Proc. Nad. Acad. Sci. USAVol. 88, pp. 10402-10406, December 1991Biochemistry

Disulfide cross-linking studies of the transmembrane regions of theaspartate sensory receptor of Escherichia coliBERKLEY A. LYNCH* AND D. E. KOSHLAND, JR.Department of Molecular and Cell Biology, 401 Barker Hall, University of California, Berkeley, CA 94720

Contributed by D. E. Koshland, Jr., August 27, 1991

ABSTRACT The Escherichia coli aspartate receptor, adimer of identical subunits, has two transmembrane regions(TM1, residues 7-30; TM2, residues 189-212) of 24 residueseach. To study the relative placement and orientation of theregions, cysteine residues were introduced individually into thecenter of each: at positions 17, 18, and 19 in TM1; and atpositions 198, 199, 200, and 201 in TM2. Based on the patternsof disulfide cross-linking observed between subunits in themutant receptors, there appears to be close contact between theTM1 and TM1' regions at the dimer interface but no suchdirect interaction between the TM2 and TM2' regions. Thecross-linking results are consistent with an a-helical structureextending across the transmembrane region up through at leastresidue 36, which lies on the periplasmic side of TM1. Theability ofan 18-18' cross-linked dimer to transmit an aspartate-induced transmembrane signal is also supportive of such anextended helix. The changes in relative rates of disulfidecross-linking provide experimental evidence of a conforma-tional change transmitted through the transmembrane domainduring signaling. Once formed, disulfides between the trans-membrane regions are unusually resistant to reduction by lowmolecular weight thiols in the presence of denaturants like SDS.These targeted disulfide cross-links can be used to revealstructural and dynamic aspects of protein function.

-pI la '

I:,.J

FIG. 1. Simple model of the aspartate receptor of chemotaxis(adapted from ref. 9). This model shows the dimeric nature of thereceptor as well as the presence of four transmembrane regions(TM1, TM1', TM2, and TM2') in the transmembrane domain.Aspartate binds to the periplasmic domain, and a signal is transmittedacross the membrane to the cytoplasmic domain.

Many transmembrane receptors have now been identified,cloned, and sequenced, but the mechanisms of transmem-brane signaling remain obscure. Structurally, transmem-brane receptors fall into at least three categories. One classof receptors contains ion channels, which transmit a signal bythe passage of ions (1). A second class, which contains seventransmembrane regions, is exemplified by rhodopsin and the,3-adrenergic receptors, which function by interacting withGTP binding proteins (2). The third class of receptors con-tains one or two highly hydrophobic transmembrane se-quences of the type found in the epidermal growth factorreceptor (3), the insulin receptor (4), and the aspartate andserine receptors of chemotaxis (5, 6). Transmembrane re-ceptors containing one or two transmembrane sequenceshave been postulated to operate by either an association/dissociation reaction, as suggested for the epidermal growthfactor receptor (7), or by a ligand-induced conformationalchange through an individual subunit, as suggested for theaspartate receptor (8-10).To investigate the properties of the aspartate receptor, a

new tool, targeted disulfide cross-linking, was developed (9).Using site-specific mutagenesis, cysteine residues were in-troduced at various locations throughout the aspartate re-ceptor. The rates of formation of disulfide bonds betweenmonomers revealed properties of the juxtaposition and flex-ibility ofthe protein domains. A rate is a probability function,and therefore the rate of cross-linking is related to theproximity of the residues, the flexibility of the protein, and

the environment around the sulfhydryl group, which allow itsreaction. This tool can be used in addition to other biophys-ical and chemical methods to provide information aboutprotein structure and function.

In the case of the aspartate receptor (Fig. 1), disulfidecross-linking studies revealed that no dimer association/dissociation was needed for transmembrane signaling, andthat cross-linking at the 36-36' intersubunit site produced aprotein that was fully active (9). Furthermore, cross-linkingbetween other positions, such as 106-106' or 3-3', producedinactive proteins, indicating that protein motion allowed theoccasional movement of the two cysteine side chains closeenough to permit disulfide formation, even though the disul-fide distorted the protein sufficiently to destroy its activity.The disulfide cross-links described above were obtained by

introducing cysteine residues in the nonmembrane domainsof the receptor molecule. In this paper, we extend thedisulfide cross-linking methodology to the transmembraneregions of the receptor, where it is used to study their relativeorientation in the membrane, and properties of the signalingstate.

