A GTP affinity probe for proteomics highlights flexibility in purine nucleotide selectivity

Article abstract of DOI:10.1021/ja400839e

GTP affinity probes are important tools for the study of GTP-binding proteins, and proteomic profiling is a powerful methodology well suited for the study of such a diverse class of proteins. Here, we synthesize and characterize a photoreactive GTP affinity probe that covalently photocross-links to protein targets and has an alkyne handle for click chemistry conjugation to reporter tags. The GTP-BP-yne probe facilitated identification of a variety of GTP-binding proteins by mass spectrometry, such as small GTPases and members of the GTP1/OBG family. Several ATP-binding proteins were also identified, highlighting variability in purine nucleotide selectivity of some proteins, and the probe was used to elucidate targets' relative nucleotide selectivities. The GTP-BP-yne probe will be a useful tool for the study of GTP-binding proteins, especially when targets of interest are not known a priori.

Full text of DOI:10.1021/ja400839e

Communication
pubs.acs.org/JACS
A GTP Affinity Probe for Proteomics Highlights Flexibility in Purine
Nucleotide Selectivity
Elizabeth A. George Cisar, Nhan Nguyen, and Hugh Rosen*
The Department of Chemical Physiology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037,
United States
S
* Supporting Information
ABSTRACT: GTP affinity probes are important tools for
the study of GTP-binding proteins, and proteomic
profiling is a powerful methodology well suited for the
study of such a diverse class of proteins. Here, we
synthesize and characterize a photoreactive GTP affinity
probe that covalently photocross-links to protein targets
and has an alkyne handle for click chemistry conjugation
to reporter tags. The GTP-BP-yne probe facilitated
identification of a variety of GTP-binding proteins by
mass spectrometry, such as small GTPases and members
of the GTP1/OBG family. Several ATP-binding proteins
were also identified, highlighting variability in purine
nucleotide selectivity of some proteins, and the probe was
used to elucidate targets’ relative nucleotide selectivities.
The GTP-BP-yne probe will be a useful tool for the study
of GTP-binding proteins, especially when targets of
interest are not known a priori.
Figure 1. Structures of the GTP-BP-yne probes and control
compound BP-yne.
GTP and ATP affinities of each target. Together, our results
validate the utility of the GTP-BP-yne probe for use in mass
spectrometry and gel based proteomics and for determining
relative purine nucleotide selectivity of protein targets, while
providing additional evidence6,7 that purine nucleotide
selectivity may not be as strict as it is often assumed to be.
Two structural features, a cross-linker moiety and handle for
click chemistry8,9 to append enrichment or fluorescent tags,
must be added to GTP to make an affinity probe suitable for
protemic profiling. Photoactivatable benzophenone was chosen
as the cross-linker to allow control over the cross-linking
timing, and an alkyne tag was used as the click chemistry
handle. Structure−activity relationships of previously reported
GTP analogs showed that modification of the gamma-
phosphate and 2′ and 3′ hydroxyl groups minimally impacts
binding to GTP-binding proteins.3,10−15 We appended both the
photocross-linking and alkyne moieties to the gamma
phosphate of GTP. CDI mediated coupling of 3 to GTP
yielded the 2′,3′-carbonate, 1 (Scheme S1). When CDI was
limiting and the reaction monitored to ensure the cyclic
carbonate was not formed before 3 was added, the 2′,3′-diol
(2) was formed in low yield (Materials and Methods, SI). The
2′,3′-carbonate is a minor modification relative to the bulky
fluorescent groups appended to the hydroxyl groups of
commercially available GTP analogs13,15 and commonly used
TP-binding proteins constitute a vast, ubiquitous class of
Gproteins essential for cell signaling, trafficking, cytoskele-
tal structure, nucelotide metabolism, and translation. Thus,
GTP affinity probes are broadly applicable for the study of a
variety of GTPases. Radioactive GTP analogs, such as
[35S]GTP and [a-32P]GTP azidoaniline have facilitated
characterization of GTPase signaling pathways.1,2 However,
radioactive or Western blots must be used to visualize
predetermined protein targets of these probes. Mass spectrom-
etry based protein profiling allows identification of a large
number of unknown protein targets simultaneously. Although
Kaneda et al. synthesized a GTP probe designed for
proteomics,3 it has not been characterized, and a commercially
available GTP acyl phosphate relies on an active site lysine
properly positioned for covalent linkage to its targets.4 Here, we
report the synthesis and characterization of a GTP affinity
probe, GTP-BP-yne (1, Figure 1) for proteomic profiling. We
used Multidimensional Protein Identification Technology
(MudPIT)5 to identify 33 proteins specifically labeled by 1
with high confidence, including members of several different
classes of known GTP-binding proteins. Surprisingly, two of
the highest confidence targets of 1 were ATP-binding proteins,
one of which was not previously reported to bind GTP.
Validation studies using in-gel fluorescence to visualize GTP-
BP-yne−protein conjugates served two purposes: (1) they
confirmed that targets discovered by MudPIT bind 1
specifically, and (2) they allowed direct analysis of relative
Received: January 24, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja400839e | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
for the study of nucleotide-protein interactions. And a
comparison of lysate labeling by 1 and 2 showed that both
probes have very similar profiles (Figure S1). Thus, the
synthesis of the higher yielding probe, 1, was scaled up for use
in mass spectrometry based proteomic studies.
To determine the protein targets of 1, MudPIT5 experiments
were carried out with HEK 293T cells. Briefly, cell lysates were
fractionated into soluble and insoluble fractions, and
endogenous nucleotides were removed by desalting. Proteomes
were labeled with 30 μM 1 in the presence of 60 mM MgCl2,
then irradiated with 365 nm light, followed by addition of a
biotin tag via click chemistry. To exclude nonspecific targets of
1, two control samples were generated in each sample set: (1)
excess GTP was added to compete away specific GTP-BP-yne−
protein interactions and (2) BP-yne (3, Figure 1) was added
instead of 1 to identify any targets that were labeled due to
interaction with the non-nucleotide portion of the probe or due
merely to high abundance of the target. The labeled proteomes
were reduced, alkylated, enriched for labeled targets using
streptavidin beads, and trypsinized.16 The tryptic peptides were
analyzed by MudPIT and protein targets were identified using
ProLuCID17 and quantified by spectral counting.18 True target
proteins were identified as those that met threshold criteria in
each of three replicate data sets. First, targets with fewer than 3
spectral counts were eliminated. Next, the spectral counts of all
targets in each sample were normalized based on the total
number of spectral counts of each sample. The normalized
spectral counts were used to calculate spectral count ratios of
samples labeled with GTP-BP-yne to the control samples, and
only targets with ratios of 2 or greater were considered to be
true hits. Ratios from all three data sets were then averaged.
Thirty-three proteins met the above criteria in all three data
sets (Tables 1, S1), and 77 proteins met the criteria in two out
Figure 2. Dendrogram of the GTP-binding protein targets of GTP-
BP-yne (1).
Table 1. Top Five Targets of GTP-BP-yne (1)
To confirm that 1 specifically binds target proteins’ GTP
binding sites, we transiently transfected five targets identified by
mass spectrometry in HEK 293T cells and labeled the resulting
lysates with 1. Azide-Rhodamine was conjugated to the GTP-
BP-yne−protein complexes via click chemistry, and the samples
were separated by SDS-PAGE and visualized with a
fluorescence scanner. This method allowed direct visualization
of the labeled targets in the presence and absence of excess
nucleotide competitors and immediate qualitative comparison
of the targets’ relative affinities for GTP and ATP. Atlastin-3
(ATL3, Table 1) and Mitochondrial Rho GTPase 2 (RHOT2,
Table S1), two GTPase targets identified in the insoluble
fraction, were confirmed using this method (Figure 3A).
Binding of 1 to both GTPases was competed by excess GTP
but not noticeably changed in the presence of excess ATP, as
expected. Casein kinase II alpha′ (CSNK2A2, Table 1) utilizes
both GTP and ATP as phosphate donors,19 and labeling by 1
was eliminated by excess GTP and ATP (Figure 3C).
While 1 labeled many known GTP-binding proteins as
expected, several hits bind ATP, nucleic acids, or other proteins
that bind purine nucleotides but are not known to bind GTP.
This target profile is similar to the profile of the ATP acyl
phosphate probe made by Patricelli et al. that labels targets via
nucleophilic attack by an active site lysine.4 Most of the ATP
probe targets are ATP-binding proteins, and other targets
include proteins that bind nucleic acids, NAD, and FAD. Only
1.5% of the ATP probe targets bind GTP, in contrast with the
26% of the GTP-BP-yne targets that bind ATP, but the stricter
average spectral
counts
NT
protein
1 vs 1 + GTP 1 vs 3
ligand(s)
BCS1L
GNL3
207.08
72.95
45.33
43.95
40.71
14.35
5.45
33.51
7.98
ATP
GTP
CSNK2A2
ATL3
27.95
Undef
3.18
3.06
ATP, GTP
GTP
a
23.72
3.40
OPA1
GTP
a
Undef = Undefined. The spectral counts for the control sample was
equal to zero in all three replicates.
of three data sets (Table S2), including many known GTP-
binding proteins. Thus, 1 works as designed to bind, label, and
enrich GTP-binding proteins in mass spectrometry based
proteomics. Interestingly, the GTP-binding hit proteins are
members of diverse GTP-binding protein classes (Figure 2),
including small Ras-related GTPases (e.g., Rab10), hetero-
trimeric G alpha (e.g., Gai2), unusual GPN-loop GTPases (e.g.,
GPN1), translation elongation factors (e.g., EEF1A1), and
some proteins of unknown function that have been annotated
as GTP-binding based on their sequence (e.g., GTPBP6 of the
MMR1/HSR1 family), thus demonstrating the utility of 1 for
studying a variety of GTP−protein interactions.
Furthermore, the target proteins described here were
identified using one set of conditions and thereby represent a
minimalist set of possible hits. It is possible that varying
parameters, such as divalent cation concentration, could result
in even greater coverage of the GTP-binding proteome.
B
dx.doi.org/10.1021/ja400839e | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
occur to a bystander protein within a protein−protein complex
when the GTP binding pocket orients the benzophenone in
close proximity to the binding partner. Thus, 1 may also be
useful for the study of proteins that bind GTP-binding proteins
even if they do not bind a nucleotide themselves. And BCS1L,
the hit with the highest spectral counts (Table 1), is a unique
AAA ATPase localized to the inner mitochondrial membrane
that is not known to bind GTP.23
BCS1L is essential for formation of respiratory complex III in
mitochondira, and certain mutations cause the fatal GRACILE
syndrome.24 Sequence analysis puts BCS1L in a subclass of its
own, distinct from other AAA ATPases,25 and it is required for
maturation of the Fe−S cluster containing Rieske protein in
Saccharomyces cerevisiae.23 Labeling of BCS1L by 1 was
confirmed by the appearance of a dark fluorescent band at 48
kDa, which is even observed in mock-transfected samples
(Figure 3B), consistent with the high spectral counts observed
by mass spectrometry. While GTP competition diminished
BCS1L labeling, ATP competition completely abolished it,
showing that ATP is indeed preferred over GTP but may not
be the only physiological nucleotide ligand, especially since
both ATP and GTP are known to be required for maturation of
iron−sulfur cluster proteins in mitochondria via unknown
mechanisms. Although we could not detect GTPase (or
ATPase) activity with purified BSC1L in an inorganic
phosphate assay, it is possible that other cofactors or membrane
components may be required for BCS1L activity.
Taken together, our results show that GTP-BP-yne, 1,
functions as the first GTP affinity probe demonstrated to be
compatible with proteomics, and it facilitated the identification
of proteins previously unknown to bind GTP, such as the
ATPase BCS1L, and to evaluate the purine nucleotide
selectivities of the protein targets.
ASSOCIATED CONTENT
* Supporting Information
■
S
Figure 3. Validation and relative purine nucleotide affinity of selected
targets by overexpression, labeling with 1, conjugation to azide-
rhodamine, and in-gel fluorescence. Western blots shown below each
gel confirmed target overexpression. Excess GTP or ATP was added to
assess nucleotide preference. Arrows indicate proteins of interest. (A,
B) Insoluble fraction. (C) Soluble fraction.
Supplemental tables and figures, materials and methods, and
synthetic scheme. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
hrosen@scripps.edu
■
nucleotide selectivity observed with the ATP probe may be due
to its unique labeling mechanism and structure.
Notes
Targets of 1 that are not annotated as GTP-binding proteins
include the related kinases Src and Lyn are high confidence
targets of 1 (Table S1), and Yes kinase was also identified by 1
(Table S2). Src kinase has been reported to use GTP as a
phosphate donor in vitro,20 but GTP binding by family
members Lyn and Yes has not been tested to our knowledge.
RFC1 and 5, two ATPase subunits in the large DNA replication
complex, are targets of 1 (Tables S1, S2), suggesting that the
replication factor complex may hydrolyze GTP and ATP. The
gene product of NUDT2, Ap4A hydrolase, primarily hydrolyzes
diadenosine tetraphosphate as well as other dinucleoside
phosphates,21 and our labeling studies indicate that it binds
GTP and ATP with similar affinities (Figure S2). We also
identified two high confidence targets that are not known to
bind any nucleotides, COPB2 and AP3B1 (Table S1), members
of the COPI coat complex and clathrin complex, respectively.
COPB2 binds Arf·GTP, and the crystal structure of the
complex shows that the gamma phosphate of the Arf-bound
GTP is close to COPB2.22 It is possible that cross-linking can
The authors declare the following competing financial
interest(s): HR is a consultant to Activx Biosciences and a
scientific co-founder of Receptos.
ACKNOWLEDGMENTS
■
We thank Sherry Niessen, Melissa Dix, Justin Cisar, and other
members of the laboratory of Ben Cravatt for assistance with
mass spectrometry and helpful discussions. We thank Stephan
Schurer and Caty Chung of the University of Miami for
generation of Figure 2 and Kevin Shreder of ActivX for helpful
discussions.
REFERENCES
■
(1) Bidlack, J. M.; Parkhill, A. L. Methods Mol. Biol. 2004, 237, 135.
(2) Offermanns, S.; Schultz, G.; Rosenthal, W. Methods Enzymol.
1991, 195, 286.
(3) Kaneda, M.; Masuda, S.; Tomohiro, T.; Hatanaka, Y.
ChemBioChem 2007, 8, 595.
C
dx.doi.org/10.1021/ja400839e | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
(4) Patricelli, M. P.; Szardenings, A. K.; Liyanage, M.; Nomanbhoy,
T. K.; Wu, M.; Weissig, H.; Aban, A.; Chun, D.; Tanner, S.; Kozarich,
J. W. Biochemistry 2007, 46, 350.
(5) Link, A. J.; Eng, J.; Schieltz, D. M.; Carmack, E.; Mize, G. J.;
Morris, D. R.; Garvik, B. M.; Yates, J. R., III. Nat. Biotechnol. 1999, 17,
676.
(6) Shugar, D. Acta Biochim. Pol. 1996, 43, 9.
(7) Shakya, T.; Wright, G. D. Antimicrob. Agents Chemother. 2010, 54,
1909.
(8) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B.
Angew. Chem., Int. Ed. 2002, 41, 2596.
(9) Tornoe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002,
67, 3057.
(10) Pfeuffer, T. J. Biol. Chem. 1977, 252, 7224.
(11) Hingorani, V. N.; Chang, L. F.; Ho, Y. K. Biochemistry 1989, 28,
7424.
(12) Rajagopalan, K.; Chavan, A. J.; Haley, B. E.; Watt, D. S. J. Biol.
Chem. 1993, 268, 14230.
(13) Remmers, A. E.; Posner, R.; Neubig, R. R. J. Biol. Chem. 1994,
269, 13771.
(14) Yang, M.; Leyh, T. S. Biochemistry 1997, 36, 3270.
(15) Eberth, A.; Dvorsky, R.; Becker, C. F.; Beste, A.; Goody, R. S.;
Ahmadian, M. R. Biol. Chem. 2005, 386, 1105.
(16) Speers, A. E.; Cravatt, B. F. Curr. Protoc. Chem. Biol. 2009, 1, 29.
(17) Tayo, L. L.; Lu, B.; Cruz, L. J.; Yates, J. R., III. J. Proteome Res.
2010, 9, 2292.
(18) Old, W. M.; Meyer-Arendt, K.; Aveline-Wolf, L.; Pierce, K. G.;
Mendoza, A.; Sevinsky, J. R.; Resing, K. A.; Ahn, N. G. Mol. Cell.
Proteomics 2005, 4, 1487.
(19) Pinna, L. A. Biochim. Biophys. Acta 1990, 1054, 267.
(20) Graziani, Y.; Erikson, E.; Erikson, R. L. J. Biol. Chem. 1983, 258,
2126.
(21) McLennan, A. G. Cell. Mol. Life Sci. 2006, 63, 123.
(22) Yu, X.; Breitman, M.; Goldberg, J. Cell 2012, 148, 530.
(23) Nobrega, F. G.; Nobrega, M. P.; Tzagoloff, A. EMBO J. 1992,
11, 3821.
(24) Visapaa, I.; Fellman, V.; Vesa, J.; Dasvarma, A.; Hutton, J. L.;
Kumar, V.; Payne, G. S.; Makarow, M.; Van Coster, R.; Taylor, R. W.;
Turnbull, D. M.; Suomalainen, A.; Peltonen, L. Am. J. Hum. Genet.
2002, 71, 863.
(25) Frickey, T.; Lupas, A. N. J. Struct. Biol. 2004, 146, 2.
D
dx.doi.org/10.1021/ja400839e | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX