Engineering Protein-Lipid Interactions - American Chemical Society

---

4048

Langmuir 1995,11, 4048-4055

Engineering Protein-Lipid Interactions: Targeting of Histidine-TaggedProteins to Metal-Chelating Lipid Monolayers Kingman Ng, Daniel W. Pack, Darryl Y. Sasaki,+and Frances H. Arnold* Division of Chemistry and Chemical Engineering 210-41, California Institute of Technology, Pasadena, California 91125 Received May 9, 1995@ In an effort to devise simple and robust systems that can reproduce in synthetic membranes important features of biological targeting and surface assembly, a versatile system for targeting proteins t o lipid membranes has been developed. This system utilizes metal-chelating iminodiacetate (IDA)lipids loaded with divalent metal ions (Cu2+or Ni2+)to target proteins genetically modified with a poly(histidine)fusion peptide. The new pyrene-labeled iminodiacetate lipid 2 can be used for fluorescence imaging and spectroscopicstudies of lipid reorganization induced by protein binding and assembly on lipid membranes. Metal-chelating IDA lipids 1 and 2 target the soluble domain of cytochrome b5 to lipid assemblies by sharing the metal ion with a six-histidine sequence appended to the protein C-terminus. Protein binding to Langmuir monolayers containing the IDA-Cu2+ lipids 1and 2 is observed by monitoring increases in the monolayer area at a surface pressure high enough to block nonspecific protein insertion (25 mN1m). The His-tagged cytochrome b5 binds the Cu2+-loaded2 monolayer with high affinity (& < 50 nM). No binding is observed in the absence of metal ions or for cytochrome bg without the 6-His fusion peptide. Specific protein targeting t o the monolayer loaded with Ni2+is confirmed by fluorescence microscopy of fluorescein-labeled 6-His cytochrome b5. The poly(histidine) fusion peptide, widely used for recombinant protein purification, makes this targeting approach applicable to a large number of proteins.

Introduction Membrane targeting of soluble ligands accompanied by assembly of membrane components is a fundamental feature of cellular signal transduction1 and intercellular communication.2 Targeting followed by assembly into superstructures underlies a variety of other biological processes, ranging from assembly of the prothrombinase complex and blood coagulation3 to bi~mineralization~ and the formation of two-dimensional protein crystals on bacterial cell surface^.^ Protein or lipid components provide the interactions required for targeting and specific orientation of surface-bound molecules; the membrane's fluidity allows reorganization and sampling of intermolecular contacts required for assembly into more complex structures. A goal of our research is to devise simple and robust systems that can reproduce in syntheticmembranes important features of biological targeting and assembly. In addition to providing a model for significantly more complex biological structures, protein assemblies in synthetic membranes are of considerable use in the development of biosensors6 and biomolecular devices7and synthesis of novel organic and inorganic materiah4j8 Simply controlling protein orientation on a surface, for example, has been shown to be useful for AFM imaging,s

* To whom correspondence should be addressed. Telephone: (818) 395-4162. FAX: (818) 568-8743. Current address: Sandia National Laboratory, Albuquerque, New Mexico 87185-0368. Abstract published i n Advance ACSAbstracts, September 15, 1995. (1) Ullrich, A.; Schlessinger, J. Cell 1990, 61, 203-212. (2) Singer, S. J. Science 1992,255, 1671-1677. (3) Zwaal, R. F. A. Biochim. Biophys. Acta 1978, 515, 163-205. (4) (a) Mann, S.; Archibald, D. D.; Didymus, J. M.; Douglas, T.; Heywood, B. R.; Meldrum, F. C.; Reeves, N. J. Science 1993,261,12861292. (b) Heuer, A. H.; Fink, D. J.; Laraia, V. J.; Arias, J. L.; Calvert, P. D. ;Kendall, K.; Messing, G. L.; Blackwell, J.;Rieke, P. C.; Thompson, D. H.; Wheeler, A. P.; Veis, A.; Caplan, A. I. Science 1992,255, 10981105. (c) Heywood, B. R. Microscopy Res. Technol. 1994,27,376-388. (5) Baumeister, W.; Wildhaber, J.;Phipps, B. M. Can. J . Microbiol. 1989,35, 215-227. (6) Stelzle, M.; Weissmuller, G.; Sackmann, E. J . Phys. Chem. 1993, 97, 2971-2981.

'

@

0743-746319512411-4048$09.0010

stabilization of labile biomolecules,1° obtaining novel electronic and optical proper tie^,^ and influencing subsequent recognition processes. We are also interested in targeting proteins to lipid membranes with controlled orientation in order to promote protein crystallization at the interface. A variety of proteins not normally associated with membranes have been crystallized beneath Langmuir monolayers containing lipids derivatized with specific affinity ligands, including the B1 subunit of E . coli ribonucleotide reductase bound to a lipid-linked dATP,l2 streptavidin on biotinylated lipids,13and monoclonal anti-dinitrophenyl antibody bound to dinitrophenyl lipids.'* Furthermore, it has been shown recently that 3-D streptavidin crystals can be epitaxially grown directly from biotinylated lipid layers at very low protein concentration^.'^ A more general method for targeting proteins to lipid assemblies could extend this crystallization approach to a much wider range of proteins and make it significantly easier to implement. The targeting system we have chosen utilizes specific coordination interactions between metal ions and (1) (7) Samuelson, L. A.; Miller, P.; Galotti, D. M.; Marx, K. A,; Kumar, J.; Tripathy, S. K.; Kaplan, D. L. Langmuir 1992, 8 , 604-608. Samuelson, L. A.; Wiley, B.; Kaplan, D.; Sengupta, S.; Kamath, M.; Lim, J. 0.;Cazeca, M.; Kumar, J.; Marx, K. A.; Tripathy, S. K. J . Intell. Mater. Syst. Struct. 1994, 5, 305-310. (8) Ahlers, M.; Grainger, D. W.; Herron, J. N.; Lim, K.; Ringsdorf, H.; Salesse, C. Biophys. J . 1992, 63, 823-838. (9) Ill, C. R.; Keivens, V. M.; Hale, J. E.; Nakamura, K. K.; Jue, R. A,; Cheng, S.;Melcher, E. D.; Drake, B.; Smith, M. C. Biophys. J . 1993, 64. 919-924. (10) Shen, Y.; Safinya, C. R.; Liang, K. S.; Ruppert,A. F.; Rothschild, K. J. Nature 1993, 366, 48-50. (11) McLean, M. A.; Stayton, P. S.; Sligar, S. G. Anal. Chem. 1993, 65, 2676-2678. (12) Ribi, H. 0.;Reichard, P.; Kornberg, R. D. Biochemistry, 1987, 26, 7974-7979. (13) Darst, S. A,;Ahlers, M.; Kubalek, E. W.; Meller, P.; Blankenburg, R.; Ribi, H. 0.; Ringsdorf, H.; Kornberg, R. D. Biophys. J . 1991, 59, 387-39fi -- . -- -.

(14) Uzgiris, E. E.; Kornberg, R. A. Nature 1983, 301, 125-129. (15) (a) Hemming, S. A.; Bochkarev, A,; Darst, S. A,; Kornberg, R. D.; Ala, P.; Yang, D. S. C.; Edwards, A. M. J . Mol. Biol. 1996, 246, 308-316. (b) Edwards, A. M.; Darst, S. A.; Hemming, S. A.; Li, Y.; Kornberg, R. D. Nature (Struct. Biol.) 1994, 1, 195-197.

0 1995 American Chemical Society

Engineering Protein-Lipid Znteractions

Langmuir, Vol. 11, No. 10, 1995 4049

histidines naturally occurring on the surfaces of protein,16 (2) genetically engineered metal-binding sites,17 or as reported here, (3) a poly(histidine1 sequence genetically appended to the protein N- or C-terminus.ls Widely used for the purification of recombinant proteins by metalaffinity chromatography,lg versatile “histidine tags” (or other metal-coordinating peptide sequencesz0) offer a general approach to protein targeting and orientation on metal-complexing surface^,^ including lipid membranes and monolayers. Metal-chelating lipids designed for this purpose have been described recently by this laboratoryz1and by Tamp6 and co-workers.22 Kubelak and co-workersZ3have also recently reported the two-dimensional crystallization of His-tagged reverse transcriptase from HIV (HIV-RT) on a Ni2+-nitrilotriacetate (NTA) derivatized 1,2-dioleoyl3-sn-glycerophosphatidylethanolamine monolayer (Niz+NTA-DOPE). It was proposed that the Ni2+-NTAgroup is responsible for concentrating the protein a t the lipidwater interface and ultimately for the formation of twodimensional HIV-RTcrystals.z3 Our laboratory has shown that myoglobin (with five surface histidines) is targeted with moderate affinity (K, > lo6 M-l) to liposomes containing a Cu2+-iminodiacetate (IDA) lipid, lSz1 ESR spectroscopy of Cuz+-IDA liposomes in the presence of native myoglobin and myoglobin whose surface histidines had been derivatized with DEPC (diethylpyrocarbonate) demonstrated the primary role of the histidines in protein targeting via coordination to the membrane-bound metal ions.21 Lipid 1 (DSIDA) is a simple and robust “receptor” for protein targeting to synthetic membranes via metal coordination. In addition to the targeting function provided by the headgroup metal-IDA complex, however, it is also desirable to have the capability to directly probe molecular redistribution or assembly that accompanies protein binding. Thus a new metal-chelating IDA lipid, 2 (PSIDA), has been prepared in which a fluorescent pyrene label is incorporated in the lipid tail. The excimer

\ O -

B

and monomer fluorescence emission intensities of pyrene lipids are sensitive to the physical state of the monolayer and could be used for imaging monolayer assemblies by fluorescence microscopy as well as for monitoring such properties as the probe concentration, its lateral distribution within the membrane, or its rate oflateral These features of the lipid membrane are all sensitive to protein binding and reorganization of lipid components. (16)Mallik, S.;Plunkett, S. D.; Dhal, P. K.; Johnson, R. D.; Pack, D.; Shnek, D.; Arnold, F. H. New J . Chem. 1994,18, 299-304. (17)Arnold, F. H.; Haymore, B. L. Science 1991,252,1796-1797. Zhang, J.-H.; Arnold, F. H. Trends Biotechnol. 1994,12,189-192. (18)Hochuli, E.;Bannwarth, W.; Dobeli, H.; Gentz, R.; Stuber, D. BiolTechnology 1988,6,1321-1325. (19)Arnold, F. H. BiolTechnology 1991,9,151-156. (20)Smith, M. C.;Furman, T. C.; Ingolia, T. D.; Pidgeon, C. J.Biol. Chem. 1988.263.7211-7215. (21)Shnek, D. R.;Pack, D. W.; Sasaki, D. Y.; Arnold, F. H. Langmuir 1994,10,2382-2388. (22)Schmitt, L.; Dietrich, C.; Tampe, R. J . A m . Chem. SOC.1994, 116,8485-8491. (23)Kubalek, E.W.; Le Grice, S. F. J.; Brown, P. 0. J . Struct. B i d . 1994,113, 117-123. (24)Foster, T.Angew. Chem., Int. Ed. Engl. 1969,8, 333-343.

In the course of studying ligand-induced reorganization of the fluorescent lipid 2 in membrane assemblies, we observed that metal ion binding can strongly affect the ratio of the excimer to monomer emission intensities. A simple approach to metal ion detection and quantification has been developedusing this effect.z5The pyrene-labeled IDA lipid also serves as a membrane “receptor mimic” in that binding of external ligands can induce its reorganization in the membrane and change its fluorescence proper tie^.^^,^^ Here we show that a small globular protein, the soluble domain ofcytochrome b5,can be targeted with high affinity to PSIDA-containing lipid monolayers uia metal coordination to a six-histidine peptide appended to the protein’s C-terminus (6-His cyt b5). The poly(histidine) fusion peptide and metal-chelating IDA lipid comprise a n effective and versatile system for targeting proteins to lipid assemblies.

Results Lipid Synthesis. Fluorescent PSIDA lipid 2 was synthesized according to the multistep sequence outlined in Scheme 1. PSIDA features,a glycerol backbone connecting two ether-linked alkyl tails a t the 2 and 3 positions and a hydrophilic spacer to the iminodiacetic acid headgroup on the 1 position. The octadecyl trityl-protected glycerol was prepared from solketal and 941-pyrenyll14methylsulfonyl)nonane by a modified procedure from de Bony and TocanneZ7and Sunamoto et a1.z6 The two units were coupled by ether synthesis to produce 4 in good yield. Deprotection of the trityl group and etherification with l-(methylsulfonyl)-9-(triphenylcarbinyl)-3,6,9-trio~nonane (prepared from triethylene glycol) introduced the triethylene glycol spacer group to the main lipid body (6). Subsequent functional group manipulation a t the headgroup position gives the iminodiacetic acid functionality and lipid 2 (PSIDA). Characterizationof Metal-ChelatingLipid Monolayers. The surface pressure-area (n-A)isotherms of the chelating lipids are sensitive to changes in lipid packing caused by metal and protein binding. Figure 1 shows isotherms of pure DSIDA and PSIDA monolayers spread with and without Cuz+. The x-A isotherm of the unmetalated DSIDA monolayer (Figure 1, curve a) shows apparent biphasic behavior, with a liquid-like region a t low surface pressure, a “phase transition” region a t -3 mN/m, and a condensed phase a t high pressure. In contrast, DSIDA premetalated with CuZf exhibits a condensed phase a t all pressures (curve b). Inclusion of the bulky pyrene moiety in the lipid tail of PSIDA results in very different n-A behavior. The unmetalated monolayer (curve c) exhibits a relatively high compressibility. No apparent phase transition is observed, although a subtle decrease in compressibility is seen a t surface pressures near 20-25 mN/m. The PSIDAisotherm shifts to lower area a t all pressures in the presence of Cuz+. A previous study showed that a CuC12 concentration of 0.5pM in the subphase was sufficient to saturate the IDA headgroups with metal ions.z1 Premetalated lipid spread on a buffered subphase containing no additional metal ions (Figure 1) exhibits x-A behavior identical to the unmetalated lipid spread on a buffered subphase contain(25)Sasaki, D. Y.;Shnek, D. R.; Pack, D. W.; Arnold, F. H. Angew. Chem., Int. Ed. Engl. 1996,34,905-907. (26)Shnek, D. R.;Maloney, K. M.; Arnold, F. H. Manuscript in preparation. (27)de Bony, J.;Tocanne, J . F. Chem. Phys. Lipids 1983,32,105rn, LLL.

(28)Sunamoto, J.;Kondo, H.; Nomura, T.; Okamoto, H. J.Am. Chem. SOC.1980,102,1146-1152.

CH3(CHz)i70

~ ~ + o ~ ~ . r N ~ o z

@PHZ)’”

e

$02 0

/

Z=CH2CH3 8

\

(Z=H

2

-

(a)9-(l-pyrenyl)-l-(methylsulfonyl)nonane, KOH, DMSO,80 “C,12h. (b)TsOH, THFMeOH,rt, 13h. (c) NaH, 1-(methylsulfony1)9-(triphenylcarbinyl)-3,6,9-trioxanonane, THF, reflux, 10 h TsOH, THFMeOH,rt,5 h. (d) CBr4, PPh3, THF, rt, 12 h. (e)Diethyl iminodiacetate, NEt3, CH&NPTHF, reflux, 3 days. (0 NaOH, aqueous THFMeOH, reflux, 1h. a

f

40

t“‘. b i

30

H

2

20

u

10

0

4

0

50

100

0.0

,

2 I

150

Mean molecular area (A2/molecule)

Time (minutes)

Figure 1. Surface pressure-area isotherms of pure DSIDA lipid 1 without Cu2+(curve a) and with Cu2+(b) and PSIDA lipid 2 without Cu2+(c)and with Cu2+(d)on a buffered subphase (20 mM MOPS, pH 7.5, 100 mM NaC1, 25 “C).

Figure 2. Relative monolayer area increase W Aversus time at n = 25 mN/m for (1)wild-type cyt bs with PSIDA-Cu2+,(2) &His cyt bg with PSIDA, and (3) 6-His cyt b5 with PSIDACu2+. Protein concentration in the subphase is -0.2 pM.

ing 0.5 pM CuC12. This isotherm does not change over a 10 h period (data not shown). Thus, premetalated lipid spread on a subphase containing no additional divalent metal ions is believed to yield a monolayer fully loaded with Cu2+. Protein Binding to Pure PSIDA and DSIDA Monolayers. Qualitative features of protein binding to the lipid monolayer can be discerned from changes in the monolayer surface pressure vs area behavior. These studies were carried out to demonstrate specifictargeting via metal ion coordination and distinguish it from nonspecific protein insertion into the monolayer. Nonspecificprotein insertion, observed for a variety of proteinlipid systems a t low surface pressure^,^^^^^ can be minimized by increasing the surface pressure. To identify conditions that favor specific protein targeting, a pure, metal-free PSIDA monolayer was compressed and held a t constant pressure while the monolayer relaxed (15-30 min). When 6-His cyt bg was injected into the subphase 0.2 pM) and behind the barrier (final concentration mixed by circulation of the subphase, measurable increases in monolayer surface area due to nonspecific insertion were observed only at surface pressures less than 10 mN/ m. All subsequent protein binding experiments were carried out a t 25 mN/m, well above this threshold pressure.

Specificprotein binding involves formation of a ternary complex between the IDA lipid, the metal ion, and the poly(histidine) fusion peptide. To establish that protein binding occurs via this mechanism and not through other nonspecific interactions with the membrane, the monolayer area was measured a t constant pressure for three situations: (1)wild-type protein (no 6-His) with IDACu2+lipid, (2) 6-His protein with IDA lipid (no Cu2+),and (3) 6-His protein with the IDA-Cu2+ lipid. No monolayer expansion was observed when 6-His cyt b5 was injected below PSIDA that had not been metalated (curve 2 of Figure 2))while a large increase in area is seen for the premetalated monolayer (curve 3). Subsequent addition of a n excess of a strong metal-chelating agent, ethylenediaminetetraacetate (EDTA), to the expanded monolayer causes the area to decrease (data not shown). The area returns not to the starting value but to the area of the unmetalated PSIDA monolayer. Although 6-His cyt b5does not insert into unmetalated PSIDA a t 25 mN/m (see above),it is still possible that the monolayer expansion in the presence of Cu2+ (curve 3) is due to nonspecific interactions with the monolayer, as Cu2+ changes the nature of the PSIDA interface (see Figure 1). This possibility can be excluded, however, by comparing curve 3, for the histidine-tagged protein, and curve 1,for wildtype cyt b5 (no 6-His). Without the poly(histidine) fusion peptide, the protein does not expand the Cu2+-lipid monolayer.

-

(29) Phillips, M. C.; Sparks, C. E. Ann. N.Y. Acad. Sei. 1980,348, 122-137. (30) Peschke, J.; Mohwald, H.Colloids Surf. 1987,27,305-323.

Langmuir, Vol. 11, No. 10, 1995 405 1

Engineering Protein-Lipid Interactions 2 100

90 80 h

70

Y

-8 6 0

1

.z2

3

0.0

0

100

200

300

50

400

Time (minutes)

Figure 3. Relative monolayer area increase W Aversus time for (1)DSIDA-Cu2+ and (2) PSIDA-Cu2+monolayers at n = 25 mN/m with 6-His cytochrome bj. Protein concentration in the subphase is -0.2 pM. 20 mN/m

a

a

8

a.

30 mN/m 40

I

0

0.2

,

1

0.4

,

1

.

0.6

I

0.8

.

1

Mole fraction PSIDA

Figure 5. Mean molecular area versus mole fraction PSIDA in SOPC monolayer at different surface pressures. Subphase was 20 mM MOPS, 100 mM NaC1, pH 7.5, containing (A) no divalent metal ions or (B)10 pM CuC12. would be dispersed uniformly in the monolayer, a t least initially. The zwitterionic SOPC phosphatidylcholine headgroup is expected to show little or no interaction with Time (minutes) cytochrome b5 in high-salt buffer, while the oleoyl tail Figure 4. Relative area increase M A of a pure PSIDA-Cu2+ ensures monolayer fluidity a t all surface pressures (at monolayer at z = 25 mN/m versus time for varying 6-His cyt room t e m p e r a t ~ r e ) . Finally, ~~ PSIDA, both with and b5 concentration in the subphase: (1)50 nM, (2) 100 nM, (3) without chelated metal ion, appears to be readily miscible 200 nM, (4)500 nM, and (5) 1000 nM. with SOPC a t all compositions and surface pressures. In Figure 5 the molecular areas of PSIDNSOPC monolayers Similar results were obtained for protein binding to are plotted as a function ofPSIDA mole fraction for various monolayers of pure DSIDA lipid 1,although the extent of values of the surface pressure. All data points lie on or monolayer expansion is much less than for 6-His cyt b5 near the straight line defined by A d n ) = A1(n)xl binding to PSIDA-Cu2+ (Figure 3). While a (small) Az(n)x~,where A & c )is the mean molecular area of the monolayer expansion was observed for 6-His cyt b5 binding mixture, Al(n)andA2(n)are the mean molecular areas of to DSIDA-Cu2+, untagged protein has no effect on the pure PSIDA and SOPC films, respectively, and x1 and x2 metalated monolayer, nor did the tagged protein affect are the mole fractions of PSIDA and SOPC. This behavior the area of the unmetalated monolayer (data not shown). generally indicates ideal mixing in a two-component Figure 4 shows the PSIDA-Cu2+ monolayer expansion m o n ~ l a y e r . ~In~ addition, ,~~ fluorescence microscopy of measured for different concentrations of 6-His cyt b5 in the mixed PSIDNSOPC monolayers (5, 10, 30, and 50 the subphase. Special precaution was taken to ensure mol % PSIDA) shows a homogeneously bright field for the that the same amount of PSIDA-Cu2+ lipid was spread pyrene emission at all surface pressures, with no indication in each experiment. The error in the initial monolayer of lateral phase separation (data not shown). area a t 25 mN/m is -2%. Complete monolayer expansion Surface area measurements show no monolayer expanis observed with as low as 50 nM His-tagged protein in sion when 6-His cyt b5 is injected below pure SOPC, either the subphase, suggesting a n apparent dissociation conwith or without Cu2+in the subphase (data not shown). stant, Kd, for protein targeting less than 50 nM. As the No expansion is observed, in fact, until the fraction of protein concentration is reduced from 1pM to 50 nM, the PSIDA-Cu2+ in the mixed monolayer is between 10 and final extents of monolayer expansion remain similar. The 30 mol %, and then monolayer expansion increases with time required to reach maximum adsorption, however, the fraction of PSIDA-Cu2+ (Figure 6). The absence of increases significantly. expansion for the 10 mol % PSIDA-Cu2+ monolayer is Protein Binding to PSIDNSOPC Mixed Mononotable, as there is still a significant density of metal ions layers. The influence of interfacial metal ion density on at the interface (-1 per 500 Az) a t this PSIDA-Cu2+ protein targeting was studied using mixed monolayers concentration. of PSIDA and a matrix lipid that does not chelate metal ions, l-stearoyl-2-oleoyl-sn-3-glycerophosphocholine (31) Phillips, M. C.; Hauser, H.; Paltauf, F.Chem. Phys. Lipids 1972, 8, 127-133. (SOPC). Considerations in choosing a matrix lipid were (32) Gaines, G. L. Insoluble Monolayers at Liquid-Gas Interfaces; that it (1)have little or no nonspecific interactions with John Wiley & Sons: New York, 1966. the protein, ( 2 ) maintain the fluidity of the monolayer, (33) Bacon, K. J.; Barnes, G.,T. J. Colloid Interface Sci. 1978,67, and (3) be miscible with PSIDA, so that the metal ions 70-77.

+

4052 Langmuir, Vol. 11, No. 10, 1995

Ng et al.

10 mole% 1

1

100 200 Time (minutes)

1

I

300

Figure 6. Relative monolayer expansion M A versus time for mixed PSIDA-Cu2+/SOPC monolayers containing different mole fractions of PSIDA-Cu2+. The measurements were carried out at 25 mN/m and a subphase protein concentration of -0.2 pM.

Figure 7. Fluorescence micrographs of fluorescein-labeled 6-His cyt bg in the presenceof metal-chelating lipid monolayers spread on a buffered subphase (20 mM MOPS, 100 mM NaC1, pH 7.5) atn= 25 mN/m. (A)10%PSIDA-Ni2+, (B) 10%PSIDA, no metal, (C) 30%PSIDA-Ni2+, (D) 30%PSIDA, no metal, (E) 100%PSIDA-Ni2+, and (F)100%PSIDA, no metal.

Fluorescence Microscopy of Bound Protein. Metal ion-mediated protein targeting to the PSIDNSOPC interface was confirmed by fluorescence microscopy of fluorescein-labeled 6-His cyt bg. The fluroescent pyrene moiety of PSIDA is excited at wavelengths in the near U V (-355 nm) and thus does not contribute to the protein fluorescein signal, which is excited in the 450-490 nm range. The protein and lipid layers can therefore be imaged separately. For fluorescence imaging of the protein layer, it was necessary to use Ni2+ as the coordinating metal, since Cu2+can quench the fluorescein emission. Ni2+and Cu2+both serve to bind His-tagged proteins very tightly to IDA-derivatizedsupports in metalaffinity chromatography.lg Figure 7 compares the fluorescein signals a t three mixed PSIDNSOPC monolayers

(10,30, and 100mol %PSIDA),in the presence and absence of metal ions. Care was taken to ensure identical camera settings from experiment to experiment, so that fluorescence intensity can be correlated with fluorophore, and therefore protein, density. In the presence of Ni2+ a uniformly bright field is observed. The fluorescein fluorescence intensities of the 100% and 30% PSIDA monolayers are similar, while the 10%PSIDA monolayer is somewhat less intense. In the absence of Ni2+,very little fluorescein emission is observed for any of the three monolayers. No fluroescence signal was detected from a pure SOPC monolayer in the presence of 10 pM NiCl2, indicating that the phosphatidylcholine headgroup alone is unable to target the protein to the monolayer (data not shown). Discussion Interactions between proteins and a lipid monolayer spread at the air-water interface can be discerned from changes in the surface pressure-area behavior of the monolayer. The surface activity of most proteins, however, complicates the molecular interpretation of surface pressure-area measurements. One difficulty lies in distinguishing specific binding (via metal ion coordination in the present case) from nonspecific insertion of the protein into the monolayer, commonly observed at low surface pressure^.^^*^^ Nonspecific insertion introduces protein domains in the lipid monolayer, disrupting the lipid packing and resulting in a large increase in monolayer area at constant pressure. Because 6-His cytochrome bg becomes inserted nonspecifically in DSIDA and PSIDA monolayers only below 10 mN/m, monolayer expansion measured at 25 mN/m can be attributed to a different protein binding mechanism. Our surface area measurements clearly indicate that the lipid packing in metalchelating monolayers is modified by binding the Histagged protein. The surface area measurements and the fluorescence microscopy data, together with the fact that binding is reversible with the addition of EDTA to the subphase, support the view that protein binding occurs via coordination of the fusion peptide to the IDA lipidbound metal ion. Both Cu2+ and Ni2+ are effective for protein targeting with tridentate IDA lipids, as they are for metal-affinity chromatography of His-tagged prot e i n ~ .This ~ ~ may not be the case for the tetradentate NTA chelating group, which is generally used with Ni2+ only. 18,22,23 The nature of the lipid-lipid interactions and especially the fluidity and homogeneity of the monolayer can influence protein binding and the subsequent packing or assembly of the lipid-Cu2+-protein ternary complex within the monolayer. From their n-A isotherms (Figure l), it can be seen that the two IDA lipids 1 and 2 pack very differently at the interface, especially when loaded with metal ions. At pH 7.5, where all the present experiments were carried out, IDA bears a net negative charge. Chelation of a divalent metal ion by the negatively-charged and mobile IDA headgroup displaces a proton from the IDA amine and results in a charge-netural and, most likely, more compact headgroup. Consequently, the metalated DSIDA monolayer exhibits only a condensed phase. In contrast, the bulky pyrene moiety is sufficient to maintain the metalated PSIDA monolayer in the fluid phase at all surface pressures (Figure 1). Varying the protein concentration in the subphase by a factor of 20 has little effect on the final extent of PSIDACu2+monolayer expansion (Figure 4). The lowest protein concentration, 50 nM,corresponds to -6 nmol total protein in the subphase, which is still several-fold greater than the amount required to form a monolayer beneath the

Engineering Protein -Lipid Interactions A

B

Langmuir, Vol. 11, No. 10, 1995 4053

change in monolayer area until the point a t which the protein forms a complete monolayer beneath the lipid (B). The mole fraction of PSIDA-Cu2+ in the monolayer, XIDA, a t this point equals AI/Ap, where A L is the mean molecular area of a lipid molecule and Ap is the area occupied by a singl? protein molecule. The area per lipid molecule is -52 A2 a t 25 mN/m (obtained from the n-A isotherms, Figures 1and 5 ) , while an estimate ofAp caq be obtained from the crystal structure (see above), -500 A2. Thus, for the current system AI/Ap 0.1. Indeed, at XIDA = 10 mol % PSIDA-Cu2+, protein binding did not expand the monolayer, while monolayers with higher fractions of PSIDA-Cu2+ did expand. More metal ions than required for monolayer protein coverage are available a t the interface for monolayers containing higher PSIDA-Cu2+ fractions. In order for protein to occupy these additional binding sites, area must be created a t the interface (C). At PSIDA-Cu2+ fractions greater than AI/Ap, the density of bound protein appears to remain essentially constant. The increase in final area with increasing fraction of PSIDA (Figure 6) is believed to reflect an increase in the total amount of bound protein. The maximum change in area for a particular monolayer composition reflects a balance between the energetic cost of perturbing the lipid packing and the energetic benefit of the additional IDACu2+-histidine interactions. This proposed binding mechanism is also consistent with the larger area increases observed for protein binding to metalated PSIDA than to DSIDA a t the same pressures and protein concentrations (Figure 3). As discussed above, the energetic cost of disrupting the lipid packing will be quite different for these two chelating lipids. The fluid PSIDA-Cu2+ monolayer readily allows formation of the IDA-Cu2+-protein complex at the interface. More energy would be required to effect similar reorganization of the tightly-packed DSIDA-Cu2+ monolayer. The metalated PSIDA monolayer has a higher capacity for specificprotein binding and thus is significantly more effective for targeting the His-tagged protein to the interface. Here we have demonstrated a versatile and generallyapplicable approach for targeting proteins with very high affinity and high specificity to lipid assemblies. One promising application for this targeting system is in the two-dimensional crystallization of His-taggedproteins.23 Previous studies indicate that the following conditions favor two-dimensional cry~tallization:~~ (1) fixing the molecules in a plane with a specific orientation, (2) high concentrations of the molecule a t the interface, and (3) allowing the molecules sufficient mobility within the plane to sample various intermolecular contacts. Langmuir monolayers can provide these conditions: an affinity ligand serves to concentrate the protein molecules a t the interface and fix them in the monolayer plane with a particular orientation (conditions 1 and 2), while the fluidity of the monolayer allows the bound proteins to sample different intermolecular contacts (3). The same poly(histidine)hsion peptide used to purify a recombinant protein by metal-affinity chromatographycould also serve as a general affinity “tag” for two-dimensional crystallization on monolayers of suitable metal-chelating lipids such as the PSIDA lipid 2 described here. Because formation of protein domains or crystals should alter the distribution of the PSIDA lipid in the monolayer, it should be possible to use fluorescence microscopy of the pyrene lipid to image these domains, without the need for modification of the protein with a fluorescent label.

-

Figure 8. Proposed mechanism for monolayer expansion due to protein binding. (A)XIDA < AJAP. Protein binds IDA-Cu2+ lipids without affecting monolayer surface area. (B) XIDA = AJAp. Bound protein forms a monolayer beneath the lipid monolayer. Essentially all IDA-Cu2+ binding sites are occupied, and no increase in area is observed. (C) XIDA > AJAp. Protein binding disrupts lipid packing and increases monolayer surface area at constant surface pressure.

IDA-Cu2+ lipid monolayer. (On the basis of the dimen!ions of ytochrome b5 in its crystal structure, 30 A x 22 A x 22 A,34and a monolayer area a t 25 mN/m of -6000 mm2, a protein monolayer would include -2 nmol total protein). The fact that the system reaches essentially the same final protein binding capacity a t all concentrations studied suggests that the apparent Kd for IDA-Cu2+6-His cyt b5 binding at the interface is less than 50 nM. The density of protein binding sites (metal ions) a t the interface can be controlled by mixing the IDA lipid with a suitable matrix lipid. Both the mixing curves (Figure 5) and fluorescence microscopy of the pyrene-labeled lipid indicate that PSIDA and SOPC are ideally mixed, both with and without metal ions. Fluorescence microscopy of the bound protein also shows a homogeneous field (Figure 7). Decreasingthe PSIDA-Cu2+ mole fraction in a mixed monolayer with SOPC decreases the observed total monolayer expansion caused by protein (Figure 6). However, the fluorescence microscope images indicate that the bound protein density is similar for monolayers containing 100%and 30%PSIDA (Figure 7). That the 10 mol % PSIDA-Cu2+ monolayer does not expand could indicate that the protein does not bind. This is unlikely, however, given the high affinity of the 6-His peptide for IDA-Cu2+ and IDA-Ni2+ complexes and the significant gensity of metal ions a t the interface (-1 Cu2+per 500 A2). The fluorescencemicroscopy experimentsin fact show that fluorescein-labeled 6-His cyt b5 is indeed targeted to SOPC monolayers containing 10% PSIDA-Ni2+ (Figure 7a),albeit at a lower density than to monolayers containing 30% or 100% PSIDA-Ni2+. Thus, while an increase in monolayer area reflects protein binding to the interface, protein binding does not necessarily change lipid packing and increase the surface area. An explanation for these observations is offered below and outlined schematically in Figure 8. At very low fractions of PSIDA-Cu2+ in mixed monolayers (A), the 6-His cyt b5 occupies essentially all the binding sites a t the interface. Bound protein molecules are expected to be well-separated and do not measurably alter the monolayer surface area. As the fraction of PSIDA-Cu2+ is increased, the density of bound cyt bg increases with no I

(34)Mathews, F. S.; Levine, M.; Argos, P. J . Mol. Biol. 1972, 64, 449-464.

