Báo cáo khoa học: Adaptability and flexibility of HIV-1 protease potx

Adaptability and flexibility of HIV-1 protease
Mukesh Kumar and Madhusoodan V. Hosur
Solid State Physics Division, Bhabha Atomic Research Centre, Mumbai, India
Even though more than 200 three-dimensional structures
of HIV-1 protease complexed to a variety of inhibitors
are available in the Protein Data Bank; very few struc-
tures of unliganded protein have been determined. We
have recently solved structures of unliganded HIV-1
protease tethered dimer mutants to resolutions of 1.9 A
˚
and 2.1 A
˚
, and have found that the flaps assume closed-
flap conformation even in the absence of any bound lig-
and. We report comparison of the unliganded closed-flap
structure with structures of HIV-1 protease inhibitor
complexes with a view to accurately identifying structural
changes that the ligand can induce on binding to HIV-1
protease in the crystal. These studies reveal that the least
flexible region present in the active site of HIV-1 protease
need not also be the least adaptable to external stress,
thus highlighting the conceptual difference between flexi-
bility and adaptability of proteins in general.
Keywords: adaptability; flexibility; HIV-1 protease; inhibi-
tors; structure.
The significance of human immunodeficiency virus type 1
(HIV-1) protease in the life cycle of HIV has made it a prime
therapeutic target for the development of anti-HIV drugs.
This has resulted in the determination of a large number of
structures of identical or closely related sequences with
different ligands by X-ray, NMR and theoretical molecular

modelling approaches. To date a total of 213 such structures
are available from the Protein Data Bank (PDB) [1] and
HIV-protease database ( />In spite of such a large number of studies on a single system
consisting of closely related sequences, we still lack the
proper understanding to tackle the problem of drug
resistance mutations, a typical characteristic associated with
HIV infection. This inability to understand the behaviour of
the protease demands the development of more sophisti-
cated tools for the study of protein structure. There is an
inevitable need to improve our knowledge of the inherent
flexibility and adaptability in the three-dimensional struc-
ture of proteins. These two features of proteins are as
important as the structure itself and are very often
responsible for the functional characteristics of a particular
structure under different ‘natural’ and ‘stressed’ environ-
ments. As reported by Zoete et al.[2],X-raystructures
through the B-factors, molecular dynamics simulations and
normal mode analyses give a fair idea about the fluctuations
of different residues/regions of proteins about their mean
positions. This mobility may be described as the flexibility of
the region. In contrast, adaptability of a residue is the ability
of that residue to alter its mean position itself in response to
changes in its chemical environment. Information about
residue adaptabilities can be obtained only by detailed
comparison of the protein structures in the presence and
absence of the environmental stress, which may be in the
form of a point mutation or a bound ligand. We determined
earlier the structure of the double mutant C95M/C1095A in
an unliganded state. By comparison with the isomorphous
structure of unliganded C95M single mutant, we analysed

[3] the effect of C95A mutation on the structure of the
protease itself. In addition to the changes on the dimer
interface, we observed that the catalytic aspartates Asp25/
1025 and the catalytic water move to make this catalytic site
more accessible to the substrate (Fig. 1).
This structural change in the active site was subsequently
shown to be associated with increased autolysis rates in
protease carrying the C95A mutation [3]. To analyse further
the behaviour of these catalytic aspartates, we have
extended our comparison to structures of C95A mutant
protease complexed with different inhibitors. We find that
the polypeptide chain segment 23–26, which includes the
catalytic aspartates, is structurally most adaptable even
though it is least flexible. This observation is significant and
can contribute to the fact that HIV-1 protease enzyme has
the unique property of cleaving protein substrates at eight
different amino acid sequences [4].
Materials and methods
‘Closed-flap’ structure of unliganded HIV-1 protease
The X-ray structures of C95M single mutant- and C95M/
C1095A double mutant-tethered HIV-1 protease have been
solved to 1.9 A
˚
and 2.1 A
˚
resolution, respectively, and
deposited in the PDB (files 1G6L and 1LV1). Residues in
the first monomer have been numbered 1–99 and those in
the second monomer 1001–1099. Residues in the linker are
numbered 101–105. Electron density was not visible for the

linker peptide suggesting that the linker was highly flexible.
Correspondence to M. V. Hosur, Solid State Physics Division,
Bhabha Atomic Research Centre, Mumbai, India-400085.
Fax: +91 22 5515050, Tel.: +91 22 5593614,
E-mail:
Abbreviations: HIV-1, human immunodeficiency virus type 1;
Rmsd, root mean square deviation.
Enzyme: HIV-1 protease (EC: 3.4.23.16).
(Received 2 August 2002, revised 14 November 2002,
accepted 27 January 2003)
Eur. J. Biochem. 270, 1231–1239 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03483.x
These two structures revealed for the first time the closed-
flap conformation of HIV-1 protease even in the absence of
any ligand bound in the active site [3,5]. Comparison of such
a structure with the closed-flap ligand-bound structure is
expected to be more rational and useful.
Inhibitor complex structures containing C95A HIV-1
protease
Eight structures deposited in the PDB (files 1DAZ, 1A94,
1HWR, 1HVR, 1QBS, 1DMP, 1QBR and 1QBT) are of
complexes between different types of inhibitor molecules
and HIV-1 protease containing the C95A mutation. In two
of these structures (1DAZ and 1A94), the protein is
complexed with a peptide inhibitor [6,7] and in the
remaining six (1HWR, 1HVR, 1QBS, 1DMP, 1QBR and
1QBT) the enzyme is complexed with a cyclic urea-based
inhibitor [8–13]. All the cyclic urea-based inhibitors contain
a central seven-membered ring having a urea moiety, and a
diol group. P1/P1¢ substituents are attached to C3/C6 atoms
of the central ring and P2/P2¢ substituents are attached to

urea nitrogen atoms of the ring as shown in Fig. 2A. The
seven membered ring sits in the active site cavity as a bridge
between the flaps and the catalytic aspartates. While the
urea moiety makes hydrogen bonds with main chain NH of
the flap residue Ile50 from both the monomers, the diol
group makes hydrogen bonds with the two catalytic residues
Asp25 and Asp1025. The P2 moities are all different among
the six inhibitors listed in Fig. 2B.
All of the superposition and structure comparisons were
performed using software
O
[14] and molecular images
were made using
MOLRAY
[15]. All structure alignments
were based on superposition of Ca atoms of all residues of
the entire HIV-1 protease dimer. With the whole structures
so superposed, the root mean square deviations (Rmsd
values) quoted for any polypeptide segment is then calcu-
lated for the Ca atoms of the amino acid residues in that
polypeptide segment. Error-scaled difference distance mat-
rix plots were generated using the program
ESCET
[16]. The
errors in atomic positions for error-scaling were estimated
using ‘diffraction-component precision index’ plus linear
B-scaling [17].
Results
Structural features of C95A mutation in HIV-1 protease
Fig. 3 shows a B-factor vs. residue number plot for C95M/

C1095A HIV-1 protease. It may be seen in the figure that
the residues 23–26 in the first monomer and 1023–1026 in
the second monomer located in the active site of protease
have the lowest B-factors suggesting that these residues are
relatively ‘rigid’ entities in the protein structure. The electron
density in this region is also very well defined indicating that
these residues have very well-defined conformation. Karplus
and his colleagues have drawn a similar conclusion recently,
after analysing 73 crystal structures of HIV-1 protease
complexed to a variety of inhibitors. Comparison of the
C95M/C1095A double mutant with the C95M single
mutant HIV-1 protease structure by superposition of all
198 Ca atom pairs showed that the main chain of the
residues 23–26/1023–1026 has moved towards the flap after
C1095A mutation. Interestingly, this movement includes
movement of catalytic aspartates Asp25/1025 along with the
bound catalytic water (Fig. 1) in a direction that would
enable more access to the catalytic centre for the incoming
substrate. Eventually this resulted in increased autolysis
after C1095A mutation [3]. Therefore, this subtle movement
in the catalytic centre is of prime importance from the point
of view of enzymatic activity. These atomic movements were
also verified by means of difference distance matrix plots,
which have the advantage of not depending on any
particular frame of superposition. These plots are shown
in Fig. 4 for one subunit. In Fig. 4A the upper right triangle
shows the normal difference distance matrix plot while the
lower left triangle shows error-scaled difference distance
matrix plot for the structures 1LV1 vs. 1G6L. It may be seen
that the region 23–26 is involved in significant atomic

Fig. 1. Superposition of catalytic aspartates of C95M HIV-1 protease (shown by green carbons) and C95M/C1095A HIV-1 protease (shown by yellow
carbons).
1232 M. Kumar and M. V. Hosur (Eur. J. Biochem. 270) Ó FEBS 2003
movement. The error-scaled difference distance matrix plot
also shows that this segment in 1LV1 has moved towards
the flaps and along the direction joining the active site
aspartates to the tips of the flaps.
Structural effect of inhibitor binding on HIV-1 protease
The closed-flap structure of the unliganded C1095A
mutant enables, through structural comparisons, accurate
Fig. 2. Basic structure of P2 analogues of cyclic urea inhibitors (A) and properties and structures of all inhibitors (B).
Ó FEBS 2003 Adaptability and flexibility of HIV-1 protease (Eur. J. Biochem. 270) 1233
identification of the changes, if any, brought about by
inhibitor binding. The unliganded and all six of the urea
inhibitor complex structures belong to the same space group
(Fig. 2B) further enhancing the meaningfulness of these
comparisons. Table 1 shows the Rmsd values for pair-wise
superposition of the unliganded C95M/C1095A structure
with each of the six cyclic urea-complexed structures listed
in Fig. 2B.
Fig. 3. Average B-factors for different residues
of C95M/C1095A HIV-1 protease.
Fig. 4. Difference distance matrix plot. The upper right triangle shows the normal difference distance matrix plot and the lower left triangle shows
the error-scaled difference distance matrix plot. (A) 1LV1 vs. 1G6L. (A) 1LV1 vs. 1QBS.
1234 M. Kumar and M. V. Hosur (Eur. J. Biochem. 270) Ó FEBS 2003
The corresponding Rmsd for superposition onto the
1DAZ structure, which belongs to the orthorhombic crystal
system, is 0.45 A
˚
. Thus, in spite of different crystal

environments, the overall conformation of HIV-1 protease
is very similar, with Rmsd values ranging from 0.26
to 0.45 A
˚
. However, there are subtle differences character-
istic of the bound ligand. Fig. 5 shows the superposition of
peptide complex structure 1DAZ with unliganded structure
1LV1. Interestingly, the catalytic aspartates Asp25/1025 in
the two structures (1LV1 and 1DAZ) are very much super
imposable, and the Rmsd for residues 23–26/1023–1026 is
0.11 A
˚
. Similar is the case with another peptide–inhibitor
complex (PDB code 1A94) suggesting that the linear chain
of peptide inhibitors does not cause much alteration in the
position of catalytic aspartates.
However, when the cyclic urea complex structures are
superimposed on to the unliganded structure 1LV1, the
position of the main chain of the catalytic residues
Asp25/1025 is significantly different (see Fig. 7). The shift
of the Ca atoms of Asp25/1025 ) when averaged over
all six comparisons ) is 0.37 A
˚
, and has apparently been
induced by the inhibitor to relieve what would otherwise
be bad steric contacts (2.7 A
˚
and 2.81 A
˚
)betweenthe

cyclic urea ring of the inhibitor and the aspartate side
chains, as shown in Fig. 6. This movement of catalytic
Asp25 is also revealed by the error-scaled difference
distance matrix plot (Fig. 4B) for comparison of 1LV1
and 1QBS structures. The residues 23–26 in 1QBS have
moved away from the flap-tips, and to the direction
joining active-site aspartates to flap-tips, in agreement
with results of molecular superposition (Fig. 7). It thus
appears that the position of polypeptide chain segment
23–26/1023–1026 is influenced by both mutational stress
C95A (Fig. 1) below the active site cavity and also by
steric stress from inhibitor binding in the active site
cavity (Fig. 7). Thus, of utmost interest is the fact that
what is considered the most rigid polypeptide segment in
the active site of HIV-1 protease can alter its position
when required. Structural rearrangement in the form of
movements of the side-chain atoms in the S1 pocket on
Table 1. Rmsd values for pair-wise superposition of unliganded C95M/C1095A structure with each of the six cyclic urea-complexed structures shown in
Fig. 2B.
1LV1 1HWR 1DMP 1QBS 1HVR 1QBR 1QBT
1LV1 0.00 0.39 0.37 0.34 0.32 0.35 0.35
1HWR 0.00 0.38 0.39 0.41 0.41 0.41
1DMP 0.00 0.32 0.36 0.33 0.34
1QBS 0.00 0.31 0.33 0.34
1HVR 0.00 0.37 0.37
1QBR 0.00 0.26
1QBT 0.00
Fig. 5. Superposition of peptide inhibitor-bound C95A HIV-1 protease in PDB file 1DAZ (shown by green carbons) on unliganded HIV-1 protease
structure in PDB file 1LV1 (shown by yellow carbons). The water molecule is present only in the unliganded structure.
Ó FEBS 2003 Adaptability and flexibility of HIV-1 protease (Eur. J. Biochem. 270) 1235

inhibitor binding has been recently observed crystallo-
graphically [18]. However, the movements of the main
chain containing catalytic aspartates Asp25/1025 of the
active site has been reported for the first time here.
Thus the main chain containing these catalytic aspartates,
although having small B-factors, is very much adaptable to
the external stress induced by the cyclic urea scaffold. This
observation suggests that it is important to distinguish such
adaptability from the concept of general flexibility arising
out of multiple conformations of the side chains and
thermal motions of the atoms therein.
In the six cyclic urea complex only the substituents at the
P2/P2¢ site are different as shown in Fig. 2B. While changes
induced by cyclic urea inhibitors in the S1/S1¢ pocket are
very similar (Fig. 7), changes induced in the S2/S2¢ pocket
depend on the substituents at the P2/P2¢ site. These
substituents point directly towards the polypeptide segment
27–32 of the S2/S2¢ subsite. The allyl substituent at this site
in the case of XK216 is small and therefore there is not
much change in the conformation of these residues (Fig. 8).
However, when a larger hydrophobic moiety like the
naphthyl group, as in case of XK263, is present at the P2/P2¢
Fig. 6. Superposition of cyclic urea bound HIV-1 protease in PDB file 1HWR (shown by green carbons) on unliganded HIV-1 protease (shown by
yellow carbons). The relieved steric contacts are marked.
Fig. 7. Superposition of cyclic urea inhibitor-bound HIV-1 protease (shown by green carbons) on the unliganded structure (shown by yellow carbons).
1236 M. Kumar and M. V. Hosur (Eur. J. Biochem. 270) Ó FEBS 2003
site, the main chain as well as the side chain of the
hydrophillic residue Asp30 moves away as shown in Fig. 9.
Further, when a hydrophilic group such as NH
2

replaces
one of the benzene rings of the naphthyl moiety above, as in
the case of DMP450, the aspartic acid residue Asp30 moves
toward the inhibitor and makes a hydrogen bond with this
new substituent (Fig. 10).
More bulky groups at this site, as in SD146, causes steric
hindrance near Asp30 inducing the side chain of Asp30 to
move away (Fig. 11).
Discussion and conclusion
Structure alignments are very often used to derive functional
information on proteins. Distant evolutionary relationships
are also detected exclusively by structure alignments.
Structural superpositions using complete proteins or
domains thereof, are a powerful method of identifying
ligand- or mutation-induced conformational changes. The
inferences drawn from such comparisons are reliable
especially when the structures superimposed are variants
of the same protein crystallizing in isomorphous space
groups, as is the case here. Comparison of unliganded and
ligand-bound HIV-1 protease is important for obtaining
information about adaptability of residues in the active site.
The flaps in unliganded HIV-1 protease assume two very
different conformations, open and closed, whereas the flaps
always assume a closed conformation in ligand-bound
structures. When ligand-bound and open-flap unliganded
HIV-1 protease structures are superimposed the Rmsd is
greater than 1.0 A
˚
, with changes being distributed through-
out the protein. This might suggest that concerted changes

distributed throughout the protein are needed to bring the
flaps into a closed conformation. Therefore, comparison of
closed-flap unliganded structure with ligand-bound struc-
tures would accurately reflect changes induced by ligand
binding alone. The uliganded structure we have determined
is of closed-flap conformation. Comparison of the unligan-
ded C95A mutant structure of HIV-1 protease with that
complexed to cyclic urea-based inhibitors reveals that the
residues which are known to be less flexible and are
characterized by a small B-factor are not necessarily less
Fig. 8. Superposition of XK216 bound HIV-1 protease (shown by green carbons) on the unliganded protease structure (shown by yellow carbons).
Fig. 9. Superposition of XK263 bound HIV-1 protease (shown by green carbons) on the unliganded protease structure (shown by yellow carbons).
Ó FEBS 2003 Adaptability and flexibility of HIV-1 protease (Eur. J. Biochem. 270) 1237
adaptable to environmental changes. Residues 23–26 in
both monomers of HIV-1 protease are considered to be very
‘robust’ and ‘rigid’ [2]. The electron density of these
segments is also very well defined and they have a small
B-factor, implying less flexibility of this segment. However
the present study shows that these residues are very
adaptable to internal or external stresses. This property
may have been built into HIV-1 protease to cater to the
functional requirement that the enzyme cleave substrates of
eight different sequences [4]. Cys95 at the dimer interface is
far away from the active site. Also the B-factors of residues
91–99 are significantly more than those of segment 23–26
(Fig. 3). Even so the C95A mutation does not cause much
change in the structure near the mutation site [3]. Instead it
affects the catalytic site residues 23–26 thereby suggesting
that the residues 23–26 are more adaptable than residues
around 95. Hence it is important to distinguish this type of

adaptability from the concept of flexibility. Zoete et al.[2]
have analysed the X-ray structures of 73 HIV-1 protease
complexes to look at the overall structural variability of
protease. Interestingly, they found that the pattern of
structural variability of different residues in all of these
structures is very much the same as the pattern of B-factors
of these residues in any single structure. Thus the analysis of
an ensemble of structures gives information about the
inherent flexibility of a protein. However, the information
about adaptability is lost. Only analysing and comparing
structures on a one to one basis would provide information
about adaptability.
Here in this paper we are trying to distinguish between the
concept of flexibility and that of adaptability in protein
structures. Very often, these two terms are used inter-
changeably in the literature. However with the growing
amount of structural information available about proteins,
there arises a need to analyse and understand the structural
features in a more comprehensive and accurate way. One
must have a well-defined tool to understand different
properties of protein structure. We thus feel that the word
‘flexibility’ with reference to protein structure should be
reserved to describe the conformational variability of a
Fig. 11. Superposition of SD146 bound HIV-1 protease (shown by green carbons) on the unliganded protease structure (shown by yellow carbons).
Fig. 10. Superposition of DMP450 bound HIV-1 protease (shown by green carbons) on the unliganded protease structure (shown by yellow carbons).
1238 M. Kumar and M. V. Hosur (Eur. J. Biochem. 270) Ó FEBS 2003
residue as well as to include the effects of thermal motions of
atoms therein. On the other hand, the word ‘adaptability’
should be used to describe the ability of a residue/region to
adjust and accommodate itself in response to a stress/

change in its environment. This stress could be either
internal, as in case of mutation of a nearby residue, or
external, as in case of presence of an inhibitor or other
nonprotein molecule in its surrounding. This distinction is
important: we see in the present analysis that a less flexible
region need not necessarily be less adaptable.
Acknowledgements
We thank the National Facility for Macromolecular Crystallography,
BARC for providing all the X-ray and biochemistry equipment used in
this investigation. We are thankful to S. K. Sikka for encouragement
and support. We thank K. K. Kannan, B. Pillai and V. Prashar for
scientific discussions and S.R. Jadhav for technical help.
References
1. Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N.,
Weissig,H.,Shindyalov,I.N.&Bourne,P.E.(2000)TheProtein
Data Bank. Nucl. Acids Res. 28, 235–242.
2. Zoete, V., Michielin, O. & Karplus, M. (2002) Relation between
sequence and structure of HIV-1 protease inhibitor complexes: a
model system for the analysis of protein flexibility. J. Mol. Biol.
315, 21–52.
3. Kumar, M., Kannan, K.K., Hosur, M.V., Bhavesh, N.S., Chat-
terjee, A., Mittal, R. & Hosur, R.V. (2002) Effects of remote
mutation on the autolysis of HIV-1 PR: X-ray and NMR
investigations. Biochem. Biophys. Res. Commun 294, 395–401.
4. Henderson, L.E., Copeland, T.D., Sowder, R.C., Scultz, A.M. &
Oroszlan, S. (1988) In Human Retrovirus, Cancer and AIDS:
Approaches to Prevention and Therapy, pp. 135–147. New York:
Liss.
5. Pillai, B., Kannan, K.K. & Hosur, M.V. (2001) 1.9 A
˚

x-ray study
shows closed flap conformation in crystals of tethered HIV-1 PR.
Proteins 43, 57–64.
6. Mahalingam,B.,Louis,J.M.,Reed,C.C.,Adomat,J.M.,Krouse,
J., Wang, Y.F., Harrison, R.W. & Weber, I.T. (1999) Structural
and kinetic analysis of drug resistant mutants of HIV-1 protease.
Eur. J. Biochem. 263, 238–245.
7. Wu, J., Adomat, J.M., Ridky, T.W., Louis, J.M., Leis, J., Harri-
son, R.W. & Weber, I.T. (1998) Structural basis for specificity of
retroviral proteases. Biochemistry 37, 4518–4526.
8. Lam, P.Y.S., Jadhav, P.K., Eyermann, C.J., Hodge, C.N., Ru, Y.,
Bacheler, L.T., Meek, J.L., Otto, M.J., Rayner, M.M., Wong,
Y.N., Chang, C.H., Weber, P.C., Jackson, D.A., Sharpe, T.R. &
Erickson-Viitanen, S. (1994) Rational design of potent,
bio-available, non-peptide cyclic ureas as HIV-1 protease
inhibitors. Science 263, 380–384.
9. Lam, P.Y., Ru, Y., Jadhav, P.K., Aldrich, P.E., DeLucca, G.V.,
Eyermann, C.J., Chang, C.H., Emmett, G., Holler, E.R., Dane-
ker, W.F., Li, L., Confalone, P.N., McHugh, R.J., Han, Q., Li, R.,
Markwalder, J.A., Seitz, S.P., Sharpe, T.R., Bacheler, L.T., Ray-
ner, M.M., Klabe, R.M., Shum, L., Winslow, D.L., Kornhauser,
D.M.,Hodge,C.N.et al. (1996) Cyclic HIV protease inhibitors:
synthesis, conformational analysis, P2/P2¢ structure-activity rela-
tionship, and molecular recognition of cyclic ureas. J. Med. Chem.
39, 3514–3525.
10. Hodge, C.N., Aldrich, P.E., Bacheler, L.T., Chang, C.H., Eyer-
mann, C.J., Garber, S., Grubb, M., Jackson, D.A., Jadhav, P.K.,
Korant, B., Lam, P.Y., Maurin, M.B., Meek, J.L., Otto, M.J.,
Rayner, M.M., Reid, C., Sharpe, T.R., Shum, L., Winslow, D.L.
& Erickson-Viitanen, S. (1996) Improved cyclic urea inhibitors of

the HIV-1 protease: synthesis, potency, resistance profile, human
pharmacokinetics and X-ray crystal structure of DMP 450. Chem.
Biol. 3, 301–314.
11. Yamazaki, T., Hinck, A.P., Wang, Y X., Nicholson, L.K.,
Torchia, D.A., Wingfield, P., Stahl, S.J., Kaufman, J.D., Chang,
C H., Domaille, P.J. & Lam, P.Y.S. (1996) Three dimensional
solution structure of the HIV-1 protease complexed with DMP323,
a novel cyclic urea typw inhibitor, determined by nuclear magnetic
resonance spectroscopy. Protein Sci. 5, 495–506.
12. Ala,P.J.,Huston,E.E.,Klabe,R.M.,Jadhav,P.K.,Lam,P.Y.&
Chang, C.H. (1998) Counteracting HIV-1 protease drug
resistance: structural analysis of mutant proteases complexed with
XV638 and SD146, cyclic urea amides with broad specificities.
Biochemistry 37, 15042–15049.
13. Ala, P.J., DeLoskey, R.J., Huston, E.E., Jadhav, P.K., Lam,
P.Y.S., Eyermann, C.J., Hodge, C.N., Schadt, M.C., Lewan-
dowski, F.A., Weber, P.C., McCabe, D.D., Duke, J.L. & Chang,
C.H. (1998) Molecular recognition of cyclic urea HIV-1 protease
inhibitors. J. Biol. Chem. 273, 12325–12331.
14. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991)
Improved methods for building protein models in electron density
maps and the location of errors in these models. Acta Crystallogr.
A47, 110–119.
15. Harris,M.&Jones,T.A.(2001)Molray–awebinterfacebetween
O and the POV-Ray ray tracer. Acta Crystallogr. D57, 1201–1203.
16. Schneider, T.R. (2002) A genetic algorithm for the identification of
conformationally invariant regions in protein molecules. Acta
Crystallogr. D58, 195–208.
17. Schneider, T.R. (2000) Objective comparison of protein structures:
error-scaled difference distance matrices. Acta Crystallogr. D56,

714–721.
18. King, N.M., Melnick, L., Prabu-Jeyabalan, M., Nalivaika, E.A.,
Yang,S.S.,Gao,Y.,Nie,X.,Zepp,C.,Heefner,D.L.&Schiffer,
C.A. (2002) Lack of synergy for inhibitors targeting a multi-drug-
resistant HIV-1 protease. Protein Sci. 11, 418–429.
Ó FEBS 2003 Adaptability and flexibility of HIV-1 protease (Eur. J. Biochem. 270) 1239