When Size Is Important
1998; Elsevier BV; Volume: 273; Issue: 44 Linguagem: Inglês
10.1074/jbc.273.44.28994
ISSN1083-351X
AutoresAnders Malmendal, Johan Evenäs, Eva Thulin, Garry P. Gippert, Torbjörn Drakenberg, Sture Forsén,
Tópico(s)NMR spectroscopy and applications
ResumoThe accommodation of Mg2+ in the N-terminal domain of calmodulin was followed through amide1H and 15N chemical shifts and line widths in heteronuclear single-quantum coherence spectroscopy NMR spectra. Mg2+ binds sequentially to the two Ca2+-binding loops in this domain, with affinities such that nearly half of the loops would be occupied by Mg2+ in resting eukaryotic cells. Mg2+ binding seems to occur without ligation to the residue in the 12th loop position, previously proven largely responsible for the major rearrangements induced by binding of the larger Ca2+. Consequently, smaller Mg2+-induced structural changes are indicated throughout the protein. The two Ca2+-binding loops have different Mg2+ binding characteristics. Ligands in the N-terminal loop I are better positioned for cation binding, resulting in higher affinity and slower binding kinetics compared with the C-terminal loop II (k off = 380 ± 40 s–1compared with ∼10,000 s−1 at 25 °C). The Mg2+-saturated loop II undergoes conformational exchange on the 100-μs time scale. Available data suggest that this exchange occurs between a conformation providing a ligand geometry optimized for Mg2+ binding and a conformation more similar to that of the empty loop. The accommodation of Mg2+ in the N-terminal domain of calmodulin was followed through amide1H and 15N chemical shifts and line widths in heteronuclear single-quantum coherence spectroscopy NMR spectra. Mg2+ binds sequentially to the two Ca2+-binding loops in this domain, with affinities such that nearly half of the loops would be occupied by Mg2+ in resting eukaryotic cells. Mg2+ binding seems to occur without ligation to the residue in the 12th loop position, previously proven largely responsible for the major rearrangements induced by binding of the larger Ca2+. Consequently, smaller Mg2+-induced structural changes are indicated throughout the protein. The two Ca2+-binding loops have different Mg2+ binding characteristics. Ligands in the N-terminal loop I are better positioned for cation binding, resulting in higher affinity and slower binding kinetics compared with the C-terminal loop II (k off = 380 ± 40 s–1compared with ∼10,000 s−1 at 25 °C). The Mg2+-saturated loop II undergoes conformational exchange on the 100-μs time scale. Available data suggest that this exchange occurs between a conformation providing a ligand geometry optimized for Mg2+ binding and a conformation more similar to that of the empty loop. calmodulin residues 6–20 residues 29–38 residues 45–56 residues 65–75 heteronuclear single-quantum coherence spectroscopy first macroscopic binding constant second macroscopic binding constant microscopic binding constant of loop I microscopic binding constant of loop II off-rate on-rate residues 39–44 residues 20–31 residues 56–67 nuclear Overhauser effect spectroscopy total correlation spectroscopy the recombinant N-terminal domain of calmodulin (Ala1–Asp80) with an additional N-terminal Met the recombinant C-terminal domain of calmodulin (Met76–Lys148). Mg2+ is an essential ion in biological systems, with structural and catalytic functions (1Birch N.J. Magnesium and the Cell. Academic Press Ltd., London1993Google Scholar, 2Cowan J. The Biological Chemistry of Magnesium. VCH publishers Inc., New York1995Google Scholar). It is the most abundant divalent metal ion in mammalian cells, with the cytosolic free concentration kept nearly constant at 0.5–2.0 mm (3Ebel H. Gunther T. J. Clin. Chem. Clin. Biochem. 1980; 18: 257-270PubMed Google Scholar). In this milieu, Ca2+ is able to regulate a vast number of cellular activities through transient increases in cytosolic concentration from less than 0.1 μm in a resting cell to 1–10 μm in an activated cell (4Evenäs J. Malmendal A. Forsén S. Curr. Opin. Chem. Biol. 1998; 2: 293-302Crossref PubMed Scopus (61) Google Scholar). Thus, the primary protein targets of Ca2+, in many cases calmodulin (CaM)1 or other EF-hand proteins, must be able to respond in a 100–10,000-fold excess of Mg2+. Due to the high abundance of Mg2+, intracellular Mg2+-specific proteins need no structural discrimination against Ca2+ (5Needham J.V. Chen T. Falke J.J. Biochemistry. 1993; 32: 3363-3367Crossref PubMed Scopus (74) Google Scholar). In contrast, Ca2+-binding proteins may accomplish discrimination against Mg2+ by taking advantage of the larger ionic radius of Ca2+ and its less stringent demands on the number (often 6–8) and spatial arrangement of coordinating oxygen ligands, as compared with Mg2+, which has a strong preference for 6-fold coordination in an octahedral symmetry (6Falke J.J. Drake S.K. Hazard A.L. Peersen O. Q. Rev. Biophys. 1994; 27: 219-290Crossref PubMed Scopus (336) Google Scholar, 7Linse S. Forsén S. Means A.R. Calcium Regulation of Cellular Function. 30. Raven Press, New York1995: 89-152Google Scholar). For example, the Mg2+ affinities of the two sites in toad parvalbumin are about a factor of 6000 lower than the Ca2+ affinities (8Tanokura M. Imaizumi M. Yamada K. FEBS Lett. 1986; 209: 77-82Crossref PubMed Scopus (21) Google Scholar). However, the high cytosolic Mg2+ concentration implies that many Ca2+ sites are occupied by Mg2+ in resting cells. The EF-hand family of Ca2+-binding proteins may be divided into distinct subfamilies, e.g. CaM, troponin C, parvalbumins, and S100 proteins (9Kawasaki, H., and Kretsinger, R. H. (1994) Protein Profile1Google Scholar). In these proteins, Ca2+ binds in the loop region of a 29-residue-long EF-hand helix-loop-helix motif (10Kretsinger R.H. Nockolds C.E. J. Biol. Chem. 1973; 248: 3313-3326Abstract Full Text PDF PubMed Google Scholar). This motif, which is among the five most common protein motifs in animal cells (11Henikoff S. Greene E.A. Pletrokovski S. Bork S. Attwood T.K. Science. 1997; 278: 609-614Crossref PubMed Scopus (302) Google Scholar), usually appears in pairs, where cooperative Ca2+ binding frequently is observed (7Linse S. Forsén S. Means A.R. Calcium Regulation of Cellular Function. 30. Raven Press, New York1995: 89-152Google Scholar). The consensus EF-hand loop comprises 12 residues arranged to coordinate the Ca2+ with pentagonal bipyramid symmetry, with the seven ligands provided by five side chain carboxylate oxygens, one backbone carbonyl oxygen, and one water oxygen (12Strynadka N.C.J. James M.N.G. Annu. Rev. Biochem. 1989; 58: 951-998Crossref PubMed Google Scholar). Two of the side chain ligands are provided by a conserved, bidentate Glu in the 12th and last loop position (Fig. 1 a). Calmodulin, the ubiquitous regulatory Ca2+-binding protein in eukaryotic cells, consists of two distinct domains connected by a flexible tether. The two domains are structurally similar, and each has two EF-hands packed in a roughly parallel fashion with a short β-sheet connecting the Ca2+-binding loops (Fig. 1). The eight helices and four binding loops are denoted A–H and I–IV, respectively. Within each domain, the two EF-hands are connected by a short linker, i.e. between helices B and C and between F and G. Each domain binds two Ca2+ with positive cooperativity (13Linse S. Helmersson A. Forsén S. J. Biol. Chem. 1991; 266: 8050-8054Abstract Full Text PDF PubMed Google Scholar, 14Teleman A. Drakenberg T. Forsén S. Biochim. Biophys. Acta. 1986; 873: 204-213Crossref PubMed Scopus (31) Google Scholar). Upon Ca2+ binding, the secondary structure in both domains remains essentially unchanged, while the relative orientations of the helices change in such a way that the domains go from a relatively compact, "closed" structure (Fig. 1 b) to an "open" structure with well defined hydrophobic patches where target proteins may bind (Fig. 1 c) (15Chattopadhyaya R. Meador W.E. Means A.R. Quiocho F.A. J. Mol. Biol. 1992; 228: 1177-1192Crossref PubMed Scopus (616) Google Scholar, 16Kuboniwa H. Tjandra N. Grzesiek S. Ren H. Klee C.B. Bax A. Nat. Struct. Biol. 1995; 2: 768-776Crossref PubMed Scopus (613) Google Scholar, 17Zhang M. Tanaka T. Ikura M. Nat. Struct. Biol. 1995; 2: 758-767Crossref PubMed Scopus (645) Google Scholar, 18Finn B.E. Evenäs J. Drakenberg T. Waltho J.P. Thulin E. Forsén S. Nat. Struct. Biol. 1995; 2: 777-783Crossref PubMed Scopus (278) Google Scholar, 19Nelson M. Chazin W. Protein Sci. 1998; 7: 270-282Crossref PubMed Scopus (125) Google Scholar). The two domains of CaM can be expressed and produced independently (20Finn B.E. Drakenberg T. Forsén S. FEBS Lett. 1993; 336: 368-374Crossref PubMed Scopus (35) Google Scholar), fold independently (18Finn B.E. Evenäs J. Drakenberg T. Waltho J.P. Thulin E. Forsén S. Nat. Struct. Biol. 1995; 2: 777-783Crossref PubMed Scopus (278) Google Scholar, 21Bentrop D. Bertini I. Cremonini M.A. Forsén S. Luchinat C. Malmendal A. Biochemistry. 1997; 36: 11605-11618Crossref PubMed Scopus (91) Google Scholar), and have Ca2+ binding characteristics similar to intact CaM (13Linse S. Helmersson A. Forsén S. J. Biol. Chem. 1991; 266: 8050-8054Abstract Full Text PDF PubMed Google Scholar). These protein "fragments" were originally produced by trypsin cleavage of CaM in presence of Ca2+ and are named TR1C and TR2C, respectively (22Drabikowski W. Kuznicki J. Grabarek Z. Biochim. Biophys. Acta. 1977; 485: 124-133Crossref PubMed Scopus (76) Google Scholar, 23Walsh M. Stevens F.C. Kuznicki J. Drabikowski W. J. Biol. Chem. 1977; 252: 7440-7443Abstract Full Text PDF PubMed Google Scholar). The Mg2+ dissociation constants of CaM are in the millimolar range (24Drabikowski W. Brzeska H. Venyaminov S.Yu. J. Biol. Chem. 1982; 257: 11584-11590Abstract Full Text PDF PubMed Google Scholar, 25Tsai M.-D. Drakenberg T. Thulin E. Forsén S. Biochemistry. 1987; 26: 3635-3643Crossref PubMed Scopus (80) Google Scholar), and Mg2+ has generally been assumed to bind to the same sites as Ca2+ (25Tsai M.-D. Drakenberg T. Thulin E. Forsén S. Biochemistry. 1987; 26: 3635-3643Crossref PubMed Scopus (80) Google Scholar, 26Seamon K.B. Biochemistry. 1980; 19: 207-215Crossref PubMed Scopus (165) Google Scholar) but to induce only small structural rearrangements (24Drabikowski W. Brzeska H. Venyaminov S.Yu. J. Biol. Chem. 1982; 257: 11584-11590Abstract Full Text PDF PubMed Google Scholar, 26Seamon K.B. Biochemistry. 1980; 19: 207-215Crossref PubMed Scopus (165) Google Scholar). This was recently verified by Ohki et al. using1H–15N NMR (27Ohki S. Ikura M. Zhang M. Biochemistry. 1997; 36: 4309-4316Crossref PubMed Scopus (95) Google Scholar). The Mg2+-loaded form of CaM is reported to cause only negligible activation of CaM target proteins (28Ohki S. Iwamoto U. Aimoto S. Yazawa M. Hikichi K. J. Biol. Chem. 1993; 268: 12388-12392Abstract Full Text PDF PubMed Google Scholar, 29Chao S.-H. Suzuki Y. Zysk J.R. Cheung W.Y. Mol. Pharmacol. 1984; 26: 75-82PubMed Google Scholar). At the time of writing, x-ray structures of Mg2+-loaded EF-hand sites are only available for pike parvalbumin (30Declercq J.P. Tinant B. Parello J. Rambaud J. J. Mol. Biol. 1991; 220: 1017-1039Crossref PubMed Scopus (129) Google Scholar), myosin regulatory light chain (31Houdusse A. Cohen C. Structure. 1996; 4: 21-32Abstract Full Text Full Text PDF PubMed Scopus (191) Google Scholar), and calbindin D9k (32Andersson M. Malmendal A. Linse S. Ivarsson I. Forsén S. Svensson L.A. Protein Sci. 1997; 6: 1139-1147Crossref PubMed Scopus (78) Google Scholar). In parvalbumin and myosin regulatory light chain, the only difference between Mg2+ and Ca2+ligation is that the residues in the 12th loop positions serve as monodentate ligands in the Mg2+ structures but bidentate in the Ca2+ structures. In calbindin D9k, the Glu in the 12th position is not used for direct Mg2+ ligation. Instead, a water molecule is inserted between the side chain and Mg2+. The Glu in the 12th loop position has been shown to be very important for the structural rearrangements from a "closed" to an "open" conformation occurring upon Ca2+ binding (33Evenäs J. Thulin E. Malmendal A. Forsén S. Carlström G. Biochemistry. 1997; 36: 3448-3457Crossref PubMed Scopus (70) Google Scholar, 34Evenäs J. Malmendal A. Thulin E. Carlström G. Forsén S. Biochemistry. 1998; 37: 13744-13754Crossref PubMed Scopus (67) Google Scholar, 35Gagné S.M. Li M.X. Sykes B.D. Biochemistry. 1997; 36: 4386-4392Crossref PubMed Scopus (113) Google Scholar). In the present study, the TR1C fragment of vertebrate CaM was titrated by Mg2+ and followed by1H–15N NMR, in order to address the questions regarding the detailed Mg2+ binding characteristics of this CaM domain and the structural and dynamic nature of protein states at different levels of Mg2+ saturation. The synthetic gene for TR1C was constructed from overlapping oligonucleotides, 2P. Brodin, unpublished results.essentially as described for calbindin D9k (36Brodin P. Grundström T. Hofmann T. Drakenberg T. Thulin E. Forsén S. Biochemistry. 1986; 25: 5371-5377Crossref PubMed Scopus (75) Google Scholar). The TR1C gene was cloned into the pRCB1 plasmid. Unlabeled and uniformly 15N-labeled TR1C was expressed in Escherichia coli and purified as reported previously for the TR2C fragment (18Finn B.E. Evenäs J. Drakenberg T. Waltho J.P. Thulin E. Forsén S. Nat. Struct. Biol. 1995; 2: 777-783Crossref PubMed Scopus (278) Google Scholar). 1H and15N chemical shifts of (Mg2+)2-TR1C were assigned at 25 °C, pH 7.2, on 4 mm protein samples of unlabeled and15N-labeled TR1C in H2O with 10% D2O, 30 mm MgCl2, 10 mmKCl, and 100 μm NaN3. The assignments were obtained from COSY (37Aue W.P. Batholdi E. Ernst R.R. J. Chem. Phys. 1976; 64: 2229-2246Crossref Scopus (3082) Google Scholar), R-COSY (38 ms) (38Wagner G. J. Magn. Reson. 1983; 55: 151-156Google Scholar), 2Q (30 ms) (39Braunschweiler L. Bodenhausen G. Ernst R.R. Mol. Phys. 1983; 48: 535-560Crossref Scopus (427) Google Scholar), TOCSY (110 ms) (40Braunschweiler L. Ernst R.R. J. Magn. Reson. 1983; 53: 521-528Crossref Scopus (3104) Google Scholar, 41Bax A. Davis D.G. J. Magn. Reson. 1985; 65: 355-360Google Scholar) with DIPSI-2rc mixing (42Cavanagh J. Rance M. J. Magn. Reson. 1992; 96: 670-678Google Scholar), and NOESY (120 ms) (43Macura S. Ernst R.R. Mol. Phys. 1980; 41: 95-117Crossref Scopus (1582) Google Scholar) spectra acquired on the unlabeled sample. Sensitivity-enhanced and gradient-selected (44Zhang O. Kay L.E. Oliver J.P. Forman-Kay J.D. J. Biomol. NMR. 1994; 4: 845-858Crossref PubMed Scopus (610) Google Scholar) two-dimensional 15N HSQC-TOCSY with DIPSI-2rc mixing (110 ms) and three-dimensional 15N NOESY-HSQC (150 ms) spectra were acquired on the15N-labeled sample. Water was suppressed by weak presaturation (1.3 s) in the 1H NMR experiments, and by water-flip-back pulses (45Grzesiek S. Bax A. J. Am. Chem. Soc. 1993; 115: 12593-12594Crossref Scopus (1009) Google Scholar) in the 1H-15N experiments. The MgCl2 titration on TR1C monitored by1H-15N HSQC was performed at 25 °C, pH 7.5, on a 0.46 mm15N-labeled TR1C sample in H2O with 10% D2O, 150 mmKCl, 100 μm NaN3, and 10 μmdimethylsilapentanesulfonic acid. Aliquots of Mg2+ were added as solutions of MgCl2. The resulting Mg2+concentrations were 0, 0.2, 0.4, 0.8, 1.2, 2.0, 3.6, 7.4, 18, 40, 81, and 190 mm. The final protein and Mg2+concentrations were determined by amino acid hydrolysis and atomic absorption spectrophotometry, respectively. At each titration point, pH was adjusted by microliter additions of 0.1 m HCl or KOH. The (Ca2+)2-TR1C sample consisted of 0.2 mm15N-labeled TR1C in H2O with 10% D2O, 2 mmCaCl2, 150 mm KCl, 100 μmNaN3, and 10 μm dimethylsilapentanesulfonic acid at pH 7.5. Amide 1H and 15N chemical shifts were followed using sensitivity-enhanced and gradient-selected two-dimensional HSQC spectra, recorded with spectral widths of 1600 and 7692 Hz, sampled over 256 and 2048 complex data points in the 15N and1H dimension, respectively. Using 18 scans pert 1-increment and a relaxation delay of 1.5 s, the total experimental time was 3.5 h/spectrum. 15N nuclei were decoupled during acquisition using the GARP-1 sequence (46Shaka A.J. Barker P.B. Freeman R. J. Magn. Reson. 1985; 64: 547-552Google Scholar). All NMR spectra were recorded on a Varian Unity Plus spectrometer at a 1H frequency of 599.89 MHz. 1H chemical shifts were referenced to dimethylsilapentanesulfonic acid at 0 ppm and15N chemical shifts indirectly via the 1H frequency using the frequency ratio (15N/1H) of 0.101329118 (47Wishart D.S. Bigham C.G. Yao J. Abildgaard F. Dyson H. Oldfield J. Markley J.L. Sykes B.D. J. Biomol. NMR. 1995; 6: 135-140Crossref PubMed Scopus (2059) Google Scholar). Amide chemical shifts were measured in the HSQC spectra at different Mg2+ concentrations. The spectra were processed for either resolution or sensitivity, using Lorentzian-Gaussian or Lorentzian line-broadening window functions in ω2, and Kaiser or sine squared window functions in ω1. After zero filling in ω1, the matrix size was 1024 × 512 real points. Amide line widths were measured from HSQC spectra processed using a Lorentzian line-broadening window function in both dimensions and zero-filling to 1024 points in ω1. The final matrix size was 1024 × 1024. NMR line widths were determined from the HSQC spectra using the in house curve fitting software CFIT. 3G. P. Gippert, unpublished results., 4Source code and instructions for GENXPK and CFIT are available on the World Wide Web athttp://www.fkem2.lth.se/∼garry/programs.html.Fitting was performed by minimizing the error squared sum between a one-dimensional slice taken through the peak center and a pure Lorentzian line shape, as exemplified for Gly33 HN in Fig. 2. The Levenberg-Marquardt algorithm (48Press W.H. Flannery B.P. Teukolsky S.A. Vetterling W.T. Numerical Recipes: The Art of Scientific Computing. Cambridge University Press, Cambridge1986Google Scholar) was used, and starting parameters were obtained through an automatic search procedure. Sequential assignments of 1H and15N resonances for (Mg2+)2-TR1C were obtained following standard procedures (49Chazin W.J. Wright P.E. J. Mol. Biol. 1988; 202: 603-622Crossref PubMed Scopus (73) Google Scholar, 50Wüthrich K. NMR of Proteins and Nucleic Acids. John Wiley & Sons, Inc., New York1986Crossref Google Scholar), using the FELIX 95 software (MSI Inc.), GENXPK (51Gippert G.P. New Computational Methods for 3D NMR Data Analysis and Protein Structure Determination in High-dimensional Internal Coordinate Space.Ph.D. thesis. The Scripps Research Institute, La Jolla, CA1995Google Scholar),4 and the in house assignment tool ASSAR. The assignment procedure was facilitated by the close similarity of the chemical shifts to those of the ion-free state of intact CaM, kindly provided by Ad Bax. The complete 1H and15N resonance assignments for (Mg2+)2-TR1C at 25 °C, pH 7.2, are deposited in BioMagResBank. 1H and 15N chemical shifts of ion-free TR1C were assigned using the HSQC-TOCSY spectra and comparisons with the chemical shifts of ion-free intact CaM. Similarly, the Ca2+-loaded form of the protein was assigned using the chemical shifts of Ca2+-loaded intact CaM (52Ikura M. Kay L.E. Bax A. Biochemistry. 1990; 29: 4659-4667Crossref PubMed Scopus (880) Google Scholar). The chemical shifts of the amide resonances in the HSQC spectra at different Mg2+ concentrations were assigned at increasing Mg2+ concentrations using the ion-free assignment and at decreasing Mg2+ concentrations using the (Mg2+)2-TR1C assignment. The binding kinetics of Mg2+occur on the fast to intermediate NMR chemical shift time scale. Nuclei for which the chemical shift changes induced by Mg2+-binding, Δδ (ppm), result in small resonance frequency changes compared with the exchange rate between two states,k ex, are in the fast exchange regime; 2π ν0 Δδ ≪ k ex, where ν0 (MHz) is the spectrometer frequency of the nucleus observed. For these nuclei, the observed chemical shift is a population-weighted average of the shifts of the ion-free and Mg2+-bound forms,δ=δionfree+ΔδI pI+ΔδII pII+ΔδI,II pI,II(Eq. 1) where p I, p II, and p I,II are the relative populations of the protein with an Mg2+ in site I, site II, and both site I and II, respectively, at a given Mg2+ concentration; δion-free is the chemical shift of the ion-free state; and ΔδI, ΔδII, and ΔδI,II are the chemical shift changes induced by Mg2+ binding to sites I, II, and both I and II, respectively. Nuclei experiencing resonance frequency changes of the same order of magnitude as the exchange rate (2π ν0 Δδ ≈k ex) are in the intermediate exchange regime, where resonances are severely broadened by Mg2+ exchange. Their chemical shifts depend not only on the populations but also on the binding kinetics (53Sandström J. Dynamic NMR Spectroscopy. 1st Ed. Academic Press, Inc., London1982Google Scholar). Therefore, only resonances experiencing no or only moderate line broadening were used in the binding constant calculations. For a nucleus experiencing intermediate to fast exchange the contributions to the line width from the exchange process, Δν1/2,ex, can be calculated as follows,Δν1/2,ex=4π ν02(Δδ)2pApBkex(Eq. 2) wherekex=kon[Mg2+]+koff(Eq. 3) and where p A and p B are the relative populations of the two states, k off is the off-rate,k on is the on-rate, and [Mg2+] is the free Mg2+ concentration. In the case where Mg2+ binding to a protein site is studied by adding Mg2+ to a given protein solution, the equation is readily rearranged to the following,Δν1/2,ex=4π ν02(Δδ)2koff×K [Mg2+](1+K [Mg2+])3(Eq. 4) where K is the binding constant to the site (54Campbell I.D. Dobson C.M. Methods Biochem. Anal. 1979; 25: 1-134Crossref PubMed Google Scholar). If line broadening is an effect of fast to intermediate conformational exchange within a certain state, the contribution to the total line width is approximately as follows,Δν1/2,ex=pΔν1/20(Eq. 5) where p is the relative population of the state and is the line width at 100% of that state, which may be calculated from Equation 2. Due to the generally larger changes in resonance frequency, ν0 Δδ, for 1H compared with15N, a larger number of 1H resonances than15N resonances are broadened during the titration. Broadening in the 1H dimension renders the evaluation of15N line widths uncertain. Therefore, 15N line widths will generally not be discussed in this paper. Mg2+ binding constants were derived from chemical shifts and line widths using a simulated annealing algorithm similar to that used previously (33Evenäs J. Thulin E. Malmendal A. Forsén S. Carlström G. Biochemistry. 1997; 36: 3448-3457Crossref PubMed Scopus (70) Google Scholar). In the present study, however, the binding constants were determined for individual residues, and the average was calculated. The microscopic binding constants of loop I and II (K I and K II) were determined from chemical shifts of 12 residues. The ion-free shifts,δ ion-free, were taken directly from the15N HSQC spectrum of the ion-free state. The two microscopic binding constants, K I and K II, and the chemical shift changes induced by binding an Mg2+ to loop I, II, and both loops, ΔδI, ΔδII and ΔδI,II, were determined minimizing the following expression, χ2=∑in(δionfree+ΔδI pI(i)+ΔδII pII(i)+ΔδI,II pI,II(i)−δobs(i))2(Eq. 6) where p I(i),p II(i) and p I, II(i) are the relative populations of the Mg2+ bound to loop I, II, and both I and II, respectively, calculated from the binding constants and the protein concentration, and δobs(i) is the observed chemical shift for the nucleus at Mg2+ concentration i. All of the nuclei chosen had a Mg2+-induced chemical shift change of between 0.04 and 0.15 ppm for 1H and 0.1 and 0.4 ppm for15N and showed no or only moderate broadening. The uncertainties were estimated as the maximal deviation causing a doubling of χ2. The microscopic binding constant of loop I (K I) was also determined from line widths of six residues at or near this loop that experience moderate line broadening (5–20 Hz).K I, the line width without exchange broadening before and after this binding event (ν1/2,nat), and the ratio of the squared Mg2+-induced chemical shift change and the off-rate ((Δδ)2/k off) were determined by minimizing the following expression, χ2=∑inν1/2,nat+4π ν02(Δδ)2koff×KI[Mg2+](1+KI[Mg2+])3−ν1/2,obs2(Eq. 7) where [Mg2+] is the free Mg2+concentration calculated from the fitted K I and a fixed K II, and ν 1/2,obs is the measured line width. The rates of the dynamic processes in loop II were estimated from a comparison of experimental and calculated line shapes. A four-site exchange program, based on the Bloch-McConnell equations (55McConnell H.M. J. Chem. Phys. 1958; 28: 430-431Crossref Scopus (1276) Google Scholar), was used in this analysis. Chemical shifts of backbone and side chain1H–15N pairs were determined from HSQC spectra at 12 Mg2+ concentrations ranging from 0 to 190 mm, in 150 mm KCl at 25 °C, pH 7.5. The titration provides evidence for two binding events, both characterized by dissociation constants in the millimolar range. As shown in Fig. 3, the major chemical shift changes occur in or around the N-terminal parts of the Ca2+ binding loops. The chemical shift changes in these regions are very similar to the Ca2+-induced changes, clearly identifying the location of the Mg2+ binding sites to the Ca2+ binding loops. The first binding event is characterized by effects that are intermediate on the chemical shift time scale; the amide signals of residues Asp20—Ile27 in the N-terminal part of loop I become broadened beyond detection at intermediate Mg2+ concentrations, and a large number of other signals are significantly broadened (Figs. 4 and 5 b). This exchange broadening can be attributed purely to binding to loop I, because all chemical shift changes of the same size do not result in the same degree of line broadening. When Mg2+ binds to loop II, no line width maxima appear at intermediate Mg2+ occupancies, but some signals are continuously broadened as the loop is filled (Fig. 5 c). This indicates faster binding kinetics of this loop and a conformational exchange within the Mg2+-bound form. Sequential binding of two Mg2+ is possible to demonstrate using the present method, because the two events are characterized by different degrees of exchange broadening and, for some signals, chemical shift changes of different signs, cf.Leu69 HN in Fig. 5 a. At the highest Mg2+ concentration, a number of signals that are visible at lower salt concentrations become broadened beyond detection. This may be the result of changes in the time scale of the exchange processes within the (Mg2+)2 state, nonspecific effects of the very high ionic strength, and/or transient aggregation.Figure 4Backbone trace of the ion-free N-terminal domain of calmodulin (16Kuboniwa H. Tjandra N. Grzesiek S. Ren H. Klee C.B. Bax A. Nat. Struct. Biol. 1995; 2: 768-776Crossref PubMed Scopus (613) Google Scholar). Residues with protons experiencing significant exchange broadening during Mg2+ binding to loop I are colored red; those with protons experiencing significant exchange broadening due to conformational exchange in the Mg2+-saturated loop II are colored blue; and those with protons subjected to both effects are coloredgreen. The two loops and the four helices are labeled. Thisfigure was generated using UCSF software Midas Plus (60Ferrin T.E. Huang C.C. Jarvis L.E. Langridge R. J. Mol. Graphics. 1988; 6: 13-27Crossref Scopus (928) Google Scholar).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Figure 5Titration curves of chemical shifts (a) and line widths (b) as a function of MgCl2 concentration and line widths (c) as a function of calculated occupancy of loop II. Symbolsrepresent the measured values, with the error bars showing the uncertainty in the measurement, and lines represent fitted values. a, chemical shift fits are shown for Ser17 HN (open circles, solid line), Thr29 HN (filled boxes, dashed line), Ala57 HN (filled circles, dotted line), and Leu69 HN (open boxes, dotted and dashed line). b, line width fits are shown for Ser17 HN (open circles, solid line), Thr29 HN (filled boxes, dashed line), Val35 HN (filled circles, dotted line), and Leu69 HN (open boxes, dotted and dashed line). c, line width fits are shown for Ala57 HN (filled circles, dotted line), Gly59 HN (open circles, solid line), Gly61 HN (filled boxes, dashed line), and Ile63 HN (open boxes, dotted and dashed line).View Large Image Figure ViewerDownload Hi-res image Download (PPT) The microscopic Mg2+ binding constants (K I and K II) at high salt (150 mm KCl) for loops I and II in TR1C were calculated using data exemplified in Fig. 5, a and b.K I was calculated from the Mg2+-induced line broadening (Fig. 5 b), and K I and K II were calculated from the Mg2+-induced chemical shift changes (Fig. 5 a). The line shape calculations were based on the assumption that contributions to the signals included in the optimization from Mg2+ binding to loop II could be neglected. The chemical shift-based calculations were made using a number of different models, some including cooperative interactions. However, from the present data no additional information was obtained using more complicated models than a model with two independent binding loops. Binding constants obtained from chemical shifts and line shapes agree well. Since the precision of K I was better using line shape analysis, this value was used to calculateK II from the shift changes. The calculated microscopic Mg2+ binding constants are log10 K I = 3.07 ± 0.04 and log10 K II = 2.7 ± 0.2 (Table I). These values agree well with values obtained from Mg2+/Ca2+ competition studies, 5A. Malmendal, unpublished results. and earlier b
Referência(s)