4054 Langmuir, Vol. 11, No. 10, 1995

Experimental Section Lipids. Synthesis of the DSIDA lipid (1) was described previously.z1 SOPC was obtained from Avanti Polar Lipids, Inc. (Alabaster, AL) and was used as received without further purification. Synthesis of PSIDA Lipid 2. General. All reactions were performed in oven-dried (160 "C) glassware under positive N2 atmosphere. Tetrahydrofuran (THF) was distilled from sodium benzophenone, and DMSO was dried over 3 A molecular sieves. Methylsulfonyl chloride, triethylamine (NEtS), solketal, and diethyl iminodiacetate (Kodak) were dried and distilled prior to use. All other reagents were used as received. All 'H and NMR were performed in CDC13 on a Nicolet QE-300 NMR spectrometer, and infrared (IR)spectra were obtained on a PerkinElmer 1600 Series FTIR spectrometer. Elemental analyses were performed by Desert Analytics (Tucson, AZ). All flash column chromatography was performed with Merck grade 60 silica gel. Melting point determinations were done on a Laboratory Devices Mel-Temp I1 and are uncorrected. 3-(Octadecyloxy)-2-((9-(1-pyrenyl)nonyl)oxy-1- ((triphenylcarbiny1)oxy)propane (4). l-Octadecyl-3-(triphenyl)carbinylglycerol, 3, (2.78 g, 4.74 mmol) and KOH (0.80 g, 14.3 mmol) were placed in DMSO (50 mL) and stirred at 80 "C for 20 min. 941-pyrenyll1-(methylsulfony1)nonane[prepared using a modified procedure from Tocannez71(2.00 g, 4.74 mmol) was then added as a solid with a reaction time of 12 h. The solution was then cooled to room temperature, and the product flash column chromatographed with CHZClfiexanes (50% v/v, Rf = 0.30). Product 4 was obtained as a yellow oil (2.25 g, 52%). IH NMR 6: 8.28 (d, J = 9.3 Hz, lH, Py-H), 8.16-7.95 (m, 7H, Py-H), 7.86 (d, J = 7.8 Hz, lH, Py-H), 7.46 (m, 6H, Ph-H), 7.30-7.18 (m, 9H, PhH),3.55-3.49 (m, 5H, C(H)O), 3.38 (t, J = 6.6 Hz, 2H, CHzO), 3.32 (t,J = 7.8 Hz, 2H, Py-CHz), 3.16 (m, 2H, CH20CPh3), 1.84 (m, 2H, PY-CH~CH~), 1.58-1.24 (m, 44H), 0.87 (t, J = 6.5 Hz, 3H, CH3). NMR 6: 144.13, 137.33, 130.92, 129.64, 128.73, 127.92, 127.70, 127.53, 127.24, 127.06, 126.87, 126.45, 125.73,

Ng et al. 129.63, 127.50, 127.20, 127.04, 126.42, 125.70, 125.02, 124.73, 124.57, 123.49, 77.82, 72.51, 71.64, 71.35, 70.78, 70.61, 70.55, 70.32,61.72,33.62,31.95,31.93,30.04,29.82,29.70,29.66,29.56,

29.50, 29.37, 26.11, 26.06, 22.70, 14.14. IR (neat): 3448, 3040, 2922, 2852, 1466, 1350, 1134, 844 cm-l. Anal. Calcd for C5~H8206: C, 77.76; H, 10.29. Found: C, 77.87; H, 10.42. 8-[l-0ctadecyl-2-(9-(1-pyrenyl)nonyl)-rac-glyceroyl]-3,6-dioxaoctyl 1-Bromide (7). 6 (0.60 g, 0.75 mmol) was dissolved in THF (10 mL) and the solution cooled to 5 "C. Carbon tetrabromide (0.37 g, 1.1 mmol) and triphenylphosphine (0.29 g, 1.1mmol) were added to the solution, and the mixture was stirred for 12 h. The solution was then concentrated in vacuo and flash chromatographed with EtzOhexanes (40% v/v, Rf = 0.26). Product 7 was obtained as a yellow oil (0.61 g, 94%). lH NMR 6: 8.28 (d, J = 93. Hz, lH, Py-H), 8.17-7.98 (m, 7H, Py-H), 7.86 (d, J = 7.8 Hz, lH, Py-H), 3.78 ( t , J =6.2 Hz, 2H, CHZBr), 3.64-3.40 (m, 19H, C(H)-O), 3.33 (t,J = 7.8 Hz, 2H, py-CHz), 1.85 (m, 2H, Py-CHzCHz), 1.54-1.24 (m, 44H), 0.87 (t, J = 6.6 Hz,3H,CH3). 13CNMR6: 137.31,131.42,130.91,129.66,127.52, 127.23, 127.06, 126.45, 125.73, 124.75, 124.59, 123.52, 77.82, 71.66,71.39,71.19,70.86,70.70,70.59,33.64,31.98,31.94,30.30, 30.08,29.84,29.71,29.67,29.57,29.52,29.38,26.13,26.08,22.70,

14.16. IR(neat): 3037,2923,2852, 1465, 1350,1134,844cm-l. Anal. Calcd for C&8105Br: C, 72.11; H, 9.43. Found: C, 72.04; H, 9.27. N-[8-[1-Octadecyl-2-(9-(l -pyrenyl)nonyl)-rac-glyceroyll-3,6-dioxaoctylliminodiacetic Acid Diethyl Ester (8). 7 (0.60 g, 0.69 mmol), diethyl iminodiacetate (0.52 g, 2.7 mmol), and triethylamine (0.40 mL, 2.9 mmol) were placed in CH3CN (10 mL)A'HF (4 mL) solution and refluxed for 3 days. The reaction was cooled to room temperature, concentrated in vacuo, and flash chromatographed with EtOAdhexanes (40%v/v, Rf= 0.26). A yellow oil was obtained and identified as 8 (0.26 g, 39%). lH NMR 6: 8.27 (d, J = 93. Hz, lH, Py-H), 8.16-7.97 (m, 7H, Py-H), 7.86 (d, J = 7.8 Hz, lH, Py-H), 4.15 (q, J = 7.1 Hz, 4H, (C=O)CHzCHd, 3.61-3.39 (m, 23H, CUI)-0, N-CHz(C=O)), 3.32 (t, J = 7.8 Hz, 2H1, Py-CHz), 2.96 (t,J = 5.6 Hz, 2H, CHzCHz-N), 125.05,124.75,124.59,123.53,86.58,78.28,71.61,71.15,70.68, 63.55,33.64,31.97,30.12,29.85,29.72,29.56,29.39,26.15,26.11, 1.84 (m, 2H, Py-CHzCHz), 1.54-1.24 (m, 50H, aliphatic CHZ, (C=O)CH2CH3),0.87(t,J=6.6Hz, 3H,CH3). WNMR6: 171.37, 22.71, 14.17. IR (neat): 3037, 2924, 2852, 1686, 1490, 1448, 137.29, 131.39, 130.89, 129.61, 127.50, 127.20, 127.04, 126.42, 1117, 843, 705 cm-l. Anal. Calcd for C65H8403: C, 85.48; H, 125.70,125.02,124.72,124.56,123.49,77.80,71.62,71.35,70.80, 9.27. Found: C, 85.88; H, 9.23. 70.75,70.51,70.32,60.42,55.87,53.55,33.62,31.95,30.06,29.83, 3-(Octa&cyloxy)-2- ((9-(1-pyrenyl)nony1)oxy)propan-1-01 (5). Into 29.69,29.54,29.50,29.36,26.10,22.69, 14.24, 14.14. IR (neat): a mixture of THF (20 mL)/MeOH (20 mL) was dissolved 4 (2.20 3040,2924,2853,1745,1466,1186,1116,1032,844 cm-l. Anal. g, 2.41 mmol) and TsOH.Hz0 (70 mg, 0.37 mmol). The solution CalcdforC60H9~N09:C, 73.96; H, 9.83; N, 1.44. Found: C, 73.58; was stirred for 13 h, and triethylamine (0.10 mL) was added H, 9.67; N, 1.55. prior to solution concentration in vacuo. Flash column chroN-[8-[1 -Octadecyl-2-(9-(1-pyrenyl)nonyl)-rac-glyceroyl]-3,6-dimatographywas performed withEtzO/CHZClz (2%v/v,Rf=0.18). oxaoctylliminodiacetic Acid) (2). In a solution of THF (6 mL)/ Yellow crystals of5 were recovered (1.51 g, 93%, mp 35-36 "C). MeOH (6 mL)/HzO (1 mL) were placed 8 (200 mg, 0.206 mmol) lH NMR 6: 8.27 (d, J = 9.3 Hz, lH, Py-HI, 8.16-7.95 (m, 7H, and crushed NaOH (50 mg). The homogeneous mixture was Py-H), 7.86 (d, J = 7.8 Hz, 1H Py-HI, 3.70 (br s, lH, CH-01, refluxed for 1h, cooled to room temperature, and then acidified 3.63-3.39 (m, 8H, CHz-O), 3.32 (t, J = 7.7 Hz, 2H, Py-CHZ), to pH 1. The solution was solvent stripped in uacuo and then 2.19 (br s, lH, OH), 1.84 (m, 2H, Py-CHzCHz), 1.56-1.24 (m, taken up in Et20 (30 mL) and aqueous saturated NaCl(20 mL). 44H), 0.87 (t, J = 6.4Hz, 3H, CH3). 13CNMR6: 137.29, 131.41, The mixture was shaken, the layers were separated, and the 130.90, 128.56, 127.52, 127.23, 127.06, 126.45, 125.73, 124.75, aqueous emulsion was extracted with fresh Et20 (2 x 30 mL). 124.58, 123.50, 78.17, 71.84, 70.88, 70.35, 63.10, 33.63, 31.95, The final extraction removed the emulsion, and two cleanly 30.04,29.71,29.62,29.48,29.39,26.09,22.71, 14.16. IR(KBr): separated phases were observed. The organics were combined 3444, 3037, 2917,2850, 1468, 1118,840 cm-I. Anal. Calcd for and dried over anhydrous MgS04 and filtered, and the solvent C46H7003: C, 82.33; H, 10.51. Found: C, 82.58; H, 10.25. S-[l-Octadecyl-2-(9-(1-pyrenyl)nonyl)-rac-glyceroyll-3,6-di- was removed. The waxy material was then taken up in CHZC12 and run through a small plug of silica gel. Removal of solvent oxaoctan-1-01 (6). 5 (1.40 g, 2.09 mmol) in THF (20 mL) was in uacuo yielded product 2 as a yellow, waxy material (160 mg, syringe injected into a suspension of NaH (0.33 g of a 60% oil 85%). lH NMR 6 : 10.18 (br s, 2H, COOH), 8.25 (d, J = 9.3 Hz, dispersion, 8.3 mmol) in THF (20 mL), and the mixture was l H , Py-H), 8.14-7.96 (m, 7H, Py-HI, 7.84 (d, J = 7.8 Hz, lH, stirred for 30 min at room temperature. l-(Methylsulfonyl)-9Py-HI, 4.01 (br s, 4H, N-CHz-COzH), 3.75 (br s, 2H, CHZCHZ(triphenylcarbinyl)-3,6,9-trioxynonane (1.28 g, 2.72 mmol) was N), 3.60-3.38 (m, 19H, O-C(H)), 3.30 (t, J = 7.7 Hz, 2H, Pythen added as a solid and the reaction brought to reflux for 10 CHZ),1.82 (m, 2H, Py-CHzCHz), 1.52-1.22 (m, 44H), 0.86 ( t , J h. The solution was then cooled to room temperature, worked =6.5Hz,3H,CH3). 13CNMR6: 169.56,137.30,130.88, 129.61, up, and concentrated in vacuo. 128.53, 127.51, 127.21, 127.06, 126.43, 125.72, 125.02, 124.74, The resulting oils were taken up in THF (10 mL)/MeOH (10 124.58, 123.50, 77.76, 71.72, 71.23, 70.95, 70.53, 70.38, 70.23, mL). TsOHaH20 (70 mg) was added and the mixture stirred for 56.86,56.78,33.62,31.96,30.00,29.83,29.72,29.55,29.38,26.09, 5 h. Triethylamine (0.10 mL) was added, and the solution was 26.06, 22.70, 14.16. IR (neat): 3044, 2923, 2852, 1734, 1466, concentrated in vacuo and flash chromatographed with EtOAd 1244,1115,843,720 em-'. Anal. Calcd for C56HagN010: C, 71.84; hexanes (40%v/v, Rf= 0.13). Product 6 was obtained as a yellow H, 9.58; N, 1.50. Found: C, 71.53; H, 9.13; N, 1.57. oil (1.47 g, 88%). 'H NMR 6: 8.27 (d, J = 9.3 Hz, lH, Py-H), 8.16-7.97(m,7H,Py-H),7.85(d,J=7.8Hz, lH,Py-H),3.72Protein Preparation. The soluble fragment of rat liver 3.39 (m, 21H, C(H)-O), 3.32 (t, J = 7.8 Hz, 2H, Py-CH2), 2.64 cytochrome b~(wild type and avariant with a 6-histidine sequence fused to the C-terminus) was expressed inEscherichia coli strain (br s, l H , OH), 1.84 (m, 2H, Py-CHzCHz), 1.54-1.24 (m, 44H0, 0.87(t,J=6.6Hz,3H,CH3). 13CNMR6: 137.29,131.40,130.89, TB-1 from synthetic genes generously provided by Dr. S. G.

Langmuir, Vol. 11, No. 10, 1995 4055

Engineering Protein-Lipid Znteractions Sligar.35 The cells were grown in LB broth with 200 mg/L ampicillin a t 37 "C in several inoculation steps. Cells were harvested by centrifugation and lysed in 50 mM Tris-HC1 buffer, pH 8.0, with 1mg/mL lysozyme by sonication. Cell debris was removed by centrifugation. Wild-type cytochrome bs was purified according to the procedures described by Beck von Bodman et al.35 For the 6-histidine variant, the supernatant was applied to a NTA-Nickel column (QIAGEN Inc.) equilibrated in 50 mM Tris-HC1, pH 8.0, with 20 mM imidazole. The column was extensively washed with 20 mM imidazole, and the 6-His cyt bg was eluted with 250 mM imidazole. A final DEAE chromatography step with a linear gradient ofNaCl(0-0.5 M) gave protein homogenous by SDS-PAGE. Cyt bg was labeled with fluorescein succinimidyl ester (FSE, Molecular Probes, Inc., Eugene OR) following the procedure of B r i n k l e ~ Briefly, . ~ ~ 10 mM FSE in dry DMF was added dropwise, with stirring, to a solution of ca. 100pM cyt bg (in 0.1 MNaHC03, 0.1 M NaC1, pH 8.3) to yield a FSE/protein ratio of 2:l. The reaction mixture was incubated at room temperature for 1 h. Labeled protein was separated from unreacted dye by gel filtration on prepacked Sephadex G-25 columns (PD-10, Pharmacia, Uppsala, Sweden) preequilibrated with 20 mM MOPS, 100 mM NaC1, pH 7.5. Absorbance measurements using extinction coefficients of 130 mM-'.cm-' at 413 nm and 68 mM-'.cm-l at 495 nm for cyt bg and fluorescein, respectively, indicate a fluoresceirdcyt bg ratio of approximately 1:l. The resulting labeled 6-His cyt bg behaves similarly to unlabeled 6-His cyt bs in analytical immobilized metal-ion-affinity chromatography on a Cuz+-IDA support (data not shown), which indicates that the labeling reaction does not interfere with the ability of the 6-His tail to interact with immobilized Cu2+-IDA. Monolayer Measurements. Monolayer studies were carried out on a computer-controlled Langmuir film balance (KSV Instruments, Helsinki) with a surface area of 75 x 230 mm2.21 The entire film balance was placed inside a plexiglass box, and the temperature was controlled by a circulating water bath. The surface pressure was measured by the Wilhelmy plate method using plates cut from filter paper (Whatman, No. 1). The plate was rinsed thoroughly with chloroform and calibrated with stearic acid prior to use. Lipid solutions were prepared in chloroform (A.C.S. HPLC grade, Aldrich). IDA lipid-SOPC mixtures were prepared by mixing appropriate volumes of stock solutions prior to spreading. The IDA lipids were premetalated by adding a n aliquot of concentrated CuClz (99.999%, Aldrich) in methanol to the lipid solutions. Cuz- excess in the lipid solution is -10-15%. (35) Beck van Bodman, S.;Schules, M. A.; Jollie, D. R.; Sligar, S. G. Proc. Natl. Acad. Sci. U S A . 1986,83, 9443-9447. (36) Brinkley, M. Bioconjugate Chem. 1992,3,2-13.

Protein binding experiments were carried out on a subphase buffered with 20 mM MOPS, 100 mM NaC1, pH 7.5. Monolayers (molecule*min)-i.A were compressed a t a constant rate of 3 15 min waiting period prior to compression allowed for solvent evaporation. Once the desired surface pressure was attained, the instrument was switched to constant pressure mode, and the monolayer was allowed to equilibrate for 15-30 min. The desired amount ofprotein was injected into the subphase behind the barrier and mixed using a peristaltic pump. Changes in surface area were recorded as a function of time. After each experiment, the trough was thoroughly cleaned with detergent, Milli-Q water, and chloroform. All measurements were performed at 25 "C. FluorescenceMicroscopy, The Langmuir instrument was placed above the stage of an inverted epifluorescence microscope (Zeiss Axiovert 35). Acircular quartz window (-5 cm in diameter) was incorporated into the bottom of the trough. An HBO 100 W/2 Hg arc lamp was used as the excitation source. A long working distance reflectingobjective(36x, NAO.5,8 "working distance, Ealing Electro-Optics, Holliston, MA) transmits light from the far UV to the near IR, allowing UV excitation of the pyrene fluorophore. Excitation and emission wavelengths were selected using appropriate narrow-band interference filters. Images of fluorescein-labeled protein were obtained using excitation in the range 450-490 nm and a 515-565 nm band pass emission filter (Zeiss set 487910). Pyrene fluorescence was excited by selection of the 334 nm line of the Hg lamp via a narrow band-pass (10 nm fwhm, Melles Griot, Irvine, CA) interference filter. By the choice of appropriate filters it is possible to selectively observe pyrene monomer and excimer signals (405DF35 and 470DF40, respectively, Omega Optical, Inc., Brattleboro, VT). The images were obtained via a variable-gain image intensifier (Hamamatsu C2400-68)with a solid-state CCD camera (Hamamatsu C2400-60) and were recorded on a video cassette recorder. Fluorescence images were digitized on a Macintosh Quadra 660AV and were manipulated using the program NIH Image (written by Wayne Rasband a t the U.S. National Institutes of Health and available from the Internet by anonymous ftp from zippy.nimh.nih.gov or on floppy disk from NTIS, 5285 Port Royal Rd., Springfield, VA 22161, part number PB93-504868).

Az

Acknowledgment. This research w a s supported by the Office ofNaval Research (N00014-92-5-1178), D.W.P. is a Landau Fellow and is supported by a training fellowship from the National Institute of General Medical Sciences, NRSA Award 1 T32 GM 08346-01. LA950357U