© 2003 Nature Publishing Group http://www.nature.com/naturebiotechnology
LETTERS
Discovering potent and selective reversible inhibitors of
enzymes in complex proteomes
Donmienne Leung1,2, Christophe Hardouin1, Dale L Boger1 & Benjamin F Cravatt1,2
To realize the promise of genomics-based therapeutics, new
methods are needed to accelerate the discovery of small
molecules that selectively modulate protein activity. Toward
this end, advances in combinatorial synthesis have provided
unprecedented access to large compound libraries of
considerable structural complexity and diversity1,2, shifting the
bottleneck in drug discovery to the development of efficient
screens for protein targets3. Screening for reversible enzyme
inhibitors typically requires extensive target-specific work,
including protein expression and purification, as well as the
development of specific substrate assays. Here we report a
proteomic method for the discovery of reversible enzyme
inhibitors that avoids these steps. We show that competitive
profiling of a library of candidate serine hydrolase inhibitors in
complex proteomes with activity-based chemical probes4–6
identifies nanomolar reversible inhibitors of several enzymes
simultaneously, including the endocannabinoid-degrading
enzyme fatty acid amide hydrolase (FAAH)7, triacylglycerol
hydrolase (TGH)8 and an uncharacterized membraneassociated hydrolase that lacks known substrates. The strategy
tests inhibitors against numerous enzymes in parallel,
assigning both potency and selectivity factors to each agent.
In this way, promiscuous inhibitors were readily rejected in
favor of equally potent compounds with 500-fold or greater
selectivity for their targets.
Challenges associated with enzyme inhibitor screening have thus far
confined this pursuit to a limited number of well-validated targets,
contributing to at least two undesired trends. First, the selectivity of
inhibitors can be overestimated, because counterscreening is conducted against only a modest number of related enzymes available in
pure form. Second, novel enzymes can be excluded from analysis
because of a lack of functional information required for assay development. A chemical proteomics strategy, activity-based protein profiling (ABPP)4,9, has been introduced to address these shortcomings.
In this approach, active site–directed chemical probes that target
many members of a given enzyme class (or classes) are used to evaluate the activity of candidate inhibitors directly in complex proteomes6,10,11. ABPP methods offer the advantage of testing enzymes
in their native environment, thus eliminating the need for recombinant expression, purification and the development of a specific substrate assay. Because inhibitors are screened against multiple enzymes
in parallel, both the potency and selectivity of these agents can be
concurrently evaluated.
Thus far, ABPP has been applied to identify irreversible enzyme
inhibitors10,11. Although useful as experimental tools for certain
enzyme classes, irreversible inhibitors, owing to their inherent reactivity, typically display poor target selectivity in vivo and are therefore
less desirable than reversible inhibitors as lead compounds for drug
design. Consequently, to discover compounds of therapeutic relevance, ABPP must be adapted to allow the identification of reversible
enzyme inhibitors.
The ABPP method for the discovery of irreversible inhibitors
involves preincubation of cell or tissue proteomes with enzyme
inhibitors followed by addition of activity-based probes that bear one
or more reporter groups (such as radioisotopes, biotin or fluorophores). Enzymes modified by irreversible inhibitors are incapable
of reacting with probes and can be detected by a reduction in signal
intensity from the reporter group10,11. However, this strategy is not
directly applicable to the discovery of reversible inhibitors, for which
the kinetics of probe-proteome reactions must be taken into
account6. Indeed, reversible inhibitors will only affect probe labeling
of a given enzyme for a restricted period of time, depending on both
the affinity of the inhibitor and the rate of probe reactivity.
Accordingly, for ABPP to serve as an effective proteomic screen for
reversible enzyme inhibitors, assay conditions are required that permit the rates of probe labeling for the majority of enzymes in a proteome to be monitored collectively. With these considerations in
mind, we examined the reactivity profiles of mouse tissue proteomes
with a rhodamine-tagged fluorophosphonate (FP-rhodamine) probe
that targets the serine hydrolase superfamily5,12 and identified conditions under which the extent of probe labeling for the majority of
enzymes could be monitored at a single, kinetically relevant time
point (Fig. 1a,b). Under such conditions of incomplete probe labeling, the binding of competitive reversible inhibitors to enzymes
should slow the rate of probe reaction and be detected as a decrease in
fluorescence signal intensity.
A library of candidate reversible serine hydrolase inhibitors was
synthesized as described13,14 and tested in competitive proteomic
profiling assays with FP-rhodamine. These electrophilic ketone agents
were originally evaluated as inhibitors of FAAH7 using conventional
substrate assays13,14, but their activity toward other serine hydrolases
has remained unexplored. Because compounds possessing electrophilic carbonyls typically inhibit serine hydrolases by formation of
The Skaggs Institute for Chemical Biology and the Departments of 1Chemistry and 2Cell Biology, The Scripps Research Institute, 10550 N. Torrey Pines Road, La
Jolla, California 92037, USA. *Correspondence should be addressed to B.F.C. (cravatt@scripps.edu).
NATURE BIOTECHNOLOGY VOLUME 21 NUMBER 6 JUNE 2003
687
LETTERS
of 45- to 50-kDa brain enzymes (Fig. 1d, left
panel, double arrowhead). Likewise, an
IC50 values (nM), 95% confidence limits
examination of other tissue proteomes identified a 60-kDa heart membrane enzyme that
60-kDa brain
45- and 50-kDa
60-kDa heart
exhibited an inhibitor sensitivity profile disenzyme (FAAH)
brain enzyme
enzyme (TGH)
tinct from the brain enzymes (Fig. 1d, right
(KIAA1363)
panel, arrowhead). Thus, these initial studies
identified several serine hydrolases that were
N
4
1.1
>100,000
1.4
N
(0.72–1.7)
(1.1–1.9)
targeted by the compound library.
O
ABPP of membrane proteomes from
O
FAAH-knockout mice17 confirmed that the
60-kDa brain enzyme was FAAH and estabN
5
0.81
50,000
83
N
lished that the 60-kDa heart enzyme was a
(0.56–1.1)
(21,000–
(69–100)
120,000)
O
distinct protein (Fig. 2a). Using biotinylated
O
FP probes and avidin chromatography–mass
spectrometry (MS) procedures6,12, we identiN
6
44
62,000
1,400
N
fied the 60-kDa heart enzyme as TGH8, an
(31–62)
(23,000–
(880–2,400)
160,000)
enzyme that mobilizes triacylglycerol storage
O
O
pools for the assembly of very-low-density
lipoprotein particles, and the 45- to 50-kDa
brain enzymes as mouse orthologs of the preN
N
7
150
>100,000
>100,000
dicted human serine hydrolase, KIAA1363.
(84–280)
O
The migration of KIAA1363 as two distinct
O
bands may be a result of heterogeneous glycosylation, because treatment with peptideCF 3
9
4,500
1,100
4,800
N-glycosidase F (PNGase F) converted the
(3,100–6,500)
(730–1,500)
(1,900–12,000)
O
45- and 50-kDa versions of this enzyme to a
single 40-kDa band (Fig. 2b).
CF3
15
11,000
470
61
Recombinant mouse TGH and KIAA1363
(6,200–21,000)
(320–690)
(47–79)
O
enzymes expressed in COS-7 cells (Fig. 2c)
CF 3
were found to exhibit equivalent inhibitor
18
5,300
770
0.47
(3,300–8,300)
(570–1,100)
(0.10–2.2)
sensitivity profiles to those of their natively
O
expressed counterparts (Fig. 2d). For example, both recombinant and native KIAA1363
CF 3
25
15,000
52
320
O
were potently inhibited by bulky trifluo(8,100–28,000)
(38–71)
(230–450)
O
O
romethyl ketones (TFMKs) 25 and 28, but
were less affected by a shorter chain TMFK
CF3
26
30,000
1,500
1.8
26 (Fig. 2d, left panels). In contrast, recombi(19,000–48,000) (1,300–1,700)
(1.0–3.1)
O
nant and native TGH showed the opposite
CF 3
profile, being more strongly inhibited by
28
37,000
110
160
(16,000–84,000) (86–150)
(86–310)
compound 26 than by either 25 or 28 (Fig. 2d,
O
right panels). Notably, neither TGH nor
KIAA1363 was inhibited by α-keto heterocyN
30
>100,000
>100,000
>100,000
N
cle (α-KH) 7 (Fig. 2d), which was a potent
inhibitor of FAAH (Table 1, IC50 = 150 nM).
O
The IC50 values for FAAH inhibition calculated by ABPP were generally within 5- to
20-fold of the earlier reported Ki values for
Data for entire compound library are available as Supplementary Table 1 online.
members of the compound library13,14
(Supplementary Table 1 online). Additiona covalent reversible intermediate with the conserved serine nucle- ally, ABPP provided a relative rank order for FAAH inhibitors that
ophile15,16 (Fig. 1c), these agents may interact with multiple members paralleled the order determined by standard substrate assays (Fig. 3a).
of this large enzyme class. Each inhibitor was initially tested for its To explore further the relationship between the IC50 values detereffect on the FP-labeling profile of the mouse brain membrane pro- mined by ABPP and the Ki values calculated from substrate assays, we
teome at concentrations ranging from 1 nM to 100 µM (Fig. 1d; screened a subset of the compound library for inhibition of recombishown for 1 µM). From these data, IC50 values were calculated for nant TGH using p-nitrophenyl laurate as a substrate18. The Ki values
each agent and its corresponding target(s) in the proteome (Table 1; for these TGH inhibitors were all found to be within five-fold of the
see Supplementary Table 1 online for complete results). As expected, IC50 values determined by ABPP (Supplementary Table 2 online).
several compounds showed strong activity toward a 60-kDa protein Collectively, these results indicate that ABPP can accurately measure
predicted to be FAAH (Fig. 1d, left panel, arrowhead). Notably, how- both the relative and absolute potencies of enzyme inhibitors in comever, a subset of the inhibitors displayed higher activity toward a pair plex proteomes.
© 2003 Nature Publishing Group http://www.nature.com/naturebiotechnology
Table 1 IC50 values determined by ABPP for representative members of the compound library
688
VOLUME 21 NUMBER 6 JUNE 2003 NATURE BIOTECHNOLOGY
© 2003 Nature Publishing Group http://www.nature.com/naturebiotechnology
LETTERS
Analysis of IC50 values with a hierarchical
b
a
clustering algorithm19 uncovered distinct
classes of potent inhibitors for FAAH, TGH
and KIAA1363 (Fig. 3b). Re-clustering of
these data after normalization to discern
inhibitor selectivity independent of potency
revealed sets of inhibitors that were selective
for FAAH, TGH and KIAA1363 (Fig. 3c).
FAAH-selective inhibitors represented a mixture of high- and moderate-potency agents,
such as compound 7, a 150-nM inhibitor that
showed no detectable activity against other
hydrolases in the proteomes examined,
c
including TGH and KIAA1363 (Table 1).
TGH-selective inhibitors included short-chain
alkyl/aryl TFMKs 26 and 31, whereas the
bulkier TFMKs 25 and 27 represented excellent lead inhibitors for KIAA1363 (Table 1 and
Supplementary Table 1 online). ABPP also
uncovered compounds with strong activity
toward multiple enzymes, including α-KHs
d
2–4, which potently inhibited both FAAH and
TGH (IC50s < 5 nM for both targets; Table 1
and Supplementary Table 1 online).
To further evaluate inhibitor selectivity
among different enzyme classes, we analyzed
the compound library in competitive screens
using a second class of ABPP probes bearing
a sulfonate ester reactive group20,21. These
sulfonate probes profile several distinct
enzyme classes, including aldehyde dehydrogenases, glutathione S-transferases and enoyl
CoA-hydratases. Although most compounds
were inactive against these other enzyme
Figure 1 ABPP of complex proteomes with the serine hydrolase–directed probe FP-rhodamine.
classes, a subset of TFMKs did exhibit mod- (a) Kinetic analysis of FP-rhodamine labeling of the mouse brain membrane proteome (conditions:
erate potency against enoyl CoA-hydratase-1 100 nM FP-rhodamine, 1 mg/ml protein, 50 mM Tris-HCl, pH 8.0). Numbers 1–4 signify four
(ECH-1) (for example, 15, 18, 29; IC50 values representative FP-rhodamine targets. (b) Labeling progress curves for FP-rhodamine targets 1–4
0.7–30 µM; see Supplementary Table 3 and in a. Red dashed line highlights a 10-min time point at which the labeling reactions for the majority
Supplementary Fig. 1 online). Notably, com- of FP-rhodamine targets had proceeded to a sufficient extent to permit protein visualization by in-gel
pound 29 was at least five-fold more active fluorescence scanning, but had not yet reached completion. (c) General mechanism for covalent
reversible and irreversible inhibition of serine hydrolases by electrophilic ketone (right) and rhodamine
against ECH-1 than any of the serine hydro(Rh)-tagged fluorophosphonate (left) reagents, respectively. Ketone inhibitors form a reversible
lases examined. Collectively, these findings hemiketal intermediate with the serine nucleophile, which is also stabilized by the oxyanion hole15,16
demonstrate that ABPP can identify potent, (shown as a hydrogen bond with the negatively charged oxygen of the carbonyl). In competitive
reversible inhibitors of multiple enzymes profiling experiments, the presence of a reversible ketone inhibitor will reduce the rate of
within the confines of the proteomes in phosphonylation of an enzyme by FP-rhodamine. (d) Competitive proteomic profiling of a library of
which they are naturally expressed and, at the candidate serine hydrolase inhibitors (1 µM) with FP-rhodamine (100 nM). Inhibitor-sensitive targets
same time, discern which of these inhibitors in the mouse brain (single and double arrowheads, left panel) and heart (single arrowhead, right
panel) membrane proteomes are highlighted.
are selective.
Here, we have shown that ABPP can be
used to discover potent and selective
It is noteworthy that, despite their overlapping inhibitor sensitivreversible inhibitors of many enzymes directly in complex proteomes.
This strategy does not require that enzymes be recombinantly ity profiles, FAAH, TGH and KIAA1363 share no significant
expressed or purified before analysis. Moreover, by using active sequence identity. These findings highlight that nonhomologous
site–directed probes with broad target selectivity, ABPP is essentially enzymes may have related active site structures, and thus empha‘substrate-free’, permitting the simultaneous identification of size the value of proteome-wide screens that can detect unanticiinhibitors for enzymes with diverse substrate preferences (such as pated ‘off-target’ activities for inhibitors. Indeed, further analysis of
FAAH and TGH) and novel enzymes that lack known substrates the compound library using sulfonate ester ABPP probes20,21
(such as KIAA1363). Finally, because ABPP tests compounds against uncovered a group of TFMKs that targeted the mechanistically dismany enzymes in the proteome, selective inhibitors can be readily dis- tinct enzyme ECH-1. Thus, ABPP can be used to screen for
tinguished from promiscuous agents, thus clarifying which com- inhibitor activity and selectivity not only within, but also among,
pounds are most suitable for pharmacological studies in vivo.
enzyme classes.
NATURE BIOTECHNOLOGY VOLUME 21 NUMBER 6 JUNE 2003
689
LETTERS
© 2003 Nature Publishing Group http://www.nature.com/naturebiotechnology
a
b
Figure 2 Identification and characterization of
serine hydrolase targets of the inhibitor library.
(a) ABPP of tissues from FAAH-WT and KO mice
confirmed the identity of the 60-kDa FP-reactive
brain membrane protein as FAAH (left two lanes)
and revealed that the 60-kDa FP-reactive heart
membrane protein is a distinct enzyme (right two
lanes). (b) Deglycosylation of the FP-rhodaminelabeled brain membrane proteome was
accomplished with the glycosidase PNGaseF,
causing the 45- and 50-kDa targets (single
arrowhead) to reduce in size to a single 40-kDa
band (double arrowhead). (c) FP-rhodamine
labeling of recombinant mouse TGH and
KIAA1363 enzymes expressed by transient
transfection in COS-7 cells. Mock cells were
transfected with empty pcDNA3 vector.
(d) Recombinant mouse KIAA1363 and TGH
enzymes expressed in COS-7 cells exhibited
similar inhibitor sensitivity profiles to their natively
expressed counterparts (from brain and heart
membrane proteomes, respectively). See Table 1
for IC50 values for the indicated inhibitors.
c
d
FAAH is the primary enzyme responsible for regulating endogenous cannabinoid signaling in the brain17,22, and therefore specific
inhibitors of this enzyme may be of value for the treatment of pain
and related neural disorders. Of the compounds that inhibited
FAAH, α-KH 7 was of particular interest in that it displayed excellent
potency (IC50 = 150 nM) and selectivity (>500-fold). This inhibitor
was the only member of the compound library that has an unfused
pyridyl oxazole moiety, which may be uniquely accommodated by
FAAH because this enzyme possesses a bifurcated active site with
extended channels on either side of the serine nucleophile23. This
suggests a strategy for designing second-generation FAAH inhibitors
in which the acyl chain substituents of potent inhibitors like 1–5 are
merged with the unfused pyridyl oxazole of 7.
The ABPP approach for discovering reversible enzyme inhibitors
has a few limitations. First, because ABPP probes target only a portion
of the proteome, the inhibitor selectivity factors assigned by this
method are not representative of proteome-wide specificity.
Nonetheless, as long as ABPP probes provide broad coverage within
the enzyme family under examination, they should allow promiscuous inhibitors to be rejected, thereby circumventing potential ‘mechanism-based’ toxicity in vivo caused by inhibitors that target several
enzymes from the same family24. Another potential shortcoming of
Figure 3 Cluster analysis of the inhibitor sensitivity profiles of FAAH, TGH
and KIAA1363. (a) Comparison of the IC50 and Ki values for FAAH inhibition
as determined by ABPP and substrate assays, respectively. The relative
potencies that members of the inhibitor library displayed for FAAH as
determined by ABPP matched closely the relative potencies determined
by substrate assays13,14. For ABPP, IC50 values ranged from 0.8 nM to
>100 µM; for substrate assays, Ki values ranged from 0.14 nM to >100 µM.
(b) Potency cluster analysis for inhibitors of FAAH, TGH and KIAA1363.
Clustering of the IC50 values identifies four general classes of inhibitors:
highly potent FAAH inhibitors (FAAH1, magenta numbers), moderately potent
FAAH inhibitors (FAAH2, green), potent TGH inhibitors and potent KIAA1363
inhibitors. (c) Selectivity cluster analysis for inhibitors of FAAH, TGH and
KIAA1363. Clustering of the IC50 values after normalization to the most
potently affected target distinguished FAAH inhibitors that show high
selectivity (such as 1, 5–7) from those that are not selective for this enzyme
(for example, 2–4). Selective inhibitors of TGH and KIAA1363 are also noted.
See Supplementary Methods for details on data treatment for cluster analysis.
690
ABPP is that active site–directed probes compatible with whole-proteome analysis are currently available for only a select number of
enzyme classes. However, efforts are underway to expand the proteomic coverage of ABPP, and probes for more than ten enzyme
classes have already been identified9,25. The strategy reported herein
to screen for potent and selective reversible inhibitors of these
enzymes, as well as others that may become amenable to analysis by
ABPP, within the confines of the proteomes in which they are natu-
a
b
c
VOLUME 21 NUMBER 6 JUNE 2003 NATURE BIOTECHNOLOGY
© 2003 Nature Publishing Group http://www.nature.com/naturebiotechnology
LETTERS
rally expressed should accelerate the discovery of new chemical agents
for both basic research and therapeutic pursuits.
Received 7 November 2002; accepted 24 February 2003
Published online 12 May 2003; doi:10.1038/nbt826
METHODS
1. Maly, D.J., Huang, L. & Ellman, J.A. Combinatorial strategies for targeting protein
families: application to the proteases. Chembiochem. 3, 17–37 (2002).
2. Houghton, R.A. Parallel array and mixture-based synthetic combinatorial chemistry:
tools for the next millenium. Annu. Rev. Pharmacol. Toxicol. 40, 273–282 (2000).
3. Fernandes, P.B. Technological advances in high-throughput screening. Curr. Opin.
Chem. Biol. 2, 597–603 (1998).
4. Liu, Y., Patricelli, M.P. & Cravatt, B.F. Activity-based protein profiling: the serine
hydrolases. Proc. Natl. Acad. Sci. USA 96, 14694–14699 (1999).
5. Patricelli, M.P., Giang, D.K., Stamp, L.M. & Burbaum, J.J. Direct visualization of serine hydrolase activities in complex proteomes using fluorescent active site-directed
probes. Proteomics 1, 1067–1071 (2001).
6. Kidd, D., Liu, Y. & Cravatt, B.F. Profiling serine hydrolase activities in complex proteomes. Biochemistry 40, 4005–4015 (2001).
7. Cravatt, B.F. et al. Molecular characterization of an enzyme that degrades neuromodulatory fatty acid amides. Nature 384, 83–87 (1996).
8. Dolinsky, V.W., Sipione, S., Lehner, R. & Vance, D.E. The cloning and expression of a
murine triacylglycerol hydrolase cDNA and the structure of the corresponding gene.
Biochem. Biophys. Acta 1532, 162–172 (2001).
9. Adam, G.C., Sorensen, E.J. & Cravatt, B.F. Chemical strategies for functional proteomics. Mol. Cell. Proteomics 1, 781–790 (2002).
10. Greenbaum, D. et al. Chemical approaches for functionally probing the proteome.
Mol. Cell. Proteomics 1, 60–68 (2002).
11. Greenbaum, D. et al. Small molecule affinity fingerprinting. A tool for enzyme family
subclassification, identification, and inhibitor design. Chem. Biol. 9, 1085–1094
(2002).
12. Jessani, N., Liu, Y., Humphrey, M. & Cravatt, B.F. Enzyme activity profiles of the
secreted and membrane proteome that depict cancer cell invasiveness. Proc. Natl.
Acad. Sci. USA 99, 10335–10340 (2002).
13. Boger, D.L. et al. Exceptionally potent inhibitors of fatty acid amide hydrolase: the
enzyme responsible for degradation of endogenous oleamide and anandamide. Proc.
Natl. Acad. Sci. USA 97, 5044–5049 (2000).
14. Boger, D.L. et al. Trifluoromethyl ketone inhibitors of fatty acid amide hydrolase: a
probe of structural and conformational features contributing to inhibition. Bioorg.
Med. Chem. Lett. 9, 265–270 (1999).
15. Liang, T.C. & Abeles, R.H. Complex of α-chymotrypsin and N-acetyl-L-leucyl-L-phenylalanine trifluoromethyl ketone: structural studies with NMR spectroscopy.
Biochemistry 26, 7603–7608 (1987).
16. Trimble, L.A. et al. NMR structural studies of the tight complex between a trifluoromethyl ketone inhibitor and the 85 kDa human phospholipase A2. Biochemistry
32, 12560–12565 (1993).
17. Cravatt, B.F. et al. Supersensitivity to anandamide and enhanced endogenous
cannabinoid signaling in mice lacking fatty acid amide hydrolase. Proc. Natl. Acad.
Sci. USA 98, 9371–9376 (2001).
18. Alam, M., Vance, D.E. & Lehner, R. Structure-function analysis of human triacylglycerol hydrolase by site-directed mutagenesis: identification of the catalytic triad and a
glycosylation site. Biochemistry 41, 6679–6687 (2002).
19. Eisen, M.B., Spellman, P.T., Brown, P.O. & Botstein, D. Cluster analysis and display
of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95, 14863–14868
(1998).
20. Adam, G.C., Cravatt, B.F. & Sorensen, E.J. Profiling the specific reactivity of the proteome with non-directed activity-based probes. Chem. Biol. 8, 81–95 (2001).
21. Adam, G.C., Sorensen, E.J. & Cravatt, B.F. Proteomic profiling of mechanistically distinct enzyme classes using a common chemotype. Nat. Biotechnol. 20, 805–809
(2002).
22. Kathuria, S. et al. Modulation of anxiety through blockade of anandamide hydrolysis.
Nat. Med. 9, 76–81 (2003).
23. Bracey, M.H., Hanson, M.A., Masuda, K.R., Stevens, R.C. & Cravatt, B.F. Structural
adaptations in a membrane enzyme that terminates endocannabinoid signaling.
Science 298, 1793–1796 (2002).
24. Leung, D., Abbenante, G. & Fairlie, D.P. Protease inhibitors: current status and future
prospects. J. Med. Chem. 43, 305–341 (2000).
25. Borodovsky, A. et al. Chemistry-based functional proteomics reveals novel members
of the deubiquitinating enzyme family. Chem. Biol. 9, 49–59 (2002).
Competitive proteomic profiling of FAAH inhibitors with FP-rhodamine.
Mouse tissues were Dounce-homogenized in Tris buffer (50 mM Tris-HCl,
pH 8.0) with 320 mM sucrose and membrane proteomes isolated by centrifugation at 4 °C (22,000g), washed, resuspended in Tris buffer, and
adjusted to a protein concentration of 1 mg/ml. Proteomes were preincubated with inhibitors (10–100,000 nM; DMSO stocks) for 10 min and then
treated with FP-rhodamine (100 nM, DMSO stock; Activx Biosciences5) at
room temperature for 10 min (final DMSO concentration 2%). Reactions
were quenched with SDS-PAGE loading buffer, subjected to SDS-PAGE, and
visualized in-gel using a flatbed fluorescence scanner (MiraBio). Labeled
proteins were quantified by measuring integrated band intensities (normalized for volume); control samples (DMSO alone) were considered 100%
activity and inhibitor-treated samples were expressed as a percentage of
remaining activity. Potent inhibitors (IC50 values < 10 nM) were also tested
at 0.5, 1 and 5 nM with proteomes adjusted to 0.1 mg protein/ml, so that the
estimated concentration of target enzymes was kept at least five-fold below
the inhibitor concentration (enzyme concentrations were estimated by
ABPP as described12). IC50 values were determined from dose-response
curves from three trials at each inhibitor concentration using Prism software
(GraphPad). See Supplementary Methods online for information on how
IC50 data were transformed for cluster analyses (Fig. 3).
Isolation, identification and recombinant expression of targets of the
inhibitor library. Enzyme targets were affinity-isolated using a biotinylated FP
probe and avidin-agarose beads (Sigma) as described6,12. Tryptic peptide maps
of the isolated proteins were analyzed by a combination of MALDI (Kratos
Analytical Axima CFR Reflector MALDI-TOF MS instrument) and LC-ESI
tandem MS (Finnigan LCQ Deca ion trap MS; Thermo Finnigan). The MS data
were used to search public databases to identify enzymes. cDNAs for mouse
TGH and KIAA1363 were purchased as expressed-sequence tags (ESTs;
Invitrogen), sequenced, subcloned into the eukaryotic expression vector
pcDNA3 and transiently transfected into COS-7 cells as described4. Membrane
fractions from transfected cells were isolated by centrifugation and analyzed as
described earlier.
Inhibition studies of TGH using substrate assays. TGH inhibition was assayed
using COS-7–expressed enzyme and the substrate p-nitrophenyl laurate as
described in the Supplementary Methods online.
Note: Supplementary information is available on the Nature Biotechnology website.
ACKNOWLEDGMENTS
We thank members of the Cravatt group for helpful discussions, G. Adam for
sulfonate probes and J. Leszyk and J. Wu for assistance with MS analysis. This work
was supported by National Institutes of Health grants CA87660 (B.F.C.), DA13173
(B.F.C.) and DA15648 (D.L.B.), by Activx Biosciences and by the Skaggs Institute
for Chemical Biology.
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests (see the Nature Biotechnology
website for details).
NATURE BIOTECHNOLOGY VOLUME 21 NUMBER 6 JUNE 2003
691