A photoactivatable prenylated cysteine designed to study isoprenoid recognition

Article abstract of DOI:10.1021/ja0012016

Protein prenylation, involving the alkylation of a specific C-terminal cysteine with a C₁₅ or C₂₀ isoprenoid unit, is an essential posttranslational modification required by most GTP-binding proteins for normal biological activity. Despite the ubiquitous nature of this modification and numerous efforts aimed at inhibiting prenylating enzymes for therapeutic purposes, the function of prenylation remains unclear. To explore the role the isoprenoid plays in mediating protein-protein recognition, we have synthesized a photoactivatable, isoprenoid-containing cysteine analogue (2) designed to act as a mimic of the C-terminus of prenylated proteins. Photolysis experiments with 2 and RhoGDI (GDI), a protein which interacts with prenylated Rho proteins, suggest that the GDI is in direct contact with the isoprenoid moiety. These results, obtained using purified GDI as well as Escherichia coli (E. coli) crude extract containing GDI, suggest that this analogue will be an effective and versatile tool for the investigation of putative isoprenoid binding sites in a variety of systems. Incorporation of this analogue into peptides or proteins should allow for even more specific interactions between the photoactivatable isoprenoid and any number of isoprenoid binding proteins.

Full text of DOI:10.1021/ja0012016

VOLUME 123, NUMBER 19
MAY 16, 2001
© Copyright 2001 by the
American Chemical Society
A Photoactivatable Prenylated Cysteine Designed to Study Isoprenoid
Recognition
Tamara A. Kale,† Conrad Raab,‡ Nathan Yu,‡ Dennis C. Dean,‡ and Mark D. Distefano*,†
Contribution from the Department of Chemistry, UniVersity of Minnesota, Minneapolis, Minnesota 55455,
and Department of Drug Metabolism, Merck Research Laboratories, Rahway, New Jersey 07065
ReceiVed April 6, 2000
Abstract: Protein prenylation, involving the alkylation of a specific C-terminal cysteine with a C15 or C20
isoprenoid unit, is an essential posttranslational modification required by most GTP-binding proteins for normal
biological activity. Despite the ubiquitous nature of this modification and numerous efforts aimed at inhibiting
prenylating enzymes for therapeutic purposes, the function of prenylation remains unclear. To explore the role
the isoprenoid plays in mediating protein-protein recognition, we have synthesized a photoactivatable,
isoprenoid-containing cysteine analogue (2) designed to act as a mimic of the C-terminus of prenylated proteins.
Photolysis experiments with 2 and RhoGDI (GDI), a protein which interacts with prenylated Rho proteins,
suggest that the GDI is in direct contact with the isoprenoid moiety. These results, obtained using purified
GDI as well as Escherichia coli (E. coli) crude extract containing GDI, suggest that this analogue will be an
effective and versatile tool for the investigation of putative isoprenoid binding sites in a variety of systems.
Incorporation of this analogue into peptides or proteins should allow for even more specific interactions between
the photoactivatable isoprenoid and any number of isoprenoid binding proteins.
Introduction
which catalyze isoprenoid addition.5 Another means of halting
oncogenic signal transduction may involve disruption of the
interaction between prenylated proteins and other proteins
involved in signaling processes. Whereas prenyltransferase
inhibition affects all suitable G-protein substrates, disruption
of prenylated protein-protein interactions may be of greater
therapeutic interest since it may be possible to design compounds
to interfere with key contacts within a specific prenylated
protein-protein system.
Elimination of C-terminal prenylation, addition of an incorrect
isoprenoid, or substitution of the prenyl group with various
alternatives have been shown to either eliminate or alter normal
(4) (a) Du, W.; Lebowitz, P. F.; Prendergast, G. C. Mol. Cell. Biol. 1999,
19, 1831-1840. (b) Lin, R.; Cerione, R. A.; Manor, D. J. Biol. Chem. 1999,
274, 23633-23641. (c) Mira, J.-P.; Benard, V.; Groffen, J.; Sanders, L.
C.; Knaus, U. G. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 185-189.
(5) For recent reviews see: (a) Qian, Y.; Sebti, S. M.; Hamilton, A. D.
Biopoly. 1997, 43, 25-41. (b) End, D. W. InVest. New Drugs 1999, 17,
241-258.
Prenylated GTP-binding proteins (G-proteins) play a key role
in the progression of cancer with Ras oncoproteins being present
in roughly 30% of human tumors.1 For both normal and
cancerous activity, Ras proteins require addition of a C15
(farnesyl) isoprenoid to a specific C-terminal cysteine via a
thioether linkage,2 yet most members of the Ras-like superfamily
of G-proteins undergo geranylgeranylation (C20 addition) for
biological function.3 Other Ras superfamily members have also
been implicated in tumor development and require prenylation
for this activity as well.4 Efforts to prevent prenylation have
been pursued by developing inhibitors of the prenyltransferases
† University of Minnesota.
‡ Merck Research Laboratories.
(1) Barbacid, M. Annu. ReV. Biochem. 1987, 56, 779-827.
(2) Magee, T.; Marshall, C. Cell 1999, 98, 9-12.
(3) Chow, M.; Der, C. J.; Buss, J. E. Curr. Opin. Cell Biol. 1992, 4,
629-636.
10.1021/ja0012016 CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/21/2001

4374 J. Am. Chem. Soc., Vol. 123, No. 19, 2001
Kale et al.
biological activity in a wide variety of cases: Ras proteins,2,6
the γ-subunits of heterotrimeric G-proteins,7 and the yeast
a-factor mating dodecapeptide pheromone.8 These results and
other studies9-11 support the idea that isoprenoid binding sites
exist within proteins which interact with prenylated substrates.
The presence of such a pocket would impart a specificity
requirement for the types of proteins and molecules a G-protein
effector, activator, or receptor protein recognizes, thereby paving
the way for synthetic analogues of pharmaceutical relevance to
be produced. With the goal of understanding how prenylated
proteins are recognized by other proteins, we have developed a
photoaffinity labeling reagent designed to probe the interaction
between prenylated proteins and their cognate receptors.
To examine the effectiveness of using a photoaffinity labeling
compound for isoprenoid binding site investigation, photolysis
experiments were performed using Rho GDP-dissociation
inhibitor (GDI). This is a cytosolic protein which performs a
variety of regulatory activities on Rho proteins, including their
solubilization from membranes.12,13 Rho proteins are members
of the Ras superfamily of proteins which are farnesylated or,
more commonly, geranylgeranylated, and have been shown to
be necessary for oncogenic Ras transformation.4,14 The proposed
mechanism of Rho protein membrane extraction involves first
a binding step between GDI and Rho followed by a slower
isomerization step where the geranylgeranyl moiety moves from
the lipid membrane into an isoprenoid binding pocket of GDI.13
NMR studies of an amino-terminally truncated GDI suggest that
N-acetyl farnesylated cysteine (AFC, 1a) binds to the isoprenoid
binding site in GDI,10a,b and this type of interaction was recently
confirmed with an X-ray crystal structure, showing the geranyl-
geranyl portion of Cdc42 (a geranylgeranylated Rho protein)
buried in a hydrophobic pocket of GDI.11 Given these results,
we thought GDI was an appropriate choice for evaluating the
binding capabilities of our synthetic, photoactivatable isoprenoid
mimic.
Flash chromatography silica gel (60-120 mesh) was obtained from E.
M. Science or Scientific Adsorbents, Inc. CH2Cl2 was distilled from
CaH2. Deuterated NMR solvents were used as obtained from Cambridge
Isotope Laboratories, Inc. NMR spectra were obtained at 300 MHz
(1H) or 75 MHz (13C) on Varian instruments. Chemical shifts are
reported in ppm, and J values are given in Hertz. Protein concentrations
were determined by the Biorad method, which employs the Bradford
procedure,15 using BSA as a standard. Molecular masses of GDI and
GST-GDI were determined by comparison with high-molecular weight
standards (GIBCO BRL Life Technologies: myosin (200 kDa),
phosphorylase b (97.4 kDa), BSA (68 kDa), ovalbumin (43 kDa),
carbonic anhydrase (29 kDa), â-lactoglobulin (18.4 kDa), lysozyme
(14.3 kDa)) using SDS-polyacrylamide gel electrophoresis (PAGE)
analysis. Unless otherwise noted, HPLC analyses were carried out using
a Beckman model 127/166 instrument equipped with a diode array UV
detector and a Varian Dynamax C18 column (8.0 µm, 4.6 × 250 mm)
equipped with a 5 cm guard column (flow rate: 1.0 mL/min, 500 µL
injection loop, 5 to 100% B in 40 min. Solvent A: 95% H2O, 5%
CH3CN, 0.2% TFA; solvent B: 100% CH3CN, 0.2% TFA) with
monitoring at 220 and 260 nm. Phosphorimaging analysis was
performed with either a Molecular Dynamics STORM 840 instrument
and ImageQuaNT version 4.1 software or with a Bio-Rad Molecular
Imager FX instrument using Quantity One version 4.1.0 software.
(E,E)-8-O-(3-Benzoylbenzyl)-3,7-dimethyl-2,6-octadiene-1-bro-
mide (4). This method follows the basic published procedure which
starts from 3-methylbenzophenone and arrives at compound 4.16
However, the following changes were made to generate the bromide
from the alcohol precursor ((E,E)-8-O-(3-benzoylbenzyl)-3,7-dimethyl-
2,6-octadiene-1-ol, 3): polymer-supported PPh3 beads17 (Aldrich; 2
equiv) were soaked in 3 mL of dry CH2Cl2 along with 3 (1 equiv) for
30 min to allow the beads to swell. CBr4 (1.2 equiv) was then added
and the reaction stirred for 5 h. TLC analysis (5:2 toluene/EtOAc, Rf
) 0.88) showed the reaction was complete at this time, and the beads
were removed by filtration. The filtrate was evaporated, and purification
was not necessary; product was present in >90% purity as determined
1
1
by H NMR. H NMR (CDCl3) δ 1.65 (s, 3H), 1.71 (s, 3H), 2.07-
2.21 (m, 4H), 3.91 (s, 2H), 3.95 (d, J ) 8.4, 2H), 4.48 (s, 2H), 5.38 (t,
J ) 6.5, 1H), 5.51 (t, J ) 8.4, 1H), 7.25-7.80 (m, 9H). 13C-DEPT
NMR (CDCl3) δ 14.1, 16.0 (primary), 25.8, 29.6, 39.1, 71.1, 76.4
(secondary), 120.9 (worth 2 C), 127.6, 128.3, 128.4, 129.2, 129.3, 130.1,
130.4, 131.7, 132.7 (tertiary), 137.6, 137.7, 139.1, 143.1, 196.6
(quarternary). HR-FAB-MS calcd for C24H28O2Br [M + H]+ 427.1265,
found 427.1251.
(7) (a) For a review see: Gautam, N.; Downes, G. B.; Yan, K.; Kisselev,
O. Cell. Signal. 1998, 10, 447-455. (b) Myung, C.-S.; Yasuda, H.; Liu,
W. W.; Harden, T. K.; Garrison, J. C. J. Biol. Chem. 1999, 274, 16595-
16603 and references within.
(8) Epand, R. F.; Xue. C.-B.; Wang, S.-H.; Naider, F.; Becker, J. M.;
Epand, R. M. Biochemistry 1993, 32, 8368-8373.
(9) (a) Cox, A. D.; Der, C. J.; Curr. Opin. Cell Biol. 1992, 4, 1008-
1016. (b) See footnote 6a. (c) Thissen, J. A.; Casey, P. J. J. Biol. Chem.
1993, 268, 13780-13783. (d) Marshall, C. J. Science 1993, 259, 1865-
1866. (e) Siddiqui, A. A.; Garland, J. R.; Dalton, M. B.; Sinensky, M. J.
Biol. Chem. 1998, 273, 3712-3717. (f) Zhang, L.; Tschantz, W. R.; Casey,
P. J. J. Biol. Chem. 1997, 272, 23354-23359.
(10) (a) Gosser, Y. Q.; Nomanbhoy, T. K.; Aghazadeh, B.; Manor, D.;
Combs, C.; Cerione, R. A.; Rosen, M. K. Nature 1997, 387, 814-819. (b)
Keep, N. H.; Barnes, M.; Barsukov, I.; Badii, R.; L-Yun, L.; Segal, A. W.;
Moody, P. C. E.; Roberts, G. C. K. Structure 1997, 5, 623-633. (c)
Aharonson, Z.; Gana-Weisz, M.; Varsano, T.; Haklai, R.; Marciano, D.;
Kloog, Y. Biochim. Biophys. Acta. 1998, 1406, 40-50.
(11) Hoffman, G. R.; Nassar, N.; Cerione, R. A. Cell 2000, 100, 345-
356.
(12) For a recent review of RhoGDIs, see: Olofsson, B. Cell. Signal.
1999, 11, 545-554.
Experimental Section
(13) Nomanbhoy, T. K.; Erickson, J. W.; Cerione, R. A. Biochemistry
1999, 38, 1744-1750.
Materials and Methods. All synthetic reactions were conducted
under a nitrogen atmosphere, stirred magnetically, and carried out at
room temperature unless otherwise noted. Analytical TLC was per-
formed on precoated (250 µm) silica gel 60 F-254 plates from E. Merck.
Visualization of plates was done under UV irradiation or by staining
with an ethanolic phosphomolybdic acid solution followed by heating.
(14) (a) Qiu, R.-G.; Chen, J.; McCormick, F.; Symons, M. Proc. Natl.
Acad. Sci. U.S.A. 1995, 92, 11781-11785. (b) Khosravi-Far, R.; Solski, P.
A.; Clark, G. J.; Kinch, M. S.; Der, C. J. Mol. Cell. Biol. 1995, 15, 6443-
6453.
(15) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.
(16) Gaon, I.; Turek, T. C.; Distefano, M. D. Tetrahedron Lett. 1996,
37, 8833-8836.
(6) McGeady, P.; Porfiri, E.; Gelb, M. H. Bioorg. Med. Chem. Lett. 1997,
7, 145-150.
(17) Bernard, M.; Ford, W. T. J. Org. Chem. 1983, 48, 326-332.

PhotoactiVatable Isoprenoid
J. Am. Chem. Soc., Vol. 123, No. 19, 2001 4375
Cysteine (S-(E,E)-8-O-(3-Benzoylbenzyl)-3,7-dimethyl-2,6-octa-
diene) Methyl Ester (5). Following literature precedent,18 bromide 4
(85.1 mg, 0.20 mmol, 1 equiv) was dissolved in 180 µL of DMF and
allowed to stir. Cysteine methyl ester hydrochloride (51.1 mg, 0.30
mmol, 1.5 equiv) was then added followed by 55.8 µL of DIEA (0.32
mmol, 1.6 equiv). The reaction continued to stir for roughly 20 h at
which time it was diluted with ethyl acetate and washed twice with
water. The organic layers were combined and dried over anhydrous
Na2SO4, and the solvent was evaporated. After flash chromatography
(1:1 toluene/EtOAc, Rf ) 0.23), product was recovered in 40% yield
(38.5 mg). 1H NMR (CDCl3) δ 1.66 (s, 6H), 2.07-2.18 (m, 4H), 2.67
(dd, J ) 13.5, 7.8, 1H), 2.87 (dd, J ) 15.0, 4.5, 1H), 3.16 (m, 2H),
3.65 (m, 1H), 3.73 (s, 3H), 3.91 (s, 2H), 4.49 (s, 3H), 5.23 (t, J ) 7.5,
1H), 5.40 (m, 1H) 7.43-7.50 (m, 3H), 7.57-7.59 (m, 2H), 7.68-7.71
(m, 1H), 7.77-7.80 (m, 3H). 13C-DEPT NMR (CDCl3) δ 14.0, 14.2,
52.2 (primary), 26.1, 29.8, 36.5, 39.2, 70.9, 76.5 (secondary), 120.3,
128.0, 128.3 (worth 2 C), 129.1, 129.3, 130.3 (worth 2 C), 131.7, 132.5
(tertiary), 132.2, 137.6, 137.7, 139.0, 139.1, 174.6, 196.6 (quaternary).
