Ultramicroscopy 42-44 (1992) 1125-1132 North-Holland

Streptavidin binding observed with an atomic force microscope A.L. W e i s e n h o r n a,1, F.-J. S c h m i t t b, W. Knoll b a n d P.K. H a n s m a a a Department of Physics, University of California, Santa Barbara, CA 93106-9530, USA b Max-Planck-lnstitut fiir Polymerforschung, Postfach 3148, D-6500 Mainz, Germany Received 12 August 1991

A n atomic force microscope (AFM) was used to investigate a specific recognition reaction: the binding of streptavidin to a biotinylated lipid bilayer. Prior to the recognition reaction, the phase coexistence of the lipid bilayer was clearly observed: fluid domains were lower than the crystalline domains. After introducing to the bilayer a very dilute solution of streptavidin to give a final concentration of ~ 0.5p, M, the recognition reaction was imaged in real time. Several hours later, we observed a contrast reversal, i.e., the previously lower fluid domains grew so much in height that they became higher than the crystalline domains. We found that the streptavidin molecules bound almost exclusively to the biotin in the fluid domain ( < 0.25% coverage of the crystalline domains). The apparent structure of the few streptavidin molecules bound to the crystalline domain of the bilayer is shown to depend on the applied force. Finally, in a 2-dimensional quasi-crystal in which the streptavidin molecules were compressed at the air-water interface molecular resolution was achieved.

I. Introduction

Streptavidin is a bacterial protein produced by streptomyces auidinii and has a molecular weight of 60000 g / m o l [1]. It is homologous to avidin and has similar properties. As a tetramer, it consists of four identical subunits, each with one binding site for biotin. The binding is similar to an antibody-antigen binding: non-covalent, but of comparable strength. The dissociation constant for each individual biotin binding is gdiss = 10-15 mol/1 [2]. The 3-dimensional structure of crystalline streptavidin and of the streptavidin-biotin complex has been resolved by X-ray analysis [3]. It was found that the biotin is nearly incorporated in the binding site. Biotin (vitamin H) may be chemically modified in various ways [4]. In particular, it can be bound to phospholipids, which form monolayers at the air-water interface. Recently, it was shown that streptavidin still recognizes these biotinylated 1 P e r m a n e n t address: Institut d'Histologie, P6rolles, CH-1700 Fribourg, Switzerland.

UniversitY,

monolayers and may form 2-dimensional crystals [5]. The unit cell has been determined by electron diffraction [6], and the size of a single molecule can be estimated to be ~ 5 . 0 x 4 . 5 x 4 . 5 nm. Even after transfer of a biotinylated monolayer to a solid support, the biotin receptor remains accessible and reactive to the streptavidin. In a binary mixture with non-labelled lipids, a phase transition yields inhomogeneous distribution of the biotin lipid: It is enriched in the fluid domains and "squeezed out" of the pure, crystalline domains. Therefore, subsequently injected streptavidin binds preferably to the fluid domains [7]. This has been monitored in real time with plasmon-surface-polaritions microscopy, an optical technique which can monitor thickness changes of 0.2 nm with a lateral resolution of ~ 5 /zm

[8,9]. The atomic force microscope (AFM) [10,11] has not only a very high vertical resolution (0.02 nm); it has also a very high lateral resolution (atoms and molecules, i.e. subnanometer resolution). The AFM has successfully imaged lipid bilayer with molecular resolution [12-15].

0304-3991/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved

1126

A.L. Weisenhorn et al. / AFM obsercation o[ streptacidin binding

Furthermore, binding of antibodies [13] to a bilayer with active antigen binding sites has been observed. Previously, streptavidin microcrystals have been imaged with a scanning tunneling microscope (STM) [16]. Here we report on the binding of streptavidin molecules to a biotinylated bilayer using an AFM. We also investigated the structure of the streptavidin imaging at very low forces ( ~ 10-11 N).

0 II

DMPE / ~ / x / x ,

~

o

o

t

o

..

L0_P_0/N / C1/,- c h a i n

z

OH

DPPE-biotin

(a)

o~i 2. Material and methods

/N/N/N/N/N/N/N/N 0 ~ 0

2.1. Lipids and streptavidin

H I N

0 II

C16- chain

OHI

0 II 011

i/I

H I S,~

N~---O I

H T h e lipids L-c~-dimyristoyl phosphatidyl ethanolamine (DMPE) [17] and L-a-dipalmitoyl phosphatidyl ethanolamine (DPPE) [17] and the biotinylated lipid D P P E - b i o t i n [18] were used as purchased (see fig. la). Note the e-amino caproyl spacer between the D P P E moiety and the biotin. Streptavidin and fluorescein labelled streptavidin were kindly provided by Boehringer Mannheim G m b H (Werk Tutzing), Germany.

2.2. Monolayer Freshly cleaved, atomically flat mica [19] was used as a solid support. As a first monolayer, DPPE was deposited using the Langmuir-Blodgett (LB) technique [20,21]. At a pressure of 35 m N / m , the mica was dipped vertically (mica relative to air-water interface) through the crystalline DPPE layer at the air-water interface. The second, biotinylated monolayer was transferred by horizontal dipping (Langmuir-Schaefer technique) onto the now hydrophobic D P P E / mica support (see fig. lb). We used 5 mol% mixture of the D P P E - b i o t i n in DMPE, either compressed to the phase coexistence region (17 m N / m ) between the fluid and the crystalline state or to the crystalline, solid state (35 m N / m ) .

I

DPPE

I spacer-J--bi°tin

I

cantilever with tip Buff ---~

~ ~

× ~U[~HUHHHH ~

M

i

Streptavidin

HHHHUUj]~UHH~H DPPE c

(b)

a

Fig. 1. (a) Models of a DMPE molecule and a DPPE-biotin molecule. (b) Schematic of a D M P E / D P P E - b i o t i n on D P P E / m i c a layer system in buffer. Left: without streptavidin molecules in the buffer; right: with streptavidin in the buffer and bound to the DPPE-biotin molecules. Note that the microfabricated cantilever and the AFM tip are completely immersed in the buffer.

streptavidin, as in ref. [6]. Bright, fluorescing areas were observed with a fluorescence microscope where protein crystals were formed at the air-water interface. However, these crystals did not completely cover the whole interface. The transfer was performed by horizontal dipping, as described above.

2.3. Protein quasi-crystal 2.4. Atomic force microscope For the preparation of the 2-dimensional quasi-crystal, pure D P P E - b i o t i n was spread in a flat petri dish and incubated with fluorescein

We used the commercial AFM Nanoscope II [22] with a liquid cell and S i 3 N 4 microfabricated

A.L. Weisenhorn et al. / AFM observation of streptaeidin binding

cantilever with pyramidal tips [23] (for more details of AFM schematic, see ref. [24]). The liquid cell (total volume ~ 0.5 ml), sealed with an Oring, kept the bilayer and the streptavidin quasicrystal submersed in buffer while imaged with the AFM. The streptavidin solution was introduced into the liquid cell, without removing the bilayer. For an estimation of the imaging force, the spring constant of the microfabricated cantilever was calculated from the geometrical dimensions and material properties [25]. Since they can be determined only within a certain range, the accuracy of the spring constant and therefore the accuracy of the imaging force is estimated to be within _+50%.

3. Experiment and results The top layer of a lipid bilayer can be compressed and hence immobilized enough to get a surface that is very stable for imaging with an AFM. Modification (here: biotinylation) of the lipid heads can make the bilayer reactive in a specific way. Fig. 2a shows a typical area of the lipid bilayer ( D M P E / D P P E - b i o t i n 19 : 1 on D P P E / m i c a in 0.5M aqueous NaCl solution with the top layer being in fluid-crystalline coexistence (~-= 17 m N / m ) . The white areas are the crystalline domains, and the darker areas are the fluid domains. Even the small crystalline islands (diameter ~ 100 nm) were not damaged or pushed around by continued scanning with the AFM tip. This reactive bilayer can be used to study recognition reaction such as the specific binding of streptavidin to biotin. About 70 min after fig. 2a was taken, 30 ~1 of a 6/~M streptavidin solution in 0.5M NaCI was introduced into the liquid cell to give a resulting concentration of ~ 0.5/~M and after 50 min fig. 2b was taken. The domain structure is the same as in fig. 2a. the borders between crystalline and fluid domains are not as pronounced as in fig. 2a. Furthermore, there is a horizontal band in the upper third, where binding of streptavidin to the fluid biotinylated lipids seems to have occurred. Note there is no binding visible in the crystalline domains.

1127

About 4 h later the liquid cell was flushed with 1 ml of a 0.5~M streptavidin solution, and fig. 2c was taken after another 5 h. The crystalline domains still show the same pattern, but now they are lower than the fluid domain with the bound streptavidin. No individual streptavidin was resolved in the fluid domain, since the lipids are fluid. Again, in the crystalline domain there was no streptavidin binding visible. In certain areas, however, we have observed individual streptavidin bound to some of the crystalline domains (area of coverage < 0.25%). This small coverage cannot account for the coverage seen by surface plasmon/fluorescence microscope technique that has a lateral resolution of about 5 /zm and therefore cannot resolve fluid domains in large crystalline domains and vice versa [7]. Newer surface plasmon experiments on highly compressed bilayers ( ~ - = 40 m N / m ) showed no measurable binding of streptavidin. (The least detectable difference in height for the surface plasmon technique is 0.2 nm, which yields a least detectable coverage of 4%, assuming 5 nm height of streptavidin). With the AFM we observed again less than 0.25% coverage of the crystalline domain. To quantitatively analyze the binding of streptavidin, depth histograms were calculated of the same area for figs. 2a-2c; they are shown in figs. 2d-2f. The value of the depth was measured in the middle of the bell-like curve (not necessarily at the maximum); for the bell-like curve of the fluid domains the measuring line is indicated. The difference in height (i.e. depth) changes from 2.1 nm (fig. 2d) to 1.5 nm (fig. 2e) and to - 0 . 3 nm (fig. 2D. The large difference in height (fig. 2d) compared to the measurements with surface plasmon technique (height difference: 0.7 nm [9]) is caused by the difference in rigidity of the crystalline and fluid domains. The lipids in the fluid domains are being pushed aside by the scanning tip ("plowing" through the lipids). This can also be seen in fig. 2d, where the bell-like curve of the fluid domain is wider. In fig. 2e the width is even more increased since there are now also streptavidin molecules bound to the fluid lipids. The height difference in fig. 2f ( - 0 . 3 nm) is too small considering a complete coverage with

1128

A.L. Weisenhorn et al. / AFM obsercation of streptaL,idin binding

Fig. 2. A F M images of D M P E / D P P E - b i o t i n 19:1 on D P P E / m i c a in 0.5M aqueous NaCI solution at room temperature. Image size: 1200 rim, image height: 5 nm. (a) Before introducing streptavidin, (b) about 50 min after introducing 30/xl of 6/~M solution of streptavidin to give a final concentration of ~ 0.5/zM, (c) about 8 h later, (Note that 4 h 45 min after the first streptavidin introduction, the liquid cell was flushed with 1 ml of a 0.5/~M streptavidin solution.) (d)-(f) Histograms of height of a selected area for (a)-(c). The differences in height of the crystalline domain to the fluid domain are for (d) 2.1 nm, (e) 1.5 nm, and (f) - 0 . 3 nm (minus sign indicates contrast reversal). Note the broadening of fluid-domain peak from (d) to (e).

A.L. Weisenhorn et al. / A F M observation of streptavidin binding

streptavidin (size of streptavidin: ~ 5.0 x 4.5 × 4.5 nm). The softness of streptavidin might be the reason for it. The minus sign in the height difference indicates contrast reversal, i.e., the proteins bound to the fluid domain appear higher than the crystalline domains after the binding was completed. There are enough biotinylated lipids in the fluid domains (at least twice as much as is needed for a complete coverage) and streptavidin molecules in the solution (about 100 times as much as is needed for a complete coverage) to make the binding complete. In contrast to fig. 2, in which the crystalline domains were islands in a fluid domain, in fig. 3 the fluid domains of the same bilayer are enclosed by the crystalline domain. There are also a few crystalline islands in the fluid. Again fig. 3a was taken prior to introducing the streptavidin solution. Only about 5 min after streptavidin was introduced, fig. 3b was taken without removing the tip from the surface or without stopping the scanning. In fluid domains the binding of streptavidin is clearly visible; again no individual molecules were resolved in the fluid domains. Figs. 3a and 3b were imaged at about 50 pN. Increasing the force to > 200 pN makes the structure of the bound molecules disappear (fig. 3c). We believe that the tip is "plowing" through the protein lipid at this high force. As was seen in fig. 3c it is important to control the imaging force for soft molecules such as proteins. Theoretical calculation predicts deformation of globular proteins for forces > 10 pN [26]. A streptavidin protein bound to the crystalline domain was imaged at different forces. In fig. 4a the protein was imaged at 30 pN. Substructure of the protein can be seen and is consistent with a structural model for streptavidin based on electron diffraction and computer modelling [6]. However, the maximum height of the protein is with 1.12 nm still too small. This effect has previously been observed [14,27]. Increasing the force to 60 pN (fig. 4b) and even to 150 pN (fig. 4c) reduces the maximum height to 0.65 and 0.25 nm, respectively. A crude estimation of the compressibility of the protein gives the following: The compressibility of streptavidin in the vertical direction measured in the surface force apparatus (SFA) (see fig. 5 of ref. [28] or fig. 4 of ref. [29]) is on order of 0.2 n m / M P a (Ad/p, where Ad is the

1129

Fig. 3. A F M images of a different region of the same lipid bilayer as in fig. 2. Image size: 2000 nm, image height: 6 nm. (a) Before introducing streptavidin, (b) about 5 min after introducing 30 p.l of a 6/zM solution of streptavidin, (c) about 15 min later. Note that in (c) the force was increased from about 50 pN to > 200 pN. Therefore the streptavidin molecules are pushed to the side and are not visible anymore.

A.L. Weisenhorn et al. / AFM obseruation of streptavidin binding

1130

C

/ f~3

am

/'~'

~!

;::

Fig. 4. Measuring the compressibility of the protein streptavidin that is bound to the crystalline domain of the bilayer by imaging the protein at different forces and comparing the images to the streptavidin model; (a) was imaged at about 30 pN, (b) at about 60 pN, and (c) at about 150 pN. The maximal height of the protein relative to the surroundings (measured in the raw data) is 1.12 nm for (a), 0.65 nm for (b), and 0,25 nm for (c).

vertical displacement and using a radius R ~ 1 cm, /zm 2, and A d - - 1 nm. In force of > 200 pN (fig. 4c)

p the pressure) [30], a contact area of 10 the A F M an imaging acting on one strepta-

vidin with an area of 20 nm 2 yields > 10 MPa pressure. This would result in a deformation of > 2 nm, which is the right order of magnitude of deformation that is observed in the A F M images. Reducing the force to 30 pN reduces also the contact area and hence the pressure is not reduced accordingly. Therefore the deformation can still be on the order of a few nm. Instead of imaging proteins that are bound to an already immobilized, crystalline domain of a monolayer, one can study a monolayer of proteins that are bound to the biotinylated lipids before immobilization at the a i r - w a t e r interface. This technique was previously used to image Fab fragments [14,31]. The mobility is decreased because the proteins are densly packed and in a quasicrystalline state (the Fourier transformation did not reveal a clear spectrum; therefore, the proteins were not in a pure crystalline state). Fig. 5a shows an area of the protein quasi-crystal transferred to D P P E / m i c a . Plateaus of different heights can clearly be seen. The lowest (darkest) plateau (#1 in fig. 5a) is mica, i.e. areas where the bottom monolayer D P P E is incomplete. The next plateau (#2) is the D P P E bottom monolayer. On the highest (brightest) plateau (#3) structure caused by the streptavidin can be seen. A cross-section (see fig. 5b) gives about 5 nm height difference between area #1 and #3, 2.9 nm between #1 and # 2 (the literature value of 2.7 nm agrees well [29]), and 2.1 nm between # 2 and #3, which is again too small for the size of the streptavidins. Note that the D P P E - b i o t i n does not contribute to the height of the streptavidin quasi-crystal, since it is very dilute. The compression of the streptavidin seems to be consistent with the value of figs. 2c and 2f. Area #1 was imaged in fig. 5c; the correct periodicity of mica was measured with Fourier transformation. In fig. 5d (area #2) no molecular structure of the D P P E monolayer was resolved. The roughness of this monolayer is about 0.2 nm. In fig. 5e (#3) the roughness is clearly increased (1.9 nm) due to the size of the streptavidin proteins. A high-magnification image of a protein did not reveal the submolecular structure as clearly as in fig. 4a, because the streptavidin proteins were not sufficiently immobilized.

A.L. Weisenhorn et al. / A F M observation of streptauidin binding

_~(~)~P'~Im'~

~

_ .

1131

Buffer- - .

.

.

( ~

_

~#7~}/~}~/I~ - Streptavidin

Mica

Fig. 5. Streptavidin quasi-crystal. (a) Three different domains can be seen: The lowest area is the substrate mica (#1 in figure), the middle area is the bottom monolayer (DPPE) (#2) and the highest area is the streptavidin quasi-crystal (#3). Image size: 1000 nm, image height: 15 nm. (b) Top: section of (a) along black line. Bottom: schematic of section with the three areas mica (#1), bottom monolayer (#2), and the quasi-crystal (#3). (c) Mica from the area #1. The hexagonal structure of mica can clearly be seen. Image size: 20 nm. (d) Bottom monolayer from the area (#2). Image size: 100 nm, image height: 2 nm. (e) Protein quasi-crystal from the area #3. Image size: 250 nm, image height: 2 nm. Note the corrugation is larger for the quasi-crystal than the bottom monolayer.

1132

A.L. Weisenhorn et al. / AFM obsere,ation of streptavidin binding

4. Conclusion

In conclusion, we have observed the domains of reactive bilayers under buffer with an AFM, and followed the specific binding of water soluble proteins in real time. This system is important as a model for biomembranes and as a model for recognition reaction which might lead to applications like biosensors. Furthermore, we have imaged individual streptavidin proteins with submolecular resolution at 10 -11 N and estimated the compressibility of these proteins varying the imaging force, which could lead to better understanding of mechanical properties of proteins. In a 2-dimensional quasi-crystal domains of proteins and their degree of ordering can clearly be distinguished, which is important for the quality control of these crystals.

Acknowledgements

We would like to thank J.N. Israelachvili for the use of his lab and his Langmuir trough, I. Hale for letting us use the fluorescence microscope, C.A. Helm for helpful discussion, and J. Cleveland, G.L. Kelderman and E.T. Martzen for technical support. This work was supported by an IBM Manufacturing fellowship (ALW), by the Bundesministerium fi.ir Forschung und Technologie, Germany (0.3-KN 2 MPG 6) (FJS, WK); by Boehringer Mannheim, GmbH, Germany (FJS, WK); and a National Science Foundation - Solid State Physics Grant DMR89-17164 (PKH).

References [1] R.M. Buckland, Nature 320 (1986) 557. [2] N.M. Green, Adv. Protein Chem. 29 (1975) 85. [3] P.C. Weber, D.H. Ohlendorf, J.J. Wendoloski and F.R. Salemme, Science 243 (1989) 85. [4] M. Wilchek and E.A. Bayer, Anal. Biochem. 171 (1988) 1. [5] R. Blankenburg, P.H. Meller, H. Ringsdorf and C. Salesse, Biochemistry 28 (1989) 8212. [6] S.A. Darst, M. Ahlers, P.H. Meller, E.W. Kubalek, R. Blankenburg, H.O. Ribi, H. Ringsdorf and R.D. Kornberg, Biophys. J. 59 (1991) 387. [7] F.-J. Schmitt and W. Knoll, Biophys. J. 60 (1991) 716.

[8] B. Rothenh~iusler and W. Knoll, Nature 332 (1988) 615. [9] W. Hickel, D. Kamp and W. Knoll, Nature 339 (1989) 186. [10] G. Binnig, C.F. Quate and Ch. Gerber, Phys. Rev. Lett. 56 (1986) 930. [11] D. Rugar and P.K. Hansma, Phys. Today (Oct. 1990) 23. [12] A.L. Weisenhorn, M. Egger, F. Ohnesorge, S.A.C. Gould, S.-P. Heyn, H.G. Hansma, R.L. Sinsheimer, H.E. Gaub and P.K. Hansma, Langmuir 7 (1991) 8. [13] A.L. Weisenhorn, H.E. Gaub, H.G. Hansma, R.L. Sinsheimer, G.L. Kelderman and P.K. Hansma, Scanning Microsc. 4 (1990) 511. [14] A.L. Weisenhorn, B. Drake, C.B. Prater, S.A.C. Gould, P.K. Hansma, F. Ohnesorge, M. Egger, S.-P. Heyn and H.E. Gaub, Biophys. J. 58 (1990) 1251. [15] H.G. Hansma, A.L. Weisenhorn, S.A.C. Gould, R.L. Sinsheimer, H.E. Gaub, G.D. Stucky, C. Zaremba and P.K. Hansma, J. Vac. Sci. Technol. B 9 (1991) 1282. [16] L. H~iussling, B. Michel, H. Ringsdorf and H. Rohrer, Angew. Chem. 103 (1991) 568. [17] Sigma Chemical Co., St. Louis, MO. [18] Molecular Probes, P.O. Box 22010, 4849 Pitchford Avenue, Eugene, OR 97402. [19] Muscovite mica from Asheville-Schoemaker, P.O. Box 318, Newport News, VA. [20] G.L. Gaines, Jr., Insoluble monolayers at liquid-air interface (Interscience, New York, 1966). [21] V.K. Agarwal, Phys. Today (June 1988) 40-46. [22] Digital Instruments, Inc., 6780 Cortona Drive, Santa Barbara, CA 93117. [23] T.R. Albrecht, S. Akamine, T.E. Carver and C.F. Ouate, J. Vac. Sci. Technol. A 8 (1990) 3386. [24] S.A.C. Gould, B. Drake, C.B. Prater, A.L. Weisenhorn, S. Manne, G.L. Kelderman, H.-J. Butt, H.G. Hansma, P.K. Hansma, S. Magonov and H.-J. Cantow, Ultramicroscopy 33 (1990) 93. [25] A.L. Weisenhorn, Atomic Force Microscopy in Liquids, Dissertation, Department of Physics, University of California, Santa Barbara, CA 93106 1991. [26] B.N.J. Persson, Chem. Phys. Lett. 141 (1987) 366. [27] S.A.C. Gould, B. Drake, C.B. Prater, A.L. Weisenhorn, S. Manne, H.G. Hansma, P.K. Hansma, J. Masse, M. Longmire, V. Elings, B. Dixon Northern, B. Mukergee, C.M. Peterson, W. Stoeckenius, T.R. Albrecht and C.F. Ouate, J. Vac. Sci. Technol. A 8 (1990) 369.. [28] C.A. Helm, F.-J. Schmitt,.J.N. lsraelachvili and W. Knoll, Makromol. Symp. 46 (1991) 103. [29] C.A. Helm, W. Knoll and J.N. Israelachvili, Proc. Natl. Acad. Sci. 88 (1991) 8169. [30] Since the streptavidin molecules cannot expand laterally in the setup of the SFA, the compressibility of 0.2 n m / M P a is a lower-limit estimation for the setup of the AFM, where they can expand laterally while being compressed vertically (C.A. Helm, private communication). [31] M. Egger, F. Ohnesorge, A.L. Weisenhorn, S.-P. Heyn, B. Drake, C.B. Prater, S.A.C. Gould, P.K. Hansma and H.E. Gaub, J. Struct. Biol. 103 (1990) 89.

Streptavidin binding observed with an atomic force microscope.

An atomic force microscope (AFM) was used to investigate a specific recognition reaction: the binding of streptavidin to a biotinylated lipid bilayer...
2MB Sizes 0 Downloads 0 Views