LETTERS
The site-specific incorporation of p-iodo-L-phenylalanine
© 2004 Nature Publishing Group http://www.nature.com/naturebiotechnology
into proteins for structure determination
Jianming Xie1, Lei Wang1, Ning Wu1, Ansgar Brock2, Glen Spraggon2 & Peter G Schultz1,2
A recently developed method makes it possible to genetically
encode unnatural amino acids with diverse physical, chemical
or biological properties in Escherichia coli1 and yeast2. We
now show that this technology can be used to efficiently and
site-specifically incorporate p-iodo-L-phenylalanine (iodoPhe)
into proteins in response to an amber TAG codon. The
selective introduction of the anomalously scattering iodine
atom into proteins should facilitate single-wavelength
anomalous dispersion3,4 experiments on in-house X-ray
sources. To illustrate this, we generated a Phe153-iodoPhe
mutant of bacteriophage T4 lysozyme and determined its
crystal structure using considerably less data than are needed
for the equivalent experiment with cysteine and methionine.
The iodoPhe residue, although present in the hydrophobic core
of the protein, did not perturb the protein structure in any
meaningful way. The ability to selectively introduce this and
other heavy atom–containing amino acids into proteins should
facilitate the structural study of proteins.
The anomalous signal (df ¢¢) of iodine at the CuKa wavelength
typically used with in-house generators (1.5418 Å) is 6.85 e–, twelve
times that of sulfur (0.56 e–)5. Therefore, the selective introduction of
an iodine atom into proteins should reduce the high data redundancy
and high solvent content necessary for sulfur single-wavelength
anomalous dispersion (SAD) phasing6. The genetic incorporation of
iodoPhe offers a number of advantages over current approaches
for introducing heavy atoms for SAD phasing. One such method
involves substituting bound water molecules at the surface of the
protein with either halides or metal ions by soaking the crystal in a
solution containing the relevant ions7,8. Unfortunately this approach
can produce a number of low-occupancy sites whose positions must
be determined before phases can be derived. Another method, tell-
uromethionine incorporation9,10, provides a measurable anomalous
signal at the CuKa wavelength (df ¢¢ ¼ 6.4 e–), but is limited by the
extreme sensitivity of telluromethionine to oxidation, toxicity to the
host organism and the difficulty in achieving quantitative replacement
of methionine with telluromethionine. In contrast, by genetically
encoding iodoPhe with a unique codon, tRNA and aminoacyl-tRNA
synthetase, this amino acid can be quantitatively and efficiently
incorporated at any desired site in a protein. Furthermore, the
1Department of Chemistry and Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, USA.
2Genomics Institute of the Novartis Research Foundation, 10675 John Jay Hopkins Drive, San Diego, California 92121, USA. Correspondence should be addressed to
P.G.S. (schultz@scripps.edu) or G.S. (spraggon@gnf.org).
Published online 19 September 2004; doi:10.1038/nbt1013
NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 10 OCTOBER 2004
substitution of large hydrophobic residues with iodoPhe, at either
surface or internal sites, is likely to cause minimal perturbation of the
protein structure.
To selectively incorporate iodoPhe (Fig. 1a) into proteins in
response to an amber codon, we used a previously generated Metha-
Tyr
nococcus jannaschii tRNACUA-tyrosyl-tRNA synthetase (TyrRS) pair.
It has been shown that this tRNA-synthetase pair is orthogonal to all
tRNA-synthetase pairs in E. coli1; that is, neither the M. jannaschii
tyrosyl tRNA nor the synthetase crossreacts with any of the endoge-
neous tRNA or synthetases in E. coli. This condition ensures that
iodoPhe is incorporated into proteins with high translational fidelity.
To alter the specificity of the TyrRS synthetase to selectively recognize
iodoPhe, we generated a library of M. jannaschii TyrRS mutants by
randomizing five residues (Tyr32, Glu107, Asp158, Ile159 and Leu162)
in the tyrosine binding pocket of TyrRS (Fig. 1b)1; we based our
choice of residues on the crystal structure of the homologous Bacillus
stearothermophilus TyrRS–tyrosyl adenylate complex11. The mutant
TyrRS library was then subjected to alternating positive and negative
selections as described12. The positive selection was based on the
suppression of an amber mutation at a permissive site (Asp112) in the
gene encoding type I chloramphenicol acetyltransferase in the pre-
sence of iodoPhe and the orthogonal M. jannaschii tyrosyl tRNA-
synthetase pair. The negative selection was based on the suppression of
amber mutations at permissive sites (Gln2, Asp44 and Gly65) in the
gene encoding toxic barnase in the absence of iodoPhe. Only synthe-
tases that efficiently incorporate iodoPhe, and no endogeneous amino
acid, in response to the amber codon can survive both selections.
After five rounds of alternating positive and negative selections, we
identified one synthetase variant that survives in 120 mg/ml of
chloramphenicol in the presence of iodoPhe, but dies in 20 mg/ml
of chloramphenicol in the absence of iodoPhe. When the mutant
synthetase was used to suppress a Tyr7-TAG mutant of the
Z-domain protein (with a C-terminal His6 tag)12 in the presence of
iodoPhe, full-length protein was produced (Fig. 1c). In the absence of
either iodoPhe or the mutant synthetase, no Z-domain protein was
detected by silver staining on an SDS-PAGE gel (Fig. 1c). The selective
incorporation of iodoPhe was further verified by electrospray ioniza-
tion Fourier transform–ion cyclotron resonance mass spectrometry
(FT-ICR MS). For iodoPhe-substituted Z-domain protein lacking the
first methionine, the calculated and the experimental monoisotopic
1297
 LETTERS

Figure 1 Evolution of a mutant M. jannaschii

TyrRS that specifically incorporates iodoPhe into a b
Leu180 (Leu162)
c
proteins in response to an amber stop codon in Phe177 (Ile159)
E. coli. (a) Structure of p-iodo-L-phenylalanine Asp176 (Asp158)
Asn123 (Glu107)
(iodoPhe). (b) Genetic selections were carried Tyr34 (Tyr32)
out with a library of M. jannaschii tyrosyl-tRNA
synthetase mutants in which five residues (in
parentheses, that is, Tyr32, Glu107, Asp158,
Ile159 and Leu162) were randomized. Residues H2N COOH
were selected based on observed contacts p -iodo-L-phenylalanine
© 2004 Nature Publishing Group http://www.nature.com/naturebiotechnology

between the homologous B. stearothermophilus

TyrRS residues (Tyr34, Asn123, Asp176, Phe177 and Leu180) and tyrosyl adenylate in the crystal structure of the B. stearothermophilus TyrRS-tyrosyl
adenylate complex. (c) Verification of iodoPhe incorporation in response to amber codon by SDS-PAGE and silver staining of the expressed amber mutant
Z-domain protein. Proteins were purified by Ni2+ affinity chromatography.

masses are 7902.720 Da and 7902.728 Da, respectively. The yield of
iodoPhe-substituted Z-domain protein was 3.8 mg/l in minimal
media. For comparison, the yield of Z-domain protein was 4.6 mg/l
when the wild-type M. jannaschii TyrRS was expressed instead of the
mutant TyrRS (which results in incorporation of tyrosine in the
protein). The iodoPhe-specific synthetase has the following mutations:
Y32L, E107S, D158P, I159L and L162E. The Y32L and D158P
mutations may disrupt the hydrogen bonds with the hydroxyl group
of tyrosine to create a hydrophobic pocket that accommodates
iodoPhe. From these results, we conclude that iodoPhe is incorporated
into proteins in E. coli site specifically and quantitatively. In a related
experiment, another group used a mutant E. coli phenylalanyl-tRNA
synthetase to introduce iodoPhe into proteins in a phenylalanine-
auxotrophic E. coli strain. However, this approach resulted in
only partial substitution of phenylalanine by iodoPhe at all sites in
the proteome13.
To determine the utility of iodoPhe incorporation for SAD phasing
in protein crystallography, we used bacteriophage T4 lysozyme as a
model system. Cys54 and Cys97 in T4 lysozyme were mutated to
threonine and alanine, respectively , to enhance the stability of the
protein, as previously reported14. The codon corresponding to Phe153
of T4 lysozyme, a residue in the hydrophobic core of the large lobe of
T4 lysozyme, was mutated to TAG. The evolved orthogonal synthe-
tase–tRNA pair was then used to suppress the amber codon with
iodoPhe in E. coli. The expressed mutant T4 lysozyme was purified by
cation-exchange chromatography and size-exclusion chromatography.
The yield after purification was 5.7 mg/l in minimal media. Crystals
were grown under the same conditions reported for crystallization of
T4 lysozyme and its mutants14.
The structure of the T4 lysozyme was determined from data
collected on an in-house X-ray generator at the CuKa wavelength
(1.5418 Å). In all, 3601 of data were collected at 1001 K to a maximum

Table 1 Summary of data collection statistics and RESOLVE model auto-building

Unit cell a ¼ 59.67 Å, b ¼ 59.67 Å, c ¼ 97.57 Å, a ¼ 90.01, b ¼ 90.01, g ¼ 1201

Space group P3221

Resolution (Å) 30.0–2.0

Degrees collected (1) 45 60 90 135 180 360

Rmerge 0.051 0.055 0.054 0.059 0.060 0.078
Completeness 0.93 0.97 0.99 0.98 0.99 0.99
Anomalous completeness 0.66 0.80 0.82 0.83 0.86 0.86
Mean redundancy 2.4 3.2 4.4 6.8 9.0 19.4
Determined substructure No No No Yes Yes Yes
Mean FOM – – – 0.36 0.37 0.39
Mean DF 0.034 0.031 0.026 0.037 0.033 0.024
Mean DF/sDF 1.45 1.48 1.50 2.64 2.77 6.25
No. cycles auto-building needed – – – 18 17 7
No. residues built (%) – – – 138 (84) 139 (85) 151 (92)
No. side chains placed (%) – – – 122 (74) 120 (73) 133 (81)

Structure refinement

Rcryst (Rfree) 0.157 (0.207)

r.m.s.d. bonds (Å) 0.012
r.m.s.d. angles (1) 1.258
Mean B-factor (Å2) 17.18
No. protein atoms 1322
No. waters 187
No. hetero-atoms 20

Rmerge ¼ S(S|Ii – oIi4| |/S|Ii|) where Ii is the scaled intensity of the ith measurement, and oIi4 is the mean intensity for that reflection.
Rcryst ¼ S| |Fobs| – |Fcalc| |/S|Fobs| where Fcalc and Fobs are the calculated and observed structure factor amplitudes, respectively.
Rfree ¼ as for Rcryst, but for 4.6% of the total reflections chosen at random and omitted from refinement.

1298 VOLUME 22 NUMBER 10 OCTOBER 2004 NATURE BIOTECHNOLOGY

 LETTERS

resolution of 2.0 Å, with no attempt being

made to collect Friedel mates on the same
a z = 0.3333 to 0.3333 z = 0.3333 to 0.3333
b
45 degrees V 60 degrees V

image or close in time. To establish the

U
U
minimum amount of data necessary for Cl1
structure determination, we split the data M106
M120
Cl2
set into six different oscillation ranges repre- z = 0.3333 to 0.3333 z = 0.3333 to 0.3333
90 degrees V 135 degrees V
senting different values of data redundancy
(Table 1). As would be expected, the sub-

U
stitution of iodoPhe for phenylalanine appre-
© 2004 Nature Publishing Group http://www.nature.com/naturebiotechnology

Cl3
ciably increased the anomalous signal (df ¢¢)
z = 0.3333 to 0.3333 z = 0.3333 to 0.3333
for T4 lysozyme. The average ratio of Bijovet 180 degrees V 360 degrees V
lodoPhe153 2–hydroxyethyl–disulfide
pairs (o|DF|4/oF4) was approximately

U
U
3% for all data sets (Table 1), in close M102
M6
agreement with the 4% calculated using the M1
method of Hendrickson and Teeter3. This is
in contrast to the 0.9% computed for native Figure 2 Patterson maps and difference Fourier maps for the anomalous data sets. (a) First Harker
T4 lysozyme with 1,309 atoms and 7 sulfur section of Patterson maps of the iodoPhe incorporated crystal for 451, 601, 901, 1351, 1801, 3601 of
data collected. The iodine self peaks appear after 901 of data are collected; additional peaks present on
atoms. In addition to an enhanced signal, the the 1801 and 3601 sets are the self peaks from chlorine ions present in the crystal. The graphics were
ability to place a fixed number of iodines produced with FFT and NPO from the CCP4 program package21. Plots are contoured at 3s; subsequent
within the hydrophobic core of a protein contours are plotted in 3s steps. (b) Difference Fourier using phases from lysozyme structure 1L6315
provides a substantial advantage relative to and anomalous difference amplitudes. The iodine site as well as a number of sulfurs and ions can be
other techniques such as iodine soaks7,8, clearly distinguished. The electron density map is contoured at 3s above the mean and was calculated
21 27–29.
where multiple low occupancy sites may with programs from the CCP4 package ; the figure was produced by BobScript and Raster3D
confound substructure solution.
Before phase determination, Patterson maps
and anomalous difference Fourier maps were calculated to character- data needed for an inverse beam experiment. Additional redundancy,
ize the anomalous signal. Self vectors for iodine atoms were clearly and thus anomalous signal, does allow faster convergence and the
evident in Harker sections of the Patterson maps in all data sets with a placement of more residues with the auto-building algorithm of
mean redundancy 44.4 (901) (Fig. 2a). Above a redundancy of 9.0 RESOLVE17, primarily because of the generation of better starting
(1801 of data), self peaks for additional chloride ions become evident phases (Table 1).
at the 3s level, but no sulfur vectors were ever visible. The anomalous The structure of the iodinated T4 lysozyme conforms to the
difference Fourier maps calculated from phases derived from a T4 canonical viral lysozyme structure15,18 (Fig. 3a), a bi-lobal mixed
lysozyme structure (Protein Data Bank (PDB) code 1L6315) (Fig. 2) a/b fold. Lobe 1 of the structure is entirely a-helical, whereas lobe 2
showed electron density for all data sets, not only for the iodine but consists of an antiparallel b-sheet inserted between three a-helices.
also for all five methionine sulfurs present in the protein, three The only difference between the native and modified structure is the
putative chlorine ions and a 2-hydroxyethyl disulfide molecule incorporation of the iodoPhe residue at position 153 in the center of
included in the crystallization conditions (Fig. 2b). lobe 1 (Fig. 3a). As would be expected for a residue within the
When SOLVE16 was used to determine the iodine position and hydrophobic core, the iodoPhe is completely buried and makes no
calculate the initial phases, only data sets with a mean redundancy contact with the outside solvent (calculated with Tunneller). The
46.8 (1351 of data) provided suitable starting phases (Figs. 2 and 3). iodoPhe residue is only slightly distorted relative to the native
This redundancy, corresponding to a oDF4/osDF4 of 2.6, is Phe153, the chi1 angle being rotated by 231 relative to the native
probably close to the absolute minimum signal necessary for sub- structure, presumably to accommodate the large iodine atom
structure determination and phase derivation, and represents for this (Fig. 3c). All other residues in the protein are largely unchanged
space group (P3221) just over two times the amount of data one relative to the native structure. If a comparison is made of the
would collect for a complete native data set (601) or the amount of positions of shared atoms contained within the hydrophobic core of
iodoPhe T4 lysozyme and wild-type T4 lysozyme (PDB code 1L63
(ref. 15)), the 264 common atoms contained in a 10-Å sphere around
Figure 3 Structures and representative electron the iodine atom have a root mean squared deviation (r.m.s.d.) of
density of the iodoPhe. (a) Ribbon diagram of T4 1.08 Å, and the 164 aligned Ca atoms in the structures have an r.m.s.d.
lysozyme with an iodoPhe at position 153. The
ribbon is colored blue at the N-terminal to red at
the C-terminal. The iodoPhe is presented in a
ball-and-stick representation. (b) Initial electron a b c
density around iodoPhe after phase refinement of
initial phases generated by SOLVE with RESOLVE
for the data set representing 1351 of data. The
final refined model is shown in a ball-and-stick
representation for comparison. (c) Comparison
of hydrophobic core with native15 and iodoPhe
T4 lysozyme. The iodoPhe structure is colored
green whereas the native (1L63) protein is
shaded red. Figures produced by Bobscript
and Raster3D27–29.

NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 10 OCTOBER 2004 1299

 LETTERS
of 0.282 Å. If we compare this with two ‘native’ lysozyme structures
(PDB codes 120L and 1L63), the r.m.s.d. of atoms contained within
the same sphere is 1.05 Å and that of the 162 aligned Ca atoms is
0.151 Å. This, together with the isomorphous nature of the crystals,
shows that the incorporation of the iodine does not perturb the
structure to any degree (Fig. 3c).
In conclusion, this study describes an approach to preparing iodine-
containing proteins for SAD phasing. It overcomes the limitations of
current methods by introducing an iodine atom into proteins site speci-
© 2004 Nature Publishing Group http://www.nature.com/naturebiotechnology
fically and quantitatively. With T4 lysozyme, a protein of 164 amino
acid residues, a single iodoPhe substitution was enough to carry out
SAD phasing and solve the structure with an in-house X-ray source.
Moreover, the iodine atom did not perturb the protein structure.
Considerably less data were required than for the corresponding experi-
ment using sulfur as the anomalous scatterer. Because this approach can
also be used to introduce two or more iodoPhe into proteins in res-
ponse to amber codons (P.G.S., unpublished data), its application to
SAD phasing should not be limited by the protein size. The technique
will also be useful for proteins that contain few or no methionine
residues, for which the selenomethionine derivation method is not
applicable. And finally, this method should also be applicable to higher
organisms. Indeed, we2 and others19 have shown that 3-iodo-L-tyrosine
and p-iodo-L-phenylalanine can be incorporated into proteins in
response to an amber codon in mammalian cells and in yeast.
METHODS
Expression and purification of iodoPhe-substituted T4 lysozyme. Plasmid
pT4L153TAG was used to express a Phe153-TAG mutant cysteine-free T4
lysozyme under the control of a bacteriophage T5 promoter and t0 terminator,
Tyr
and the tRNACUA gene under the control of the lpp promoter and rrnC
terminator. Plasmid pBK-iodoPheRS encoded the iodoPhe-specific tRNA-
aminoacyl synthetase under the control of the constitutive E. coli GlnRS
promoter and terminator. Electro-competent BL21 (DE3) cells cotransformed
with pT4L153TAG and pBK-IodoPheRS were grown in minimal medium
containing 1% glycerol and 0.3 mM leucine with 50 mg/ml kanamycin,
34 mg/ml of chloramphenicol and 1.0 mM p-iodo-L-phenylalanine at 37 1C.
When cells reached an OD600 of 0.5, isopropyl-b-D-thiogalactopyranoside was
added to a final concentration of 1 mM to induce protein expression. Cells
were grown an additional 8 h at 30 1C, pelleted and resuspended in 28 ml lysis
buffer (30 mM Tris
HCl, pH 7.6, one tablet of Complete (protease inhibitor
cocktail tablets, Roche Applied Science) for 50 ml solution). The cells were
lysed by sonication and pelleted again. The supernatant was then applied to a
cation exchange column (Mono S HR 10/10 column; Amersham Biosciences)
previously equilibrated with 30 mM Tris
HCl, pH 7.6. The proteins were eluted
with a linear gradient from 0 to 0.28 M NaCl. Peak fractions were analyzed by
SDS-PAGE. Fractions from a major peak that eluted at 0.25 M NaCl were
pooled together and applied to gel filtration (Superdex 75 HR 10/30 column;
Amersham Biosciences). Proteins were eluted with 25 mM Tris
HCl, 100 mM
NaCl, pH 7.6. The final purified T4 lysozyme was dialyzed against 100 mM
NaH2PO4, 0.55 M NaCl, pH 6.7 and then concentrated to 25 mg/ml for
crystallization. The concentration of protein was measured by Bradford assay
(BCA kit; Biorad).
Crystallization of iodoPhe-substituted T4 lysozyme. Mutant T4 lysozyme was
crystallized using the hanging-drop vapor diffusion method under conditions
similar to those described for other T4 lysozyme mutants14. The crystallization
solution consisted of 2.0–2.2 M aqueous sodium/potassium phosphate
buffer (pH 6.7–7.1), with 15 mM hydroxyethyl disulfide. One microliter of a
25 mg/ml protein solution was mixed with 1 ml of crystallization solution on a
silanized coverslip, which was inverted and placed above a reservoir containing
0.5 ml of the crystallization solution. Crystals grew in 2 weeks at 4 1C.
Before data collection, crystals were soaked in a cryoprotectant solution
(2.3 M sodium/potassium phosphate buffer, 0.25 M NaCl, 25% glycerol) and
cryo-cooled to 100 K.
1300
Data collection. Data were collected at 100 K on a standard in-house Rotating
RU300 X-ray generator (Rigaku/MSC) incorporating Osmic mirrors and an
RAXISIV++ imaging plate system at the CuKa wavelength of 1.5418 Å. The
crystal was mounted in a random orientation, no attempt being made to collect
Bijovet pairs on the same image, and data were collected in 0.51oscillations for a
total rotation of 3601 using an exposure time of 5 min per frame. Data were
collected to a maximum resolution of 2.0 Å and reduced and scaled with the
HKL2000 package20. Crystals belong to the trigonal space group P3221 and are
isomorphous with standard T4 lysozyme crystals15(Table 1). To determine the
minimum data needed to determine the substructure and derive initial phases,
the data were split into six different sets constituting rotations of 3601, 1801,
1451, 901, 601 and 451, statistics for which are presented in Table 1. Optimum
completeness for the data sets was calculated using the strategy option
of Mosflm21. Further crystallographic manipulations were done with the
CCP4 package21.
Structure solution and refinement. All substructure solution and determina-
tion of initial phases were carried out with SOLVE16 using the standard SAD
phasing script, and local scaling of the data. This was followed by solvent
flattening and auto-building using RESOLVE16 with a solvent content of 50%.
For the three data sets (451, 601, 901) that SOLVE was unable to determine, the
iodine position SHELXS and SHELXD22 were also tried with no success.
Refinement and rebuilding of the model was carried out with all data between
30.0 and 2.0 Å, using Refmac5 (ref. 23), incorporating a parameter file for the
p-iodo-L-phenylalanine generated by PROGRG24 and O25. Automatic water
building was carried with the ARP/WARP package26. Subsequently, three water
molecules were replaced by three putative chlorine ions, added on the basis of
inspection of the anomalous difference Patterson maps and the low B-factors of
the waters. In addition, a 2-hydroxyethyl disulfide moiety was also added as
observed in the anomalous difference (DF,Fcalc-90) Fourier maps and in most
other T4 lysozyme structures. The final model converged at an Rcryst and Rfree
of 0.157 and 0.207 respectively, the stereochemical properties were excellent
and all residues were contained within allowed regions of the Ramachandran
plot (Table 1).
Accession number for structure. The structure has been deposited with the
Research Collaboratory for Structural Bioinformatics Protein Data Bank (PDB)
and can be accessed as PDB code 1T6H (http://www.rcsb.org/pdb/).
ACKNOWLEDGMENTS
We would like to thank Heidi Erlandsen and Raymond Stevens for technical
assistance. J.X. thanks Jeremy Mills, Jun Yin and Yan Zhang for helpful
discussions. This work is supported by a grant from the National Institutes
of Health (GM62159) and the Skaggs Institute for Chemical Biology. This is
manuscript number 16547-CH of the Scripps Research Institute.
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
Received 13 May; accepted 2 August 2004
Published online at http://www.nature.com/naturebiotechnology/
1. Wang, L., Brock, A., Herberich, B. & Schultz, P.G. Expanding the genetic code of
Escherichia coli. Science 292, 498–500 (2001).
2. Chin, J.W. et al. An expanded eukaryotic genetic code. Science 301, 964–967 (2003).
3. Hendrickson, W. & Teeter, M. Structure of the hydrophobic protein crambin determined
directly from the anomalous scattering of sulfur. Nature 290, 107–113 (1981).
4. Debreczeni, J.E., Bunkoczi, G., Ma, Q., Blaser, H. & Sheldrick, G.M. In-house
measurement of the sulfur anomalous signal and its use for phasing. Acta Crystallogr.
D Biol. Crystallogr. 59, 688–696 (2003).
5. Dauter, Z., Dauter, M. & Dodson, E. Jolly SAD. Acta Crystallogr. D Biol. Crystallogr. 58,
494–506 (2002).
6. Dauter, Z., Dauter, M., de La Fortelle, E., Bricogne, G. & Sheldrick, G.M. Can
anomalous signal of sulfur become a tool for solving protein crystal structures?
J. Mol. Biol. 289, 83–92 (1999).
7. Dauter, Z., Dauter, M. & Rajashankar, K.R. Novel approach to phasing proteins:
derivatization by short cryo-soaking with halides. Acta Crystallogr. D Biol. Crystallogr.
56, 232–237 (2000).
8. Nagem, R.A., Dauter, Z. & Polikarpov, I. Protein crystal structure solution by fast
incorporation of negatively and positively charged anomalous scatterers. Acta Crystal-
logr. D Biol. Crystallogr. 57, 996–1002 (2001).
9. Boles, J.O. et al. Bio-incorporation of telluromethionine into buried residues of
dihydrofolate reductase. Nat. Struct. Biol. 1, 283–284 (1994).
VOLUME 22 NUMBER 10 OCTOBER 2004 NATURE BIOTECHNOLOGY

10. Budisa, N. et al. Bioincorporation of telluromethionine into proteins: a promising new
approach for X-ray structure analysis of proteins. J. Mol. Biol. 270, 616–623 (1997).
11. Brick, P., Bhat, T.N. & Blow, D.M. Structure of tyrosyl-tRNA synthetase refined at 2.3 Å
resolution. Interaction of the enzyme with the tyrosyl adenylate intermediate. J. Mol.
Biol. 208, 83–98 (1989).
12. Wang, L., Zhang, Z., Brock, A. & Schultz, P.G. Addition of the keto functional group to
the genetic code of Escherichia coli. Proc. Natl. Acad. Sci. USA 100, 56–61 (2003).
13. Kirshenbaum, K., Carrico, I.S. & Tirrell, D.A. Biosynthesis of proteins incorporating a
versatile set of phenylalanine analogues. Chembiochem 3, 235–237 (2002).
14. Eriksson, A.E., Baase, W.A. & Matthews, B.W. Similar hydrophobic replacements of
Leu99 and Phe153 within the core of T4 lysozyme have different structural and
thermodynamic consequences. J. Mol. Biol. 229, 747–769 (1993).
© 2004 Nature Publishing Group http://www.nature.com/naturebiotechnology
15. Nicholson, H., Anderson, D.E., Dao-pin, S. & Matthews, B.W. Analysis of the interac-
tion between charged side chains and the alpha-helix dipole using designed thermo-
stable mutants of phage T4 lysozyme. Biochemistry 30, 9816–9828 (1991).
16. Terwilliger, T.C. & Berendzen, J. Automated MAD and MIR structure solution. Acta
Crystallogr. D Biol. Crystallogr. 55, 849–861 (1999).
17. Terwilliger, T.C. Maximum-likelihood density modification using pattern recognition of
structural motifs. Acta Crystallogr. D Biol. Crystallogr. 57, 1755–1762 (2001).
18. Weaver, L.H. & Matthews, B.W. Structure of bacteriophage T4 lysozyme refined at 1.7 Å
resolution. J. Mol. Biol. 193, 189–199 (1987).
19. Sakamoto, K. et al. Site-specific incorporation of an unnatural amino acid into proteins
in mammalian cells. Nucleic Acids Res. 30, 4692–4699 (2002).
NATURE BIOTECHNOLOGY VOLUME 22 NUMBER 10 OCTOBER 2004
LETTERS
20. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation
mode. Methods Enzymol. 276, 307–326 (1997).
21. Collaborative Computational Project. N The CCP4 suite: programs for protein crystal-
lography. Acta Crystallogr. D Biol. Crystallogr. 50, 760–763 (1994).
22. Schneider, T.R. & Sheldrick, G.M. Substructure solution with SHELXD. Acta Crystal-
logr. D Biol. Crystallogr. 58, 1772–1779 (2002).
23. Murshudov, G.N., Vagin, A.A. & Dodson, E.J. Refinement of macromolecular structures
by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255
(1997).
24. van Aalten, D.M. et al. PRODRG, a program for generating molecular topologies and
unique molecular descriptors from coordinates of small molecules. J. Comput. Aided
Mol. Des. 10, 255–262 (1996).
25. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Improved methods for building
protein models in electron density maps and the location of errors in these models.
Acta Crystallogr. A 47, 110–119 (1991).
26. Lamzin, V.S. & Wilson, K.S. Automated refinement of protein models. Acta Crystallogr.
D Biol. Crystallogr. 49, 129–147 (1993).
27. Kraulis, P.J. MOLSCRIPT: a program to produce both detailed and schematic plots of
protein structures. J. Appl. Crystallogr. 24, 946–950 (1991).
28. Merritt, E.A. & Murphy, M.E.P. Raster3D Version 2.0. A program for photorealistic
molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 50, 869–873 (1994).
29. Esnouf, R.M. An extensively modified version of MolScript that includes greatly
enhanced coloring capabilities. J. Mol. Graph. Model. 15, 132–134 (1997).
1301