HR-FAB-MS calcd for C28H36NO4S [M + H]+ 482.2356, found
482.2362.
N-Methanesulfonyl Cysteine (S-(E,E)-8-O-(3-Benzoylbenzyl)-3,7-
dimethyl-2,6-octadiene) Methyl Ester (6). The following procedure
is a modification of a published protocol.19 Cysteine (S-(E,E)-8-O-(3-
benzoylbenzyl)-3,7-dimethyl-2,6-octadiene) methyl ester 5 (189.9 mg,
0.39 mmol, 1 equiv), was dissolved in 200 µL of freshly distilled
CH2Cl2 followed by addition of Et3N (109 µL, 0.78 mmol, 2 equiv).
Methanesulfonyl chloride was then added slowly (51.3 µL, 0.66 mmol,
1.7 equiv) over a period of 1 min, and the reaction quickly became
more yellow. Additional CH2Cl2 (0.5 mL) was added, and TLC analysis
showed that the reaction was essentially complete in 10 min. After 1
h, the reaction was diluted with CH2Cl2 and washed two times with
saturated NaHCO3. The aqueous layers were combined and washed
one time with CH2Cl2, and all organic layers were combined, dried
over anhydrous Na2SO4, and evaporated. Flash chromatography (5:2
toluene/EtOAc, Rf ) 0.42) was performed, and 154.8 mg of product
were recovered, a 71% yield. 1H NMR (CDCl3) δ 1.67 (s, 6H), 2.03-
2.21 (m, 4H), 2.83 (dd, J ) 13.8, 6.3, 1H), 2.90 (dd, J ) 13.9, 5.2,
1H), 3.00 (s, 3H), 3.15 (dd, J ) 13.1, 7.6, 1H), 3.22 (dd, J ) 13.2,
7.8, 1H), 3.76 (s, 3H), 3.91 (s, 2H), 4.32 (dt, J ) 8.6, 5.8, 1H), 4.51
(s, 2H), 5.21 (t, J ) 7.7, 1H), 5.39 (t, J ) 6.3, 1H), 5.48 (d, J ) 8.4,
1H), 7.42-7.49 (m, 3H), 7.55-7.60 (m, 2H), 7.68-7.70 (m, 1H), 7.78-
7.80 (m, 3H). 13C-DEPT NMR (CDCl3) δ 14.0, 16.2, 41.9, 53.0
(primary), 26.0, 30.0, 34.4, 39.2, 71.0, 76.5 (secondary), 55.8, 119.8,
127.9, 128.3 (worth 2 C), 129.2, 129.4, 130.1 (worth 2 C), 131.8, 132.5
(tertiary), 132.3, 137.6, 137.7, 139.1, 139.9, 171.2, 196.8 (quaternary).
HR-FAB-MS calcd for C29H38NO6S2 [M + H]+ 560.2131, found
560.2106. Purity analysis by analytical scale reversed-phase HPLC
indicated >85% pure material when the reaction was performed at this
scale (tR ) 40.5 min). An extinction coefficient (254 nm) value of
14 800 M-1 cm-1 was determined in CH3CN.
129.7 (worth 2 C), 131.7, 132.5 (tertiary), 131.9, 137.5 (worth 2 C),
138.6, 139.2, 174.1, 197.1 (quaternary). HR-FAB-MS calcd for C28H36-
NO6S2 [M + H]+ 546.1975, found 546.1975; for [M + Na]+: calcd
568.1795, found 568.1781. Purity analysis by analytical scale reversed-
phase HPLC indicated a purity of >90% when the reaction was
performed at this scale (tR ) 37.0 min). An extinction coefficient (256
nm) value of 17 700 M-1 cm-1 was determined in MeOH.
[35S]N-Methanesulfonyl Cysteine (S-(E,E)-8-O-(3-Benzoylbenzyl)-
3,7-dimethyl-2,6-octadiene) Methyl Ester. The title compound was
prepared using the procedure of Dean et al.19 A solution of [35S]methane-
sulfonyl chloride (24 mCi, 1100 Ci/mmol) in anhydrous CH2Cl2 (9
mL) was dried over Na2SO4 for 1 h, filtered, then concentrated by
atmospheric distillation to 120 µL. This [35S]MsCl concentrate was
added to a solution of cysteine (S-(E,E)-8-O-(3-benzoylbenzyl)-3,7-
dimethyl-2,6-octadiene) methyl ester (4.8 mg, 9.9 µmol) 5 and Et3N
(5 µL) in anhydrous CH2Cl2 (100 µL). After stirring 1 h, the reaction
was quenched by the addition of saturated aqueous NaHCO3 (2 mL)
and diluted with CH2Cl2 (2 mL). The layers were separated, and the
aqueous layer was extracted with CH2Cl2 (4 × 2 mL). The combined
organic layers were concentrated, and the residue was dissolved in
MeOH (2.5 mL) and the amount of radioactivity determined using a
liquid scintillation counter (7.9 mCi). Analysis of the solution by
analytical HPLC (Zorbax RX-C18 4.6 × 250 mm column, 1 mL/min,
30 °C, 210/254 nm, Packard flow monitor, A ) 0.1% aqueous TFA,
B ) CH3CN, 50A/50B linear gradient to 0A/100B over 30 min, tR
amine ) 8.5 min, tR [35S]sulfonamide ) 19.5 min) showed the
radiochemical purity of the [35S]sulfonamide to be 37%. The identity
of the product was confirmed by coelution with an authentic sample
of unlabeled sulfonamide. The crude [35S]N-methanesulfonyl cysteine
(S-(E,E)-8-O-(3-benzoylbenzyl)-3,7-dimethyl-2,6-octadiene) methyl es-
ter was used without further purification in the next step.
[35S]N-Methanesulfonyl Cysteine (S-(E,E)-8-O-(3-Benzoylbenzyl)-
3,7-dimethyl-2,6-octadiene) ([35S]2). To the above methanolic solution
of [35S]N-methanesulfonyl cysteine (S-(E,E)-8-O-(3-benzoylbenzyl)-
3,7-dimethyl-2,6-octadiene) methyl ester was added THF (3.0 mL),
followed by a solution of LiOH‚H2O (30 mg) in H2O (1.0 mL). The
reaction mixture was stirred for 3.5 h at which stage HPLC indicated
that reaction was complete. The reaction mixture was acidified with
0.15 M HCl (20 mL) and diluted with EtOAc (20 mL). The layers
were separated, and the aqueous layer was extracted with EtOAc (4 ×
5 mL). The combined organic layers were concentrated, and the residue
was dissolved in CH3CN (5 mL), counted (4.46 mCi), and analyzed
by HPLC (Zorbax RX-C18, 4.6 × 250 mm, 5 µm, 1 mL/min, 30 °C,
210/254 nm, Packard flow monitor, A ) 0.1% aqueous TFA, B )
CH3CN, 50A/50B linear gradient to 0A/100B over 30 min, tR
[35S]sulfonamide ester ) 19.5 min, tR [35S]2 ) 14.9 min, 38%
radiochemical purity). Purification was effected by RP-HPLC (Zorbax
SB-Phenyl, 9.4 × 250 mm, 5 µm, 5 mL/min, 20 °C, 210/254 nm,
Packard flow monitor, A ) 0.1% aqueous TFA, B ) CH3CN, 52A/
48B isocratic, tR [35S]2 ) 27 min). The purified [35S]2 was isolated by
SepPak, concentrated, taken up in 2:1 CH3CN/H2O (1.5 mL), counted
(390 µCi), and analyzed by HPLC (Zorbax SB-Phenyl, 4.6 × 250 mm,
5 µm, 1 mL/min, 30 °C, 210/254 nm, Packard flow monitor, A ) 0.1%
aqueous H3PO4, B ) CH3CN, 46A/54B isocratic, tR [35S]2 ) 16.5 min,
97.5% radiochemical purity). The identity of the [35S]N-methanesulfonyl
cysteine (S-(E,E)-8-O-(3-benzoylbenzyl)-3,7-dimethyl-2,6-octadiene)
was confirmed by coelution with an authentic sample of unlabeled 2.
N-Methanesulfonyl Cysteine (S-(E,E)-8-O-(3-Benzoylbenzyl)-3,7-
dimethyl-2,6-octadiene) (2). N-Methanesulfonyl cysteine (S-(E,E)-8-
O-(3-benzoylbenzyl)-3,7-dimethyl-2,6-octadiene) methyl ester 6 (132.7
mg, 0.24 mmol, 1 equiv) was dissolved in a 0.5 M LiOH solution
consisting of 1:1 iPrOH/H2O (4.75 mL, 2.4 mmol LiOH, 10 equiv).
After stirring for 24 h, the reaction was acidified to pH 5 with 10%
HCl and then dissolved in EtOAc, followed by three washings with
water. The organic layer was dried over anhydrous Na2SO4 and then
evaporated, leaving the product as 81.6 mg of a thick, dark yellow-
Ac-N-KKSRRC(S-Geranylgeranyl)-OH. The unmodified peptide
was synthesized and cleaved from the resin as a C-terminal carboxylic
acid at the Microchemical Facility at the University of Minnesota,
Minneapolis, MN. Alkylation, using geranylgeranyl bromide, was
performed as previously described.20 Purity analysis by analytical scale
reversed-phase HPLC (Vydac C4 column (8.0 µm, 4.6 × 250 mm)
equipped with a 5 cm guard column (flow rate: 1.0 mL/min, 500 µL
injection loop, 5 to 100% B in 40 min. Solvent A: 95% H2O, 5%
CH3CN, 0.2% TFA; solvent B: 100% CH3CN, 0.2% TFA) with
monitoring at 220 and 260 nm) indicated a purity of >85% (tR ) 27.1
min). MALDI-TOF-MS calcd for C52H95N1409S [M + H]+ 1091.71,
1
colored oil (60% yield). H NMR (CD3OD) δ 1.64 (s, 3H), 1.67 (s,
3H), 2.13 (m, 4H), 2.73 (dd, J ) 13.8, 7.2, 1H), 2.88 (dd, J ) 13.8,
5.1, 1H), 2.96 (s, 3H), 3.18 (dd, J ) 12.9, 7.8, 1H), 3.24 (dd, J )
12.9, 8.0, 1H), 3.91 (s, 2H), 4.05 (m, 1H), 4.47 (s, 2H), 5.22 (t, J )
7.8, 1H), 5.39 (t, J ) 7.0, 1H), 7.46-7.70 (m, 9H). 13C-DEPT NMR
(CD3OD) δ 12.8, 14.9, 39.9 (primary), 25.6, 29.3, 34.4, 38.9, 70.2,
76.1 (secondary), 57.4, 120.7, 128.1, 128.2 (worth 2 C), 128.8, 128.9,
(18) Yang, C.-C.; Marlowe, C. K.; Kania, R. J. Am. Chem. Soc. 1991,
113, 3177-3178.
(19) Dean, D. C.; Nargund, R. P.; Pong, S.-S.; Chaung, L.-Y. P.; Griffin,
P.; Melillo, D. G.; Ellsworth, R. L.; Van Der Ploeg, L. H. T.; Patchett, A.
A.; Smith, R. G. J. Med. Chem. 1996, 39, 1767-1770.
(20) Xue, C.-B.; Becker, J. M.; Naider, F. Tetrahedron Lett. 1992, 33,
1435-1438.

4376 J. Am. Chem. Soc., Vol. 123, No. 19, 2001
Kale et al.
found 1091.44. ESI-MS for [M + H]+ found 1091.6; [M + 2H]2+ calcd
546.3, found 546.6; [M + 3H]3+ calcd 364.6, found 364.8. ESI-MS-
MS resulted in the prenyl group fragmenting first, leaving the original
peptide (calcd [M + H]+ 819.5, found 819.4). The entire peptide was
sequenced using multiple MS-MS (MS3, MS4) fragmentation methods
(see Supporting Information).
buffer solutions (20 mM HEPES, pH 8.0, 5 mM MgCl2, 1 mM NaN3,
100 mM NaCl) and added to the photolysis mixtures as needed. The
percent DMSO in all photolysis reactions remained under 10%. GDI
buffer was used for reactions. All reagents (buffer, protein, [35S]2,
competitor as applicable) were combined, mixed gently, and im-
mediately photolyzed. Irradiation times up to 10.7 h resulted in no
decomposition of GDI as detected by SDS-PAGE, and cross-linking
only occurred when the mixtures were irradiated. Following photolysis,
unless otherwise indicated, loading buffer (50 mM Tris-HCl, pH 6.8,
100 mM dithiothreitol, 2% SDS, 0.1% bromophenol blue, 10% glycerol)
was added to each sample, and samples were heated for 2 min at 100
°C followed by analysis by SDS-PAGE using Tris-glycine gels (12 or
15% acrylamide). Unincorporated [35S]2 passed through the gel and
did not interfere with analysis of most cross-linked protein products,
but upper and lower portions of each gel were removed before
autoradiography in order to facilitate analysis. Gels were then stained
with Coomassie brilliant blue or SYPRO stain (Bio-Rad), dried, and
subjected to autoradiography. Intensities of radiolabeled products were
determined by phosphorimaging analysis of the dried gels.
Immunoprecipitation of Cross-Linked GST-GDI. Two samples
of crude GST-GDI-induced cell lysate described above (30-40 µg
each) were combined with the typical amount of [35S]2 (2.9 pmol) in
total volumes ranging from 30 to 50 µL using RIPA buffer (10 mM
Tris-HCl, pH 7.8, 150 mM NaCl, 600 mM KCl, 5 mM EDTA, 3 mM
PMSF, 1 µg/mL aprotinin, 2% Triton X-100 (v/v)) as desired. After
photolysis (0.5 h), an equivolume amount of loading buffer was added
to one sample, and following heating of the sample for 2 min at 100
°C, the material was stored at -20 °C. The second sample was
transferred to a small microcentrifuge tube, and a small magnetic stir
bar was added. Anti-GST antibodies (Amersham Pharmacia Biotech;
4 µL at 5 mg/mL) were added, and the material was slowly stirred for
6 h at 4 °C.
Preparation of GST-GDI Escherichia coli (E. coli)/pGEX-KG
Crude Cell Lysates; Purification of GDI. GST-GDI was expressed
in E. coli using a modified protocol,21 and the following represents a
typical procedure (cell growth time and final cell densities varied with
each preparation; see text). A culture plate, supplemented with
ampicillin at 100 µg/mL, was streaked with a frozen culture of JM-
109 cells containing the pGEX-KG GDI plasmid and incubated
overnight at 37 °C. A 2 mL overnight culture was grown from a selected
colony, and following overnight incubation at 37 °C with vigorous
shaking (275 rpm), the saturated culture was diluted into 100 mL of
LB media (10 g/L NaCl, 10 g/L tryptone, 5 g/L yeast extract) containing
100 µg/mL ampicillin. This culture was incubated at 37 °C with
vigorous shaking (275 rpm) until the OD560 reached 0.7 at which time
one-half of the material was removed for later experiments and termed
the “uninduced sample.” Isopropyl-â-D-thiogalactopyranoside (IPTG)
was added to a final concentration of 100 µM, and the incubation was
continued for an additional 2-3 h. Uninduced and induced cells were
harvested by centrifugation (10 min at 5000g and then resuspended in
lysis buffer (20 mM Tris-HCl, pH 8.0, 50 mM EDTA, 0.2 mM
phenylmethanesulfonyl chloride (PMSF), 10 µg/mL aprotinin, 10 µg/
mL leupeptin, 10 µM benzamidine), followed again by centrifugation.
Cells of both types not being immediately used were flash frozen with
N2 (l) and stored at -80 °C until further use. Cell lysis of both types
was carried out by sonication (3 × 30 s) in fresh lysis buffer followed
by centrifugation (10 min at 12000g, thus producing the crude soluble
protein extract.
During this time, Gammabind Sepharose (Amersham Pharmacia
Biotech) resin (100-200 µL as a 50/50 (v/v) slurry) was washed
sequentially with the following: 0.5 M AcOH, pH 3.0; 1.0 M AcOH,
pH 2.5; binding buffer (0.01 M NaH2PO4, 0.15 M NaCl, 0.01 M EDTA,
pH 7.0) and then equilibrated in RIPA buffer. Equilibrated resin was
then added to the crude lysate/antibody mixture. Stirring then continued
for an additional 16 h at 4 °C at which time the stir bar was removed,
and the vessel was centrifuged (10 min at 12000g), leaving the
immunoprecipitated GST-GDI on resin as a pellet. Resin was washed
with RIPA buffer, and the material was centrifuged again, followed
by removal of the supernatant.
Photolysis of Commercially Available Proteins. A mixture of
commercially available proteins (GIBCO BRL Life Technologies) was
prepared at either 0.1 or 0.4 mg/mL concentrations and photolyzed in
the presence of [35S]2 (58 nM) in GDI buffer (50 µL total volume) for
0.5 h. The samples were treated as described in the General Photolysis
Procedure although cutting of the upper and lower portions of the gel
removed the following proteins and prevented their analysis by
autoradiography: myosin, phosphorylase b, and lysozyme (see Sup-
porting Information).
Purification of the fusion protein, when desired, was performed using
Glutathione Sepharose 4B resin (Amersham Pharmacia Biotech; 0.5
mL) equilibrated first with PBS buffer (137 mM NaCl, 2.7 mM KCl,
4.3 mM Na2HPO4, 1.4 mM KH2PO4, pH 7.3) and then TEDA buffer
(20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM DTT, 1 mM NaN3)
plus 100 mM NaCl and 0.2 mM PMSF. Crude induced soluble extract
(250-500 µL) was applied to the column resin, and undesired proteins
were eluted by washing with TEDA, 100 mM NaCl, and 0.2 mM
PMSF. The fusion protein was then eluted using TEDA, 100 mM NaCl,
0.2 mM PMSF, and 10 mM glutathione.
In cases where isolation of GDI was desired, thrombin (Amersham
Pharmacia Biotech; 1 unit/µL activity) was used to cleave GDI while
GST-GDI was still bound to the glutathione affinity resin after
unwanted proteins were removed by washing with TEDA. Cleavage
was monitored by SDS-PAGE. To further purify GDI, literature
precedent was followed,21 employing a Mono-Q HR 5/5 ion-exchange
column (Amersham Pharmacia Biotech).
Preparation of E. coli HB101 Crude Cell Lysate. E. coli HB101
competent cells were purchased from Transfinity, BRL Life Technolo-
gies, Inc. Colony development, culture growth, centrifugations, cell
lysis, and preparation of the crude extract for photolysis were performed
in the same manner as described for uninduced GST-GDI except for
the removal of ampicillin from the procedure.
General Photolysis Procedure. All reactions (50 µL final volume
unless otherwise noted) were performed in silinized quartz test tubes
covered with Parafilm at 4 °C in a UV Rayonet minireactor equipped
with 8 RPR-3500° lamps and a circulating platform that allowed for
irradiation of up to eight samples simultaneously. The amount of [35S]2
added to photolysis mixtures, unless otherwise noted, came from stock
solutions in DMSO to yield 2.9 pmol of material (58 nM in 50 µL).
Competitor stock solutions were also prepared as either DMSO or GDI
Determination of [35S]2 Percent Incorporation Following Pho-
tolysis with GDI. A solution of unlabeled 2 was prepared in DMSO
and added to 2.9 pmol of [35S]2 (the typical amount used for photolysis;
see General Photolysis Procedure) such that the final specific activity
was 8 Ci/mmol. This solution was used to prepare a photolysis reaction
mixture containing an equimolar ratio of GDI to total compound 2 (each
at 4.0 µM). Following the General Photolysis Procedure, the phosphor-
imaged gel was compared to a phosphorimaged standard which
consisted of 2.9 pmol of [35S]2, unincubated with protein and
unphotolyzed, applied to a separate gel of the same percentage
acrylamide. This unincorporated material was allowed to run partway
through the gel and, by phosphorimaging analysis, migrated as a single
spot. Imaging of this gel in the same manner (i.e., the same cassette)
as the gel containing the photolyzed protein allowed for direct
comparisons to the standard by phosphorimaging analysis.
(21) We used a truncated GDI whose first 25 amino-terminal residues
had been removed. This modification increases the stability of the protein
(G. R. Hoffman, personal communication) and behaves similarly to the wild-
type protein (Platko, J. V.; Leonard, D. A.; Adra, C. N.; Shaw, R. J.; Cerione,
R. A.; Lim, B. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 2974-2978). For
expression and purification of GST-GDI and GDI alone, see: Leonard,
D.; Hart, M. J.; Platko, J. V.; Eva, A.; Henzel, W.; Evans, T.; Cerione, R.
A. J. Biol. Chem. 1992, 267, 22860-22868.
Results
Design and Synthesis. Prenylated cysteine analogues includ-
ing AFC (1a) have been used to examine prenylated protein-

PhotoactiVatable Isoprenoid
J. Am. Chem. Soc., Vol. 123, No. 19, 2001 4377
Scheme 1
protein behavior10 and in several cases, have been shown to
significantly alter these interactions.22 These results, along with
the desire to create a universal probe which could be used with
a wide variety of proteins thought to interact with prenylated
proteins, led us to design compound 2. Benzophenone-contain-
ing isoprenoid analogues have been shown to be effective
mimics of the isoprenoidal portion of prenyl diphosphates;23
recently, a crystal structure of such an analogue bound in the
farnesyl diphosphate binding site of farnesyltransferase has been
obtained.24 Thus, we incorporated this photoactivatable group,
with a C10 isoprenoid spacer, into a cysteine residue via a
thioether linkage. Cross-linking between 2 and a putative
isoprenoid binding protein should provide evidence for protein-
isoprenoid interactions and may also allow for the eventual
identification of amino acid residues important for isoprenoid
binding.25 To facilitate identification of these cross-linked
products, an 35S-labeled sulfonamide was introduced in lieu of
the N-acetyl group used in many analogues.10,22 Compound 2
was prepared using a geranyl bromide derivative (4) containing
the benzoylbenzyl moiety linked to it via a stable ether linkage
(see Scheme 1).16 This linkage has the advantage of withstand-
ing a variety of conditions, an important feature not only for
the purposes of generating 2 and compounds similar to it, but
also for the eventual process of identifying cross-linked residues.
Compound 3 was prepared as previously reported16 and
converted to 4 using a modified procedure employing resin-
bound PPh3.17 We have found this reagent extremely useful for
the preparation of allylic bromides since the only major impurity,
resin-bound triphenylphosphine oxide, is easily removed by
filtration in lieu of chromatography. Cysteine methyl ester was
Figure 1. Analysis of photolabeling of GDI by [35S]2 by SDS-
polyacrylamide gel electrophoresis (PAGE). Lanes 1 and 2 show protein
staining, while lanes 1′ and 2′ show the radiolabeled proteins. Lanes 1
and 1′ contain a sample of GDI (4.0 µM) irradiated in the presence of
[35S]2 (116 nM); lanes 2 and 2′ contain a sample of GDI (4.0 µM)
irradiated in the presence of 1a (10 mM) and [35S]2 (116 nM).
Irradiation time was 0.5 h.
alkylated with the allylic bromide 4, followed by sulfonami-
dation of the free amine and saponification of the methyl ester.
Using methane[35S]sulfonyl chloride to form the sulfonamide
linkage,19 [35S]2 was synthesized with a specific activity of 1100
Ci/mmol from the free amine precursor and purified using
reversed-phase HPLC chromatography. The product was ob-
tained in 97.5% radiochemical purity and matched the retention
time of unlabeled probe as determined by reversed-phase HPLC.
Preliminary Photolabeling Experiments with AFC Ana-
logue [35S](2). To investigate the cross-linking ability of our
probe to GDI, [35S]2 was first photolyzed in the presence of a
purified sample of GDI.21 Photolysis experiments using 58-
116 nM probe in the presence of 4.0 µM GDI resulted in
successful cross-linking (Figure 1, lanes 1 and 1′). Inclusion of
10 mM AFC (1a) in the photolysis reaction (Figure 1, lanes 2,
2′) resulting in a 5-7-fold decrease in the amount of radiola-
beling (ca. 15% of original, lanes 1, 1′). AFC was initially
chosen as a competitor with [35S]2 due to its perturbation of
putative prenyl binding site residues of GDI in the NMR study
(22) (a) Scheer, A.; Gierschik, P. Biochemistry 1995, 34, 4952-4961.
(b) Parish, C. A.; Brazil, D. P.; Rando, R. R. Biochemistry 1997, 36, 2686-
2693. (c) Kilic, F.; Dalton, M. B.; Burrell, S. K.; Mayer, J. P.; Patterson,
S. D.; Sinensky, M. J. Biol. Chem. 1997, 272, 5298-5304.
(23) (a) Gaon, I.; Turek, T. C.; Weller, V. A.; Edelstein, R. L.; Singh,
S. K.; Distefano, M. D. J. Org. Chem. 1996, 61, 7738-7745. (b) Turek, T.
C.; Gamache, D.; Distefano, M. D. Bioorg. Med. Chem. Lett. 1997, 7, 2125-
2130. (c) Marecak, D. M.; Horiuchi, Y.; Arai, H.; Shimonaga, M.; Maki,
Y.; Koyama, T.; Ogura, K.; Prestwich, G. D. Bioorg. Med. Chem. Lett.
1997, 7, 1973-1978. (d) Zhang, Y.-W.; Koyama, T.; Marecak, D. M.;
Prestwich, G. D.; Maki, Y.; Ogura, K. Biochemistry 1998, 37, 13411-
13420.
(24) Turek, T. C.; Gaon, I.; Distefano, M. D.; Strickland, C. L. J. Org.
Chem. In press.
(25) Dorman, G.; Prestwich, G. D. Biochemistry 1994, 33, 5661-5673.

4378 J. Am. Chem. Soc., Vol. 123, No. 19, 2001
Kale et al.
Figure 2. Analysis of the effect of irradiation time on the photolabeling
of GDI by [35S]2. Reactions were performed using GDI (4.0 µM) and
[35S]2 (58 nM), and the extent of cross-linking is expressed as the
amount of radioactivity determined by phosphorimaging analysis
(phosphorimage counts, pc).
Figure 4. Phosphorimaging analyses following photolabeling of GDI
by [35S]2 in the presence of various potential competitors. Reactions,
irradiated for 0.5 h, were performed using GDI (4.0 µM) and [35S]2
(58 nM). Shaded bars represent cross-linking obtained with 200 µM
competitor (aside from lane 1), while white bars represent cross-linking
obtained using the peptides indicated. Lane assignments, from left to
right, are as follows: no competitor, 200 µM AFC, 200 µM AFCE,
200 µM AGGC, 200 µM AGGCE, 5 µM Ac-KKSRRC(S-geranyl-
geranyl) (CR6-gg), 50 µM CR6-gg, 100 µM VSSSTNPFRPQKVC(S-
geranylgeranyl) (γ5-gg).
effects occurring with AFC at high concentrations,22b we
performed additional competitions at lower concentrations.
Interestingly, photolysis employing the methyl ester of 1a as a
competitor (AFCE, 1b) produced somewhat less cross-linking
than was obtained in the presence of 1a at the same concentra-
tion (100 µM, Figure 3). Rando and co-workers recently found
similar results with GDI: carboxymethylated cysteine deriva-
tives bound with higher affinity to GDI than their acid
counterparts.26
Figure 3. Phosphorimaging analyses following photolabeling of GDI
by [35S]2 in the presence of various potential competitors. Reactions,
irradiated for 0.5 h, were performed using GDI (4.0 µM) and [35S]2
(58 nM). Lane assignments, from left to right, are as follows: no
competitor, 200 µM FPP, 200 µM GGPP, 10 mM farnesol, 10 mM
AC, 10 mM AFC, 100 µM AFC, 100 µM AFCE.
Examining this phenomenon further, GDI and [35S]2 were
photolyzed in the presence of equimolar concentrations of AFC,
AFCE, and two other prenylated cysteine derivatives: geranyl-
geranylated N-acetyl cysteine and its methyl ester, AGGC and
AGGCE, respectively (Figure 4). Of the four prenylated cysteine
derivatives, AGGCE was the most effective competitor, in
keeping with Rando’s observed trend. Their studies utilized a
fluorescently labeled geranylgeranylated cysteine methyl ester
probe to determine dissociation constants for the binding of
various prenylated cysteines to GDI.26 AGGCE bound with
greater affinity than AGGC, and AFCE exhibited an improve-
ment in binding over AFC. Similarly, our data show that
differences in the amount of competition between acid and ester
is more pronounced in the geranylgeranyl system than in the
farnesyl system as noted by Rando and co-workers.26 These
results may have biological significance since many (but not
all) prenylated proteins are esterified in subsequent posttrans-
lational events.27
Also shown in Figure 4 are the competition results with
geranylgeranylated peptides. The 14-mer, VSSSTNPFRPQKVC-
(S-geranylgeranyl) (γ5-gg), mimics the C-terminal residues of
a γ-subunit of a geranylgeranylated heterotrimeric protein known
to interact with the M2 muscarinic receptor and terminates in a
carboxylic acid.28 At 100 µM, the peptide does not compete as
well as AGGC or AGGCE, but is a more effective competitor
than AFC (Figure 4). At 200 µM peptide, cross-linking is
reduced to roughly 30% that of GDI alone, a value comparable
to the competition seen with AGGCE at the same concentration.
by Rosen and co-workers.10a Thus, it appears that 1a and 2 bind
to a common site on GDI, suggesting that [35S]2 is a good probe
to study isoprenoid binding by GDI. Initially, photolysis was
performed for 2.7 h but later it was found that at 0.5 h, cross-
linking appears to reach a maximum, as shown in Figure 2.
It should be noted that no cross-linking occurred in the
absence of light and that the amount of cross-linking was
reduced upon addition of unlabeled 2. However, we were unable
to completely block radiolabeling by [35S]2 because addition
of higher concentrations of unlabeled probe (g100 µM) created
problematic degradation of the protein and prevented us from
completely suppressing the cross-linking by [35S]2 (data not
shown). Finally, photoaffinity labeling of GDI with [35S]2 is
efficient. Photolysis of an equimolar ratio (4.0 µM) of GDI and
[35S]2, followed by electrophoretic separation and quanti-
tation via phosphorimaging, indicates that 21% of the GDI is
labeled.
Competition Studies. Structural determinants important for
decreasing the cross-linking of [35S]2 were examined next in a
series of competition experiments. Photolysis of GDI and [35S]2
in the presence of the natural prenyl diphosphates, geranyl-
geranyl diphosphate (GGPP) and farnesyl diphosphate (FPP),
showed that GGPP was a more effective competitor than FPP,
suggesting a preference in the binding site for the longer prenyl
group (Figure 3). High levels (10 mM) of farnesol or N-acetyl
cysteine (AC) resulted in no significant decrease in cross-linking,
indicating that these fragments bind with markedly less avidity
than 1a (Figure 3). Because of the possibility of detergent-like
(26) Mondal, M. S.; Wang, Z.; Seeds, A. M.; Rando, R. R. Biochemistry
2000, 39, 406-412.
(27) For a review of prenylation and further processing events, see:
Zhang, F. L.; Casey, P. J. Annu. ReV. Biochem. 1996, 65, 241-269.
(28) Azpiazu, I.; Cruzblanca, H.; Li, P.; Linder, M.; Zhuo, M.; Gautam,
N. J. Biol. Chem. 1999, 274, 35305-35308.

PhotoactiVatable Isoprenoid
J. Am. Chem. Soc., Vol. 123, No. 19, 2001 4379
While the peptide γ5-gg clearly competed with [35S]2 for the
GDI isoprenoid binding site, this peptide does not contain the
same C-terminal sequence present in typical Rho proteins. To
address this issue, a peptide, acetyl-KKSRRC(S-geranylgeranyl)
(CR6-gg), a C-terminal mimic of Rho proteins,10a was prepared.
A farnesylated version of this peptide (CR6-f) was previously
used by Rosen and co-workers in their NMR study of GDI.10a
In their work performed at mM concentrations of CR6-f, GDI
was fully folded and several specific NOEs between CR6-f and
GDI were reported. At 5 µM CR6-gg, some competition with
[35S]2 was observed, while at 50 µM, the cross-linking by [35S]2
was reduced to 30% of that obtained in the absence of any
competitor (Figure 4). Clearly, CR6-gg is the most effective
competitor used in this study.
Figure 5. Analysis of photolabeling of GDI by [35S]2 by SDS-
polyacrylamide gel electrophoresis (PAGE). Lanes 1, 2, and 3 show
protein staining while lanes 1′, 2′, and 3′ show the radiolabeled proteins.
Lanes 1 and 1′ contain a sample of GDI (4.0 µM) irradiated in the
presence of [35S]2 (58 nM); lanes 2 and 2′ contain a sample of GDI
(4.0 µM) irradiated in the presence of Ac-KKSRRC(S-geranylgeranyl)
(CR6-gg) (5 µM) and [35S]2 (58 nM); lanes 3 and 3′ contain a sample
of GDI (4.0 µM) irradiated in the presence of CR6-gg (50 µM) and
[35S]2 (58 nM). Irradiation time was 0.5 h.
In summary, the results of the competition experiments with
prenyl cysteine derivatives and short prenylated peptides
demonstrate that the majority of cross-linking of GDI by [35S]2
can be suppressed using ligands known to bind to the isoprenoid
binding site of GDI. Particularly in the case of CR6-gg, a
biologically relevant competitor,10a the reduction in cross-linking
occurs at a concentration of CR6-gg (50 µM) too low to be
ascribed as a nonspecific detergent effect in which the competi-
tor is reducing the cross-linking either by denaturing the protein
or by sequestering [35S]2 in a micellular phase.29 As noted from
the structural work cited above, GDI remains folded both in
the presence of mM concentrations of CR6-f and saturating
concentrations of AFC (1a).10a In light of this, our results suggest
that [35S]2 does bind with significant specificity to GDI.
However, some residual cross-linking of GDI does occur even
in the presence of a variety of competitors. This may be due to
low levels of nonspecific protein binding (see specificity studies
described below) or may reflect the rapid on/off kinetics of
competitor binding to GDI;30 in all likelihood, both of these
phenomena are occurring. Consistent with the kinetic explana-
tion, it is interesting to note that many of the competitors
employed here have only modest affinities for GDI. For
example, the reported dissociation constant of AFC for GDI is
>20 µM.26 Thus, the bound and unbound states are likely to be
in rapid equilibrium enabling irreversible cross-linking to occur
even in the presence of excess competitor. It is also noteworthy
that while some nonspecific labeling is probably occurring in
these experiments, it is still possible to clearly differentiate
between high versus low affinity binding of [35S]2. As reported
above, high concentrations of AFC (10 mM) are necessary to
reduce labeling of GDI by [35S]2; these high levels also decrease
the background labeling (compare lanes 1′ and 2′, Figure 1).
However, a similar same extent of protection can be ac-
complished using a much lower concentration (50 µM) of a
more specific competitor, CR6-gg. Under those conditions, there
is no reduction in background labeling (compare lanes 1′ and
3′, Figure 5).
and the results were, in both cases, compared to an amount of
photolyzed GDI also identical to that used in the competition
experiments above (4.0 µM, 0.1 mg/mL GDI; 58 nM [35S]2).
When the individual test proteins were present at concentrations
comparable to GDI, no detectable labeling of the random
proteins occurred. Increasing the amount of proteins 3.5-5.4-
fold over GDI resulted observable amounts of cross-linking in
the protein mix ranging from 1% (BSA) to a high of 12%
(ovalbumin) of that of GDI. These results suggest that [35S]2
has an affinity 10-100 fold greater for GDI than for the other
proteins. Thus, probe 2 shows specificity for interacting with
isoprenoid binding proteins, even in the presence of higher
concentrations of other proteins.
Application of [35S]2 to Crude Extracts. After establishing
that the labeling of GDI is preferential to that of several known
nonisoprenoid binding proteins, we next examined the ability
of [35S]2 to label GDI in the presence of a complex mixture of
other, unknown cellular proteins: compound [35S]2 was pho-
tolyzed in a crude E. coli cell extract expressing GDI as a
glutathione-S-transferase (GST)-GDI fusion protein under the
control of an inducible promoter.21 The products of those
photolysis reactions after fractionation via SDS-polyacrylamide
gel electrophoresis (PAGE) are shown in Figure 6. Lane 1
illustrates that large a number of proteins are present in the crude
extract. Despite this plethora of targets, [35S]2 effectively labels
GST-GDI (lane 1′).31 Additionally, the labeling of GST-GDI
can be suppressed by the inclusion of 1a (lane 2′), analogous
to results obtained with purified GDI (Figure 1).
To obtain a more quantitative view of the effectiveness of
[35S]2 as a probe, photolysis reactions were performed on crude
cellular extracts obtained from E. coli lacking the expression
plasmid for GST-GDI and samples of E. coli harboring this
plasmid before and after induction with IPTG. An important
difference between the initial experiments with crude extracts
Specificity Studies. To evaluate the binding selectivity of
our isoprenoid mimicking probe, [35S]2 was photolyzed in the
presence of a mixture of commercially available proteins:
myosin, phosphorylase b, bovine serum albumin (BSA), oval-
bumin, carbonic anhydrase, â-lactoglobulin, and lysozyme.
These proteins are not known to bind isoprenoids and were
collectively irradiated with a concentration of [35S]2 comparable
to that used in the GDI labeling experiments above (58 nM).
Two experiments were performed using this protein mixture,
(31) It should be noted that cleavage of the GST-GDI fusion protein
with thrombin either before or after photolysis with [32S]2 yielded GDI
and GST fragments that were both radiolabeled. This is not surprising given
the affinity of GST for hydrophobic compounds (Wang, J.; Bauman, S.;
Colman, R. F. Biochemistry 1998, 37, 15671-15679). While we were able
to determine that the level of GST cross-linking by [32S]2 is less than that
of GDI, for technical reasons it was not possible to determine a precise
value for this ratio. This was one of the reasons we photolyzed [32S]2 in
the presence of several nonisoprenoid binding proteins as noted above.
(29) As a comparison, a recent study has shown that the critical micellular
concentration of AGGC is 340 µM: Desrosiers, R. R.; Gauthier, F.; Lanthier,
J.; Be´liveau, R. J. Biol. Chem. 2000, 275, 14949-14957.
(30) Chowdry, V.; Westheimer, F. H. Annu. ReV. Biochem. 1979, 48,
293-325.

4380 J. Am. Chem. Soc., Vol. 123, No. 19, 2001
Kale et al.
Figure 6. Analysis of photolabeling of a crude E. coli/pGEX-KG cell
extract of GST-GDI after induction with IPTG by [35S]2 by SDS-
polyacrylamide gel electrophoresis (PAGE). Lanes 1 and 2 show protein
staining, while lanes 1′ and 2′ show the radiolabeled proteins. Lanes 1
and 1′ contain a sample of crude extract (26 µg) irradiated in the
presence of [35S]2 (58 nM); lanes 2 and 2′ contain a sample of GDI
(26 µg) irradiated in the presence of 1a (10 mM) and [35S]2 (58 nM).
Irradiation time was 0.5 h.
described above and subsequent ones is that the crude lysates
in the latter experiments were isolated after shorter growth
periods. This was done to decrease the levels of proteolytic
degradation of GST-GDI32 and to decrease the expression of
GST-GDI to render its concentration closer to other proteins
in the extract. Densitometric analyses of these reactions fol-
lowing SDS-PAGE are given in Figure 7. Photolysis of a crude
lysate generated from cells containing no plasmid for the GST-
GDI protein (HB101) resulted in only very low levels of protein
labeling. Similarly, prior to induction with IPTG, only faint
levels of cross-linked products were observed. However, pho-
tolysis of [35S]2 in extracts obtained from cells after induction
of expression resulted in the marked appearance of a signifi-
cantly radiolabeled protein attributable to GST-GDI.33
To unambiguously identify the predominant cross-linked
species,34 an immunological approach was employed. Following
irradiation of the crude induced lysate with [35S]2, addition of
anti-GST antibodies reduced the protein amount of putative
GST-GDI as well as the amount of corresponding radioactivity
similarly (ca. 50% less protein and 50% less labeling). A
densitometric analysis of the cross-linked products separated
by SDS-PAGE before and after immunoprecipitation is given
in Figure 8. This analysis clearly shows that the major cross-
linked product is GST-GDI.
Figure 7. Densitometric representations of protein staining (above)
and radiolabeling (below) by [35S]2 of three crude cell extracts after
fractionation by SDS-polyacrylamide gel electrophoresis (PAGE): E
coli. HB101 (no plasmid control) (s s s), uninduced GST-GDI E.
coli/pGEX-KG (contains plasmid for GST-GDI expression) (- - -),
and induced GST-GDI E. coli/pGEX-KG (solid line). The arrow
indicates the band corresponding to GST-GDI. The extent of cross-
linking is expressed as the amount of radioactivity determined by
phosphorimaging analysis (phosphorimage counts, pc). Data has been
normalized to represent equivalent amounts of protein for each extract;
58 nM [35S]2, 0.5 h irradiation. See Supporting Information for a more
detailed analysis of these data.
other proteins. The ability of 2 to cross-link to GDI suggests
the presence of an isoprenoid binding site on GDI which is
consistent with recent structural studies of this protein.10a,b,11
Given the results described here with purified proteins and crude
cellular extracts, it should be possible to use 2 in many types
of studies involving pure or partially pure proteins. Experiments
employing 2 with membrane proteins that are refractory to
conventional methods of structural analysis may be particularly
fruitful. However, for experiments targeting proteins at low
abundance present in crude extracts, it is unlikely that 2
manifests sufficient specificity. This may not be a significant
limitation of this approach since the photoactivatable isoprenoid
substructure present in 2 could also be incorporated into peptides
via chemical synthesis using conditions similar to those
employed here for the alkylation of cysteine. This would
undoubtedly result in more selective probes that could be used
to study specific subclasses of prenylated protein-protein
interactions.
Concluding Remarks
The results reported here illustrate the utility of 2 as a probe
for investigating the interaction between prenylated proteins and
(32) Various homologues of GDIs are proteolytically sensitive: footnote
10a and G. R. Hoffman, personal communication.
(33) Sheffield, P.; Garrard, S.; Derewenda, Z. Protein Expression Purif.
1999, 15, 34-39.
(34) In addition to the major cross-linked product, several less pronounced
products at both higher and lower mass than GST-GDI were observed in
Figures 7 and 8. All of these species must contain portions of GST since
they react with the anti-GST antibodies (Figure 8). As noted above, the
smaller cross-linked products are likely to be proteolytic fragments of GST-
GDI (see footnote 32). The identity of the higher mass species is somewhat
less clear. One possibility is that they are incompletely reduced disulfide-
linked dimers of GST-GDI or fragments thereof. It is noteworthy that the
amount of these species appears to vary in different experiments (compare
Figure 7, lower panel with Figure 8).
Important examples of prenylated protein-protein interactions
that are currently under study include the prenyl-specific
protease27,35 and methyltransferase;27,36 both of these are possible

PhotoactiVatable Isoprenoid
J. Am. Chem. Soc., Vol. 123, No. 19, 2001 4381
containing the photoactivatable moiety present in 2 may also
reveal previously unknown proteins that interact with prenylated
proteins. Moreover, the effectiveness of 2 as an isoprenoid
mimic suggests that it may be possible to prepare inhibitors of
prenylated protein-protein interactions by combining modified
isoprenoid units with peptides or other peptidomimetic moieties.
Benzophenone-containing compounds appear to be good struc-
tural mimics of isoprenoids39 and are effective inhibitors of
farnesyltransferase with KI’s as low as 45 nM.16 As a final point,
it is important to emphasize the benefits of the 35S radiolabel
in these experiments. The [35S]CH3SO2Cl reagent developed
by Dean and co-workers19 allows for the facile introduction of
a high specific activity radiolabel into amine-containing mol-
ecules that provides greater sensitivity than 3H and 14C without
the health risks associated with 125I. Thus, the novel probe
presented here and variations thereof offer a versatile and
sensitive means of obtaining structural information about the
isoprene binding region of virtually any protein thought to
contain such a site.
Figure 8. Densitometric autoradiographic representation of the SDS-
polyacrylamide gel electrophoretic fractionation of crude E. coli/pGEX-
KG cell lysate overexpressing [35S]2-labeled GST-GDI before (dashed
line) and after (solid line) treatment with anti-GST antibodies (Ab).
The extent of cross-linking is expressed as the amount of radioactivity
determined by phosphorimaging analysis (phosphorimage counts, pc).
Data has been normalized and is representative of equal amounts of
protein for both sets of conditions; 58 nM [35S]2, 0.5 h irradiation.
Acknowledgment. We thank Dr. Richard A. Cerione for
the provision of purified GDI and the GST-GDI plasmid and
Gregory R. Hoffman for fusion protein purification guidance.
Geranylgeraniol was a gift from P. J. Casey. VSSSTNPFR-
PQKVC(S-geranylgeranyl) was a gift from Drs. Inaki Azpiazu
and N. Gautum. This work was supported by a grant from the
National Institutes of Health (GM58442).
targets for cancer therapy. Additional examples involving pos-
sible prenylated protein-protein interactions include hetero-
trimeric G-protein receptor complexes,7,28,36 a variety of proteins
that interact with the smaller GTP-binding proteins including
Ras,6,9e,35,36 and others.9a-d,f,37,38 The use of 2 or compounds
Supporting Information Available: Synthesis and charac-
terization of compounds used in this study and additional figures
and details concerning the photolysis reactions with [32S]2
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org
(35) (a) Otto, J. C.; Kim, E.; Young, S. G.; Casey, P. J.; J. Biol. Chem.
1999, 274, 8379-8382. (b) Kim, E.; Ambroziak, P.; Otto, J. C.; Taylor,
B.; Ashby, M.; Shannon, K.; Casey, P. J.; Young, S. G. J. Biol. Chem.
1999, 274, 8383-8390. (c) Chen, Y.; Ma, Y. T.; Rando, R. R. Biochemistry
1996, 35, 3227-3237.
(36) Bergo, M. O.; Leung, G. K.; Ambroziak, P.; Otto, J. C.; Casey, P.
J. J. Biol. Chem. 2000, 275, 17605-17610.
(37) Higgins, J. B.; Casey, P. J. J. Biol. Chem. 1994, 269, 9067-9073.
(38) Kilic, F.; Johnson, D. A.; Sinensky, M. FEBS Lett. 1999, 450,
61-65.
JA0012016
(39) Crystal structures of farnesyltransferase complexed with a benzo-
phenone-containing farnesyl diphosphate analogue (footnote 24) and a
geranylgeranyl diphosphate analogue (M. D. Distefano, unpublished results)
demonstrate that the benzophenone unit in each case is within the isoprenoid
binding site of the protein.