Characterization of the FKBP-rapamycin-FRB ternary complex

Article abstract of DOI:10.1021/ja043277y

Rapamycin is an important immunosuppressant, a possible anticancer therapeutic, and a widely used research tool. Essential to its various functions is its ability to bind simultaneously to two different proteins, FKBP and mTOR. Despite its widespread use, a thorough analysis of the interactions between FKBP, rapamycin, and the rapamycin-binding domain of mTOR, FRB, is lacking. To probe the affinities involved in the formation of the FKBP-rapamycin-FRB complex, we used fluorescence polarization, surface plasmon resonance, and NMR spectroscopy. Analysis of the data shows that rapamycin binds to FRB with moderate affinity (K_d = 26 ± 0.8 μM). The FKBP12-rapamycin complex, however, binds to FRB 2000-fold more tightly (K_d = 12 ± 0.8 nM) than rapamycin alone. No interaction between FKBP and FRB was detected in the absence of rapamycin. These studies suggest that rapamycin's ability to bind to FRB, and by extension to mTOR, in the absence of FKBP is of little consequence under physiological conditions. Furthermore, protein-protein interactions at the FKBP12-FRB interface play a role in the stability of the ternary complex.

Full text of DOI:10.1021/ja043277y

Published on Web 03/09/2005
Characterization of the FKBP‚Rapamycin‚FRB Ternary
Complex
Laura A. Banaszynski, Corey W. Liu,† and Thomas J. Wandless*,‡
Contribution from the Department of Chemistry, Stanford Magnetic Resonance Laboratory, and
Department of Molecular Pharmacology, Stanford UniVersity, Stanford, California 94305
Received November 8, 2004; E-mail: wandless@stanford.edu
Abstract: Rapamycin is an important immunosuppressant, a possible anticancer therapeutic, and a widely
used research tool. Essential to its various functions is its ability to bind simultaneously to two different
proteins, FKBP and mTOR. Despite its widespread use, a thorough analysis of the interactions between
FKBP, rapamycin, and the rapamycin-binding domain of mTOR, FRB, is lacking. To probe the affinities
involved in the formation of the FKBP‚rapamycin‚FRB complex, we used fluorescence polarization, surface
plasmon resonance, and NMR spectroscopy. Analysis of the data shows that rapamycin binds to FRB with
moderate affinity (Kd ) 26 ( 0.8 µM). The FKBP12‚rapamycin complex, however, binds to FRB 2000-fold
more tightly (Kd ) 12 ( 0.8 nM) than rapamycin alone. No interaction between FKBP and FRB was detected
in the absence of rapamycin. These studies suggest that rapamycin’s ability to bind to FRB, and by extension
to mTOR, in the absence of FKBP is of little consequence under physiological conditions. Furthermore,
protein-protein interactions at the FKBP12-FRB interface play a role in the stability of the ternary complex.
Introduction
Rapamycin (Figure 1A) is a 31-membered macrolide anti-
fungal antibiotic that was first isolated from Streptomyces
hygroscopicus in an Easter Island soil sample in 1975.1-4
Interest in the compound intensified in 1987, following the
discovery of FK506, a potent immunosuppressant with striking
structural similarity.5 Rapamycin binds with high affinity (Kd
) 0.2 nM) to the 12-kDa FK506 binding protein (FKBP12,
hereafter called FKBP)6-8 as well as to a 100-amino acid domain
(E2015 to Q2114) of the mammalian target of rapamycin
(mTOR) known as the FKBP-rapamycin binding domain
(FRB).9-12 Rapamycin is an FDA-approved drug which is used
clinically as an immunosuppressant for organ transplant patients.
Rapamycin is also important scientifically, as it is a potent and
Department of Chemistry.
† Stanford Magnetic Resonance Laboratory.
‡ Department of Molecular Pharmacology.
(1) Ve´zina, C.; Kudelski, A.; Sehgal, S. N. J. Antibiot. 1975, 28, 721-726.
(2) Sehgal, S. N.; Baker, H.; Ve´zina, C. J. Antibiot. 1975, 28, 727-732.
(3) Swindells, D. C. N.; White, P. S.; Findlay, J. A. Can. J. Chem. 1978, 56,
2491-2492.
(4) Findlay, J. A.; Radics, L. Can. J. Chem. 1980, 58, 579-590.
(5) Tanaka, H.; Kuroda, A.; Marusawa, H.; Hatanaka, H.; Kino, T.; Goto, T.;
Hashimoto, M. J. Am. Chem. Soc. 1987, 109, 5031-5033.
(6) Siekierka, J. J.; Hung, S. H. Y.; Poe, M.; Lin, C. S.; Sigal, N. H. Nature
1989, 341, 755-757.
(7) Harding, M. W.; Galat, A.; Uehling, D. E.; Schreiber, S. L. Nature 1989,
341, 758-760.
(8) Bierer, B. E.; Mattila, P. S.; Standaert, R. F.; Herzenberg, L. A.; Burakoff,
S. J.; Crabtree, G.; Schreiber, S. L. Proc. Natl. Acad. Sci. U.S.A. 1990, 87,
9231-9235.
