Anal. Chem. 2004, 76, 4343-4348
Thermodynamic Analysis of Cyclosporin A Binding to Cyclophilin A in a Lung Tumor Tissue Lysate Michael Z. Wang,† Jagat T. Shetty,† Brandon A. Howard,‡ Michael J. Campa,‡ Edward F. Patz, Jr.,‡ and Michael C. Fitzgerald*,†
Department of Chemistry, Duke University, Box 90346, Durham, North Carolina 27708, and Department of Radiology, Duke University Medical Center, Durham, North Carolina 27710
We report on the application of SUPREX (stability of unpurified proteins from rates of H/D exchange) to the analysis of a protein-ligand binding interaction under the ex vivo solution conditions of a human lung tumor tissue lysate. A SUPREX-derived binding free energy (i.e. ∆∆Gf value) and dissociation constant (i.e., Kd value) were determined for the binding of cyclosporin A (CsA) to a cyclophilin A (CypA) sample in which the protein was a component of a tissue lysate derived from fresh frozen lung tumor. The ∆∆Gf and Kd values determined by SUPREX for CsA binding to CypA in this unpurified protein sample, 4.7 ( 0.8 kcal/mol and 77 ( 17 nM, respectively, were comparable to the those obtained when SUPREX was used to analyze the binding of CsA to a highly purified CypA sample, 4.2 ( 1.0 kcal/mol and 32 ( 20 nM, respectively. Moreover, the SUPREX-derived Kd values determined in this work were both in the range of those previously reported for the CypA-CsA complex. The results of this proof-of-principle work validate the extension of SUPREX to the thermodynamic analysis of proteins and protein-ligand binding interactions in the unpurified, ex vivo conditions of human tissue lysates, and they represent the first Kd measurement on a proteinligand complex under such conditions Methods for the thermodynamic analysis of protein-ligand binding interactions typically rely on the use of various optical spectroscopies or the use of calorimetric methods. Such methods generally require large amounts of a highly purified protein sample. Thus, before the thermodynamic properties of a particular protein-ligand complex can be studied, a successful protein purification strategy must be devised. At best, this is a laborious and time-consuming exercise. The inability of conventional methods to provide for the thermodynamic analysis of proteinligand binding interactions in unpurified samples is not only an experimental inconvenience, but it has meant that most proteinligand binding affinities have been measured under solution conditions lacking the proteins, nucleic acids, and other molecules naturally found in biological samples. Therefore, the potential impact of such additional molecules on the thermodynamic * Corresponding author. Tel: 919-660-1547. Fax: 919-660-1605. E-mail:
[email protected]. † Department of Chemistry. ‡ Department of Radiology. 10.1021/ac049536j CCC: $27.50 Published on Web 06/26/2004
© 2004 American Chemical Society
properties of protein-ligand binding interactions has been difficult to assess. We have recently developed a new technique, termed SUPREX (stability of unpurified proteins from rates of H/D exchange), that can be used to make thermodynamic measurements of a protein’s conformational stability and to assess the binding affinity of protein ligands.1-7 The SUPREX technique exploits the H/D exchange properties of proteins and relies on a matrix-assisted laser desorption/ionization (MALDI) readout to evaluate solution-phase folding free energies (i.e., ∆G values) of protein folding reactions. By making such ∆G value determinations in the presence and in the absence of a protein ligand, the SUPREX technique can be used to evaluate solution-phase binding affinities (i.e., Kd values) of protein-ligand complexes.4,6,7 One advantage of the MALDI readout in SUPREX is that it enables the analysis of small amounts (picomole quantities) of both purified and unpurified proteins.2,8 Therefore, the SUPREX technique provides a means by which to measure the binding affinities of protein ligands under solution conditions that include the proteins, nucleic acids, and other molecules that are naturally found in biological samples. Described here is the application of SUPREX to characterize the binding interaction of a protein-ligand complex under the ex vivo solution conditions of a human lung tumor tissue lysate. The model protein-ligand complex in this study was the complex formed between the 165-amino acid protein cyclophilin A (CypA) and the immunosuppressive drug cyclcosporin A (CsA). CypA has been shown to have peptidyl-prolyl cis-trans isomerase activity and was recently shown to be overexpressed in lung tumors.9-11 The binding of CsA, a cyclic peptide from fungus, to purified forms (1) Ghaemmaghami, S.; Fitzgerald, M. C.; Oas, T. G. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 8296-8301. (2) Powell, K. D.; Fitzgerald, M. C. Anal. Chem. 2001, 73, 3300-3304. (3) Powell, K. D.; Wales, T. E.; Fitzgerald, M. C. Protein Sci. 2002, 11, 841851. (4) Powell, K. D.; Ghaemmaghami, S.; Wang, M. Z.; Ma, L. Y.; Oas, T. G.; Fitzgerald, M. C.J. Am. Chem. Soc. 2002, 124, 10256-10257. (5) Powell, K. D.; Wang, M. Z.; Silinski, P. S.; Ma, L.; Wales, T. E.; Dai, S. Y.; Warner, A. H.; Yang, X.; Fitzgerald, M. C. AUTHOR, PLEASE PROVIDE THE NAME OF THE JOURNAL. 2003, 496, 225-232. (6) Powell, K. D., Fitzgerald, M. C. Biochemistry 2003, 42, 4962-4970. (7) Ma, L.; Fitzgerald, M. C. Chem. Biol. 2003, 10, 1205-1213. (8) Ghaemmaghami, S.; Oas, T. G. Nat. Struct. Biol. 2001, 8, 879-882. (9) Fisher, G.; Wittmannliebold, B.; Lang, K.; Kiefhaber, T.; Schmid, F. X. Nature 1989, 337, 476-478. (10) Takahashi, N.; Hayano, T.; Suzuki, M. Nature 1989, 337, 473-475. (11) Campa, M. J.; Wang, M. Z.; Howard, B.; Fitzgerald, M. C.; Patz, E. F. Cancer Res. 2003, 63, 1652-1656.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004 4343
of CypA has been previously characterized, and Kd values in the range of 30-200 nM have been measured for the binding of CsA to purified sample preparations of CypA using conventional binding experiments that employed titration and direct detection by either radioactivity or by fluorescence.12 As part of this work, we report SUPREX-derived Kd values for the binding of CsA to CypA in two different samples, one in which the protein was highly purified and one in which the protein was a component of a tissue lysate derived from a fresh frozen lung tumor. EXPERIMENTAL SECTION Materials. The purified CypA used in this study was purchased as a lyophilized white solid from either Sigma-Aldrich Co. (St. Louis, MO) or Affinity Bioreagents, Inc. (Golden, CO). The CypA protein in this purified sample was purified from a recombinant DNA-based preparation of the protein. CsA was obtained from Sigma-Aldrich Co.. The molecular weight standards, bovine trypsinogen and bovine ubiquitin, were purchased from Sigma. The lung tumor cell lysates were prepared as previously described.11 The fresh frozen samples were obtained following surgical resection for lung cancer. All patients enrolled in the study signed an informed consent approved by our Institutional Review Board. Briefly, a 10-mg portion of tissue was washed in a phosphate-buffered saline solution, combined with 70 µL of mammalian protein extraction reagent (Pierce, Rockford, IL), crushed with a plastic pestle, and shaken for 30 min at 4 °C. Cellular debris was removed by centrifugation, and the supernatant was recovered and used for analysis. The deuterated exchange buffers (20 mM sodium phosphate, pD 7.4) were prepared using deuterium oxide (99.9% atom D), sodium deuterioxide (40 wt % in D2O, 99.9% atom D) (Aldrich, Milwaukee, WI), phosphoric acid-d3 (99% atom D) (Cambridge Isotope Laboratories, Inc., Andover, MA), and deuterated guanidinium chloride (GdmCl). Guanidinium chloride was purchased from EM Science (Merck KgaA, Darmstadt, Germany). Deuterated guanidinium chloride was prepared by repeated dissolution and lyophilization of protonated guanidinium chloride in D2O until the calculated deuterium content was >99%. The concentration of guanidinium chloride in each H/D exchange buffer was determined with a Baush & Lomb refractometer as previously described.13 pH measurements were performed on a Jenco microcomputer pH-Vision 6072 pH meter equipped with a Futura calomel pH electrode from Beckman. A pH correction for the isotope effect was made by adding 0.4 pH unit to each pH value measured.14 Trifluoroacetic acid (TFA) was from Halocarbon, and acetonitrile (MeCN) and methanol (MeOH) were from Fisher. Sinapinic acid (SA) was obtained from Sigma. Mass Spectrometry. All MALDI-TOF mass spectra were acquired on a Voyager DE-Pro Biospectrometry workstation (Applied Biosystems) in the linear mode using a nitrogen laser (337 nm). SA, prepared in 72% MeCN, 0.1% TFA as a saturated aqueous solution, was used as the matrix in this study. All mass (12) (a) Handschumacher, R. E.; Harding, M. W.; Rice, J.; Drugge, R. J. Science 1984, 226, 544-547. (b) Harding, M. W.; Handschumacher, R. E. Transplantation 1988, 46, 29S-35S. (c) Holzman, T. F.; Egan, D. A.; Edalji, R.; Simmer, R. L.; Helfrich, R.; Taylor, A.; Burres, N. J. Biol. Chem. 1991, 266, 2474-2479. (d) Lui, J.; Albers, M. W.; Chen, C.-M.; Schreiber, S. L.; Walsh, C. Proc. Natl. Acad. Sci., U.S.A. 1990, 87, 2304-2308. (13) Nazaki, Y. Methods Enzymol. 1972, 26, 43-50. (14) Glasoe, P. K.; Long, F. A. J. Phys. Chem. 1960, 64, 188-190.
