Conformational restraint is a critical determinant of unnatural nucleotide recognition by protein kinases

Article abstract of DOI:10.1016/S0960-894X(02)00616-9

This report describes the synthesis of N⁴-(benzyl) AICAR triphosphate, a conformationally restrained analogue of N⁴-(benzyl) ribavirin triphosphate. Both of these nucleotides were evaluated as phosphodonors for wild-type p38 MAP kina

Full text of DOI:10.1016/S0960-894X(02)00616-9

Bioorganic & Medicinal Chemistry Letters 12 (2002) 3223–3227
Conformational Restraint is a Critical Determinant of Unnatural
Nucleotide Recognition by Protein Kinases
Scott M. Ulrich,a Nathan A. Salleeb and Kevan M. Shokata,c,*
aDepartment of Cellular and Molecular Pharmacology, University of California, San Francisco, CA 94143-0450, USA
bProgram in Chemistry and Chemical Biology, University of California, San Francisco, CA 94143-0450, USA
cDepartment of Chemistry, University of California, Berkeley, CA 94720, USA
Received 11 April 2002; accepted 27 April 2002
Abstract—This report describes the synthesis of N4-(benzyl) AICAR triphosphate, a conformationally restrained analogue of
N4-(benzyl) ribavirin triphosphate. Both of these nucleotides were evaluated as phosphodonors for wild-type p38 MAP kinase and
T106G p38 MAP kinase, a designed mutant with expanded nucleotide specificity. The conformationally restrained nucleotide,
N4-(benzyl) AICAR triphosphate, is orthogonal to (not accepted as a substrate by) wild-type p38 MAP kinase, in contrast to
N4-(benzyl) ribavirin triphosphate. Furthermore, N4-(benzyl) AICAR triphosphate, is accepted as a substrate by T106G p38 MAP
kinase, in contrast to N4-(benzyl) ribavirin triphosphate. We hypothesize that the presence of an internal hydrogen bond in
N4-(benzyl) AICAR and itsabesnce in N4-(benzyl) ribavirin triphosphate is the main determinant for their differing structure–
activity relationships.
# 2002 Elsevier Science Ltd. All rights reserved.
11ꢀ13
Protein kinases are a large enzyme family that catalyze
phosphoryl group transfer from ATP to tyrosine or
serine/threonine residues on cellular proteins. Protein
phosphorylation is a central mechanism of cellular sig-
nal transduction and controls pathways such as insulin
signaling, T- and B-cell activation, cellular differentia-
tion, mitogenic signaling, and many others.1ꢀ3
Our laboratory hasdeveloped a chemical-genetic
strategy to identify the direct substrates of protein
kinases. This method involves engineering a protein
kinase to utilize a synthetic analogue of ATP [N6-(benzyl)
ATP] that is orthogonal to, or a poor substrate of, all
wild type kinases. The benzyl substituent of N6-(benzyl)
ATP contacts I338 in v-Src, causing a severe steric
clash.14 Truncating the I338 residue to glycine in v-Src15
Identification of the direct substrate(s) of an individual
protein kinase within a complex signaling cascade
involving many kinases has been difficult to accomplish
with existing experimental techniques. Efforts to eluci-
date protein kinase substrates using peptide libraries4ꢀ7
do not account for the secondary and tertiary structure
of substrate proteins, known to be a critical determinant
of protein kinase substrate specificity.8 Screening cDNA
libraries expressed on phage for protein kinase sub-
strates has also been used to identify candidate sub-
strates of MAP kinases9 and Src.10 However, it isclear
that these experimental conditions are vastly different
from conditionsfound in the cell, where protein–protein
interactionsand cellular localization are both critical
determinants of protein kinase substrate specificity.1
(or the corresponding mutation in other kinases16,17
)
allows N6-(benzyl) ATP to bind and support phos-
phoryl group transfer. Adding g-[32P]-N6-(benzyl) ATP
to a complex cell preparation such as a cell lysate or
immunoprecipitation containing the mutant kinase
results in specific labeling of the direct substrates of only
the mutant kinase, even in the presence of all cellular
wild type kinases.15
Live or intact cellsare the bets context in which to
identify bona fide kinase substrates, since subcellular
localization and protein–protein interactionsare key
determinants of kinase signaling specificity.1 Labeling
kinase substrates in vivo would require transient per-
meablization of the cell membrane to allow impermeant
18
ATP analoguesto enter.
Thislimited amount of
g-[32P] labeled ATP analogue would then compete with
the high (millimolar) cellular concentration of ATP for
the mutant kinase active site. In this case, it would be
*Corresponding author. Tel.: +1-415-514-0472; fax: +1-415-502-
8644; e-mail: shokat@cmp.ucsf.edu
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00616-9

3224
S. M. Ulrich et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3223–3227
Figure 2. Nucleotides discussed in this report.
Figure 3. Wild type p38 (Row 1) and T106G p38 (Row 2) kinase assay
with no nucleotide (Column 1), ATP (Column 2) and N4-(benzyl)
ribavirin triphosphate (Column 3), using a western blot assay which
detects phospho-ATF-2, a substrate of p38.
corresponding I338G kinase mutation) should, if possible,
remain in place (Fig. 1c).
Figure 1. Design goals of natural and unnatural nucleotides with wild
type and mutant protein kinases.
We previously described21 a nucleotide that was
designed to meet these criteria based on ribavirin, a
clinical antiviral agent.22 Ribavirin hasthe tsructural
featuresthat we require to fit a sterically occluded pro-
tein kinase active site, yet is capable of presenting a
benzyl substituent to the I338G pocket and satisfies
hydrogen bonding patterns seen in most kinase/ligand
crystal structures (Fig. 2). We synthesized N4-(benzyl)
ribavirin triphosphate and tested this compound as a
phosphodonor for wild type p38 mitogen-activated
protein kinase (MAP kinase)23 and T106G p38y MAP
kinase. We expressed p38 and T106G p38 carrying an
N-terminal 6-histidine purification tag from bacteria as
described.24 Activating transcription factor-2 (ATF-2),
a known p38 substrate25 wasused in conjunction with a
phospho-ATF-2 specific antibody (Cell Signaling, Bev-
erly, MA){ to detect p38-dependent phosporylation of
ATF-2 by western blotting (Fig. 3).
desirable to engineer a kinase allele that solely uses
analoguesof ATP, but not ATP, GTP, or any other
naturally occurring nucleotide triphosphate. A kinase
that is insensitive to ATP will provide a higher signal-
to-noise ratio of substrate labeling, which is likely to be
critical for the identification of low abundance kinase
substrates in intact cells.
Here we report progress towards an ATP analogue that
is suited for a protein kinase that is insensitive to ATP
according to the following design strategy. N6-(benzyl)
ATP is a poor substrate for wild type kinases, because it
is sterically excluded from the active site (Fig. 1a). A
complementary space-creating mutation (I338G in
v-Src) isknown to allow recognition and catalysiswith
N6-(benzyl) ATP (Fig. 1b).19,20 We reasoned that mak-
ing small-to-large amino acid substitutions in the
kinase active site would sterically exclude ATP from
the active site. Synthetic ATP analogues that have
atoms‘deleted’ from the adenine scaffold could avoid a
steric clash with this second mutation and bind the
pocket. Additionally, in order to label substrates specific
to a mutant kinase, such ATP analogues would have to
be orthogonal to the set of wild type protein kinases in
the cell; thusthe original N6-benzyl substituent (and
In agreement with similar experiments using v-Src and
I338G v-Src,21 the data indicate that N4-(benzyl) ribavirin
yT106 in p38 corresponds to I338 in v-Src by sequence alignment.
{Purified p38 or T106G p38 (100 ng), ATF-2 (2 mg), were incubated
with the appropriate nucleotide (100 mM) for 20 min at 23 ꢁC in kinase
buffer (50 mM TrispH 8, 10 mM MgCl 2, 100 mM NaCl) in a final
volume of 30 mL. Proteins were resolved by PAGE, transferred to a
nitrocellulose membrane and probed by Western blotting.

S. M. Ulrich et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3223–3227
3225
phosphate was specifically designed to conformationally
restrain the benzyl substituent and properly orient it
towardsthe T106G pocket, rendering it orthogonal to
wild type p38 and a competent substrate for T106G p38
(Fig. 1c). The 5-amino group, present in AICAR and
absent in ribavirin, should constrain the benzyl sub-
stituent to a conformation similar to N6-(benzyl) ATP
by steric hindrance aswell asthrough an intramolecular
hydrogen bond to the amide carbonyl group (Fig. 4b).
The crystal structure of AICAR shows this hydrogen
bond between the amide carbonyl and 5-amino group,
supporting this idea.29 Furthermore, studies on the
mechanism of AICAR transformylase showed that
N4-(allyl) AICAR and N4-(methyl) AICAR can bind
enzyme active sites that require the syn rotamer, but not
enzymesthat require the anti rotamer (Fig. 4b).30
Therefore, we believe that N4-(benzyl) AICAR tripho-
sphate should similarly adopt primarily the syn con-
formation shown in Fig. 4. This is the conformation
that resembles N6-(benzyl) ATP, and should make
N4-(benzyl) AICAR triphosphate orthogonal to wild
type kinases and a substrate of kinases with a mutation
corresponding to T106G in p38. However, N4-(benzyl)
AICAR triphosphate is similar to N4-(benzyl) ribavirin
triphosphate in that it may accommodate mutations
that block ATP binding, our ultimate design goal.21
Figure 4. N4-(benzyl) ribavirin triphosphate (a) can readily inter-
convert between two conformerswith repsect to the amide-triazole
bond. Syn N4-(benzyl) ribavirin triphosphate mimics N6-(benzyl) ATP,
yet binds T106G p38 (a kinase that efficiently uses N6-(benzyl) ATP) in
a non-productive conformation. N4-(benzyl) ribavirin triphosphate,
however, supports catalysis with wild type p38, presumably through
the anti conformation. Both of these facts are in contrast to the more
rigid nucleotide, N6-(benzyl) ATP. We therefore designed N4-(benzyl)
AICAR triphosphate (b), an analogue of N4-(benzyl) ribavirin tripho-
sphate that is restricted to the syn rotamer. Therefore, N4-(benzyl)
AICAR triphosphate is a nucleotide that should replicate the phos-
phodonor propertiesof N6-(benzyl) ATP, yet share with N4-(benzyl)
ribavirin triphosphate the ability to bind a mutant kinase active site
that excludesATP.
We synthesized the AICAR nucleoside by benzylation
of inosine31 followed by deformylation in base32 to yield
compound 2. Addition of the triphosphate moiety was
carried out according to the procedure of Ludwig33
(Scheme 1).34
triphosphate is not orthogonal to wild type p38. Fur-
thermore, N4-(benzyl) ribavirin triphosphate is not a
competent substrate for T106G p38. This is the exact
opposite of the structure–activity relationship we
desired.
We then evaluated N4-(benzyl) AICAR triphosphate as
a phosphodonor for wild type p38 and T106G p38 (Fig.
5) using the same assay used for N4-(benzyl) ribavirin
triphosphate described above. The result of this experi-
ment indicatesthat the 5-amino group of N4-(benzyl)
AICAR triphosphate has a profound effect on the
phosphodonor properties with respect to wild type p38
and T106G p38. N4-(benzyl) AICAR triphosphate is
not utilized by wild type p38 (Fig. 5, Row 1), yet isa
substrate for T106G p38 (Fig. 5, Row 2). Densitometry
measurements of the relative amounts of phospho-
ATF2 indicate that ATP is12-fold more efficient than
N4-(benzyl) AICAR triphosphate as a T106G p38 sub-
strate.x This result confirms our hypothesis that con-
straining the conformation of the benzyl substituent is a
key factor in generating nucleotidesthat are: (1) ortho-
gonal to wild type kinases and (2) competent substrates
for mutant kinases such as I338G v-Src and T106G p38.
We hypothesized that N4-(benzyl) ribavirin triphosphate
failed to meet our design criteria because the amide
bearing the benzyl substituent is conformationally flex-
ible, and can adopt a conformation (anti in Fig. 4a) that
does not clash with the T106 active site residue as we
observe in the case of N6-(benzyl) ATP.14 Thus,
N4-(benzyl) ribavirin triphosphate is not orthogonal to
wild type p38. The second problem is that N4-(benzyl)
ribavirin triphosphate is not a viable substrate for
T106G p38. Although N4-(benzyl) ribavirin triphos-
phate can also adopt a conformation similar to
N6-(benzyl) ATP (syn in Fig. 4a), we found it bindsthe
pocket exposed by the I338G mutation in v-Src much
more tightly than N6-(benzyl) ATP does, and appar-
ently in a non-productive orientation.21 Previousstudies
on the base pairing preferences of deoxyribavirin incor-
porated into a DNA duplex utilizing NMR measure-
mentsaswell asesmi-empirical calculationsof the
In conclusion, we have demonstrated that N4-(benzyl)
AICAR triphosphate is orthogonal to wild type kinases
and is a substrate of T106G p38, although it is utilized
less efficiently than ATP. This structure–activity rela-
tionship is critically dependent on the conformational
rigidity of N4-(benzyl) AICAR triphosphate. In con-
trast, N4-(benzyl) ribavirin triphosphate, a related
nucleotide that isconformationally mobile, isneither
orthogonal to wild type kinases nor a substrate of
amide conformation26,27 indicate that rotation about
the amide bond isextremely facile, and furthermore,
there isno isgnificant energetic difference between the
syn and anti rotamers.
To remedy these issues, we designed an analogue of
N4-(benzyl) ribavirin triphosphate that is a benzyl sub-
stituted derivative of AICAR triphosphate (5-amino-
imidazole-4-carboxamide ribotide), an intermediate in
purine biosynthesis (Fig. 2).28 N4-(benzyl) AICAR tri-
xMeasured on an Alpha Innotech Chemimager 5500.

3226
S. M. Ulrich et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3223–3227
Scheme 1. Synthesis of N4-(benzyl) AICAR triphosphate.
(MCB-9874587). We thank Sasha Buzko for assistance
in computational conformational analysis, and mem-
bersof the Shokat laboratory for critical reading of the
manuscript.
References and Notes
1. Pawson, T.; Nash, P. Genes. Dev. 2000, 14, 1027.
2. Hunter, T. Cell 2000, 100, 113.
3. Cohen, P. Trends. Biochem. Sci. 2000, 25, 596.
4. Songyang, Z.; Blechner, S.; Hoagland, N.; Hoekstra, M. F.;
Piwnica-Worms, H.; Cantley, L. C. Curr. Biol. 1994, 4, 973.
5. Songyang, Z.; Lu, K. P.; Kwon, Y. T.; Tsai, L. H.; Filhol,
O.; Cochet, C.; Brickey, D. A.; Soderling, T. R.; Bartleson, C.;
Graves, D. J.; DeMaggio, A. J.; Hoekstra, M. F.; Blenis, J.;
Hunter, T.; Cantley, L. C. Mol. Cell. Biol. 1996, 16, 6486.
6. Songyang, Z.; Cantley, L. C. Methods Mol. Biol. 1998, 87,
87.
Figure 5. Wild type p38 (Row 1) and T106G p38 (Row 2) kinase assay
with no nucleotide (Column 1), ATP (Column 2), and N4-(benzyl)
AICAR triphosphate (Column 3), using a western blot assay which
detects phospho-ATF-2, a substrate of p38.
7. Obata, T.; Yaffe, M. B.; Leparc, G. G.; Piro, E. T.; Mae-
gawa, H.; Kashiwagi, A.; Kikkawa, R.; Cantley, L. C. J. Biol.
Chem. 2000, 275, 36108.
8. Hawkins, J.; Zheng, S.; Frantz, B.; LoGrasso, P. Arch.
Biochem. Biophys. 2000, 382, 310.
9. Fukunaga, R.; Hunter, T. Embo. J. 1997, 16, 1921.
10. Lock, P.; Abram, C. L.; Gibson, T.; Courtneidge, S. A.
EMBO J. 1998, 17, 4346.
11. Alaimo, P. J.; Shogren-Knaak, M. A.; Shokat, K. M.
Curr. Opin. Chem. Biol. 2001, 5, 360.
12. Shogren-Knaak, M. A.; Alaimo, P. J.; Shokat, K. M.
Annu. Rev. Cell. Dev. Biol. 2001, 17, 405.
13. Bishop, A.; Buzko, O.; Heyeck-Dumas, S.; Jung, I.;
Kraybill, B.; Liu, Y.; Shah, K.; Ulrich, S.; Witucki, L.; Yang,
F.; Zhang, C.; Shokat, K. M. Annu. Rev. Biophys. Biomol.
Struct. 2000, 29, 577.
kinases with mutations corresponding to T106G in p38.
N4-(benzyl) AICAR triphosphate is a good candidate
substrate for a kinase which is insensitive to ATP
because it can potentially bind a kinase active site that
sterically excludes ATP. Thus, N4-(benzyl) AICAR tri-
phosphate represents the first step in our ultimate goal
of designing a unique nucleotide triphosphate/protein
kinase pair. This will extend the scope of experiments
designed to label the direct substrates of protein kinases
to include most relevant context, the intact cell.
Acknowledgements
14. Witucki, L. A.; Huang, X.; Shah, K.; Liu, Y.; Kyin, S.;
Eck, M. J.; Shokat, K. M. Chem. Biol. 2002, 9, 25.
15. Shah, K.; Shokat, K. M. Chem. Biol. 2002, 9, 35.
This work was supported by the National Institutes of
Health (CA70031) and the National Science Foundation

S. M. Ulrich et al. / Bioorg. Med. Chem. Lett. 12 (2002) 3223–3227
3227
16. Habelhah, H.; Shah, K.; Huang, L.; Burlingame, A. L.;
Shokat, K. M.; Ronai, Z. J. Biol. Chem. 2001, 276, 18090.
17. Habelhah, H.; Shah, K.; Huang, L.; Ostareck-Lederer, A.;
Burlingame, A. L.; Shokat, K. M.; Hentze, M. W.; Ronai, Z.
Nat. Cell. Biol. 2001, 3, 325.
18. Chitu, V.; Demydenko, D.; Toth, G. K.; Hegedus, Z.;
Monostori, E. Biotechniques 1999, 27, 435.
19. Liu, Y.; Shah, K.; Yang, F.; Witucki, L.; Shokat, K. M.
Chem. Biol. 1998, 5, 91.
20. Shah, K.; Liu, Y.; Deirmengian, C.; Shokat, K. M. Proc.
Natl. Acad. Sci. U.S.A. 1997, 94, 3565.
21. Ulrich, S. M.; Buzko, O.; Shah, K.; Shokat, K. M. Tetra-
hedron 2000, 56, 9495.
28. Voet, D.; Voet, J. G. Biochemistry; 2nd ed.; Wiley: New
York, 1995.
29. Adamiak, D. A.; Saenger, W. Acta Crystallogr. 1979, B35,
924.
30. Wall, M.; Shim, J. H.; Benkovic, S. J. Biochemistry 2000,
39, 11303.
31. Szarek, W. A.; Pinto, B. M.; Iwakawa, M. Can. J. Chem.
1985, 63, 2149.
32. Fujii, T.; Saito, T.; Hisata, H.; Shinbo, K. Chem. Pharm.
Bull. 1990, 38, 3326.
33. Ludwig, J. Acta Biochim. Biophys. Acad. Hung. 1981, 16,
131.
1
34. Compound 2: H NMR (400 MHz, DMSO-d6) d 7.89 ppm
22. Sidwell, R. W.; Robins, R. K.; Hillyard, I. W. Pharmacol.
Ther. 1979, 6, 123.
23. English, J. M.; Cobb, M. H. Trends Pharmacol. Sci. 2002,
23, 40.
24. Khokhlatchev, A.; Xu, S.; English, J.; Wu, P.; Schaefer,
E.; Cobb, M. H. J. Biol. Chem. 1997, 272, 11057.
25. Gupta, S.; Campbell, D.; Derijard, B.; Davis, R. J. Science
1995, 267, 389.
26. Klewer, D. A.; Zhang, P.; Bergstrom, D. E.; Davisson,
V. J.; LiWang, A. C. Biochemistry 2001, 40, 1518.
27. Zhang, P.; Johnson, W. T.; Klewer, D.; Paul, N.; Hoops,
G.; Davisson, V. J.; Bergstrom, D. E. Nucleic Acids Res. 1998,
26, 2208.
(t, 1H, J=6.4 Hz), 7.29 ppm (s, 1H), 7.23 ppm (m, 5H),
5.88 ppm (br s, 2H), 5.43 ppm (d, 1H, J=6.4 Hz), 5.32 ppm (d,
1H, J=6.5 Hz), 5.19 ppm (t, 1H, J=4.8 Hz), 5.11 ppm (d, 1H,
J=4.4 Hz), 4.33 ppm (d, 2H, J=6.4 Hz), 4.24 ppm (m, 1H),
4.00 ppm (m, 1H), 3.85 ppm (m, 1H), 3.54 ppm (m, 2H).
13C NMR (400 MHz, DMSO-d6) d 165.22, 143.27, 141.29,
129.42, 128.79, 127.81, 127.12, 113.33, 88.32, 86.00, 73.55,
70.90, 61.78, 41.98 ppm. HRMS calcd for C16H20N4O5:
348.1434. Found 348.1436.
Compound 3: 31P NMR (161 MHz, D2O) d=ꢀ5.7 ppm (d,
1P, J=19.61 Hz), ꢀ10.80 ppm (d, 1P, J=19.81 Hz),
ꢀ21.94 ppm (vt, 1P, J=19.62 Hz). MS (FAB+) calcd for
C16H23N4O14P3: 588. Found 588, 610 (MꢀH+Na).