1: %\documentstyle[twocolumn,aps,prl,epsf]{revtex}
2: %\documentstyle[preprint,aps,prd,epsf]{revtex} % power law ref if prb
3: \documentclass[preprint,aps,prb,epsf]{revtex4} % power law ref if prb
4: %\topmargin 0in
5: \include{amsmath}
6: \include{epsfx}
7: \usepackage{graphicx}
8: \begin{document}
9: \title{{\bf Stretching of Proteins in a Uniform Flow}}
10: \author{{\bf P. Szymczak $^1$ and Marek Cieplak$^2$}}
11:
12: \address{
13: $^1$Institute of Theoretical Physics, Warsaw University,
14: ul. Ho\.za 69, 00-681 Warsaw, Poland\\
15: $^2$Institute of Physics, Polish Academy of Sciences,
16: Al. Lotnik\'ow 32/46, 02-668 Warsaw, Poland}
17:
18:
19:
20: \vskip 40pt
21: \noindent {\bf
22: Keywords: protein stretching, protein folding, manipulation of
23: proteins, Go model,
24: molecular dynamics, ubiquitin, integrin}
25:
26: \noindent {PACS numbers: 82.37.Rs, 87.14.Ee, 87.15.-v}
27:
28: \vspace*{1cm}
29:
30:
31: \begin{abstract}
32: {Stretching of a protein by a fluid flow is compared to that in a
33: force-clamp apparatus. The comparison is made
34: within a simple topology-based dynamical model of a protein in
35: which the effects of the flow are implemented using Langevin dynamics.
36: We demonstrate that unfolding induced by a uniform flow shows a richer
37: behavior than that in the force clamp. The dynamics of unfolding
38: is found to depend strongly on the selection of the amino acid,
39: usually one of the termini, which is anchored. These features offer
40: potentially wider diagnostic tools to investigate structure of proteins
41: compared to experiments based on the atomic force microscopy.}
42: \end{abstract}
43:
44: \maketitle
45:
46:
47: \section{Introduction}
48:
49: The deformation of polymers in a flow has been a subject of active research for
50: at least seventy years (see e.g. \cite{kuhn,kramers,gennes}).
51: A recent renewed interest in this topic \cite{smith,perkins1,perkins2,rzehak,cheon}
52: arose due to
53: development of precise experimental techniques allowing for studies of
54: conformations
55: at a single-molecule level. In particular, biological macromolecules
56: such as DNA have been intensely studied in this way
57: \cite{larson,abramchuk,shaqfeq,blossey}.
58: Stretching by a flow is also at the heart of the manipulation technique known as
59: molecular combing used in genomic studies
60: \cite{Strick,Bensimon,BensBens,Austin1,Austin}
61: and in nano-electronics \cite{Inganas}.\\
62:
63: In this paper, we concentrate on analysis of protein unfolding in uniform flow
64: and compare it with unfolding in a force clamp \cite{clampober}, i.e.
65: under the condition of a constant force applied to one of the termini.
66: Theoretical studies on protein stretching in a flow are scarce
67: \cite{lemak1,lemak2} and
68: limited to the minimalist $\beta$ - barrel model.
69: Here, we present a theoretical method to study flow induced
70: deformations of, in principle, any protein and we illustrate it by
71: considering ubiquitin and integrin. These two proteins were chosen
72: because of their contrasting dynamical behavior, as established through
73: simulations \cite{cs1}, when unfolding in a force clamp:
74: ubiquitin unfolds as a rule in a single kinetic step whereas integrin
75: -- in multiple steps, i.e. with several intermediates. Additionally,
76: we also consider synthetic $\alpha$-helices, which are homopolymers,
77: and show a very different behaviour than that seen in complex proteins.\\
78:
79: The model we propose here is coarse grained and it involves no explicit solvent.
80: It is an extension of the Go-like modelling \cite{Goabe} that we have
81: employed in the past
82: to study folding \cite{Hoang,biophysical} and stretching at constant velocity
83: \cite{thermtit,Pastore,ubiq}. In short, a protein is represented by
84: a chain of C$^{\alpha}$ atoms that are tethered by harmonic potentials
85: with minima at 3.8 {
86: \AA}.
87: The effective self-interactions between the atoms are either purely
88: repulsive or are
89: minimum-endowed-contacts of the Lennard-Jones type. The parameters of
90: these potentials
91: reflect existence of the surrounding solvent in the sense that their minima
92: correspond to the experimentally determined distances between the
93: C$^{\alpha}$ atoms of a
94: protein in water.\\
95:
96: The dynamics of the protein is governed by the Langevin equation
97: \begin{equation}
98: m \ddot{{\bf r_i}} = -\gamma (\dot{{\bf r}}_i-{\bf u}({\bf r}_i)) +
99: F^c_i + \Gamma \;\;.
100: \label{lang}
101: \end{equation}
102:
103: Here, ${\bf r_i}$ is the position of $i$'th aminoacid, $F^c_i$ is the
104: net force on it
105: due to contact potentials, $\gamma $ is the friction coefficient, and
106: $\Gamma$ is a white
107: noise term
108: with the dispersion of $\sqrt{2\gamma k_B T}$, where $k_B$ is the
109: Boltzmann constant. Finally, ${\bf u}({\bf r}_i)$ denotes the solvent
110: flow field.\\
111:
112:
113:
114: In principle the model allows for more general description in which
115: both $\gamma$ and $m$
116: are amino acid dependent and, in particular, friction is reduced
117: for hydrophobic residues since hydrophobicity leads
118: to a slip \cite{knudsenprl,lotus}. In this paper, we stay with the spirit
119: of the traditional Go-like modelling in which all features of the
120: protein are assumed to be contained in the native geometry and consider
121: uniform masses and friction coefficients. Also, we neglect the effects of hydrodynamic
122: interactions,
123: which corresponds
124: to the free- draining limit. This is a serious approximation, since the
125: hydrodynamic forces between the particles contain long-range terms decaying as
126: $R^{-1}$ with the interparticle distance. However, since the number of residues
127: in the considered proteins is relatively large (100- 200), the inclusion of
128: hydrodynamic interactions (HI) into the Langevin dynamics scheme would add
129: considerably to the numerical complexity of the problem, rendering an accurate
130: calculation of mean unfolding times unfeasible, particularly in the small-force
131: regime where those times are exceedingly long.\\
132:
133: In fact, the hydrodynamic effects are very rarely taken into account in the
134: numerical simulations of protein folding and unfolding: in the all-atom MD
135: simulations sometimes an explicit solvent is used; but this restricts severely
136: the largest feasible trajectory length. The coarse-grained models, in
137: principle, would be a best starting point for the analysis of the impact of HI
138: on the protein dynamics. However, to the best knowledge of the authors, no
139: detailed studies of the impact of HI on protein folding and unfolding were
140: performed, even though it was argued \cite{Tanaka} that such an effect is
141: expected to be non-negligible.\\
142:
143: Keeping the above in mind, we nevertheless believe that the free-draining case
144: may still provide useful insights on protein dynamics in a flow. This is
145: partially confirmed by the results of the analysis of DNA streching in a
146: uniform flow \cite{perkins2}, where it has been observed that the change in the
147: extension vs flow dependence is very modest when no hydrodynamic interactions
148: are included in the model (see also the respective discussion in
149: \cite{shaqfeq}). A similar conclusion is drawn by Hsieh et al. \cite{hsieh} who
150: compare the results of extensive simulations of the bead-spring DNA models in
151: extensional flow to the experimental data by Perkins et al. \cite{perkins1}.
152: They note that deformation-dependent HI has very little effect on the
153: extensional flow properties of DNA molecules, whereas the rates of unraveling
154: of single long molecule of DNA observed optically in an extensional flow can be
155: even quantitatively predicted by bead–spring models that neglect HI.\\
156:
157: Naturally, due to their highly heterogeneous structure, proteins are much more
158: complex than a DNA chain, thus one cannot expect those results to apply
159: directly to the protein streching in a flow. Still, it seems that a free-
160: draining case may provide a good starting point for understanding, at least on
161: the qualitative level, the properties of the protein in the flow.\\
162:
163:
164: With the use of this simplified model, we demonstrate that flow may stretch
165: proteins to partially unravelled stationary conformations that depend on the
166: flow rate and on the selection of the terminus which is anchored. This is in
167: contrast to stretching in a force clamp, in which the set of intermediate
168: states stays the same whether we fix the C terminus and pull on the N one or do
169: it the other way around. This difference is caused by the fact that a flow
170: generates a non-uniform tension in the polymer.
171: A simple explanation of this phenomenon is
172: presented in Figure 1 for the case of a linearly positioned chain of $N$ beads connected by
173: bonds
174: (in the absence of hydrodynamic interactions):
175: the bond which is most distant from the anchor is pushed by $1/N$ of the
176: force that is experienced by the first bond since the latter accumulates all
177: individual pushes. It follows that, as the flow velocity is increased,
178: the contacts near the anchored end of the protein are broken first and
179: the protein
180: unwinds segment by segment starting from the fixed end. Similar phenomena are
181: observed in the experiments and simulations of polymers subject to a uniform
182: flow (see \cite{rzehak} and references therein) and the corresponding
183: shape of the partially
184: unwound polymer was called
185: ``stem and a flower''\cite{bro1,bro2} or ''ball and a string"
186: \cite{Haupt,Maurice,homop}.
187: Thus exploration of stationary conformations corresponding to various flow
188: rates should offer a more telling diagnostic of elastic properties of
189: the protein than the
190: force clamp.\\
191:
192:
193: We present the model in Section 2 and consider the case of the helix in
194: Section 3. Stretching in a constant flow is analyzed in Section 4 for
195: various ways of
196: choosing
197: the anchoring point and for tandem arrangement of proteins.
198: In Section 5, we discuss an example of a non-uniform flow: elongational one.
199: Finally, in Section 6, we consider refolding after stopping the flow.\\
200:
201: \section{The model}
202:
203: The effective interactions between the C$^{\alpha}$ amino acids are
204: split into two classes: native and non-native. The distinction
205: is done by checking for native overlaps of all atoms in aminoacids when
206: represented by enlarged van der Waals spheres as proposed in
207: reference \cite{Tsai}. The amino acids, $i$ and $j$ that do overlap in
208: this sense
209: are endowed with the effective Lennard-Jones potential
210: $V_{ij} = 4\epsilon \left[ \left( \frac{\sigma_{ij}}{r_{ij}}
211: \right)^{12}-\left(\frac{\sigma_{ij}}{r_{ij}}\right)^6\right]$.
212: The length parameters $\sigma _{ij}$ are chosen so that the potential
213: minima correspond, pair-by-pair, to the experimentally established
214: native distances between
215: the C$^{\alpha}$ atoms in amino acids in the pair. The repulsive
216: interactions are described
217: by the $r_{ij}^{-12}$ part of the Lennard-Jones potential combined
218: with a constant shift term
219: that makes the potential vanish smoothly at $\sigma =5$ {\AA}.
220: It should be noted that the specificity of a protein is contained in
221: the length parameters
222: $\sigma _{ij}$. The energy parameter, $\epsilon$, is taken to be
223: uniform and its effective
224: value
225: for titin and ubiquitin appears to be of order 900 K so the reduced temperature,
226: $\tilde{T}=k_BT/\epsilon \;$ of 0.3
227: ($k_B$ is the Boltzmann constant and $T$ is temperature)
228: should be close to the room temperature
229: value \cite{Pastore,ubiq}. All of the simulations
230: reported here were performed at this temperature.
231: In our stretching simulations, the anchored terminus of the protein is
232: attached to a harmonic spring of elastic
233: constant $k$=0.06 $\epsilon /${\AA}$^2$.\\
234:
235:
236: As explained in the Introduction,
237: thermostating and mimicking some effects of the solvent
238: are provided by the Langevin dynamics, Eq. 1.
239: The friction coefficient $\gamma$ is taken to be equal to $2m/\tau$
240: where $\tau =\sqrt{m \sigma^2 / \epsilon} \approx 3 ps$
241: is the characteristic time scale of oscillations in the Lennard-Jones well.
242: The selected value of $\gamma$ corresponds to a situation in which the
243: inertial effects are small \cite{Hoang} but the damping action is not
244: yet as strong as in water.\\
245:
246:
247: The equations of motion are solved by a fifth order predictor-corrector scheme.
248: In the course of stretching, the native contacts are being ruptured. A
249: contact between
250: amino acids $i$ and $j$ is said to be ruptured if the corresponding
251: distance $r_{ij}$ becomes
252: larger
253: than $1.5 \sigma _{ij}$ (close to the inflection point of the
254: Lennard-Jones potential) for
255: the last time.
256: When studying folding, we consider establishing contacts
257: starting from an unfolded state. A contact is said to be established
258: when the corresponding value of $r_{ij}$ crosses the threshold value
259: for the first
260: time. Folding is considered to be achieved when {\em all} contacts are
261: established simultaneously.\\
262:
263: When simulating the force clamp, the force $F$ is applied to the spring that
264: pulls one of the termini (the choice of the terminus is irrelevant in
265: this case).
266: In the case of the flow, we discuss the results in terms of the net
267: hydrodynamic force, ${\bf F}=\gamma \sum_{i=1}^{N} {\bf u}({\bf r}_i)$, that is
268: experienced at the anchor point. For uniform flows, $F=N\gamma u$. The
269: dimensionless force, $F\sigma/\epsilon$, will be denoted by $\tilde{F}$.
270: The conformations will be characterized by the end-to-end
271: distance $L$.\\
272:
273: The relative strength of convective and diffusive effects in the dynamics of a
274: protein is given by the Peclet number $$ Pe = \frac{U R_g}{D} $$ where $U$ is
275: the characteristic flow magnitude, $R_g$ - radius of gyration and $D$ - the
276: diffusion coefficient of the protein. Numerically, one may estimate $D$ by the
277: analysis of the mean square displacement of the protein as a function of
278: time. For example, for ubiquitin, the calculations give
279: $D \approx 0.2 {\sigma^2 / \tau}$, whereas $R_g=2.3 \sigma$. The flow rates
280: used in the simulations lie in the range $U=0.02 - 0.07 \sigma / \tau$ ,
281: what gives $\mbox{Pe} \approx 0.2 - 0.7$.\\
282:
283: Since the Peclet number is dimensionless, it can be used to relate the
284: simulation to the experimental setup. Namely, as reported in \cite{diff}, the
285: diffusion coefficient for ubiquitin is $D \approx 1.7 \cdot 10^{-6}$ cm$^2$/s, whereas
286: $R_g=1.15 \cdot 10^{-7} cm$. Thus the above mentioned Peclet number range
287: corresponds to the flow rates of $4-13$ cm/s. Such speeds are about three
288: orders of magnitude faster than those needed to unravel DNA molecules. This is
289: because proteins contain larger clusters of bonds that need to be ruptured
290: simulateneously and are also smaller in size.
291: The above comparison may also be used to relate the numerical time scale $\tau$ to the
292: experimental time scales.
293: Namely, from the fact that $U=0.02 \sigma / \tau$ corresponds to
294: 4 cm/s and $\sigma = 5\; \AA$ one concludes that $\tau$ corresponds to approximately
295: 0.25 ns of the `real' time. This time scale is of the same order of magnitude
296: as the one that Veitshans et al. \cite{veit} arrived at (3 ns) by using an entirely
297: different argument.\\
298:
299:
300: \section{Flow-induced stretching of homopolymers}
301:
302: A synthetic helix provides an example of a homopolymer since none of its
303: parts, except at the termini, is distinguished.
304: The dependence of the end-to-end distance in the stretched helix on the total
305: stretching force is shown in Figure 2. The fractional extension is seen to be a
306: smooth function of the applied force without any stationary or quasistationary
307: stages.
308: The steady-state conformations corresponding to different values of $F$ clearly
309: show the ``stem and a flower'' phenomenon -- the helix unwinds from
310: the fixed end
311: and the unwound length depends on the net hydrodynamic force, i.e. on
312: the flow rate.
313: It is also seen that the dependence of the fractional change in $L$ on $F$
314: does not change with the total number of residues in the helix.
315: This finding is consistent with a similar and well-established result
316: \cite{rzehak,bro1,bro2} pertaining to homopolymers in the free-draining limit.\\
317: The other part of Figure 2 shows the dependence of the mean unfolding
318: time of the helix
319: on the net hydrodynamic force. For the purpose of making this figure,
320: we consider
321: the helix to be unfolded when its total length exceeds 90\% of the
322: maximum extension length
323: of $(N-1) \times 3.8 \AA$. It is seen that in the small-force regime,the
324: unfolding time exponentially decreases
325: as a function of the force. For larger forces the dependence
326: of the unfolding time on the force becomes much weaker. An analogous
327: phenomenon is observed in the
328: simulations of stretching of proteins in the force clamp \cite{cs1}.
329:
330:
331: \section{ Flow-induced stretching of proteins}
332:
333: Figures 3 and 4 show the end-to-end distance versus time for integrin
334: unfolding in a uniform flow. In
335: Figure 3, terminus C is anchored, whereas in Figure 4 it is terminus N
336: that is anchored. Several
337: trajectories corresponding to different values of the total
338: hydrodynamic force are shown. Since the
339: tension is strongest near the anchoring point, the protein unfolds
340: from the fixed end towards the free
341: one. In contrast to homopolymer, here the unfolding pathway traverses
342: through a number of intermediate
343: states, corresponding to the unzipping of subsequent structures from
344: the bulk of the protein. We observe
345: that if the flow rate is sufficiently low, the protein may remain
346: trapped in one of these states for the
347: duration of the simulation. In contrast to the simple helix, complex
348: proteins are cross-linked and
349: inhomogeneous and yield to inhomogeneous tension in a way which is
350: specific to the stretching protocol. Thus the
351: steady-state conformation in which the protein is found after a long
352: time depends not only on the value
353: of the force but also on the choice of the terminus. In particular, as
354: it is seen in the Figures, the set
355: of intermediates is much richer for the case of fixed C terminus than
356: vice versa. Also, in the former
357: case the full unwinding of integrin chain requires a smaller net
358: force. This suggests that the strongest
359: bonds in the native structure of integrin are located nearer to the C
360: terminus.\\
361:
362: The differences between
363: unfolding with different termini fixed are further highlighted by
364: analysis of the so called unfolding scenarios
365: \cite{Hoang}, in which one plots an average time when a given contact
366: is broken against the contact
367: order, i.e. against the sequential distance, $|j - i|$, between the
368: amino acids that form a native
369: contact. Figure 5 compares the unfolding scenarios for different
370: anchorings of the integrin chain. Again,
371: it is seen that anchoring at the C terminus gives rise to a much
372: richer unfolding dynamics, including
373: several intermediates, than the anchoring at the other terminus.