4344
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
spectra were collected in the positive ion mode using an acceleration voltage of 25 kV, a grid voltage of 23.5 kV, a guide wire voltage of 18.75 V, and a delay time of 650 ns. Each mass spectrum obtained was the sum of 32 laser shots. The raw data in each mass spectrum were smoothed using a 19-point Gaussian smoothing routine supplied by the instrument manufacturer’s software package before mass calibration using proteins of known mass as internal calibrants (i.e., bovine trypsinogen MW 8565 and bovine ubiquitin MW 23 979). SUPREX Analyses. SUPREX analyses were performed on both purified CypA taken from a 14 µM stock solution and unpurified CypA contained in the biological samples from the lung tumor cell lysates described above. The purified and unpurified CypA protein samples in this study were also subject to SUPREX analyses in the presence of CsA. The CsA was prepared as a concentrated stock solution in DMSO and added to the H/D exchange buffers just prior to the addition of protein. The final concentration of CsA in each exchange buffer was 90 µM, and the concentration of DMSO was less than 10% v/v. In each SUPREX analysis, 1.0-µL aliquots from either the purified CypA stock solution or the cell lysate containing the unpurified CypA were diluted 11-fold into a series of H/D exchange buffers containing increasing concentrations of the chemical denaturant, GdmCl. The resulting solutions were each incubated at room temperature and allowed to exchange for the same amount of time. After this time, the H/D exchange reactions were quenched by adding 1.6 µL of 3% TFA in H2O. The specific H/D exchange times used in these experiments varied from 0.2 to 23 h (see Table 1 below). After the H/D exchange reactions were quenched, either C4 ZipTip pipet tips (Millipore) or C4 SuproTip pipet tips (Nest Group, Inc.) were employed to desalt, to concentrate, and to ultimately deposit the samples onto the MALDI sample target as previously described.2 Each sample was subjected to MALDI analysis in order to determine the number of deuterons that had exchanged into the protein (i.e., a ∆Mass value) in each GdmCl-containing H/D exchange buffer. A single MALDI sample was prepared from each buffer, and 5-10 replicate mass spectra were acquired on each MALDI sample. The protein molecular weight determinations from these 5-10 replicate MALDI mass spectra were then used to calculate 5-10 ∆Mass values for the protein in each GdmClcontaining H/D exchange buffer. Ultimately, an average ∆Mass value was determined for the protein in each GdmCl-containing H/D exchange buffer, and this value was used to generate a SUPREX curve (i.e., a plot of ∆Mass versus [GdmCl]) at a specific exchange time. Multiple SUPREX curves in which the H/D exchange time was varied were generated for the purified and unpurified CypA samples both in the absence and in the presence of CsA. The data points in each SUPREX curve were fit to a four-parameter sigmoidal equation (eq 1) using a nonlinear regression procedure
∆Mass ) ∆M0 +
a 1+e
-(([GdmCl]-CSUPREX )/b) 1/2
(1)
in SigmaPlot (SPSS Inc.) in order to determine a CSUPREX value 1/2 (i.e., the denaturant concentration at the transition midpoint) for each curve.
Table 1. SUPREX Data and Thermodynamic Parameters for CypA Samples a CSUPREX 1/2 [GdmCl](M)
∆Gfb (kcal/mol)
0.20 0.58 1.00 2.17 8.00
1.28 1.19 1.15 1.08 0.82
11.3 ( 0.7
0.20 0.58 3.00 14.00 26.27 0.20 0.58 2.17 8.00
2.39 2.34 2.11 1.84 1.75 1.64 1.33 1.03 0.80
0.58 2.17 8.00 26.27
2.27 2.16 1.88 1.67
exchange time (h)
protein sample purifed CypA
purified CypA +CsA
unpurified CypA
unpurified CypA +CsA
15.5 ( 0.7
9.4 ( 0.2
14.1 ( 0.8
∆∆Gfc (kcal/mol)
SUPREX Kdd (nM)
4.2 ( 1.0
77 ( 17
-
4.7 ( 0.8
-
32 ( 20
a The CSUPREX values were taken from the best fit of each SUPREX curve at different exchange times. Errors were typically (0.1 M. b Values 1/2 defined by the y intercepts of the -RT ln(〈kint〉t/0.693 - 1)versus CSUPREX plots generated in Figure 3. Errors are the standard errors of fitting 1/2 generated by SigmaPlot. c ∆∆Gf values were calculated from the ∆Gf values of CypA in the presence and in the absence of CsA. d Kd values were calculated according to eq 3. Values represent the average and standard deviation from at least four replicate evaluations (see Experimental Sections for details).
In eq 1, ∆M0 is the change in mass measured before the globally protected protons in the protein exchanged with solvent deuterons, ∆Mass is the change in mass compared with fully protonated protein (i.e., number of protons in the protein that exchanged with solvent deuterons under certain conditions), [GdmCl] is the molar concentration of denaturant guanidinium chloride, CSUPREX is the denaturant concentration at the transi1/2 tion midpoint of a SUPREX curve, a is the transition amplitude of a SUPREX curve, and b is a parameter that describes the steepness of the transition region of a SUPREX curve. In fitting our data to eq 1, all parameters were allowed to float in order to find the best fit. The CSUPREX values determined using eq 1 were ultimately 1/2 used in eq 2 to calculate m and ∆Gf values.3,8 In eq 2, R is the gas
[ ( )] (
)
〈kint〉t -1 0.693 -RT ln n n [P]n-1 2n-1
) mCSUPREX + ∆Gf 1/2
(2)
constant, T is the temperature (in kelvin), 〈kint〉 is the average intrinsic exchange rate of an amide proton, t is the H/D exchange time, n is the number of subunits in the protein (for CypA in this study, n ) 1), [P] is the protein concentration in terms of n-mer, m is defined as δ∆G/δ[Denaturant], and ∆Gf is the folding free energy in the absence of denaturant. In our data analysis, the term on the left side of the eq 2 was plotted as a function of the CSUPREX values extracted from our SUPREX curves recorded at 1/2 different H/D exchange times (t in eq 2). Ultimately, a linear leastsquares analysis of the data in these plots yielded the equation of a line in which the slope and y-intercept corresponded to m and ∆Gf, respectively. We note that a 〈kint〉 value of 7.08 s-1 was used
in all our evaluations of the term on the left side of eq 2. This 〈kint〉 value was estimated for the pH conditions of our buffer (pH 7.0) using the SPHERE program and the primary amino acid sequence of CypA.15,16 Kd Values Determinations. Dissociation constants, Kd values, for the protein-ligand complexes in this work were determined according to the eq 3.17 In eq 3 [L] is the free ligand concentration,
Kd ) [L]/(e-∆∆Gf/nRT - 1)
(3)
∆∆Gf is the difference in folding free energy of the protein measured in the presence of ligand and in the absence of ligand, and n is the number of independent binding sites (n ) 1 for the CypA-CsA complex). Since the total ligand concentrations during the H/D exchange were significantly greater than the protein concentrations used in our studies, the total ligand concentration [L]0 was used to approximate the free ligand concentration [L]. Replicate Kd values were calculated from the multiple SUPREX curves we generated for the CypA and CypA-CsA samples. In these calculations, eq 2 was used to determine ∆Gf values from the individual SUPREX curves we recorded at the different exchange times for purified and unpurified samples of CypA and CypA-CsA. In these ∆Gf calculations using eq 2, the 〈kint〉 and m values used were, 7.08 s-1 and 3.7 kcal/(mol M), respectively. The 〈kint〉 value was calculated from SPHERE as described above, and the m value was the average m value extracted from the four RT ln(〈kint〉)t/0.693 - 1) versus CSUPREX value plots in Figure 3. 1/2 The resulting ∆Gf values were used to calculate multiple ∆∆Gf (15) Zhang, Y.-Z. Ph.D. Thesis, Structural Biology and Molecular Biophysics, University of Pennsylvania, 1995. (16) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, S. W. Proteins 1994, 20, 4-14. (17) Schellman, J. A. Biopolymers 1975, 14, 999-1018.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4345
Figure 1. MALDI analysis of (A) a purified commercial preparation of CypA and (B) a lung tumor cell lysate in which CypA was a natural component. The insets show the two CypA isoforms detected (indicated by arrows) in our experiments. The asterisk denotes the doubly charged CypA ion signal.
values (four values in the case of the purified sample and five values in the case of the unpurified sample), and the resulting ∆∆Gf values were ultimately used to calculate multiple Kd values (four values in the case of the purified sample and five values in the case of the unpurified sample). RESULTS AND DISCUSSION SUPREX Analysis of Cyclophilin A Samples. Representative MALDI-TOF mass spectra of the purified and unpurified CypA samples used in our SUPREX analyses are shown in Figure 1. Two isoforms of CypA were detected in each sample. The existence of such isoforms of CypA has been previously noted.12 The average molecular weights of the two isoforms detected in our purified sample were 17 997 ( 6 and 18 049 ( 3, and the average molecular weights of the two isoforms detected in our unpurified sample were 17 884 ( 6 and 17 920 ( 7. Differences in the masses determined for the purified and unpurified forms of the two isoforms are likely due to minor variations in the primary amino acid sequences of the CypA in the two different samples. For example, it is likely that the isoforms of the purified CypA sample contain an extra N-terminal Met residue as it was generated by recombinant DNA methods. The ability to generate and detect an appropriately resolved MALDI ion signal is an important prerequisite for the SUPREX technique. Such MALDI ion signals are generally easy to obtain on protein analytes in highly purified samples. However, the generation and detection of MALDI ion signals from protein analytes in multicomponent protein mixtures can be complicated by the fact that the different protein components of a complex mixture can experience preferential desorption and/or ionization during the MALDI process.18 We have previously noted that CypA is overexpressed in lung tumors and that the protein is readily (18) (a) Amado, F. M. L.; Dominigues, P.; Santa-Marques, M. G.; Ferrer-Correia, A. J.; Tomer, K. B. Rapid Commun. Mass Spectrom. 1997, 11, 1347-1352. (b) Kratzer, R.; Eckerskorn, C.; Karas, M.; Lottspeich, F. Electrophoresis 1998, 19, 1910-1919. (c) Beavis, R. C.; Chait, B. T. Methods Enzymol. 1996, 270, 519-551.
4346 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 2. Representative SUPREX curves obtained for (A) the purified CypA sample and (B) the unpurified CypA from the lung tumor cell lysate. The data points and error bars represent the average and 2 standard deviations of replicate ∆Mass measurements at each [GdmCl]. The open circles represent the data obtained in the absence of CsA, and the closed squares represent the data obtained in the presence of CsA. The lines represent the best fit of each data set to eq 1. The dashed lines and arrows indicate the transition midpoint (CSUPREX ) of each SUPREX curve. The same 35-min H/D exchange 1/2 time was used to generate each SUPREX curve.
detected in the direct MALDI analysis of human lung tumor tissue lysates.11 This clearly facilitated the SUPREX analyses in this work. Theoretically, both isoforms in the CypA samples could be analyzed simultaneously in the SUPREX experiment. Practically, however, it was difficult to accurately determine the mass of the larger isoform of CypA after H/D exchange due to the limited resolution of our mass spectrometer. Therefore, only the lower MW isoform of CypA was used to generate the SUPREX data in this study. Our inability to baseline resolve the isotopic envelopes of the two CypA isoforms in this study introduces a potential source of error in the ∆Mass measurements used in our SUPREX analyses. This source of error is likely responsible for some of the scatter in the data points that define the pre- and posttransitions of our SUPREX curves. However, we note that the scatter observed was small (i.e.,
