1: %\documentclass[aps]{revtex}
2: %\documentstyle[aps]{revtex}
3: %\documentstyle[12pt]{article}
4: %\documentclass[11pt,a4paper]{article}
5: %\documentstyle[aps]{revtex}
6: %\documentstyle[aps,preprirnt]{revtex}
7:
8:
9:
10: \documentclass[prb,aps,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
11: \usepackage{amsmath,amssymb,array}
12:
13: %\textwidth 16 true cm
14: %\textheight 25 true cm
15: %\baselineskip=15pt
16: %\textwidth 16 true cm
17: %\textheight 22 true cm
18: %\hoffset=-10 mm
19: %\voffset=10 mm
20: %\parindent 10pt
21: %---------------------------------------------------------------------------
22: %\renewcommand{\baselinestretch}{1.0}
23: %\renewcommand{\theequation}{\arabic{equation}}
24:
25: \begin{document}
26:
27:
28: \def\brho{{\mbox{\boldmath $\rho $}}}
29: \def\bomega{{\mbox{\boldmath $\Omega $}}}
30: \def\bomicron{{\mbox{\boldmath $\omega $}}}
31: \def\ul#1#2{\textstyle{\frac#1#2}}
32: %----------------------------------------------------------------------------
33: %----------------------------------------------------------------------------
34:
35: \title{Density waves theory of the capsid structure of small icosahedral viruses}
36:
37: \author{V.L. Lorman$^\ddagger$ and S.B. Rochal$^{\dagger,\ddagger}$}
38:
39: \noindent
40:
41: \affiliation{$^\ddagger$Laboratoire de Physique Theorique et Astroparticules, CNRS - Universite
42: Montpellier 2, Place Eugene Bataillon, 34095 Montpellier, France \\
43: $^\dagger$Physical Faculty, Rostov State University, 5 Zorge Str.,
44: 344090 Rostov-on-Don, Russia}
45:
46:
47: %\draft
48:
49: %\twocolumn[\hsize\textwidth\columnwidth\hsize\csname@twocolumnfalse\endcsname
50: \noindent
51:
52: \begin{abstract}
53: \small
54: %-------------------------------------------------------------------------
55: We apply Landau theory of crystallization to explain and to
56: classify the capsid structures of small viruses with spherical
57: topology and icosahedral symmetry. We develop an explicit method
58: which predicts the positions of centers of mass for the proteins
59: constituting viral capsid shell. Corresponding density
60: distribution function which generates the positions has universal
61: form without any fitting parameter. The theory describes in a
62: uniform way both the structures satisfying the
63: well-known Caspar and Klug geometrical model for capsid
64: construction and those violating it. The quasiequivalence of
65: protein environments in viral capsid and peculiarities of the
66: assembly thermodynamics are also discussed.
67: %-------------------------------------------------------------------------
68: \end{abstract}
69: %-------------------------------------------------------------------------
70: \pacs{64.70.Dv, 64.70.Nd, 87.68.+z, 61.44.Br}
71: \maketitle
72: %====================================
73:
74:
75: Viruses represent rather simple biologocal systems which can be
76: studied by different chemical and physical methods. Their
77: organization and functionning show a number of universal features.
78: The viral protein shell (capsid) encloses the genetic material (either
79: desoxyribonucleic acid (DNA) or ribonucleic acid (RNA) [1])
80: responsible for the infective properties of the viruses. The capsid
81: serves both to preserve and to transmit the genetic material to an
82: appropriate host cell. Soon after the transmission the host cell
83: starts the reproduction of the viral DNA (or RNA) and capsid
84: proteins. From shell proteins and replicated genomes, new
85: identical copies of the viruses spontaneously assemble.
86: Though the final infective virus structure formation involves
87: biologically specific events, some steps of the self-assembly demonstrate
88: properties typical for ordering in passive physical systems.
89: The host cell is not necessary for the viral capsid formation.
90: The self-assembly does not need active local energy consumption
91: like ATP hydrolisis and the process can be reversible [2,3].
92: Moreover, in many cases the viral shells assembling does
93: not even need genomes and proceeds in vitro in purified protein solutions [1].
