-13=C-14 double bond (Thr-46, Tyr-57,Asp-85, Thr-89, Tyr-185, and Asp-212), can have profoundeffects on light and dark adaptation and increase the fractionof 13-cis retinal present in the light-adapted state of bacterio-rhodopsin (20-25). However, the effects of these substitutionsare to change the kinetics of the steps associated with protontransport in the photocycle or to reduce the efficiency ofproton transport under steady-state illumination (9, 10).

Leu-93 is located in close proximity to the methyl grouplocated on C-13 of retinal (ref. 26; Fig. 5), a key region thatundergoes changes in structure following light absorption(27-29). Comparison of the effects of replacing Leu-93 withalanine vs. threonine or valine provides a further insight intothe mechanism of protein-driven retinal reisomerization (refs.12 and 17; J.K.D. and S.S., unpublished data). As in the caseof the Leu-93 -> Ala mutant, the photocycle of the Leu-93 ->Thr mutant displays long-lived N and 0 intermediates (v 1.2

A

B

l N.

11122 Biophysics: Delaney et aL

Dow

nloa

ded

by g

uest

on

May

30,

202

1

Proc. Natl. Acad. Sci. USA 92 (1995) 11123

A/A:j0>t-o X, --

RETINAL

FIG. 5. Expanded view of the atomic structural model for bacte-riorhodopsin (25) showing van der Waals surfaces for all-trans retinal,Leu-93 (third transmembrane helix), and Val-49 (second transmem-brane helix) side chains in the initial state of bacteriorhodopsin. Thearrow identifies the 13-methyl group on retinal. Light absorption leadsto the isomerization of retinal at the C-13-C-14 double bond. Theisomerization is thought to result in an upward motion of retinal nearthe C-13==C-14 bond, leading to the displacement of the C-13 methylgroup toward Leu-93 and to subsequent changes in protein tertiarystructure. Completion of the photocycle involves coupling of proteinstructural changes to the reisomerization of retinal to the startingall-trans state. The results presented here suggest that the surfacecontact of Leu-93 and the 13-methyl group is important for thiscoupling.

sec) whose formation coincides with proton uptake and isaccelerated in the presence of background illumination. Incontrast, the photocycle of the Leu-93 -> Val mutant is onlyslightly (1.7-fold) slower than in wild-type bacteriorhodopsin.Since the side-chain volumes (30) of alanine (91.5 A3) andthreonine (122.1 A3) are smaller than those ofvaline (141.7 A3)and leucine (167.9 A3), we infer that the steric contact of theC-13 methyl group of retinal with a bulky nonpolar side chainat residue 93 is important for the rapid completion of step IIIas monitored by the decay of the 0 intermediate. The increasein the activation barrier for the decay of the 0 intermediate isalso consistent with a role for Leu-93 in retinal reisomeriza-tion. We conclude that the van der Waals interactions betweenretinal and the side chain at residue 93 play a key structural rolein the course of the photocycle.

Val-49 is the only other residue in the retinal binding pocketwhose replacement (by alanine but not by leucine) has beenshown to prolong the lifetime of the N/O intermediates (by-10-fold; refs. 13 and 31). As in the case of Leu-93 -- Ala

mutant, the photocycling time in the Val-49 -> Ala mutant was

accelerated from 150 msec to 70 msec in the presence ofbackground illumination (J.K.D. and S.S., unpublished data).Val-49 is located within van der Waals distance of Leu-93 (Fig.5), and the replacement of Val-49 by alanine is expected toresult in a looser packing of the Leu-93 side chain. Together,these observations further support the hypothesis that thecontact between retinal and Leu-93 regulates the rate of retinalreisomerization.The mechanism proposed above implies that the van der

Waals contact between retinal and Leu-93 is likely to bedifferent in the late stages of the photocycle as compared withthe early stages. Spectroscopic studies (32-34) and diffractionexperiments (35, 36) provide strong evidence for proteinstructural changes in the late stages (i.e., II and III) of thephotocycle. Indirect evidence for structural changes in thevicinity of the Leu-93 side chain comes from linear dichroism(37) and neutron diffraction studies (38), which show that theangle of the C-S to C-13 portion of the polyene chain isincreased by 110 60 with respect to the membrane normal in

the M intermediate. Given the steric constraints at the ,B-ion-one ring (i.e, the C-S end) of the chromophore, this pivotingimplies an upward motion of retinal near the C-13=C-14double bond, leading to the displacement of the C-13 methylgroup toward Leu-93. Hence, the interaction of Leu-93 withthe 13-methyl group is likely to play a central role in the returnof retinal to the starting all-trans configuration. These findingssuggest that the Leu-93 side chain is involved in retinalreisomerization by providing a point of contact at which theprotein may "push" on retinal. Whether the contact betweenthe Leu-93 side chain and retinal is only confined to the C-13methyl group or whether it involves other neighboring sites onretinal remains to be determined.

It is instructive to compare the long-lived N and 0 inter-mediates in the photocycle of the Leu-93 -- Ala mutant withcorresponding intermediates in the photocycle of wild-typebacteriorhodopsin. A variety of visible and vibrational spec-troscopic analyses (9, 11) suggest that under physiologicalconditions, the sequence of intermediates in the late stages ofthe wild-type photocycle can be described as follows:

M ± N-1 No 0 > bR13-cis 13-cis 13-cis twisted all-trans

all-trans

in which proton uptake occurs during the transition from theN-1 state to the NO state, and the transition from NO to bR(bacteriorhodopsin) involves completion of retinal reisomer-ization. If the same photocycle model described the photocycleof the Leu-93 -* Ala mutant, then one possible scenario is thatthe observation of a long-lived 13-cis intermediate in thephotocycle is due to contributions from the N intermediatethat exists in equilibrium with a twisted all-trans 0 interme-diate. If so, the presence of a long-lived 0 intermediate in theLeu-93 -> Ala mutant must be due to a defect in stepsassociated with the conversion of retinal from the twistedall-trans state to the initial all-trans state. An alternative kineticmodel for the photocycle of the mutant, consistent with thedata presented here is as follows:

M ±N No± 0 -° bR13-cis 13-cis 13-cis 13-cis all-trans

in which the 0 intermediate that accumulates in the Leu-93 -Ala mutant contains retinal in the 13-cis state. If this modelcorrectly describes the photocycle, the presence of the long-lived 0 intermediate in the mutant must reflect a defect inisomerization of retinal from the 13-cis to either twistedall-trans or the all-trans state. The data presented here do notallow us to distinguish unequivocally between these two mod-els. However, they strongly suggest that the latter model bestdescribes the photocycle of the Leu-93 -- Ala mutant. It isimportant to note that the accumulation of a 13-cis 0 inter-mediate in the photocycle of the Leu-93 -> Ala mutant doesnot exclude the presence of an additional all-trans 0 interme-diate. Interestingly, the presence of a 13-cis 0 intermediate inthe wild-type photocycle has also been postulated (6). How-ever, the conclusion that the van der Waals contact of Leu-93with retinal is important in reisomerization is independent ofwhich one of these scenarios best describes the photocycle orwhether the interaction between retinal and Leu-93 is ulti-mately important for other protein conformational changesthat occur during the course of decay of the 0 intermediate.[We note that the Leu-93 -* Ala mutation does not alter theM ¢ N--± N0 equilibrium, since at higher pH values, protonuptake is biphasic as observed for wild-type bacteriorhodop-sin, and a slowly decaying component of the M intermediateis observed (J.K.D. and S.S., unpublished data).]

It is important to qualify the above discussion with thecaveat that assigning detailed retinal configuration/protein

Biophysics: Delaney et al.

Dow

nloa

ded

by g

uest

on

May

30,

202

1

Proc. Natl. Acad. Sci. USA 92 (1995)

states observed in the wild-type photocycle to specific opticalintermediates in the photocycles of a site-specific mutant is notalways reliable. In the case of the Asp-96 -* Asn mutant, thelong-lived M-like state has a Amax at 405 nm but has an N-likeprotein conformation as determined by vibrational spectros-copy (39). Thus, the presence of an intermediate with ared-shifted Ama does not necessarily imply that it has the sameprotein conformation and retinal configuration as that ob-served for the corresponding intermediate in the photocycle ofwild-type bacteriorhodopsin. If indeed there is a 13-cis 0intermediate in the photocycle of the Leu-93 -> Ala mutant,as suggested here, it challenges the assumption that the redshift of the Am. upon transition of N to 0 is primarily due tothe isomerization of retinal from the 13-cis to the all-transstate.The identification of a mutant in which an intermediate can

be accumulated at a stage in the photocycle after protonuptake provides a powerful tool to investigate light-drivenstructural changes in bacteriorhodopsin. Structural studieswith the Asp-96 -> Gly mutant in which the photocycle isslowed down between the proton release and uptake steps (i.e.,stages I and II) have already identified significant structuralchanges in helices F and G following light-driven protonrelease (36). The results presented here also provide a struc-tural basis to understand previous findings on the importanceof the methyl groups on retinal in the function of differentrhodopsins. When bacteriorhodopsin is regenerated with aretinal analog lacking the 13-methyl group, the ratios of 13-cisand all-trans retinal in the dark-adapted state are greatlyaltered, and the photocycles associated with both isomer formsdisplay long-lived intermediates (40). Similar experiments withsensory rhodopsin I (41) and bovine rhodopsin (42) suggestthat specific methyl groups of retinal act as a "steric trigger"in light-driven activation. Thus, sensory rhodopsin I-mediatedphototaxis in Halobacterium salinarium is not detected whenretinal is exchanged for 13-desmethylretinal. Regeneration ofvertebrate rhodopsin with a retinal analog lacking the 9-methylgroup impairs formation of metarhodopsin II, the key inter-mediate required for activation of guanine nucleotide-bindingproteins in the photoreceptor cell. High-resolution structuresare not currently available for any other bacterial or visualrhodopsin. Our results suggest that the critical role of retinalmethyl groups in light transduction is because they are directlyinvolved in van der Waals contacts with regions of the proteinundergoing structural changes. Such interactions may alsorepresent a general theme for protein-ligand coupling duringsteps in signal transduction by other seven-helix membranereceptors.

We thank Dr. T. Woolf for assistance with molecular graphics, Dr.J. Lanyi for generously providing the Leu-93 -> Thr and Val-49 -> Alamutants, and Dr. D. Oesterhelt for helpful comments. This work wassupported by grants from the National Eye Institute and the SearleScholars Program/Chicago Community Trust (to S.S.) and by aNational Eye Institute National Research Service Award fellowship(to J.K.D.).

1. Stoeckenius, W., Lozier, R. H. & Bogomolni, R. A. (1979)Biochim. Biophys. Acta 505, 215-278.

2. Stryer, L. (1986) Annu. Rev. Neurosci. 9, 87-119.3. Wald, G. (1968) Nature (London) 219, 800-807.4. Hamdorf, K. (1979) in Handbook of Sensory Physiology, ed.

Autrum, H. (Springer, Berlin), Vol. 7/6A, pp. 145-224.5. Lozier, R. H., Bogomolni, R. A. & Stoeckenius, W. (1975)

Biophys J. 15, 955-963.6. Ebrey, T. G. (1993) in Thermodynamics of Membrane Receptors

and Channels, ed. Jackson, M. (CRC, Boca Raton, FL), pp.353-387.

7. Grzesiek, S. & Dencher, N. (1986) FEBS Lett. 208, 337-342.8. Fodor, S. P., Ames, J. B., Gebhard, R., van den Berg, E. M.,

Stoeckenius, W., Lugtenburg, J. & Mathies, R. A. (1990) Bio-chemistry 27, 7097-7101.

9. Lanyi, J. K. (1993) Biochim. Biophys. Acta 1183, 241-261.10. Krebs, M. P. & Khorana, H. G. (1993) J. Bacteriol. 175, 1555-

1560.11. Smith, S. O., Pardoen, J. A., Mulder, P. P. J., Curry, B., Lugten-

burg, J. & Mathies, R. (1983) Biochemistry 22, 6141-6148.12. Subramaniam, S., Greenhalgh, D. A., Rath, P., Rothschild, K. J.

& Khorana, H. G. (1991) Proc. Natl. Acad. Sci. USA 88, 6873-6877.

13. Greenhalgh, D. A., Farrens, D. L., Subramaniam, S. & Khorana,H. G. (1993) J. Biol. Chem. 268, 20305-20311.

14. Ferrando, E., Schweiger, U. & Oesterhelt, D. (1993) Gene 125,41-47.

15. Oesterhelt, D. & Stoeckenius, W. (1974) Methods Enzymol. 31,667-678.

16. Scherrer, P., Mathew, M. K., Sperling, W. & Stoeckenius, W.(1989) Biochemistry 28, 829-834.

17. Cao, Y., Brown, L. S., Needleman, R. & Lanyi, J. K. (1993)Biochemistry 32, 10239-10248.

18. Varo, G. & Lanyi, J. K. (1991) Biochemistry 30, 5016-5022.19. Varo, G. & Lanyi, J. K. (1991) Biochemistry 30, 7165-7171.20. Marti, T., Otto, H., Mogi, T., Rosselet, S., Heyn, M. P. &

Khorana, H. G. (1991) J. Biol. Chem. 266, 6919-6927.21. Needleman, R., Chang, M., Ni, B., Varo, G., Fornes, J., White,

S. H. & Lanyi, J. K. (1991) J. Biol. Chem. 266, 11478-11484.22. Rothschild, K. J., He, Y.-W., Sonar S., Marti, T. & Khorana,

H. G. (1992) J. Biol. Chem. 267, 1615-1622.23. Sonar, S., Krebs, M. P., Khorana, H. G. & Rothschild, K. J.

(1993) Biochemistry 32, 2263-2271.24. Govindjee, R., Kono, M., Balashov, S., Imasheva, E., Sheves, M.

& Ebrey, T. G. (1995) Biochemistry 34, 4828-4838.25. Balashov, S. P., Govindjee, R., Kono, M., Imasheva, E., Luka-

shev, E., Ebrey, T., Crouch, R. K., Menick, D. R. & Feng, Y.(1993) Biochemistry 32, 10331-10343.

26. Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beck-mann, E. & Downing, K. H. (1990) J. Mol. Biol. 213, 899-929.

27. Schulten, K. & Tavan, P. (1978) Nature (London) 272, 85-86.28. Tsuda, M., Glaccum, M. & Ebrey, T. G. (1980) Nature (London)

287, 351-353.29. Gartner, W., Towner, P., Hopf, H. & Oesterhelt, D. (1983)

Biochemistry 22, 2637-2644.30. Creighton, T. (1984) Proteins (Freeman, New York), p. 242.31. Brown, L. S., Gat, Y., Sheves, M., Yamazaki, Y., Maeda, A.,

Needleman, R. & Lanyi, J. K. (1994) Biochemistry 33, 12001-12011.

32. Rothschild, K. J., Marti, T., Sonar, S., He, Y.-W., Rath, P.,Fischer, W. & Khorana, H. G. (1993) J. Biol. Chem. 268, 27046-27052.

33. Souvignier, G. & Gerwert, K. (1992) Biophys. J. 63, 1393-1405.34. Steinhoff, H.-J., Mollaaghababa, R., Altenbach, C., Hideg, K.,

Krebs, M., Khorana, H. G. & Hubbell, W. (1994) Science 266,105-107.

35. Koch, M. H. J., Dencher, N., Oesterhelt, D., Plohn, H.-J., Rapp,G. & Buldt, G. (1991) EMBO J. 10, 521-526.

36. Subramaniam, S., Gerstein, M., Oesterhelt, D. & Henderson, R.(1993) EMBO J. 12, 1-8.

37. Heyn, M. & Otto, H. (1992) Photochem. Photobiol. 56, 1105-1112.

38. Hauss, T., Buldt, G., Heyn, M. P. & Dencher, N. A. (1994) Proc.Natl. Acad. Sci. USA 91, 11854-11858.

39. Sasaki, J., Schichida, Y., Lanyi, J. K. & Maeda, A. (1992) J. Biol.Chem. 267, 20782-20786.

40. Gartner, W., Oesterhelt, D., Vogel, J., Maurer, R. & Schneider,S. (1988) Biochemistry 22, 3497-3502.

41. Yan, B., Nakanishi, K. & Spudich, J. L. (1991) Proc. Natl. Acad.Sci. USA 88, 9412-9416.

42. Ganter, U. M., Schmid, E. D., Perez-Sala, D., Rando, R. R. &Siebert, F. (1989) Biochemistry 28, 5954-5962.

11124 Biophysics: Delaney et al.

Dow

nloa

ded

by g

uest

on

May

30,

202

1