Ultramicroscopy 42-44 (1992) 1200-1203 North-Holland
Imaging of proteins by scanning tunnelling microscopy T i m o t h y N.C. W e l l s a, M a r g a r e t S t e d m a n b a n d R o b i n J. L e a t h e r b a r r o w c Glaxo Institute for Molecular Biology, S.A., 14 ch. des Aulx, 1228 Plan-les-Ouates, Geneva, Switzerland b Division of Mechanical and Optical Metrology, National Physical Laboratory, Teddington T W l l OLW, UK c Department of Chemistry, Imperial College of Science Technology and Medicine, London SW7 2AZ, UK Received 12 August 1991
Scanning tunnelling microscopy has been used to examine the structure of proteins deposited on a graphite surface. Three molecules have been studied; immunoglobulin G (IgG), Complement component lq (Clq) and ATP-citrate lyase (ACL). The images show IgG as a tri-lobed molecule, consistent with the known 3D structure as determined by X-ray crystallography. The C l q images differ from the well known "tulip bunch" model derived by electron microscopy, but are consistent with the model if it is assumed that the six globular heads have aggregated. Molecules of A C L are visible as discrete units, with some hints of substructure. These results highlight the potential of STM in studying protein structures, but also illustrate the difficulties of interpreting micrographs of proteins whose structure is currently unknown.
1. Introduction The choice of proteins for our initial STM studies was based on availability of proteins with defined three dimensional structures, which would allow us to evaluate the utility of STM on protein samples. I g G has a characteristic tri-lobed structure [1] and C l q which appears as a " b o u q u e t of tulips" [2].
2. Experiment 1 /zl of 1 m g / m l solution of protein in 50mM Tris-HC1 p H 8.0, 10mM MgCI 2 was applied to an H O P G surface (Union Carbide) and dried using a constant stream of filtered compressed air. Molecules were located on the surface using a large scan area, with a scan frequency of 5-10 Hz, bias voltage 20 mV and 0.5-2.0 nA current. Larger scan speeds increased the tendency of the molecules to move across the surface during the imaging process. With C l q separation of the globular heads from the stems was observed at
high scan rates, presumably due to breakage of the collagen-like helix.
3. Results and discussion Fig. 1 shows a representative IgG molecule at high magnification from a region of the H O P G surface which contained several molecules [3]. Its size is in reasonable, but not precise agreement with the dimensions found by X-ray crystallographic studies. The length and the width of the Fab arms in the X-ray structure are 8.5 × 6 nm, whereas in these images it has a size of 17.5 x 11.5 nm. The difference in size could be caused by a number of factors such as solvent remaining on the surface, deformation of the protein by the tip, convolution of the tip profile with the molecular shape [4], or flattening due to interactions with the H O P G surface. More detailed investigations into the effect of imaging conditions on the apparent size of protein molecules by STM are planned. Without these results a detailed analysis of the absolute sizes should be carried out with
0304-3991/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
caution. Information along the z axis is limited due to the complex nature of the tunnelling effect in large molecules, but a height of 3 nm is consistent with X-ray data. C l q is an important component in the complement cascade, which is a major part of the mechanism by which the immune system destroys foreign microorganisms. The complex itself has a molecular mass of 460 kDa and from electron microscopy the dimensions are approximately 30 nm x 30 nm [5]. The images obtained (fig. 2) are different from those seen in electron microscopy. Although the size is reasonably close (35 nm along the longest dimension), the globular Cterminal regions, which normally appear as 6 distinct domains are not individually resolved. This suggests that these head regions, which are known to be attached to flexible stalks, have aggregated during the sample preparation. This raises important questions, such as whether increased ionic strength during deposition would help keep the domains apart, and whether different images are obtained when the protein remains in solution. We plan to use SFM to test these ideas. In addition a clear gap can be seen
1201
between the head and the stem region of the molecule. This is spanned by collagen helices which are presumably too small to image in these samples. If the scan rate is increased then the C l q can be cleaved in this region, leaving globular domains fixed to the graphite surface. Generally for our C l q images, the molecules are oriented in the same direction on the H O P G surface (fig. 2b). We have observed this orientation effect with other molecules, and hypothesize that it may be caused by preferential binding of a hydrophobic surface area on C l q to the H O P G . A C L is an important enzyme in the cellular biosynthesis of both cholesterol and fatty acids and thus A C L inhibitors may have potential as cholesterol lowering agents. It is a tetramer, showing "half of sites" reactivity, and has a molecular weight of 484000 (see references in ref. [6]). There is no three-dimensional structural information on this protein. In this case our interest was whether discrete, reproducible molecules could be imaged, and whether any substructure could be seen. (On the basis of biochemical data either two- or four-fold symmetry could be expected.) The images obtained (fig. 3) show a
Fig. 1. (a) Image of IgG on HOPG graphite. The size of the image is 45 nmx45 nm. The image was recorded on a Nanoscope II (Digital Instruments, Inc.) in air using a platinum/iridium tip. The tunnelling current was 0.5 nA and the bias voltage was 20 mV. (b) Model built from the x-ray coordinates of IgG(Dob) [1l. Several images with this characteristic shape have been recorded [3].
1202
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
molecule with d i m e n s i o n s approximately 10 n m x 14 nm, which is consistent with its m o l e c u l a r mass, a n d the images o b t a i n e d for the o t h e r two proteins. T h e large scan area (fig. 3a) shows m a n y molecules imaged at low resolution. A large n u m -
b e r of these a p p e a r to have a l o n g i t u d i n a l cleft, indicating a two-fold symmetry axis. T h e I g G a n d C l q e x p e r i m e n t s are excellent systems with which to test the quality of information derived from S T M b e c a u s e of their charac-
b i
~
i
I
_
_
Fig. 2. (a) Clq imaged on HOPG. The image size is 35 nmx35 nm, and the scanning conditions are the same as in fig. 1. The globular head domains are clearly separated from the stalk. (b) Two Clq molecules (image.size 90 nm x 90 nm). (c) A model based on electron microscopy[5] is shown for comparison.
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
1203
Fig. 3. (a) ATP-citrate lyase imaged using a large scan area, (250 nm x 250 nm). (b) Detail of an individual molecule. Scanning conditions are the same as in fig. 1. teristic q u a t e r n a r y structure. T h e strong correspondence between the images obtained and the known structure of I g G suggest that this technique will be of great value in studying other protein systems. T h e results from C l q and A C L highlight two of the major issues facing our research. First, the question of sample preparation. W h a t effect do the substrate, the buffers, the m e t h o d of deposition have on the type of images p r o d u c e d ? H e r e it will be important to c o m p a r e the results with those obtained from electron microscopy. It will also be important to add nonprotein size markers as internal controls for the microscope calibration. Second, as can be seen in the case of A C L , reproducible images can be obtained of proteins without a highly characteristic quaternary structure. In such cases, the interpretation of such images should be carried out cautiously. S T M has a useful role in the under-
standing of protein structure, especially for relatively large multi-domain complexes. T h e experiments need submicrogram quantities of protein and high-resolution images can be r e c o r d e d in several minutes.
References [1] E.V. Silverton, M.A. Navia and D.R. Davies, Proc. Natl. Acad. Sci. USA 74 (1977) 5140. [2] R.R. Porter and K.B.M. Reid, Adv. Protein Chem. 33 (1979) 1. [3] R.J. Leatherbarrow, M. Stedman and T.N.C. Wells, J. Mol. Biol. 221 (1991) 361. [4] M. Stedman, J. Microscopy 152 (1988) 611. [5] C.J. Strong, R.C. Siegel, M.L. Philips, P.H. Poon and V.N. Schumaker, Proc. Natl. Acad. Sci. USA 79 (1982) 486. [6] T.N.C. Wells, Eur. J. Biochem. 199 (1991) 163.
Imaging of proteins by scanning tunnelling microscopy T i m o t h y N.C. W e l l s a, M a r g a r e t S t e d m a n b a n d R o b i n J. L e a t h e r b a r r o w c Glaxo Institute for Molecular Biology, S.A., 14 ch. des Aulx, 1228 Plan-les-Ouates, Geneva, Switzerland b Division of Mechanical and Optical Metrology, National Physical Laboratory, Teddington T W l l OLW, UK c Department of Chemistry, Imperial College of Science Technology and Medicine, London SW7 2AZ, UK Received 12 August 1991
Scanning tunnelling microscopy has been used to examine the structure of proteins deposited on a graphite surface. Three molecules have been studied; immunoglobulin G (IgG), Complement component lq (Clq) and ATP-citrate lyase (ACL). The images show IgG as a tri-lobed molecule, consistent with the known 3D structure as determined by X-ray crystallography. The C l q images differ from the well known "tulip bunch" model derived by electron microscopy, but are consistent with the model if it is assumed that the six globular heads have aggregated. Molecules of A C L are visible as discrete units, with some hints of substructure. These results highlight the potential of STM in studying protein structures, but also illustrate the difficulties of interpreting micrographs of proteins whose structure is currently unknown.
1. Introduction The choice of proteins for our initial STM studies was based on availability of proteins with defined three dimensional structures, which would allow us to evaluate the utility of STM on protein samples. I g G has a characteristic tri-lobed structure [1] and C l q which appears as a " b o u q u e t of tulips" [2].
2. Experiment 1 /zl of 1 m g / m l solution of protein in 50mM Tris-HC1 p H 8.0, 10mM MgCI 2 was applied to an H O P G surface (Union Carbide) and dried using a constant stream of filtered compressed air. Molecules were located on the surface using a large scan area, with a scan frequency of 5-10 Hz, bias voltage 20 mV and 0.5-2.0 nA current. Larger scan speeds increased the tendency of the molecules to move across the surface during the imaging process. With C l q separation of the globular heads from the stems was observed at
high scan rates, presumably due to breakage of the collagen-like helix.
3. Results and discussion Fig. 1 shows a representative IgG molecule at high magnification from a region of the H O P G surface which contained several molecules [3]. Its size is in reasonable, but not precise agreement with the dimensions found by X-ray crystallographic studies. The length and the width of the Fab arms in the X-ray structure are 8.5 × 6 nm, whereas in these images it has a size of 17.5 x 11.5 nm. The difference in size could be caused by a number of factors such as solvent remaining on the surface, deformation of the protein by the tip, convolution of the tip profile with the molecular shape [4], or flattening due to interactions with the H O P G surface. More detailed investigations into the effect of imaging conditions on the apparent size of protein molecules by STM are planned. Without these results a detailed analysis of the absolute sizes should be carried out with
0304-3991/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
caution. Information along the z axis is limited due to the complex nature of the tunnelling effect in large molecules, but a height of 3 nm is consistent with X-ray data. C l q is an important component in the complement cascade, which is a major part of the mechanism by which the immune system destroys foreign microorganisms. The complex itself has a molecular mass of 460 kDa and from electron microscopy the dimensions are approximately 30 nm x 30 nm [5]. The images obtained (fig. 2) are different from those seen in electron microscopy. Although the size is reasonably close (35 nm along the longest dimension), the globular Cterminal regions, which normally appear as 6 distinct domains are not individually resolved. This suggests that these head regions, which are known to be attached to flexible stalks, have aggregated during the sample preparation. This raises important questions, such as whether increased ionic strength during deposition would help keep the domains apart, and whether different images are obtained when the protein remains in solution. We plan to use SFM to test these ideas. In addition a clear gap can be seen
1201
between the head and the stem region of the molecule. This is spanned by collagen helices which are presumably too small to image in these samples. If the scan rate is increased then the C l q can be cleaved in this region, leaving globular domains fixed to the graphite surface. Generally for our C l q images, the molecules are oriented in the same direction on the H O P G surface (fig. 2b). We have observed this orientation effect with other molecules, and hypothesize that it may be caused by preferential binding of a hydrophobic surface area on C l q to the H O P G . A C L is an important enzyme in the cellular biosynthesis of both cholesterol and fatty acids and thus A C L inhibitors may have potential as cholesterol lowering agents. It is a tetramer, showing "half of sites" reactivity, and has a molecular weight of 484000 (see references in ref. [6]). There is no three-dimensional structural information on this protein. In this case our interest was whether discrete, reproducible molecules could be imaged, and whether any substructure could be seen. (On the basis of biochemical data either two- or four-fold symmetry could be expected.) The images obtained (fig. 3) show a
Fig. 1. (a) Image of IgG on HOPG graphite. The size of the image is 45 nmx45 nm. The image was recorded on a Nanoscope II (Digital Instruments, Inc.) in air using a platinum/iridium tip. The tunnelling current was 0.5 nA and the bias voltage was 20 mV. (b) Model built from the x-ray coordinates of IgG(Dob) [1l. Several images with this characteristic shape have been recorded [3].
1202
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
molecule with d i m e n s i o n s approximately 10 n m x 14 nm, which is consistent with its m o l e c u l a r mass, a n d the images o b t a i n e d for the o t h e r two proteins. T h e large scan area (fig. 3a) shows m a n y molecules imaged at low resolution. A large n u m -
b e r of these a p p e a r to have a l o n g i t u d i n a l cleft, indicating a two-fold symmetry axis. T h e I g G a n d C l q e x p e r i m e n t s are excellent systems with which to test the quality of information derived from S T M b e c a u s e of their charac-
b i
~
i
I
_
_
Fig. 2. (a) Clq imaged on HOPG. The image size is 35 nmx35 nm, and the scanning conditions are the same as in fig. 1. The globular head domains are clearly separated from the stalk. (b) Two Clq molecules (image.size 90 nm x 90 nm). (c) A model based on electron microscopy[5] is shown for comparison.
T.N.C. Wellset al. / Imaging of proteins by scanning tunnelling microscopy
1203
Fig. 3. (a) ATP-citrate lyase imaged using a large scan area, (250 nm x 250 nm). (b) Detail of an individual molecule. Scanning conditions are the same as in fig. 1. teristic q u a t e r n a r y structure. T h e strong correspondence between the images obtained and the known structure of I g G suggest that this technique will be of great value in studying other protein systems. T h e results from C l q and A C L highlight two of the major issues facing our research. First, the question of sample preparation. W h a t effect do the substrate, the buffers, the m e t h o d of deposition have on the type of images p r o d u c e d ? H e r e it will be important to c o m p a r e the results with those obtained from electron microscopy. It will also be important to add nonprotein size markers as internal controls for the microscope calibration. Second, as can be seen in the case of A C L , reproducible images can be obtained of proteins without a highly characteristic quaternary structure. In such cases, the interpretation of such images should be carried out cautiously. S T M has a useful role in the under-
standing of protein structure, especially for relatively large multi-domain complexes. T h e experiments need submicrogram quantities of protein and high-resolution images can be r e c o r d e d in several minutes.
References [1] E.V. Silverton, M.A. Navia and D.R. Davies, Proc. Natl. Acad. Sci. USA 74 (1977) 5140. [2] R.R. Porter and K.B.M. Reid, Adv. Protein Chem. 33 (1979) 1. [3] R.J. Leatherbarrow, M. Stedman and T.N.C. Wells, J. Mol. Biol. 221 (1991) 361. [4] M. Stedman, J. Microscopy 152 (1988) 611. [5] C.J. Strong, R.C. Siegel, M.L. Philips, P.H. Poon and V.N. Schumaker, Proc. Natl. Acad. Sci. USA 79 (1982) 486. [6] T.N.C. Wells, Eur. J. Biochem. 199 (1991) 163.
