A self-assembled monolayer for the binding and study of histidine-tagged proteins by surface plasmon resonance

Article abstract of DOI:10.1021/ac9504023

This paper reports the generation of a self-assembled monolayer (SAM) that selectively binds proteins whose primary sequence terminates with a His-tag: a stretch of six histidines commonly incorporated in recombinant proteins to simplify purification. The SAM was prepared by the adsorption onto a gold surface of a mixture of two alkanethiols: one thiol that terminated with a nitrilotriacetic acid (NTA) group, a group that forms a tetravalent chelate with Ni(II), and a second thiol that terminated with a tri(ethylene glycol) group, a group that resists protein adsorption. His-tagged proteins bound to the SAM by interaction of the histidines with the two vacant sites on Ni(II) ions chelated to the surface NTA groups. Studies with model proteins showed the binding was specific for His-tagged proteins and required the presence of Ni(II) on the surface. Immobilized His-tagged proteins were kinetically stable in buffered saline at pH 7.2 but could be desorbed by treatment with 200 mM imidazole. Surface plasmon resonance studies for two model systems showed that His-tagged proteins adsorbed on the NTASAM retained a greater ability to participate hi binding interactions with proteins in solution than proteins immobilized in a thin dextran gel layer by covalent coupling.

Full text of DOI:10.1021/ac9504023

Anal. Chem. 1996, 68, 490-497
A Self-Assembled Monolayer for the Binding and
Study of Histidine-Tagged Proteins by Surface
Plasmon Resonance
George B. Sigal,† Cynthia Bamdad,§ Alcide Barberis,‡ Jack Strominger,‡ and George M. Whitesides*,†
Departments of Chemistry and Biochemistry and the Committee for Higher Degrees in Biophysics, Harvard University,
Cambridge, Massachusetts 02138
molecule of interest (for example, a protein) in this interfacial
region occurring by adsorption to the interface or association with
an immobilized ligand can be monitored as a function of time by
measuring Θm.
A variety of techniques have been used to immobilize proteins
on silver or gold films for studies using SPR: the two major classes
are (i) physical adsorption of the protein either on the metal
surface6 or on a hydrophobic film spin-coated on the metal surface7
and (ii) covalent attachment of the protein to thin functionalized
dextran gel layers3 or to layers of silica deposited on silver and
modified with (aminopropyl)trimethoxysilane.2
We and others have developed procedures to modify the
surface of gold films by formation of self-assembled monolayers
(SAMs) of alkanethiolates.8,9 One class of SAMs useful for
immobilizing proteins uses alkanethiolates terminally functional-
ized with a biotin moiety; the biotin binds streptavidin, which in
turn permits the immobilization of biotin-labeled proteins.10 This
system is successful and widely used, but shares the disadvantages
of most immobilization schemes requiring chemical modification
of a protein: (i) chemical modification may lead to denaturation
or loss of activity and (ii) the presence of multiple sites on the
protein available for modification results in loss of control over
the orientation of the protein after immobilization. Here we
describe the generation of a SAM functionalized with the nitrilo-
triacetic (NTA) group 1 ; this group, when complexed with Ni-
This paper reports the generation of a self-assembled
monolayer (SAM) that selectively binds proteins whose
primary sequence terminates with a His-tag: a stretch of
six histidines commonly incorporated in recombinant
proteins to simplify purification. The SAM was prepared
by the adsorption onto a gold surface of a mixture of two
alkanethiols: one thiol that terminated with a nitrilotri-
acetic acid (NTA) group, a group that forms a tetravalent
chelate with Ni(II), and a second thiol that terminated with
a tri(ethylene glycol) group, a group that resists protein
adsorption. His-tagged proteins bound to the SAM by
interaction of the histidines with the two vacant sites on
Ni(II) ions chelated to the surface NTA groups. Studies
with model proteins showed the binding was specific for
His-tagged proteins and required the presence of Ni(II)
on the surface. Immobilized His-tagged proteins were
kinetically stable in buffered saline at pH 7 .2 but could
be desorbed by treatment with 2 0 0 mM imidazole.
Surface plasmon resonance studies for two model systems
showed that His-tagged proteins adsorbed on the NTA-
SAM retained a greater ability to participate in binding
interactions with proteins in solution than proteins im-
mobilized in a thin dextran gel layer by covalent coupling.
Surface plasmon resonance (SPR) is a useful technique for
measuring the kinetics of association and dissociation of ligands
from proteins in aqueous solution.1-3 It is particularly useful for
processes occurring at or near interfaces. The active sensing
element is a thin (40-50 nm) gold or silver film deposited on a
glass substrate. Monochromatic, p-polarized light is reflected from
the glass-gold interface from the back side. A plot of reflected
intensity as a function of the angle of incidence (Θ) shows a
minimum (Θm) corresponding to the excitation of surface plas-
mons at the gold-solution interface.4 The value of Θm shifts with
changes in the refractive index of the interfacial region near the
surface of the gold (within approximately a wavelength of the
incident light). For thin organic films (<100 nm) and light with
a wavelength of 760 nm, the shift in Θm is roughly proportional
to the thickness of the film.5 Changes in the concentration of a
(II), selectively binds proteins whose sequence terminates with a
stretch of six histidines (His-tag).11,12 His-tags are commonly
(5) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. J. Colloid Interface
Sci. 1 9 9 1 , 143, 513-526.
(6) Flanagan, M. T.; Pantell, R. H. Electron. Lett. 1 9 8 4 , 20, 968-970.
(7) Davies, J.; Roberts, C. J.; Dawkes, A. C.; Sefton, J.; Edwards, J. C.; Glasbey,
T. O.; Haymes, A. G.; Davies, M. C.; Jackson, D. E.; Lomas, M.; Shakesheff,
K. M.; Tendler, S. J. B.; Wilkins, M. J.; Williams, P. M. Langmuir 1 9 9 4 ,
10, 2654-2661.
(8) Dubois, L. H.; Nuzzo, R. G. Annu. Rev. Phys. Chem. 1 9 9 2 , 43, 437-463.
(9) Whitesides, G. M.; Gorman, C. G. In Handbook of Surface Imaging and
Visualization; Hubbard, A. T., Ed.; CRC Press: Boca Raton, FL, 1995; pp
713-733.
† Department of Chemistry.
‡ Department of Biochemistry.
§ Committee for Higher Degrees in Biophysics.
(1) Liedberg, B.; Nylander, C.; Lundstro¨ m, I. Sens. Actuators 1 9 8 3 , 4, 299-
304.
(2) Daniels, P. B.; Deacon, J. K.; Eddowes, M. J.; Pedley, D. G. Sens. Actuators
1 9 8 8 , 16, 11-18.
(3) Lo¨ fås, S.; Johnsson, B. Chem. Commun. 1 9 9 0 , 1526-1528.
(4) Raether, H. Phys. Thin Films 1 9 7 7 , 9, 145-261.
(10) Spinke, J.; Liley, M.; Guder, H.-J.; Angermaier, L.; Knoll, W. Langmuir 1993,
9, 1821-1825.
490 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996
0003-2700/96/0368-0490$12.00/0 © 1996 American Chemical Society

incorporated into the primary sequence of recombinant proteins
to facilitate purification. The objective of this work was to develop
a method of immobilization of proteins that would (i) control the
orientation of the immobilized protein such that the active site
would be accessible to molecules in solution, (ii) create a surface
that would specifically immobilize a protein of interest while
resisting nonspecific binding of other proteins, and (iii) avoid the
requirement for nonspecific, covalent modification of the protein.
Protein adsorption experiments were carried out on SAMs
incubated sequentially in 1 mM aqueous. NaOH for 5 min, and
40 mM aqueous nickel sulfate for 1 h to adsorb Ni(II) to the
surface NTA groups. The samples were washed with ∼1 mL of
HBS, followed by ∼5 mL of water, and dried in a stream of
nitrogen. Proteins were adsorbed from solutions in HBS. To
determine binding in the absence of Ni(II) or other heavy metals
on the surface, the step involving incubation with the solution of
nickel sulfate was eliminated, and 5 mM of EDTA was added to
the protein solutions.
Surface P lasmon Resonance Measurements. SPR mea-
surements were made with a BIACore instrument (Pharmacia
Biosensor).5 Plastic cassettes holding a gold-coated glass sub-
strate derivatized with a thin carboxydextran gel layer were
purchased from Pharmacia Biosensor. For experiments using
mixed monolayers of thiols 2 and 3 , the glass substrates
supporting the carboxydextran films were removed with a razor
blade. SAMs were prepared on gold films evaporated on No. 2
coverslips. The coverslips were cut to size and glued in place on
the plastic cassettes using a two-part epoxy (5 Minute Epoxy,
Devco Corp.).
EXPERIMENTAL SECTION
Reagents. Hepes-buffered saline (HBS) is 10 mM 4-(2-
hydroxyethyl)-1-piperazineethanesulfonic acid and 150 mM NaCl
in water adjusted to pH 7.2 with NaOH. Lysozyme (EC 3.2.1.17,
grade III from chicken egg white, Sigma) was used as received.
The preparation of the His-tag containing model proteins and their
protein ligands is described elsewhere.13,14 The Gal 4 construct
used (Gal 4 1-147 + AH) comprised the Gal 4 DNA binding and
dimerization domains attached to a short activating peptide (AH,
amphipathic helix) and a His-tag. The His-tagged Gal 11 construct
used (mini-Gal 11) was a shortened form of Gal 11 that contained
a His-tag and a myc epitope near the carboxy terminus.
P reparation of Gold Substrates. Gold substrates were
prepared by evaporating 1 nm of Ti, followed by 40 nm of Au,
onto either silicon wafers (Silicon Sense) for ellipsometric
measurements or No. 2 glass coverslips (Corning Glass) for SPR
measurements. Substrates were coated with metals by electron
beam evaporation at pressures of less than 5 × 10-7 Torr and
evaporation rates of 0.2 nm/ s. Gold substrates were broken into
smaller pieces after scribing with a diamond stylus.
P reparation of SAMs. Stock solutions of thiols 2 and 3 in
(1 mM in 95% ethanol) were combined in glass scintillation vials
to give mixtures with a total thiol concentration of 1 mM. Gold
substrates were incubated between 12 and 20 h in the solutions
of thiols, rinsed with 95% ethanol, and dried in a stream of
nitrogen. SAMs used in SPR measurements were prepared from
a solution of thiols 2 and 3 containing a mole fraction of thiol 2 ,
SPR experiments were conducted with a constant 5 µL/ min
flow of solution over the surfaces. Protein adsorption on the NTA-
SAM was carried out by sequential injections of 25-35 µL of a 40
mM aqueous solution of nickel sulfate and then 35 µL of the
protein solution diluted in HBS. The surface was washed with
HBS after each injection. Protein adsorption resulted in a shift
in the resonance angle that was reported in resonance units (RU;
10 000 RU ) 1.0°). To determine binding of proteins in the
absence of Ni(II) or other heavy metals on the surface, the nickel
sulfate injection was omitted and 5 mM EDTA was included in
the protein solution. Covalent attachment of proteins to the
carboxydextran-derivatized surface after activation with N-ethyl-
N′-[(dimethylamino)propyl]carbodiimide hydrochloride in the
presence of N-hydroxysuccinimide (EDC-NHS) was carried out
according to established procedures.3
Synthesis of the NTA-Terminated Thiol 2. N-[5 -[[[(3,6 ,9-
Trioxaeicos-1 9 -en-1 -yl)oxy]carbonyl]amino]-1 -carboxypen-
tyl]iminodiacetic Acid (4). Carbonyldiimidazole (3.8 g, 2 equiv)
was added while stirring to 3.5 g (11 mmol) of undec-1-en-11-
yltri(ethylene glycol) (prepared according to ref 19) dissolved in
35 mL of methylene chloride. After stirring for 2 h, the solution
was applied to a column containing 300 g of silica gel 60 (230-
400 mesh, E. Merck) equilibrated with ethyl acetate and the
imidazole carbamate eluted with 1 L of ethyl acetate. Evaporation
of the solvent under reduced pressure gave 4.6 g (100%) of the
imidazole carbamate as an oil.
ø2 ) 0.1.
soln
X-ray P hotoelectron Spectroscopy (XP S). XPS spectra
were obtained using an SSX-100 spectrometer (Surface Sciences
Instruments). The spectra were accumulated at a take-off angle
of 35° relative to the surface and at pressures less than 1 × 10-8
Torr. Peaks were fitted and integrated using software from
Surface Science Instruments.
Ellipsometry. Ellipsometric measurements were made with
a Rudolf Research Type 43603-200E manual thin-film ellipsometer
using a He-Ne laser (632.8 nm) at an angle of incidence of 70°.
The PCSA (polarizer-compensator-sample-analyzer) configu-
ration was used with the compensator set to -45°. Ellipsometric
constants for surfaces were measured before and after adsorption
of thiols or proteins. The thickness of adsorbed layers were
calculated using a planar, three-layer, isotropic model15 with
assumed refractive indices of 1.00 for air and 1.45 for both protein
and SAM.16
Amine 1 (8.0 g, 31 mmol), prepared according to ref 11, was
dissolved in 100 mL of water. The solution was titrated to pH
10.2 with 12 N NaOH, and 130 mL of dimethylformamide was
added. The imidazole carbamate (4.5 g, 11 mmol) in 10 mL of
dimethylformamide was added dropwise to the amine while
stirring. After 12 h, the solution was added to 500 mL of water
and washed three times with 250 mL portions of ethyl acetate
(using gentle stirring to avoid the formation of an emulsion). The
aqueous phase was acidified with 6 N HCl to pH 1.5 and extracted
with ethyl acetate (4 × 250 mL). The combined extracts were
(11) Hochuli, E.; Do¨ beli, H.; Schacher, A. J. Chromatogr. 1 9 8 7 , 411, 177-184.
(12) Gentz, R.; Chen, C.-H.; Rosen, C. A. Proc. Natl. Acad. Sci. U.S.A. 1 9 8 9 , 86,
821-824.
(13) Chung, S.; Wucherfennig, K.; Freidman, S.; Hafler, D.; Strominger, J. Proc.
Natl. Acad. Sci. U.S.A. 1 9 9 5 , 91, 12654-12658.
(14) Barberis, A.; Pearlberg, J.; Simkovich, N.; Farell, S.; Reinagel, P.; Bamdad,
C.; Sigal, G.; Ptashne, M. Cell 1 9 9 5 , 81, 359-368.
(16) The refractive indices of proteins usually fall between 1.35 and 1.55. The
effect of error in the assumed refrective index on the calculated thickness
is described in detail in ref 18.
(15) McCrackin, F. L.; Passaglia, E.; Stromberg, R. R.; Steinberg, H. L. J. Res.
Natl. Bur. Stand., Sect. A 1 9 6 3 , 67, 363-377.
Analytical Chemistry, Vol. 68, No. 3, February 1, 1996 491

Scheme 1. Synthesis of the NTA-Terminated Thiol 2a
a
(a) Aqueous NaOH, 100 °C; (b) CDl, CH2Cl2; (c) water-DME, pH 10.2; (d) CH3COSH, AIBN, THF, hν; (e) NaOH, O2, water-DME;
PEt3.
washed with saturated NaCl and dried over MgSO4. The solvent
was removed under reduced pressure to give 3.6 g (54%) of olefin
5 as a hydroscopic white solid: 1H NMR (CD3OD, 400 MHz) δ
5.80 (m, 1H), 5.00 (dd, 1H), 4.92 (dd, 1H), 4.13 (t, 2H), 3.60 (br
m, 14H), 3.47 (m, 3H), 3.08 (t, 2H), 2.02 (m, 2H), 1.78 (m, 1H),
1.65 (m, 1H), 1.52 (br m, 4H), 1.28 (br m, 14H).
N-[5 -[[[[2 0 -(Acetylthio)-3 ,6 ,9 -trioxaeicos-1 -yl]oxo]car-
bonyl]amino]-1 -carboxypentyl]iminodiacetic Acid (5 ). The
olefin 4 (3.6 g, 6.1 mmol) was dissolved in 15 mL of distilled
tetrahydrofuran, together with 1.8 mL of thiolacetic acid (4 equiv)
and 300 mg of azobis(isobutyronitrile) (AIBN). The solution was
irradiated for 4 h under a 450 W medium-pressure mercury lamp
(Ace Glass). The solvent was removed under reduced pressure,
and the crude product was triturated with hexane. Recrystaliza-
tion from ethyl acetate-hexane gave 4.0 g (98%, calculated as a
pure compound) of thioacetate as a hydroscopic tan solid 5 : 1H
NMR (CD3OD, 400 MHz) δ 4.14 (t, 2H), 3.63 (br m, 14H), 3.44
(m, 3H), 3.09 (t, 2H), 2.85 (t, 2H), 2.29 (s, 3H), 1.78 (m, 1H), 1.66
(m, 1H), 1.54 (br m, 8H), 1.29 (br m, 14H).
N-[5 -[[[(2 0 -Mercapto-3 ,6 ,9 -trioxaeicos-1 -yl)oxo]carbon-
yl]amino]-1 -carboxypentyl]iminodiacetic Acid (2 ). To thio-
acetate 5 (4.0 g, 6.0 mmol) in 20 mL of dimethoxyethane was
added 17 mL of water followed by 20 mg of I2. After the addition
of 3 mL of 2 N NaOH, the solution was stirred for 4 h while O2
was bubbled through it. Addition of 100 mL of water and acidi-
fication to pH 1.5 with 6 N HCl led to precipitation of the product
as the disulfide. The disulfide was filtered, washed with water
and dried, under vacuum to give 2.9 g (78%) of a white powder.
The disulfide was reduced to the thiol 2 with triethylphos-
phine.17 To the disulfide (1.0 g, 0.80 mmol) in 36 mL of methanol
containing 4 mL of water under an atmosphere of nitrogen was
added 1.0 g (10 equiv) of triethylphosphine. The solution was
stirred for 5 h and concentrated to an oil under reduced pressure.
The residue was dissolved in 40 mL of degassed water and
acidified to pH 1.5 with 6 N HCl. The product was extracted three
times with 20 mL portions of ethyl acetate. The combined organic
phases were washed with saturated NaCl, dried over MgSO4, and
concentrated under reduced pressure to give thiol 2 as a tan
hydroscopic solid (0.84 g, 84%). Purification was achieved by
elution from a column of silica gel 60 (230-400 mesh, E. Merck)
with a gradient of 20-80% (v/ v) 2-propanol in hexane plus 2% (v/
v) acetic acid, followed by a gradient of 0-8% (v/ v) water in 80%
(v/ v) 2-propanol, 2% (v/ v) acetic acid in hexane: 1H NMR (DMSO,
400 MHz) δ 7.18 (t, 1H), 4.01 (t, 2H), 3.49 (br m, 14H), 3.34 (m,
3H), 2.91 (q, 2H), 2.44 (q, 2H), 2.21 (t, 1H), 1.51 (br m, 5H), 1.22
(br m, 17H); 13C NMR (DMSO, 400 MHz) δ 173.91, 173.14, 156.10,
70.31, 69.78, 69.71, 69.46, 68.89, 64.21, 62.96, 53.22, 40.05, 33.39,
29.28, 29.20, 29.14, 29.00, 28.86, 28.50, 27.74, 25.64, 23.75, 22.97;
HRMS-FAB [M + Na]+ Calcd for C28H51N2O11S 647.3190. Found
647.3163.
RESULTS AND DISCUSSION
The approach used in this work to generate a surface
appropriate for immobilizing His-tagged proteins for study by SPR
was to prepare a surface analogous to that used for nickel-affinity
chromatography, commonly used to purify His-tagged proteins.11,12
This technique utilized the NTA derivative 1 to coordinate Ni-
(II), leaving two vacant coordination sites on the nickel ion for
chelation by His-tag imidazole side chains. In nickel-affinity
chromatography, chelate
1 is attached to chromatography
beads through the primary amine.11 For SPR, chelate 1 was
attached to gold surfaces by formation of mixed monolayers from
two alkanethiols: the first, 2 , was terminally functionalized with
the nickel ligand; the second, 3 , was terminated with tri(ethylene
glycol)sa functional group that resists nonspecific adsorption of
protein.18
The synthesis of thiol 2 is outlined in Scheme 1. The synthesis
is analogous to the previously described synthesis of the tri-
(ethylene glycol)-terminated alkane thiol 3 ,19 with the addition of
a carbonyldiimidazole coupling step to link the NTA derivative 1
to the terminal hydroxyl group. In order to make purification
simpler, the product from deacetylation of the thioacetate was
isolated in the form of a disulfide and reduced to the thiol in a
separate step.
To determine the optimal concentration of ligand on the surface
for protein binding, mixed monolayers20 of thiols 2 and 3 were
prepared on gold films by adsorption from solutions containing a
mixture of the two thiols (at a total thiol concentration of 1.0 mM)
in 95% ethanol. To bind Ni(II) to the NTA groups, the SAMs were
(17) Overman, L. E.; Smoot, J.; Overman, J. D. Synthesis 1 9 7 4 , 59-60.
(18) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1 9 9 3 , 115, 10714-10721.
(19) Pale-Grosdemange, C.; Simon, E. S.; Prime, K. L.; Whitesides, G. M. J. Am.
Chem. Soc. 1 9 9 1 , 113, 12-20.
(20) Laibinis, P. E.; Nuzzo, R. G.; Whitesides, G. M. J. Phys. Chem. 1 9 9 2 , 96,
5097-5105.
492 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

Figure 2. Ellipsometric thickness of mixed monolayers of thiols 2
and 3 on gold. ø(2)soln is the mole fraction of thiol 2 in the solution of
thiols 2 and 3 used to prepare the mixed monolayer.
Figure 1. XPS analysis of SAM prepared from thiol 2. SAM was
treated sequentially with 1 mM aqueous NaOH, 40 mM aqueous NiII-
SO4, 150 mM aqueous NaCl, and water and dried before analysis.
treated with 1.0 mM NaOH, followed by 40 mM NiIISO4 in water,
and then washed with 150 mM NaCl and water. The SAMs were
characterized by XPS and ellipsometry. Figure 1 shows an XPS
spectrum of a SAM containing only the thiolate from 2 , after
treatment with Ni(II) and washing. Peaks are present for both
nitrogen and nickel. The integrated areas of the Ni 2p3 and N 1s
peaks, after correcting for the different photoelectric cross sections
of the orbitals,21 are consistent with the expected 2:1 stochiometry
of nitrogen to nickel, indicating that the Ni(II)-binding sites on
the surface were active. Incubation of this surface in HBS solution
for 1 h led to no observable loss of nickel (as measured by XPS
after washing the sample). Measurement of the thickness of
mixed monolayers of 2 and 3 by ellipsometry demonstrated an
approximately linear increase with the mole fraction of 2 in
solution [ø(2 )soln] and established that the ratio of 1 :2 on the
surface corresponded roughly to that in solution (Figure 2).22
Ellipsometry was used to measure the average thickness of
the protein layer that resulted from adsorption of proteins on the
mixed monolayers and to test for the selectivity of these surfaces.
As a model for a His-tagged protein, we used a three-domain
single-chain T-cell receptor construct (ABC scTCR) with a (His)6
tag incorporated near the C-terminus, and a molecular mass of
42 kDa;13 this protein was available from other studies. To test
for nonspecific binding of proteins to the negatively charged SAMs
by an ion-exchange mechanism, we measured the adsorption of
a highly positively charged model protein, lysozyme (pI ∼11).
Figure 3 shows the results of these binding studies. Surfaces
that were not pretreated with Ni(II) did not bind either protein
as long as the buffer included EDTA to scavenge metal ion
Figure 3. Ellipsometric thickness of adsorbed protein layers on
mixed monolayers of thiols 2 and 3 as a function of the mole fraction
of thiol 2. Lysozyme and the His-tag-containing protein scTCR were
adsorbed to SAMs for 1 h from 0.3 mg/mL solutions in HBS. “No
Ni(II)” refers to surfaces without chelated Ni(II), “Ni(II)” refers to
surfaces that were pretreated with a 40 mM aqueous solution of NiII-
SO4, and “Ni(II) + Imid.” refers to protein solutions containing 5 mM
imidazole that were adsorbed onto surfaces pretreated with NiIISO4.
impurities that could complex with the NTA group. The lack of
nonspecific adsorption of lysozyme indicated that adsorption of
even strongly positively charged proteins to the negatively charged
surface through charge-charge interactions was not significant
under these conditions. Pretreatment of the surfaces with
aqueous Ni(II) led to strong binding of the scTCR. The binding
increased with increasing surface concentration of the NTA group
until a maximal value was reached at ø(2 )soln ) 0.2 (presumably
from formation of a complete monolayer of protein). The binding
was concentration dependent. Half-maximal binding required an
scTCR concentration of ∼1.5 µM (Figure 4). Figure 3 shows some
binding of lysozyme was also observed when Ni(II) was present,
through histidines on the surface of the protein,23 but the thickness
of this layer was much less than observed for the scTCR. The
addition of low concentrations of imidazole to solutions of protein
enhanced the selectivity of the surfaces for His-tagged proteins:
5 mM imidazole eliminated nonspecific binding of lysozyme while
having little or no effect on the binding of the scTCR. Immobiliza-
(21) Corrections were applied by the Surface Science Instruments software
according to: Scofield, J. H. J. Electron Spectrosc. 1 9 7 6 , 8, 129-137.
(22) Since ellipsometry measures an average thickness over a macroscopic area
(∼1 mm2), the data do not rule out the possibility that the two thiols are
not homogeneously mixed on the surface. Recently, Stranick et al. have
shown, by scanning tunneling microscopy, that mixed monolayers formed
from CH3(CH2)15SH and CH3O2C(CH2)15SH phase segregate into micro-
domains with length scales of approximately 1-10 nm: Stranick, S. J.;
Parikh, A. N.; Tao, Y.-T.; Allara, D. L.; Weiss, P. S. J. Phys. Chem. 1 9 9 4 , 98,
7636-7646. Disorder in the tri(ethylene glycol) chains attached to thiols 2
and 3 could lead to better mixing of these thiols; however, the process of
phase separation in SAMs is not yet understood well enough to make
predictions with confidence.
(23) Zhao, Y.; Sulkowski, E.; Porath, J. Eur. J. Biochem. 1 9 9 1 , 202, 1115-1119.
Analytical Chemistry, Vol. 68, No. 3, February 1, 1996 493

Figure 6. SPR signal on adsorption of the His-tagged scTCR
followed by desorption with imidazole. The surface was treated
sequentially with 25 µL of a 40 mM aqueous solution of NiIISO4, 35
µL of a 0.1 mg/mL solution of the scTCR, and 25 µL of a 200 mM
solution of imidazole in HBS, pH 7.2. The surface was washed with
HBS after each injection. This cycle was repeated three times.
Figure 4. Ellipsometric thickness of His-tag-containing scTCR
adsorbed on Ni(II)-NTA-SAM as a function of the protein concentra-
tion. Solutions of the scTCR in HBS were adsorbed for 1 h on a mixed
monolayer of thiols 2 and 3 [ø(2)soln ) 0.1] that had been pretreated
with a 40 mM aqueous solution of NiIISO4.
of refraction in solution of 0.0011), followed by a slower increase
as the His-tag-modified protein bound to the surface (the formation
of a protein film with a refractive index of 1.45 will give an SPR
response of 700 RU/ nm of film thickness).25 On reinjection of
buffer into the flow cell, there was an immediate drop in the SPR
signal due to the change in refractive index, but the signal
remained elevated compared to the original baseline because of
the bound scTCR. The mass of adsorbed protein is proportional
to the difference between the SPR signal before injection of the
protein solution and the SPR signal after reinjection of buffer.5
The surface presenting Ni(II) ions bound more protein than one
that was free of Ni(II) by more than a factor of 10; this observation
confirms the requirement for Ni(II) on the surface for protein
binding.
In order to use SPR to study the interaction of bound His-
tagged proteins with molecules in solution, the rate of dissociation
of His-tagged proteins from the surface must be slow. There
should, however, also be a way to desorb His-tagged proteins
quickly from the surfaces so that they can be used in multiple
experiments. Adsorbed layers of the scTCR bound to the NTA-
Ni(II) surface were kinetically stable in HBS. After flowing HBS
over adsorbed scTCR for 1 h, the SPR signal indicated that greater
than 95% of the protein remained on the surface. Despite the
stability of the adsorbed protein layers in HBS, the NTA surface
was easily regenerated by treatment with 200 mM imidazole, pH
7.2, for 5 min. Three cycles of adsorption of the scTCR followed
by desorption with imidazole are shown in Figure 6. About 90%
of the adsorbed protein was dissociated from the surface during
the first treatment with imidazole. In subsequent cycles, each
treatment with the scTCR left the same total mass of protein on
the surface, but the SPR signal due to protein remaining on the
Figure 5. SPR signal on injection of the scTCR His-tag protein over
a mixed SAM comprising NTA (2) and (EG)3OH (3) groups. The solid
line is the signal after pretreatment with 25 µL of 40 mM aqueous
NiIISO4 before injection of the protein. The dashed line is the signal
in the absence of Ni(II) on the surface.
tion of the scTCR was reversed by high concentrations of
imidazole: treatment of the adsorbed protein layer with 200 mM
imidazole in HBS, pH 7.2, for 30 min led to complete dissociation
of the bound protein from the surface.
For SPR studies, we used SAMs prepared with a low mole
fraction of thiol 2 [ø(2 )soln ) 0.1] to ensure that adsorbed proteins
would not be densely packed on the surface.24 Figure 5 shows
that the adsorption of the scTCR can be followed by SPR under
conditions of continuous flow and that the adsorption requires
Ni(II) to be present on the surface. Solutions of the scTCR were
passed over two surfaces: one had been pretreated with a solution
of NiIISO4; the other had no metal ions bound to the NTA groups.
Upon injection of protein-containing solution over each surface,
there was a rapid response due to differences in the refractive
indices of the buffer and protein solution (1000 RU is a change in
resonance angle of 0.1° and corresponds to a change in the index
(25) The theoretical SPR response to changes in the index of refraction of the
bulk liquid and to deposition of thin protein films was determined by
calculating the reflection of p-polarized light from a stratified, planar, isotropic
structure, as described by: Azzam, R. M. A.; Bashara, N. M. Ellipsometry
and Polarized Light; North-Holland: New York, 1977. The model used
consisted of three layers with finite thicknesses (gold, SAM, and protein)
between two semiinfinite media (glass and solution). The indices of refraction
for the gold (0.17 + 4.93I), glass (1.511), and water (1.329) layers were
taken from ref 5. We used an index of refraction of 1.45 for both the SAM
and protein layers (see ref 16).
(24) Adsorbed proteins should not be densely packed as long as the NTA-
terminated thiolates are randomly distributed or phase segregated into
protein size microdomains (see ref 22). Another possibility that we cannot
rule out is that the thiolates phase segregated into domains much larger
than a protein. This possibility could lead to islands of densely packed
proteins surrounded by areas free of protein.
494 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

Scheme 2. Procedures for Immobilizing Proteins on Surfaces To Test for the Ability of Immobilized
Proteins To Interact with Ligands in Solutiona
a
(a, top) Receptors are immobilized through the interaction of a His-tag group with Ni(II) chelated to an NTA-SAM. (b, bottom)
Receptors are immobilized by covalent coupling to a carboxylated dextran gel layer. In this procedure, treatment of the surface with
EDC in the presence of NHS leads to formation of NHS-activated esters on the surface. Proteins are immobilized on this surface by
reaction of protein amino groups with the activated esters. Ethanolamine (EA) is added to quench unreacted NHS esters.
surface after each treatment with imidazole gradually increased.
We have not yet identified the origin of this apparent irreversibility
and heterogeneity in binding.26 It may be possible to reduce the
Table 1. SPR Signal from Adsorption of Proteins to
Ni(II) on NTA-SAMa
protein
His-tag
conc (mg/ mL)
∆RU
amount of residual protein further by increasing the concentration
of imidazole or by increasing the time of treatment; we have not,
however, optimized this procedure.
ABC scTCR
Gal 11
hu TBP
TFIIB
Gal 4
Gal 4
yes
yes
yes
yes
yes
no
0.30
0.05
0.23
0.23
0.25
0.50
1928
673
2265
1344
2685
47
To ensure that His-tagged proteins other than the scTCR
bound well to nickel on the NTA-SAM, we tested four other His-
tagged proteins that were available from other studies. These
included human TATA box binding protein (huTBP), the tran-
scriptional activator Gal 4, and two components of the yeast RNA
polymerase II holoenzymesTFIIB and Gal 11.14 Gal 4 without a
His-tag group was also available and was tested as a control to
ensure that binding of Gal 4 was through the His-tag. Table 1
lists the results of the binding experiments. All the His-tag labeled
proteins bound well. The Gal 4 construct without a His-tag did
not bind.
a
ø(2 )soln ) 0.10. Surface was pretreated with a 40 mM aqueous
solution of NiSO4 for 7 min and then with protein solution in HBS for
7 min.
We used SPR to compare the effect of the immobilization
technique on the ability of immobilized proteins to interact with
proteins in solution. As shown in Scheme 2, His-tagged proteins
were immobilized either by adsorption to Ni(II) on an NTA-SAM
or by chemically coupling to a carboxylated dextran surface after
activation with EDC in the presence of NHS.27 We tested
(26) Some possible explanations for the heterogeneity in the dissociation of
adsorbed scTCR by imidazole include the following: (i) some fraction of
adsorbed protein may have bound to multiple Ni(II) ions and been less
sensitive to treatment with imidazole, (ii) there may have been a strongly
adsorbing impurity in the scTCR preparation, or (iii) some denaturation of
protein may have occurred on the surface.
(27) This procedure is commonly used to immobilize proteins on the carboxy-
dextran-modified substrates sold by Pharmacia for use with the BIACore
SPR instrument (see ref 3 for details).
Analytical Chemistry, Vol. 68, No. 3, February 1, 1996 495

Figure 8. Interaction of mini-Gal 11 with SRB2 and a monoclonal
antibody (myc-R) detected by SPR when mini-Gal 11 was immobilized
by metal affinity on a Ni(II)-containing NTA-SAM but not detected
when Gal 11 was covalently immobilized on a carboxydextran gel
layer. (a, top) His-tag mini-Gal 11 (0.05 mg/mL) was immobilized on
the Ni(II)-NTA-SAM by two consecutive injections because of the
low protein concentration. SRB2 (0.25 mg/mL) and myc-R bound to
the immobilized protein as evidenced by significant increases in the
SPR signal. (b, bottom) Mini gal 11 was covalently attached to the
carboxydextran surface as described in Figure 6. Although more
protein was immobilized in the dextran gel layer than on the SAM,
the protein immobilized on the SAM was not recognized by SRB2 or
myc-R. The large refractive index changes during the protein injections
are due to the presence of glycerol in the stock protein solutions.
Figure 7. SPR-detected enhanced binding to the His-tagged
scTCR, by two monoclonal antibodies (C1 and âF1) recognizing
different epitopes, when the scTCR was immobilized on a Ni(II)-
NTA-SAM compared to when the scTCR was immobilized by covalent
coupling to a carboxydextran gel layer. (a, top) Plot shows four
sequential injections of solutions over the NTA-SAM: NiIISO4 (40 mM
aqueous solution, 35 µL), scTCR (0.3 mg/mL, 35 µL), C1 (0.2 mg/mL
35 µL), and âF1 (0.2 mg/mL, 35 µL). The surface was washed with
HBS after each injection. The proteins were diluted in HBS containing
20 mM imidazole. (b, top) Plot shows five sequential injections over
the carboxydextran surface: an aqueous solution containing 200 mM
EDC and 50 mM NHS (15 µL), scTCR (0.3 mg/mL, 35 µL),
ethanolamine (1 M, 35 µL), C1 (0.2 mg/mL, 35 µL), and âF1 (0.2 mg/
mL, 35 µL). The proteins were diluted in HBS.
to both C1 and âF1 added sequentially (∆RUC1/ ∆RUscTCR ) 0.37,
∆RUâF1/ ∆RUscTCR ) 0.20). In contrast, Figure 7b shows that when
the scTCR was chemically coupled to a carboxylated dextran
surface only a small fraction of the protein attached to the surface
antibodies against two epitopes on the scTCR for their ability to
bind to scTCR immobilized by the two techniques (Figure 7).
Monoclonal antibody C1 is specific for an epitope of the Vâ17 TCR
gene segment present in the scTCR, while the monoclonal
antibody âF1 recognizes an epitope present in the constant region
of the â-chain (called Câ).13 Comparison of the SPR signal
(∆RUscTCR) on immobilization of the scTCR to the Ni(II)-NTA
and carboxylated dextran surfaces indicates that much more
scTCR immobilized to the dextran layer. The dextran surface
exhibited a higher immobilization capacity because the thick
(∼100 nm) hydrogel layer can incorporate the weight equivalent
of many monolayers of protein,5 while the Ni(II)-NTA-SAM can
adsorb a maximum of one full monolayer. Despite the larger
immobilization capacity of the dextran surface, scTCR adsorbed
on the Ni(II)-NTA-surface was better able to bind antibodies in
solution (as indicated by the ratio ∆RUAb/ ∆RUscTCR, which is
proportional to the percentage of immobilized scTCR molecules
recognized by the antibodies). Figure 7a shows a large portion
of the scTCR immobilized to Ni(II) on the surface was reactive
could bind to either C1 or âF1 (∆RUC1/ ∆RUscTCR ) 0.03, ∆RUâF1
/
∆RUscTCR ) 0.11). Either the dextran matrix or the transforma-
tions involved in covalent coupling appeared to mask the epitopes
recognized by these antibodies. A negative control antibody
(myc-R) did not bind to the scTCR on either surface (data not
shown). This observation confirmed that the measured interac-
tions were specific. Possible explanations for the differences in
behavior for the two systems include the following: (i) the
covalent immobilization procedure may have led to chemical
modification of the antibody epitopes or to changes in the three-
dimensional structure of the scTCR, (ii) the scTCR may have been
linked to the dextran matrix in an orientation that made the
epitopes inaccessible to the antibodies, or (iii) the dextran matrix
itself may have interfered with the ability of the antibodies to
access the scTCR.
496 Analytical Chemistry, Vol. 68, No. 3, February 1, 1996

Similarly, mini-Gal 11,14 produced with a His-tag at the amino
terminus, was immobilized either through Ni(II) to the NTA-SAM
or by EDC-NHS coupling to the carboxylated dextran surface
(Figure 8a,b, respectively). SRB2, another component of the yeast
RNA polymerase II holoenzyme,28 and myc-R, an antibody to an
epitope included in the mini- Gal 11 construct,14 bound to Gal 11
immobilized through Ni(II) to the NTA-SAM (∆RUSRB2/ ∆RUGal 11
) 2.80, ∆RUmyc-R/ ∆RUGal 11 ) 0.71) but not to Gal 11 chemically
covalent attachment to a carboxylated dextran layer. The surface
of the gold modified by inclusion of thiolate 1 allows immobiliza-
tion of protein containing His-tag with a higher percentage of
protein recognizable by antibodies and other proteins than can
be achieved by covalent modification.29
This method of immobilization may also be useful for other
bioanalytical techniques that require or are compatible with
proteins immobilized on metal surfaces, for example: interfer-
31
immobilized to the carboxylated dextran surface (∆RUSRB2
∆RUGal 11 ) 0.00, ∆RUmyc-R/ ∆RUGal 11 ) 0.02).
/
ometry,30 surface acoustic wave sensing,
ellipsometry,32 am-
perometric detection,33 and electrochemiluminescence.34
CONCLUSIONS
ACKNOWLEDGMENT
Many proteins of biological interest are prepared by overex-
pression in cells. We have taken advantage of a modification
commonly included in the constructs encoding these proteins to
aid in their purification (His-tag) to develop a method to im-
mobilize them on the surface of gold for study by SPR. This
method offers advantages over methods that have historically been
used for immobilizing proteins on metal surfaces, including
This study was supported by NIH (GM 30367, CA 47554, GM
22526) and ONR/ ARPA (N00014-93-1-0498). C.B. was supported
by
a training program in Molecular Biophysics (MRSA
2T32GM08313-06). We thank Prof. Don Wiley (Dept. of Bio-
chemistry and Molecular Biology, Harvard University) for the use
of his Biacore instrument, Dr. Steve Friedman (Dept. of Medicine,
Cornell University Medical College) for providing the C1 mono-
clonal antibody, Luisella Barberis-Maino (Dept. of Bochemistry
and Molecular Biology, Harvard University) for technical as-
sistance, and Prof. Mark Ptashne (Dept. of Biochemistry and
Molecular Biology, Harvard University) for helpful discussions.
(28) Koleske, A. J.; Young, R. A. Nature 1 9 9 4 , 368, 466-469.
(29) The Ni-NTA surface we describe here has already been applied in probing
the interactions between an array of defined proteins and protein fragments
involved in gene regulation in yeast (ref 14). Without exception, a strong
correlation was found between the strengths of the interactions observed
in vitro and the known biological activity of the molecules. These data
NOTE ADDED IN PROOF
indicate that, at least within the range of the interaction strengths tested
-7
(10-6-10
M), the surface can be used to quantitate certain protein-
Subsequent to the submission of this paper, Gershon and
protein interactions accurately. More experience with the method is,
however, required to evaluate its generality.
Khilko published a report describing the preparation of a car-
boxydextran film with immobilized NTA groups for use in SPR
studies.35
(30) Schlatter, D.; Barner, R.; Fattinger, Ch.; Huber, W.; Hubscher, J.; Hurst, J.;
Koller, H.; Mangold, C.; Muller, F. Biosens. Bioelectron. 1 9 9 3 , 8, 109-116.
(31) Dahint, R.; Grunze, M.; Josse, F. Anal. Chem. 1 9 9 4 , 66, 2888-2892.
(32) Steibel, Ch.; Brecht, A.; Gauglitz, G. Biosens. Bioelectron. 1 9 9 4 , 9, 139-
146.
(33) Janata, J.; Bezegh, A. Anal. Chem. 1 9 8 8 , 60, 62-74.
(34) Gudibande, S. R.; Kenten, J. H.; Link, J.; Friedman, K.; Massey, R. J. Mol.
Cell. Probes 1 9 9 2 , 6, 495-503.
Received for review April 26, 1995. Accepted October 16,
1995.X
AC9504023
(35) Gershon, P. D.; Khilko, S. J. Immunol. Methods 1 9 9 5 , 183, 65-76.
X Abstract published in Advance ACS Abstracts, December 1, 1995.
Analytical Chemistry, Vol. 68, No. 3, February 1, 1996 497