94:
95: The problem of the capsid structure formation attracts the
96: attention of physicists since fifty years. In their pioneer work
97: Crick and Watson [4] stated that spherical viruses should have the
98: symmetry (but not necessarily the structure) of one of regular
99: polyhedra with the faces formed by identical perfect polygons.
100: Later in 1962, Caspar and Klug (CK) argued that spherical capsids
101: adopt icosahedral point symmetry [5]. They have seen the physical
102: reason why the Nature prefers this type of symmetry in the fact
103: that the icosahedron has the largest volume-to-surface ratio among
104: the regular polyhedra. Besides, CK obtained four prominent results
105: [5]: i) The capsid symmetry is lower than that of the regular
106: icosahedron since the proteins are asymmetric. Identical asymmetric
107: building blocks can compose the structures with rotational symmetry
108: elements only, excluding inversion and mirror planes. ii) The
109: asymmetric proteins can be located only in regular (trivial)
110: 60-fold positions of the rotational icosahedral point group,
111: therefore the total number of proteins in a capsid is always equal
112: to 60N, where N is a positive integer number. iii) CK concluded,
113: for the first time, that 'the self-assembly is a process akin to
114: crystallisation and is governed by the laws of statistical
115: mechanics'. iv) They proposed a geometrical model for the viral
116: capsid construction based on the properties of the almost regular
117: mapping of the 2D hexagonal structure on the icosahedron surface.
118: Specific properties of the model impose the selection rules for the
119: value of N (and, consequently, for the total number of proteins in
120: the shell). Only the values which satisfy the relation $N= h^2+
121: k^2+hk$, where h and k are non-negative integers are allowed by the
122: CK selection rules. All four points and their direct consequences
123: resulted in the principle formulated by CK and
124: put in the basis of modern virology. Though a big number of virus
125: capsid structures are in a good agreement with all the points of
126: the CK scheme, there is a growing number of experimentally resolved
127: structures which do not satisfy the CK selection rules nor their
128: predictions about local proteins arrangment [6]. These facts show that point iv) of the
129: principle is not universal and needs to be generalized.
130:
131: In recent years the investigation of capsid structures has undergone a real
132: burst due to the progress of the X-ray and cryoelectron microscopy
133: techniques and micromechanical experiments [7].
134: From the theoretical point of view the main
135: effort was done in two directions (see [8]). On the one hand, the
136: mean-field studies of simple model systems were performed
137: in order to approach the thermodynamics of the self-assembly process.
138: On the other hand, the mechanical properties of capsids and their
139: relation to the capsid shape were investigated. Along the first line,
140: the free energy of the viral structure has been approximated by that
141: of a model system consisting of two types of disks located on the
142: spherical surface [9]. The proposed pair potential of the disk
143: interaction favors the icosahedral symmetry of the disk packing [9]
144: provided an optimisation of several model parameters. Along the second line,
145: the possible buckling instability of the spherical capsid structure
146: was studied in the frame of the nonlinear physics of thin elastic
147: shells [10]. The results of this study explain why
148: the relatively small viruses are always spherical while the larger
149: ones have a more angular or faceted shape. In addition, for large
150: viruses the use of continuum elasticity approximation can be
151: justified. This makes the predictions of the mechanical properties [11] of
152: viral capsids and their large-scale shape details [12] more universal.
153: Nevertheless, the results obtained depend crucially on the
154: model assumptions concerning the explicit form of interaction
155: between proteins or groups of proteins (capsomers). Let us also
156: note that all recent theoretical works on the capsid structure do not
157: take into account the asymmetry of capsid proteins nor the restrictions
158: on the capsid symmetry formulated in points i) and ii) of the CK
159: principle. By contrast, the nonuniversal CK selection rules (point iv))
160: are taken as an ingredient in all models.
161:
162: In the present work we propose to apply the Landau theory of
163: crystallization to the problem of small capsid formation. Resulting
164: approach to the icosahedral virus structure accounts explicitly for
165: the protein symmetry and satisfies points i)-iii) of the CK principle but it is free of
166: nonuniversal CK selection rules. It allows us to describe in a
167: uniform way all experimentally observed small spherical viruses
168: including those which can not be obtained using the CK geometrical
169: model (e.g. L-A virus, Dengue virus, West Nile virus, Murine
170: Polyoma virus, etc.)
171:
172:
173: Both the experimental data and the theoretical consideration [10]
174: show that the shape of small viruses with the icosahedral symmetry
175: is close to the spherical one. This fact gives the possibility to
176: consider the crystallization on a spherical surface and to avoid
177: the problems arising in the CK geometrical construction during the
178: mapping of planar hexagonal structures upon the icosahedron
179: surface. Like in the case of usual 3D crystal solidification [13] Landau
180: theory of the assembly process gives simple and clear predictions
181: in the vicinity of crystallization point. In this region the
182: probability density $\rho$ of protein distribution in the capsid
183: structure is presented as:
184: \begin{equation}
185: \rho= \rho_0+\Delta \rho,
186: \end{equation}
187: where $\rho_0$ is an isotropic density in the solution and $\Delta
188: \rho$ corresponds to the density deviation induced by the ordering.
189: The symmetry breaking during the crystallisation is associated with
190: one critical order parameter which spans an irreducible
191: representation of the symmetry group of the disordered state. In
192: addition, in the vicinity of crystallization point, the structure
193: of the ordered state (defined by $\Delta \rho$) is determined by
194: the critical order parameter only, the contribution of non-critical
195: degrees of freedom being negligible in this region. For the
196: crystallization process the order parameter represents a critical
197: system of density waves (CSDW) with the wave vectors of the
198: \textit{same length} and the transition free energy is an invariant
199: function of the CSDW amplitudes [13]. The symmetry of crystals
200: which condense from the isotropic state coincides exactly with that
201: of the corresponding CSDW. For crystals of metals (and especially
202: for the crystals of elements) the atomic positions in the vicinity
203: of crystallization point can be then associated with the positions
204: of maxima of the CSDW.
205:
206: The same
207: principles are applied here to the assembly process on a sphere. The
208: critical part $\Delta \rho_l$ of the density is determined by a
209: CSDW with the \textit{same wave number} $l$. The spherical harmonics
210: $Y_{lm}$ constituting CSDW on a sphere span one irreducible
211: representation (IR) of the SO(3) symmetry group of the disordered
212: state, thus $\Delta \rho_l$ is given by:
213:
214: \begin{equation}
215: \Delta \rho_l(\theta, \phi)= \sum_ {m=-l}^{m=l} A_{lm} Y_{lm}(\theta, \phi),
216: \end{equation}
217: where $l$ is the IR number, $A_{lm}$ are the amplitudes of the spherical
218: harmonics $Y_{lm}$ and $\theta$ and $\phi$ are the conventional angular
219: variables of the spherical coordinate system.
220:
221: According to points i) and ii) of the CK principle the ordered
222: distribution of proteins in the viral capsid has the symmetry group
223: $I$ of the icosahedron rotations which does not contain spatial
224: inversion nor mirror planes. This restriction is of major
225: importance in the proposed theory. It selects the parity of the
226: 'active' IR's of the SO(3) symmetry group which induce the assembly
227: of icosahedral shells of asymmetric proteins. Thus the spherical
228: harmonics $Y_{lm}$ with \textit{even} $l$ numbers cannot form critical
229: density (2) for viral capsids. The restriction affects also the free
230: energy expansion of the assembly process taken in a standard for
231: the crystallization theory form [13] $F=F_0+F_2+F_3+F_4+...$ and
232: containing invariant terms
233: \begin{eqnarray}
234: F_2&=&A(T,c)\sum_{m=-l}^{m=l}A_{l,m}A_{l,-m},\nonumber \\
235: F_3&=&B(T,c)\sum_{m_l,m_2,m_3}a_{m_l,m_2,m_3}
236: A_{l,m_1}A_{l,m_2}A_{l,m_3}\delta(m_l+m_2+m_3)\equiv 0,\\
237: F_4&=&\sum_{k}C_k(T,c)\sum_{m_l,m_2,m_3,m_4}a_{m_l,m_2,m_3,m_4}^k
238: A_{l,m_1}A_{l,m_2}A_{l,m_3}A_{l,m_4}\delta(m_l+m_2+m_3+m_4),
239: \nonumber
240: \end{eqnarray}
241: where $a_i$ are weight coefficients of the SO(3)
242: group (e.g. Clebsch-Gordan coefficients for the third order term $F_3$),
243: $\delta(0)=1$, $\delta(i\neq 0)=0$, $A(T,c)$, $B(T,c)$, and $C^k(T,c)$ are temperature- and
244: composition-dependent coefficients of the Landau theory. For any
245: \textit{odd} wave number $l$ the third-order term $F_3$
246: is identically zero. This fact makes the thermodynamics of
247: asymmetric proteins assembly quite different with respect to the
248: thermodynamics of 3D icosahedral atomic clusters formation [14]
249: in spite of several common points in formal description.
250:
251: Next restriction on the choice of order parameters of the capsid
252: formation comes from the fact that $\Delta \rho_l$ function with $I$
253: symmetry can be constructed not for all but for particular odd $l$ numbers only.
254: The analysis based upon the theory of invariants shows that
255: any critical order parameter which drives the icosahedral assembly
256: of asymmetric proteins has the wave number $l$ satisfying the
257: relation:
258: \begin{equation}
259: l=15+6i+10j,
260: \end{equation}
261: where $i$ and $j$ are positive integers or zero. Eq. (4) defines
262: the list of $l$ numbers for which the restriction of an IR of the
263: SO(3) group on the icosahedral group $I$ contains at least one
264: totally symmetric representation. The sequence of the permitted
265: values of the wave number $l$ is given by: $l=(15, 21, 25, 27, 31,
266: 33, 35...)$. As we show below this sequence determines
267: possible capsid shell structures for small icosahedral viruses.
268: Selection rule (4) gives the possibility to obtain the explicit
269: form of critical density (2). Then the protein centers are associated with the
270: positions of maxima of $\Delta\rho_l$ function (2). Thus the
271: density wave approach replaces nonuniversal geometrical model iv)
272: of the CK principle.
273:
274: The explicit form of the critical density function $\Delta
275: \rho_l(\theta, \phi)$ is given by the basis
276: functions $f_l^i (\theta, \phi)$ ($i=1,2...n_t$) of all $n_t$
277: totally symmetric representations of the icosahedral group $I$
278: in the restriction of the 'active' IR of the SO(3). The CSDW
279: is a linear combination of these functions invariant with respect to the $I$ group:
280: \begin{equation}
281: \Delta \rho_l (\theta, \phi)= \sum_ {i=1}^{n_t} B_i f_l^i (\theta,
282: \phi),
283: \end{equation}
284: where $B_i$ are arbitrary coefficients.
285:
286: Their number
287: $n_t$ is equal to the number of integer non-negative solutions
288: $(i,j)$ of Eq. (4) for a fixed permitted value of $l$. Another way to calculate $n_t$ is to use the
289: well-known relations of characters [15]:
290: \begin{equation}
291: n_t=1/|G|\sum_G\xi(\hat{g})
292: \end{equation}
293: where the sum runs over the elements $\hat{g}$ of the $I$ group,
294: $|G|=60$ is the $I$ group order, and $\xi(\hat{g})$ is the
295: character of the SO(3) group element which reads as [15]:
296: $$
297: \xi(l,\alpha)=\frac{sin((l+1/2)\alpha)}{sin (\alpha/2)},
298: $$
299: where $l$ is the IR number and the angle $\alpha$ is determined by the element $\hat{g}$.
300: Then the explicit form of (6) becomes:
301: \begin{equation}
302: n_t(l)=\frac{1}{60}(2l+1 +15\xi(l,\pi) +20\xi(l,2\pi/3) +12\xi(l,2\pi/5)
303: +12\xi(l,4\pi/5)).
304: \end{equation}
305:
306: For small icosahedral capsids the practical construction of the
307: protein density distribution is simplified because the CSDW (5)
308: contains only one function $f_l(\theta,\phi)$. Indeed, according to
309: Eq. (4) and/or Eq. (7) $n_t=1$ for all $l\leq 43$. In this simplest case
310: $\Delta\rho_l(\theta,\phi)=B f_l(\theta, \phi)$, where $B$ is a
311: single arbitrary coefficient. The positions of maxima of the
312: density function do not depend on the value of $B$. They are
313: generated by a single universal function $f_l(\theta, \phi)$ which
314: has no any fitting parameter. In the following consideration the
315: functions $f_l(\theta, \phi)$ possessing this properties are called
316: irreducible icosahedral density functions and the structures
317: generated by $f_l(\theta, \phi)$ are mentioned as irreducible
318: icosahedral structures. The explicit form of the irreducible
319: density function $f_l(\theta, \phi)$ for a given value of $l$ is
320: obtained by averaging of $Y_{lm}(\theta, \phi)$ harmonics over the
321: $I$ symmetry group [16].
322: \begin{equation}
323: f_l (\theta, \phi) =\frac{1}{60}\sum_{G} Y_{l,m}(\hat{g}(\theta,
324: \phi)).
325: \end{equation}
326: For any fixed value of $m$, procedure (8) gives either the same
327: function $f_l (\theta, \phi)$ we are looking for, or zero.
328: Functions which differ by a constant complex multiplier are
329: considered the same.
330:
331: Fig. 1 resumes the irreducible density functions $f_l(\theta,
332: \phi)$ permitted by selection rule (4) for the five smallest
333: icosahedral capsids (Fig. 1(a-e)); the function $f_{37}(\theta,
334: \phi)$ (Fig. 1(f)) is added as an example illustrating protein density
335: distribution with higher $l$. The value of $f_l(\theta,\phi)$ is
336: represented using false color image: variation of colors from red
337: to violet corresponds to the function growth. Note that all
338: $f_l(\theta,\phi)$ functions are anti-symmetric : they change their
339: sign under the inversion of all coordinates or under the action of
340: mirror planes of a regular icosahedron. Thus, for the sake of
341: clariry, we present the positive part $f_l(\theta,\phi)>0$ only.
342: The number of maxima of the density functions is equal to $60N$,
343: where $N$ is the number of different regular 60-fold positions of
344: the $I$ group. In the viral capsid $N$ corresponds to the number of
345: different positions occupied by the proteins. Let us stress that in
346: a sharp contrast with the CK geometrical model the crystallization
347: theory predicts the existence of capsids with all positive integer
348: values of $N$ and not only for $N=h^2+hk+k^2$. Functions
349: $f_l(\theta,\phi)$ generate in a uniform way protein distributions
350: which can be obtained by the CK mapping of the hexagonal lattice on
351: an icosahedron and those which can not. On the one hand, the
352: distributions in Fig. 1(a) ($l=15$, $N=1$), Fig. 1(d) ($l=27$,
353: $N=3$), and Fig. 1(e) ($l=31$, $N=4$) give classical CK structures.
354: The positions of protein centers in a big number of viral capsids
355: are described by these structures. Fig. 2 shows the correspondence
356: between the maxima of $f_{15}$, $f_{27}$, and $f_{31}$ and the
357: protein arrangement in Satellite Tobacco Mosaic virus (Fig. 2(a)),
358: Cowpea Chlorotic Mottle virus (Fig. 2(b)) and Sindbis virus (Fig.
359: 2(c)), respectively. On the other hand, the distributions in Fig. 1(b) ($l=21$,
360: $N=2$) and in Fig. 1(f) ($l=37$, $N=6$) do not satisfy the CK
361: selection rules for $N$ number. The distribution in Fig. 1(c)
362: ($l=25$, $N=3$) shows no hexagonal arrangements of protein
363: positions and can not be obtained by the CK geometrical model,
364: though the number of protein positions $N$ satisfies the CK
365: selection rules. In addition, the comparison of distributions in
366: Fig. 1(c) and Fig. 1(d) illustrates another striking result of the
367: crystallization theory : there exist qualitatively different capsid
368: structures (induced by $f_l(\theta,\phi)$ functions with different
369: $l$) but with the same number $N$ of protein positions.
370:
371: The X-ray and cryomicroscopy data show the existence of a whole
372: series of viral capsids which violate the CK geometrical model but
373: correspond to the distributions generated by density functions
374: $f_l(\theta,\phi)$. Fig 3(a) illustrates the correspondence between
375: the positions of maxima of $f_{21}(\theta,\phi)$ and the structure
376: [17] of L-A virus with $N=2$; Fig 3(b) relates the maxima of
377: $f_{25}(\theta,\phi)$ to the structure [17] of Dengue virus with
378: $N=3$ and in Fig 3(c) the maxima of $f_{37}(\theta,\phi)$ are
379: compared with the protein distribution [17] of Murine Polyoma virus
380: with $N=6$.
381:
382: As an additional comment note, that in our opinion (contrary to the
383: opinion of Ref. [18]) the above structures violating the CK
384: geometrical model do not violate the CK idea about the
385: quasiequivalence of proteins in the viral capsid. CK stated [5]
386: that since the proteins are identical their environments in the
387: viral structure should be similar. Initially, this idea was used by
388: CK to justify their geometrical model. The hexagonal planar
389: crystalline structure proposed by CK to be the first step of the model
390: contains six proteins per unit cell. All
391: proteins of the structure are symmetry-equivalent since they belong to the same
392: regular orbit of the corresponding planar symmetry group. On
393: the second step of the CK model, after the mapping of the
394: planar structure on the icosahedron surface, the same
395: proteins belong to different 60-fold orbits of the I-group.
396: In any 3D icosahedral capsid structure the
397: proteins which belong to different positions can not be
398: symmetry-equivalent. Nevertheless, the CK geometrical construction
399: ensures approximate structural equivalence of proteins from
400: different orbits. This "quasiequivalence" means that the local
401: order around any protein (the distances between proteins, the
402: number of nearest neighbors) is more or less the same. The latter
403: property is intrinsic not only to the CK structures but also for
404: the capsid structures violating the CK geometrical model including those shown
405: in Fig 3. Indeed, each position (location of maximum) in all these
406: structures have five or six nearest neighbors and the distances to
407: these neighbors are approximately equal. In other words if the
408: asymmetrical identical building blocks can be slightly deformed (
409: it is also assumed in the original CK theory) then there is no
410: problem to put them together in the structure in slightly different
411: local environments.
412:
413: Let us finally briefly discuss particular features of the assembly
414: thermodynamics. Due to the absence of the cubic term in free energy
415: (3) the icosahedral capsid assembly can be second order phase
416: transition. Thermodynamic processes of this type have two
417: advantages for the assembly optimization: they need no latent heat
418: to be involved in; and they take place without nucleation process.
419: The latter feature is confirmed experimentally for a number of
420: small viruses [19] : at equilibrium, either intact virus shell or
421: free proteins are dominant species while assembly intermediates
422: (capsid germs) are found in trace concentration.
423:
424: We would like to stress that irreducible icosahedral density
425: function $f_l(\theta,\phi)$ contains much more physical
426: information than simple positions of proteins centers. The full
427: density distribution generated by $f_l(\theta,\phi)$ is very
428: useful for understanding of biologically important properties like
429: virus infectivity. Recent advances in virology have shown that
430: that infectivity promoted by interaction of cell receptors with
431: virus surface depends not only on bio-specific binding properties but
432: also on the capsid proteins distribution. Along this line, the
433: relation can be established between the minima of $f_l(\theta,\phi)$
434: and binding sites on the capsid surface. One-to-one correspondence
435: of the deepest minima of $f_{25}$ (Fig. 1(c)) and the binding
436: sites for the carbohydrate recognition domains of the dendritic
437: cell receptors on the Dengue virus surface [Fig 2 in Ref. 20] can
438: be taken as an illustration of the relation.
439:
440: \begin{center}
441: {\Large Figure captions}
442: \end{center}
443: \bigskip
444:
445:
446: Fig 1. (a)-(e) : The first five irreducible icosahedral density functions
447: with the wave numbers $l=$15, 21, 25, 27, and 31, respectively. Corresponding numbers of
448: different 60-fold positions of density maxima are $N=$1, 2, 3, 3, and
449: 4. (f): Function with $l=$37 and $N=$6.
450:
451: Fig 2. Comparison of the positions of proteins centers predicted by
452: our model (left panel) with the experimental viral structures [17]
453: (right panel) for the capsids satisfying selection rules of the CK
454: geometrical model. Capsids of Satellite Tobacco Mosaic virus (a),
455: Cowpea Chlorotic Mottle Virus (b), and Sindbis virus (c) are presented. The
456: corresponding density functions for $l=15$,
457: $l=27$, and $l=31$, respectively, are shown in Fig. 1(a), (d), and (e).
458:
459: Fig 3. Comparison of the positions of proteins centers predicted by our model
460: (left panel) with the experimental viral structures [17] (right panel)
461: for the capsids which can not be explained by the CK geometrical
462: model. Capsids of L-A virus (a), Dengue
463: virus (b), Murine Polyoma virus (c) are presented. The corresponding
464: density functions for $l=21$, $l=25$, and
465: $l=37$, respectively are shown in Fig. 1(b), (c), and (f).
466:
467:
468: \begin{thebibliography}{20}
469: \bibitem{1}
470: S. J. Flint et. al., \textit{Principles of Virology:
471: Molecular Biology, Pathogenesis, and Control}
472: (ASM Press,Washington, 2000).
473: \bibitem{2}
474: H. Fraenkel-Conrat and R.C.Williams, Proc. Natl. Acad.
475: Sci. U.S.A. {\bf 41}, 690 (1955).
476: \bibitem{3}
477: J. B. Bancroft, \textit{Advances of Virus Research} (Academic,
478: New York), Vol. 16, p. 99. (1970).
479: \bibitem{4}
480: F. H. C. Crick and J. D.Watson, Nature (London) {\bf 177}, 473
481: (1956).
482: \bibitem{5}
483: D. L. D. Caspar and A. Klug, Cold Spring Harbor Symp.
484: Quant. Biol. {\bf 27}, 1 (1962).
485: \bibitem{6}
486: H. Naitow et al., Nature Struct. Biol. {\bf 9}, 725 (2002); I.B.
487: Rayment et al., Nature {\bf 295}, 110 (1982); R.C.
488: Liddington et al., Nature {\bf 354}, 278 (1991); R.J.
489: Kuhn et al., Cell {\bf 108}, 717 (2002); S Mukhopadhyay
490: et al., Science {\bf 302}, 248 (2003).
491: \bibitem{7}
492: T.S. Baker, N.H. Olson, and S.D. Fuller, Microbiol. Mol. Biol.
493: Rev. {\bf 63}, 862 (1999) and references therein.
494: \bibitem{8}
495: A. Zlotnick, Proc. Natl. Acad.
496: Sci. U.S.A., {\bf 101}, 15549 (2004).
497: \bibitem{9}
498: R.F. Bruinsma, W.M. Gelbart, D. Reguera, J. Rudnick, and R.Zandi,
499: Phys. Rev. Lett., {\bf 90}, 248101 (2003).
500: \bibitem{10}
501: J. Lidmar, L. Mirny, and D. R. Nelson,
502: Phys. Rev. E, {\bf 68}, 051910 (2003).
503: \bibitem{11}
504: R. Zandi and D. Reguera,
505: Phys. Rev. E, {\bf 72}, 021917 (2005).
506: \bibitem{12}
507: T. T. Nguyen, R. F. Bruinsma, and W. M. Gelbart
508: Phys. Rev. Lett., {\bf 96}, 078102 (2006).
509: \bibitem{13}
510: L. D. Landau, Phys. Zs. Sowjet., {\bf 11}, 545 (1937);
511: L. D. Landau, Phys. Zs. Sowjet., {\bf 11}, 26 (1937);
512: S. Alexander and J. McTague,
513: Phys. Rev. Lett. {\bf 41}, 702 (1978);
514: \bibitem{14}
515: P.J. Steinhardt, D.R. Nelson, and M. Ronchetti, Phys. Rev. B
516: {\bf 28}, 784 (1983).
517: \bibitem{15}
518: J.P. Elliot and P.G. Dawber, \textit{ Symmetry in Physics}
519: (Macmillan Press, London), (1979).
520: \bibitem{16}
521: A computer program composed to calculate the irreducible
522: icosahedral structures can be found at http://www.lpta.univ-montp2.fr/article.php3?id\_article=132.
523: \bibitem{17}
524: Experimental structures are reproduced using the UCSF Chimera
525: package. E.F. Pettersen, T.D. Goddard, C.C. Huang, G.S. Couch,
526: D.M. Greenblatt, E.C. Meng, and T.E. Ferrin, J.
527: Comput. Chem. {\bf 25} 1605 (2004).
528: \bibitem{18}
529: R. Twarock, J. Theor. Biol., {\bf 226}, 477 (2004).
530: \bibitem{19}
531: A. Zlotnik, J. Mol. Biol., {\bf 241}, 59 (1994).
532: \bibitem{20}
533: E. Pokidysheva et. al., Cell, {\bf 124}, 485 (2006).
534:
535: \end{thebibliography}
536:
537: \end{document}
538:
539: