©
1996 Oxford University Press
2767-2773 Footnote
Analysis of co-crystal structures to identify the stereochemical determinants of the
orientation of TBP on the TATA box
Analysis of co-crystal structures to identify the stereochemical determinants of the orientation of TBP on the TATA box
Masashi
Suzuki
,
Mark D.
Allen
,
Naoto
Yagi
1
and
John T.
Finch
2,
*
AIST-NIBHT Structural Biology Centre, Higashi 1-1,
Tsukuba
305,
Japan
,
1
Tohoku University, School of Medicine, Seiryo-machi,
Sendai
980-77,
Japan
and
2
MRC Laboratory of Molecular Biology, Hills Road,
Cambridge
CB2 2QH,
UK
Received February 27, 1996;
Revised and Accepted May 23, 1996
ABSTRACT
Possible stereochemical determinants of the orientation of TBP on the TATA box
are discussed using the crystal coordinates of TBP-TATA complexes, which have been determined by other groups. The C-terminal half of the TBP
[beta]
-sheet interacts with the TATA site of the DNA, and the N-terminal half with the A-rich site, so that the two sites with distinct curvatures
produce a unique fit. Although chemical contacts take place between one side of
the
[beta]
-sheet and the DNA minor groove, the interaction seems to be facilitated
indirectly by the characteristics of the other side of the
[beta]
-sheet and the DNA major groove. Thus, Ala71, Leu162 and Pro190
differentiate the curvature of the
[beta]-sheet in the N- and C-halves. The methyl positions in the DNA major groove modulate
the bendability of the two DNA sites by using differences in the rolling
capacity of TA and AT compared with PyT, and in the shifting capacity of AT
compared with TT. The deformations of the first steps (TA and PyT) in the two
sites are the largest and thus are important for the overall bending of the
DNA. The differences between the two DNA sites are greatest at the second steps
(AT and TT) and so these are important for determining the orientation of TBP.
INTRODUCTION
In most eukaryotic systems of protein transcription, the TATA box is positioned
upstream of the transcription initiation site (
1
-
9
). Half of its sequence, 5'-TATA-3', is conserved well (referred to as the TATA site in
this paper), while the other half is less so but generally has many adenine
bases (referred to as the A-rich site).
Upon initiating transcription, the TATA box-binding protein, TBP, binds to the DNA, contacting the TATA site by its C-terminal domain and the A-rich site by its N-terminal domain. This polarity determines the direction
of transcription through further interaction between TBP and RNA polymerase II.
It is possible that some other proteins which interact with TBP might help for
fixing the orientation, but it is more likely that TBP by itself is able to do
so, since the orientation is kept in the same way in the two TBP-DNA complexes co-crystallised in the absence of such a protein [the two structures
are referred to as the Yale (
10
) and Rockefeller (
11
) structures].
It has been noticed that van der Waals contacts are important for TBP-TATA interaction (
10
,
11
), and it has been suggested that possible differences in the flexibility of two
DNA sites might be important for fixing the orientation (
10
-
12
). However, no stereochemical characteristic of the DNA or TBP involved in such
a mechanism has been specified. On the contrary, the structures have provoked a
number of puzzling questions.
The N- and C-domains of TBP have the same composition of secondary structural
elements and very similar three dimensional structures, and thus it is not
immediately obvious why the two domains can choose different partner sites. In
fact, the chemical contacts between the TBP domains and the DNA sites are quite
similar in the two sets. In addition, the TATA and A-rich sites are moulded into similar structures, while the two sequences
are expected to behave in very different ways (
12
,
14
-
16
).
In this paper to answer the above questions we re-analyse the co-crystal structures using their atomic coordinates. The first step is
to examine whether the two halves of the complexes are indeed as similar as
they appear.
COMPARISON OF THE TWO HALF STRUCTURES
Each domain of TBP is composed of two [alpha]-helices and five [beta]-strands. The ten [beta]-strands of the two domains fold into a
single [beta]-sheet, and eight [beta]-strands among the ten fit around the minor groove of the
DNA closely (Fig.
1
a).
Figure 1
.
Interaction of the TBP [beta]-sheet and the TATA box. (
a
) DNA binding mode of the TBP [beta]-sheet. Four strands in the N-terminal domain, N1, N3, N4 and N5, bind to the TATA site (the
base sequence is shown in upper case characters), another four strands in the C-terminal domain, C1, C3, C4 and C5 bind to the A-rich site (the base sequence is shown in lower case characters) in
the minor (m) groove. The DNA sequence shown here is that of the Rockefeller
structure. (
b
and
c
) Methyl-methyl distances (in Å) in the Yale (b) and Rockefeller (c) structures. The subfigures
are drawn looking into the major (M) groove of the DNA. In each structure the
asymmetric unit contains two molecules, and thus the averaged numbers are
shown. (
d
) Amino acid residues positioned on the protein (H2) side of the [beta]-sheet. The amino acid residues which are larger or smaller than
their neighbours on the DNA side are indicated. The rest have sizes similar to
their neighbours. For the comparison of sizes, the small (s), medium (m), large
(l) classification in ref. 30 is used-if an `m' residue is neighboured by an `m' residue and an `s' residue on
the other side, it is regarded as larger etc. (
e
) The shortest van der Waals' distances between side-chain atoms of amino acids in the (Yale/Rockefeller) structures in are a,
3.9/4.2
4.3/4.2
3.5/3.9
4.2/4.2
4.1/4.0
4.2/4.2
3.7/4.7
4.6/3.8
3.8/4.0
4.3/3.2
4.3/3.8
f
) Amino acid residues positioned on the DNA side. Amino acid residues which are
different from those occupying the equivalent position in the other half of the
[beta]-sheet are marked with asterisks. The six residues which contact C2'H atoms of the A bases are enclosed into two groups by broken
lines.
The three dimensional structures of the two halves of the complexes were
compared with each other by duplicating each crystal structure. One of the two
identical models (shown in yellow in Fig.
2
a and b) was rotated around the pseodo-dyad axis, and the N-terminal eight residues of [beta]-strand 1 (residues 24-31, here the numbers are given by those of the
Rockefeller structures, unless stated otherwise) and [alpha]-helix 2 (residues 87-98) in the N-domain were optimally superimposed onto their counter
parts (residues 24-31, and 178-189, respectively) in the C-domain of the other model (blue) and vice versa.
Figure 2
. Comparison of N- and C-terminal halves of TBP (
a
and
b
) and that of the two DNA sites (
c
and
d
). Two models were created using the same set of TBP-TATA co-crystal coordinates (those of the Rockefeller structure were used
in a and c, while those of the Yale structure in b and d). One of the two
identical models was rotated around the pseudo-dyad axis (shown by the broken line), by best overlapping the N-terminal eight residues of [beta]-strand 1 (residues 24-31) and [alpha]-helix 2 (residues 87-98) in the N-half onto their counter
parts (residues 114-121, 178-189) in the C-half of the other model. In (a) and (b) the DNA structures
are omitted. Pro190 and Pro323 in the C-domains are indicated. In (c) and (d) the TBP structures are omitted. The
green DNA molecule binds to the yellow TBP, placing the TATA site on the left,
and the A-rich site on the right, and the red DNA binds the blue TBP. The figure was
prepared by using the program Molscript (31). M and m: the major and minor
grooves, respectively.
In all the four structures (each of the Yale and Rockefeller structures has two
molecules in the asymmetric unit) the curvature of the TBP [beta]-sheet is larger in the C-domain than in the N-domain, and, as a consequence, the C-terminal half of the [beta]-sheet is curved more towards the DNA
(Fig.
2
a and b). For better fitting with the C-domain, the TATA site of the DNA bends more around the major groove than
the A-rich site (Fig.
2
c and d). We therefore propose that the different curvature of the [beta]-sheet in the two domains is used to choose their native partner
sites by detecting the different bendability of the two DNA sequences.
Three key amino acid positions in TBP
The difference in curvature at the C- and N-terminal halves of the [beta]-sheet is kept essentially the same even in the absence
of DNA (data not shown, see also
21
-
23
), and thus it is likely to be fixed by the protein itself.
Alternate positions along the [beta]-strands face the DNA on one side of the [beta]-sheet (here referred to as the DNA side, Fig.
1
f) and the rest face the [alpha]-helix 2 on the other side (here referred to as the H2 side, Figs
1
d and e, 2a and b). If each pair of equivalent amino acid positions on the two
sides were occupied by the same types of amino acid residues, the two sides
would be identical, and thus the [beta]-sheet would not curve around the DNA. In reality, the sizes of
residues on the H2 side are generally larger than those on the DNA side in the
C-domain, while smaller in the N-domain (Fig.
1
d). By such positioning, the curvature of the [beta]-sheet at the C-half is likely to be enhanced, while that at the N-half is relaxed.
Not much difference is found between the two halves at amino acid positions on
the DNA side except in strands 1 (Fig.
1
f), but on the H2 side, most positions in the C-terminal half are occupied by amino acid residues larger than those found
at the equivalent positions in the N-terminal half. In particular, closer inspection suggests that Ala71 and
Leu162 on the H2 side are likely to be the key residues (doubly circled in Fig.
1
e). The smallest van der Waals' distances from the side-chain atoms of Leu162 to the nearby six positions are shorter, by >= 0.5 Å, than the equivalent distances measured from Ala71, except for
the Leu162-Ile164 distance (4.3-4.5 Å ), which is slightly longer than its equivalent, Ala71-Ile73 (4.2 Å ) (Fig.
1
e).
Leu162 itself is a large residue, and is surrounded by other large residues,
Tyr153 and Met155, and branched residues, Ile160, Ile164, Ile170 and Ile172. It
is the tight packing of these residues that appears to create the higher
curvature. In contrast, most of the residues surrounding Ala71 are smaller than
the residues at the equivalent positions in the C-terminus. In brief, Leu162 and Ala71 are found at the centres of patches,
respectively, of tightly packed larger residues, and of loosely packed smaller
residues.
Helix 2 packs against the [beta]-sheet (Fig.
2
). In the C-domain, helix 2 kinks towards the DNA at Pro190 (Pro232 in the Yale
structure), while in the N-half it is straighter (Fig.
2
a and b). The proline is well conserved in the C-domain but not found at the equivalent position in the N-domain. Thus Pro190 is another key residue which is likely to assist
the increased curvature of the [beta]-sheet by kinking helix 2 towards another key residue Leu162 (Ile122
in helix 2 contacts Leu162). In the N-domain the straighter helix 2 faces Ala71, which is smaller (Tyr97 in
helix 2 contacts Ala71).
Positions of methyl groups in the DNA major groove
As a whole the DNA is helically untwisted and considerably bent around the major
groove, exposing the minor groove to TBP (Figs
1
a and
2
). Here the bases are numbered as follows: 5'-T
1
A
2
T
3
A
4
(T/A)
5
A
6
A
7
(A/G)
8
-3'/5'-(T/C)
1
T
2
T
3
(A/T)
4
T
5
A
6
T
7
A
8
-3'.
An A[middot]T base pair has a CH group (C2'H of A) near the centre in the minor groove, which can be used
for hydrophobic contact, and two hydrogen bond acceptors (N3' of A and O2' of T) on both sides of the CH. Thus it seems very difficult to
discriminate between A[middot]T and T[middot]A base pairs on their minor groove sides. Clear differences are,
however, found on the major groove sides-i.e. in the position of the methyl group of the T base. The bulky methyl
group of a T base is found at the far end of the major groove edge and
restricts the DNA conformation in many ways (
19
,
20
).
Bending the TATA box around the major groove pushes the methyl groups towards
each other within the groove (Fig.
2
c and d). This seems to be easily done at the TATA site, since the methyl groups
are positioned on alternating strands along the groove (Fig.
1
b and c). The methyl-methyl distances change from 8.6-11.1 in the standard B-conformation to 6.6-10.6 in the TBP structures. In an A-tract the methyl groups of neighbouring T bases
are only 4.8 apart even in the standard B-conformation, and become even closer, 3.8-4.2 apart, upon binding TBP. Thus even though the A-rich site is bent to a lesser extent than the TATA site, it
nevertheless appears to have been bent up to the limit of the structure.
Dinucleotide steps crucial for adapting to the TBP surface
Six parameters are used for describing the geometrical relationship between two
neighbouring base pairs (
24
; Fig.
3
of this paper). Among the three rotational angles (helical twist, roll and
tilt) the roll parameter is the most important for understanding the bending of
DNA (
25
; Fig.
4
i of this paper).
Figure 3
.
Dinucleotide step parameters. Those of the Yale (Y, -) and Rockefeller (R, [circle]) structures. The values expected for the standard B-DNA (36o for helical twist etc.) are shown by horizontal lines. In
(a), (b), (e) and (f) the T
1
A
2
/T
7
A
8
and (A/G)
7
A
8
/T
1
(T/C)
2
steps, and the A
2
T
3
/A
6
T
7
and A
6
A
7
/T
2
3
o , the sign becomes reversed. For direct comparison of the two half
sites those parameters for the right half are plotted according to the reversed
axis shown on the right. If, for example, the tilt angle of A
2
T
3
is +10o and if that of T
6
T
7
is -10o , that of A
2
A
3
is +10o , and thus the two steps are identical. Calculation of the parameters was
carried out using a computer program (32,33).
Figure 4
.
Shifting and rolling of dinucleotide steps. `M' and `m' indicate, respectively,
the sides of the major and minor grooves. Signs, `-' and `+', show, positive and negative directions of the movement. (
a
and
d
) Negative shifting of a TT step (`1' in d) would move the sugar-phosphate backbone (P) closer to the nearby methyl group (`2'), and thus
is tolerated only to a small extent. When it takes place, to avoid the clash,
the methyl group moves in the direction shown by a white arrow (a), which
creates small positive sliding. (
b
and
e
) Negative shifting of an AT step (`1' in e) and the approach of the sugar-phosphate backbone to the methyl group (`2' in e) can be tolerated by
negative sliding (`3' in e). In other words by helically untwisting (shown by a
white arrow in b) the methyl group moves away from the sugar-phosphate backbone, and thereby create negative sliding and negative
shifting of the step. (
c
and
f
) A TA step can adopt the same tactics as in (b) and (e). Alternatively, it can
roll in the positive direction (`1' in f), which is accompanied by helically
untwisting (`2' in f) and thereby pushing the sugar-phosphate backbone away from the methyl groups. (g and h) Shifting of the
A-rich (g) and TATA (h) sites and rolling of a dinucleotide step (i). The
numbers shown are those of average of the Yale and Rockefeller structures.
The T
1
A
2
step shows the highest rolling (of ~48 degrees). The A
7
(A/G)
8
step (or the (T/C)
1
T
2
step) rolls to a lesser extent (~42 degrees). The neighbouring A
2
T
3
step rolls higher than its equivalent T
2
N
3
(20o versus 15o ). Together, the two pairs of steps differ by roughly 12o in rolling.
The third step in the A-rich site, T
3
(A/T)
4
, rolls higher than T
3
A
4
, but this might not compensate for the above difference in rolling appreciably,
since, because of the accumulation of helical twisting of the steps, the phase
of the third step (the direction of bending) is not close to that of the first
step.
In general the rolling of a TA step is easier than that of AT or TT/AA (
20
). The methyl groups of the two T bases in TA are separated from each other. In
addition, helical untwisting of the step is coupled with the rolling (
26
), and it moves the sugar-phosphate backbones further away from the methyl groups (Fig.
4
f). Thus no serious steric hindrance is expected for the methyl groups. In an AT
step, however, the T bases tend to stack tightly onto the A bases in the
neighbouring base pairs (Fig.
4
b) and the methyl to sugar-phosphate distances are shorter than those in a TA step (compare Fig.
4
b with c), which makes rolling of AT more difficult. In a TT step the two methyl
groups and the nearby sugar-phosphate group are interlocked with each other and thus any movement is
difficult (Fig.
4
a and d). The above explains the different degree of rolling found at T
1
A
2
and (T/C)
1
T
2
, and that of A
2
T
3
and T
2
T
3
.
Among the three translational distances (rise, slide and shift) the shift
parameter can contribute best for adapting to the curved TBP surface (Fig.
4
g and h). The A
2
T
3
step shifts more (~1.3 Å, Fig.
4
g) than the T
2
T
3
step (~0.5 , Fig.
4
h) in the negative direction. In the A-rich site, shifting of the neighbouring (T/C)
1
T
2
step (by ~0.54 Å) is also observed. However, shifting at a single step (Fig.
4
h) can create a curvature larger than shifting of a similar length divided
between two successive steps (Fig.
4
g).
Shifting of the A
2
T
3
step is accompanied by sliding of the step (Fig.
3
). Negative shifting (`1' in Fig.
4
e) moves the sugar-phosphate backbone towards the nearby methyl group (`2'), and this will
slide the step in the negative direction (`3'). Movement of this type is
possible for a TA step but seems difficult at a TT step since the other methyl
group is blocking the way. The above can explain the differences found in the
shifts of A
2
T
3
and T
2
T
3
.
In summary the differences in the curvature at the two DNA sites are created by
the differences in the rolling capacity of TA and (T/C)T, and of AT and TT, and
the differences in the shifting capacity of AT and TT. The deformation of the
first steps is the largest and determines the overall structure of the DNA. The
differences between the two DNA sites are greater at the second steps and thus
are important for the orientation of TBP. It should be noted in this context
that the three key residues are positioned closest to the second step (Fig.
1
).
SOME OTHER CHARACTERISTICS OF THE DNA
The T
3
A
4
(Rockefeller and Yale) and T
3
A
4
(Yale) steps show positive high sliding. A gap is found between two lines of
the hydrophobic residues (circled in Fig.
1
c) which contact the C2'H atoms of A bases-i.e. Val(aa39, aa119), Val(aa80, aa171), and Leu(aa72, aa163). To
cross over this gap the positive sliding of the dinucleotide steps seem to be
used.
In general, TA is the only step which can slide to a large extent in the
positive direction among the A/T-rich sequences (
20
; see also Fig.
5
d, e and f to understand how positive sliding would clash the methyl group
against the nearby sugar-phosphate backbone at TT and AT). This explains the high sliding found
for the TA steps. Smaller sliding of T
3
T
4
is discussed later in this section.
Figure 5
.
Rolling of a step and propeller twisting of the base pair. (
a
,
c
and
d
) Positive rolling of a dinucleotide step (a) is followed by positive (c) or
negative (d) propeller twisting depending on which the DNA strand the T base is
to avoid the methyl group approaching the neighbouring base pair. (
b
) Propeller twist angles of base pairs calculated for the Yale (Y) and
Rockefeller (R) DNA structures.
An A[middot]T base pair has a tendency to show high propeller twisting (
27
). At the two highest positively rolled steps, T
1
A
2
and (T/C)
1
T
2
, the A
2
[middot]T
7
base pair has high positive propeller twisting, and T
7
[middot]A
2
has negative propeller twisting (Fig.
5
c). Thus independent of whichever strand the T base is on, an A[middot]T base pair propeller twists so that the T base moves away from the
nearby base pair on the major groove (Fig.
5
c and d). As a consequence, the partner A base becomes closer to the nearby base
pair but this is less problematic as the A base is slimmer.
The nucleotide sequences of the TATA box which are found in the real eukaryotic
transcription systems (
28
) and those which are strongly bound by TBP
in vitro
(
13
) are similar as a whole, except for position 5, which is occupied frequently by
an A base
in vivo
but by a T base
in vitro
. Such a discrepancy might not be so surprising, since the
in vitro
sequence is important only for binding but the
in vivo
sequence is important also for fixing the orientation of TBP. Thus a symmetric
and flexible sequence, TATATATA, is a good
in vitro
binding site but is found less frequently
in vivo
.
The DNA sequence in the Yale structure is of the
in vitro
type, TATA
T
in vivo
type, TATA
A
5
A
6
and A
5
A
6
in Fig.
3
). Thus, for transcriptional regulation the less symmetric Rockefeller sequence
seems to be preferable.
The A
4
T
5
and T
5
A
6
steps in the Yale structure have conformations distinctly different from each
other-the former mainly rotates, while the latter slides (Fig.
3
). However, the A
4
A
5
and A
5
A
6
steps in the Rockefeller structure behave in similar ways, probably up to the
limit of the freedom in movement by an AA step. The above arguments can explain
why the TATATAA(G/A) sequence is a better binding site but is less frequently
used
in vivo
.
CONCLUSION
The minor groove side of an A[middot]T base pair is smoother than that of G[middot]C (note N2'H
2
of G). Thus, for making van der Waals contacts with TBP an A/T-rich sequence is appropriate. For fixing the right orientation of TBP, the
two halves of the TATA box are differentiated, one half to the flexible TATA
sequence, and the other to a less flexible A-rich sequence by arranging T bases differently. Positioning of the methyl
groups on the inner surface of the complex is correlated with the positioning
of large/small amino acid residues on the outer surface through the
bendability/curvature of the two molecules. Many characteristics of the TBP-binding sites can be explained consistently by focusing attention on
steric hindrance of the methyl groups of the T bases.
ACKNOWLEDGEMENTS
We thank Dr C. Chothia for stimulative discussion on the curvature of [beta]-sheet. We thank Ms M. Iimura for her help in preparing figures. The
coordinates of the Rockefeller structure were kindly provided by Prof. Burley,
and we thank also the Yale group whose coordinates are already deposited to
Protein Data Bank (29), PDB code 1YTB.
REFERENCES
1 Van Dyke ,M.W. , Roeder,R.G. and Sawadogo,M. (1988 ) Science 241 1335 -1338.MEDLINE Abstract
2 Buratowski ,S. , Hahn,S., Guarente,L. and Sharp,P.A. (1989 ) Cell 56 549 -561.MEDLINE Abstract
3 Hoey ,T. , Dyniacht,B.D., Peterson,M.G., Pugh,B.F. and Tjian,R. (1990 ) Cell 61 1179 -1186.MEDLINE Abstract
4 Sharp ,P.A. (1992 ) Cell 68 819 -821.MEDLINE Abstract
5 White ,R.J. and Jackson,S.P. (1992 ) Trends Genet. 8 284 -288.MEDLINE Abstract
6 Rigby ,P.W.J. (1993 ) Cell 72 7 -10.MEDLINE Abstract
7 Hahn ,S. , Buratowski,S., Sharp,P.A. and Guarente,L. (1989 ) Proc. Natl. Acad. Sci. USA 86 5718 -5722.MEDLINE Abstract
8 Yamamoto ,T. , Horikoshi,M., Wang,J., Hasegawa,S., Weil,P.A. and Roeder,R.G. (1992 ) Proc. Natl. Acad. USA 89 2844 -2848.
9 Reddy ,P. and Hahn,S. (1991 ) Cell 65 349 -357.MEDLINE Abstract
10 Kim ,Y. , Geiger,J.H., Hahn,S. and Sigler,P.B. (1993) Nature , 365
11 Kim ,J.L. , Nikolov,D.B. and Burley,S.K. (1993 ) Nature 365 520 -527.MEDLINE Abstract
12 Klug ,A. (1993 ) Nature 365 486 -487.MEDLINE Abstract
13 Wong ,J.M. and Bateman,E. (1994 ) Nucleic Acids Res. 22 1890 -1896.MEDLINE Abstract
14 Klug ,A. , Jack,A., Viswamitra,M.A., Kennard,O., Shakked,Z. and Steitz,T.A. (1979 ) J. Mol. Biol. 131 669 -680.MEDLINE Abstract
15 Nelson ,H.C. , Finch,J.T., Luisi,B.F. and Klug,A. (1987 ) Nature 330 221 -226.MEDLINE Abstract
16 Goodsell ,D.S. , Kaczor-Grzeskowiak,M. and Dickerson,R.E. (1994 ) J. Mol. Biol. , 234 79 -96.
17 Chothia ,C. (1984 ) Annu. Rev. Biochem. 53 537 -572.MEDLINE Abstract
18 Janin ,J. and Chothia,C. (1980 ) J. Mol. Biol. 143 95 -128.MEDLINE Abstract
19 Hunter ,C.A. (1993 ) J. Mol. Biol. 230 1025 -1054.MEDLINE Abstract
20 Suzuki ,M. , Yagi,N. and Finch,J.T. (1996 ) FEBS Lett. 379 148 -152.
21 Kim ,J.L. and Burley,S.K. (1994 ) Nature Struct. Biol. 1 638 -653.
22 Nikolov ,D.B. , Hu,S.-H., Lin,J., Gasch,A., Hoffman,A., Horikoshi,M., Chua,N.-H., Roeder,R.G. and Burley,S.K. (1992 ) Nature 360 40 -46.MEDLINE Abstract
23 Chasman ,D.I. , Flaherty,K.M., Sharp,P.A. and Kornberg,R.D. (1993 ) Proc. Natl. Acad. Sci. USA 90 8174 -8178.MEDLINE Abstract
24 Dickerson ,R.E. , Bansal,M., Calladine,C.R., Diekmann,S., Hunter,W.N. and Kennard,O. (1989 ) EMBO J. 8 1 -4.MEDLINE Abstract
25 Suzuki ,M. and Yagi,N. (1995 ) Nucleic Acids Res. 23 2083 -2091.MEDLINE Abstract
26 Suzuki ,M. , Yagi,N. and Gerstein,M. (1995 ) Protein Engng , 8 329 -338.
27 Levitt ,M. (1978 ) Proc. Natl. Acad. Sci. USA 75 640 -644.MEDLINE Abstract
28 Prestridge ,D.S. (1995 ) J. Mol. Biol 249 923 -932.MEDLINE Abstract
29 Bernstein ,F.C. , Koetzle,T.F., Williams,G.J.B., Meyer,E.F. Jr, Brice,M.D., Rodgers,J.R., Kennard,O., Shimanouchi,T. and Tasumi,M. (1977 ) J. Mol. Biol. 112 525 -542.
30 Suzuki ,M. (1993 ) Structure 2 317 -326.MEDLINE Abstract
31 Kraulis ,P.J. (1991 ) J. Appl. Crystallogr 24 946 -950.
32 Babcock ,M.S. , Pednault,E.P.D. and Olson,W.K. (1993 ) J. Biomol. Struct. Dynamics 11 597 -628.
33 Babcock ,M.S. , Pednault,E.P.D. and Olson,W.K. (1994 ) J. Mol. Biol. 237 125 -156.MEDLINE Abstract
Return
*
To whom correspondence should be addressed
