A Robust Method for Determining DNA Binding Constants Using Capillary Zone Electrophoresis

ANALYTICAL BIOCHEMISTRY ARTICLE NO.

263, 72–78 (1998)

AB982791

A Robust Method for Determining DNA Binding Constants Using Capillary Zone Electrophoresis Chunze Li and Lenore M. Martin Department of Biomedical Sciences, College of Pharmacy, University of Rhode Island, 41 Lower College Road, Kingston, Rhode Island 02881-0809

Received January 13, 1998

Capillary zone electrophoresis (CZE or CE) with online UV detection was utilized to measure the binding constants between purified calf thymus DNA and a library of designed tetrapeptides which had been constructed using unnatural amino acids with thiazole ring side chains. Mixtures containing a constant amount of a tetrapeptide, the neutral marker (mesityl oxide), and varying concentrations of DNA were prepared and equilibrated at 8°C for 12 h. CE was then utilized to separate unbound tetrapeptides from the DNA–peptide complex. The UV absorbance of the peak representing unbound tetrapeptide decreased incrementally as a result of increasing the concentration of DNA in the equilibrium mixture. The absorbance of the peak corresponding to the unbound tetrapeptide was obtained directly from the electropherogram and used in the calculation of the DNA–peptide binding constants. The binding constant for each tetrapeptide to calf thymus DNA was obtained from the negative slope of a Scatchard plot and a comparison of the binding constants for different peptides showed that the tetrapeptides in the library have DNA-binding affinities ranging from 102 to 106 M21. © 1998 Academic Press Key Words: capillary electrophoresis; DNA; binding constants; peptides; ethidium bromide; thiazoles.

To enhance our ability to screen combinatorial peptide libraries for DNA-binding affinity, we found it valuable to develop a fast and efficient method for the determination of DNA binding constants using capillary electrophoresis (CE).1 In this paper, we introduce a new CE method, distinct from affinity capillary electrophoresis (ACE), which we used to screen for DNA-binding activity in a 1

Abbreviations used: CE, capillary electrophoresis; ACE, affinity capillary electrophoresis; BOC, tert-butyloxycarbonyl; EOF, electroendoosmotic flow; bs, binding sites; bp, basepair; HF, hydrogen fluoride. 72

library containing 15 synthetic tetrapeptides. We have adapted an experimental procedure used for the quantification of a stable DNA–ligand complex so that the stability of the complex during the analysis is no longer a requirement for our assay. Under our conditions, the macromolecule (DNA) comigrates with the electroendoosmotic flow. If the DNA– ligand complex dissociates during a run, the intensity of the ligand peak still reflects the concentration of free ligand at time zero, since the ligand migrates toward the detector faster than does either the DNA or the DNA– ligand complex. We focus on the analysis of the ligand to be screened, in a background of macromolecule, and no buffer additives are required. In our experiments, we observed that the peak areas of cationic tetrapeptides decreased as a result of increasing the amount of purified calf thymus DNA present in the initial equilibrium mixtures. By switching the polarity of the electrodes during electrophoresis, the method should also work for negatively charged ligands. ACE represented the adaptation to CE of a technique used to study macromolecule–ligand interactions in gels. Using slab gel electrophoretic methods such as “electrophoretic retardation” (1) or “affinity electrophoresis” (2), many DNA-binding proteins have been successfully identified. Since the charge-to-mass ratio of a DNA–ligand complex is generally quite different from that of the unbound DNA and ligand separately, the complex migrates separately from the free ligands. Compared with traditional gel electrophoresis, ACE exhibited short analysis time, low-volume sample requirements, a high efficiency, and convenience in quantification using the on-column UV detector (3). The diverse applications of CE in the analysis of a very wide range of structurally different compounds, including small biomolecules, makes it a more practical approach for the study of molecular interactions than slab gel electrophoresis. 0003-2697/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

DETERMINING DNA BINDING CONSTANTS

There are two methods currently used for ACE, each based on the stability of the complex. If a complex is fairly stable during the time required for electrophoresis (typically 10 min or more), then assessment of binding constants may be performed as follows (i.e., if the half-life for dissociation, t1/2 5 ln 2/k21, then more than one-half of the complex will dissociate during the run if k21 . 0.07 s21): Mixtures with different ratios of DNA and ligand are loaded into the capillary and run using capillary electrophoresis to separate the free ligand from the DNA–ligand complex (4). As the intensity of the ligand peak decreases, that of the complex peak generally increases. The binding constants may be calculated from either the decreases in the ligand peak or the increases in the complex peak intensities. This method is ideal for strong binding (.108 M21) but is not generally applicable. If the complex dissociates very quickly compared with the time required for electrophoresis, binding constants will not be accurately measured via the first method. Instead, the intensity of the DNA-ligand complex peak will decrease due to dissociation as the complex migrates through the capillary. Hummel and Dryer pioneered a method, later widely adapted for CE (5–12), which prevents the dissociation of the complex during the separation by including the macromolecule in the running buffer. Weaker binding constants may be determined using this method, based on the change in the migration time of the sample as the concentration of macromolecule in the buffer is increased. This method is widely used in CE, but requires a large quantity of the macromolecule to obtain adequate data. In addition, incorporating the macromolecule as a component of the buffer may lead to unexpected effects on the electrophoretic separation. Our method of screening for DNA-binding activity in a peptide library is different from the two methods described above because it does not depend upon knowledge of binding kinetics nor does it depend upon a detailed knowledge of the nature of the peptide–DNA complex. MATERIALS AND METHODS

Apparatus A Dionex capillary electrophoresis instrument (CES-1) with both positive and negative voltage sources and an on-column UV and fluorescence detector was purchased from Dionex Corp. (CA). Fused silica capillaries of 50 mm i.d. were obtained from Polymicro Technologies Inc. (Phoenix, AZ). The capillary was initially conditioned by washing it with 1 N sodium hydroxide for 30 min, followed by a 15-min wash with 0.1 N sodium hydroxide. Then it was extensively rinsed with deionized water and running buffer before the capillary was put into use. Samples were injected using a gravity injection method with a relative height difference of 50 mm between the

73

SCHEME 1. The design of the experiment. The sample is injected at the right end of a 50-mm capillary using gravity (hydrodymanic) injection, and then the capillary end is placed into the anode buffer solution and the voltage is applied. All samples in an uncoated capillary (negatively charged surface) migrate toward the detector and the anode (apparent flow, app) due to the electroendoosmotic flow (eof). The eof is usually greater than the electrophoretic mobility (ep) at pHs greater than 4. The order of migration seen in the electropherogram (view from the detector) will then correspond to the order shown, left to right: cations, then neutral species, followed by anions.

inlet and outlet vials for 10 s. The electropherograms were recorded and analyzed using a Model 1022 Personal Integrator (Perkin–Elmer Corp., CT). Water (18.3 MV) was obtained from a Millipore MilliQ water-purification system (Millipore Corp., MA) (Scheme 1). Chemicals Tris–HCl, Tris base, mesityl oxide (used as a neutral marker in CE), and calf thymus DNA (double-stranded, 10 mg/mL, sonicated to 580–830 bp, purified by phenol/ chloroform extraction and ethanol precipitation) were obtained from Fluka Chemie AG (Switzerland). Potassium chloride was from Sigma (St. Louis, MO), and magnesium chloride was from Fisher Scientific (NJ). Argon gas (prepurified grade) was from Linde Specialty gases and nitrogen gas was from Corp Brothers, Inc. (RI). The tetrapeptides used were synthesized in this lab using the solid-phase method (13, 14) on a 4-methylbenzhydrylamine resin (15) and the unnatural amino acids D-3(4-thiazolyl)alanine 1 and L-3-(4-thiazolyl)alanine 2 (obtained from Synthetech, Inc., Albany, OR) (Fig. 1) on a simultaneous multiple peptide synthesizer (16). Peptide syntheses were carried out using the tert-butyloxycarbonyl (BOC) protecting group strategy; BOC deprotection and amino acid coupling reactions were monitored by the ninhydrin reaction (17). The high hydrogen fluoride (HF) method of cleavage of the product peptides from the solid support (18) was performed without adding any scavenger. The purity of the peptide products was confirmed to

74

LI AND MARTIN

be greater than 90% by RP-HPLC (C18, Vydac, 5 mm, 0.1% trifluoroacetic acid in H2O/CH3CN) and capillary zone electrophoresis (15 mM Tris (1.6 g Tris–HCl/L and 1.6 g Tris base/L), 10 mM MgCl2, 1 mM dithiothreitol, 30 mM KCl, pH 7.9, at 20°C), and products were identified by 1H NMR analysis. The crude products were used for CE binding studies without further purification. Tetrapeptide No. 4, LDDD-NH2, was not available in sufficient quantities for testing. The sequences of the tetrapeptides used to test our CE method are shown in Table 1. Procedure Mixtures containing different ratios of the tetrapeptides to purified calf thymus DNA (42% GC) (19) were prepared in a buffer of 1.5 mM Tris and Tris–HCl, 3.0 mM KCl, and 1.0 mM MgCl2 at pH 7.9 and were equilibrated at 8°C (to prevent bacterial growth) for 12 h. Each sample in a series contained the same amount of a given tetrapeptide (6.2 3 1027 to 5.7 3 1028 M) and a neutral marker (mesityl oxide, 0.01%, v/v), and the concentration of DNA was varied (from 1.75 3 1024 to 8.76 3 1023 M). In a typical experiment the stock solutions were prepared as follows: the peptides were dissolved in water at a concentration of 1 mg/mL, neutral marker was mixed with water (1 mL in 1 mL), DNA stock solutions were prepared ranging from 10 mg/mL to 0.1 mg/mL, and the running buffer was diluted (1 mL into 7 mL). Samples were prepared by mixing several microliters of the peptide stock solution with either dilute buffer or a known concentration of DNA dissolved in dilute buffer to achieve identical peptide concentrations in the incubation mixture. CE was utilized to separate the unbound tetrapeptides from the DNA–tetrapeptide complexes at room temperature (;20°C) based on their different electrophoretic mobilities (Figs. 2 and 3). The conditions for electrophoretic separation of each peptide from its

FIG. 1. The chemical structures of the two amino acid monomers used to construct the enantiomeric peptide library.

TABLE 1

The Stoichiometric Binding Constants of the 15 Tetrapeptides Calculated from the Scatchard Plotsa ID No.

Peptide sequence

Ka(1) (M21)b

Ka(2) (M21)c

12 9 11

LLLL-NH2 DLLL-NH2 LLDL-NH2

2.1 3 106 4.2 3 105 1.9 3 105

2.1 3 106 2.0 3 103 1.0 3 103

8 15 7 2 6 3 14 10 13 16

LDLD-NH2 LLLD-NH2 LDLL-NH2 DDDL-NH2 DDLD-NH2 LDDL-NH2 DLLD-NH2 DLDL-NH2 DLDD-NH2 LLDD-NH2

5.5 3 104 5.2 3 104 2.8 3 104 2.4 3 104 2.3 3 104 2.0 3 104 1.8 3 104 1.5 3 104 1.4 3 104 1.4 3 104

1.3 3 103

1 5

DDDD-NH2 DDLL-NH2

2.5 3 103 1.7 3 103

Note. The peptides are ordered so that Ka(1) ranges from highest affinity to lowest. Peptides are separated into three different categories, having high, medium, and low binding constants, respectively. a Nonlinear Scatchard plots were not fitted. Instead, the data was observed to be biphasic and was assumed to represent two hypothetical binding sites having high and low affinities. b Ka(1) is the stoichiometric equilibrium binding constant near saturation. c Ka(2) is the binding constant at high [DNA]/[peptide] ratios.

DNA–peptide complex, using an uncoated capillary, were as follows: length to the detector (LD), 70.51 cm; total capillary length (LT), 75.81 cm; applied voltage, 30 kV; measured current, 80 mA. The running buffer (15 mM Tris and Tris–HCl, pH 7.9 at 20°C, 10 mM MgCl2, 30 mM KCl), was chosen for comparison with CE studies of the binding of transcription factors with DNA (20, 21). The concentration of the buffer used in the incubation mixture was 10 times less than that of the running buffer to facilitate sample stacking during electrophoresis. Sample stacking sharpens analyte bands by slowing the mobility at the front of the sample band when the sample makes the transition into the running buffer of higher ionic strength. All buffer solutions were degassed via sonication and vacuum, and were filtered through 0.22-mm-pore-size sterile filters prior to analysis. Since the sample size was small (10 ml), the temperature at which the binding constants were sampled was effectively 20°C. The concentration of unbound tetrapeptide in each sample was quantified by UV absorbance at maximum absorbance (238 nm) in the detector of the CE. The binding constant of each tetrapeptide was determined by plotting the results of a series of electrophoretic runs. Each sample was analyzed in duplicate to check the reproducibility of the method.

75

DETERMINING DNA BINDING CONSTANTS

The method was tested using ethidium bromide (1.20 3 1023 M) and calf thymus DNA at concentrations ranging from 1.25 3 1024 to 1.25 3 1022 M, for comparison with the results from our peptides. RESULTS

By convention (22), the net migration of all species in a positive polarity capillary zone electrophoresis experiment is toward the cathode in an uncoated capillary. Our peptides were positively charged at pH 7.9, and so the elution of the peptides past the detector was quite fast, because it represented the sum of the electrophoretic mobility toward the cathode and electroendoosmotic flow (EOF). High EOF is observed in uncoated capillaries at buffer pHs greater than 4, due to the unmasking of negative charges on the wall of the capillary. At pH .4, the velocity of the EOF becomes greater than most electrophoretic migration velocities and dominates the separation. Neutral molecules do not have any electrophoretic mobility and thus migrate at the same velocity as the EOF. Anions such as DNA and the peptide–DNA complexes actually electrophoretically migrate against the EOF (toward the an-

FIG. 3. Electropherogram showing the peptide 12 peak intensity decrease which occurs after equilibration with purified calf thymus DNA, plotting the absorbance at 238 nm (the maximum absorbance of the peptide) as a function of migration time to the detector. The tops of the peptide peaks are aligned to facilitate comparison of peak heights. The CE runs were performed at 30 kV with a field strength of 396 V/cm at pH 7.85 in 15 mM Tris–HCl, 30 mM KCl, 10 mM MgCl2.

ode and the injector), but in our experiments they were dragged along by the EOF toward the cathode as well. The molar concentrations of unbound tetrapeptides in equilibrium mixtures with DNA were readily determined using the UV absorbance intensity of the peptide peaks in the electropherograms. The amount of unbound tetrapeptide remaining after equilibration with DNA in a given mixture was calculated using the height of the peak. In CE, peak heights are more universally quantifiable than peak areas due to the varying speeds (mobilities) at which different analytes pass through the detector. DNA molarity is expressed per base pair using the extinction coefficients determined by LePecq and Paoletti for calf thymus DNA (23). The extinction coefficient of the tetrapeptides was determined to be 6495 cm21 M21 at the wavelength of maximum absorptivity (238 nm). The binding constants were then determined from the best-fit lines on our plots (Fig. 4) (24 –26). Data Analysis T p 1 DNAbs º Complex

FIG. 2. Electropherogram showing an injection of DNA alone (above), peptide 7 alone (middle), and DNA in a complex with peptide 7 (below) monitored at 260 nm (resulting in decreased sensitivity). The CE runs were performed at 30 kV with a field strength of 396 V/cm at pH 7.85 in 15 mM Tris–HCl, 30 mM KCl, 10 mM MgCl2.

Equation [1] shows the ratio of the concentration of a peptide–DNA complex to the concentration of unbound binding sites (bs) on duplex DNA, and the concentration of unbound tetrapeptide at equilibrium. Ka 5

[Complex] [DNAbs]f [T p]f

[1]

76

LI AND MARTIN

In our method, measurements include an oft-neglected part of the sigmoidal binding curve, the region where high concentrations of ligand are present in the mixture relative to the macromolecule. We begin with ligand:DNA ratios which ensure the saturation of all possible DNA-binding sites and proceed to add DNA until no further change in the ligand peak height is observed. Although the experiments are performed at ligand concentrations near saturation, the concentration of unoccupied DNA base pairs was found to be much greater than the concentration of ligand, and so [DNAbp]free was substituted for ([DNAbp]free 2 [Tp]b). To prepare a variation of the Scatchard plot for the estimation of the binding constant of each DNA–tetrapeptide complex using our CE method, the Eq. [3] was developed. r/[DNAbp]free 5 K a 2 K a r FIG. 4. Scatchard plots of peptides 3 LDDL (diamonds), 5 DDLL (circles) 6 DDLD (triangles), and 11 LLDL (squares) showing the results using the CE method for high-affinity 11, medium-affinity 3 and 6, and low-affinity 5 binding interactions.

If n is the number of consecutive DNA bases covered by the ligand (binding site size), then [DNAbp]free 5 n[DNAbs]free and Ka is the equilibrium binding constant (association constant) for the complexation reaction at any one type of binding site. Due to the possibility of the presence of multiple types of binding site, the binding site size was arbitrarily assumed to be one base pair of duplex DNA (n 5 1) to facilitate quantitation of the DNA using UV absorbance and due to the absence of additional structural information about the complex. By the law of mass action, the net molar concentration of all DNA–peptide complexes ([Complex]) is equal to the concentration of bound tetrapeptide ([Tp]b), and the net concentration of bound tetrapeptide in all DNA–peptide complexes is equal to the total tetrapeptide concentration ([Tp]total) less the concentration of free tetrapeptide ([Tp]f) remaining at equilibrium. The ratio r is calculated by dividing the amount of bound tetrapeptide [Tp]b 5 ([Tp]total) 2 ([Tp]f) by the total amount of peptide added to the mixture ([Tp]total). Reversible binding of any tetrapeptide to DNA may then be expressed as Eq. [2]. r5

K a[DNAbp]f @T p# b 5 @T p# total ~1 1 K a[DNAbp]f!

[2]

The concentration of free DNA base pairs in the equilibrium mixture was set equal to the total concentration of DNA (base pairs) added to the mixture minus the amount of bound DNA in the complex (equal to [Tp]bound).

[3]

The ratio, r, of bound ligand to the total ligand concentration, was plotted vs r divided by the free DNA concentration (Fig. 4). The stoichiometric binding constant Ka was calculated from the slope of the curve fit (slope 5 2Ka). DISCUSSION

Upon examination of the electropherogram shown in Fig. 2, it is easy to understand why we recommend the ligand peak for quantification purposes. Under the conditions used in this method, calf thymus dsDNA comigrates with both the neutral marker and the peptide–DNA complex. The DNA was mainly propelled along the capillary by the EOF. This is a reasonable result, since no sieving matrix is used in the capillary buffer to prevent the DNA from being swept along. We kept the tetrapeptide concentrations constant and varied the DNA concentrations in our equilibrium mixtures, because it is more informative to compare the binding affinities of different tetrapeptides in a library when identical DNA solutions are added to several peptides. This is the reverse of the usual manner of determining binding constants at low ligand saturation of the DNA, and our approach is particularly suited to CE, where the ligand must be at a concentration where it can be quantitated in the detector. Due to the migration speed of the tetrapeptide toward the detector (approx 1 cm/min faster than the EOF), the tetrapeptide peak intensity is unaffected by any dissociation of the complex which might occur during a run, since the tetrapeptides liberated by dissociation would not catch up to the original tetrapeptide peak. Based on Beer’s Law, using the extinction coefficient of the tetrapeptides (6495 cm21 M21 at 238

