Embed Size (px)
Citation preview
An Immuno-Chemo-Proteomics Method for Drug Target
Deconvolution
Chaitanya Saxena,* Eugene Zhen, Richard E. Higgs, and John E. Hale*
Integrative Biology, Greenfield Laboratories, Eli Lilly and Company, Greenfield, Indiana 46140
Received March 24, 2008
Chemical proteomics is an emerging technique for drug target deconvolution and profiling the toxicityof known drugs. With the use of this technique, the specificity of a small molecule inhibitor toward itspotential targets can be characterized and information thus obtained can be used in optimizing leadcompounds. Most commonly, small molecules are immobilized on solid supports and used as affinitychromatography resins to bind targets. However, it is difficult to evaluate the effect of immobilizationon the affinity of the compounds to their targets. Here, we describe the development and applicationof a soluble probe where a small molecule was coupled with a peptide epitope which was used toaffinity isolate binding proteins from cell lysate. The soluble probe allowed direct verification that thecompound after coupling with peptide epitope retained its binding characteristics. The PKC-R inhibitorBisindolylmaleimide-III was coupled with a peptide containing the FLAG epitope. Following incubationwith cellular lysates, the compound and associated proteins were affinity isolated using anti-FLAGantibody beads. Using this approach, we identified the known Bisindolylmaleimide-III targets, PKC-R,GSK3-�, CaMKII, adenosine kinase, CDK2, and quinine reductase type 2, as well as previouslyunidentified targets PKAC-R, prohibitin, VDAC and heme binding proteins. This method was directlycompared to the solid-phase method (small molecule was immobilized to a solid support) providingan orthogonal strategy to aid in target deconvolution and help to eliminate false positives originatingfrom nonspecific binding of the proteins to the matrix.
Keywords: chemical proteomics • Bisindolylmaleimide III • PKC inhibitor • drug target deconvolution• peptide coupled small molecule • mass spectrometry • affinity chromatography
IntroductionIn the postgenomic era, the perceived ‘failure’ of target-based
drug discovery1 has led to a resurgence of system-biology basedtools2 where test compounds are screened based on their abilityto elicit phenotypic changes in model cell and evolutionarylower organism systems. This approach of ‘phenotypic screen-ing’ poses new challenges of identifying targets which areresponsible for observed phenotypic changes upon test com-pound administration. The process of identifying targets, frompossibly several thousands of biomolecules working in the cell,is termed as target deconvolution.2 Identification of targets isimportant for elucidating the biological mechanism of disease,mechanism of drug action, rational drug design, and patientstratification. In addition, it also provides a target-specifictoxicity profile for given test compounds.
Chemical proteomics is one of the many techniques fortarget deconvolution where affinity chromatography is used toidentify the cellular target proteins that interact with smallmolecules.3 Traditionally, a small molecule of interest ismodified in such a way that it can be immobilized on a solidsupport through a hydrophilic linker. After incubating this
immobilized small molecule with protein extract for a briefperiod, a series of washing steps are performed to remove theunbound protein from the mixture. At this stage, specific bufferconditions capable of disrupting small molecule-protein in-teractions are applied to recover the target proteins. At the end,proteins are typically identified by mass spectrometry orimmunodetection methods. Although this approach providessome level of target deconvolution and has been successfullyused,3,5 it remains limited because of the nature of the probewhere ‘solid support’ is used to immobilize the small molecules.Problems associated with steric hindrance6 and limited mobilityof the probe molecule may alter its binding to targets. A largeresin surface area also increases the chance of nonspecificbinding and may lead to the identification of false positives.Coupling small molecules with solid supports makes it difficultto characterize binding affinity toward the protein targets afterimmobilization. Typically, immobilized molecules on solidsupport exhibit weaker affinity for target proteins comparedto their free state. This could lead to unacceptable losses oftarget proteins during the washing steps.
To circumvent some of these shortcomings associated withthe small molecule/solid support approach, we developed ageneralized strategy for target deconvolution. Here, we describethe development and application of an orthogonal techniquefor target deconvolution where the test molecule Bisindolyl-
* To whom correspondence should be addressed. Chaitanya Saxena:phone, 317-651-1539; fax, 317-277-0173; e-mail, [email protected] E. Hale: phone, 317-277-5373; fax, 317-277-0173, e-mail, [email protected].
3490 Journal of Proteome Research 2008, 7, 3490–3497 10.1021/pr800222q CCC: $40.75 2008 American Chemical SocietyPublished on Web 07/01/2008
maleimide III (Bis-III), a derivative of a known Protein KinaseC (PKC) inhibitor,7 was coupled with FLAG peptide. The FLAG-tagged kinase inhibitor was demonstrated to retain inhibitoryactivity and was then incubated with cell lysate. Inhibitor-proteincomplexes were isolated with an anti-FLAG affinity resin. Incomparison, the same compound was immobilized to a solidsupport using the traditional approach, and was then incubatedwith cell lysate. We provide a comparative analysis of bothtechniques. The development of the peptide-coupled-small-molecule approach provided us an extra peptide-handle toperform various related but essential tasks of binding affinitymeasurements and target interaction validation studies. In theprocess, we not only could identify previously unknown targetsof Bis-III but also established a robust process for deconvo-luting targets from seemingly uncountable false positivesoriginating from different types of probes.
Material and Methods
Reagents. Cell culture media was purchased from Invitrogen(Carlsbad, CA). Epoxy-activated Sepharose 6B resin was ob-tained from Amersham Biosciences (GE Healthcare) and Bis-III was purchased from Alexis Biochemical (San Diego, CA).Modified FLAG peptide (NH2-DYKDDDDKC-COOH) with anextra cysteine residue at the C-terminal end was custom-synthesized at Genscript, Inc. (NJ). A heterobifunctional cross-linker succinimidyl-4-[N-maleimidomethyl]cyclohexane-1-car-boxy-[6-amidocaproate] (LC-SMCC) capable of linking asulfhydryl group at one end and an amine at the other endwas purchased from Pierce (Rockford, IL). Protease inhibitorcocktail was purchased from Calbiochem (EMD Biosciences,San Diego, CA). All other reagents were obtained from Sigma(St. Louis, MO).
Antibodies used were rabbit polyclonal anti-PKC-R antibody(Cell Signaling Technology, Beverly, MA), mouse monoclonalanti-Glycogen Synthase Kinase 3 R/� (GSK3 R/�) antibody(Abcam, Cambridge, MA). Agarose based anti-FLAG M2 AffinityGel was purchased from Sigma. Active recombinant GlycogenSynthase Kinase 3� (GSK-3 �) was purchased from BiovisionResearch Product (Mountain View, CA). Tau protein, as GSK-3� substrate and Phospho-Tau specific antibody were purchasedfrom Sigma. Anti-FLAG M2 affinity resins were purchased fromSigma.
Preparation of FLAG-Coupled-Bis-III. Stock solutions of 13mM of Bis-III and LC-SMCC were prepared in DMSO. Finalconcentrations of 1 mM Bis-III and 1 mM LC-SMCC wereadded to 0.6 M phosphate buffer (pH 7.2) in a 1 mL prepara-tion. The reactant solution was kept under constant orbitalmixing for 45 min in the dark at room temperature. From stocksolution of 13 mM, final concentration of 1 mM of modifiedFLAG peptide (NH2-DYKDDDDKC-COOH) was added to thesolution and mixing was continued for another 45 min. Theresultant solution was subjected to gel-filtration using SephadexG-25 (Sigma) to remove unreacted Bis-III and LC-SMCC. Gel-filtration flow-through containing unreacted FLAG, FLAGreacted to LC-SMCC and FLAG coupled with Bis-III throughLC-SMCC were separated over a 250 × 4.6 mm Kromasil C18HPLC column using a linear gradient of water/acetonitrile from95:5 to 55:65 developed over a period of 45 min with a flowrate of 1 mL/min using an Agilent 1100 HPLC pump. Fractionswere collected every 3 min and were analyzed using a MALDI-TOF (4700 Proteomic analyzer, Applied Biosystems). Fractionsshowing peaks at m/z 1832.83 were pooled together andlyophilized. The concentration of thus obtained ‘FLAG-coupled-
Bis-III’ was measured by acquiring absorption spectra of theBis-III, which is the only optically active moiety in the visiblespectral range.
Immobilization of Bisindolylmaleimide-III on Solid Sup-port. Immobilization of Bis-III on a solid support was ac-complished as described.8 Briefly, epoxy-activated 6B resinswere washed extensively using distilled water and were resus-pended in 2 vol of 20 mM Bis-III dissolved in 50% dimethyl-formamide/0.1 M Na2CO3, pH 11. NaOH was added to thereaction mixture to 10 mM final concentration and couplingwas performed overnight in the dark at room temperature withconstant orbital mixing. The next day, resins were washed fivetimes with 50% dimethylformamide/0.1 M Na2CO3 to removeunbound Bis-III. Remaining epoxy reactive groups on resinswere blocked by adding 1 M ethanolamine, pH 11, andcontinuing the orbital shaking for another 4 h. Resins werefinally washed with at least three cycles of alternating pH asper the manufacturer’s instruction. Each cycle consisted of awash with 0.1 M acetate buffer, pH 4.0, containing 0.5 M NaClfollowed by a wash with 0.1 M Tris-HCl buffer, pH 8.0,containing 0.5 M NaCl. Control-resins were prepared by directlyincubating the drained epoxy-activated Sepharose 6B resinswith 1 M ethanolamine and were washed as described above.The beads were used right after preparation.
Cell Culture and Lysis. HeLa cells were cultured in abioreactor to yield more than 1 kg of wet weight cells for variouspurposes. Cells were frozen immediately after harvesting andwere thawed a day before the lysis at 4 °C in an ice-watermixture. Ten grams of wet weight cells was lysed using aDounce homogenizer in buffer containing 50 mM Hepes (pH7.5), 150 mM NaCl, 0.25% Triton X-100, 10% glycerol, 1 mMEDTA, 10 mM sodium pyrophosphate, 1 mM DTT, and proteaseinhibitor cocktail. Lysate was centrifuged at 500g for 15 min at4 °C to remove the cellular debris. Pellets were discarded andthe supernatant was further centrifuged at 40 000g at 4 °C.Pellets were discarded and supernatant thus obtained wasaliquoted, quickly frozen in liquid nitrogen and stored at -70°C. Protein concentration of the aliquot was measured usingthe Bradford method.9 On the day of the experiment, requirednumbers of aliquots were thawed in ice at 4 °C. To get finalprotein concentration of 2 mg/mL, 128 µL of protein stockaliquots (15.6 mg/mL) was diluted to 1000 µL in high salt buffer(50 mM Hepes (pH 7.5), 1 M NaCl, 0.1% NP-40, 1 mM EDTA)immediately before the in vitro association experiments. Twomilligrams of protein was used for all subsequent experiments.
Affinity Chromatography with FLAG-Coupled-Bis-III. Ahundred micromolar of FLAG-coupled-Bis-III or control-FLAGpeptide (NH2-DYKDDDDKC-COOH) was incubated with 1 mLof high salt cell lysate for 2 h at 4 °C. In the preincubationexperiments, 1 mM free Bis-III was added to the lysates priorto the addition of FLAG-coupled-Bis-III. After incubation, 200µL of drained anti-FLAG M2 affinity resin washed with the highsalt buffer was added to the mixture. The mixtures wereincubated overnight at 4 °C with constant orbital mixing. Thenext day after brief centrifugation, the supernatant was col-lected and the anti-FLAG affinity resins were washed four timesusing 500 µL of the high salt buffer each time. The resins wereeluted with 3 vol of elution buffer containing either (a) 1 mMBis-III in high salt buffer, (b) 10 mM FLAG peptide in high saltbuffer, or (c) 0.1 M Glycine (pH 3.5). Because of poor solubilityof Bis-III in aqueous solution, a stock solution of Bis-III wasprepared in DMSO and the required amount of Bis-III was
Immuno-Chemo-Proteomics For Target Deconvolution research articles
Journal of Proteome Research • Vol. 7, No. 8, 2008 3491
diluted in the high salt buffer to prepare the elution buffers;thus, elution buffer contained less than 1% of DMSO.
Affinity Chromatography with Resin Immobilized Bis-III.A hundred microliters of Bis-III-resins or control-resins withno Bis-III immobilized was equilibrated with high salt bufferand later incubated with 1 mL of cell lysate in high salt bufferfor 3 h at 4 °C. In preincubation experiments, 1 mM free Bis-III was added to the lysates prior to the addition of Bis-III resins.Resins were washed four times using 500 µL of the high saltbuffer each time. The resins were eluted with 3 vol of elutionbuffer containing 1 mM Bis-III in 50 mM Hepes (pH 7.5), 1 MNaCl, 0.1% NP-40, and 1 mM EDTA.
Protein Sample Preparation for Analysis. Eluted proteinswere concentrated in a Speed-Vac system and were subse-quently precipitated using a chloroform-methanol precipita-tion method.10 Precipitated proteins were dissolved in the LDSsample buffer (Invitrogen), separated using 4-12% SDS/PAGEgel, and visualized using Coomassie blue stain or silver stainor were transferred to a nitrocellulose membrane and immu-noblotted with the specific antibodies.
Mass Spectrometry. Coomassie-stained gel bands from thewhole lanes were cut into 1 mm slices and the proteins werereduced, alkylated and trypsin-digested.11 Peptides were ex-tracted from the gel by incubating the trypsin-digested bandswith 1 M urea in 50 mM NH4HCO3. Peptides were desaltedand concentrated using µ-C18 Ziptips (Millipore). Peptidefragments thus obtained were injected onto a 5 cm × 7.5 µmC-18 reverse-phase column, and were eluted with a gradientof 5-50% CH3CN developed over 60 min. the eluate wasinfused into an IT mass spectrometer (LTQ, Thermo Finnigan)using a nanoelectrospray source. and data were collected inthe triple-play mode. MS/MS spectra were searched against anonredundant protein database with SEQUEST and X! Tandemfor the identification of proteins.12 Supplementary Informationpages provide the details of the protein identification process.
In Vitro Kinase Assay. GSK3-� inhibition assay was per-formed at 37 °C in a total volume of 20 µL. One microgram ofactive recombinant GSK3-� protein was taken up in 25 mMTris-HCl, pH 7.5, 5 mM �-glycerol phosphate, 12 mM MgCl2, 2mM DTT, 0.1 M Na3VO4, and 200 µm ATP and this mixturewas incubated with varying concentration of Bis-III and FLAG-coupled-Bis-III for 5 min. Tau protein (100 ng) was used asthe substrate for GSK3-�. The kinase reaction proceeded for30 min at 37 °C. Afterward, LDS running buffer was added tothe mixture and proteins were separated over SDS-PAGE.Proteins were transferred to a PVDF membrane and immuno-blotted with anti-phosphoTau antibody. The primary antibod-ies was rabbit anti-human-phospho-Tau (pSer396 and pSer202)(Sigma, 1:100) and secondary antibody was FITC-conjugatedanti-rabbit IgG (Sigma, 1:100). Membrane was imaged for thespecific fluorophore using a Li-Cor Odyssey IR imaging system.Decrease in fluorescence intensity with increasing concentra-tion of Bis-III and/or FLAG-coupled-Bis-III represented inhibi-tion of GSK3-� kinase activity and was quantified to obtain theIC50 values.
Results
Development and Characterization of a Soluble PeptideCoupled Small Molecule Probe. As an alternative to im-mobilizing Bis-III on solid support, we sought an affinity handlewhich could carry the small molecule in a freely diffusable form.The other condition in handle selection was that it should beeasily isolatable once it has interacted with the protein matrix.
FLAG-tag, or FLAG octapeptide (NH2-DYKDDDDK-COOH), isa polypeptide protein tag that is often added to a recombinantexpressedproteinforidentificationandpurificationpurposes.13,15
The epitope of FLAG-tag and commercially available anti-FLAGantibody are well-characterized and researchers use it for awide variety of purposes. Thus, we selected FLAG-tag as the“handle” to be attached to Bis-III. A modified form of the FLAG-tag where a cysteine residue was added to its C-terminus wascustom-synthesized. A heterobifunctional linker LC-SMCC wasused to couple the primary amine group of Bis-III with thesulfhydryl of the modified FLAG-peptide. Increased hydrophi-licity to reduce nonspecific binding and extended spacing armof 16.2 Å for efficient capture of target proteins dictated thechoice of linker in the form of LC-SMCC. Figure 1A show thecoupling scheme of the modified FLAG-tag with Bis-III. Toreduce the reaction of the primary amine of the modified FLAG-peptide with the NHS moiety of the LC-SMCC, Bis-III wasreacted first with LC-SMCC and the modified FLAG-peptidewas added later in the reaction (see Material and Methods).FLAG-coupled-Bis-III was purified by gel-filtration chromatog-raphy followed by RP-HPLC. Coupling was characterized byacquiring the mass spectra of the purified FLAG-coupled-Bis-III which displays a m/z value at 1832.83 (Figure 1B). Toascertain that FLAG-coupled-Bis-III was still an active kinaseinhibitor, a kinase activity inhibition assay of GSK3-� was
Figure 1. Preparation and characterization of FLAG-coupled-Bis-III probe. (A) Structure of PKC-R inhibitor Bis-III and the linker(LC-SMCC) used to couple Bis-III with the modified FLAG peptide.(B) Mass spectrum of the purified FLAG-coupled-Bis-III showingm/z peak at 1832.83. (C) GSK3-� kinase activity inhibition assayin presence of Bis-III and FLAG-coupled-Bis-III.
research articles Saxena et al.
3492 Journal of Proteome Research • Vol. 7, No. 8, 2008
performed in the presence of free Bis-III or FLAG-coupled-Bis-III since GSK3-� is a known target of Bis-III.16,17 Free Bis-IIIinhibited GSK3-� with an IC50 value of ∼0.14 µm. The FLAG-coupled-Bis-III inhibited GSK3-� with an IC50 value of 4 µm(Figure 1C). Although FLAG-coupled-Bis-III showed lowerkinase inhibition efficiency, we concluded that the interactionof FLAG-coupled-Bis-III with its binding partner would beretained when we incubated this probe with the protein matrix.Therefore, we knew the exact probe used to isolate proteinbinding partners was active.
Optimization of Affinity Chromatography and Captureof Cellular Binding Partner Using FLAG-Coupled-Bis-II. Theassumption that, with 4 µm IC50 value, FLAG-coupled-Bis-IIIwill still bind to its target proteins was confirmed by performingin vitro association experiments as shown in Figure 2. GSK3-�in the HeLa total cell lysate specifically interacted with theprobe and was enriched after anti-FLAG chromatography, whileno binding was visible in the control experiments where Bis-III was not conjugated to FLAG peptide. Preincubation of celllysate with 1 mM Bis-III prior to the addition of FLAG-coupled-Bis-III eliminated the binding of GSK3-� to the FLAG-coupled-Bis-III probe (data not shown). In optimizing affinity chro-matographic conditions, we found that the amount of antibodyand the incubation time were critical in capturing all FLAG-coupled-Bis-III probes once it had interacted with the proteinmatrix. Optical absorption of Bisindolylmaleimide was mea-sured in the supernatant to conclude that FLAG-coupled-Bis-III has been completely captured by anti-FLAG antibody (datanot shown). Once the FLAG-coupled-Bis-III along with theinteracting proteins was bound to anti- FLAG resins, there weremultiple options for elution. We eluted bound proteins using
only 1 mM free Bis-III, or eluated the whole FLAG-coupled-Bis-III-protein complex with 10 mM modified-FLAG peptideor 100 mM glycine (pH 3.5). All eluted proteins were separatedover SDS-PAGE and visualized using Coomassie blue stain(Figure 2). Although the starting protein concentration, theamount of FLAG-coupled-Bis-III probe, the amount of FLAGantibody resins and buffer conditions were the same, the elutedprotein profiles using different elution buffer was different(Figure 2). A 1 mM Bis-III brought most proteins off the anti-FLAG resins; however, the amount of protein eluted from thecontrol in a similar experiment was also higher. Overall, wefound that all of these conditions recovered the known targetGSK3-� from the anti-FLAG resins, as shown by immunoblot-ting in Figure 2.
After establishing that FLAG-coupled-Bis-III probe can cap-ture the target protein and 1 mM Bis-III in elution buffer canelute most of the proteins off the anti-FLAG resin, we repeatedthis experiment three times. Eluted proteins from each replicatesample were separated over SDS-PAGE and identified usingeither Western blot or mass spectrometric techniques. Wholegel lanes were cut in 12-15 bands, proteins were in-gel trypticdigested and were subjected for mass-spectrometric analysis.Along with PKC-R (Figure 3) and GSK3-�/GSK3-R (Figures 2and 3), we could specifically identify previously known targetsof Bis-III in samples incubated with FLAG-coupled-Bis-III.These include identification of calcium/calmodulin dependentprotein kinase II-delta, gamma (CaMKII ∆, γ), adenosinekinase, cell division protein kinase 2 (CDK2) and ribosyldihy-dronicotinamide dehydrogenase (NQO2). We also identifiedsome unreported targets of Bis-III which include cAMP-dependent protein kinase type I-alpha (PKAC-R), prohibitin,
Figure 2. Elution profile of proteins from anti-FLAG antibodyresins. Different elution conditions were applied for the elution,and the eluted proteins were concentrated before SDS-PAGEanalysis. Flow through is the unbound fraction to FLAG-coupled-Bis-III. Proteins were separated over SDS-PAGE and visualizedusing Coomassie blue stain. Corresponding samples were ana-lyzed with immunoblotting technique using a monoclonal anti-body against GSK3-R/� protein.
Figure 3. Comparative elution profile of proteins using twotechniques. Proteins were eluted after affinity chromatographyand concentrated 100-fold to cell lysate and flow through.Proteins were separated over SDS-PAGE and visualized usingsilver stain. Corresponding samples were analyzed with immu-noblotting technique using monoclonal antibody against GSK3-R/� and PKC-R.
Immuno-Chemo-Proteomics For Target Deconvolution research articles
Journal of Proteome Research • Vol. 7, No. 8, 2008 3493
Tab
le1.
Pro
tein
Iden
tifi
edU
sin
gT
wo
Tec
hn
iqu
esa
pro
tein
IDan
no
tati
on
no
.d
isti
nct
pep
tid
esto
pm
atch
edp
epti
des
pep
tid
ep
-val
ue1
2p
epti
de
q-va
lue1
2Se
qu
est
Xc
tan
dem
eva
lue
FLA
G-C
ou
ple
d-B
is-I
IIan
dR
esin
-Bis
-III
IPI0
0219
129.
8R
ibo
syld
ihyd
ron
ico
tin
amid
eD
ehyd
roge
nas
e(N
QO
2)30
VLC
QG
FA
FD
IPG
FY
DSG
LLQ
GK
e0.
000
e0.
000
5.31
0.00
0IP
I002
3436
8.1
Ad
eno
sin
eK
inas
e(A
K)
20V
MP
YV
DIL
FG
NE
TE
AA
TF
AR
e0.
000
e0.
000
5.51
0.00
0IP
I000
1733
4.1
Pro
hib
itin
12A
AE
LIA
NSL
AT
AG
DG
LIE
LRe
0.00
0e
0.00
06.
070.
000
IPI0
0172
636.
3C
alci
um
/cal
mo
du
lin-d
epen
den
tP
rote
inK
inas
eII
del
ta(C
aMK
II∆
)10
DLV
TG
GE
LFE
DIV
AR
e0.
000
e0.
000
5.59
0.00
5
IPI0
0172
450.
2C
alci
um
/cal
mo
du
lin-d
epen
den
tP
rote
inK
inas
eII
gam
ma
(CaM
KII
γ)9
ITE
QLI
EA
INN
GD
FE
AY
TK
e0.
000
e0.
000
5.45
0.00
0
IPI0
0028
570.
2G
lyco
gen
Syn
thas
eK
inas
e-3
bet
a(G
SK3�
)6
IQA
AA
STP
TN
AT
AA
SDA
NT
GD
Re
0.00
0e
0.00
06.
020.
000
IPI0
0024
145.
1V
olt
age
Dep
end
ent
An
ion
Sele
ctiv
eC
han
nel
Pro
tein
2(V
DA
C-2
)5
SCSG
VE
FST
SGSS
NT
DT
GK
e0.
000
e0.
000
5.03
0.00
0IP
I002
1630
8.4
Vo
ltag
eD
epen
den
tA
nio
nSe
lect
ive
Ch
ann
elP
rote
in1
(VD
AC
-1)
7T
DE
FQ
LHT
NV
ND
GT
EF
GG
SIY
QK
e0.
000
e0.
000
5.16
0.00
0
FLA
G-C
ou
ple
d-B
is-I
IISp
ecifi
cIP
I000
2183
1.1
cAM
P-d
epen
den
tP
rote
inK
inas
ety
pe
Ial
ph
a(P
KA
C-R
)16
IVV
QG
EP
GD
EF
FII
LEG
SAA
VLQ
Re
0.00
0e
0.00
06.
090.
000
IPI0
0027
252.
6P
roh
ibit
in-2
16IY
LTA
DN
LVLN
LQD
ESF
TR
e0.
000
e0.
000
5.49
0.00
0IP
I002
9100
5.7
Mal
ate
Deh
ydro
gen
ase
9D
VIA
TD
KE
DV
AF
Ke
0.00
0e
0.00
04.
430.
002
IPI0
0010
810.
1E
lect
ron
Tra
nsf
ers
Fla
vop
rote
inSu
bu
nit
alp
ha
8D
PE
AP
IFQ
VA
DY
GIV
AD
LFK
e0.
000
e0.
000
6.54
0.00
0IP
I002
9056
6.1
T-c
om
ple
xP
rote
in1
Sub
un
ital
ph
a8
MLV
DD
IGD
VT
ITN
DG
AT
ILK
e0.
000
e0.
000
4.50
0.00
0IP
I001
4806
3.1
Hem
eB
ind
ing
Pro
tein
16
GIG
MG
MT
VP
ISF
AV
FP
NE
DG
SLQ
Ke
0.00
0e
0.00
03.
160.
021
IPI0
0031
681.
1C
ell
Div
isio
nP
rote
inK
inas
e2
(CD
K2)
5F
MD
ASA
LTG
IPLP
LIK
e0.
000
e0.
000
4.52
0.00
0
Res
in-B
is-I
IISp
ecifi
cIP
I003
3477
5.5
Hea
tSh
ock
Pro
tein
9030
GF
EV
VY
MT
EP
IDE
YC
VQ
QLK
e0.
000
e0.
000
5.51
0.00
0IP
I003
0259
2.2
Fila
min
Aal
ph
a25
IPE
ISIQ
DM
TA
QV
TSP
SGK
e0.
000
e0.
000
5.72
0.00
0IP
I000
2282
7.1
STE
20-l
ike
Seri
ne/
Th
reo
nin
eP
rote
inK
inas
e(S
LK)
22D
TIL
QT
VD
LVSQ
ET
GE
Ke
0.00
0e
0.00
05.
850.
000
IPI0
0000
874.
1P
ero
xire
do
xin
-118
HG
EV
CP
AG
WK
PG
SDT
IKP
DV
QK
e0.
000
e0.
002
5.23
0.00
0IP
I000
0934
2.1
Ras
GT
Pas
eA
ctiv
atin
g-lik
eP
rote
inIQ
GA
P1
17E
ELQ
SGV
DA
AN
SAA
YQ
Re
0.00
0e
0.00
06.
120.
000
IPI0
0027
350.
2P
erix
ired
oxi
n-2
14K
EG
GLG
PLN
IPLL
AD
VT
Re
0.00
0e
0.00
25.
460.
000
IPI0
0013
004.
1P
yrid
oxa
lK
inas
e12
GQ
VLN
SDE
LQE
LYE
GLR
e0.
002
e0.
008
5.16
0.00
0IP
I004
1958
5.8
Pep
tid
yl-p
roly
lci
s-tr
ans
iso
mer
ase
A8
HT
GP
GIL
SMA
NA
GP
NT
NG
SQF
FIC
TA
Ke
0.00
0e
0.00
04.
290.
000
IPI0
0398
625.
5H
orm
erin
8G
EQ
HG
SSSG
SSSS
YG
QH
GSG
SRe
0.00
0e
0.00
04.
040.
000
IPI0
0031
523.
3H
eat
Sho
ckP
rote
in86
8H
ND
OE
QY
AW
ESS
AG
GSF
e0.
000
e0.
000
5.08
0.00
0IP
I000
1814
6.1
14-3
-3P
rote
inth
eta
7A
VT
EQ
GA
ELS
NE
ER
e0.
000
e0.
000
4.97
0.00
3IP
I000
2626
0.1
Nu
cleo
sid
eD
iph
osp
hat
eK
inas
eB
7G
DF
CIQ
VG
Re
0.00
0e
0.00
02.
930.
310
IPI0
0012
048.
1N
ucl
eosi
de
Dip
ho
sph
ate
Kin
ase
A6
FM
QA
SED
LLK
e0.
000
e0.
000
3.57
0.08
7IP
I000
1206
6.1
po
ly(r
C)-
bin
din
gP
rote
in2
4E
STG
AQ
VQ
VA
GD
MLP
NST
ER
e0.
009
e0.
011
4.59
0.00
0IP
I000
2743
4.1
Rh
o-r
elat
edG
TP
-Bin
din
gP
rote
inR
ho
C4
SID
SPD
SLE
NIP
EK
e0.
000
e0.
000
3.68
0.00
6IP
I000
8462
5.2
Vo
ltag
eD
epen
den
tA
nio
nSe
lect
ive
Ch
ann
elP
rote
in4
(VD
AC
4)2
LTLS
ALL
DG
Ke
0.00
0e
0.00
53.
570.
002
aP
rote
ins
iden
tifi
edu
sin
gm
ass
spec
tro
met
ryw
her
eb
ait
com
po
un
dB
is-I
IIis
eith
erF
LAG
-co
up
led
or
imm
ob
ilize
do
nre
sin
s.P
leas
ere
fer
toth
eSu
pp
ort
ing
Info
rmat
ion
sect
ion
1an
dH
iggs
etal
.12
for
the
defi
nit
ion
of
‘p-v
alu
e’an
d‘q
-val
ue’
.P
KC
-Ris
no
tlis
ted
sin
cein
all
the
sam
ple
sit
was
iden
tifi
edu
sin
gW
este
rnb
lot
anal
ysis
(Fig
ure
3).
research articles Saxena et al.
3494 Journal of Proteome Research • Vol. 7, No. 8, 2008
voltage dependent anion-selective channel protein (VDAC) andheme binding protein which were consistently present in allthree replicates (Table 1). PKAC-R and heme binding proteinboth contain a nucleotide binding site and Bis-III is a knownATP competitive inhibitor; hence, Bis-III interaction with theseproteins can be explained. An earlier report where Bis-III wasimmobilized over solid support did not capture CDK2 in thepull-down experiment.8 However, in an in vitro experiment,Bis-III was shown to inhibit CDK2 activity with an IC50 valueof ∼2 µm.8 It is possible that due to lower structural strain,FLAG-coupled-Bis-III was able to capture a lower affinitybinding protein from the protein matrix. In a single experiment,we observed the presence of CDK3 and CDK4 in FLAG-coupled-Bis-III samples. Activities of these cell division protein kinasesare inhibited by Bis-III but with lower affinity.9 Out of all theother eluted proteins, identification of VDAC was unexpected.VDAC is a major constituent of the outer mitochondrialmembrane where it constitutes the pore-forming protein‘porin’.18 VDAC interacts with kinases 19,20 and it is possiblethat VDAC was pulled out as a part of a complex. However,none of the known VDAC interacting kinases were identified.Other studies have shown that the protein kinase C inhibitorRo 31-8220 inhibits voltage dependent sodium (anion) chan-nels21 suggesting that FLAG-coupled-Bis-III may directly in-teract with the VDAC protein.
Comparison of Target Binding Using Immobilized Bis-IIIon Solid Support to FLAG-Coupled-Bis-III. Cell lysate wasincubated with immobilized Bis-III-matrix and bound proteinswere eluted using 1 mM free Bis-III. Proteins were separatedover SDS-PAGE and visualized using silver stain. Capture ofthe known target proteins PKC-R and GSK3-� was confirmedby Western blot analysis (Figure 3). Proteins isolated with FLAG-coupled-Bis-III were also run on a parallel gel and proteins wereidentified using mass spectrometry. We identified most of theproteins previously reported by Daub et al. 8 using the solidsupport method with Bis-III. Consistently in all the repeatedthree experiments, we observed the presence of CaMKII-∆, γ,SLK, NQO2, GSK3- � and Adenosine kinase. Additionally, wefound that pyridoxal kinase, VDAC, prohibitin, nucleosidediphosphate kinase A and B, hydroxyacyl-Coenzymme A de-hydrogenase type II, Ras GTPase-activating-like protein (IQ-GAP1) were also consistently present in all the three experi-ments. Table 1 lists all the identified proteins. Both techniquescaptured several of the same proteins including PKC-R, GSK3-Ra/�, CaMKII-∆, γ, NQO2, VDAC and prohibitin. Pyridoxalkinase, SLK and IQGAP1 were identified only with Bis-IIIimmobilized on the solid support not through FLAG-coupled-Bis-III. At the same time, PKAC-R, CDK2 and heme bindingprotein were specifically found through FLAG-coupled-Bis-IIIapproach.
Bis-III efficiently inhibits the activities of PKC-R, GSK3- �and NQO2.8,17 Bis-III also inhibits the activity of rCaMKII(species-rat) with an IC50 values of 0.86-5.3 µm (unpublisheddata). Although inhibition activity values can not be directlyconverted to binding affinity values, higher inhibition activityshould correlate with higher binding affinity of the smallmolecule ligand. Thus, we hypothesize that the higher bindingaffinity proteins were captured using both the techniques.Additionally, high-abundance protein with lower affinity mayalso have been captured. However, the different couplingmethods of Bis-III captured different proteins of lower bindingaffinity (Figure 4), perhaps due to differences in steric inhibition.
Strategies for Target Deconvolution. One of the majordrawbacks to chemo-proteomic strategies is the presence ofhigh levels of nonspecific protein binding. To identify proteinsspecifically interacting with Bis-III, we designed the followingstrategy (Figure 5). All experiments were repeated three timesunder identical conditions. Proteins identified with at least twodistinct peptides having identification q-values <0.10 (‘q-value’represents the false-discovery rate and reflects the confidencein the peptide identification12) in each experimental replicatewere considered candidates from a particular bait (eitherimmobilized on solid support or FLAG-coupled). The specificityof these candidates was further evaluated by comparing anestimate of the relative protein concentration in the experi-mental replicates to the relative protein concentration in thecontrol replicates. Relative protein concentrations were esti-mated by the weighted sum of the number of peptidesidentified, where the weighting reflects the confidence in thepeptide identification (the identification q-value). The sum SE
) ∑ -log10(identification q-values) across the three experi-mental replicates was compared to the corresponding sum SC
) ∑ -log10(identification q-values) across the control replicatesusing the ratio SE/(SE + SC) as an indicator of the total observedprotein attributed to the experimental replicates. Proteinsexclusively identified in the experimental replicates have a ratioof 1, while proteins not selectively identified in the experimentalreplicates are identified by small (e.g., <0.6) ratios. For clarity,we termed this ratio as ‘binding specificity ratio’. Table 1 liststhe proteins identified from both methods with bindingspecificity ratios >0.8. A complete list of the identified proteinswith binding specificity ratios is provided as SupportingInformation (section 2). Using a cutoff binding specificity ratio>0.8 reduced the list of proteins to 1/50 of the original at thefinal stage of deconvolution.
Discussion
The use of immobilized small molecules as affinity reagentsfor the identification of drug targets is gaining popularity.
Figure 4. Comparative protein identification using two tech-niques. The modified Ven diagram shows the comparativenumber of proteins identified with particular technique. ‘N’ is thenumber of protein identified with a cutoff binding specificity ratio>0.8 (see Strategies for Target Deconvolution). Decreasingbinding arrows represent the hypothesis that higher affinitybinding proteins with Bis-III or higher abundance proteins werecaptured using both techniques; however, lower affinity bindersshowed steric specificity in binding to either one or other probe.
Immuno-Chemo-Proteomics For Target Deconvolution research articles
Journal of Proteome Research • Vol. 7, No. 8, 2008 3495
Immobilization of a small molecule may introduce stericinterference which may inhibit some of the binding interactionsof the molecule. In this study, we described the developmentof an orthogonal technique for capturing cellular bindingpartners of a small molecule kinase inhibitor. This orthogonalmethod affords the ability to verify proteins identified throughthe solid-phase method but also to identify protein interactionsthat may be impaired by the solid phase geometry. The epitope-small molecule coupled strategy may offer several advantages.As demonstrated, the affinity of the tagged molecule to knowntargets may be measured directly verifying that the compoundretains its binding characteristics. Protein interactions with thebait molecule in the form of epitope-small molecule will bediffusion limited allowing enhanced rates of association oversolid-phase immobilized molecules.
Small molecule affinity chromatographic experiments aredesigned to enrich the small molecule interacting targetproteins, but in practice, nonspecific binders and high-abundance proteins tend to dominate the list of proteinsidentified which complicates the drug target deconvolutionprocesses.22 Here, we provided a strategy for deconvoluting thetrue drug targets from a large number of proteins identifiedusing mass spectrometry: (1) repeating the experiments mul-tiple times and selecting only those proteins which werepresent each time; (2) subtracting those also present in thecontrol samples; and (3) combining all proteins meeting thiscriteria from orthogonal affinity techniques. We believe thatthis strategy provides robust criteria for the identification ofproteins which are specifically interacting with small molecules.
Conclusions
With the present work, we successfully demonstrated thatusing peptide-coupled-small molecules in conjunction with
subsequent enrichment with an antipeptide chromatographycan efficiently capture cellular binding partners of a smallmolecule. A comparative analysis of proteins captured usingthis technique and the method with solid-phase immobilizationof the small molecule was performed. It was demonstrated thatboth techniques efficiently capture a subset of the sameproteins; however, additional proteins captured were found tobe specific to the individual method selected perhaps due todiffering steric interactions. Further, a target deconvolutionstrategy was developed and employed for the reduction of falsepositives which helped in identifying the true binding targetsfor small molecules. We believe that this immuno-chemo-proteomics approach will become a useful tool for the futuredrug target deconvolution strategies.
Acknowledgment. We thank Laura J. Bloem and JoanH. Carter for performing GSK3-� kinase inhibition assay forBis-III and FLAG-coupled Bis-III. We also thank Michael D.Knierman for his technical assistance during massspectrometric data acquisition.
Supporting Information Available: (1) Data analysisand (2) table of identified proteins using mass spectrometrywith the binding specificity ratios This material is available freeof charge via the Internet at http://pubs.acs.org.
References(1) Sams-Dodd, F. Drug Discovery Today 2005, 10, 139–147.(2) Terstappen, G. C.; Schlupen, C.; Raggiaschi, R.; Gaviraghi, G. Nat.
Rev. Drug Discovery 2007, 6, 891–903.(3) Godl, K.; Wissing, J.; Kurtenbach, A.; Habenberger, P.; Blencke, S.;
Gutbrod, H.; Salassidis, K.; Stein-Gerlach, M.; Missio, A.; Cotten,M.; Daub, H. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 15434–15439.
Figure 5. Target deconvolution strategy. The specificity of the identified proteins was evaluated with the calculation of the bindingspecificity ratio (see text). Proteins identified with a binding specificity ratio <0.8 were listed as the interacting proteins with the Bis-IIIprobe (either immobilized or FLAG-coupled).
research articles Saxena et al.
3496 Journal of Proteome Research • Vol. 7, No. 8, 2008
(4) Knockaert, M.; Gray, N.; Damiens, E.; Chang, Y. T.; Grellier, P.;Grant, K.; Fergusson, D.; Mottram, J.; Soete, M.; Dubremetz, J. F.;Le Roch, K.; Doerig, C.; Schultz, P.; Meijer, L. Chem. Biol. 2000, 7,411–422.
(5) Guiffant, D.; Tribouillard, D.; Gug, F.; Galons, H.; Meijer, L.;Blondel, M.; Bach, S. Biotechnol. J. 2007, 2, 68–75.
(6) Hahn, R.; Berger, E.; Pflegerl, K.; Jungbauer, A. Anal. Chem. 2003,75, 543–548.
(7) Toullec, D.; Pianetti, P.; Coste, H.; Bellevergue, P.; Grand-Perret,T.; Ajakane, M.; Baudet, V.; Boissin, P.; Boursier, E.; Loriolle, F.;Duhamelll, L.; Charon, D.; Kirilovsky, J. Biol. Chem. 1991, 266,15771–15781.
(8) Brehmer, D.; Godl, K.; Zech, B.; Wissing, J.; Daub, H. Mol. Cell.Proteomics 2004, 3, 490–500.
(9) Bradford, M. M. Anal. Biochem. 1976, 72, 248–254.(10) Wessel, D.; Flugge, U. I. Anal. Biochem. 1984, 138, 141–143.(11) Hale, J. E.; Butler, J. P.; Gelfanova, V.; You, J. S.; Knierman, M. D.
Anal. Biochem. 2004, 333, 174–81.
(12) Higgs, R. E.; Knierman, M. D.; Bonne Freeman, A.; Gelbert, L. M.;Patil, S. T.; Hale, J. E. J. Proteome Res. 2007, 6, 1758–1767.
(13) Witzgall, R.; O’Leary, E.; Bonventre, J. V. Anal. Biochem. 1994, 223,291–298.
(14) Knappik, A.; Pluckthun, A. BioTechniques 1994, 17, 754–761.(15) Chubet, R. G.; Brizzard, B. L. BioTechniques 1996, 20, 136–141.(16) Hers, I.; Tavare, J. M.; Denton, R. M. FEBS Lett. 1999, 460, 433–
436.(17) Davies, S. P.; Reddy, H.; Caivano, M.; Cohen, P. Biochem. J. 2000,
351, 95–105.(18) Colombini, M. Nature 1979, 279, 643–645.(19) Nakashima, R. A. J. Bioenerg. Biomembr. 1989, 21, 461–470.(20) Mannella, C. A. J. Bioenerg. Biomembr. 1997, 29, 525–531.(21) Lingameneni, R.; Vysotskaya, T. N.; Duch, D. S.; Hemmings, H. C.,
Jr. FEBS Lett. 2000, 473, 265–268.(22) Mano, N.; Sato, K.; Goto, J. Anal. Chem. 2006, 78, 4668–4675.
PR800222Q
Immuno-Chemo-Proteomics For Target Deconvolution research articles
Journal of Proteome Research • Vol. 7, No. 8, 2008 3497
