Biochemistry 1994, 33, 7797-7810
7797
Resolution of Structural Changes Associated with Calcium Activation of Calmodulin Using Frequency Domain Fluorescence Spectroscopy? Yihong Yao,t Christian Sch6neich,%and Thomas C. Squier’*t Departments of Biochemistry and Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas 66045-21 06 Received January 6, 1994; Revised Manuscript Received April 1 1 , 1994”
ABSTRACT: Structural changes associated with the calcium-dependent activation of wheat germ calmodulin
(CaM) were assessed through measurements of steady-state and time-resolved changes in the fluorescence associated with (1) the unique tyrosine (Tyr139) located in calcium binding loop IV or (2) N-(1-pyreny1)maleimide (PM) or 4-(iodoacetamido)salicylic acid (IASA) covalently attached to Cys2, present in calcium binding loop I. These fluorophores permit the measurement of calcium-dependent changes in (i) the solvent accessibilityand rotational dynamics associated with calcium binding loops I and IV and (ii) the hydrodynamic properties of the entire protein. Specific nitration of the unique tyrosine (Tyr139) in calcium binding loop IV permits the use of fluorescence resonance energy transfer to measure both the average spatial separation and distance heterogeneity between Cys27 and Tyr139,providing a direct measurement of the conformational flexibility of the central helix. Upon calcium binding, (i) the solvent accessibility and rotational dynamics of both PM and IASA (covalently bound to Cys27) and Tyr139increase, (ii) overall protein rotational motion decreases, (iii) the average separation between the chromophores a t Cys27 and nitrotyrosine 139 decreases, and (iv) the conformational flexibility associated with the central helix decreases. Therefore, upon calcium binding, the central helix becomes more extended and rigid, while the globular domains adopt a more open tertiary conformation that brings Cys27 and Tyr139 into closer proximity. This calcium-dependent structural change functions to expose the hydrophobic binding sites located within the globular domains, and to enhance the probability of binding target sequences through a reduction in conformational heterogeneity.
Calmodulin (CaM) is a ubiquitous eukaryoticCa2+-binding protein that regulates numerous cellular processes, including muscle contraction, neurotransmission, neuronal plasticity, cytoskeletal assembly, and a host of reactions involved in the energy and biosynthetic metabolism of the cell [reviewed by Wylie and Vanaman (1988)l. The crystal structure of the calcium-liganded form of CaM has recently been refined to 1.7 A (Chattopadhyayaet al., 1992) and shows twostructurally homologous, globular domains connected by an eight-turn central cy-helix (Babu et al., 1985, 1988). Both globular domains consist of two Ca2+-bindingsites, each of which shows a helix-loophelix motif known as an EF-hand (Kretsinger & Nockolds, 1973). Binding of Ca2+ is cooperative with respect to each globular domain, and this binding functions to induce a conformational change that is thought to expose
an otherwise inaccessible nonpolar surface involved in target enzyme binding, which leads to the activation of various target enzymes (Tanaka & Hidaka, 1980). Calcium binding sites I1 and I11 are associated with the central helix, suggesting that structural changes involving the central helix may play a critical role in the cooperative binding of calcium (Babu et al., 1985). However, while there is evidence that the two opposing domains of CaM do interact [Johnson, 1983;Yoshida et al., 1983; Thulin, et al., 1984; reviewed by Klee (1988)], a cooperative enhancement in calcium binding between these two domains has only been observed in the presence of target peptides (Yazawa et al., 1992),which involves the association of both opposing globular domains in CaM (Meador et al., 1992; Ikura et al., 1992b; Meador et al., 1993). Therefore, it remains unclear whether structural changes associated with the central helix play any role in the calcium-dependent activation of CaM. Supported in part by grants from the National Institutes of Health X-ray structural data is available only for the calcium(GM46837), the Muscular Dystrophy Association, and the University liganded form of CaM at acidic pH (Le., pH 5.0 and 5.6) in of Kansas. * Correspondence should be addressed to Dr. Thomas C. Squier (913the presence of helicogenic solvents (i.e., nonaqueous solvents 864-408 1). that appear to induce additional a-helix; Bayley et al., 1988). * Department of Biochemistry. Therefore, it remains unclear as to the nature of both (i) the Department of Pharmaceutical Chemistry. structural changes associated with calcium activation and (ii) Abstract published in Advance ACS Abstracts, June 1, 1994. * Abbreviations: IAEDANS, 5 4 [2-[(iodoacetyl)amino]ethyl]amino]the average structure of CaM in solution under physiological naphthalene-1-sulfonic acid; CaM, calmodulin; DTNB, 5,Y-dithiobisconditions. In fact, significant differenceshave been observed (2-nitrobenzoicacid); DTT, dithiothreitol; EDTA, ethylenediaminetetin both the secondary and tertiary structures of CaM as a raacetic acid; FRET, fluorescence resonance energy transfer; FURA-2, 1-[ [2-(5-carboxyoxazol-2-yl)-6-aminobenzofuan-5-yl]oxy]-2-(2’-amino- function of both pH and salt concentration (Tbr6ket al., 1992). Hydrodynamic measurementsusing both optical and magnetic 5’-methylphenoxy)ethane-N,N,N’,N’-tetraacetic acid; HEPES, N42hydroxyethyl)piperazine-N‘-2-ethanesulfonicacid; GnHC1, guanidinium resonance techniques suggest that CaM is more compact at hydrochloride;HOMOPIPES,homopiperazine-N,”-bis(2-ethanesulfonic physiological pH relative to acidic conditions analogous to acid); HPLC, high-performanceliquid chromatography;IASA, 4-(iodothose used to obtain thecrystal structure (Wang, 1989; Small acetamido)salicylic acid; PM, N-( 1 -pyrenyl)maleimide; SDS, sodium & Anderson, 1988; Heidorn & Trewhella, 1988;T6rbk et al., 1 -piperididodecyl sulfate; TEMPAMINE, 4-amino-2,2,6,6-tetramethylnyloxyl; TFA, trifluoroacetic acid; TNM, tetranitromethane. 1992). Likewise, it has been suggested that CaM becomes @
0006-2960/94/0433-7797$04.50/0 0 1994 American Chemical Society
7798 Biochemistry, Vol. 33, No. 25, 1994 extended upon calcium binding at physiological pH [ LaPorte et al., 1981; Seaton et al., 1985; Small & Anderson, 1988; Heidorn & Trewhella, 1988; reviewed by Trewhella (1992)], although possible structural linkages between the two globular domains (in the absence of target peptides) remain controversial [Barbato et al., 1992; Torok et al., 1992; reviewed by Williams (1992); reviewed by Kretsinger (1992a)l. Subsequent to calcium activation, the critical element associated with binding target enzymes appears to involve the flexibility of the central helix, as the association of the calcium-activated form of CaM with target peptides results in only minimal changes in the conformation of the opposing globular domains (Meador et al., 1992;Ikura et al., 1992b; Meador et al., 1993). Therefore, central to the question of how calcium activation of CaM facilitates its specific interaction with target sequences is information regarding the structure of the globular domains as well as that of the central helix. In this regard, fluorescence spectroscopy is a sensitive tool for studying the structure and dynamics of proteins [reviewed by Lakowicz and Gryczynski (1991)]. Using fluorescence anisotropy and quenching experiments we are able to specifically probe both the global conformation of CaM (Le., hydrodynamic properties) and the localized structure associated with the globular domains. In addition, through the use of fluorescence resonance energy transfer (FRET) we are able to measure the conformational heterogeneity associated with the central linker helix that connects the two opposing globular domains [reviewed by Cheung (1991)]. This study takes advantage of the single cysteine and tyrosine present in wheat germ CaM, which can be modified to provide fluorescence signals used to assess the calcium-dependent structural changes of CaM. Wheat germ CaM has 94% sequence identity with bovine brain CaM (Klee et al., 1980; Toda et al., 1985) and previously has been shown to fully activate a range of target proteins from human and animal sources (Strasburg et al., 1988). Our measurements involve the specific labeling of the unique cysteine group (Cys~7) available in wheat germ CaM with two different sulfhydryl group directed chromophores, which serve as energy-transfer donors. Likewise, we are able to use the intrinsic fluorescence associated with the single tyrosine (Tyr139) in the carboxyl domain to assess structural changes associated with calcium binding. Alternatively, the selective nitration of this tyrosine with tetranitromethane (TNM) permits its use as an energytransfer acceptor, permitting direct measurements of the spatial separation between the globular domains of CaM. This latter measurement provides direct information relating to both the average structure and conformational heterogeneity of the central helix. Together, these complementary measurements of both the local and global conformation and dynamics associated with calcium activation permit us to distinguish between alternative models relating to the structural changes in both the globular domains and central helix associated with calcium binding to CaM. EXPERIMENTAL PROCEDURES Materials. IASA [4-(iodoacetamido)salicylic acid] and PM [N-(1-pyrenyl)maleimide] were obtained from Molecular Probes, Inc. (Junction City, OR). TEMPAMINE (4-amino2,2,6,6-tetramethyl- 1-piperidinyloxyl) and TNM (tetranitromethane) were obtained from Aldrich (Milwaukee, WI). Immediately prior to its use, TNM was purified by extracting four times with a 20-fold excess volume of water (Riordan & Vallee, 1972). Purified FURA-2 was generously provided by Professor J. David Johnson (OhioStateUniversity). All other chemicals were of the purest grade commercially available.
Yao et al.
4-Iodoacetamidosalicylic acid
Pyrene-Maleimide
FIGURE 1: Chemical structures: (a) 4-(iodoacetamido)salicylic acid (IASA) and (b) N - ( 1-pyreny1)maleimide (PM).
CaM was purified from wheat germ using the procedure outlined by Strasburg et al. (1988). Purity was assessed using both SDS-PAGE, where CaM migrates as a single band, and HPLC. Purified wheat germ CaM was stored at -70 OC. Human erythrocyte ghosts were prepared essentially as described by Niggli et al. (1979). Specific Derivatization of Cys27. Prior to chemical derivatization, CaM is first dissolved in 6 M GnHCl, 25 mM HEPES (pH 7 . 9 , 50 mM DTT, and 1 mM EDTA and incubated at 25 OC for 2 h in order to eliminate intermolecular cross-linking. DTT was removed by dialysis of the protein against deionized water prior to lyophilization. Sulfhydryl content was measured by the method described by Ellman (1959). Specificity of labeling was assured by (i) titration of remainingsulfhydryl groups using DTNB (€412= 13 600 M-I cm-I), (ii) quantitation of bound fluorophore using the molar extinction coefficient (see below), and (iii) identification of labeled peptide(s) subsequent to proteolytic digestion using HPLC (see below). Chemical derivatization of Cys2-i with pyrenylmaleimide (PM) (Figure 1) was carried out in the dark. The concentration of PM was kept below 20 pM to avoid nonspecific probe aggregation. Briefly, 12 pM PM was added to a solution containing 6 pM CaM (Le., 0.1 mg/mL) in 0.1 mM CaC12, 25 mM HEPES (pH 7.5) at 25 OC. Thereaction was quenched after 1 h by the addition of 0.1 mM DTT to the reaction mixture, which was incubated on ice for 10 min. The labeled CaM was separated from unreacted PM using a Sephadex G-25 column (1.6 X 23 cm) and was subsequently lyophilized. The specific labeling of Cys27 with 4-(iodoacetamido)salicylic acid (IASA) was essentially identical to that of PM, except that a 10-fold molar excess of IASA was added to CaM in the presence of 6 M GnHC1, essentially as described by Strasburg et al. (1988) for the labeling of CaM with IAEDANS. Incorporation of IASA and PM was determined using their molar extinctioncoefficients, € 3 0 2 = 9500 M-I cm-I and €343 = 36 000 M-' cm-I for IASA and PM, respectively (Haugland, 1992). Protein concentration was routinely determined using the micro BCA assay (Pierce). A stock solution of CaM, whose concentration was determined using the published extinction coefficient for CaM (Strasburg et al., 1988), was used as a protein standard. Selective Nitrationof Tyrljg. Nitration with freshly purified TNM was carried out at 25 OC in the presence of 60 pM CaM (Le., 1.0 mg/mL) in 1 M NaC1, 0.5 mM CaC12, and 0.1 M TRIS (pH 7.9), essentially as previously described (Richman & Klee, 1978). The nitration of Tyrl39 was initiated by the addition of saturating concentrations of TNM in ethanol (Le., 15%,v/v) to give a final concentration of 100-fold molar excess TNM relative to CaM. Subsequent to the addition of TNM,
Calcium Activation of Calmodulin the aqueous phase was immediately separated from the nonsoluble oil. After 1 h the reaction was quenched by adding excess 8-mercaptoethanol and left on ice for 10 min prior to the separation of the derivatizedCaM using a Sepharose G-25 column (1.6 X 23 cm). Derivatized CaM was stable at -70 OC for 2 months, as evidenced by both the ability to activate the erythrocyte Ca-ATPase and fluorescence measurements relating to protein structural changes (see Results). The extent of nitration was measured in 6 M GnHCl, 1 mM EDTA, and 25 mM HEPES (pH 10) using the molar extinction coefficient t428 = 4200 M-' cm-' for nitrotyrosine (Richman & Klee, 1978). Enzymatic Assays. The Ca-ATPase activity of erythrocyte ghosts was determined using the methods described by Lanzetta et al. (1979), essentially as described previously for sarcoplasmic reticulumCa-ATPase (Squier & Thomas, 1988). When applicable, the free calcium concentration was calculated using a modified version of the computer program previously described (Fabiato & Fabiato, 1979;Fabiato, 1988), which calculates the multiple equilibria between all ligands in solution. The free calcium concentrationwas checked using the calcium indicator dye FURA-2 (LX = 340 nm; A,, = 510 nm). In all cases we find an apparent dissociation constant of 4 0 f 5 nM, in close agreement with literature values [KD = 0.1 mM; Haugland (1992)]. Identification of Tryptic Fragments. In order to access sites associated with chemical derivatization, calmodulin (60 pM) was subjected to exhaustive tryptic digestion (0.6 pM trypsin in 50 mM potassium phosphate at pH 8.0 for 9 h at 37 "C). Digestion was stopped upon the addition of 1.8 pM trypsin inhibitor. Tryptic fragments were separated on HPLC using a Vydac C4 reverse phase column employing a linear gradient varying from 0.1% trifluoroacetic acid (TFA) to 0.1% TFA in 80% acetonitrile/20% water at a rate of l%/min. The respective peaks were monitored at 214 nm. The detected peaks were pooled, lyophilized, and subjected to FAB mass spectrometry. Details relating to the assignments of these peaks will be published elsewhere. Spectroscopic Measurements. Fluorescence emission spectra were recorded using an ISS K2 fluorometer in the ratio mode. Excitation was set at 302 nm for IASA-labeled CaM and 35 1 nm for PM-labeled CaM. Emission slits were adjusted to a 4-nm bandwidth, and the sample temperature was maintained at 25 "C. Fluorescence lifetime and anisotropy measurements were performed using an ISS K2 frequency domain fluorometer, whose design has previously been described in detail (Gratton & Limkeman, 1983). This instrument is equipped with a Marconi signal generator (2022A & C) and EN1 broad-band amplifiers (325LA and 403LA) which operate in conjunction with a Pockels cell to obtain intensity-modulated light from either a 300-W xenon arc lamp (ILC Technology PS300-1) or alternatively a CW argon ion laser equipped with extended UV capabilities (Coherent Innova 400/2/0.3, Santa Clara, CA). Fluorescence quenchingexperimentswere performed using an ISS K2 fluorometer in the ratio mode. TEMPAMINE was added in microliter increments to 2 mL of either 1.1 pM PM- or IASA-labeled CaM dissolved in 0.1 M KCl, 25 mM HEPES (pH7.5),andeitherO.l mMCaClzor0.1 mMEDTA. Excitation was at 351 and 302 nm, respectively. Fluorescence emission was collected using a Schott GG400 long-pass filter. Alternatively, analogous experiments involving tyrosine quenching used 5.9 pM native CaM, with 275-nm excitation. In the latter case fluorescence emission was collected through an Oriel band-pass filter centered at 320 nm (full width halfmaximum is 10 nm), and necessary corrections were made for
Biochemistry, Vol. 33, No. 25, 1994 7799
the contribution to the observed fluorescence intensity originating from Raman scattering. For comparison purposes, KI (1 M stock) was also used as a collisional quencher. Analysis of the collisional quenching was carried out essentially as described by Lehrer and Leavis (1978) and is described in the footnotes below Table 2. Decays of Fluorescence Intensity. The time-dependent decay I ( t ) of any fluorophore can always be described as a sum of exponentials,
where ai are the preexponential factors, Ti are the excitedstate decay times, and n is the number of exponential components required to describe the decay. The intensity decay law is obtained from the frequency response of the amplitude-modulated light and is characterized by the frequency-dependentvalues of the phase shift and the extent of demodulation, whose respective errors are assumed to be 0.2O and 0.005. The parameters describing the decay law are compared with the calculated values from an assumed decay law, and the parameters chosen are those that minimize the squared deviation. Explicit expressions have been provided that permit the ready calculation of ai and Ti (Weber, 1981). The parameter values are determined by minimizing the x~~ (the F statistic), which serves as a goodness-of-fit parameter that provides a quantitative comparison of the adequacy of different assumed models (Lakowicz & Gryczynski, 1991). Data are fitted using the method of nonlinear least squares to a sumof exponential decays (Bevington, 1969). Subsequent to the measurement of the intensity decay, one typically calculates the average lifetime, i,which is weighted by the amplitudes associated with each of the preexponential terms, where
is directly related to the average time the fluorophore is in the excited state, and the amplitude weighting implies a direct relationship between ;and the quantum yield of the fluorophore (Luedtke et al., 1981). Alternatively, for the case of resonance energy transfer measurements more realistic physical models were used involving a distribution of distances (see below), as previously described (Lakowicz et al., 1988). Calculationof Molecular Distances UsingFRET. Utilizing fluorescence resonance energy transfer (FRET) to measure thedistance between any fluorophore (donor; D) and a suitable acceptor (A) chromophore, one can measure distances in the 10-1 00-8,range and directly recover structural information concerning biological macromolecules (Stryer, 1978). The efficiency of energy transfer, E, and the apparent donoracceptor distance, rapp, are calculated from the Fbster equations (Fairclough & Cantor, 1978), where
Fda and Fd are the steady-state fluorescence intensities in the presence and absence of the acceptor (nitrotyrosine), ;da and i d are the average fluorescence lifetimes (see eq 2) of the donor in the presence and absence of the acceptor (nitrotyrosine), and Ro is the Fbrster critical distance that defines the distance for a given donor-acceptor pair where the efficiency ofresonanceenergy transfer is 50%. ;da and ;d are determined from frequency domain measurements relating to the fluo-
7800 Biochemistry, Vol. 33, No. 25, 1994
Yao et al.
~~
~
~~
Table 1 sample
DH
D + Ca2+
7.5
DA
+ Ca2+
7.5
D - Ca2+
7.5
DA - Ca2+
7.5
D + Ca2+
5.0
DA
+ Ca2+
D + Ca2+in 6 M GnHCl DA + Ca2+in 6 M GnHCl
5.0 7.5 7.5
sample
DH
D + Ca2+
7.5
A. Lifetime Data for PM-CaM in the Presence and Absence of Nitrotyrosine' ffl TI (ns) ff2 7 2 (ns) ff3 7 3 (ns) 0.788 (0.031) 0.808 (0.025) 0.61 1 (0.021) 0.552 (0.17) 0.632 (0.02) 0.727 (0.03) 0.474 (0.008) 0.544 (0.010)
3.72 (0.04) 3.45 (0.03) 3.45 (0.02) 2.66 (0.17) 3.43 (0.08) 3.34 (0.12) 1.36 (0.06) 1.27 (0.07)
0.136 (0.008) 0.141 (0.01 1) 0.251 (0.009) 0.352 (0.01 1) 0.279 (0.012) 0.200 (0.01 3) 0.507 (0.009) 0.438 (0.009)
18.8 (0.6) 18.1 (0.4) 14.8 (0.3) 11.9 (0.3) 20.5 (0.5) 18.8 (0.8) 4.7 1 (0.18) 4.69 (0.17)
0.076 (0.004) 0.03 1 (0,001) 0.138 (0.003) 0.096 (0.002) 0.089 (0.004) 0.073 (0.06) 0.019 (0.001) 0.018 (0.001)
Ta,r, (ns)
115.4 (3.2) 107.2 (2.3) 96.9 (2.2) 92.8 (2.9) 125.4 (4.3) 125.0 (7.3) 42.7 (0.7) 36.8 (0.6)
13.5 (0.2) 8.66 (0.1) 19.2 (0.1) 14.6 (0.2) 19.2 (0.3) 15.3 (0.3) 3.84 (0.06) 3.40 (0.04)
B. Lifetime Data for IASA-CaM in the Presence and Absence of NitrotyrosineC CY1 T I (ns) ff? r2 (ns) 1 ~ ~(ns) 7,
XR2
1.8 (10.8) 2.2 (34.7) 1.1 (1 6.9) 1.8 (27.2) 1.4 (12.2) 1.3 (25.4) 0.5 (8.8) 0.6 (10.2) x12 d
~~
DA
+ Ca2+
7.5
D - Ca2+
7.5
DA - Ca2+
7.5
+ Ca2+ DA + Ca2+ D
5.0 5.0
0.930 (0.00 1) 0.95 1 (0.003) 0.915 (0.001) 0.920 (0.003) 0.897 (0.003) 0.944 (0.003)
0.48 1 (0.006) 0.332 (0.009) 0.430 (0.006) 0.418 (0.006) 0.431 (0.009) 0.408 (0.012)
0.070 (0.00 1) 0.049 (0.002) 0.086 (0.00 1) 0.080 (0.003) 0.103 (0.002) 0.056 (0.002)
4.51 (0.06) 3.23 (0.10) 3.88 (0.06) 2.97 (0.05) 4.87 (0.09) 6.04 (0.15)
0.704 (0.008) 0.474 (0.007) 0.727 (0.009) 0.622 (0.010) 0.888 (0.009) 0.727 (0.01 1)
c. Lifetime Data for Tyr139-CaMe sample +Ca2+
PH 7.5
+EGTA
7.5
+Ca2+
5.0
ff1
0.90 (0.01) 0.86 (0.02) 0.85 (0.01)
71
(ns)
ff2
0.30 (0.05) 0.60 (0.04) 0.30 (0.06)
0.10 (0.01) 0.14 (0.02) 0.15 (0.01)
72
(ns)
2.4 (0.2) 2.6 (0.2) 2.4 (0.2)
Lair, (ns) 0.51 (0.06) 0.88 (0.06) 0.62 (0.08)
XR2
'
1.6 (95) 0.74 (43) 2.0 (1 17)
0 Average amplitudes (ai) and lifetimes ( T J , obtained from three-exponential fits to frequency domain data collected for donor only (D or PM-CaM) and donor-acceptor (DA or PM-nitrotyrosine-CaM) calmodulin, represent the average of six different measurements, where uncertainties in the measured lifetime parameters (shown in parentheses) represent the standard error of the mean. The X R for ~ a two-exponential fit to the data is shown in parentheses for comparison purposes. Experimental conditions: 25 mM HEPES (or 25 mM HOMOPIPES at pH 5.0), 0.1 M KCI, and either 0.1 mM CaC12 (+Ca2+) or 0.1 mM EDTA (-Ca2+) at 25 OC. Average amplitudes (ai) and lifetimes ( r i ) ,obtained from two-exponential fits to frequency domain data collected for donor only (D or IASA-CaM) and donor-acceptor (DA or IASA-nitrotyrosine-CaM) calmodulin, represent the average of five different measurements, where uncertainties in the measured lifetime parameters (shown in parentheses) represent the standard error of the mean. The X R for ~ a one-exponential fit to the data is shown in parentheses for comparison purposes. Experimental conditions: 25 mM HEPES (or 25 mM HOMOPIPES at pH 5.0), 0.1 M KCI, and either 0.1 mM CaC12 (+Ca2+) or 0.1 mM EDTA (-Ca2+) at 25 OC. e Average amplitudes (ai) and lifetimes (q)obtained from two-exponential fits to frequency domain data collected for native CaM (Lx= 275 nm). Uncertainties in measured lifetime parameters (shown in parentheses) are determined from a global error analysis of the parameters, as described in Experimental Procedures and represent 1 standard deviation about the mean. /The x~~for a one-exponential fit to the data is shown in parentheses for comparison purposes. Experimental conditions: 25 mM HEPES (or 25 mM HOMOPIPES at pH 5.0), 0.1 M KCI, and either 0.1 mM CaC12 (+Ca2+) or 0.1 mM EDTA (-Ca2+) at 25 OC.
rescence intensity decay of either PM-CaM or IASA-CaM in the absence and presence of nitrotyrosine (see above). Ro is given by:
R, = 9.79 x 10-5(n4K2+d~1/6 in cm
(4)
where n is the refractive index, K~ is the orientation factor, J is the spectral overlap integral, and +d is the quantum yield of the donor in the absence of acceptor. In our experiments, n is estimated to be 1.40 (Fairclough & Cantor, 1978); K~ is assumed to be 2/3, which assumes that donor and acceptor chromophores undergo rapid isotropic rotational motion relative to the lifetime of the donor (see below); @d is determined by numerical integration of the fluorescence emission spectrum of either PM-CaM or IASA-CaM, using quinine sulfate as a standard, which has a quantum yield of 0.70 in 0.1 N H2.304 (Scott et al., 1970); and Jis calculated by numerical integration
from the spectra of IASA- and PM-CaM and the absorption spectrum of nitrotyrosine CaM, as described previously for other donor-acceptor pairs (Squier et al., 1987). Color effects relating to the wavelength-dependent properties of the photomultiplier were corrected using an algorithm provided by ISS Inc. (Urbana-Champaign). In the presence of saturating calcium (pH 7.5),thequantum yields (+) for PM-CaM, IASACaM, and Tyr139 in native CaM are 0.175, 0.193, and 0.022, respectively. The latter measurement is consistent with previous estimates obtained using octopus CaM, which also contains a single tyrosine in calcium binding site I V (Kilhoffer et al., 1981). The overlap integral, J , for the two donoracceptor pairs PM-nitrotyrosine-CaM and IASA-nitrotyrosine-CaM are 2.80 X M-I cm3 and 2.76 X M-' cm3, respectively. These constants were remeasured under the different experimental conditions used in this study and are reflected in the measured values of Ro (see Table 3).
Biochemistry, Vol. 33, No. 25, 1994 7801
Calcium Activation of Calmodulin Table 2: Stern-Volmer Quenching Constants"
freeprobeb +Ca2+ -Ca2+ +Ca2+ +Ca2+ in 6 M GnHCl freeprobeb +Ca2+ -Ca2+ +Ca2+
Ksv kq pH (M-l)c i (ns)d (M-I s - ' ) ~ kq/kq(freeprobe)c A. PM-CaM 7.5 385 3.9 99X lo9 1 .o 0.14 7.5 182 13.5 14 X lo9 0.036 3.6 X lo9 7.5 69 19.2 6.0 X lo9 0.061 5.0 116 19.2 0.4Y 3.8 42 X lo9 7.5 163
7.5 7.5 7.5 5.0
605 468 248 350
B. IASA-CaM 0.30 20 X loL1 0.70 6.7 X 10" 0.73 3.4 X 10" 0.89 3.9 X 10"
1 .o
0.33 0.17 0.20
C. T Y ~ 1 .o 1.05 3.3 X 1012 freeprobeb 7.5 3520 0.78 0.51 2.6 X 10l2 +Ca2+ 7.5 1330 0.87 0.6 X 10l2 0.18 -Ca2+ 7.5 520 0.10 0.62 0.3 X 10l2 +Ca2+ 5.0 206 Protein conformational changes are revealed by changes in solvent accessibility of water soluble quenchers to fluorophores associated with CaM. Free probe is PM-Cys (A), IASA-Cys (B), or L-tyrosine (C). Kw is obtained from the Stern-Volmer relationship Fo/F = 1 + K,[TEMPAMINE], where FOis the initial fluorescenceintensity in the absence of added quenchers. kq KgV/i,where i Zarisi (see Table 1). CThe quenching efficiency of the chromophore bound to CaM was normalized to that of the free probe. kq/kq(freeprobe) corrected for the viscosity of 6 M GnHCl ( q = 1.624 cP) relative to pure water ( q = 0.89 cP) is 0.70. The concentration of fluorescently labeled calmodulin is 1.13pM. Theconcentrationofnativecalmodulinis5.9pM. Themedium buffer contained 25 mM HEPES (or 25 mM HOMOPIPES at pH 5.0), 0.1 MKC1,andeitherO.l mMCaC12(+Ca2+)or0.1mMEDTA (-Ca2+). The temperature was 25 'C.
The above analysis assumes a unique donor-acceptor separation in the calculation of molecular distances, and it ignores any conformational heterogeneity associated with the molecular dynamics of CaM. However, the intensity decay associated with the donor in the presence of an acceptor permits one to recover the conformational heterogeneity (or distribution of distances) associated with CaM [Haas et al., 1978; Beechem & Haas, 1989; reviewed by Cheung (1991)l. To minimize the number of parameters, a uniform Gaussian distribution of donor to acceptor distances, P(r), is generally assumed:
where R,is theaveragedistanceand ais thestandarddeviation of the distribution. The width of the distribution is reported as the full width at half-maximum (half-width, hw), which is given by hw = 2.354~.The adequacy of the Gaussian model is assessed through a comparison of the goodness of the fit (i.e., X R ~see ; below) relative to the fit using the more general multiexponential model which assumes no physical model. Decays of Fluorescence Anisotropy. Timeresolved anisotropies were measured both from the phase angle difference and from the ratio of the amplitudes of the parallel and perpendicular components of the modulated emission, as previously described (Lakowicz & Gryczynski, 1991; Johnson & Faunt, 1992).
RESULTS Specificity of Calmodulin Labeling. Critical to the evaluation of the fluorescence data associated with calcium activation of CaM are (i) the specificity of chemical derivatization and (ii) an assurance that the chemical derivatization did not perturb the structure or function of CaM. Therefore
we have characterized the site-directed labeling of wheat germ CaM. Using DTNB to titrate the sulfhydryl content of wheat germ CaM we find 0.95 f 0.05 mol of SH/mol of CaM, in close agreement with the reported unique Cys27 (Yoshida et al., 1983). This indicates that essentially no disulfide crosslinks are present. Chemical derivatization of Cys2.lwith either IASA or PM (Figure 1) quantitatively blocks subsequent derivatization with DTNB, suggesting that essentially all cysteines have been derivatized. Quantitation of IASA or PM incorporation using their extinction coefficients (see Experimental Procedures) indicates that 0.88 f 0.06 mol of IASA/mol of CaM and 0.95 f 0.03 mol of PM/mol of CaM are covalentlyincorporated. The incorporation of either IASA or PM onto CaM subsequent to labeling with DTNB [fluorophorecontent determined following the removal of the thio(nitrobenz0ate) blocking group] was negligible (Le., less than 0.06 mol of IASA or 0.04 mol of PM per mole of DTNBtreated CaM), suggesting that there are no significant labeling sites other than the single cysteine (Cys27). The specificityof labeling was confirmed using reverse phase HPLC to purify the peptides derived either subsequent to an exhaustive tryptic digestion of native CaM or subsequent to chemical derivatization with PM (data not shown). The retention time of only one peptide is modified following PM labeling of CaM, indicating that a single peptide is modified. Subsequent identification of the modified peptide using mass spectroscopy indicates that tryptic peptide T2 is specifically labeled. T2 contains AA15-AA38, which bracket the single cysteine in wheat germ CaM, consistent with the specific labeling of Cys27 with PM (Toda et al., 1985). In order to measure the spatial separation between Cys17 and Tyr139 using fluorescence resonance energy transfer (FRET) it is necessary to use a pair of chromophores that have overlapping absorption and fluorescenceemission spectra (Stryer, 1978; Haas et al., 1978). In our experiments this requirement is met by modification of Tyrl3~with TNM and of Cys27 with either IASA or PM (Figure 2). The extent of nitration was quantitative (i.e., 0.95 f 0.05 mol of nitrotyrosine/mol of CaM) as judged by the absorbance spectrum of the modified CaM (see Experimental Procedures). There are no structural or functional perturbations subsequent to nitration as judged by the ability of the modified CaM to activate the erythrocyte Ca-ATPase (see below). Furthermore, any nonspecific nitration of other amino acids will not affect our ability to use nitrotyrosine as an acceptor as the absorbance spectrum of Tyr139 is uniquely sensitive to modification by TNM in wheat germ CaM (since there are no tryptophans). However, in order to quantitatively assess the specificity of nitration we subjected the TNM-modified PM-CaM to exhaustive tryptic digestion and compared the resulting tryptic map with that previously generated from PM-CaM. We observe the selective disappearance of T11 (AA128-AA144; Toda et al., 1985), which can no longer be chromatographically detected (data not shown). After the nitration reaction the tryptic maps show an additional set of peaks eluting around 27.0 f 0.5 min which have spectral characteristics analogous to nitrotyrosine. The intensities and retention times of all other fragments remain unaffected, indicating that the unique tyrosine in wheat germ calmodulin is specifically nitrated. We have used SDS-PAGE to monitor any cross-linking or fragmentation that may result from either the chemical derivatization of CaM or exposure to ultraviolet light (associated with fluorescence measurements; see below). We observe no changes in the mobility of CaM on SDS-PAGE gels as a result of its covalent modification or exposure to high
7802 Biochemistry, Vol. 33, No. 25, 1994
Yao et al.
Table 3: Donor-Acceptor Separation between Chromophores on Opposing Globular Domains of Calmodulinn PH E%b R O (A). 'npp (AId R a v (A)' A. PM Donor 22.4 7.5 35.6 20.6 22.7 +Ca2+ -Ca2+
7.5
+Ca2+
5.O
+Ca2+ in 6 M GnHCl
7.5
+Ca2+
7.5
