© 2000 Nature America Inc. • http://structbio.nature.com
letters
Nitric oxide binding to
nitrophorin 4 induces
complete distal pocket
burial
Andrzej Weichsel, John F. Andersen, Sue A. Roberts
and William R. Montfort
© 2000 Nature America Inc. • http://structbio.nature.com
Department of Biochemistry,
Arizona 85721, USA.
University
of
Arizona,
Tucson,
The nitrophorins comprise an unusual family of proteins that
use ferric (Fe(III)) heme to transport highly reactive nitric
oxide (NO) from the salivary gland of a blood sucking bug to
the victim, resulting in vasodilation and reduced blood coagulation. We have determined structures of nitrophorin 4 in complexes with H2O, cyanide and nitric oxide. These structures
reveal a remarkable feature: the nitrophorins have a broadly
open distal pocket in the absence of NO, but upon NO binding,
three or more water molecules are expelled and two loops fold
into the distal pocket, resulting in the packing of hydrophobic
groups around the NO molecule and increased distortion of
the heme. In this way, the protein apparently forms a
‘hydrophobic trap’ for the NO molecule. The structures are
very accurate, ranging between 1.6 and 1.4 Å resolutions.
Nitric oxide (NO), a very reactive molecule, has the unlikely
role of signal transduction in virtually all vertebrate cells1. Among
the many cellular responses to NO are neurotransmission,
smooth muscle relaxation and inhibition of platelet aggregation.
The focus of the present work is on NO binding to nitrophorin 4
(NP4), one of four NO transport proteins found in the saliva of
the blood sucking insect Rhodnius prolixus. R. prolixus, like all
blood sucking insects, faces the difficult task of overcoming
hemostasis, the host’s defense mechanism against blood loss. To
obtain a meal, the insect has evolved specialized appendages for
tapping into blood vessels, and specialized proteins to assist in
promoting blood flow2,3. It also spreads the parasite Trypanosoma
cruzi while feeding, which causes Chagas’ disease, an incurable
heart disease afflicting 16–18 million Latin Americans4. The
nitrophorins transport NO from the insect saliva, using a ferric
(Fe(III)) heme center, to the victim’s tissue, where it is released for
binding by soluble guanylate synthase (sGC), the major target for
NO signaling and a ferrous (Fe(II)) hemoprotein5, resulting in
vasodilation and inhibition of platelet aggregation6. The proteins
also sequester histamine and one (NP2) blocks blood coagulation.
The ferric heme centers in the nitrophorins lead to NO dissociation constants in the micromolar range7. For efficient transport to occur, the nitrophorins must stabilize the heme against
NO dependent reduction since the resulting ferrous heme would
have a picomolar affinity for NO (ref. 8). Such stabilization is
absent, for example, in the Fe(III) center in methemoglobin9,10 or
metmyoglobin at higher pH (ref. 11), which are readily reduced
by NO to the ferrous state.
The Rhodnius NPs are all ∼20 kDa and composed of a β-barrel
with heme inserted into one end of the barrel, a unique fold for
hemoproteins12,13 (Fig. 1). The proteins, however, fall into the
lipocalin family and, surprisingly, are among a growing number of
insect salivary proteins that are functionally distinct but have
lipocalin folds3. The NO binds to the distal side of the heme, which
is part of a large, open cavity in the unliganded protein. The proxnature structural biology • volume 7 number 7 • july 2000
Fig. 1 Ribbon drawing of the NP4–NO structure. The loops that move on
NO binding (loop A-B, residues 31–37, and loop G-H, residues 125–132)
are colored red, the heme in blue (stick representation), the disulfide
bonds in yellow, and the linear NO orientation in blue and red (ball-andstick representation).
imal side of the heme is occupied by a histidine. The kinetic rate
constants for NO association with the Rhodnius NPs range
between 2 and 33 µM-1 sec-1 (refs 7,14,33), values that are similar
to that for elephant metmyoglobin but 30- to 500-fold larger than
that for whale metmyoglobin15, consistent with relatively unimpeded binding. In contrast, NO release is biphasic7,14,33, pH dependent6,7,33 and slower than that for elephant and whale
metmyoglobins (0.01–1.4 s-1 versus 14–40 s-1; refs 7,15,33). These
factors suggest that NO release is somehow blocked or otherwise
limited, and that this activity is increased at lower pH. However, in
several previously reported structures of NP1 and NP4, including
complexes with NH3, cyanide, histamine, and NO bound to
reduced (ferrous) NP1, no changes in structure were found that
could account for the NO release kinetics12,13,16.
NP4–NO structure
Obtaining a ferric NP–NO complex has proved challenging due
to the volatility and reactivity of NO, and a tendency for NO
dependent photoreduction to occur during data measurement16.
We have overcome this difficulty at least in part by obtaining crystals of NP4 at lower pH, where NO binding is tighter, and by finding conditions in which the NP4–NO crystal complex could be
trapped and stabilized through flash-freezing. Data to 1.4 Å resolution were measured from one such crystal (Table 1), and the
structure determined using a nearly isomorphous room temperature structure of NP4 at high pH (ref. 12). The resulting NP4–NO
structure displays clear electron density for the nitric oxide moiety, although the density is consistent with multiple orientations
for NO in the complex. The structure also displays a surprising
conformational change with respect to other nitrophorin complexes that leads to complete burial of the NO binding pocket. In
what follows, we first examine possible explanations for the NO
coordination in the crystal, and then provide a detailed description of the NO-induced conformational change.
Multiple NO orientations
The model that best fits the distal pocket electron density is one in
which the NO occupies two orientations in the crystal, a ‘linear’
orientation, as expected for a ferric heme–NO complex, and an
unexpected ‘bent’ or ‘side-on’ orientation (Fig. 2). Such a model
left no unexplained difference electron density peaks, whereas
modeling just one NO orientation resulted in difference electron
551
© 2000 Nature America Inc. • http://structbio.nature.com
letters
© 2000 Nature America Inc. • http://structbio.nature.com
a
Fig. 2 Electron density maps for NO and heme. a, Stereo
view of the Final 2Fo - Fc map superimposed with the model
for the heme, the four vinyl positions, and the two NO orientations. The map is contoured at a level of 1σ.
b, Fo - Fc map where the ‘bent’ NO conformer has been omitted from the model and phases calculated after removal of
model phase bias through simulated annealing. Contoured
at 3σ. c, As in (b) except the ‘linear’ NO conformer was
omitted. Carbon is shown colored green, nitrogen blue,
oxygen red, and iron black.
crowding between Leu 130 and NO in the linear orientation (3.5 Å between the NO oxygen and the
leucine side chain), and the existence of a small
hydrophobic cavity that lies in the direction of the
c
b
NO bending, which could stabilize the bent or completely dislodged NO molecule. That moderate steric
factors can bend a metal–NO bond has been demonstrated for a ruthenium nitrosyl porphyrin complex18, and suggested for the nitrosyl cytochrome c
peroxidase complex19. Other considerations include
the possibility that the 10° off-axis canting of proximal ligand His 59 induces NO bending20, or that we
density corresponding to the other orientation (Fig. 2). The high are observing NO in a photo-excited state where side-on binding
resolution of the structure allowed the NO moiety to be refined in is favored21. Additional experiments should clarify this issue.
the two orientations without restraining iron–NO distances or
angles. This led to a final model with an Fe–N distance of 1.5 Å, NO-induced loop ordering in NP4
and an Fe–N–O angle of 177° for orientation 1, and an Fe–N dis- Quite surprisingly, NO binding to NP4 leads to a major confortance of 2.0 Å, an Fe–O distance of 2.6 Å, and an angle of 110° for mational change in the protein. On binding NO, burial of the
orientation 2. In the second orientation, NO lies near a small distal pocket occurs through the shifting of loop G-H and the
hydrophobic cavity at the back of the binding pocket. The ‘linear’ ordering of loop A-B (Figs 1, 3, 4). Closure involves the expulorientation is consistent with model ferriporphyrin–NO complex- sion of at least three ordered solvent molecules, the burial of Asp
es, which mostly display Fe–N bond distances of between 1.6 and 30, the formation of an extensive hydrogen bonding network,
1.7 Å, and Fe–N–O bond angles of between 174 and 178° (ref. 17). and the packing of hydrophobic atoms around NO. Residues
The ‘bent’ conformer, however, is neither consistent with the ferri- 125–132 shift toward the distal pocket, allowing Leu 130 to pack
porphyrin complexes, nor with the ferriporphyrin–NO complexes directly against the NO molecule (Fig. 3). Residues 31–37
(Fe–N bond distances 1.7–1.8 Å and Fe–N–O angles 138–144°)17. become ordered and cover the distal pocket entrance, allowing
Two explanations for the ‘bent’ NO orientation were explored. Val 36 to pack against Leu 130 and Leu 133, further burying the
The possibility that the crystal contains a mixture of ferric and fer- bound NO. The new positions for these loops are stabilized
rous hemes was initially attractive since NP1, which is 90% identi- through an extensive hydrogen bonding network involving Asp
cal to NP4, can be photoreduced in the presence of NO and X-rays. 30, Glu 32, Asp 35, Asp 129, and the N-terminus, and require the
A ferrous NP1–NO complex was apparently converted from ferric peptide bond between Leu 130 and Glu 131 to flip over (∆φ =
NP1–NO during data measurement16, but that protein was in the -170° for Glu 131). A total of 13 new hydrogen bonds are formed
open conformation and at room temperature in the experiment, in the closed conformer, and 9 that occur in the open conformer
where reaction with hydroxide or water could lead to reduced are lost (Fig. 4). Asp 30 and Asp 35 are apparently protonated in
heme11. In the closed, frozen NP4–NO structures, this mechanism the closed conformer, and therefore have elevated pKas, based on
seems precluded. Two additional NP4–NO crystals were exam- their hydrogen bonding arrangement (Fig. 4). The new arrangeined, one at room temperature (pH 5.6) and one very quickly (3 h ment also leads to small changes in the heme geometry, which
total exposure time), and both gave similar NO geometries that are are described below.
inconsistent with reduction of Fe being the source of NO bending.
The packing of hydrophobic groups around the NO fills the
A more attractive possibility is that the NP4–NO complex con- distal pocket and, coupled with the release of solvent from the
tains a mixture of bound and unbound or loosely ligated NO
molecules. Support for this derives from Fourier transform
infrared spectroscopy (FTIR) studies of the NP1–NO complex, in
which two NO stretching bands are observed16, and from kinetic
measurements where two release rates are observed for all of the
nitrophorins7,14,33. The bent orientation may arise from steric
Fig. 3 Loop ordering on NO binding. Space filling views of the distal
–
pockets in the NP4–NO (left) and NP4–CN (right) structures. The view is
from above the distal pocket, with heme shown in black, heme oxygens
–
in orange, NO in magenta, CN in cyan, five water molecules in red, and
loops A-B and G-H in green. Binding of NO leads to reordering of the
A-B and G-H loops, expulsion of the water molecules, burying of NO and
Asp 30, and packing of Leu 130 against the NO molecule.
552
nature structural biology • volume 7 number 7 • july 2000
© 2000 Nature America Inc. • http://structbio.nature.com
letters
© 2000 Nature America Inc. • http://structbio.nature.com
Fig. 4 Hydrogen bonding in the mobile loops. a, Stereo view
of the distal pocket in NP4–NO. b, Stereo view of the N-terminus in NP4–NO. c, Stereo view of the closed (red lines) and
open (green lines) conformers of NP4, after superimposing
them. In (a) and (b), bonds are open for Asp 30 and filled for
the other residues. Nitrogens are indicated by large open
spheres, carbons by small open spheres, oxygens by shaded
spheres, and hydrogen bonds by dashed lines.
a
now more hydrophobic distal pocket, most likely provides the driving force for the NO induced conformational change. NO is 70 times more soluble in
n-hexane than in water22,23. NP4 is apparently bal- b
anced to sense this change in hydrophobicity and uses
it to bury the distal pocket and protect the bound NO
from further reaction. The numerous hydrogen
bonds that exchange during the conformational
change help to set this balance.
We determined the NP4–H2O and NP4–CN(pH 5.6) structures to further probe the factors necessary for a ligand-induced conformational change
(Table 1). Both complexes were in the open conformation with fully hydrated distal pockets (Fig. 3), c
much like that for the reported NP4 and NP1 complexes with ammonia12,13 and the NP1–cyano complex13, except that the cyano group was less well
ordered in NP4 than in NP1. Thus, although similar
in binding geometry and size to NO, CN-does not
induce a conformational change in NP4, consistent
with our proposal that the hydrophobicity of NO is
the driving force for stabilizing the closed NP4 conformer. However, the structure of the NP1–NO
complex, which was photoreduced during data measurement and did not display a closed conformation16
is not consistent with this proposal. Based on the NP1
kinetic behavior, which is very similar to that of NP47,12,33, it
seems likely that NP1 also undergoes a conformational change in
solution on binding NO, and the reason for the absence of a conformational change in the crystalline complex is due to unfavorable crystalline contacts that interfere with the conformational
change.
Heme conformational change
The high resolutions of the present structures have allowed a careful look at the NP4 heme conformation, which is highly distorted.
For the unligated protein, the bonds between the heme pyrroles
and Fe are quite regular, and the plane defined by the pyrrole
nitrogens and heme iron quite flat. The pyrrole rings, however,
rotate 4–9° about the Fe–N bonds, giving rise to a highly non-planar heme exhibiting a ‘ruffled’ conformation. On binding NO, the
ruffling becomes more pronounced, with the pyrrole rings rotating as much as 14°. The Fe–N bond lengths do not change, but the
iron moves out of the pyrrole nitrogen plane (by 0.07 Å) and into
the distal pocket, giving rise to a slightly ‘domed’ heme conformation. A more detailed description of the heme geometry, based on
a 1.15 Å NP4 structure, is presented elsewhere (S.A. R., A. W., and
W.R. M., unpublished observations).
The high resolution structures also show the heme to be discretely disordered in the crystal, with about 50% of the heme
refining as ‘right-side-up’ and 50% as ‘upside-down’, such that
all four possible vinyl positions (two for up and two for down)
are partially occupied. It is unlikely that this disorder has any
functional consequence, since the overall heme position and
nature structural biology • volume 7 number 7 • july 2000
conformation are unchanged in the two orientations, much like
the case for cytochrome b5 (see, for example, ref. 24).
Mechanism of NO transport
Taken together, these structures provide a deeper understanding of
how the nitrophorins use heme for NO transport, while sGC and
NO synthase (NOS) use heme for NO signaling and NO synthesis.
The stabilization of the NP4–NO complex to autoreduction, which
is not observed in metmyoglobin16, is apparently due to at least two
factors. First, the nitrophorin heme has a lower reduction potential
than metmyoglobin, both in the presence and absence of NO16,33.
We postulate that the high number of negative charges near the
heme iron, and possibly also the highly ruffled heme25, are responsible for the increased stability of ferric heme. These factors may be
enhanced in the NO bound closed conformer due to increased ruffling and increased hydrophobicity, which may lead to an increase
in the local negative electrostatic potential by lowering the local
dielectric constant. Second, the heme bound NO is protected by
distal pocket closure from agents such as hydroxide or excess NO
that can participate in heme reduction11.
NO dissociation constants are ∼15–75,000-fold smaller for the
nitrophorins than for the metmyoglobins at lower pH, depending
on the proteins compared7,11,14,15,33. These drop to ∼2–4,000-fold
smaller above pH 7, when the NO affinity for the nitrophorins is
reduced to enable its release. The NP4–NO structure suggests that
the NO induced conformational change is responsible for these
kinetic differences. In the absence of NO, the distal pocket is open
and NO must only displace a weakly bound water molecule to
553
© 2000 Nature America Inc. • http://structbio.nature.com
letters
Table 1 Crystallographic results
© 2000 Nature America Inc. • http://structbio.nature.com
Complex
pH / Temperature (K)
Wavelength (Å)
Cell (space group C2)
NP4–NO
5.6 / 140
1.54
a = 70.32 Å
b = 42.64 Å
c = 52.57 Å
β = 94.19°
Resolution (Å)
15–1.4
Total / unique reflections 56,842 / 28,023
Completeness1
91 / 79
Mean I / σI1
9.9 / 2.1
Rsym1,2
0.11 / 0.30
Rcryst / Rfree2
0.17 / 0.24
R.m.s. deviations
Bonds (Å)
0.01
Angles (Å)
0.039
Data / parameter
1.9
Most favored φ/ψ (%)
91
Disallowed3
0
Number of solvent atoms 160
Residues in
multiple positions4
5
<B> protein / ligand5
24.9 / 19.7
–
NP4–H2O
5.6 / 140
1.54
a = 70.18 Å
b = 42.52 Å
c = 52.86 Å
β = 94.35°
30–1.4
66,833 / 25,227
80 / 79
10.3 / 4.2
0.12 / 0.15
0.20 / 0.25
NP4–CN
5.6 / 140
1.54
a = 70.04 Å
b = 42.65 Å
c = 52.56 Å
β = 94.18°
11–1.6
31,409 / 19,393
87 / 76
11.5 / 3.5
0.05 / 0.16
0.21 / 0.26
0.01
0.030
3.6
88
0
181
0.01
0.035
2.9
91
0
162
5
15.1 / 12.1
3
19.5 / 16.0
Overall / outer shell.
2R-factors were calculated in the usual manner, using all data.
3Disallowed region in Ramachandram plot.
4Discretely disordered residues.
5Average B factor.
1
isomorphous with crystals grown at pH 7.5 in ammonium
phosphate12. The NO complex was obtained with a 2 h equilibration of a NP4 crystal in solution A (50% w/v PEG 4000,
100 mM sodium citrate, pH 5.6, saturated with argon), followed by 1 h in a similar solution saturated with NO. The
cyano complex was obtained by equilibrating a crystal for 2
h in solution B (30% w/v PEG 4000, 100 mM sodium citrate,
10 mM KCN, pH 5.6). Both crystals changed from pale red to
bright red. Crystals were picked up in a cryoloop (Hampton)
and flash-frozen in a liquid nitrogen gas stream for data
measurement on a FAST area detector and programs
MADNES27, PROCOR28, and CCP429. The structures were
modeled starting with the pH 7.5 crystal structure12 and
programs O30, REFMAC29, X-PLOR31, and SHELXL32. Final
models for NP4–NO and NP4–H2O were refined with the
conjugate gradient approach of SHELXL, which led to a
lower free R-factor. Heme and axial ligands were left unrestrained. The NO complex was refined with anisotropic
temperature factors, yielding a slight improvement in Rfree.
The aqua complex was refined isotropically except for the
iron and sulfur atoms. Each heme was refined with four
vinyls having 50% occupancy.
Coordinates. Coordinates and structure factors have been
deposited with the Protein Data Bank (accession codes:
–
1ERX, NP4–NO, pH 5.6; 1D3S, NP4–H2O; 1EQD, NP4–CN ).
Acknowledgments
We thank C. Balfour for protein purification. Supported in part by grants
from NIH, ACS, and ADCRC to W.R.M., and from the NIH to J.F.A.
Correspondence should be addressed to W.R.M. email:
[email protected]
Received 6 January, 2000; accepted 20 April, 2000.
bind to the heme iron. In sperm whale metmyoglobin, a well
ordered water molecule bound to heme and hydrogen bonded to
the distal histidine must be displaced before NO can bind, leading
to slower on-rates than occur in either elephant myoglobin or the
nitrophorins, neither of which have a stabilized aqua complex15,26.
To release NO, the NP4 ordered loops must become disordered,
whereas in metmyoglobin, no such barrier exists. That the NP4
loop transitions are indeed rate limiting for NO release is supported by preliminary data for two mutants with altered A-B
loops (D30A and D30N); both mutants release NO faster than the
wild type protein.
The mechanism by which the loops interfere with NO release is
not yet clear but forms a fascinating question in our view, since
the loops, which display higher than average temperature factors
in the crystal, are inherently dynamic, and the Fe–NO bond does
not appear to be stronger than in other ferrinitrosyl complexes,
based on IR measurements16. Apparently, the hydrophobic nature
of the NO complex somehow impedes NO release, a mechanism
we refer to as ‘hydrophobic trapping’. Defining how such a trap
might function is the subject of experiments in progress.
Methods
NP4 expression and characterization. NP4 was overexpressed as
described7,12. Release of NO by NP4 at low pH in an open container
(15 µM NO, 10 µM NP4, 20 mM NaOAc, pH 5.0) was monitored spectroscopically12 and found to be unchanged after 5 days. Upon raising
the pH (2× dilution with 250 mM Tris-HCl, pH 8.0), the solution
returned to the unliganded spectrum over a period of hours, consistent with the presence of ferric heme in the protein.
Crystal structure determinations. Crystals at pH 5.6 were
obtained by the hanging drop method at room temperature (20 mg
ml-1 NP4 in 10 mM Tris-HCl, pH 7.5 mixed 1:1 with a well buffer of
25% w/v PEG 4000, 100 mM Na citrate, pH 5.6). These crystals were
554
1. Bredt, D.S. & Snyder, S.H. Annu. Rev. Biochem. 63, 175–195 (1994).
2. Law, J., Ribeiro, J.M.C. & Wells, M. Annu. Rev. Biochem. 61, 87–111 (1992).
3. Montfort, W.R., Weichsel, A. & Andersen, J.F. Biochim. Biophys. Acta, in the
press (2000).
4. Kirchhoff, L.V. N. Engl. J. Med. 329, 639–644 (1993).
5. Sharma, V.S. & Magde, D. Methods 19, 494–505 (1999).
6. Ribeiro, J.M.C., Hazzard, J.M.H., Nussenzveig, R.H., Champagne, D.E. & Walker,
F.A. Science 260, 539–541 (1993).
7. Andersen, J.F. et al. Biochemistry 36, 4423–4428 (1997).
8. Traylor, T.G. & Sharma, V.S. Biochemistry 31, 2847–2849 (1992).
9. Sharma, V.S., Isaacson, R.A., John, M.E., Waterman, M.R. & Chevion, M.
Biochemsitry 22, 3897–3902 (1983).
10. Addison, A.W. & Stephanos, J.J. Biochemistry 25, 4104–4113 (1986).
11. Hoshino, M., Maeda, M., Konishi, R., Seki, H. & Ford, P.C. J. Am. Chem. Soc. 118,
5702–5707 (1996).
12. Andersen, J.F., Weichsel, A., Balfour, C.A., Champagne, D.E. & Montfort, W.R.
Structure 6, 1315–1327 (1998).
13. Weichsel, A., Andersen, J.F., Champagne, D.E., Walker, F.A. & Montfort, W.R.
Nature Struct. Biol. 5, 304–309 (1998).
14. Kaneko, Y., Yuda, M., Iio, T., Murase, T. & Chinzei, Y. Biochim. Biophys. Acta 1431,
492–499 (1999).
15. Sharma, V.S., Traylor, T.G., Gardiner, R. & Mizukami, H. Biochemistry 26,
3837–3843 (1987).
16. Ding, X.D. et al. J. Am. Chem. Soc. 121, 128–138 (1999).
17. Scheidt, W.R. & Ellison, M.K. Acc. Chem. Res. 32, 350–359 (1999).
18. Hodge, S.J. et al. Chem. Commun. 19, 2283–2284 (1996).
19. Edwards, S.L., Kraut, J. & Poulos, T.L. Biochemistry 27, 8074–8081 (1988).
20. Jewsbury, P., Yamamoto, S., Minato, T., Saito, M. & Kitagawa, T. J. Am. Chem. Soc.
116, 11586–11587 (1994).
21. Carducci, M.D., Pressprich, M.R. & Coppens, P. J. Am. Chem. Soc. 119, 2669–2678
(1997).
22. Shaw, A.W. & Vosper, A.J. J. Chem. Soc. Faraday Trans. 8, 1239–1244 (1977).
23. Kerwin, J.F., Jr., Lancaster, J.R., Jr. & Feldman, P.L. J. Med. Chem. 38, 4343–4362
(1995).
24. Walker, F.A., Emrick, D., Rivera, J.E., Hanquet, B.J. & Buttlaire, D.H. J. Am. Chem.
Soc. 110, 6234–6240 (1988).
25. Shelnutt, J.A. et al. Chem. Soc. Rev. 27, 31–41 (1998).
26. Rohlfs, R.J. et al. J. Biol. Chem. 265, 3168–3176 (1990).
27. Messerschmidt, A. & Pflugrath, J.W. J. Appl. Crystallogr. 20, 306–315 (1987).
28. Kabsch, W. J. Appl. Crystallogr. 21, 916–934 (1988).
29. CCP4. Acta Crystallogr. D 50, 760–763 (1994).
30. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjelgard, M. Acta Crystallogr. A 47, 110–119
(1991).
31. Brunger, A.T. X-PLOR: a system for X-ray crystallography and NMR (Yale
University Press, New Haven, Connecticut; 1992).
32. Sheldrick, G.M. & Schneider, T.R. Methods Enzymol. 277, 319–343 (1997).
33. Andersen, J.F. et al. Biochemistry in the press (2000).
nature structural biology • volume 7 number 7 • july 2000