1: %% 05sbd0423.tex for Biophysical Journal
2:
3: \documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
4:
5: \usepackage{bm}
6: \usepackage{epsfig}
7:
8: \newcommand{\nn}{\nonumber}
9:
10: \newcommand{\gdw}{\Leftrightarrow}
11: \newcommand{\tav}[1]{\left\langle #1\right\rangle} %thermischer MW
12: \newcommand{\cav}[1]{\left\langle\left\langle #1\right\rangle\right\rangle} %Kumulanten MW
13: \newcommand{\dav}[1]{\overline{#1}} %Unordnungs MW
14: \newcommand{\ket}[1]{\left| #1\right\rangle}
15: \newcommand{\bra}[1]{\left\langle #1\right|}
16: \newcommand{\scp}[2]{\left<#1\right|\left.#2\right>}
17: \newcommand{\bok}[3]{\left<#1\right|#2\left|#3\right>} %bra-operator-ket
18: \newcommand{\comm}[2]{\left[#1,#2\right]}
19: \newcommand{\gb}[1]{\left[#1\right]}
20: \renewcommand{\Pr}[1]{\mathrm{Pr}\left\{#1\right\}}
21: \newcommand{\csch} {\mathrm{csch}}
22: \newcommand{\sgn} {\mathrm{sgn}}
23: \newcommand{\Tr} {\mathrm{Tr}}
24:
25: %\renewcommand{\baselinestretch}{4}
26: %\newcommand{\mat}[1]{\underline{\mathbf{#1}}}
27:
28: \newcommand{\comment}[1]{$\qquad\bullet$ \texttt{ #1}\\}
29:
30: \begin{document}
31: \begin{widetext}
32: %%% \noindent{file: 05sbd0314.tex}
33: %%% \vskip 2 mm
34:
35: \noindent{\Large\bf Folding of the Protein Domain hbSBD}
36: \vskip 5 mm
37:
38:
39: %%% \title{Folding of the hbSBD domain}
40:
41: \noindent{\bf Maksim Kouza$^{1,2}$,
42: Chi-Fon Chang$^{3}$, Shura Hayryan$^2$, Tsan-hung Yu$^4$,\\ Mai Suan Li$^1$,
43: Tai-huang Huang$^{4,5}$, and Chin-Kun Hu$^{2,6}$}
44: \vskip 2 mm
45:
46: \noindent{\it $^1$Institute of Physics, Polish Academy of Sciences,
47: Al. Lotnikow 32/46, 02-668 Warszawa, Poland}
48:
49: \noindent{\it $^2$Institute of Physics, Academia Sinica, Nankang,
50: Taipei 11529, Taiwan}
51:
52: \noindent{\it $^3$Genomics Research Center, Academia Sinica, Nankang,
53: Taipei 11529, Taiwan}
54:
55: \noindent{\it $^4$Institute of Biomedical Sciences, Academia Sinica, Nankang,
56: Taipei 11529, Taiwan}
57:
58: \noindent{\it $^5$Department of Physics, National Taiwan Normal University,
59: Taipei 11718, Taiwan.}
60:
61: \noindent{\it $^6$National Center for Theoretical Sciences at
62: Taipei, Physics Division, National Taiwan University, Taipei
63: 10617, Taiwan}
64:
65: \end{widetext}
66:
67:
68:
69:
70: %%% \author{Maksim Kouza$^{1,2}$,
71: %%% Chi-Fon Chang$^{3}$, Shura Hayryan$^2$, Tsan-hung Yu$^4$,
72: %%% Mai Suan Li$^1$, T. H. Huang$^{4,5}$, and Chi-Kun Hu$^{2,6}$}
73:
74: %%% \vskip 10 mm
75: %%% \affiliation{$^1$Institute of Physics, Polish Academy of Sciences,
76: %%% Al. Lotnikow 32/46, 02-668 Warszawa, Poland}
77: %%% \affiliation{$^2$Institute of Physics, Academia Sinica, Nankang,
78: %%% Taipei 11529, Taiwan}
79: %%% \affiliation{$^3$Genomics Research Center, Academia Sinica, Nankang,
80: %%% Taipei 11529, Taiwan}
81: %%% \affiliation{$^4$Institute of Biomedical sciences, Academia Sinica, Nankang,
82: %%% Taipei 11529, Taiwan}
83: %%% \affiliation{$^5$ Department of Physics, National Taiwan
84: %%% Normal University, Taipei, Taiwan.}
85: %%% \affiliation{$^6$National
86: %%% Center for Theoretical Sciences at Taipei, Physics Division,
87: %%% National Taiwan University, Taipei 10617, Taiwan}
88: \vskip 5 mm \noindent{\bf ABSTRACT.
89: %%% \begin{abstract}
90: %% The coil-helix (native state) transitions of single domain
91: %% $\alpha$ proteins have attracted much attention because they can
92: %% be either one-state or two-state folding.
93: The folding of the $\alpha$-helice domain hbSBD of the mammalian
94: mitochondrial branched-chain $\alpha$-ketoacid dehydrogenase
95: (BCKD) complex is studied by the circular dichroism technique in
96: absence of urea. Thermal denaturation is used to evaluate various
97: thermodynamic parameters defining the equilibrium unfolding, which
98: is well described by the two-state model with the folding
99: temperature $ T_F = 317.8 \pm 1.95$ K and the enthalpy change
100: $\Delta H_G = 19.67 \pm 2.67$ kcal/mol. The folding is also
101: studied numerically using the off-lattice coarse-grained Go model
102: and the Langevin dynamics. The obtained results, including the
103: population of the native basin, the free energy landscape as a
104: function of the number of native contacts and the folding
105: kinetics, also suggest that the hbSBD domain is a two-state
106: folder. These results are consistent with the biological function
107: of hbSBD in BCKD.}
108: %% The satisfactory agrrement between the experiments and
109: %% simulations was achieved for the structural
110: %% cooperativity of the thermal denaturation transition.}
111: %%% \end{abstract}
112:
113: %%% \maketitle
114: % HERE
115: %%% \section{Introduction}
116: %% \vskip 2 mm
117: %% \noindent
118: %% {\it Keywords}: protein folding; native state; circular dichroism;
119: %% Go model; two-state folding.
120: %% \vskip 5 mm
121:
122:
123: %% {\bf Abbreviations used:} CD, circular dichroism; BCKD,
124: %% branched-chain $\alpha$-ketoacid dehydrogenase;
125: %% hbLBD, human-bearing lipoyl-bearing domain;
126: %% hbSBD, human-bearing subunit-binding domain;
127: %% DS, denaturated state;
128: %% FS, folded state;
129: %% TS, transition state.
130:
131:
132: \newpage
133:
134: \noindent
135: {\Large \bf Introduction}
136: \vskip 2 mm
137:
138: \noindent
139: Understanding the dynamics and mechanism of protein
140: folding remains one of the most challenging problems in molecular
141: biology \cite{Fersht03}. Single domain $\alpha$ proteins attract
142: much attention of researchers because most of them fold faster
143: than $\beta$ and $\alpha\beta$ proteins \cite{Jackson98,Eaton04}
144: due to relatively simple energy landscapes and one can, therefore,
145: use them to probe main aspects of the funnel theory
146: \cite{Wolynes95}. Recently, the study of this class of proteins
147: becomes even more attractive because the one-state or downhill
148: folding may occur in some small $\alpha$ proteins
149: \cite{Munoz02,Munoz04,Ferguson04}.
150:
151: The mammalian mitochondrial branched-chain $\alpha$-ketoacid
152: dehydrogenase (BCKD) complex catalyzes the oxidative decarboxylation
153: of branched-chain $\alpha$-ketoacids derived from leucine, isoleucine
154: and valine to give rise to branched-chain acyl-CoAs.
155: In patients with inherited maple syrup urine disease,
156: the activity of the BCKD complex is deficient, which
157: is manifested by often fatal acidosis and mental retardation \cite{ccf1}.
158: The BCKD multi-enzyme complex (4,000 KDa in size) is organized about
159: a cubic 24-mer core of dihydrolipoyl transacylase (E2), with multiple
160: copies of hetero-tetrameric decarboxylase (E1), a homodimeric
161: dihydrogenase (E3), a kinase (BCK) and a phosphatase attached
162: through ionic interactions. The E2 chain of the human BCKD
163: complex, similar to other related multi-functional enzymes
164: \cite{ccf2}, consists of three domains: The amino-terminal
165: lipoyl-bearing domain (hbLBD, 1-84), the interim E1/E3
166: subunit-binding domain (hbSBD, 104-152) and the carboxy-terminal
167: inner-core domain. The structures of these domains serve as bases
168: for modeling interactions of the E2 component with other
169: components of $\alpha$-ketoacid dehydrogenase complexes. The
170: structure of hbSBD (Fig. 1) has been determined by NMR
171: spectroscopy, and the main function of the hbSBD is to attach both
172: E1 and E3 to the E2 core \cite{ccf3}. The two-helix structure of
173: this domain is reminiscent of the small protein BBL
174: \cite{Ferguson04} which may be a good candidate for observation of
175: downhill folding \cite{Munoz02,Munoz04}. So the study of hbSBD is
176: interesting not only because of the important biological role of
177: the BCKD complex in human metabolism but also for illuminating
178: folding mechanisms.
179:
180: From the biological point of view, hbSBD could be less stable than
181: hbLBD and one of our goals is, therefore, to check this by the
182: circular dichroism (CD) experiments. In this paper we study the
183: thermal folding-unfolding transition in the hbSBD by the CD
184: technique in the absence of urea and pH=7.5. Our thermodynamic
185: data do not show evidence for the downhill folding and they are
186: well fitted by the two-state model. We obtained folding
187: temperature $ T_F = 317.8 \pm 1.95$ K and the transition enthalpy
188: $\Delta H_G = 19.67 \pm 2.67$ kcal/mol. Comparison of such
189: thermodynamic parameters of hbSBD with those for hbLBD shows that
190: hbSBD is indeed less stable as required by its biological
191: function. However, the value of $\Delta H_G$ for hbSBD is still
192: higher than those of two-state $\alpha$ proteins reported in
193: \cite{Eaton04}, which indicates that the folding process in the
194: hbSBD domain is highly cooperative.
195:
196: % FIGURE 1
197:
198: From the theoretical point of view it is very interesting to
199: establish if the two-state foldability of hbSBD can be captured by
200: some model. The all-atom model would be the best choice for a
201: detailed description of the system but the study of hbSBD requires
202: very expensive CPU simulations. Therefore we employed the
203: off-lattice coarse-grained Go-like model \cite{Go,Clementi00}
204: which is simple and allows for a thorough characterization of
205: folding properties. In this model amino acids are represented by
206: point particles or beads located at positions of $C_{\alpha}$
207: atoms. The Go model is defined through the experimentally
208: determined native structure \cite{ccf3}, and it captures essential
209: aspects of the important role played by the native structure
210: \cite{Clementi00,Takada99}.
211:
212: It should be noted that
213: the Go model by itself can not be employed to ascertain the two-state behavior
214: of proteins.
215: However, one can use it in conjunction with experiments providing
216: the two-state folding because this model does not {\it always} provide
217: the two-state behavior as have been clearly shown in the seminal work
218: of Clementi {\it et al.} \cite{Clementi00}. In fact,
219: the Go model correctly captures not only the two-state folding of
220: proteins CI2 and SH3 (more two-state Go folders may be found
221: in Ref. \cite{Koga01})
222: but also intermediates of the three-state folder
223: barnase, RNAse H and CheY \cite{Clementi00}.
224: The reason for this is that
225: the simple Go model ignores the energetic frustration but it still takes
226: the topological frustration into account.
227: Therefore, it can capture intermediates
228: that occur due to topological constraints but not those
229: emerging from the frustration of the contact interactions.
230:
231: With the help of Langevin dynamics
232: simulations and the histogram method \cite{Ferrenberg89} we have
233: shown that, in agreement with our CD data, hbSBD is a two-state
234: folder with a well-defined transition state (TS) in the free
235: energy landscape. The two helix regions were found to be
236: highly structured in the TS. The two-state behavior of hbSBD
237: is also supported by our kinetics study
238: showing that the folding kinetics follows the single exponential scenario.
239: The two-state folding obtained in our simulations suggests that for hbSBD
240: the topological frustration is more important than the energetic factor.
241:
242: The dimensionless quantity, $\Omega _c$
243: \cite{KlimThirum98FD}, which characterizes the structural
244: cooperativity of the thermal denaturation transition was computed
245: and the reasonable agreement between the CD experiments and Go
246: simulations was obtained. Incorporation of side chains may give a
247: better agreement \cite{KlimThirum98FD,Li_Physica05} but this
248: problem is beyond the scope of the present paper.
249:
250: \vskip 6 mm
251: %%% \section{Materials and Methods}
252: \noindent{\Large \bf {Materials and Methods}} \vskip 2 mm
253:
254: \noindent{\bf Sample Preparation}
255:
256: \vskip 2 mm
257:
258: \noindent
259: hbSBD protein was purified from the BL21(DE3) strain of
260: \textit{E. coli }containing a plasmid that carried the gene of
261: hbLBD(1-84), a TEV cleavage site in the linker region, and hbSBD
262: (104-152), generously provided to us by Dr. D.T. Chuang of the
263: Southwestern Medical Center, University of Texas. There is an
264: extra glycine in front of Glu104 which is left over after TEV
265: cleavage, and extra leucine,
266: glutamic acid at the C-terminus before six histidine residues.
267: The protein was purified by Ni-NTA affinity chromatography, and the
268: purity of the protein was found to be better than 95\%,
269: based on the Coomassie blue-stained gel. The complete sequence of
270: $N=52$ residues for hbSBD is\\
271: (G)EIKGRKTLATPAVRRLAMENNIKLSEVVGSGKDGRILKEDILNYLEKQT(L)(E).
272:
273:
274: \vskip 2 mm
275:
276: \noindent{\bf Circular Dichroism}
277:
278: \vskip 2 mm
279:
280: \noindent
281: CD measurements were carried out in Aviv CD spectrometer model 202
282: with temperature and stir control units at different temperature
283: taken from
284: 260nm to 195nm. All experiments were carried at 1 nm bandwidth
285: in 1.0 cm quartz square cuvette thermostated to $\pm 0.1^o$C.
286: Protein concentration ($\sim$ 50 uM) was determined by UV absorbance at 280nm
287: using $\epsilon _{280nm}$=1280 M$^{-1}$cm$^{-1}$ with 50mM phosphate buffer at pH7.5.
288: Temperature control was achieved using a circulating water bath system,
289: and the equilibrium time was three minutes for each temperature point.
290: The data
291: was collected at each 2K increment in temperature. The study was
292: performed at heating rate
293: of 10$^{o}$C/min and equilibration time of 3 minutes.
294: The volume changes as a result of thermal expansion as well
295: as evaporation of water were neglected.
296:
297: \vskip 2 mm \noindent {\bf Fitting Procedure}
298:
299: \vskip 2 mm
300:
301: \noindent
302: Suppose the thermal denaturation is a two-state
303: transition, we can write the ellipticity as
304: \begin{equation}
305: \theta \; \, = \; \, \theta _D + (\theta _N - \theta _D)f_N \, ,
306: \label{theta_fN_eq}
307: \end{equation}
308: where $\theta _D$ and $\theta _N$ are values for the denaturated
309: and folded states. The fraction of the folded conformation
310: $f_N$ is expressed as \cite{Privalov79}
311: \begin{eqnarray}
312: f_N \; \, &=& \; \, \frac{1}{1 + \exp (-\Delta G_T/T)} \, ,\nonumber \\
313: \Delta G_T \; \, &=& \; \, \Delta H_T - T\Delta S_T \; = \;
314: \Delta H_G\left(1 - \frac{T}{T_G}\right) \nonumber \\
315: &+&\Delta C_p \left[(T-T_G) - T \ln \frac{T}{T_G}\right] \, .
316: \label{fN_twostate_eq}
317: \end{eqnarray}
318: Here $\Delta H_G$ and $\Delta C_p$ are jumps of the enthalpy
319: and heat capacity at the mid-point temperature $T_G$ (also known
320: as melting or
321: folding temperature) of thermal transition, respectively.
322: Some other thermodynamic characterization of stability
323: such as the temperature of maximum stability ($T_S$), the
324: temperature with zero enthalpy ($T_H$), and the conformational
325: stability ($\Delta G_S$) at $T_S$ can be computed
326: from results of regression analysis \cite{Becktel87}
327: \begin{eqnarray}
328: \ln \frac{T_G}{T_S} \; &=& \; \frac{\Delta H_G}{T_G\Delta C_p}, \\
329: T_H \; &=& \; T_G - \frac{\Delta H_G}{\Delta C_p}, \\
330: \Delta G_S \; &=& \; \Delta C_p (T_S - T_H).
331: \label{parameters_eq}
332: \end{eqnarray}
333: Using Eq. (\ref{theta_fN_eq}) - Eq. (\ref{parameters_eq})
334: we can obtain all thermodynamic parameters from CD data.
335:
336: It should be noted that the fitting of Eq. \ref{fN_twostate_eq}
337: with $\Delta C_p > 0$ allows for an additional cold denaturation
338: \cite{Privalov90} at temperatures much lower than the room
339: temperature . The temperature of such a transition, $T_G'$, may be
340: obtained by the same fitting procedure with an additional
341: constraint of $\Delta H_G <0$. Since the cold denaturation
342: transition is not seen in Go models, to compare the simulation
343: results to the experimental ones we also use the approximation in
344: which $\Delta C_p=0$.
345:
346: \vskip 2 mm \noindent {\bf Simulation} \vskip 2 mm
347:
348: \noindent
349: We use coarse-grained continuum representation for hbSBD
350: protein, in which only the positions of 52 C$_{\alpha}$-carbons
351: are retained. We adopt the off-lattice version of the Go model
352: \cite{Go} where the interaction between residues forming native
353: contacts is assumed to be attractive and the non-native
354: interactions - repulsive. Thus, by definition for the Go model the
355: PDB structure is the native
356: structure with the lowest energy.
357: The advantage of this model is its simplicity which allows one to
358: study model proteins in detail.
359: Following Ref. \onlinecite{Clementi00}, we write
360: the energy of the Go-like model as
361: \begin{eqnarray}
362: &E& \; = \; \sum_{bonds} K_r (r_{i,i+1} - r_{0i,i+1})^2 + \sum_{angles} K_{\theta}
363: (\theta_i - \theta_{0i})^2 \nonumber \\
364: &+& \sum_{dihedral} \{ K_{\phi}^{(1)}
365: [1 - \cos (\Delta \phi_i)] +
366: K_{\phi}^{(3)} [1 - \cos 3(\Delta \phi_i)] \} \nonumber\\
367: &+& \sum_{i>j-3}^{NC} \epsilon_1 \left[ 5 R_{ij}^{12} - 6 R_{ij}^{10}\right] +
368: \sum_{i>j-3}^{NNC} \epsilon_2 \left(\frac{C}{r_{ij}}\right)^{12} .
369: \label{Hamiltonian}
370: \end{eqnarray}
371: Here $\Delta \phi_i=\phi_i - \phi_{0i}$, $R_{ij}={r_{0ij}}/{r_{ij}}$;
372: $r_{i,i+1}$, $\theta_i$ and $\phi_i$ stand for the $i$th bond length between
373: the $i$th and $(i+1)$th residues, the bond angle
374: between the $(i-1)$th and $i$th bonds
375: and the dihedral angle around the $i$th bond, respectively.
376: $r_{ij}$ is the distance between the $i$th and $j$th residues.
377: Subscript ``0'', ``NC'' and ``NNC'' refer to the native
378: conformation, native contacts and non-native contacts,
379: respectively. The first harmonic term keeps the chain
380: connectivity, while the second and third terms represent the local
381: angular interactions. Two last terms are non-local interactions, where
382: the former includes native contact interactions and the latter is
383: nonspecific repulsion between non-native pairs. We choose $K_r =
384: 100 \epsilon_H$, $K_{\theta} = 20 \epsilon_H,
385: K_{\phi}^{(1)} = \epsilon_H,
386: K_{\phi}^{(3)} = 0.5\epsilon_H, \epsilon_1 = \epsilon_H
387: , \epsilon_2 = \epsilon_H$ and $C = 4$ \AA, where $\epsilon_H$ is the
388: characteristic hydrogen bond energy \cite{Clementi00}.
389:
390: The nativeness of any configuration is measured by the number of
391: native contacts $Q$. We define that the $i$th and $j$th residues
392: are in the native contact if $r_{0ij}$ is smaller than a cutoff
393: distance $d_c$ taken to be $d_c = 7.5$ \AA,
394: where $r_{0ij}$ is the distance between the $i$th and $j$th residues in
395: the native conformation. Using this definition and the native
396: conformation of Ref. \onlinecite{ccf3}, we found that the total
397: number of native contacts $Q_{total}$ is $62$. To study the
398: probability of being in the native state we use the following
399: overlap function \cite{Camacho93PNAS}
400: \begin{equation}
401: \chi \; = \frac{1}{Q_{total}} \sum_{i<j+1}^N \,\;
402: \theta (1.2r_{0ij} - r_{ij}) \Delta_{ij}
403: \label{chi_eq}
404: \end{equation}
405: where
406: $\Delta_{ij}$ is equal to 1 if residues $i$ and $j$ form a native
407: contact and 0 otherwise and $\theta (x)$ is the Heaviside
408: function. The argument of this function guarantees that
409: a native contact between $i$ and $j$ is classified as formed
410: when $r_{ij}$ is shorter than 1.2$r_{0ij}$ \cite{Clementi00}
411:
412: The overlap function $\chi$, which is one if the
413: conformation of the polypeptide chain coincides with the native
414: structure and zero for unfolded conformations, can serve as an
415: order parameter for the folding-unfolding transition. The
416: probability of being in the native state, $f_N$, which can be
417: measured by the CD and other experimental techniques, is defined
418: as $f_N = <\chi>$, where $<...>$ stands for a thermal average.
419:
420: The dynamics of the system is obtained by integrating the following Langevin
421: equation \cite{Allen_book}
422: \begin{equation}
423: m\frac{d^2\vec{r}}{dt^2} \; \; = \; \;
424: - \zeta \frac{d\vec{r}}{dt} + \vec{F}_c + \vec{\Gamma},
425: \label{DynaEq_eq}
426: \end{equation}
427: where $m$ is the mass of a bead, $\zeta$ is the friction coefficient,
428: $\vec{F}_c = dE/d\vec{r}$. The random force $\vec{\Gamma}$ is a white noise,
429: i.e. $<\Gamma_i(t) \Gamma_j(t')> = 2\zeta k_BT\delta_{ij}\delta(t-t')$,
430: where $i$ and $j$ refer to components $x,y$ and $z$.
431: It should be noted that the folding thermodynamics
432: does not depend on the enviroment viscosity (or on $\zeta$)
433: but the folding kinetics depends
434: on it \cite{KlimThirumPRL97}. We chose the dimensionless parameter
435: $\tilde{\zeta} = (\frac{a^2}{m\epsilon_H})^{1/2}\zeta = 8$, where
436: $m$ is the mass of a bead and $a$ is the bond length between successive beads.
437: One can show that this value of $\tilde{\zeta}$ belongs to the
438: interval of the viscosity where the folding
439: kinetics is fast. We have tried other values of $\tilde{\zeta}$
440: but the results
441: remain unchanged qualitatively.
442:
443: We measure time in units of $\tau_L = (ma^2/\epsilon_H)^{1/2}$.
444: Using the typical value $m = 3.10^{-25}$ kg \cite{Veitshans97},
445: $a = 4$\AA $~$ and $\epsilon_H = 0.91$ kcal/mol
446: (this choice of $\epsilon_H$ follows from the requirement
447: that the simulated folding temperature coincides
448: with the experimental one, see below)
449: we obtained $\tau_L \approx 3$ ps.
450: The dynamics Eq. (\ref{DynaEq_eq}) was solved by the Verlet
451: algorithm \cite{Swope82} with the time step $\Delta t = 0.005 \tau_L$.
452: All thermodynamic quantities are obtained by the histogram method
453: \cite{Ferrenberg89}.
454:
455: % FIGURE 2
456:
457:
458:
459: %%% \section{Results}
460: \vskip 6 mm
461: \noindent{\Large \bf Results}
462: \vskip 2 mm
463:
464: %%% \subsection{Circular Dichroism experiments}
465: \noindent{\bf CD Experiments} \vskip 2 mm
466:
467: \noindent
468: The structure of hbSBD
469: is shown in Figure 1. Its conformational stability is investigated
470: in present study by analyzing the unfolding transition induced by
471: temperature as monitored
472: by CD, similar to that described previously \cite{Naik02,Naik04}.
473: The reversibility of thermal denaturation was ascertained by monitoring
474: the return of the CD signal upon cooling from 95$^{o}$C to 22 $^{o}$C;
475: immediately after the conclusion of the thermal transition.
476: The transition was found to be more than 80\% reversible.
477: Loss in reversibility to greater extent was observed on prolonged
478: exposure of the sample to higher temperatures.
479: This loss of reversibility is presumably due to irreversible
480: aggregation or decomposition. Figure 2
481: shows the wavelength dependence
482: of mean residue molar
483: ellipticity of hbSBD at various temperatures between 278K and 363K.
484: In a separate study, the thermal unfolding transition as monitored
485: by ellipticity at 228 nm was found to be independent of hbSBD
486: concentration in the range of 2 uM to 10 uM. It was also found to be
487: unaffected by change in heating rate between 2$^{o}$C/min to 20$^{o}$C/min.
488: These observations suggest absence of stable intermediates in heat
489: induced denaturation of hbSBD. A valley at around 220 nm,
490: characteristics of the helical secondary structure is evident for
491: hbSBD.
492:
493:
494: %FIGURE 3
495:
496:
497:
498:
499: Figure 3 shows the temperature dependence of the
500: population of the native conformation, $f_N$, for wave lengths
501: $\lambda = 208, 212$ and 222 nm. We first try to fit these data to
502: Eq. (\ref{fN_twostate_eq}) with $\Delta C_p \ne 0$. The fitting
503: procedure gives slightly different values for the folding (or
504: melting) temperature and the enthalpy jump for three sets of
505: parameters. Averaging over three values, we obtain $ T_G = 317.8
506: \pm 1.95$ K and $\Delta H_G = 19.67 \pm 2.67$ kcal/mol. Other
507: thermodynamic quantities are shown on the first row of Table 1.
508: The similar fit but with $\Delta C_p=0$ gives the
509: thermodynamic parameters shown on the second row of this table.
510: Since the experimental data are nicely fitted to the two-state
511: model we expect that the downhill scenario does not applied to the
512: hbSBD domain.
513:
514: %%% Table
515:
516: For the experimentally studied temperature interval two types of
517: the two-state fit (\ref{fN_twostate_eq}) with $\Delta C_p=0$ and
518: $\Delta C_p \ne 0$ give almost the same values for $T_G$, $\Delta
519: H_G$ and $\Delta S_G$. However, pronounced different behaviors of
520: the population of the native basin, $f_N$, occur when we
521: interpolate results to the low temperature region (Fig. 4).
522: For the $\Delta C_p=0$ case, $f_N$ approaches
523: the unity as $T \rightarrow 0$ but it goes down for $\Delta C_p
524: \ne 0$. This means that the $\Delta C_p \ne 0$ fit is valid if the
525: second cold denaturation transition may occur at $T_G$'. This
526: phenomenon was observed in single domains as well as in
527: multi-domain globular proteins \cite{Privalov90}. We predict that
528: the cold denaturation of hbSBD may take place at $T_G' \approx
529: 212$ K which is lower than $T_G' \approx 249.8$ K for hbLBD
530: shown on the 4th row of Table 1.
531: It would be of great interest to carry out the cold denaturation
532: experiments in cryo-solvent to elucidate this issue.
533:
534: To compare the stability of the hbSBD domain with the hbLBD domain
535: which has been studied in detail previously \cite{Naik04} we also
536: present the thermodynamic data of the latter on Table 1. Clearly,
537: hbSBD is less stable than hbLBD by its smaller $\Delta G_S$ and
538: lower $T_G$ values. This is consistent with their respective
539: backbone dynamics as revealed by $^{15}$N-T$_1$, $^{15}$N-T$_2$,
540: and $^{15}$N-$^1$H NOE studies of these two domains using
541: uniformly $^{15}$N-labeled protein samples (Chang and Huang,
542: unpublished results). Biologically, hbSBD must bind to either E1
543: or E3 at different stages of the catalytic cycle, thus it needs to
544: be flexible to adapt to local enviroments of the active sites of
545: E1 and E3. On the other hand, the function of hbLBD is to permit
546: its Lys44 residue to channel acetyl group between donor and
547: acceptor molecules and only the Lys44 residue needs to be flexible
548: \cite{Chang_JBC02}. In addition, the NMR observation for the
549: longer fragment (comprising residues 1-168 of the E2 component)
550: also showed that the hbLBD region would remain structured after
551: several months while the hbSBD domain could de-grate in a shorter
552: time.
553:
554:
555: % FIG 4
556:
557:
558:
559: %%% \subsection{Simulations}
560:
561: %%% \noindent{\bf Simulations}
562: \vskip 2 mm \noindent {\bf Folding Thermodynamics from
563: Simulations}
564:
565: \vskip 2 mm
566:
567: \noindent
568:
569: In order to calculate the thermodynamics quantities we have collected
570: histograms for the energy and native contacts
571: at six values of temperature: $T = 0.4, 0.5, 0.6, 0.7,0.8$
572: and 1.0 $\epsilon_H/k_B$. For sampling,
573: at each temperature 30 trajectories
574: of $16\times 10^7$ time steps have been generated with initial
575: $4\times 10^7$ steps discarded for thermalization.
576: The reweighting histogram method \cite{Ferrenberg89} was used
577: to obtain the thermodynamics parameters at all temperatures.
578:
579: Figure 4 (open circles) shows the temperature
580: dependence of population of the native state, defined as the
581: renormalized number of native contacts (see Material and Methods)
582: for the Go model. Since there is no cold denaturation for this model,
583: to obtain the thermodynamic parameters we fit $f_N$ to the
584: two-state model (Eq. \ref{fN_twostate_eq}) with $\Delta C_p=0$.
585:
586: The fit (black curve) works pretty well around the transition
587: temperature but it gets worse at high $T$ due to slow decay of
588: $f_N$ which is characteristic for almost all of theoretical
589: models. In fitting we have chosen the hydrogen bond energy
590: $\epsilon_H = 0.91$ kcal/mol in Hamiltonian (\ref{Hamiltonian}) so
591: that $T_G = 0.7 \epsilon_H/k_B$ coincides with
592: the experimental value 317.8 K. From the
593: fit we obtain $\Delta H_G = 11.46$ kcal/mol which is smaller than
594: the experimental value indicating that the Go model is less
595: stable compared to the real hbSBD.
596:
597: Figure 5 shows the temperature dependence of derivative of the
598: fraction of native contacts with respect to temperature $df_N/dT$
599: (we also call this value the structural susceptibility) and the
600: specific heat $C_v$ obtained from the Go simulations. The collapse
601: temperature $T_{\theta}$, defined as the temperature at which
602: $C_v$ is maximal, almost coincides with the folding temperature
603: $T_F$ (at $T_F$ the structural susceptibility has maximum).
604: According to Klimov and Thirumalai \cite{KlimThirum96PRL},
605: the dimensionless parameter $\sigma = \frac{|T_{\theta}-T_F|}{T_F}$
606: may serve as an indicator for foldablity of proteins. Namely,
607: sequences with $\sigma \leq 0.1$ fold much faster that
608: those which have the same number of residues but with $\sigma$
609: exceeding 0.5. From this perspective, having $\sigma \approx 0$ hbSBD
610: is supposed to be a
611: good folder {\it in silico}. However, one has to be cautious about
612: this conclusion because
613: the pronounced correlation between folding times $\tau _F$
614: and the equilibrium parameter
615: $\sigma$, observed for simple on- and off-lattice models
616: \cite{KlimThirum96PRL,Veitshans97} may be not valid for proteins in laboratory
617: \cite{Plaxco_Rev04}. In our opinion, since the data collected from
618: theoretical and
619: experimental studies are limited, further studies are required to
620: clarify the relationship between $\tau _F$ and $\sigma$.
621:
622: Using experimental values for
623: $T_G$ (as $T_F$) and $\Delta H_G$ and the two-state model with $\Delta C_p
624: =0$ (see Table 1) we can obtain the temperature dependence of the
625: population of native state $f_N$ and, therefore, $df_N/dT$ for
626: hbSBD (Fig. 5). Clearly, the folding-unfolding transition
627: {\it in vitro} is sharper than
628: in the Go modeling. One of possible reasons is that our Go
629: model ignores the side chain which can enhance the cooperativity of
630: the denaturation transition \cite{KlimThirum98FD}.
631:
632:
633: % FIGURE 5
634:
635: The sharpness of the fold-unfolded transition might be characterized
636: quantitatively
637: via the cooperativity index $\Omega _c$ which is defined as follows
638: \cite{Li04}
639: \begin{equation}
640: \Omega_c=\frac{T_F^2}{\Delta T}
641: \biggl(\frac{df_N}{dT}\biggr)_{T=T_F},
642: \end{equation}
643: where $\Delta T$ is the transition width. From Fig. 5, we obtain
644: $\Omega_c = 51.6$ and 71.3 for the Go model and CD experiments,
645: respectively. Given the simplicity of the Go model used here the
646: agreement in $\Omega_c$ should be considered reasonable. We can
647: also estimate $\Omega _c$ from the scaling law suggested in Ref.
648: \onlinecite{Li04a}, $\Omega _c = 0.0057 \times N^{\mu}$, where
649: exponent $\mu$ is universal and expressed via the random walk
650: susceptibility exponent $\gamma$ as $\mu = 1+\gamma
651: \approx 2.22 (\gamma
652: \approx 1.22$). Then we get $\Omega_c \approx 36.7$ which is lower
653: than the experimental as well as simulation result. This means
654: that hbSBD {\em in vitro} is, on average, more cooperative than
655: other two-state folders.
656:
657: Another measure for the cooperativity is
658: $\kappa _2$ which is defined as \cite{Chan00PRL}
659: $\kappa _2 = \Delta H_{vh}/\Delta H_{cal}$, where
660: $\Delta H_{vh} \; = \; 2T_{max}\sqrt{k_B C_V(T_{max})}$
661: and $\Delta H_{cal} \; = \; \int_0^{\infty} C_V(T)dT$,
662: are the van't Hoff and the calorimetric enthalpy, respectively,
663: $C_V(T)$ is the specific heat. Without the baseline substraction
664: in $C_V(T)$ \cite{Chan_ME04}, for the Go model of hbSBD we
665: obtained $\kappa _2 \approx 0.25$. Applying the baseline
666: substraction
667: as shown in the lower part of Fig. 5
668: we got $\kappa _2 \approx 0.5$ which is still much lower than
669: $\kappa _2 \approx 1$ for a trully all-or-none transition.
670: Since $\kappa _2$ is an extensive parameter, its low value
671: is due to the shortcomings of the off-lattice
672: Go models but not due to the finite size effects.
673: More rigid lattice models give better results for
674: the calorimetric cooperativity \cite{Li_Physica05}.
675: Thus, for the hbSBD domain the Go model gives
676: the better agreement with our CD experiments for the
677: structural cooperativity
678: $\Omega_c$ than for the calorimetric measure $\kappa _2$.
679:
680: \vskip 2 mm
681: \noindent{\bf Free Energy Profile}
682: \vskip 2 mm
683:
684: \noindent
685: To get more evidence that hbSBD is a two-state folder we
686: study the free energy profile using some quantity as a reaction
687: coordinate. The precise reaction coordinate for a
688: multi-dimensional process such as protein folding is difficult to
689: ascertain. However, Onuchic and coworkers \cite{Nymeyer98PNAS}
690: have argued that, for minimally frustrated systems such as Go
691: models, the number of native contact $Q$ may be appropriate. Fig.
692: 6(a) shows the dependence of free energy on $Q$ for $T=T_F$. Since
693: there is only one local maximum corresponding to the transition
694: state (TS), hbSBD is a two-state folder. This is not unexpectable
695: for hbSBD which contains only helices. The fact that the simple Go model
696: correctly captures the two-state behavior as was observed in the CD
697: experiments, suggests that the energetic frustration ignored in this model
698: plays a minor role compared to the topological frustration
699: \cite{Clementi00}.
700:
701: % FIGURE 6
702:
703: We have sorted out structures of the denaturated state (DS), TS
704: and the folded state (FS) at $T=T_F$ generating $10^4$
705: conformations in equilibrium. The distributions of the RMSD,
706: $P_{\rm RMSD}$,
707: of these states are
708: plotted in Fig. 6(b). As expected, $P_{\rm RMSD}$ for
709: the DS spreads out more than that for the TS and FS. According to
710: the free energy profile in Fig. 6(a), the TS conformations
711: have 26 - 40 native contacts. We have found that the size (number
712: of folded residues) \cite{Bai04} of the TS is equal to 32. Comparing this size
713: with the total number of residues ($N=52$) we see that the fraction of
714: folded residues in the TS is higher than the typical value
715: for real two-state proteins
716: \cite{Bai04}. This is probably an artefact of Go models \cite{LiKouza05}.
717: The TS conformations are relatively compact having
718: the ratio $<R_g^{TS}>/R_g^{NS} \approx 1.14$, where $<R_g^{TS}>$
719: is the average radius of gyration of the TS ensemble and $R_g^{NS}$
720: is the radius of gyration of the native conformation shown in Fig. 1.
721: Since the RMSD, calculated only for two helices, is about 0.8 \AA
722: the structures of two helices in the TS
723: are not distorted much. It is also evident from
724: the typical structure of the TS shown in Fig.6(b) where
725: the helix regions H$_1$ and H$_2$ involve residues 13 - 19 and 39 - 48,
726: respectively (a residue is considered to be in the helix state if its
727: dihedral angle is about 60$^o$).
728: Note that H$_1$ has two residues less compared to
729: H$_1$ in the native conformation (see the caption to Fig. 1)
730: but H$_2$ has even one bead more than its native state counterpart.
731: Overall, the averaged RMSD of the TS conformations from the
732: native conformation (Fig. 1) is about
733: 4.9 \AA$~$ indicating that the TS is not close to the native one. As seen
734: from Figs. 6(a) and 1, the main difference comes
735: from the tail parts. The most probable conformations
736: (corresponding to maximum of $P_{\rm RMSD}$ in Fig. 6(b) of the FS
737: have RMSD about 2.5 \AA. This value is reasonable from the point
738: of view of the experimental structure resolution.
739:
740: \vskip 2 mm
741: \noindent {\bf Folding Kinetics}
742: \vskip 2 mm
743:
744: \noindent
745: The two-state foldability, obtained from the thermodynamics simulations
746: may be also probed by studying
747: the folding kinetics. For this purpose we monitored the
748: time dependence of the fraction of unfolded trajectories $P_u(t)$ defined
749: as follows \cite{Klimov99}
750: \begin{equation}
751: P_u(t) \, = \, 1 - \int_0^t P^{\textstyle{N}}_{fp}(s)ds,
752: \label{Pu_eq}
753: \end{equation}
754: where $P^{\textstyle{N}}_{fp}$ is the distribution of first passage folding
755: times
756: \begin{equation}
757: P^{\textstyle{N}}_{fp} \, = \, \frac{1}{M} \sum_{i=1}^{M}
758: \delta (s - \tau _{f,1i}).
759: \label{Pn_eq}
760: \end{equation}
761: Here $\tau _{f,1i}$ is time for the $i$th trajectory
762: to reach the native state for the first time,
763: $M$ is the total number of trajectories used in simulations.
764: A trajectory is said to be folded if all of native contacts form. As seen from
765: Eqs \ref{Pu_eq} and \ref{Pn_eq},
766: $P_u(t)$ is the fraction of trajectories which do not reach
767: the native state at time $t$.
768: In the two-state scenario the folding becomes triggered after
769: overcoming only one free energy barrier between the transition state
770: and the denaturated one. Therefore,
771: $P_u(t)$ should be a single exponential,
772: i.e. $P_u(t) \sim \exp(-t/\tau_F)$ (a multi-exponential
773: behavior occurs in the case when the folding proceeds via intermediates)
774: \cite{Klimov99}. Since the function $P_u(t)$ can be measured
775: directly by a number of experimental techniques \cite{Greene04,Dyson05},
776: the single exponential kinetics of two-state folders
777: is supported by a large body of experimental work (see, i.e. Ref.
778: \cite{Naik02} and references there).
779:
780: Fig. 7 shows the semi-logarithmic plot for $P_u(t)$ at $T=T_F$ for
781: the Go model.
782: Since the single exponential fit works pretty well, one can expect
783: that intermediates do not occur on the folding pathways.
784: Thus, together with the thermodynamics data our kinetic study supports
785: the two-state behavior of the hbSBD domain as observed
786: on the CD experiments.
787:
788: From the linear fit in Fig. 7 we obtain the folding time
789: $\tau_F \approx 0.1 \mu$s.
790: This value is consistent with the
791: estimate of the folding time defined as the average value of the
792: first passage times.
793: If we use the empirical formula for the folding time
794: $\tau_F = \tau_F^0 \exp(1.1N^{1/2})$, where prefactor
795: $\tau_F^0=0.4 \mu$s and $N$ is a number of amino acids \cite{Li04}
796: then $\tau_F = 1.1\times 10^3 \mu$s for $N=52$. This value is
797: about four orders of magnitude larger than that obtained from the
798: Go model.
799: Thus the Go model can capture the two-state feature of
800: the denaturation transition for hbSBD domain but not folding
801: times.
802:
803:
804: % FIGURE 7
805:
806:
807: %%% \section{conclusion}
808: \vskip 6 mm
809: \noindent{\Large \bf Discusion}
810: \vskip 2 mm
811:
812: We have used CD technique and the Langevin dynamics to study the
813: mechanism of folding of hbSBD. Our results suggest that this
814: domain is a two-state folder. The CD experiments reveal that the
815: hbSBD domain is less stable than the hbLBD domain in the same
816: BCKD complex, but it is more stable and cooperative compared to
817: other fast folding $\alpha$ proteins.
818:
819: Both the thermodynamics and
820: kinetics results, obtained from the Langevin dynamics
821: simulations, show that the simple Go model correctly captures
822: the two-state feature of folding.
823: It should be noted that the two-state behavior is not the natural
824: consequence of the Go modeling because it allows for fishing folding
825: intermediates caused by the topological frustration. From this standpoint
826: it may be used to decipher the foldability of model proteins
827: for which the topological frustration dominates.
828: The reasonable agreement between
829: the results obtained by the Go modeling
830: and our CD experiments,
831: suggests that the native state topology of hbSBD is more important
832: than the energetic factor.
833:
834: The theoretical model gives the reasonable
835: agreement with the CD experimental data for the structural
836: cooperativity $\Omega _c$. However, the calorimetric cooperativity
837: criterion $\kappa _2 \approx 1$ for two-state folders is hard to
838: fulfill within the Go model. From the $\Delta C_p \ne 0$ fitting
839: procedure we predict that the cold denaturation of hbSBD may occur
840: at $T \approx 212$ K and it would be very interesting to verify
841: this prediction experimentally. We are using the package SMMP
842: \cite{smmp,smmpnew} and a parallel algorithm \cite{jcc01} to
843: perform all-atom simulation of hbSBD to check the relevant
844: results. \vskip 2 mm
845: %%\noindent
846: %% {=================================================}
847: %% \vskip 4 mm
848: %% \noindent{\Large \bf Acknowledgements} \vskip 2 mm
849:
850:
851: This work was
852: supported by KBN grant number 1P03B01827 and
853: by National Science Council in Taiwan under grant numbers
854: No. NSC 93-2112-M-001-027 (to CKH) and No. NSC 92-2113-M-001-056
855: (to THH).
856: %% The structure of hbSBD was determined by NMR method.
857: The NMR spectra used to determine hbSBD structure
858: were obtained at the High-field NMR Core Facility,
859: National Research Program for Genomic Medicine (NRPGM),
860: Taiwan, Republic of China.
861:
862: \vspace{0.2cm}
863:
864: %%% \noindent{\Large \bf References}
865:
866:
867:
868: \begin{thebibliography}{10}
869:
870:
871: \bibitem{Fersht03} Daggett, V. \& Fersht A. R. (2003)
872: Is there a unifying mechanism for protein folding? {\it Trends in
873: Biochem. Sci.} {\bf 28}, 18-25.
874:
875: \bibitem{Jackson98} Jackson, S. E. (1998)
876: How do small single domain proteins fold? {\it Des} {\bf 3},
877: R81-R91.
878:
879: \bibitem{Eaton04} Kubelka, J., Hofrichter, J. \& Eaton, W. A. (2004)
880: The protein folding 'speed limit', {\it Curr Opin Struct Biol}
881: {\bf 14}, 76-88.
882:
883: \bibitem{Wolynes95} Bryngelson,J., Onuchic,J. N., Socci, N. D.
884: \& Wolynes, P. G. (1995)
885: Funnels, pathways, and the energy landscape of protein folding -
886: synthesis.
887: {\it Proteins: Struct. Funct. Genet.} {\bf 21}, 167-195.
888:
889:
890: \bibitem{Munoz02} Garcia-Mira, M. M., Sadqi, M., Fischer, N.
891: Sanchez-Ruiz, J. M. \& Munoz, V. (2002) Experimental
892: identification of downhill protein folding. {\it Science }{\bf
893: 298}, 2191-2195.
894:
895: \bibitem{Munoz04} Oliva, F. Y. \& Munoz, V. (2004)
896: A simple thermodynamic test to discriminate between two-state and
897: downhill folding. {\it J. Am. Chem. Soc.} {\bf 126}, 8596-8597.
898:
899: \bibitem{Ferguson04} Ferguson,N., Schartau, P. J.,
900: Sharpe, T. D., Sato, S. \& Fersht, A. R. (2004) One-state
901: downhill versus conventional protein folding. {\it J. Mol. Biol.}
902: {\bf 344}, 295.
903:
904:
905: \bibitem{ccf1} Chuang, D.T. \& Shih, V. E. (2001) in
906: The Metabolic and Molecular Basis of inherited Disease,
907: eds. Scriver, C.R., Beaudet, A.L., Sly, W.S. \& Valle, D.
908: ( McGraw-Hill, New York), pp.1971-2006.
909:
910: \bibitem{ccf2} Perham, R.N (2000)
911: Swinging Arms and Swinging Domains in
912: Multifunctional Enzymes: Catalytic Machines for Multistep Reactions.
913: {\it Annu. Rev. Biochem.} {\bf 69}, 961-1004.
914:
915: \bibitem{ccf3} C.-F. Chang, Y.-J. Lin, T.A. Yu, J.L. Chuang,
916: D. T.Chuang , and T.-h. Huang (2005) in preparation.
917:
918: \bibitem{Go} Go, N. (1983)
919: Theoretical studies of protein folding.
920: {\it Ann. Rev. Biophys. Bioeng.} {\bf 12} No. 1, 183-210.
921:
922:
923: \bibitem{Clementi00} Clementi, C., Nymeyer H. \& Onuchic, J. (2000)
924: Topological and energetic factors: What determines the structural
925: details of the transition state ensemble and ``en-route''
926: intermediates for protein folding? An investigation
927: for small globular proteins.
928: {\it J. Mol. Biol.} {\bf 298}, 937-953.
929:
930: \bibitem{Takada99} Takada, S. (1999) {\it Go}-ing for
931: the prediction of protein folding mechanisms.
932: {\it Proc. Natl. Acad. Sci. USA} {\bf 96},
933: 11698-11700.
934:
935: \bibitem{Koga01} Koga, N. \& Takaga, S. (2001) Roles of native topology
936: and chain-length scaling in protein folding: A simulation study with
937: a Go-like model. {\it J. Mol. Biol.} {\bf 313}, 171-180.
938:
939: \bibitem{Ferrenberg89} Ferrenberg, A. M. \& Swendsen, R. H. (1989)
940: Optimized Monte-Carlo data analysis.
941: {\it Phys. Rev. Lett.} {\bf 63}, 1195-1198.
942:
943:
944: \bibitem{KlimThirum98FD} Klimov, D. K. \& Thirumalai, D. (1998)
945: Cooperativity in protein folding: from lattice models with
946: sidechains to real proteins. {\it Fold. Des.} {\bf 3}, 127-139.
947:
948: \bibitem{Li_Physica05} Li, M. S., Klimov, D. K. \& Thirumalai, D. (2005)
949: Finite size effects on calorimetric cooperativity of two-state proteins.
950: {\it Physica A} {\bf 350}, 38-44.
951:
952: \bibitem{Privalov79} Privalov, P. L. (1979)
953: Stability of proteins: Small globular proteins.
954: {\it Adv. Prot. Chem. } {\bf 33}, 167-241.
955:
956: \bibitem{Becktel87} Becktel, W. J. \& Schellman, J. A. (1987)
957: Protein stability curves.
958: {\it Biopolymers} {\bf 26}, 1859-1877.
959:
960: \bibitem{Privalov90} Privalov, P. L. (1990)
961: Cold denaturation of proteins.
962: {\it Crit. Rev. Biochem. Mol. Biol.} {\bf 25}, 281-305.
963:
964:
965: \bibitem{Camacho93PNAS} Camacho, C. J. \& Thirumalai, D. (1993)
966: Kinetics and thermodynamics of folding in model proteins.
967: {\it Proc. Natl. Acad. Sci. USA} {\bf 90} 6369-6372.
968:
969: \bibitem{Allen_book} Allen, M. P. \& Tildesley, D. J. (1987)
970: {\it Computer simulations of liquids}, (Oxford Science Pub., Oxford, UK).
971:
972: \bibitem{KlimThirumPRL97} Klimov, D. K. \& Thirumalai, D.
973: (1997) Viscosity dependence of the folding rates of proteins.
974: {\it Phys. Rev. Lett.} {\bf 79}, 317-320.
975:
976: \bibitem{Veitshans97} Veitshans, T., Klimov, D. K. \& Thirumalai, D.
977: (1997) Protein folding kinetics: time scales, pathways and energy landscapes
978: in terms of sequence dependent properties. {\it Folding and Design}
979: {\bf 2}, 1-22.
980:
981: \bibitem{Swope82} Swope, W. C., Andersen, H. C., Berens, P. H.
982: \& Wilson, K. R. (1982) Computer simulation method for the calculation
983: of equilibrium constants for the formation of physical clusters and molecules:
984: Application to small water clusters. {\it J. Chem. Phys.}
985: {\bf 76}, 637-649.
986:
987:
988: \bibitem{Naik02} Naik,M., Chang, Y.-C. \& Huang, T.-h.
989: Folding kinetics of the lipoic acid-bearing domain of human
990: mitochondrial branched chain alpha-ketoacid dehydrogenase complex.
991: (2002) {\it FEBS Lett.} {\bf 530}, 133-138.
992:
993: \bibitem{Naik04} Naik, M. \& Huang, T.-h. (2004)
994: Conformational stability and thermodynamic characterization of
995: the lipoic acid bearing domain of human mitochondrial branched
996: chain alpha-ketoacid dehydrogenase.
997: {\it Protein Sci.} {\bf 13}, 2483-2492.
998:
999: \bibitem{Chang_JBC02} Chang, C.-F., Chou, H.-T., Chuang,J. L., Chuang, D. T.
1000: \& Huang, T.-h. (2002) Solution structure and dynamics of the lipoic
1001: acid-bearing domain of human mitochondrial branched-chain alpha-keto
1002: acid dehydrogenase complex.
1003: {\it J. Bio. Chem} {\bf 277}, 15865-15873.
1004:
1005: \bibitem{KlimThirum96PRL} Klimov, D. K. \& Thirumalai, D. (1996)
1006: Criterion that determines the foldability of proteins.
1007: {\it Phys. Rev. Lett.} {\bf 76}, 4070-4073.
1008:
1009: \bibitem{Plaxco_Rev04} Gillepse, B. \& Plaxco, K. W. (2004)
1010: Using protein folding rates to test protein folding theories.
1011: {\it Annu. Rev. Biochem.} {\bf 73}, 837-859.
1012:
1013:
1014: \bibitem{Li04} Li, M. S., Klimov, D.K. \& Thirumalai,D. (2004)
1015: Thermal denaturation and folding rates of single domain proteins:
1016: size matters. {\it Polymer} {\bf 45}, 573-579.
1017:
1018: \bibitem{Li04a} Li, M. S., Klimov, D. K. \& Thirumalai, D. (2004)
1019: Finite size effects on thermal denaturation of globular proteins.
1020: {\it Phys. Rev. Lett.} {\bf 93}, 268107.
1021:
1022: \bibitem{Chan00PRL} Kaya H \& Chan, H. S. (2000)
1023: Energetic components of cooperative protein folding.
1024: {\it Phys. Rev. Lett.} {\bf 85}, 4823-4826.
1025:
1026: \bibitem{Chan_ME04} Chan, H. S., Shimizu, S. \& Kaya, H (2004)
1027: Cooperativity principles in protein folding. {\it Methods in
1028: Enzymology} {\bf 380}, 350-379.
1029:
1030: \bibitem{Nymeyer98PNAS} Nymeyer, H., Garcia, A. E. \& Onuchic,J.N. (1998)
1031: Folding funnels and frustration in off-lattice minimalist protein landscapes.
1032: {\it Proc. Natl. Acad. Sci. USA} {\bf 95 }, 5921-5926.
1033:
1034: \bibitem{Bai04} Bai, Y., Zhou, Y., Zhou, H. (2004)
1035: Critical nucleation size in the folding of small apparently two-state
1036: proteins. {\it Protein Sci.} {\bf 13}, 1173-1181.
1037:
1038: \bibitem{LiKouza05} Li, M. S. \& Kouza, M. (2005) in preparation.
1039:
1040: \bibitem{Klimov99} Klimov, D. K. \& Thirumalai, D. (1999)
1041: Deciphering the timescales and mechanisms of protein
1042: folding using minimal off-lattice models.
1043: {\it Curr. Opin. Struct. Biol.} {\bf 97}, 97-107.
1044:
1045: \bibitem{Greene04} Greene, L. H. ed. (2004). Elsevier Science.
1046: Investigating Protein Folding, Misfolding and Nonnative States:
1047: Experimental and
1048: Theoretical Methods. {\it Methods} {\bf 34}
1049:
1050: \bibitem{Dyson05} Dyson, H. J. \& Wright, P. E. (2005)
1051: Elucidation of the protein folding landscape by
1052: NMR.
1053: {\it Methods Enzymol.} {\bf 394}, 299-321.
1054:
1055: \bibitem{smmp} Eisenmenger,F., Hansmann, U.H.E., Hayryan, S. \&
1056: Hu,C.-K. (2001) [SMMP] A modern package for simulation of proteins.
1057: {\it Comput. Phys. Commun.} {\bf 138}, 192-212.
1058:
1059: \bibitem{smmpnew} Eisenmenger,F., Hansmann, U.H.E., Hayryan, S. \&
1060: Hu,C.-K. (2005) in preparation.
1061:
1062: \bibitem{jcc01} Hayryan,S., Hu,C.-K., Hu, S.-Y. \& and R.-J. Shang, R. J.
1063: (2001) Multicanonical parallel simulations of proteins with
1064: continuous potentials. {\it J. Comput. Chem.} {\bf 22}, 1287-1296.
1065:
1066: \end{thebibliography}
1067: \newpage
1068: %%% \begin{widetext}
1069: %% \noindent{\bf Supporting Information}
1070: %% \vskip 10 mm \noindent {\bf
1071: %% The complete sequence of amino acids for hbSBD is\\}
1072: %% (G)EIKGRKTLATPAVRRLAMENNIKLSEVVGSGKDGRILKEDILNYLEKQT(L)(E),
1073: %% which has 52 residues.
1074: \vskip 20 mm
1075:
1076: %% \centerline{\Large \bf Table Caption}
1077: \vskip 10 mm
1078:
1079: %% Table 1
1080: \begin{figure}
1081: \epsfxsize=6.5in
1082: %% \epsfxsize=5.0 in
1083: \vspace{0.3in}
1084: \centerline{\epsffile{hbSBDtable1.eps}} \vspace{0.2in}
1085: \end{figure}
1086: \noindent {\bf TABLE 1.} Thermodynamic parameters obtained from
1087: the CD experiments and simulations for hbSBD domain. The results
1088: shown on the first and fourth rows were obtained by fitting
1089: experimental data to the two-state equation (\ref{fN_twostate_eq})
1090: with $\Delta C_p \ne 0$. The second and third rows corresponding
1091: to the fit with $\Delta C_p = 0$. The results for hbLBD are taken
1092: from Ref. \onlinecite{Naik04} for comparison.
1093:
1094:
1095:
1096: %% \begin{table}[h]
1097:
1098: %% \label{parameters_tab}
1099: %% \end{table}
1100:
1101:
1102: \vskip 2 mm
1103:
1104:
1105: \newpage
1106:
1107: \centerline{\Large \bf Figure Captions} \vskip 5 mm
1108:
1109: \noindent {\bf FIGURE 1.} Ribbon representation of the structure
1110: of hbSBD domain. The helix region H$_{1}$ and H$_2$ include residues
1111: Pro12 - Glu20 and Lys39 - Glu47, respectively. \vskip 5 mm
1112:
1113: \noindent {\bf FIGURE 2.} Dependence of the mean residue molar
1114: ellipticity on the wave length for 18 values of temperatures
1115: between 278 and 363 K. \vskip 5 mm
1116:
1117: \noindent {\bf FIGURE 3.} Temperature dependence of the fraction
1118: of folded conformations $f_N$, obtained from the ellipticity
1119: $\theta$ by Eq. (\ref{fN_twostate_eq}), for wave lengths $\lambda$
1120: = 208 (blue circles), 212 (red squares) and 222 nm (green
1121: diamonds). The solid lines corresponds to the two state fit given
1122: by Eq. \ref{fN_twostate_eq} with $\Delta C_p \ne 0$. We obtained
1123: $T_G=T_F= 317.8 \pm 1.9$ K, $\Delta H_G = 19.67 \pm 2.67$ kcal/mol
1124: and $\Delta C_p = 0.387\pm0.054$.\vskip 5 mm
1125:
1126: \noindent {\bf FIGURE 4.} The dependence of $f_N$ for various sets
1127: of parameters. The blue and red curves correspond to the
1128: thermodynamic parameters presented on the first and the second
1129: rows of Table 1, respectively. Open circles refer to simulation
1130: results for the Go model. The solid black curve is the two-state
1131: fit ($\Delta C_p=0$) which gives $\Delta H_G= 11.46$ kcal/mol and
1132: $T_F=317.9$.\vskip 5 mm
1133:
1134: \noindent
1135: {\bf FIGURE 5.} The upper part refers to the temperature
1136: dependence of $df_N/dT$ obtained by the simulations (red) and the
1137: CD experiments (blue). The experimental curve is plotted using
1138: two-state parameters with $\Delta C_p = 0$ (see, the second row on
1139: Table 1). The temperature dependence of the heat capacity $C_V(T)$
1140: is presented in the lower part. The dotted lines illustrate the
1141: base line substraction. The results are averaged over 20 samples.
1142: %%% \label{cv_chi_fig}
1143: \vskip 5 mm
1144:
1145: \noindent {\bf FIGURE 6.} (a) The dependence of free energy on the
1146: number of native contacts $Q$ at $T=T_F$. The typical structures
1147: of the denaturated state , transition state and folded
1148: state are also drawn. The helix regions
1149: H$_1$ (green) and H$_2$ (orange) of the TS structure involve residues 13 - 19
1150: and 39 - 48, respectively. For the FS structure H$_2$ is the same as for
1151: the TS structure but H$_1$ has two residues more (13 - 21).
1152: (b) Distributions of RMSD for three
1153: ensembles shown in (a). The average values of RMSD are equal to 9.8,
1154: 4.9 and 3.2 \AA$~$ for the DS, TS and FS, respectively.
1155:
1156: \vskip 5 mm
1157:
1158: \noindent {\bf FIGURE 7.} The semi-logarithmic plot of the time
1159: dependence of the fraction of unfolded trajectories at $T=T_F$.
1160: The distribution $P_u(t)$
1161: was obtained from first passage times of 400 trajectories, which
1162: start from random conformations.
1163: The straight line corresponds to the fit ln $P_{\rm u}(t) =
1164: -t/\tau_F$, where $\tau _F = 0.1 \mu$s.
1165:
1166: \newpage
1167:
1168:
1169: %% fig. 1
1170: \begin{figure}
1171: \epsfxsize=4.5in \epsfxsize=3.8 in
1172: \centerline{\epsffile{hbSBDfig1.eps}}
1173: %\centerline{\epsffile{hbSBDfig1_new.eps}}
1174: %%\centerline{\epsffile{sbdmean.eps}}
1175: %\vspace{1.2in}
1176: \centerline{\bf FIGURE 1.}
1177: %%% \label{SBD_str_fig}
1178: \end{figure}
1179:
1180: \vskip 2 mm
1181: %% fig. 2
1182: \begin{figure}
1183: \epsfxsize=3.5in \vspace{0.6in}
1184: %\centerline{\epsffile{CD_wl.eps}}
1185: \centerline{\epsffile{hbSBDfig2.eps}} \vspace{0.2in}
1186: %%% \label{CD_wl_fig}
1187: \end{figure}
1188: \centerline{\bf FIGURE 2.}
1189:
1190: \newpage
1191: \vskip 2 mm
1192: %% fig. 3.
1193: \begin{figure}
1194: \epsfxsize=3.5in \vspace{0.3in}
1195: %\centerline{\epsffile{ellip_exp.eps}}
1196: \centerline{\epsffile{hbSBDfig3.eps}} \vspace{0.4in}
1197: %%% \label{ellip_exp_fig}
1198: \end{figure}
1199: \centerline{\bf FIGURE 3.}
1200: \newpage
1201: \vskip 2 mm
1202: %% fig. 4.
1203: \begin{figure}
1204: \epsfxsize=3.2in \vspace{0.3in}
1205: \centerline{\epsffile{hbSBDfig4.eps}} \vspace{0.2in}
1206: %%% \label{fn_sim_fig}
1207: \end{figure}
1208: \centerline{\bf FIGURE 4.}
1209: \newpage
1210: \vskip 2 mm
1211: %% fig. 5
1212: \begin{figure}
1213: \epsfxsize=3.5in \vspace{0.3in}
1214: \centerline{\epsffile{hbSBDfig5.eps}} \vspace{0.2in}
1215: \end{figure}
1216: \centerline{\bf FIGURE 5.}
1217: %%% \label{cv_chi_fig}
1218: \vskip 2 mm
1219:
1220: \newpage
1221:
1222: %% Fig. 6
1223: \begin{figure}
1224: \epsfxsize=6.2in \epsfxsize=5.0 in \vspace{0.3in}
1225: \centerline{\epsffile{hbSBDfig6.eps}} \vspace{0.2in}
1226: \label{free_fig}
1227: \end{figure}
1228: \centerline{\bf FIGURE 6.}
1229:
1230: \newpage
1231:
1232:
1233: \vskip 2 mm
1234: %% fig. 7
1235: \begin{figure}
1236: \epsfxsize=3.5in \vspace{0.3in}
1237: \centerline{\epsffile{hbSBDfig7.eps}} \vspace{0.2in}
1238: %%% \label{kinetic_fig}
1239: \end{figure}
1240: \centerline{\bf FIGURE 7.}
1241:
1242: %% \vskip 5 mm The folding time $\tau_F$ may be computed as the
1243: %% median time of the first passage times or from the single
1244: %% exponential fit.
1245:
1246: %%% \end{widetext}
1247:
1248:
1249:
1250: \end{document}
1251:
1252: \begin{tabular}{|c|c|c|c|c|c|c|c|c|}
1253: \hline
1254: &&$\Delta{H}_G$&$\Delta{C_p}$&$\Delta{S_G}$&&&$\Delta{G_S}$&\\
1255: Domain&$T_G(K)$&(kcal/mol/K)&(kcal/mol/K)&(cal/mol/K)&$T_S(K)$&$T_H(K)$&(kcal/mol)&$T'_G(K)$\\
1256: \hline
1257: SBD(exp)&$317.8\pm1.9$&$19.67\pm2.67$&$0.387\pm0.054$&$61.64\pm7.36$&$270.9\pm2.0$&$267.0\pm2.1$&$1.4\pm0.1$&$212\pm2.5$\\
1258: \hline
1259: SBD(exp)&$317.9\pm2.2$&$20.02\pm3.11$&$0.0$&$62.96\pm9.92$&$-$&$-$&$-$&-\\
1260: \hline
1261: SBD(sim)&$317.9\pm7.95$&$11.46\pm0.29$&$0.0$&$36.05\pm1.85$&$-$&$-$&$-$&-\\
1262: \hline
1263: LBD(exp)&$344.0\pm0.2$&$78.96\pm1.28$&$1.51\pm0.04$&$229.5\pm3.7$&$295.7\pm3.7$&$291.9\pm1.3$&$5.7\pm0.2$&$249.8\pm1.1$\\
1264: \hline
1265: \end{tabular}
1266: