Академический Документы
Профессиональный Документы
Культура Документы
Dynamics
Author(s): Piet Gros, Wilfred F. van Gunsteren and Wim G. J. Hol
Source: Science, New Series, Vol. 249, No. 4973 (Sep. 7, 1990), pp. 1149-1152
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/2878040
Accessed: 18-09-2016 01:49 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
American Association for the Advancement of Science is collaborating with JSTOR to digitize,
preserve and extend access to Science
This content downloaded from 132.239.1.231 on Sun, 18 Sep 2016 01:49:54 UTC
All use subject to http://about.jstor.org/terms
4. E. J. Ginsburg, C. B. Gorman, S. R. Marder, R. H. Macromnoleciules 20, 212 (1987); K. Y. Jen, R. L. crystallographic data by restraining the
Grubbs, J. Am. Chem. Soc. 111, 7621 (1989). Oboodi, R. L. Elsenbaumer, Polytn. Mater. Sci. Etig.
53, 79 (1985); K. Y. Jen, M. Maxfield, L. W.
structure factor amplitudes taken over the
5. C. B. Gorman, E. J. Ginsburg, S. R. Marder, R. H.
Grubbs, Angew. Chem. Adv. Mater. 101, 1603 Shacklette, R. L. Elsenbaumer, J. Chetn. Soc. Chetn. whole ensemble of structures to the ob-
(1989). Commuit. 4, 309 (1987); S. D. D. V. Rughooputh, served x-ray data. In this crystallographically
6. C. Schaverien, J. Dewan, R. R. Schrock, J. Am. M. Nowak, S. Hotta, A. J. Heeger, F. Wudl, Sytnth.
Met. 21, 41 (1987); D. C. Bott et al., ibid. 14, 245
restrained MD simulation the following tar-
Chem. Soc. 108, 2771 (1986).
7. A. M. Saxman, R. Liepins, M. Aldissi, Prog. Polym. (1986); J. Edwards, W. J. Feast, D. Bott, Polymer get function is used:
Sci. 11, 57 (1985). 25, 395 (1984).
8. M. Aldissi, Inherently Conducting Polymers: Processing, 17. A. J. Frank, S. Glenis, A. J. Nelson, J. Phys. Chem.
E= Ephys + 2 E (IFo(s)l - kl(F,(s))I)2
Fabrication, Applications, Limitationts (Noyes Data, 93, 3818 (1989); M. Ozaki, D. Peebles, B. R. crX( ()
Park Ridge, NJ, 1989). Weinberger, A. J. Heeger, A. G. MacDiarmid, J.
9. H. Shirakawa, E. J. Louis, A. G. MacDiarmid, C. K. Appl. Phiys. 51, 4252 (1980); M. J. Cohen and J. S.
Chiang, A. J. Heeger, J. Chem. Soc. Chem. Comtntiti. Harris, Jr., Appl. Phys. Lett. 33, 812 (1978).
where Ephys r
1977, 578 (1977). 18. F. Gamier and G. Horowitz, Sytnth. Met. 18, 693 gle, dihedral, im
10. R. Z. Bachrach, in Metal-Semiconductor Schottky Barm- (1987). ic, and Lennard-Jones interaction potentials
erjunctionis and Their Applications, B. L. Sharma, Ed. 19. T. Skotheim, 0. Inganas, J. Prejza, I. Lundstroem,
(Plenum, New York, 1984), pp. 93-99; S. M. Sze, Mol. Cryst. Liq. Cryst. 83, 1361 (1982); T. Skoth-
(4), s is the reciprocal lattice vector, k is a
Physics of Semiconductor Devices (Wiley, New York, eim and I. Lundstroem, J. Electrochietn. Soc. 129, scale factor, and crx is a weighting factor.
1981), pp. 245-311. 894 (1982); T. A. Skotheim and 0. Inganas, ibid.
The essential difference with the MD crystal-
11. N. S. Lewis, J. Electrochem. Soc. 131, 2496 (1984). 132, 2116 (1985); G. Nagasubramanian, S. DiSte-
12. A. L. Fahrenbruch and R. H. Bube, Fundatnetntals lographic refinement procedure (5, 6) is the
ofJ. Moacanin, in Pol ymers for Advatnced Technolo-
fano,
Solar Cells (Academic Press, New York, 1983), useEd.
gies: IUPAC Internationial Sytnposium, M. Lewin, of the ensemble average of calculated
chap. 6. (VCH, New York, 1988), pp. 261-268.
13. , ibid., chap. 7. 20. Supported bv NSF grant CHE8814694 and ONR structure factors, (F,), instead of the struc-
14. D. B. Neal et al., Sytnth. Met. 28, D711 (1989); M. grant N001488K0208. C.B.G. thanks the Jet Pro- ture factors of one single structure. We
C. J. Young, R. H. Tredgold, P. Hodge, R. Soc. pulsion Laboratory, Caltech, for a research fellow- considered similar arguments as Torda et al.
Chem. Spec. Publ. 69, 354 (1989); R. H. Tredgold, ship. E.J.G. thanks IBM for a graduate fellowship,
M. C. J. Young, P. Hodge, E. Khoshdel, Thini Solid and A.K. thanks the Department of Education for a (7, 8) for combining nuclear magnetic reso-
Films 151, 441 (1987). graduate fellowship. Contribution 8134 from the nance-nuclear Overhauser effect data and
15. F. L. Klavetter and R. H. Grubbs, Synth. Met. 28, Arnold and Mabel Beckman Laboratory of Chemical
MD procedures and used a time-weighted,
D105 (1989). Synthesis and the Arthur Amos Noyes Laboratory
16. S. H. Askari, S. D. Rughooputh, F. Wudl, A. J. of Chemical Physics. or "running," average for the calculation of
Heeger, Polym. Prep. 30, 157 (1989); S. Hotta, S.
(F,) in the restraint:
D. D. V. Rughooputh, A. J. Heeger, F. Wudl, 7 Mav 1990; accepted 26 June 1990
=Fc(s)t,
TX(1=-e-t'
- /Tx
This content downloaded from 132.239.1.231 on Sun, 18 Sep 2016 01:49:54 UTC
All use subject to http://about.jstor.org/terms
A B reflections between 1.75 and 1.70 A resolu-
0.30 0.70 tion, the R-factor drops from 0.236 to
0.104. At low resolution the R-factor re-
tr -Tc (ps,ps) 0.60 mains somewhat higher, (0.159 for reflec-
0.4 0 0 0.50 tions between 8.0 and 6.3 A resolution)
00.40 because for these data the solvent continu-
um contributes significantly to the structure
10,2 0.30
0.18 -0.20
amplitudes. In this explorative study the
bulk water in the crystal was not simulated
20,4 0.10 and this omission caused errors at low reso-
0.12 4060.00 lution. During the simulation all tempera-
80 16 - 4- -' 40 48 56 64 72 80 ture factors were kept at 1 A2 and all atoms
40-80,16 t- - Time (ps) fluctuated around their average positions.
Thus a high crystallographical residual,
0.06 Fig. 1. R-factors obtained by time-averaged crys-
R - 0.65, was observed for each individual
tallographically restrained MD of BPLA2. (A)
Crystallographic residual, R = [Y. (IFo(s) I - structure of the ensemble. After averaging
the structure factors of all structures in the
0.00 ________________________________ IFmode1(S)I)/1 IF0(s)l], as a fimction ofthe resolu-
1.75 1.94 2.22 2.67 3.58 8. 00 ton, where: (i) Fmodel was calculated from the ensemble, the R-factor dropped to 0.098,
Resolution curve is indicated by t = O ps and 'r = Owhen all data with IF.1 > ( between 8 and
Resolution (A) classically refined model (10). The corresponding
ps (curve
"0, 0"). In this model individual atomic temperature factors were used (10); (ii) Fmodel1.7 wasAcalculated
resolutionas were used (see Fig. 1B),
weighted averages with Eq. 2, from the MD run at times t = 10 ps with T = 2 ps (curve "10, 2"), whereas the residual of the classically refined
t= 20ps with Ti = 4ps (curve "20,4"), t = 40ps with T = 16ps (curve "40, 16"), and t = 80ps alsomodel is 0.171 (10). This reduction in R-
with T = 16 ps (curve "80, 16"). During the run a temperature factor of 1 A2 was assigned to each
factor was obtained while a good geometry
atom, ,afd (iii) Fmijei was calculated as the average structure factors from the trajectory t = 40 to 80 ps
of the structures in the trajectory was re-
with TX = 16 ps (curve "40-80, 16"). Additional overall scaling of IFm dell to 1F.1, yielding a scale
tained (14-17).
factor of k = 1.082 and a relative overall temperature factor Bis, = 1.028 X27 reduced the residual from
0.104 to 0.098. (B) The R-factor, based on the averaged Fc, as function of time from t = 40 ps to Perhaps more important is that the en-
t = 80 ps. At t = 40 ps one single structure had a crystallographic residual of 0.65. After averaging over
semble of structures contains information
a time span of 40 ps, that is, 20,000 structures, the R-factor dropped to 0.104, without additional
on the mobility of the protein molecule in
overall scaling (14).
the crystal lattice. An impression of the
A atomic fluctuations is obtained by inspect-
S e Be ing Cot tracings of BPLA2 models occurring
at 4-ps intervals (Fig. 2A). As expected, the
Val 65 Va l 65 interior of the molecules shows only small
fluctuations around the average atomic posi-
tions. In contrast, some surface residues
appear to be highly mobile, even with re-
spect to their backbone conformation. This
motion involves residues Gly30 to Gly35,
C y~~~Cs 123 yy s 123 Val62 to Val', Ser77 to Glu81, and the
COOH-terminal loop Asn" 17 to Cys123 (Fig.
2B). These last three regions are also the
only three regions that differ significantly
between phospholipase A2 from the Crotalus
Fig. 2. Mobility of BPLA2 as re- B 70
vealed by crystallographically re- atrox venom and the bovine enzyme (13).
strained MD. (A) The Cot tracings 60 Furthermore, loops 30 to 35, 62 to 66, and
of eight structures of BPLA2 as 117 to 123 are thought to be part of the
observed at 4-ps intervals for li binding site of aggregated phospholipids
t = 40 to 68 ps. From t = 48 to 60
ps some residues from the loop 112 n 40
(18). Hence, the observed flexibility for
to 123 occupied positions of a these loops could well be important for the
neighboring molecule, which i 30 interaction of the protein with the layer of
shows that crystal contacts should polar headgroups of the aggregated phos-
be treated explicitly in these calcula- 20 i II i!
pholipids.
tions either by van der Waals re-
straints (22) in a vacuo simulation 10 I A <. In the case of BPLA2 the mobile loops
or by simulating a whole unit cell appear to have one major conformation,
with periodic boundary conditions. 0 - which is observed in the electron density
(B) Comparison between isotropic 0 20 40 60 80 100 120 map. As an example we show in Fig. 3 two
temperature factors averaged over Residue number
the main chain atoms as determrined
residues from the mobile loop consisting of
residues 77collected
in the classical refinement (10) (solid lines) and as retrieved from the trajectory to 81. Thefrom
electron density
t = 40 toof
the single
80 ps (dashed lines). The isotropic B-factors from the MD ensemble were distinct
obtained as conformation
a mean squareis repro-
fluctuation around one single average conformation of the protein, which means thatthe
duced with these B-factors
structure factorsfor
of the en-
the hypermobile regions contain contributions related to different atomic equilibrium positions and
semble in a (21F01 - I(FclI)expiaav map
therefore reflect merely their hypermobility. The correlation coefficient between the classical crystallo-
graphic temperature factors and the isotropic temperature factors derived (Fig.from
3B), where
the aav stands for is
simulation the0.45
phase of
(Fe).improves
for all 957 non-hydrogen atoms; disregarding the outliers (AB > 20 A^) An electron the
density map,, similar
correlation toto the
0.73 for 640 atoms. one shown in Fig. 3B, is obtained when
This content downloaded from 132.239.1.231 on Sun, 18 Sep 2016 01:49:54 UTC
All use subject to http://about.jstor.org/terms
only calculated data are used, that is, with crystal it may be required to perform sepa- These errors were corrected automatically in
coefficients (I(Fc)I)expiaav. This result rate MD runs with different starting models. an early stage of the simulation, that is when
shows that the large spatial distribution of Hence, when simulating a crystal explicitly T, was small and the individual isotropic
the atoms seen in Fig. 3C results in a clear temperature factors were still largely pre-
and sampling for a longer time ('r,), an even
electron density for this loop that is missingmore reliable ensemble of structures and sent. Thus it is conceivable that the tech-
only the electron density of the side chain of further agreement with the experimental
nique might also be useful in the refinement
Asn79. The appearance of electron density at data is to be expected. of the single structure itself. Briinger (21)
the position of the side chain atoms of Asn79 In the case of BPLA2 only a few minor showed that when a series of structures is
in the classical map (Fig. 3A) might be coordinate errors were present in the start- generated by classical MD refinement, large
caused by model bias from the full occupan- ing model of the simulation, for instance, fluctuations may indicate either disordered
cy of the single site isotropic model. Fur- wrongly placed CA of Leu41 or the wrongly or erroneously fitted segments of the mole-
thermore, the discrepancy in isotropic tem- placed solvent-accessible side chain of Arg43. cule. For the time-averaged restrained simu-
perature factors (Fig. 2B) for these very
mobile regions can be explained by the fact
that classically determined B-factors refer to
the main conformation only, whereas the B-
factors retrieved from the ensemble take all
major and minor conformations into ac-
count. This result means that the interpreta-
tion of the most frequently occurring dis-
tinct conformation for the loop 77 to 81 in
the conventionally refined structure is essen-
tially correct, although this conformation is
only valid for a limited fraction of structures.
An important parameter in these crystal-
lographically restrained MD simulations is
T, The crystallographic restraint (see Eq. 1)
is nonconservative and leads to heating,
which is dependent on the magnitude of t.
The longer T is, the less heating is observed
(14). According to (8), Ti should be chosen
to be as long as possible but still roughly an
order of magnitude smaller than the length
of the simulation.
The method we present can be extended
by including the solvent continuum and the
crystal lattice explicitly in the simulation,
which implies application of the periodic
boundary conditions (19) to the unit cell
and addition of the bulk water. Most likely
this should improve the correlation between
observed and calculated x-ray data, particu-
larly at low resolution. Additional improve-
ment can then be obtained by replacing the
vacuo parameter settings of the MD force
field by the physically more realistic parame-
ter set used for simulation in water (4, 20).
Furthermore, to model static disorder of a
/ ~+ '+
This content downloaded from 132.239.1.231 on Sun, 18 Sep 2016 01:49:54 UTC
All use subject to http://about.jstor.org/terms
each atom wvere gradually reduced to 1 A2 for all ensemble.
lation it remains to be seen what the effects
atoms, wvhile TX was increased from 0 to 4 ps. At 15. J. P. Ryckaert, G. Ciccotti, H. J. C. Berendsen, J.
will be when the starting model contains Comnpuit. Phys. 23, 327 (1977).
t = 20 ps T, wvas increased to 8 ps and at t = 26 ps to
large intrinsic errors. Moreover, for struc- Tx= 16 ps. Up to t= 28 ps the temperature was 16. H. J. C. Berendsen, J. P. M. Postma, W. F. van
coupled to a bath (16) of 150 K wvith a temperature- Gunsteren, A. Dinola, J. R. Haak, J. Chetn. Phys.
tures successfully refined at high resolution
relaxation time, TT, of 0.012 ps; afterwvard the bath 81, 3684 (1984).
by classical procedures, the occurrence of temperaturc was set to 200 K. During the whole run 17. D. E. Tronrud, L. ten Eyck, B. W. Marthewvs, Acta
large errors is most unlikely, and these struc- ax wvas kept at 140 (arbitrary units), which is Crystallogr. A 43, 489 (1987).
tures are therefore good starting points to roughly equivalcnt to (<(IF,I -IF,1)')"') at 18.
t =0. 0
P. Kuipers et al., Scietnce 244, 82 (1989).
ps. All of thc x-ray data betwecn 8.0 and 1.7 A 19. M. P. Allen and D. J. Tildesley, Cotmputer Sitnulation
study the anisotropy and anharmonicity of resolution with IF0I > uT0 were used in thc calcula- of Liquiids (Clarendon, Oxford, 1987).
the atomic distributions in the molecule by tions. Heating wvas observed in the presented simu- 20. J. Hermans, H. J. C. Berendsen, W. F. van Gun-
lation. The amounit of heating depends on the steren, J. P. M. Postma, Biopolymners 23, 1513
the method we present.
magnitude of Tx; for T, = 4 ps, AT = 108 K, for (1984).
Time-averaged crystallographically re- T, = 8 ps, AT = 56 K, and for T, = 16 ps, AT = 30 21. A. T. Briinger, J. Mol. Biol. 203, 803 (1988).
strained MD shows that phospholipase A2 is K, whcre AT is defined as (T) - Tbth. The average 22. M. G. Vincent an-d J. P. Priestle, J. Appl. Crystallogr.
potential energy of the simulated system observed 18, 185 (1985).
more flexible than single site isotropic mod- from t = 40 to 80 ps is -4.52 x 10- kJ/molc, 23. We thank A. E. Torda and M. Fujinaga for helpful
els obtained by classical crystallographic re- compared wvith -5.31 x 10' kJ/mole for an unre- and stimtulating discussions and B. W. Dijkstra for
straincd simtulation of BPLA2 in vacuo at 300 K. A generously supplying the x-ray diffraction data of
finement techniques would suggest. Large
related measure is the average root-mean-square phospholipase. Supported by the Dutch Foundation
deviations from the main conformation are deviations from standard gcomctry of the structures for Chemical Research (SON) with financial aid
observed, and these are likely to be impor- from the ensemble, wvhich are 0.015 A for bond from the Dutch Organization for Scientific Research
lengths, 4.2? for bond angles, and 28? for torsion (NWO).
tant for the functioning of the molecule.
angles, as calculatcd by the progranm TNT (17) from
Time-averaged crystallographically re- a subset of structures taken at 4-ps intervals from the 24 April 1990; accepted 25 July 1990
strained MD, when applied to structure
refinement, allows for more mobility and
better searching of the conformational
space, in comparison to least-squares meth-
ods and MD refinement procedures (5, 6),
and provides a more complete representa- Possible Early Pennsylvanian Ancestor
tion of a biomacromolecule by calculating of the Cycadales
the ensemble of structures based on the
experimental data. Accurate knowledge of
molecular dynamics may prove useful in
RIcHARD L. LEARY
various fields of biochemistry.
A specimen of Lesleya, previously known only as isolated sterile foliage, has been
found with two rows of ovule-bearing receptacles on opposite sides of the petiole. This
REFERENCES AND NOTES specimen is from the Lower Pennsylvanian (Namurian B or C) of western Illinois. The
1. B. Howlin, D. S. Moss, G. W. Harris, Acta Crystal- fertile portion of this specimen is similar to Phasmatocycas kansana Mamay from the
logr. A 45, 851 (1989). Lower Permian of Kansas and Texas, which Mamay interpreted as an ancestral form to
2. J. Kuriyan, G. A. Petsko, R. M. Levy, M. Karplus,
J. Mol. Biol. 190, 227 (1986).
cycads. Phasmatocycas was found attached to the base of Taeniopteris by Gillespie and
3. W. F. van Gunsteren, H. J. C. Berendsen, J. Her- Pfefferkorn. The foliar portions, Lesleya and Taeniopteris, have many features in
mans, W. G. J. Hol, J. P. M. Postma, Proc. Nati. common; the major morphologic difference is curvature of the veins. Lesleya and its
Acad. Sci. U. S. A. 80, 4315 (1983).
4. W. F. van Gunsteren and H. J. C. Berendsen, attached fertile petiole are proposed as ancestors of Taeniopteris and Phasmatocycas,
GROningen MOlecular Simulation (GROMOS) li- and ultimately, the ancestors of modern cycads. This extends the age of known
brary manual, BIOMOS Biomolecular Software
precursors of cycads back from the Early Permian to Early Pennsylvanian (about 320
(Laboratory of Physical Chemistrv, Universitv of
Groningen, Groningen, 1987). million years ago), approximately 35 million years earlier.
5. A. T. Briinger, J. Kuriyan, M. Karplus, Science 235,
458 (1987).
6. M. Fujinaga, P. Gros, W. F. van Gunsteren, J.
URING THE PAST 20 YEARS, SEVER- that cyvcads were derived from plants with
Appi. Crystallogr. 22, 1 (1989). al fossil discoveries have shed lightentire leaves and that compound leaves were
7. A. E. Torda, R. M. Scheek, W. F. van Gunsteren, on the question of the origin of derived as suggested by Mamay [figure 11
Chem. Phys. Lett. 157, 289 (1989).
8. , J. Mol. Biol., in press. cycads. Delevorvas (1) suggested an origin of (5)].
9. W. A. Pieterson, J. C. Vidal, J. J. Volwerk, within
G. H.the
de medullosan pteridosperms. He Several sites in western Illinois have yield-
Haas, Biochemistry 13, 1455 (1974).
suggested that cvcads and cycadeoids were ed significant information about Early
10. B. W. Dijkstra, K. H. Kalk, W. G. J. Hol, J. Drenth,
J. Mol. Biol. 147, 97 (1981). derived from slender plants bearing com- Pennsylvanian nonswamp floras, both on
11. B. W. Dijkstra, J. Drenth, K. H. Kalk, Natture 289, leaves. He (2) compared the Pennsyl-
pound their overall composition and on individual
604 (1981).
vanian pinnately compound Eretnopteris za- taxa within the floras (6, 7). The Allied
12. E. A. Dennis, Thie Enzytnes, P. D. Boyer, Ed.
(Academic Press, New York, 1983), vol. 16, pp. mnioides (Bertrand) Kidston and Permian Stone Company quarry in Milan, Illinois,
307-353.
Tinsleya texaia Mamnay to compound early (8) has been a major collecting site for nearly
13. R. Renetseder, S. Brunie, B. W. Dijkstra, J. Drenth,
P. B. Sigler, J. Biol. Chem. 260, 11627 (1985).
cycad foliage. 15 years and continues to produce new
14. The GROMOS (4)-MDXREF (6) program was Mamay (3), on the other hand, in describ- species and genera as well as new data on
adjusted according to Eqs. 1 and 2. For this pur-
ing Archaeocycas and Phlasinatocycas, suggested known taxa.
pose Eq. 2 was rewritten as: (F,(s))r =
that cycads were derived from plants with
exp{-At/TXJ(F.),. - A, + (1 - exp{-At/TdFC (where Specimens described here were collected
At is the tinme step in the simulation). This equation
entire leaves. This was borne out by Gilles- from Lower Pennsylvanian sedimentary
is equivalent to Eq. 2 when t' >?> r. The simula-
tion was performed in vacuo with all protein atoms
pie and Pfefferkorn (4) who found Taeniop- rocks which fill channels eroded in the De-
and 106 crystal water molecules of BPLA,. The teris attached to Phlasinatocycas. The material vonian Cedar Valley Limestone (8). The
GROMOS 37D4 force field of (4) was used with
described here provides additional evidence fossil-bearing strata consist of shale, silt-
the SHAKE algorithm to constrain the bond lengths
(15) and with time steps of 2 fs. From t = 0 ps to 16 Dcpartmcnt of Geology, Illinois State Museum, Spring- stone, sandstone, and occasional thin con-
ps the individual isotropic temperature factors of field, IL 61706. glomerate.
This content downloaded from 132.239.1.231 on Sun, 18 Sep 2016 01:49:54 UTC
All use subject to http://about.jstor.org/terms