DETERMINING DNA BINDING CONSTANTS

nm), and the inner diameter of the capillary (50 mm), the concentration of free tetrapeptide ([Tp]f) was calculated directly from the UV absorbance of the tetrapeptide peak (Fig. 3) and was the average of duplicate CE runs. The total concentration of each tetrapeptide ([Tp]total) was determined from the absorbance of the tetrapeptide peak measured in runs performed without the addition of DNA. The extinction coefficient used to determine DNA concentrations (per base pair) was 14,000 cm21 M21 (at 260 nm), which is the average of the extremes of the measured extinction coefficients for calf thymus DNA from a large number of measurements (23). The UV absorbance at 260 nm of a diluted (1:100) solution of purified calf thymus dsDNA used in our experiments was measured to be 1.75 (path length was 1 cm), and the concentration of the stock calf thymus dsDNA solution was calculated to be 1.25 3 1022 M in base pairs. The DNA stock solution was diluted in buffer and used for binding constant determinations without further characterization. Since it was difficult to predict an appropriate DNA concentration range to obtain binding measurements without prior knowledge about the order of magnitude of the binding constants and the number of binding sites, the range we used in developing this assay was much larger than needed. In addition, the salt concentration of the equilibration buffer was quite low to achieve sample stacking, and the DNA binding curves were often biphasic. Sample stacking is probably unnecessary in most cases. It is recommended that the running buffer instead be used for the equilibration step, to eliminate any nonspecific ionic binding and this adaptation is more likely to produce a single binding curve. The data we obtained within the appropriate concentration range for each peptide was not dense enough to determine the binding constants with the highest degree of accuracy. However, using these crude data, we could estimate the binding constants with a degree of confidence, the results were highly reproducible, and the method was suitable for the screening a library of potentially DNA-binding compounds. Our method yielded a single, low-affinity binding constant (1 3 104 M21) for ethidium bromide, which is in agreement with the results from prior studies of the effect of salt on DNA binding (23). Representative Scatchard plots for several of the tetrapeptides are shown in Fig. 4. The plots showing biphasic slopes for some peptides may indicate the presence of true secondary (low-affinity) ionic-type binding sites or simply represent the effects of the presence of multiple primary (high-affinity) binding sites (27). The curve in the Scatchard plots can also be an artifact produced by slower equilibration in the

77

presence of high concentrations of DNA. An overestimation of the free ligand concentration at higher DNA concentrations will cause a dip in the Scatchard plot. The binding constant results (Table 1) indicate that singly charged tetrapeptides had moderate DNA-binding affinities under the experimental conditions. The binding constants for complex formation between the 15 peptides from the tetrapeptide library and purified calf thymus DNA ranged from 102 to 106 M21. In summary, a quick method for screening peptide libraries for DNA-binding affinity with simultaneous determination of their stoichiometric equilibrium binding constants using CE is described herein. Each CE run was complete in less than 10 min, and a physiologically relevant buffer solution necessary to maintain the DNA in a double-stranded state was directly injected into the CE. The average error was less than 0.1%, and the maximum observed variation between repeat runs was 10% for a single data point. The absorbance intensities of the peptides used in these experiments ranged from 4.0 3 1023 to 3.7 3 1024 AU, corresponding to peptide concentrations ranging from 6.2 3 1027 to 5.7 3 1028 M. The resulting binding (association) constants between the tetrapeptides and calf thymus DNA fell in the range from 102 to 106 M21. The applicability of our method to the determination of both strong and weak DNA binding constants, as well as slowly and quickly exchanging complexes, makes it unique. REFERENCES 1. Lane, D., Prentiki, P., and Chandler, M. (1992) Microbiol. Rev. 56, 509 –528. 2. Horejsi, V., and Ticha, M. (1986) J. Chromatogr. A 376, 49 – 67. 3. Baker, D. (1994) Capillary Electrophoresis, Wiley, New York. 4. Heegard, N. H. H. (1994) J. Chromatogr. A 680, 405– 412. 5. Kraak, C. J., Busch, S., and Poppe, H. (1992) J. Chromatogr. A 608, 257–264. 6. Honda, S., Taga, A., Suzuki, K., Suzuki, S., and Kakahi, K. (1992) J. Chromatogr. A 597, 377–382. 7. Chu, Y.-H., Avila, L. Z., Biebuyck, H. A., and Whitesides, G. M. (1992) J. Med. Chem. 35, 2915–2917. 8. Avila, L. Z., Chu, Y.-H., Blossey, E. C., and Whitesides, G. M. (1993) J. Med. Chem. 36, 126 –133. 9. Chu, Y.-H., Lees, W. J., Stassinopoulos, A., and Walsh, C. T. (1994) Biochemistry 33, 10616 –10621. 10. Gomez, F. A., Avila, L. Z., Chu, Y.-H., and Whitesides, G. M. (1994) Anal. Chem. 66, 1785–1791. 11. Chu, Y.-H., Avila, L. Z., Gao, J., and Whitesides, G. M. (1995) Acc. Chem. Res. 28, 461– 468. 12. Karger, B. L., Chu, Y.-H., and Foret, F. (1995) Annu. Rev. Biophys. Biomol. Struct. 24, 579 – 610. 13. Merrifield, R. B. (1963) J. Am. Chem. Soc. 85, 2149 –2154. 14. Merrifield, R. B. (1986) Science 232, 341–347.

78

LI AND MARTIN

15. Pietta, P. G., Cavallo, P. F., Takahashi, K., and Marshall, G. R. (1974) J. Org. Chem. 39, 44 – 48. 16. Vagner, J., Kocna, P., and Krchnak, V. (1991) Pept. Res. 4, 284 –288. 17. Sarin, V. K., Kent, S. B. H., Tam, J. P., and Merrifield, R. B. (1981) in The Seventh American Peptide Symposium (Rich, D. H., and Gross, E., Eds.), pp. 221–224, Pierce Chemical Company. 18. Sakakibara, S., Shimonishi, Y., Kishida, Y., Okada, M., and Sugihara, H. (1967) Bull. Chem. Soc. Jap. 40, 2164 –2167. 19. Schildkraut, C. L., Marmur, J., and Doty, P. (1962) J. Mol. Biol. 4, 430. 20. Xian, J., Harrington, M. G., and Davidson, E. H. (1996) Proc. Natl. Acad. Sci. USA 93, 86 –90.

21. Rodriguez-Campos, A., Azorin, F., and Portugal, J. (1996) Biochemistry 35, 11177–11182. 22. Grossman, P. D., Colburn, J. C., Lauer, H. H., Nielsen, R. G., Riggin, R. M., Sittampalam, G. S., and Rickard, E. C. (1989) Anal. Chem. 61, 1186 –1194. 23. LePecq, J.-B., and Paoletti, C. (1967) J. Mol. Biol. 27, 87–106. 24. Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660. 25. Klotz, M. I., and Hunston, L. D. (1971) Biochemistry 10, 3065– 3069. 26. Klotz, M. I. (1982) Science 217, 1247–1249. 27. McGhee, J. D., and Hippel, P. H. V. (1974) J. Mol. Biol. 86, 469 – 489.