Figure 1. (A) Chemical structure of rapamycin showing the FKBP and
FRB binding domains. (B) Schematic diagram representing the four possible
binding events involved in the formation of the FKBP‚rapamycin‚FRB
ternary complex.
(9) Brown, E. J.; Albers, M. W.; Shin, T. B.; Ichikawa, K.; Keith, C. T.; Lane,
W. S.; Schreiber, S. L. Nature 1994, 369, 756-758.
(10) Chiu, M. I.; Katz, H.; Berlin, V. Proc. Natl. Acad. Sci. U.S.A. 1994, 91,
12574-12578.
specific inhibitor of mTOR, which has been shown to be a major
effector of cell growth and proliferation via the regulation of
protein synthesis. Due to this activity, the mTOR pathway is
(11) Sabatini, D. M.; Erdjument-Bromage, H.; Lui, M.; Tempst, P.; Snyder, S.
H. Cell 1994, 78, 35-43.
(12) Chen, J.; Zheng, X. F.; Brown, E. J.; Schreiber, S. L. Proc. Natl. Acad.
Sci. U.S.A. 1995, 92, 4947-4951.
9
10.1021/ja043277y CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 4715-4721
4715

A R T I C L E S
Banaszynski et al.
Chart 1. Chemical Structures of Rapamycin Derivatives
emerging as a critical player in cancer and other metabolic
diseases, including diabetes and obesity.13
Perhaps the most widespread use of rapamycin has been in
technology developed to take advantage of the small molecule’s
ability to heterodimerize proteins. Proteins of interest can be
expressed as fusions to FKBP or FRB, and then conditionally
dimerized by the addition of rapamycin.14-18 This approach has
been used to regulate protein expression,19-21 protein splicing,22
and glycosylation,23 to name a few examples.
While a great deal of structural information about the FKBP‚
rapamycin‚FRB ternary complex exists,24,25 a thorough bio-
physical analysis of these interactions is lacking. The interaction
between FKBP and rapamycin has been well characterized (Kd
) 0.2 nM),8 and early experiments suggest that formation of a
ternary complex including FRB is quite favorable (Kd ≈ 2.5
nM).12 These studies also showed that FKBP and FRB do not
interact in the absence of rapamycin. Analysis of the crystal
structure shows extensive interactions between rapamycin and
its two protein partners, but relatively limited interactions
between the proteins themselves. Experiments in yeast have
shown that FKBP knockouts show resistance to rapamycin,26,27
and from these experiments it has been postulated that rapa-
mycin cannot bind to mTOR in the absence of FKBP. An
alternative explanation is that rapamycin binds to mTOR in the
absence of FKBP, but that FKBP is needed to block a protein-
protein interaction between mTOR and a critical partner protein.
To provide evidence to distinguish between these two possibili-
ties, and to determine if protein-protein interactions contribute
to overall complex stability, we sought to characterize the
interactions between rapamycin and FRB in the presence and
in the absence of FKBP (Figure 1B).
rium binding analysis of the data gave a dissociation constant
of 490 ( 39 nM (Figure 2A).
A competition binding experiment was then performed in
order to determine the affinity of unmodified rapamycin for FRB
(Figure 2B). FRB and Fl-rap were allowed to associate, and
various concentrations of rapamycin were added to compete Fl-
rap from the FRB binding pocket, which was measured as a
decrease in polarization units. However, the solubility of
rapamycin in this buffer (PBS pH 7.4, 0.011% Triton X-100,
1% EtOH) is limited (∼10 µM), and this fact, coupled with
our observation that rapamycin’s affinity for FRB is weaker
than that of Fl-rap, meant that we could not achieve concentra-
tions of rapamycin necessary to fully compete Fl-rap from FRB.
The enhanced affinity of Fl-rap for FRB relative to that of
rapamycin for FRB is perhaps due to a combination of
rapamycin‚FRB binding along with nonspecific binding of the
aromatic fluorophor to the hydrophobic protein. The partial
competition data were fit using a mathematical model that
explicitly considers all species present at equilibrium using ratios
of rate constants for association and dissociation of all possible
complexes and a corresponding collection of partial differential
rate equations from which dissociation constants can be
derived.28 The best fit of the incomplete competition data
suggested a dissociation constant of 5.2 µM (R2 ) 0.65) for
the affinity of rapamycin for FRB.
Results and Discussion
Fluorescence Polarization Assays. The crystal structure of
the ternary complex shows the C40-cyclohexyl hydroxyl group
of rapamycin to be distant from FRB.24 Therefore, we chose to
modify this location when synthesizing fluorescein-rapamycin
(Fl-rap, Chart 1) as a tracer for fluorescence polarization
experiments. The affinity of Fl-rap for FRB was determined
using a standard saturation binding experiment. A fixed
concentration of Fl-rap was incubated with various concentra-
tions of FRB, and formation of the Fl-rap‚FRB complex was
quantitated using an increase in polarization units. An equilib-
(13) Hay, N.; Sonenberg, N. Genes DeV. 2004, 18, 1926-1945.
(14) Pollock, R.; Clackson, T. Curr. Opin. Biotechnol. 2002, 13, 459-467.
(15) Clemons, P. A. Curr. Opin. Chem. Biol. 1999, 3, 112-115.
(16) Muthuswamy, S. K.; Gilman, M.; Brugge, J. S. Mol. Cell. Biol. 1999, 19,
6845-6857.
(17) Otto, K. G.; Jin, L.; Spencer, D. M.; Blau, C. A. Blood 2001, 97, 3662-
3664.
(18) Janse, D. M.; Crosas, B.; Finley, D.; Church, G. M. J. Biol. Chem. 2004,
279, 21415-21420.
(19) Ho, S. N.; Biggar, S. R.; Spencer, D. M.; Schreiber, S. L.; Crabtree, G. R.
Nature 1996, 382, 822-826.
The composite surface of the FKBP‚rapamycin complex is
thought to possess a tighter affinity for FRB relative to
rapamycin alone. To test this hypothesis, we repeated the
competition binding assay in the presence of FKBP. This
competition could not be completed, not due to rapamycin
solubility issues but because addition of the fluorescein moiety
does not completely abrogate binding of Fl-rap to FKBP
(Figure S1, Supporting Information). However, a mathematical
(20) Rivera, V. M.; Clackson, T.; Natesan, S.; Pollock, R.; Amara, J. F.; Keenan,
T.; Magari, S. R.; Phillips, T.; Courage, N. L.; Cerasoli, F.; Holt, D. A.;
Gilman, M. Nat. Med. 1996, 2, 1028-1032.
(21) Schlatter, S.; Senn, C.; Fussenegger, M. Biotechnol. Bioeng. 2003, 83, 210-
225.
(22) Mootz, H. D.; Blum, E. S.; Tyszkiewicz, A. B.; Muir, T. W. J. Am. Chem.
Soc. 2003, 125, 10561-10569.
(23) Kohler, J. J.; Bertozzi, C. R. Chem. Biol. 2003, 10, 1303-1311.
(24) Choi, J.; Chen, J.; Schreiber, S. L.; Clardy, J. Science 1996, 273, 239-
242.
(25) Liang, J.; Choi, J.; Clardy, J. Acta Crystallogr. 1999, D55, 736-744.
(26) Heitman, J.; Movva, N. R.; Hall, M. N Science 1991, 253, 905-909.
(27) Koltin, Y.; Faucette, L.; Bergsma, D. J.; Levy, M. A.; Cafferkey, R.; Koser,
P. L.; Johnson, R. K.; Livi, G. P. Mol. Cell. Biol. 1991, 11, 1718-1723.
(28) Braun, P. D.; Wandless, T. J. Biochemistry 2004, 43, 5406-5413.
9
4716 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

FKBP‚Rapamycin‚FRB Ternary Complex
A R T I C L E S
Surface Plasmon Resonance. The limited solubility of
rapamycin precluded full characterization of its affinity for FRB
using fluorescence polarization. As a result, we turned to surface
plasmon resonance (SPR) as a complementary method to
interrogate this system.29,30 To validate SPR as a technique for
interrogating these protein-ligand interactions, we first im-
mobilized a GST-FKBP fusion protein to the modified dextran
surface of a CM5 chip and exposed this surface to various
concentrations of rapamycin. These experiments provided a
dissociation constant for the FKBP‚rapamycin interaction (Kd
) 0.27 nM, Figure S2, Supporting Information) that is nearly
identical to both the literature value (Kd ) 0.2 nM)8 and the
affinity measured by fluorescence polarization (Kd ) 0.35 nM,
Figure 2C).
In our initial studies of the rapamycin‚FRB interaction, we
sought to capture a GST-FRB fusion protein using an anti-GST
antibody immobilized on the dextran surface. Using this format
we could expose the FRB surface to various concentrations of
rapamycin and quantitate the interactions. This experimental
design would allow us to use a fresh protein surface for each
titration point, making it unnecessary to expose FRB to harsh
regeneration conditions between each injection of analyte.
Additionally, the only modified species in the experiment would
be the GST-FRB fusion protein. We have characterized this
fusion protein using fluorescence polarization, and have shown
that it has an affinity for Fl-rap similar to that of the isolated
FRB domain (Kd ) 960 ( 69 nM vs Kd ) 490 ( 39 nM, Figure
S3, Supporting Information). Unfortunately, the titration could
not be completed, again due to rapamycin’s limited solubility
in aqueous buffer. At high concentrations of rapamycin (>5
µM), we observed irregularities in the sensorgrams that were
indicative of a heterogeneous solution of analyte.
These results suggested that rapamycin would have to be
immobilized in order to complete the binding isotherm. We first
acylated the C40-cyclohexyl hydroxyl group with various
protected glycine derivatives. However, unmasking the amino
group led to rapid degradation of these derivatives. Therefore,
we designed a biotin-rapamycin derivative containing an amino-
hexanoic acid spacer (long chain, LC) between the two
molecules to minimize possible interactions between the two
proteins.31 Biotin-LC-rapamycin (Chart 1) proved to be stable
and displayed a high affinity for Neutravidin.
Figure 2. Fluorescence polarization assays. (A) A saturation binding
experiment between 2 nM Fl-rap and various concentrations of FRB
indicate a dissociation constant of 490 ( 39 nM. (B) Competition binding
experiments show the affinity of rapamycin for FRB in the presence (circles)
and absence (triangles) of FKBP. (C) Competition binding experiment shows
the affinity of rapamycin for FKBP (Kd ) 0.35 nM, R2 ) 0.96).
fit of the competition data, using a rapamycin‚FRB primary
dissociation constant of 5.2 µM, gave a secondary dissociation
constant for (FKBP‚rapamycin)‚FRB of 6.2 nM (R2 ) 0.97,
Figure 2B). Despite unanticipated difficulties in this experi-
mental system, these data suggest that there are significant
differences (∼1000-fold) between the affinities of rapamycin
alone and the FKBP‚rapamycin complex for FRB.
To validate both the fluorescence polarization method and
the mathematical fits of the competition data to derive dissocia-
tion constants, the affinity of rapamycin for FKBP was
determined using a competition binding assay (Figure 2C). The
experiment was performed using Fl-SLF (Fluorescein-labeled
Synthetic Ligand for FKBP, Chart S1, Supporting Information)
as a tracer.28 Fl-SLF was incubated with FKBP, along with
various concentrations of rapamycin. The data were fit using
the mathematical model, and the best fit gave a dissociation
constant of 0.35 nM (R2 ) 0.96). This value is similar to that
measured by other methods,8 validating the experimental method
and the mathematical treatment of the data.
For all SPR experiments, we used various surface loading
levels to allow for a more informed analysis of the resulting
data fit. When determining dissociation constants, it is ideal
for the concentration of the species being titrated to be at least
an order of magnitude below that of the dissociation constant,32
but with SPR, only an estimate of concentration on the surface
can be made. To be sure of the accuracy of the data fit, the
values measured across multiple loading levels and fitting
parameters (kinetic vs equilibrium binding) can be compared.
Neutravidin is a deglycosylated version of avidin with a nearly
neutral pI (6.3), which we used as a biotin-binding domain
(Figure 3A) as it yields the lowest nonspecific binding among
the known biotin-binding proteins. Neutravidin was immobilized
(29) Myszka, D. G. Methods Enzymol. 2000, 323, 325-340.
(30) Day, Y. S. N.; Baird, C. L.; Rich, R. L.; Myszka, D. G. Protein Sci. 2002,
11, 1017-1025.
(31) Wilchek, M.; Bayer, E. A. Methods Enzymol. 1990, 184, 123-138.
(32) Kenakin, T. Pharmacologic Analysis of Drug-Receptor Interaction, 3rd
ed.; Lippincott-Raven: Philadelphia, PA, 1997; Chapter 8.
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4717

A R T I C L E S
Banaszynski et al.
Figure 4. SPR analysis of the secondary dissociation constant of FRB for
GST-FKBP‚rapamycin. (A) Schematic showing the experimental design
(see text for details). (B) A representative data set showing an overlay of
both raw data and a kinetic fit for each FRB concentration. The GST-
FKBP‚rapamycin surface was exposed to various concentrations of FRB
(1-60 nM). A kinetic fit of the data for four independent experiments
provides an average dissociation constant of 12 ( 0.8 nM.
We were able to use the GST-capture method described above
to measure the affinity of FRB for the FKBP‚rapamycin
complex (Figure 4A). We prepared a GST-FKBP‚rapamycin
surface by first immobilizing an anti-GST antibody through
standard amine coupling to the dextran surface. Various levels
of GST-FKBP fusion protein were captured on active flow cells,
and matching levels of GST were captured on the reference
flow cells. The FKBP domains were then saturated with
rapamycin using 50 nM rapamycin in all buffers. Given the
dissociation constants of rapamycin for FKBP and FRB (Kd )
0.2 nM and Kd ) 26 µM, respectively), this concentration
ensures that all FKBP binding domains are saturated, while
rapamycin‚FRB complexation is negligible. Finally, various
concentrations of FRB, all with 50 nM rapamycin, were exposed
to the surface, and a representative sensorgram is shown in
Figure 4B. The titration was performed to saturation, but data
points from both high and low concentrations of analyte have
been omitted for clarity.
Figure 3. SPR analysis of the rapamycin‚FRB interaction. (A) Schematic
showing the experimental design (see text for details, NA ) neutravidin).
(B) A representative data set. Data shown is the response of the biotin-
LC-rapamycin surface to various concentrations of FRB (0.7-360 µM).
(C) Equilibrium binding analysis provides a rapamycin‚FRB dissociation
constant of 26 ( 0.8 µM.
using standard amine coupling on three active surfaces of a CM5
chip at loading levels of 250, 500, and 1000 RU. The final
surface had an immobilization level of 1000 RU Neutravidin
and served as the reference surface. The active surfaces were
then saturated with biotin-LC-rapamycin (10, 55, and 95 RU,
respectively), while the reference surface was saturated with
biotinyl-6-aminohexanoic acid. Various concentrations of FRB
were exposed to the surfaces, and a representative titration
sensorgram is shown in Figure 3B. A qualitative analysis of
the data shows characteristics of a micromolar interaction. The
association and dissociation rates are too fast to be accurately
measured by SPR, so we used an equilibrium binding analysis
to determine the dissociation constant. The data were transferred
into the BiaEvaluation software, and this analysis provided a
dissociation constant of 26 ( 0.8 µM (Figure 3C).
We prepared a second biotin-rapamycin conjugate using a
longer linker (Chart S2, Supporting Information) to determine
if tethering rapamycin too close to the surface attenuated the
affinity of FRB for the immobilized ligand. The dissociation
constant between FRB and the longer biotin-rapamycin con-
jugate was found to be nearly identical to that of biotin-LC-
rapamycin (23 µM and 26 µM, respectively, Figure S4,
Supporting Information), suggesting that surface effects do not
reduce the affinity of soluble FRB for the immobilized ligand.
A kinetic fit of the data shows that formation of trimeric
complex is quite favorable, with a dissociation constant of 12
( 0.8 nM (ka ) 1.92 × 106 M-1‚s-1, kd ) 2.2 × 10-2 s-1).
Thus, presentation of rapamycin by FKBP provides an ∼2000-
fold enhancement over the affinity of rapamycin alone for FRB.
An additional data set allowed for both an equilibrium binding
analysis and a kinetic fit of the data, with identical results (Figure
S5, Supporting Information).
It has been previously shown that when the C16-methoxy
group of rapamycin is replaced with a bulky trimethoxyphenyl
group (TMOP-rapamycin, Chart 1), the activity of the FKBP‚
TMOP-rapamycin complex in several biological assays is
9
4718 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

FKBP‚Rapamycin‚FRB Ternary Complex
A R T I C L E S
Figure 5. Effects of rapamycin and TMOP-rapamycin titrations on backbone 1H and 15N chemical shifts of FRB. Perturbations are displayed as normalized
weighted average shift differences, δavg/δmax (see Experimental Procedures). All experiments are normalized with respect to the residue that showed the
greatest perturbation in chemical shift (F2039) from the rapamycin‚FRB titration. Unassigned resonances are indicated as such by a negative chemical shift
difference. Ribbon representations were created using PyMOL.37 (A) Ribbon representation of FRB showing the rapamycin binding site (adopted from the
FKBP‚rapamycin‚FRB crystal structure, 1FAP).24 Rapamycin is shown in blue. (B) FRB ribbon representation showing residues affected by the rapamycin
titration. Side chains of residues that make van der Waals contacts with rapamycin are shown. Residues experiencing strong chemical shift perturbations
(δavg/δmax > 0.4) are shown in red and residues experiencing moderate chemical shift perturbations (0.2 < δavg/δmax < 0.4) are shown in pink. Rapamycin
has been omitted for clarity. (C) Chemical shift perturbations for the rapamycin‚FRB titration are plotted as a function of residue number. Residues that
make van der Waals contacts with rapamycin (L2031, S2035, Y2038, F2039, W2101, Y2105, and F2108) are indicated as such (×). The resonances for
E2032, S2035, L2037, and Y2038 (denoted with an O) broaden and become unobservable upon addition of rapamycin. These four residues have been
assigned a value of 1. Secondary structure is indicated at the bottom of the graph. (D) Chemical shift perturbations for the TMOP-rapamycin‚FRB titration
are plotted as a function of residue number.
reduced by at least 3 orders of magnitude, whereas the activity
of TMOP-rapamycin in a rotamase inhibition assay is decreased
only modestly.33 These results suggest that TMOP-rapamycin
retains its ability to bind to FKBP, but can no longer bind to
FRB because of unfavorable steric interactions. To demonstrate
the specificity of the rapamycin‚FRB interactions, we repeated
the experiment described above, substituting the rapamycin in
the buffer with TMOP-rapamycin. As expected, the response
levels observed from this titration are extremely low (data not
shown), demonstrating that the response observed upon FRB
titration is specific to the FKBP‚rapamycin surface.
interactions between FKBP and FRB contribute to the overall
stability of the ternary complex. To test this hypothesis, we
sought evidence of these interactions in the absence of rapa-
mycin. GST-FKBP was captured on an active flow cell, and a
matching level of GST was captured on the reference flow cell.
The FKBP surface was exposed to 5 µM FRB with no response
observed (data not shown), suggesting that the dissociation
constant between FKBP and FRB is no lower than 50 µM.
NMR Binding Assay. Chemical shifts are sensitive to local
chemical environment, and perturbations in chemical shift upon
titration are a good indication that two species are interacting.
As additional evidence of the specificity of the rapamycin‚FRB
interaction in the absence of FKBP, we used NMR spectroscopy
to monitor changes in chemical environment of FRB residues
upon complexation with rapamycin.
The affinity of the FKBP‚rapamycin complex for FRB is
approximately 2000-fold tighter than the affinity of rapamycin
alone for FRB. This observation suggests that protein-protein
(33) Luengo, J. I.; Yamashita, D. S.; Dunnington, D.; Beck, A. K.; Rozamus,
L. W.; Yen, H.-K.; Bossard, M. J.; Levy, M. A.; Hand, A,; Newman-Tarr,
T.; Badger, A.; Faucette, L.; Johnson, R. K.; D’Alessio, K.; Porter, T.;
Shu, A. Y. L.; Heys, R.; Choi, J.; Kongsaeree, P.; Clardy, J.; Holt, D. A.
Chem. Biol. 1995, 2, 471-481.
1
We used H/15N HSQC spectra to monitor rapamycin‚FRB
interactions, but first, we wanted to correlate the individual
HSQC resonances with specific residues within FRB. Three-
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4719

A R T I C L E S
Banaszynski et al.
dimensional NMR experiments were performed and analyzed
using uniformly 13C/15N-labeled FRB. Sample instability at the
higher concentrations necessary for 3D analysis by NMR in
addition to the flexibility of the R3 helix impacted the quality
of data acquired; however, a sequential assignment of 80% of
the FRB backbone was obtained (supplementary data).
Using uniformly 15N-labeled FRB, we monitored changes in
the amide chemical shifts of FRB as a function of rapamycin
concentration using 1H/15N HSQC spectra.34-36 An FRB sample
(100 µM FRB in 20 mM PBS pH 6.8, 10% D2O) was treated
with substoichiometric amounts of rapamycin (0.125 equiv) up
to a 1:1 ratio of rapamycin to FRB, and then in 0.5 equiv
increments to a final 2:1 ratio of rapamycin to FRB. HSQC
spectra were obtained for each titration point, and the spectra
were compared to monitor chemical shift perturbations.
The amide resonances of FRB displayed a range of behaviors
during the rapamycin titration, with peaks in fast, slow, and
intermediate exchange with respect to the NMR time scale.
Given the micromolar affinity of rapamycin for FRB, we might
expect that residues directly involved in a binding event would
exhibit slow exchange behavior. Our analysis of the rapamycin‚
FRB complex showed that residues that make van der Waals
contacts with rapamycin (L2031, S2035, Y2038, F2039, W2101,
Y2105, and F2108)24,25 are all either in slow exchange or are
broadened to such an extent upon binding that the resonances
are no longer observable.
The existence of the FKBP‚rapamycin‚FRB crystal struc-
ture24,25 coupled with our sequential assignment of FRB allowed
us to correlate chemical shift changes upon rapamycin binding
for 79 of 102 residues (Figure 5A-C). Figure 5A shows the
FKBP‚rapamycin‚FRB crystal structure, highlighting the rapa-
mycin-binding site. Figure 5B shows FRB in the same orienta-
tion. Rapamycin has been removed for clarity, and side chains
that make van der Waals contacts with rapamycin are explicitly
shown. Residues greatly affected by the rapamycin titration are
shown in red, and those moderately affected are shown in pink.
Chemical shift perturbations were then plotted as a function of
FRB sequence (Figure 5C). All residues known to be involved
in rapamycin binding were affected by rapamycin titration, and
the residues most affected by the titration lie along helices R1
and R4, the crossing of which forms the rapamycin binding
pocket.24
to this mixture of FRB and TMOP-rapamycin, chemical shift
perturbations similar to those observed at 1:1 rapamycin:FRB
complex were observed (Figure S6, Supporting Information),
showing the specificity of the rapamycin‚FRB interaction.
Our SPR results suggest that any interaction between FKBP
and FRB in the absence of rapamycin is weak (Kd g 50 µM).
We took advantage of NMRs ability to measure dissociation
constants in this range in an attempt to obtain a more quantitative
1
analysis. H/15N HSQC titration was performed in which 100
µM 15N-labeled FRB was titrated with FKBP in concentrations
up to 1 mM (data not shown). Even at these high protein
concentrations little perturbation in chemical shift was observed.
This finding suggests that FKBP and FRB do not appreciably
associate in the absence of rapamycin, despite the stabilizing
effects of protein-protein interactions at the interface of the
FKBP‚rapamycin‚FRB ternary complex.
Conclusions
We have used fluorescence polarization, surface plasmon
resonance, and NMR spectroscopy to characterize the inter-
molecular interactions involved in the formation of the FKBP‚
rapamycin‚FRB ternary complex. Of these four possible interac-
tions (Figure 1B), only the FKBP‚rapamycin interaction had
been previously characterized, and our measurements using two
independent techniques are in agreement with the literature value
(Kd ) 0.2 nM).8 The affinity of rapamycin for FRB in the
absence of FKBP is modest (Kd ) 26 ( 0.8 µM), and a 2000-
fold improvement is observed in the presence of FKBP (Kd )
12 ( 0.8 nM). On the basis of these measurements, one would
predict the affinity of the rapamycin‚FRB complex for FKBP
to be 100 fM. Although they appear to contribute significantly
to the stability of the ternary complex, protein-protein interac-
tions between FKBP and FRB in the absence of rapamycin were
not observed.
These results suggest that the FKBP‚rapamycin complex is
most likely the biologically relevant ligand for mTOR, as
rapamycin alone would not bind appreciably to mTOR under
physiological conditions. While FKBP is necessary for rapa-
mycin to bind to mTOR, these data do not rule out the possibility
that FKBP acts in an additional capacity to sterically block
interactions between mTOR and a functionally relevant protein
(i.e., substrate or signaling partner), with rapamycin modulating
this activity. The dramatic difference between the dissociation
constants of rapamycin alone and the FKBP‚rapamycin complex
for FRB suggests that protein-protein interactions at the
FKBP-FRB interface play a critical role in stabilizing the
ternary complex. A thorough analysis of the individual residues
at the FKBP-FRB interface will likely provide a more
quantitative and complete understanding of the individual
contributions to the stability of the ternary complex.
NMR experiments must be performed at relatively high
concentrations, so we were mildly concerned that these high
protein concentrations might lead to aggregation, exposing
hydrophobic surfaces that might interact with rapamycin non-
specifically. If rapamycin is simply binding to hydrophobic areas
of FRB, TMOP-rapamycin might be expected to exhibit the
same behavior. Conversely, if rapamycin is making specific
contacts with FRB, TMOP-rapamycin should be excluded from
this interaction for steric reasons. To ensure that the observed
chemical shift perturbations were due to a specific rapamycin‚
FRB interaction, we repeated the 1H/15N HSQC titration using
TMOP-rapamycin, with little perturbation in chemical shift
observed (Figure 5D). Upon addition of 1.0 equiv of rapamycin
Experimental Procedures
Fluorescence Polarization Binding Assays. (i) Saturation Binding
Experiments. The affinity of Fl-rap for FRB was determined as follows.
Fl-rap (2 nM) was incubated with various concentrations of purified
recombinant FRB (4 nM to 34 µM) in PBS pH 7.4, 0.011% Triton
X-100, 0.1 mg/mL bovine γ-globulin (BGG). Polarization units (mP)
were plotted against FRB concentration and fit in Kaleidagraph using
the equation F0 - (F0 - F∞)(Kd + [tracer]total + f ) - ((Kd + [tracer]total
+ f )2 - (4[tracer]total f ))0.5/(2[tracer]total), where F0 is the polarization
of free fluorescent tracer, F∞ is the polarization of bound
(34) Bodenhausen, G.; Ruben, D. J. Chem. Phys. Lett. 1980, 69, 185-189.
(35) Kay, L. E.; Keifer, P.; Saarinen, J. T. J. Am. Chem. Soc. 1992, 114, 10663-
10665.
(36) Shuker, S. B.; Hajduk, P. J.; Meadows, R. P.; Fesik, S. W. Science 1996,
274, 1531-1534.
(37) DeLano, W. L. The PyMOL Molecular Graphics System; DeLano
Scientific: San Carlos, CA, 2002 (http://www.pymol.org).
9
4720 J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005

FKBP‚Rapamycin‚FRB Ternary Complex
A R T I C L E S
tracer, [tracer]total is the total tracer concentration, and f is the measured
polarization at each protein concentration.
pH 2.2 (40 µL at 20 µL/min). The BiaEvaluation software program
was used for data fitting.
NMR Spectroscopy. 1H/15N HSQC two-dimensional spectra for the
rapamycin/FRB titration (described below) were acquired on a Varian
800 MHz Inova spectrometer. 15N-NOESY-HSQC, HNCACB, and
CBCACONH three-dimensional spectra were acquired on 13C/15N-
FRB (600 µM in 250 µL in Shigemi symmetrical microtube, 20 mM
PBS pH 6.8, 10% D2O) on Varian 600 and 800 MHz Inova
spectrometers. All experiments were acquired at 25 °C, and both
spectrometers were running VNMR v6.1C. Spectra were processed with
VNMR and analyzed utilizing Sparky.38
NMR Titration of Rapamycin into FRB. A sample of 15N-FRB
(500 µL, 100 µM) was prepared in NMR buffer (20 mM PBS pH 6.8,
10% D2O) and a 1H/15N HSQC spectrum of the free protein was
acquired. Rapamycin (1 µL of a 6.25 mM stock in ethanol, 0.125 equiv)
was added, and after 20 min incubation, an HSQC spectrum was
acquired. Rapamycin was added in 0.125 equiv aliquots (1 µL) up to
1.0 molar equiv with respect to FRB, and then two final additions (1
µL of a 25 mM stock, 0.5 equiv) were performed to achieve a final
2:1 ratio of rapamycin to FRB (total ethanol concentration of 2%).
HSQC spectra were acquired and analyzed for each titration point. A
control experiment using TMOP-rapamycin was performed in the same
manner, with an addition of 1.0 equiv of rapamycin (2 µL of a 25 mM
stock) as the final titration point. All experiments were performed at
25 °C.
(ii) Competition Binding Experiments. The affinity of rapamycin
for FRB was determined as follows. Fl-rap (10 nM) was incubated
with FRB (1 µM) in PBS pH 7.4, 0.011% Triton X-100, 0.1 mg/mL
BGG, along with various concentrations of rapamycin (14 nM to 10
µM). Polarization units were converted to fraction bound tracer and
plotted against rapamycin concentration. The data were then fit using
a quantitative mathematical model.28 The best data fit was determined
by R2 value.
The affinity of rapamycin for FKBP was determined as follows. Fl-
SLF (0.5 nM) was incubated with FKBP (5 nM) in PBS pH 7.4, 0.011%
Triton X-100, 0.1 mg/mL BGG, along with various concentrations of
rapamycin (0.05-100 nM). Polarization units were converted to fraction
bound tracer and plotted against rapamycin concentration. The data
were then fit using a quantitative mathematical model.28 The best data
fit was determined by R2 value.
(iii) Competition Binding Experiments with Presenter Protein.
The affinity of FKBP‚rapamycin for FRB was determined as follows.
Fl-rap (10 nM) was incubated with FRB (1 µM) and FKBP (1 µM) in
PBS pH 7.4, 0.011% Triton X-100, 0.1 mg/mL BGG, along with various
concentrations of rapamycin (1.4 nM to 1 µM). Polarization units were
converted to fraction bound tracer and plotted against rapamycin
concentration. The data were then fit using quantitative mathematical
model.28 The best data fit was determined by R2 value.
Surface Plasmon Resonance Assays. Surface plasmon resonance
measurements were performed at 25 °C using a Biacore 3000. All
buffers were filtered and degassed for 20 min. When changing buffers,
the system was equilibrated in the new buffer (primed 5×) before use.
All data were reference subtracted against both the reference flow cell
Chemical Shift Perturbation. The 1H/15N HSQC spectrum of FRB
complexed with rapamycin was compared with that of free FRB. To
evaluate the effects, the normalized weighted average shift difference
of the 1H and 15N resonances, δavg/δmax, for each residue was calculated.
2
1
The weighted average shift difference, δavg, was calculated as {[δ H
and a buffer injection. Kinetic data were fit to dR/dt ) kaC(Rmax
obs) - kdRobs, where C is the analyte concentration.
(i) Affinity of Biotin-LC-Rapamycin for FRB. Neutravidin was
-
+ (δ N/5)2]/2}1/2, where δ H and δ are the change in ppm between
15
1
15
N
R
free and bound chemical shifts. δmax was set to be the maximum
observed weighted average shift difference.39 All experiments were
normalized with respect to δmax from the rapamycin‚FRB titration series.
diluted in immobilization buffer (10 µg/mL in 10 mM sodium acetate,
pH 5.0) and immobilized onto three flow cells of a CM5 chip at 250,
500, and 1000 RU. Neutravidin (1000 RU) was immobilized on the
final flow cell, which served as the reference surface. Biotin-LC-
rapamycin was diluted into the running buffer (10 µg/mL in PBS pH
7.4, 0.002% Tween-20), and the three active flow cells were saturated
with this conjugate (10, 55, and 95 RU, respectively). The reference
cell was saturated with biotinyl-aminohexanoic acid. Various concentra-
tions of FRB (0.7-360 µM) were injected over the surface (50 µL/
min, 50 µL injection with 120 s wash delay). The surface was
regenerated between analyte injections with 35% EtOH (40 µL at 20
µL/min). The data were transferred into BiaEvaluation and fit using
an equilibrium binding analysis.
(ii) Affinity of GST-FKBP‚Rapamycin for FRB. Approximately
20 000 RU anti-GST antibody was immobilized on all four flow cells
of a CM5 chip using traditional amine coupling chemistry. GST-FKBP
was diluted into the running buffer (10 µg/mL in PBS pH 7.4, 0.002%
Tween-20, 50 nM rapamycin) and captured individually on two of the
antibody surfaces, at densities of 500 and 1000 RU, respectively. On
the remaining two antibody surfaces, 500 and 1000 RU of GST were
individually captured to give the corresponding reference surfaces.
Various concentrations of FRB (6 pM to 1 µM) were exposed to the
GST-FKBP‚rapamycin active surface and the GST reference surface
at a flow rate of 50 µL/min (250 µL injection, 300 s wash delay). The
surface was regenerated between analyte injections with 10 mM glycine
Acknowledgment. L.A.B. and T.J.W. thank the NIH (GM-
068589) and NSF (CHE-9985214) for funding of this work.
C.W.L. and the Stanford Magnetic Resonance Laboratory are
funded in part by the Stanford University School of Medicine.
We thank Professor Jie Chen for the pGEX-FKBP and pGEX-
FRB plasmids. We thank Professor Carolyn Bertozzi for use
of the Biacore 3000. We thank Dr. Patrick Braun and Professor
Jody Puglisi for helpful discussions.
Supporting Information Available: Supporting figures; ex-
perimental details for the synthesis of rapamycin derivatives
and spectra of the compounds; partial differential equations used
to fit competition binding data for dissociation constants;
additional details regarding SPR experiments; chemical shift
assignments for FRB backbone residues. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA043277Y
(38) Goddard, T. D.; Kneller, D. G. SPARKY 3; University of California: San
Francisco, 2002.
(39) Garrett, D. S.; Seok, Y. J.; Peterofsky, A.; Clore, G. M.; Gronenborn, A.
M. Biochemistry 1997, 36, 4393-4398.
9
J. AM. CHEM. SOC. VOL. 127, NO. 13, 2005 4721