Anal. Chem. 1999, 66,1635-1638
isas
Optical Determination of Surface Density in Oriented Metalloprotein Nanostructures Hun-Gi Hong, Paul W. Bohn,' and Stephen G. Sligar Beckman Institute and Departments of Chemistry and Biochemistry, University of Illinois a t Urbana-Champaign, 405 North Mathews, Urbana, Illinois 61801
The immobilization of proteins on a variety of solid substrates has been extensively studied for a wide range of applications, some of which include biosensors, protein chromatography, biomaterials, and biomedical analysis.'-3 Covalent binding of functional proteins to a surface can result in strong, stable linkages and high surface coverage, both of which are important in the development of functionalproteinbased devices. Originally, the cross-linking reagents synthesized for the preparation of multisubunit enzymes and protein conjugates in solution4 were also used for immobilization of enzymes on solid supporta.5 For example, Bhatia and co-workers reported that high Surface coverages were obtained by using thio-terminal silanes and heterobifunctional cross-linkers for immobilization of IgG on silica surfaces? These linker chemistries have served well in the preparation of randomly oriented protein monolayers and multilayers. We are interested in oriented arrays of mutants of cytochrome b5 and other metalloproteins as models for the fabrication of biomaterials with novel electrooptic properties.7 Oriented arrays are obtained by introducing,via site-directed mutagenesis,a unique cysteineresidue on the water-accessible surface of a protein which lacks such residues in the wildtype variant. By moving the site of the mutation around the surface, the orientation of the heme protein is controlled. Characterization of the orientation is possible from polarized absorption measurements on the x-y degenerate Soret transition.8 We have obtained large surface coverages by sequentially reacting amino-terminal silane (3-APS) and N-succinimidyl-6-maleimidocaproate (EMCS)onto glass, as shown in Scheme I. One important property of these supermolecular arrays is the protein surface coverage. Collinson and Bowdeng reported that the surface coverage of electrostatically adsorbed cyctochrome c on tin oxide could be estimated by direct visible absorption measurements, assuming identical molar absorptivities for solution and adsorbed species. Although this assumption may hold for some structures, it is clearly inappropriate for oriented arrays, inasmuch as the solutionvalues are determined from randomly oriented species. In the current work we circumvent this
* Author to whom correspondence should be addressed.
(1) (a) Mosbach, K. Methods in Enzymology; Academic Press: New York, 1988; Vol. 137. (b) Wingard, L. B., Jr.; Katchalaki-Katzir, E.; Goldstein, L. Immobilized Enzyme Principles; Academic Press: New York, 1976. (2) (a) Jennissen, H. P. Ber. Bunsenges. Phys. Chem. 1989,93,948. (b) Porath, J. Biotechnol. Prog. 1987, 3, 14. (c) Mizutani, T. J. Liq. Chromatogr. 1985, 8, 925. (3) (a) Sevastianov,V.I.Crit.Rev.Biocompat.1988,4,109. (b)Ivaraean, B.; Lundstrom, I. Crit. Reu. Biocompat. 1986, 2, 1. (c) Brash, J. L. Makromol. Chem. Suppl. 1985, 9,69.(4) (a) Das, M.; Fox, F. C. Annu. Rev. Biophys. Bioeng. 1979,8, 165. (b) Freeman, R. B. Trends Biochem. Sci. 1979,4,193. (c) Kitagawa, T.; Shimozono, T.; Aikiwa, T.; Yoshida, T.; Nishimura, H. Chem. Pharm. Bull. 1981, 29, 1130. (5) Mattiasson, B. J. Appl. Biochem. 1981, 3, 183. (6) Bhatia,S. K.;Shriver-Lake,L.C.;Prior,K. J.;Georger, J. H.;Calvert, J. M.; Bredehorst, R.; Ligler, F. S. Anal. Biochem. 1989, 178, 408. (7) (a) Stayton, P. S.; Olinger, J. M.; Jiang, M.; Bohn, P. W.; Sligar, S. G. J. Am. Chem. SOC. 1992, 114, 9298. (b) Gouterman, M. In The Porphyrins; Dolphin, D., Ed.; Academic Press: New York, 1978. (8)(a) Cropek, D. M.; Bohn, P. W. J.Phys. Chem. 1990,94,6452. (b) Hughes, K. D.; LaBuda, M. J.; Bohn, P. W. Appl. Opt. 1991,30,4406. (9) Collinson, M.; Bowden, E. F. Anal. Chem. 1992, 64, 1470. 0003-2700/93/0365-1635$04.00/0
difficulty by noting that iron-porphyrin proteins in solution have been assayed spectrophotometrically after conversion to pyridine hemochromes.lOJ1 By comparing the total adsorbance obtained from direct absorption measurements of oriented metalloprotein layers on Si02 a t the Soret resonance (410 nm in cytochrome b5) to the total number density of surface protein, obtained from subsequent pyridine hemochrome assay (PHCA) analysis, the apparent surface molar absorptivity is obtained directly. In this correspondence we report the use of the PHCA to determine the surface molar absorptivity for oriented arrays of cytochrome b5 mutants. The heme is completely dissociated from the surface cytochrome b5 and converted to the reduced pyridine hemochrome in solution. Subsequently, the absorbance of reduced species in solution is determined colorimetrically. From the correlation of the absorbance of reduced hemochrome to the standardcurve obtained from pyridine hemochrome assay of solution cytochrome65, the surface concentration is estimated. Thus, the PHCA provides a quantitative measure of the surface number density of the oriented protein layers for any particular sample, and correlation with absorption measurements permits calibration of the surface absorption cross section for any class of samples fabricated from a particular mutant. This combined technique provides a simple method to obtain the surface density and effective molar absorptivity of metalloproteins.
EXPERIMENTAL SECTION Reagents and Materials. The usual substrate for the coating of heme proteins is a 150-pm-thickTi-Zn glass coverslip (Corning Glass). (3-Aminopropyl)trimethoxysilanewas purchased from Hula America (Piscataway, NJ), and N-succinimidyl6-maleimidocaproate was purchased from Fluka (Ronkonkoma, NY). Toluene was dried over CaHz before use. Deionized water (18.2 Ma cm) was obtained from a Milli-Q system with an Organex-Q final stage (Millipore,Bedford, MA). Phosphate buffer solution (20 mM, pH 8.0) was used for preparation of the various mutant cytochrome b5 solutions. Mutant and wild-type variants of cytochrome b5 were prepared by de novo gene synthesis followed by overexpression in Escherichia coli, as described previously.78 Water-soluble forms of all mutants were obtained by removing, at the oligonucleotide synthesis step, the sequence coding for the lipophilic anchor of the in vivo protein. Cleaning of Substrates and Silanization. Glass coverslips were cleaned by immersion in a hot 1:4mixture of concentrated hydrogen peroxide and ammonia for 10 min, followed by rinsing several times with deionized water. Next, the substrates were consecutively treated with hot, concentratedsulfuric acid for 30 min twice and rinsed with and sonicated in deionized water. Finally, the substrates were dried in flowing filtered NP. Preparation of Oriented Protein Arrays. In a glovebag under Nz atmosphere, the substrates were placed in a 2 % solution of 3-APS in dry toluene for 3-6 h at room temperature. The substrates were removed from the solution and washed in consecutive portions of dry toluene and absolute ethanol. The silanized glass substrates were treated with 2 mM EMCS in absolute ethanolfor 2-8 h in a dry Nz atmosphere. The substrates were removed, rinsed with fresh absolute ethanol, and stored in (10) Paul, K. G.; Theorell, H.; Akeson, A. Acta Chem. Scand. 1953, 7,
1284.
(11) Berry, E. A.; Trumpower, B. L. Anal. Biochem. 1987, 161, 1.
0 1993 American Chemical Society
1696
ANALYTICAL CHEMISTRY, VOL. 65, NO. 11. JUNE 1, 1993
Scheme I. Immobilization Steps f o r Covalent Attachment of Cytochrome b5Using a Heterobifunctional Linker on Glass
0.25 mL of NaOH (1 M) were added together and mixed. This mixture solution was repeatedly pipeted onto both sides of substrates derivatized with cytochrome bs. A 2-mL sample of this solution was divided into two 1-mL-volumequartz cuvettes, which were placed in the sample and reference beam paths of the spectrophotometer. After baseline measurement, ca. 50 mg of dithionite was added to the sample cell to form the reduced complex, and the difference spectrum of the reduced minus oxidized pyridine hemochrome was obtained and used for quantitation.
RESULTS AND DISCUSSION
I
360
380
400 Wavelenglh(nrn)
420
440
Flgure 1. Vlslble absorption spectra of the T8C mutant of cytochrome b, in the reglon of the Soret band in phosphate buffer SOlUtiOn (solid line);the immoblllzedT8C on glass covenlip (dotted line); after pyridine hemochrome assay of the lmmobllized T8C on glass coverslip (dashed Ilne). The two spectra at the bonom were obtained from the identical sample of Immobilized cytochrome bs beforeand after treatment with
the pyridine reagent. ethanol. Wild-type and the T8C mutant of cytochrome bi were dilutedto35-50PMwith2mMphosphatehuffersolution.These
solutions were used directly for the immobilization of wild-type and T8C cytochrome bs to the substrate. A 100-rLaliquot of the concentrated (1.6 mM) T65C solution was diluted in 1 mL of phosphate buffer and treated with 1mg of dithiothreitol (DTT). Next, the T65C solution was loaded on a P4 gel filtration column (Biorad),which was equilibrated with phosphate buffer including 1mMEDTAprior touse, toseparateT65C monomerfromDTT. The T65C-containing band was retrieved from the column and diluted to 40 r M with phosphate buffer. This solution was used immediately for the immobilization of the T65C mutant. Glass substrates treated with silane and cross-linker were washed with 20 mM phosphate buffer (pH 8.01, and the lower half of each substrate was immersed in the phosphate buffer solutions of cytochromebiatroom temperature. Typical immersed area was 14.4 cmz, and incubation times varied from 2 to 60 h. After incubation,thesubstrates were rinsed extensivelywith phosphate buffer to remove physisorhed protein. After being dried with nitrogen, parallel in situ absorption and PHCA measurements were made to determine the amount of surface-bound protin. Instrumentation. Visible absorption spectra were acquired in transmission mode at normal incidence using a computercontrolled double-beam grating UV-visible spectrophotometer (Cary 3, Varian Associates). The spectral resolution was 3 nm, and the data interval was 1.5 nm. Each spectrum was obtained by signal averaging 100 scans, with a resulting p-p noise level of 2 x 10-4 absorbance unit. The total time required for the acquisition of each spectrum was ca. 4 min. The spectrum of a glass substrate, silanized with APS and EMCS but not exposed to cytochrome bi, was used for baseline correction. Pyridine Hemochrome Assay. A 2-mL aliquot of 20 mM potassium phosphate buffer (pH 8.0), 0.5 mL of pyridine, and
The surface reactions in the immobilization step are summarized in Scheme I. The 3-APS treatment provides free amino groups on the surface t o which the succinimide group of heterohifunctional cross-linker EMCS was covalently attached through an amide bond. The resulting pendant maleimide residue of EMCS was used to covalently bind the reactive thiol of the unique cysteine residue of mutant cytochrome b5 through a thioetber linkage. The maleimide functionality provides an excellent synthetic route to immobilized cytochromes due to its inherent stability and specificity to thiols. Gregory12 reported that the rate of hydrolysis of maleimide to the maleamic acid is negligible below pH 7.5, and maleimide was reported quite reactive to the thiolate anion in weak basic buffer according to Partis and eo-workers.13 Thus, the combination of an aminoterminalsilaneand the EMCS heterobifunctionallinkeroffera a convenient method for covalent attachment of cysteinecontaining proteins to silica surfaces at high concentration. Figure 1 shows visible absorption spectra of a mutant of cytochrome b5 (T8C) derivatized on SiOz. In general, the oxidized cytochrome b5 in buffer solution (solid line) shows the characteristic Soret band at 415 nm with an extinction coefficient of 130 mM-' ~m-1.1~ This strong absorption band is blue-shifted slightlyto405 nm when bound on Si02(dotted line). Ferricytochrome bs in its native state is a 6-coordinate, low-spin heme protein, in which the position and intensities of ita characteristic Soret absorption band are sensitive to the conformational state of the protein.15 Consistent with the findings of Chottard and c o - ~ o r k e r sthis , ~ observed ~~ blueshift might be due to slight modification in the heme crevice induced by adsorPtion.l6 (12) Gregory. J. D. J. Am. Chem. Soe. 1955, 77, 3922. 113) Partis, M. D.; Griffiths, D. G.; Robert, G.C.; Beeehey, R. B. J . Protern Chem. 198.3.2.263. (14) VonBodman. F.S.;Schuler, M. A.;Jollie,D.R.;Sligar,S.G.Prae. Notl. Aead. Sci. U.S.A. 1986.83.9443, (15) la) Chottard, G.; Michelon, M.: Herve, M.; Herve, G . Bioehim. Biophys. Acto 1987,916,402. (b)Smith, D. W.; Williams, R.J. P. S t r u t . Bond. 1970. 7. 1. 1161 Independent surface Raman scattering measurements in our laboratory suggest that the heme is in the high-spin s t a t e after prolonged exposure to air. This result argues for the distortion of the hene packet by weakening of the sixth ligand, an axial histidine. However, the fact that anisotropic absorption is still obtained from these layers indicates that the heme has not become completely deligated Y . Cong, S. T. Wollman, and P. W. Bohn, unpublished results.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 11, JUNE 1, 1993
1-Soh
10x1
Table I. Comparisons of the Absorbances Obtained before (A405) and after (Ass,) Treatment of the Metalloprotein Structures with the Pyridine Reagent
TBCI
reaction time (h) APS EMCS Cytbs
8-
3 3 4 7
500
520
1637
540
560
580
600
Wavelength(nm)
Flgurr 2. Vlslble absorptlon spectra of the reduced pyrldine h e m e chromes of T8C in buffer soiutlon phase (solM line) and immobliired (dotted line).
Collinson and Bowdeng reported the difference between spectroscopic and electroactive surface coverages of electrostatically adsorbed cytochrome c on tin oxide electrodes by assuming identical molar absorptivities for solution and adsorbed cytochrome c. In the general case such an assumption for adsorbed vs solution proteins is not justified, although it may be shown to be reasonable for individual protein adsorbates on a case-by-case basis. In particular it is not appropriate for our oriented samples, since the solution absorption cross sections are determined from an inherently randomly oriented sample. Furthermore, surface coverages estimated by electrochemical methods are not generally applicable to proteins: many proteins are not electroactive, and of those that are, some component of the surface population may be nonelectroactive due to improper orientation, denaturation, or other surface-specific effects. Use of the PHCA to calibrate the surface absorption cross section for the heme moieties in oriented samples removes these difficulties. In the PHCA, heme is completely dissociated from the adsorbed cytochrome and converted to pyridine hemochrome in basic pyridine buffer. This formation of pyridine hemochrome is completed within 1min in phosphate buffer.11 By measuring the absorption spectrum of the reduced pyridine hemochrome, the total surface coverage of the immobilized cytochrome bg can be calculated directly independent of the cross section of cytochromein the adsorbed state. In addition, the PHCA relies on chemical modification of a robust protoporphyrin moiety, whereas any surface measurement is alwayssubject to specific adsorption-induced protein structural effects. The absorption spectrum after PHCA treatment (dashed line) of Figure 1 shows that the Soret absorption band due to the heme moiety of the adsorbed cytochrome b5 is completely removed after treatment with the pyridine solution. This fact indicates that the heme is completely dissociated from the adsorbed cytochrome. Figure 2 shows two visible absorption spectra obtained from pyridine hemochrome assay of adsorbed and solution cytochrome b5. These two spectra show the same basic features. The absorption peaks at 557 nm are due to the reduced pyridine hemochrome. When these peak intensitiesare compared to reduced pyridine hemochrome of standard cytochrome solutions with known concentration, the total surface coverage of the adsorbed cytochrome bg can be directly calculated. Then, going back to the Soret absorbance, as displayed in Figure I, the surface absorption cross section can be determined. With the surface densities calibrated using PHCA, the apparent molar absorptivities of the adsorbed protein can be
2 6 2 11
2 18 45 24
Adon (Air,i)
T8C
T65C
WT
0.0005 0.0017(0.0012) 0.0018(0.0011) 0.0015 0.0054(0.0041) 0.0012(0.0009) 0.0067(0.0062) 0.0013(0.0006) 0.0033(0.0027) 0.0008
calculated. Table I lists a set of four different preparation conditions optimized for the preparation of the T8C mutant. It is clear that the T8C coverage grows as a function of immersion time in the protein-containing solution and that at the longest reaction times efficient surface coverage is obtained. Simple manipulation of Beer’s law reveals that the absorbance, A, is given by A = ur~2.303 (1) where u is the absorption cross section and r is the surface density. The measured quantity (1 - T , is approximately equal to the absorbance, in the low-absorbance limit. The absolute number of molecules present on the surface is measured in the PHCA, and the geometric surface area is known for a given sample. Thus, r can be calculated and the corresponding absorbance of the surface before hemochrome treatment used to calculate the absorption cross section. Surface coverages can then be calculated in fractions of a monolayer by comparison to the full tight-packed monolayer density. rsat, the saturated monolayer surface density of 1.1 X 1013 cm-2 is obtained by inverting the 900-A2 surface footprint of the globular cytochrome b5.l’ Use of data from the last row of Table I for the T8C mutant gives an absorption cross section of 6.79 Az. Use of the relationship between absorption cross section in Az and molar absorptivity in M-1 cm-I, e
(M-’ cm-’) = 2.61 X
lo4 u (A2)
(2) yields a value of 1.77 X lo5 M-l cm-l. Similar calculations can be performed for other entries in the table; however,since the reactions were only optimized for the T8C mutant, and because no precautions, such as sonicating periodically, were taken to minimize the amount of physisorbed wild-type protein these values do not have a direct physical interpretation, other than as an average for a specific sample. It is interesting to note that the solution value for the molar absorptivity for cytochrome bs at 405 nm is 1.30 X 105 M-1 cm-l. However, solution values naturally average over all possible orientational configurations. It is expected that the oriented samples under investigation here would exhibit a larger apparent molar absorptivity, if the orientation places the plane of the heme preponderantly in the plane of the substrate. Since linear dichroism measurementa show that this is precisely the case for the T8C mutant,18the direction of change of the measured molar absorptivity in the surface layer is as expected.
CONCLUSIONS The pyridine hemochrome assay can be used to determine surface coveragevalues for heme-containing metalloproteins (17) Mathews, F. S.;Levine, M.; Argos, P. J.Mol. Biol. 1972, 64,449. (18) Polarization modulation linear dichroism measurements reported in ref 7a show that the T8C mutant is oriented with the heme plane making an angle larger than 54.7O relative to the surface normal. Recent measurements using the optical waveguide linear dichroism scheme developed in ref 8 show the angle to be 59.5 f 1.3O for T8C and 48.5 f 3.8’ for T65C: H.-G. Hong and P. W. Bohn, unpublished data.
1658
ANALYTICAL CHEMISTRY, VOL. 65, NO. 11, JUNE 1, 1993
covalentlybound to surfaces from submonolayerto multilayer coverage. While direct visible absorption measurements have previously required assumptions about the effective molar absorptivity for adsorbed cytochromes, the direct measurement of surface coverage makes calibration of the surface molar absorptivity possible. Once this is measured for a given mutant in a given orientation, the much simpler direct absorption measurement can be used to give surface coverage. Finally, the change from solution to adsorbed molar absorptivity is in the expected direction for an oriented protein
layer with the heme plane oriented mostly parallel to the surface, i.e., perpendicular to the propagation direction.
ACKNOWLEDGMENT The authors acknowledgethe support of the Biotechnology Research and Development Corporation and the Department of Energy through Grant DE FG02 88ER13949.
RECEIVEDfor review November 23, 1992. Accepted February 18, 1993.