MATERIALS AND METHODSStrains, Plasmids, Mutagenesis, Expression, and Extrac-

tion. The Escherichia coli strains RP4080 (lacking cheRmethylesterase) and RP5838 (lacking all chemotaxis recep-tors) were obtained from J. S. Parkinson (University ofUtah). Site-specific mutagenesis was performed by standard

Abbreviations: DTT, dithiothreitol; 2-ME, 2-mercaptoethanol.

10402

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

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Proc. Natl. Acad. Sci. USA 88 (1991) 10403

methods (11). Mutations were introduced into the plasmidpMK650 containing the E. coli receptor. Mutagenesis wasconfirmed by dideoxynucleotide sequencing. The TarE wild-type and mutated receptors were expressed in the strainRP4080, and membranes containing receptors were preparedas reported (12) in the presence of2 mM dithiothreitol (DTT).Disulfide bonds between two cysteines at identical sequencepositions on adjacent subunits are denoted as n-n', andinteractions between two equivalent regions on differentsubunits in the dimer are referred to as interactions betweenregions a and a' (for instance TM1 and TM1').

Tryptone and Minimal Swarm Plates. Plasmids containingthe wild-type receptor (pMK650), the mutant receptor (asindicated), or no receptor (pEMBL19) were transformed intothe E. coli chemotaxis strain RP5838, which lacks all chemo-taxis receptors. Chemotaxis was measured by swarm assaysin either tryptone soft agar plates or minimal soft agar platesplus or minus 100 AM aspartate. Swarm rates are reported inTable 1 as mm/hr for tryptone plates and as the ratio betweenthe rate in mm/hr plus aspartate divided by the rate in mm/hrminus aspartate for the aspartate swarm plates.

Cross-Linking Rates. Receptors were methylated in themembranes with S-[3H]adenosylmethionine in the presenceof methyltransferase and extracted by adding 1% TritonX-100 detergent (Pierce) to the membranes for 30 min at 4°C.Cross-linking was initiated by the addition of 1.5 mM Cu(II)(1,10-phenanthroline)3 to receptor in a detergent-solubilizedbuffer system containing bacterial phospholipids and TritonX-100, because of its similarity to a system in which the E.coli aspartate receptor undergoes aspartate-stimulated in-creases in methylation rates (M. Shapiro, personal commu-nication). At the indicated times after initiation of the reac-tion, aliquots were added to Laemmli sample buffer contain-ing EDTA and N-ethylmaleimide as described (9) and frozenin liquid N2. Monomeric and cross-linked dimeric forms ofthe receptor were well separated on SDS/7.5% polyacryl-amide gels. Bands were cut out of the gels and analyzed for[3H]methyl ester incorporation by scintillation counting. Ini-tial rates of cross-linking were determined by using the slopeofthe line through the time points in the first 5 min, where therates are linear, and are reported as fraction of total receptorcross-linked per min (Table 1, Fig. 2A).

Table 1. Properties of receptors with targeted cysteinesubstitutions

TryptoneSequence Minimal swarm Cross-link Cross-linkposition of swarm rate, rate ratemutant rate ratios mm/hr (+Asp) (-Asp)

17 2.0 1.30 <0.001 <0.00118 2.6 0.68 0.038 0.01119 1.8 0.89 0.026 0.008

198 2.2 0.88 <0.001 <0.001199 2.9 1.09 <0.001 <0.001200 2.3 0.94 <0.001 <0.001201 2.6 0.90 <0.001 <0.001pMK650

(wild type) 2.4 0.98 <0.001 <0.001pEMBL19

(receptornegative) 0.7 0.17 ND ND

Minimal swarm rate ratios are the ratio between the swarm ratesof the bacteria in soft agar minimal plates with 100 1LM aspartate tothe same plates without aspartate. Tryptone swarm rates weredetermined on soft agar tryptone plates at 30°C. Cross-linking rateswere determined in an in vitro detergent-solubilized system and aregiven as a fraction of cross-linked material formed per min {[(cross-linked)/(cross-linked + reduced)]/min}. ND, not done.

0.3-

o" 0.25-

o 0.2.

$_4u

0 0.15-

0.05

0.05.

0 1 2 3 4 5 6

-0.7

-0.6 ,,.06

-0.5 E0

0.40

-0.2

n0 5 10 15 20

Time (Minutes)

FIG. 2. Rate studies on mutant receptors. (A) Cross-linking ofF18C in a detergent-solubilized system in the presence (O) andabsence (A) of 1 mM aspartate and of A19C in the presence (e) andabsence (x) of 1 mM aspartate. Aspartate causes an increase incross-linking rates for both mutant receptors. (B) Reduction ofcross-linked F18C in SDS at 90°C by 100 mM 2-ME in standardLaemmli sample buffer (o) and in 100mM DTT at pH 9.0 (A). Degreeof cross-linking or reduction is reported on the y axis as fractioncross-linked or fraction reduced {[dpm cross-linked (or reduced)]/[total dpm in cross-linked plus reduced fractions]}.

Reduction of Cross-Linked Receptors. Receptors were

methylated with S-[3H]adenosylmethionine and cross-linkedas described above. Cross-linked receptors were passedthrough a spin column, exchanging buffers and removingoxidant at the same time. The receptor was denatured in thepresence of 1% SDS and reduced with either 100 mM2-mercaptoethanol (2-ME) at pH 6.8 or 100 mM DTT at pH9.0 at 90°C for the times indicated. The different forms of thereceptor were separated on SDS/7.5% polyacrylamide gelsand incorporated radioactivity was determined (Fig. 2B).

Analysis of the Transmembrane Regions. The transmem-brane regions of the aspartate receptor were analyzed forhydrophilicity and conservation of residues, following theapproach developed from an analysis of the transmembranedomain of the Rhodopseudomonas viridis photosyntheticreaction (13). The degree of conservation of residues in thetransmembrane helices was determined by comparing thesequence ofTarE with the homologous receptors TarS, TsrE,and TapE. The average degree of conservation for an entiretransmembrane sequence is given by the value (V), which isderived by averaging the number of different amino acidsfound at each position in the transmembrane regions amongthe four receptors (using the four receptors listed above, anyresidue position is assigned a number from one to four,depending on how many different amino acids are found atthat position). A vector was derived representing the direc-tion and degree of greatest evolutionary conservation ofresidue positions perpendicular to the long axis of the trans-membrane helices (Fig. 3). Hydrophilicity was analyzed byaveraging the hydrophobicity at each position from the fourrelated receptors mentioned above, utilizing a standard hy-drophobicity scale for the amino acids (14), and generating a

vector representing the direction and degree of greatesthydrophilicity (Fig. 3). The average hydrophobicity for thesequence averaged over the four homologous receptors isrepresented by the value (H) (13).

RESULTS

Cross-Linking and Reduction of the Mutant Receptors. Thecross-linking patterns of the mutant receptors provide topo-logical information about the relative placement and orien-

A: Crosslinking 0 B: Reduction 0

F18C+Asp °O O.M DTI *

*A19C+Asp

A0 @~~~~~~00

0 SO~ ~ ~ ~~~A

F18C* 1% If 0" O.1M 2-ME

9C X A19C AAI AI

Biochemistry: Lynch and Koshland

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

10404 Biochemistry: Lynch and Koshland

189Phe

19Ws

210Tyr /

199 Leu

206 Leu

Ala195

I / Val 12

9 F;r- Gly 13016 23

Ile 197

Aa 190

Ala208

201

Val I Leu 194202 191 Gin Leu Ile

Ala205 212

209 19

TM1 TM2FIG. 3. Helical wheel diagrams ofTM1 and TM2 of the E. coli aspartate receptor. Arrows represent a vector moment indicating the direction

and degree of maximum conservation (--- -) and hydrophilicity (-). Residues conserved between TarE and TarS, TsrE, and Tap are shownboxed. Residues underlined are conserved in TarE, TarS, and Tsr.

tation of the transmembrane regions. In the absence ofdetergent, none of the cysteine-containing receptors showedany significant cross-linking by Cu(II) (1,10-phenanthroline)3in crude membranes. Only two of the seven mutant receptorsisolated, F18C and A19C in TM1, could be oxidized tocross-linked dimers in the Triton X-100 buffer system (Table1 and Fig. 2A). The mutant receptors V17C, A198C, L199C,V200C, and V201C did not cross-link to any significant extentin this system. Both the A19C mutant receptor and the F18Cmutant receptor had similar cross-linking rates, althoughA19C had somewhat lower rates than F18C, and both had an

%3-fold greater cross-linking rate in the presence of aspartatethan in its absence (Fig. 2A; Table 1). The rates of 18-18' and19-19' cross-link formation in the presence of 1 mM aspartateare similar to the rate of 106-106' formation in TarS, whichhad the lowest cross-linking rate of the nontransmembranedomain cysteine mutants previously studied (9).

Disulfides are normally readily reduced by an excess oflowmolecular weight thiol reductant such as 2-ME or DTT in thepresence ofa denaturing agent such as SDS. Treatment ofthe18-18' cross-linked receptor in a normal Laemmli samplebuffer with 100 mM 2-ME at 90'C for 15 min resulted inconversion ofonly - 1/4 ofthe cross-linked dimer back to themonomer. Even increasing the pH to 9 with 100 mM DTT at90'C could only reduce half of the cross-linked dimers tomonomers after 15 min, although treatment with 100 mMDTT at pH 9 and 100'C for 30 min reduced the dimercompletely (data not shown). We also tested for the disulfidenature of the cross-links by treatment with performic acid.The performic acid cleaves the dimer. In addition, we alsotested cross-linked dimer created by using 12, which does notact through a free-radical mechanism, and the I2-generatedcross-linked dimer is as resistant to reduction by 2-ME as

cross-linked dimer generated by using Cu(II) (1,10-phenanthroline)3. These experiments support the identifica-tion of the cross-links as disulfides, although they do notanalytically prove it.

Functionality of Receptors. All of the bacteria containingmutant receptors swarm like wild type in the tryptone platesand show an acceleration in swarming induced by aspartateon minimal plates (Table 1). In addition, the cross-linked18-18' receptor shows an aspartate-stimulated increase in

methylation that is similar to that seen in wild-type receptorand in non-cross-linked F18C receptor (Fig. 4), while cross-

linked 19-19' receptor shows no such aspartate-stimulatedincrease in methylation (data not shown).

Analysis of Transmembrane Regions. The above cross-

linking results provide information on the orientation of thetransmembrane domains of the receptor. This is extended byan analysis of the sequences of these regions. In Fig. 3 arehelical wheel diagrams of the transmembrane regions of theaspartate receptor, with vectors representing the degree anddirection of greatest hydrophilicity (solid arrow) and conser-vation (dotted arrow) over the length of each transmembraneregion. TM1 has both vectors directed to one side of thetransmembrane helix between residues 18 and 19. TM2 alsohas both vectors oriented to one side of the helix betweenresidues 200 and 201, although the vector moments forconservation and hydrophilicity are smaller in TM2 than in

2

WVild-tVpu ReducedRecept(r FlI.S(C

A. rosslinked

.5z

0

FIG. 4. Methylation of receptors. Receptors were solubilizedfrom membranes with Triton X-100, cross-linked or not, put througha G-25 spin column to remove oxidant, and methylated for 5 min inthe presence or absence of aspartate. Samples were run on poly-acrylamide gels and bands representing either the monomer or dimerposition were excised and assayed. The first two bars are wild-typemonomer. The next two bars are F18C mutant monomer. The lasttwo bars are 18-18' cross-linked dimers. Results are presentednormalized to the methylation for each sample in the absence ofaspartate. +, Plus aspartate; -, minus aspartate. SD is no greaterthan ± 10% of the mean.

10 Leu28 X

17W k

24 Ile

13

20

Proc. Natl. Acad. Sci. USA 88 (1991)7

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

Proc. Natl. Acad. Sci. USA 88 (1991) 10405

TM1. The average hydrophobicity (H), averaged over theentire transmembrane sequence for all four receptors, is 0.30for TM1 and 0.37 for TM2; thus the average hydrophobicityofTM2 is greater than TM1. The average variability ofTM1(V) (14) is 2.08, whereas it is 2.71 for TM2. TM2 thus has ahigher average variability or a lower degree of residueconservation on average than TM1.

DISCUSSIONBy introducing cysteine residues in the transmembrane re-gions ofthe receptor, we are able to determine the orientationofthe helices in the transmembrane domain. There is growingevidence that the transmembrane regions of many integralmembrane proteins are a-helical in nature. The aspartatereceptor's transmembrane regions are hydrophobic stretches-24 residues long, which is of sufficient helical length to spana membrane bilayer of average thickness. In addition, theexperimentally determined topology of the domains of thereceptor is consistent with each transmembrane region pass-ing once through the membrane (15, 16). The cross-linkingdata indicate that only two of the seven introduced cysteineresidues, F18C and A19C in TM1, lie close enough in space,or are sufficiently exposed, or both, to form disulfides acrossthe dimer interface in a detergent-solubilized state. The factthat the disulfides can form in detergent but not in crudemembranes can be explained either by the existence of adifferent conformation in detergent than in membranes or byinhibition of oxidation in membranes. We favor the latterexplanation because of the overall slow rate of cross-linkingeven in detergent, and because both the TarS and TarEreceptors can have their signaling functions reconstituted indetergent plus phospholipid systems. The overall structure ofthe transmembrane regions deduced from the cross-linkingresults is similar to that proposed previously, based oncross-linking rates in cysteine residues flanking the receptortransmembrane regions, with TM1 and TM1' being close inspace and TM2 and TM2' being further apart (Fig. 1) (9).The results from transmembrane cross-linking experi-

ments allow a more detailed model of the interactions be-tween TM1 and TM1'. This model assumes the transmem-brane regions are a-helices. Interpretation of the cross-linking results is constrained by the properties of disulfidesand a-helices. The maximum distance found in proteinsbetween a-carbons linked by disulfide bonds is -7 A, withan average distance of 5-6 A (17). The range of distancesbetween the axes of two interacting helices is 6.8-12 A, witha mean distance of 9.4 A (18). Thus, the formation of adisulfide bond between two interacting helices withoutchanging the native structure of a protein requires that thecysteines be on facing sides of the two helices. Since both theF18C and the A19C mutant receptors cross-link in either thepresence or absence of aspartate, while the V17C mutantreceptor does not cross-link at all, TM1 and TM1' probablyinteract across the faces containing residues 18 and 19, and18' and 19' (Fig. 5). The slightly lower cross-linking rates ofA19C might indicate a longer 19-19' distance than an 18-18'distance. Since the relative cross-linking rates of F18C andA19C are dependent on several factors (proximity, orienta-tion, environment, and dynamics), we cannot assume thatcross-linking rates derive solely from cysteine-cysteine dis-tances. However, the cross-linked 18-18' receptor still sig-nals (Fig. 4) while the cross-linked 19-19' receptor does not.This additional fact, in conjunction with the faster 18-18'cross-linking rate, implies a closer juxtaposition of 18 to 18'than of 19 to 19' (Fig. 5). In contrast to the case in TM1, noneof the cysteines introduced in TM2 can cross-link in thedimer, implying that there is no direct TM2-TM2' interactionacross the dimer interface in the region studied.

F919FIG. 5. Model for the interactions ofTM1 and TM1' derived from

cross-linking studies. The F18C and A19C mutant receptors cross-link in the presence or absence of aspartate, indicating that thereceptor subunits interact across the TM1 helical face containingresidues 18 and 19. The fact that 18-18' cross-links somewhat fasterthan 19-19' combined with the fact that the 18-18' cross-linkedreceptor still signals, while the 19-19' cross-linked receptor does not,suggests that the 18-18' distance is somewhat smaller than the 19-19'distance, as shown.

The model for transmembrane helical interactions derivedfrom the cross-linking studies is supported by an analysis ofthe transmembrane sequences. The structure of the reactioncenter ofRp. viridis has been determined at atomic resolution(19). The transmembrane a-helices that constitute the trans-membrane domain resemble those in globular proteins inaqueous environments; that is, they are tightly packed to-gether in a stable assembly, with clear interior and exteriorfaces. Residues in the interior of the domain (on the innersurfaces of the transmembrane helices) have a net hydro-phobicity that is similar to the interior of a globular proteindomain, while the exterior, membrane-exposed residues ofthese helices are significantly more hydrophobic than thosein the interior of an average globular protein (13). Further-more, interior residues in the transmembrane domain aremore conserved among evolutionarily related proteins thanare exterior residues-a property shared with globular pro-tein domains.The transmembrane regions of the aspartate receptor were

examined by these principles, and the results agree with thecross-linking experiments (Fig. 3). TM1 has a helical face(that containing residues 18 and 19) with a high proportion ofthe most conserved and most hydrophilic residues in thetransmembrane region, although it is by no means a hydro-philic face. TM2 has a similar sidedness, but the vectormoments for these values are smaller than in TM1. TM2 alsohas a higher average hydrophobicity ((H)) than TM1 and amuch lower sequence conservation ((V)) with related recep-tors. The combination of a greater degree and vectorialorientation of hydrophilicity and conservation of residues forTM1 relative to TM2 implies that TM1 and TM1' are moreburied in the interior of the transmembrane domain, whileTM2 and TM2' are on the outside of the domain and moreexposed to the membrane environment. The analysis ofTM1suggests that the face toward the center of the transmem-brane domain contains residues 18 and 19. This is confirmedby the cross-linking studies and suggests that, by analogy, theface of TM2, which is on the interior of the transmembranedomain, is the one containing residues 189 and 190.The cross-linking data and sequence analysis not only help

us orient the transmembrane helices, but also provide evi-dence that there is a continuous a-helix from the cytoplasmicboundary of TM1 through residue 36 in the ligand-bindingdomain. As demonstrated previously, a disulfide cross-linkbetween residues 36 and 36' in TarS forms rapidly andpermits signaling (9). If the TM1 helix extends to the vicinity

Biochemistry: Lynch and Koshland

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

10406 Biochemistry: Lynch and Koshland

of residue 36, which is 6 residues outside of the putative TM1region, residues 36 and 18 would reside on the same face ofthis a-helix (Fig. 5). The ability of cross-links to formbetween both 18-18' and 36-36', and the ability of both ofthese cross-linked receptors to signal, suggests that thetransmembrane a-helix extends at least up through residue 36in the periplasm. The 19-19' cross-linked receptor does notsignal, suggesting that the cross-link in this case distorts thenative structure of the receptor.The cross-linking studies have important implications for

the mechanism of transmembrane signaling. In the presenceof 1 mM aspartate, both the F18C and A19C receptors showan =3-fold increase in cross-linking rates. One possibleexplanation is that there is a relative motion ofTM1 and TM1'when aspartate binds to the receptor (intersubunit relativemotion), bringing the cysteines closer together, or into a morefavorable orientation, so that disulfide formation is morerapid. To explain the similar changes observed in the cross-linking rates, any such motion must affect the 18-18' and19-19' distances to an equal degree. One way this restrictioncould be accommodated is for TM1 and TM1' to moverelative to one another along a vector normal to the plane oftheir interaction. Such a hypothesis seems unlikely, sincecross-links between TM1 and TM1' should constrain motionsof this sort, and yet the 18-18' and 36-36' cross-linkedreceptors can signal. A second possibility is that aspartatebinding causes a motion of TM2 relative to TM1 (motionwithin a subunit), increasing the exposure of the cysteinepairs to oxidant. Such intrasubunit motion could involve TM2moving away from TM1 in a motion within the plane of themembrane, the transmembrane regions moving up and downperpendicular to the plane ofthe membrane, or a combinationof both sorts of motion. If the relative orientation ofTM1 andTM1' remains fixed, while TM2 and TM2' move away fromthe TM1-TM1' interface, exposing the cysteine residues tooxidant, it would account for the symmetrical increase incross-linking in the presence of aspartate. A third possibilityis that the binding of aspartate has an effect on the dynamicsof the receptor. In the absence of aspartate the cysteinesulfhydryls might spend less time at the correct distance andorientation for forming disulfides. Aspartate binding mightdecrease the dynamic motion of the receptor, allowing thedisulfides to form more rapidly. A combination of thesealternatives might account for the changes in cross-linking.Further experiments are required to understand this effect.The transmembrane cysteines and the disulfides in the

aspartate receptor exhibit some interesting chemical proper-ties. The disulfides are very stable on exposure to lowmolecular weight thiols in the presence of SDS, requiringextremes of heat and pH for full reduction. The reducedcysteines are not labeled by N-ethylmaleimide, even in thepresence of denaturants (data not shown). In the case ofbacteriorhodopsin, labeling of introduced transmembranecysteines required denaturation of the protein in 3% SDS and6 M urea (20). A lack of reactivity might be a common

property of transmembrane cysteines, possibly resultingfrom a high pKa of the side-chain sulfhydryl in the hydro-phobic environment of the membrane.The application of the method of targeted disulfide cross-

linking to other receptors, both in transmembrane and otherregions, should prove useful for testing the importance ofoligomerization in the mechanisms of these receptors, forstabilizing the oligomerized state for crystallization attempts,and for locking the receptors in specific conformations-capturing "on" and "off' states, for example-to probe themechanisms of transmembrane signaling.

We acknowledge the generous support of the William M. KeckFoundation and the National Institutes of Health (Grant DK09765).B.A.L. is a Fellow of the Jane Coffin Childs Memorial Fund forMedical Research.

1. Krueger, B. K. (1989) FASEB J. 3, 1906-1914.2. O'Dowd, B. F., Lefkowitz, R. J. & Caron, M. G. (1989) Annu.

Rev. Neurosci. 12, 67-83.3. Ullrich, A., Coussens, L., Hayflick, J. S., Dull, T. J., Gray, A.,

Tam, A. W., Lee, J., Yarden, Y., Libermann, T. A., Schles-singer, J., Downward, J., Mayes, E. L. V., Whittle, N., Wa-terfield, M. D. & Seeburg, P. H. (1984) Nature (London) 309,418-425.

4. Ullrich, A., Bell, J. R., Chen, E. Y., Herrera, R., Petruzzelli,L. M., Dull, T. J., Gray, A., Coussens, L., Liao, Y.-C.,Tsubokawa, M., Mason, A., Seeburg, P. H., Grunfeld, C.,Rosen, 0. M. & Ramachandran, J. (1985) Nature (London) 313,756-761.

5. Russo, A. F. & Koshland, D. E., Jr. (1983) Science 220,1016-1020.

6. Boyd, A., Kendall, K. & Simon, M. I. (1983) Nature (London)301, 623-626.

7. Schlessinger, J. (1988) Trends Biochem. Sci. 13, 443-447.8. Mowbray, S. L. & Koshland, D. E., Jr. (1987) Cell 50, 171-180.9. Falke, J. J. & Koshland, D. E., Jr. (1987) Science 237, 1596-

1600.10. Milligan, D. L. & Koshland, D. E., Jr. (1988) J. Biol. Chem.

263, 6268-6275.11. Kunkel, T. A. (1985) Proc. Nati. Acad. Sci. USA 82, 488-492.12. Foster, D. L., Mowbray, S. L., Jap, B. K. & Koshland, D. E.,

Jr. (1985) J. Biol. Chem. 260, 11706-11710.13. Rees, D. C., DeAntonio, L. & Eisenberg, D. (1989) Science

245, 510-513.14. Eisenberg, D., Weiss, R. M., Terwilliger, T. C. & Wilcox, W.

(1982) Faraday Symp. Chem. Soc. 17, 109-115.15. Mowbray, S. L., Foster, D. L. & Koshland, D. E., Jr. (1985)

J. Biol. Chem. 260, 11711-11718.16. Falke, J. J., Dernburg, A. F., Sternberg, D. W., Jr., Zalkin,

N., Milligan, D. L. & Koshland, D. E., Jr. (1988) J. Biol.Chem. 263, 14850-14858.

17. Katz, B. A. & Kossiakoff, A. (1986) J. Biol. Chem. 261,15480-15485.

18. Chothia, C., Levitt, M. & Richardson, D. (1981) J. Mol. Biol.145, 215-250.

19. Deisenhofer, J., Epp, O., Miki, K., Huber, R. & Michel, H.(1985) Nature (London) 318, 618-624.

20. Altenbach, C., Flitsch, S. L., Khorana, H. G. & Hubbel,W. L. (1989) Biochemistry 28, 7806-7812.

Proc. Natl. Acad. Sci. USA 88 (1991)

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0