374: Existence of these differences may offer
375: an interesting way of the experimental probing the structure of a
376: protein by analysis of unfolding
377: trajectories with different anchoring points.\\
378:
379:
380: For comparison, Figure 6 shows the unfolding trajectories for the
381: integrin pulled by a force applied at
382: the terminus only, as in the force-clamp apparatus. All the
383: intermediates present here are also seen in
384: uniform flow experiments, particularly those with C terminus fixed
385: ({\it cf.} Figure 3), but vice versa
386: is not true. Thus uniform flow unfolding appears to be richer in intermediate
387: conformations than simple pulling.\\
388:
389: It is instructive to perform a similar analysis for another protein,
390: ubiquitin. Ubiquitin
391: behaves very differently from integrin when stretched in a force-clamp
392: \cite{cs1}, since it
393: unfolds usually in a single kinetic step whereas integrin unfolding
394: involves several
395: intermediates as it is illustrated in Figure 5. However, as it is seen
396: in Figures 7 and 8, in a
397: uniform flow ubiquitin does display several intermediate steps in
398: unfolding, which confirms
399: that the unfolding in a flow shows a richer behavior than stretching
400: in a force clamp.
401: Additionally, one again observes a difference in behavior between $C$
402: and $N$ anchoring. This
403: time, it is $N$ anchoring which shows a larger number of intermediates
404: and requires a smaller
405: force for the full unfolding. Thus, in the case of ubiquitin, the
406: strongest bonds are located
407: in the neighborhood of $N$ terminus.\\
408:
409:
410: Finally, Figure 9 summarizes results on statistically averaged (over
411: 50 unfolding
412: trajectories) flow-induced and force-clamp-induced processes of
413: unfolding in ubiquitin (top
414: panel) and integrin (bottom panel). The figure shows the dependence of
415: the logarithm of the
416: median unfolding time on the total force. Just as in the case of
417: helix, we consider
418: a protein to be unfolded when its total length exceeds 90\% of the
419: maximum extension length
420: of $(N-1) \times 3.8 \AA$. Again, for the uniform flow, one should note
421: the lack of symmetry
422: between the anchoring by the N terminus and by the C terminus.
423: One should also observe that determination of
424: which choice offers more resistance to unravelling is protein dependent.
425: However, force-clamp
426: stretching generally requires a smaller force since in the
427: flow-induced case the segments
428: which are near the free end are exposed to relatively small
429: unravelling tensions and thus they unfold
430: only partially.\\
431:
432:
433:
434: \section{Stretching of polyprotein in a flow}
435:
436:
437: The experiments on protein unfolding in a force clamp are usually performed with
438: polyprotein chains consisting of several repeats of a given protein.
439: For example in the
440: studies of Fernandez group \cite{FernandezLi,Schlierf}, polyubiquitin chains
441: of 2-9 linked ubiquitin domains were investigated. When a constant
442: force is applied
443: to the terminus of such a system, the domains unfold in a staircase-like manner,
444: with each step corresponding to the unwinding of a single domain.
445: Serial unwinding
446: of polyubiquitin is also observed in molecular dynamics simulations \cite{cs1}.
447: It is observed that selection of the domain to be unravelled the first
448: is fluctuations-driven and thus random in nature.\\
449:
450: In a uniform flow, the situation is different, as it is seen in Figure 10
451: which presents the unfolding
452: pathways
453: for two-ubiquitin for various flow rates. First, because of the
454: nonuniform tension along the chain,
455: the unwinding always begins with the domain closest to the the anchoring
456: point. Also, as it is seen in
457: Figure 10, if the flow rate is high enough the unfolding is not serial
458: - one of the ubiquitin domains
459: unfolds together with a considerable piece of the other domain. For smaller flow rates, one
460: of the intermediates corresponds to the situation when one of the domains is
461: fully unfolded while the other is not. However, now it is just one of the many
462: intermediate states of 2-ubiquitin and not the unique intermediate conformation, as it is
463: the case for the force-clamp stretching.\\
464:
465:
466:
467:
468: \section{Non-terminal attachment}
469:
470: Since unfolding in a uniform flow depends considerably on the choice
471: of the anchoring terminus, it is worth exploring other possibilities
472: of anchoring. Figure 11 corresponds to an unfolding trajectory
473: of integrin chain that is tethered at lysine 148.
474: Here we monitor the end-to-end lengths, $L_1$ and $L_2$, of
475: two segments (1-148) and (148-184) respectively, at the
476: net hydrodynamic force of $\tilde{F}=4$.
477: We observe that, initially, both segments get streched side-by-side
478: and at the same rate. However, as
479: discussed in the Introduction, the tension in a longer segment is
480: higher which leads to a rapid rupture of the inter-segmental contacts.
481: From this time on, they evolve independently - the
482: longer chain unwinds quickly whereas the shorter one snaps back and folds
483: into a stationary conformation as shown in Figure 11.\\
484:
485:
486:
487: \section{Elongational flow}
488:
489: Finally, we observe that the abundance of intermediate states
490: seen in a uniform flow is not necessarily present for other kinds of flows.
491: As an illustration, we have carried out
492: simulations of integrin unfolding in an elongational flow described by:
493: \begin{equation}
494: u_x=g (x-x_0), \ \ \ u_y=-\frac{1}{2} g (y-y_0), \ \ \
495: u_z=-\frac{1}{2} g (z-z_0) \;\;\;.
496: \end{equation}
497: Here, $(x_0,y_0,x_0)$ corresponds to a location of the stagnation
498: point for the flow ({\it cf. }Figure 12)
499: and we take it to coincide with the C terminus of a protein.
500: In over 500 unfolding trajectories,
501: obtained for a broad range of the elongational rate, $g$, we have not observed
502: even a single stationary intermediate state, whereas there are at least
503: four intermediate states when flow is uniform and the anchoring is applied
504: to the C terminus. Figure 13 shows an example of the unfolding pathway in
505: this case together with the dependence of the mean unfolding time on $g$.\\
506:
507:
508: The lack of intermediate states in elongational flow, in contrast to
509: the uniform flow, has been already noticed by Lemak et al. \cite{lemak1,lemak2}.
510: However, they attributed the absence to the fact that ``in an
511: elongational flow every
512: monomer experiences a force that is high enough to delocalize it from
513: a bonding site''.
514: In our opinion, the physical mechanism here is, in fact, different.
515: We note that, in contrast to the case of the uniform flow, the total hydrodynamic force
516: acting on a protein chain in an
517: elongational flow depends on the actual total length of the chain. This is
518: so because the farther from $(x_0,y_0,z_0)$ the pulled terminus is, the
519: faster flow it experiences. Thus as the protein unravels even just a bit
520: the total hydrodynamic force also increases correspondingly.
521: This results in a positive feedback mechanism that leads to a rapid
522: rupture until the protein is unravelled fully.\\
523:
524:
525: \section{Refolding after stopping the flow}
526:
527: If the flow is stopped suddenly, the protein chain folds again. The
528: analysis of the folding trajectories
529: shows a considerable number of misfolded stationary conformations
530: arising in these
531: processes (typically about 10-20\% trajectories, depending on the initial
532: extension of the protein). Typically, the misfolded conformations lie
533: relatively close
534: to the native state, with a RMSD of $1-10 \AA$ but the escape time from the
535: misfolded conformation is often longer than the time needed
536: to reach the misfolded state.
537: A typical trajectory with a misfolding event is presented in Figure 14.
538: In this trajectory, the protein gets into a conformation
539: with essentially the same end-to-end length as
540: in the native state without establishing about 10\% of the native contacts.
541: The lower panel of
542: Figure 14 shows the distribution
543: of the folding times for ubiquitin in which the initial state
544: has been obtained by a
545: constant flow unfolding with a total hydrodynamic
546: force of $\tilde{F}=5$. In this case, almost 20\% of all trajectories
547: lead to misfolding. The trajectory in the upper panel of
548: Figure 14 corresponds to a
549: relatively short-lived misfolded state.
550: In most situations corresponding to misfolding,
551: the native conformation is not reached within the duration of the simulation.
552: Such trajectories have not been included in the histogram shown in Figure 14.
553:
554:
555: \section{Summary}
556:
557: In summary, stretching in a uniform flow provides a promising tool for probing
558: the conformational landscape of proteins. In contrast to homopolymers,
559: the steady-state conformations of proteins corresponding to various
560: flow rates form a discrete set. Unfolding usually involves several kinetic
561: transitions between subsequent intermediates and has a richer dynamics than that
562: in the force-clamp case. Moreover, the unfolding pathways depend on the
563: selection of the point of anchor. Thus, making various selections
564: provides additional information about the structure of proteins.
565: Harnessing this information may be facilitated experimentally
566: by attaching the free end of the protein to a fluorescent quantum dot \cite{Weitz}.
567: A similar technique has been succesfully used,
568: e.g., in tracking of the myosin molecular motor \cite{Warshaw}.\\
569:
570:
571:
572: M.C. appreciates stimulating discussions with Olle Inganas.
573: This work was supported by Ministry of Science in Poland
574: (Grant No. 2P03B-03225)
575: and by the European program IP NaPa through Warsaw
576: University of Technology.
577:
578:
579:
580:
581: \begin{thebibliography}{99}
582:
583:
584: \bibitem{kuhn} W. Kuhn, Kolloid Z. {\bf 68}, 2 (1934)
585:
586: \bibitem{kramers} H. A. Kramers HA,
587: %The behavior of macromolecules in inhomogeneous flow,
588: J. Chem. Phys. {\bf 14}, 415-424 (1946).
589:
590: \bibitem{gennes} P. G. de Gennes,
591: %Coil-stretch-transition of dilute flexible polymer under ultra-high
592: %velocity gradients,
593: J. Chem. Phys {\bf 60}, 5030-5042 (1974).
594:
595:
596: \bibitem{smith} D. E. Smith and S. Chu,
597: %Response of Flexible Polymers to a Sudden Elongational Flow,
598: Science {\bf 281}, 1335-38 (1998).
599:
600: \bibitem{perkins1} T. T. Perkins, D. E. Smith, S. Chu,
601: %Single Polymer dynamics in an Elongational Flow,
602: Science {\bf 276},
603: 2016-21 (1997).
604:
605: \bibitem{perkins2} T. T. Perkins, D. E. Smith, R. G. Larson, and S. Chu,
606: %Stretching of a Single Tethered Polymer in a Uniform Flow,
607: Science {\bf 268} 83-87 (1995).
608:
609: \bibitem{rzehak} R. Rzehak, W. Kromen, T. Kawakatsu and W. Zimmermann,
610: %Deformation of a Tethered Polymer in Uniform Flow,
611: European Physical Journal E {\bf 2}, 3-30 (2000).
612:
613: \bibitem{cheon} M. Cheon, I. Chang, J. Koplik, and J. R. Banavar,
614: %Chain molecule deformation in a uniform flow - A computer experiment.
615: Europhys. Lett. {\bf 58} 215-221 (2002).
616:
617: \bibitem{larson} R. G. Larson, T. T. Perkins, D. E. Smith, and S. Chu,
618: %Hydrodynamics of a DNA Molecule in a Flow Field,
619: Phys. Rev. E {\bf 55} 1794-1797 1997.
620:
621: \bibitem{abramchuk} S. S. Abramchuk, A. R. Khokhlov, T. Iwataki, H.
622: Oana and K. Yoshikawa
623: %Direct observation of DNA molecules in a convection flow of a drying droplet
624: Europhys. Lett., {\bf 55}, 294-300 (2001).
625:
626: \bibitem{shaqfeq} E. S. G. Shaqfeq,
627: %The dynamics of single-molecule DNA in flow
628: J. Non-Newtonian Fluid Mech., {\bf 130}, 1-28 (2005).
629:
630: \bibitem{blossey}
631: T. Heim, S. Preuss, B. Gerstmayer, A. Bosio, and R. Blossey,
632: %Deposition from a drop: morphologies of unspecifically bound DNA,
633: J. Phys.: Cond. Matter {\bf 17}, S703-S716 (2005).
634:
635: \bibitem{Strick}
636: T. R. Strick, M.-N. Dessinges, G. Charvin, J. F. Allemand, D.
637: Bensimon, and V. Croquette,
638: %Stretching of macromolecules and proteins.
639: Rep. Prog. Phys. {\bf 66}, 1-45 (2003).
640:
641: \bibitem{Bensimon}
642: X. Michalet, R. Ekong, F. Fougerousse, S. Rousseaux, C. Schurra, N. Hornigold,
643: M. van Slengtenhorst, J. Wolfe, S. Povey, J. S. Beckmann, A. Bensimon,
644: %Dynamic molecular combing: stretching the whole human genome for
645: %high-resolution studies,
646: Science {\bf 277}, 1518-1523 (1997).
647:
648: \bibitem{BensBens}
649: D. Bensimon, A. J. Simon, V. Croquette, and A. Bensimon,
650: %Stretching DNA with a receding meniscus: expewriments and models,
651: Phys. Rev. Lett. {\bf 74}, 4754-4757 (1995).
652:
653: \bibitem{Austin1}
654: O. B. Bakajin, T. A. J. Duke, C. F. Chou, S. S. Chan, R. H. Austin,
655: and E. C. Cox,
656: %Electrohydrodynamic stretching of DNA in confined environments,
657: Phys. Rev. Lett. {\bf 80}, 2737-2740 (1998).
658:
659: \bibitem{Austin}
660: E. Y. Chan, R. Gilmanshin, R. H. Austin, E. D. Carstea, M. Fuchs, L. Gleich,
661: N. M. Gonclaves, R. A. Haeusler, J. W. Larson, A. M. Maletta, M. P. Masera,
662: T. G. Thompson, P. S. Wellma,, G. G. Wong, and G. R. Yantz,
663: %DNA mapping technology based on microfluidic stretching and
664: %single-molecule detection of motif tags,
665: Bioph. J. {\bf 83}, 302A-302A, Part 2 Suppl. (2003).
666:
667: \bibitem{Inganas}
668: P. Bjork, A. Herland, I. G. Scheblykin, and O. Inganas,
669: %Single molecular imaging and spectroscopy of conjugated
670: %polyelectrolytes decorated on
671: %stretched aligned DNA,
672: Nano Lett. {\bf 5}, 1948-1953 (2005).
673:
674: \bibitem{clampober}
675: F. Oberhauser, P. K. Hansma, M. Carrion-Vazquez, and J. M. Fernandez,
676: %Stepwise unfolding of titin under force-clamp atomic force microscopy,
677: {\it Proc. Natl. Acad. Sci. (USA) } {\bf 98}, 468 (2001).
678:
679: \bibitem{lemak1} A. Lemak, J. R. Lepock, and J. Z. Y. Chen,
680: %Molecular dynamics simulations of a protein model in uniform and
681: %elongation flow fields,
682: Proteins: Structure, Function and Genetics {\bf 51}, 224-235 (2003).
683:
684: \bibitem{lemak2} A. Lemak, J. R. Lepock, J. Z. Y. Chen,
685: %Unfolding a protein in external fields:
686: %can we always observe folding immediate states?,
687: Phys. Rev. E {\bf 67}, 031910 (2003).
688:
689:
690: \bibitem{cs1} P. Szymczak and M. Cieplak J. Phys.: Condens. Matter {\bf 18},
691: L21-L28, (2006).
692:
693: \bibitem{Goabe}
694: H. Abe and N. Go,
695: %Noninteracting local-structure model of folding and unfolding transition in
696: %globular proteins. II. Application to two-dimensional lattice proteins.
697: {\it Biopolymers} {\bf 20} 1013 (1981).
698:
699: \bibitem{Hoang}
700: T. X. Hoang and M. Cieplak,
701: %Sequencing of folding events in Go-like proteins,
702: J. Chem. Phys. {\bf 113}, 8319-8328 (2000).
703:
704: \bibitem{biophysical}
705: M. Cieplak and T. X. Hoang,
706: %Universality classes in folding times of proteins.
707: {\it Biophysical J.} {\bf 84} 475 (2003).
708:
709: \bibitem{thermtit}
710: M. Cieplak, T. X. Hoang and M. O. Robbins,
711: %Thermal effects in stretching of Go-like models of titin and
712: %secondary structures.
713: {\it Proteins: Struct. Funct. Bio.} {\bf 56} 285 (2004).
714:
715: \bibitem{Pastore}
716: M. Cieplak, A. Pastore and T. X. Hoang,
717: %Mechanical properties of the domains of titin in a Go-like model.
718: {\it J. Chem. Phys.} {\bf 122} 054906 (2004).
719:
720: \bibitem{ubiq}
721: M. Cieplak and P. E. Marszalek,
722: %Mechanical unfolding of ubiquitin molecules,
723: J. Chem. Phys. {\bf 123}, 194903 (2005).
724:
725: \bibitem{knudsenprl}
726: M. Cieplak, J. Koplik, J. R. Banavar,
727: %Boundary conditions at a fluid-solid interface,
728: {\it Phys. Rev. Lett.} {\bf 86}, 803 (2001).
729:
730: \bibitem{lotus}
731: M. Cieplak, J. Koplik, J. R. Banavar,
732: %Nanoscale fluid flows in the vicinity of patterned surfaces,
733: Phys. Rev. Lett. {\bf 96} 114502 (2006).
734:
735: \bibitem{Tanaka} H. Tanaka
736: %Roles of hydrodynamic interactions in structure formation of soft matter: protein folding
737: %as an example
738: {J. Phys. Cond. Matt.}, {bf 17} S2795-S2803 (2005).
739:
740: \bibitem{hsieh} C-C. Hsieh, L. Li, R.G. Larson,
741: %Modeling hydrodynamic interaction in Brownian dynamics:
742: %simulations of extensional flows of dilute solutions of
743: %DNA and polystyrene
744: J. Non-Newtonian Fluid Mech., {\bf 113}, 147-191 (2003).
745:
746:
747:
748: \bibitem{bro1} F. Brochard-Wyart,
749: %Deformations of one tethered chain in strong flows,
750: Europhys. Lett. {\bf 23}, 105-111 (1993).
751:
752: \bibitem{bro2} F. Brochard-Wyart,
753: %Polymer-chains under strong flows -- stems and flowers,
754: Europhys. Lett. {\bf 30}, 387-392 (1995).
755:
756: \bibitem{Haupt}
757: B. J. Haupt, T. J. Senden, and A. M. Sevick,
758: %AFM evidence of Rayleigh instability in single polymer chains,
759: Langmuir {\bf 18} 2174-2182 (2002).
760:
761: \bibitem{Maurice}
762: R. G. Maurice and C. C. Matthai,
763: %Force-extension curves for a single polymer chain under varying
764: %solvent conditions,
765: Phys. Rev. E {\bf 60} 3165-3169 (1999).
766:
767: \bibitem{homop}
768: M. Cieplak, T. X. Hoang, and M. O Robbins,
769: %Stretching of homopolymers and contact order,
770: Phys. Rev. E {\bf 70} 011917 (2004)
771:
772: \bibitem{Tsai}
773: J. Tsai, R. Taylor, C. Chothia, and M. Gerstein,
774: %The packing density in proteins: Standard radii and volumes.
775: {\it J. Mol. Biol.} {\bf 290} 253 (1999).
776:
777: \bibitem{diff}
778: A. J. Dingley, J. P. Mackay, G. L. Shaw, B. D. Hambly and G. F. King
779: %Measuring macromolecular diffusion using heteronuclear
780: %multiple-quantum pulsed-field-gradient NMR
781: {\it J. Bio. NMR}, {\bf 10} 1 (1997).
782:
783: \bibitem{veit} T. Veitshans, D. Klimov, and D. Thirumalai.
784: % Protein folding kinetics:Timescales,
785: %pathways and energy landscapes
786: %in terms of sequence-dependent properties.
787: {\it Folding and Design}, {\bf 2} 1 (1997).
788:
789: \bibitem{FernandezLi}
790: J. M. Fernandez and H. Li,
791: %Force-Clamp Spectroscopy Monitors the Folding Trajectory of a Single Protein,
792: {\it Science} {\bf 303} 1674 (2004).
793:
794: \bibitem{Schlierf}
795: M. Schlierf, H. Li and J. M. Fernandez
796: %The unfolding kinetics of ubiquitin captured with
797: %single-molecule force-clamp techniques
798: {\it Proc. Natl. Acad. Sci. (USA)} {\bf 101}, 7299, (2004)
799:
800: \bibitem{Weitz}
801: A.R. Bausch and D.A. Weitz,
802: %Tracking the dynamics of single quantum dots:
803: %Beating the optical resolution twice
804: {\it J. Nanoparticle Res.}, {\bf 4}, 477, (2002).
805:
806: \bibitem{Warshaw}
807: D. M. Warshaw, G. G. Kennedy, S. S. Work, E. B. Krementsova, S. Beck,
808: and K. M. Trybus,
809: %Differential Labeling of Myosin V Heads with Quantum Dots Allows
810: %Direct Visualization of Hand-Over-Hand Processivity
811: {\it Biophys J.} {\bf 88}, L30, (2005).
812:
813:
814:
815: \end{thebibliography}
816:
817:
818: \newpage
819: \centerline{FIGURE CAPTIONS}
820:
821: \begin{description}
822:
823: \item[Fig. 1. ]
824: Schematic representation of stretching a polymer by fluid flow. The bead
825: denoted by 1 is free whereas the bead denoted by $N$ is attached to a
826: spring. The other
827: end of the spring is fixed. A uniform flow is directed from the right
828: to the left.
829: The tension along the chain increases linearly from the free end towards the
830: anchored end.
831:
832:
833: \item[Fig. 2. ]
834: The top-left panel shows the dependence of the fractional extension of
835: the helix in the steady state on
836: the total
837: hydrodynamic force for synthetic helices with $N=48$ (asterisks) and
838: $N=24$ residues (squares). The snapshots at the bottom show examples of the
839: stationary states for forces indicated. The top-right panel plots the
840: logarithm of
841: the median unfolding time against the total hydrodynamic force for the
842: $N$=24 helix.
843: For the purpose of making this figure, we consider
844: a helix to be unfolded when its total length exceeds 90\% of the
845: maximum extension length
846: of $(N-1) \times 3.8 \AA$.
847:
848: \item[Fig. 3. ]
849: Examples of the time evolution of the end-to-end distance in unfolding of
850: integrin in a flow for the total hydrodynamic forces as indicated.
851: The C-terminal is fixed and the conformations corresponding to the
852: plateau regimes of the unfolding pathways are shown on the right.
853: %For $\tilde{F}$=5, the protein fully unfolds.
854:
855: \item[Fig. 4. ]
856: Similar to Figure 3 but for the situation in which the N terminus is fixed.
857:
858: \item[Fig. 5. ]
859: The scenarios of unfolding of integrin in a uniform flow.
860: The top and bottom panels corresponds to anchoring of the C and N termini
861: respectively. The values of the total stretching forces are indicated.
862:
863:
864: \item[Fig. 6. ]
865: Similar to Figure 3 but for unfolding induced by applying a constant
866: force, as indicated, in a force clamp.
867:
868:
869: \item[Fig. 7. ]
870: Similar to Figure 3 but for ubiquitin.
871:
872: \item[Fig. 8. ]
873: Similar to Figure 4 but for ubiquitin.
874:
875: \item[Fig. 9. ]
876: The dependence of the logarithm of the median unfolding time on the force.
877: The top panel is for ubiquitin and the bottom panel for integrin.
878: The solid data points and solid lines (the latter are guides to the eye)
879: correspond to unfolding in a flow. The choice of the anchored
880: terminus is indicated next to the lines.
881: The open symbols and dotted lines correspond to stretching in a force clamp.
882:
883: \item[Fig. 10.]
884: Unfolding pathways of two-ubiquitin for various total forces as indicated.
885: An unfolding trajectory for
886: two-ubiquitin in a force-clamp (the dashed line) is given for comparison.
887: The full extension in force clamp is longer than in the flow
888: induced case since in the latter case the segments which are most distant from the
889: anchor experience too small force to unravel.
890: The inset shows the scenarios of unravelling events. The contacts
891: in the domain that is closer to the anchor are marked by black
892: squares and those in the more distant domain by open squares.
893: The snapshots on the right correspond to the stationary states
894: at those values of $L$ (approximately) at which the snapshots are plotted.
895: The values of dimensionless forces used in the simulations ($\tilde{F}$) are indicated.
896:
897: \item[Fig. 11.]
898: The unfolding pathway of integrin anchored at $Lys 148$ in a uniform
899: flow. The end-to-end
900: length of the segments (1-148) and (148-184) is plotted as a function
901: of time, with the corresponding
902: protein conformations shown.
903: The inset shows a protein conformation just before the contacts between the segments are
904: broken.
905:
906: \item[Fig. 12. ]
907: A schemtatic view of elongational flow with a stagnation point $(x_0,y_0)$ in the center of
908: the graph
909:
910:
911:
912: \item[Fig. 13. ]
913: Integrin: the unfolding pathway in an elongational flow and the median
914: unfolding time as a function of
915: elongational rate $\tilde{g}$ defined as $\tilde{g}=N \gamma g L_m$,
916: where $L_m$ is the maximum extension
917: length of a protein, $L_m=(N-1) \times 3.8 \AA$. Note that below $g \approx 1.6$
918: the unfolding process is essentially arrested.
919:
920: \item[Fig. 14. ]
921: The top panel shows a typical folding trajectory with a misfold for integrin.
922: The initial state was obtained by a constant flow unfolding with a
923: total hydrodynamic
924: force of $\tilde{F}=5$. The time is measured from an instant at which
925: the force is
926: suddenly reduced to zero (i.e. the flow is stopped). Throughout the process, the
927: C terminus is held anchored. The graph shows both the end-to-end
928: distance ($L$) and the
929: fraction of native contacts ($Q$).
930: Note that the final escape from the misfolded state is only seen in
931: the $Q(t)$ graph.
932: The bottom panel shows the distribution of folding times in this case.
933: The fit is to a log-normal distribution
934: $\frac{1}{\sqrt{2\pi}\sigma (t-t_0)}
935: \exp{(-\frac{ln^2(\frac{t-t_0}{m})}{2\sigma ^2})}$.
936: with $t_0 =7675 \tau$, $\sigma = 0.4$, and $m = 3065 \tau$.
937: The histogram is based on 300 trajectories all starting from the same stretched
938: conformation. Trajectories with long-lasting misfolded
939: conformations (corresponding to the unfolding times longer than $18000
940: \tau$) have not been taken into account
941: in the histogram. They make up about 20\% of all trajectories.
942: \end{description}
943:
944:
945:
946: \clearpage
947:
948: %FIGURE 1
949: \begin{figure}
950: \epsfxsize=4.6in
951: \centerline{\epsffile{521639JCP01.eps}}
952: %\vspace*{-1.5cm}
953: \caption{ }
954: \end{figure}
955:
956: \vspace*{-4cm}
957:
958: \clearpage
959:
960: \newpage
961:
962: %FIGURE 2
963: \begin{figure}
964: \epsfxsize=4.6in
965: %\centerline{\epsffile{alfig2.ps}}
966: \centerline{\epsffile{521639JCP02.eps}}
967: \vspace*{-2.5cm}
968: \caption{ }
969: \end{figure}
970:
971: \clearpage
972:
973: \newpage
974:
975: %FIGURE 3
976: \begin{figure}
977: \epsfxsize=3.8in
978: %\centerline{\epsffile{alfig3.ps}}
979: \centerline{\epsffile{521639JCP03.eps}}
980: %\vspace*{-2.0cm}
981: \caption{ }
982: \end{figure}
983:
984: \vspace*{-4.5cm}
985:
986: \clearpage
987:
988: \newpage
989:
990: %FIGURE 4
991: \begin{figure}
992: \epsfxsize=3.8in
993: %\centerline{\epsffile{alfig5.ps}}
994: \centerline{\epsffile{521639JCP04.eps}}
995: %\vspace*{-2.5cm}
996: \caption{ }
997: \end{figure}
998:
999: \clearpage
1000:
1001: \newpage
1002:
1003: %FIGURE 5
1004: \begin{figure}
1005: \epsfxsize=3.8in
1006: \centerline{\epsffile{521639JCP05.eps}}
1007: \caption{ }
1008: \end{figure}
1009: \vspace{2.5cm}
1010: \clearpage
1011:
1012: \newpage
1013:
1014: %FIGURE 6
1015: \begin{figure}
1016: \epsfxsize=3.8in
1017: %\centerline{\epsffile{alfig4.ps}}
1018: \centerline{\epsffile{521639JCP06.eps}}
1019: %\vspace*{-2.5cm}
1020: \caption{ }
1021: \end{figure}
1022:
1023: \clearpage
1024:
1025: \newpage
1026:
1027: %FIGURE 7
1028: \begin{figure}
1029: \epsfxsize=3.8in
1030: %\centerline{\epsffile{alfig10.ps}}
1031: \centerline{\epsffile{521639JCP07.eps}}
1032: %\vspace*{-2.5cm}
1033: \caption{ }
1034: \end{figure}
1035:
1036: \clearpage
1037:
1038: \newpage
1039:
1040: %FIGURE 8
1041: \begin{figure}
1042: \epsfxsize=3.8in
1043: %\centerline{\epsffile{alfig9.ps}}
1044: \centerline{\epsffile{521639JCP08.eps}}
1045: %\vspace*{-2.0cm}
1046: \caption{ }
1047: \end{figure}
1048:
1049: \clearpage
1050:
1051: \newpage
1052:
1053: %FIGURE 9
1054: \begin{figure}
1055: \epsfxsize=3.8in
1056: \centerline{\epsffile{521639JCP09.eps}}
1057: \vspace*{-0.2cm}
1058: \caption{ }
1059: \end{figure}
1060:
1061: \clearpage
1062:
1063: \newpage
1064:
1065: %FIGURE 10
1066: \begin{figure}
1067: \epsfxsize=3.8in
1068: %\centerline{\epsffile{alten.ps}}
1069: \centerline{\epsffile{521639JCP10.eps}}
1070: %\vspace*{-2.5cm}
1071: \caption{ }
1072: \end{figure}
1073:
1074: \clearpage
1075:
1076: %\newpage
1077:
1078: %FIGURE 11
1079: \begin{figure}
1080: \epsfxsize=3.8in
1081: %\centerline{\epsffile{belelev.ps}}
1082: \centerline{\epsffile{521639JCP11.eps}}
1083: \vspace*{-2cm}
1084: \caption{ }
1085: \end{figure}
1086:
1087: \clearpage
1088:
1089: \newpage
1090:
1091: %FIGURE 14
1092: \begin{figure}
1093: \epsfxsize=2.8in
1094: %\centerline{\epsffile{flow.eps}}
1095: \centerline{\epsffile{521639JCP12.eps}}
1096: %\vspace*{3cm}
1097: \caption{ }
1098: \end{figure}
1099:
1100: \clearpage
1101: \newpage
1102:
1103: %FIGURE 13
1104: \begin{figure}
1105: \epsfxsize=3.8in
1106: %\centerline{\epsffile{gig12.eps}}
1107: \centerline{\epsffile{521639JCP13.eps}}
1108: %\vspace*{3cm}
1109: \caption{ }
1110: \end{figure}
1111:
1112: \clearpage
1113:
1114: \newpage
1115:
1116: %FIGURE 14
1117: \begin{figure}
1118: \epsfxsize=3.9in
1119: %\centerline{\epsffile{gig14.eps}}
1120: \centerline{\epsffile{521639JCP14.eps}}
1121: %\vspace*{3cm}
1122: \caption{ }
1123: \end{figure}
1124:
1125: \clearpage
1126:
1127:
1128: \end{document}
1129:
1130: