Chemistry & Biology 13, 1041–1050, October 2006 ª2006 Elsevier Ltd All rights reservedDOI 10.1016/j.chembiol.2006.08.008
An Enzyme that Regulates Ether
Lipid Signaling Pathways in Cancer
Annotated by Multidimensional Profiling
Kyle P. Chiang,1,2Sherry Niessen,1,2
Alan Saghatelian,1and Benjamin F. Cravatt1,*
1The Skaggs Institute for Chemical Biology and
Departments of Cell Biology and Chemistry
The Scripps Research Institute
10550 North Torrey Pines Road
La Jolla, California 92037
Hundreds, if not thousands, of uncharacterized en-
zymes currently populate the human proteome.
Assembly of these proteins intothe metabolic and sig-
naling pathways that govern cell physiology and pa-
thology constitutes a grand experimental challenge.
Here, we address this problem by using a multidimen-
sional profiling strategy that combines activity-based
proteomics and metabolomics. This approach deter-
mined that KIAA1363, an uncharacterized enzyme
highly elevated in aggressive cancer cells, serves
as a central node in an ether lipid signaling network
that bridges platelet-activating factor and lysophos-
phatidic acid. Biochemical studies confirmed that
KIAA1363 regulates this pathway by hydrolyzing the
metabolic intermediate 2-acetyl monoalkylglycerol. In-
activation of KIAA1363 disrupted ether lipid metabo-
lism in cancer cells and impaired cell migration and
tumor growth in vivo. The integrated molecular profil-
tional annotation of metabolic enzymes in any living
Elucidation of the metabolic and signaling networks that
regulate health and disease stands as a principal goal
of postgenomic research. The remarkable complexity
of these molecular pathways has inspired the advance-
ment of ‘‘systems biology’’ methods for their character-
ization . Toward this end, global profiling technologies,
such as DNA microarrays [2, 3] and mass spectrometry
(MS)-based proteomics [4, 5], have succeeded in gener-
ating gene and protein signatures that depict key fea-
tures of many human diseases. However, extricating
from these associative relationships the roles that spe-
cific biomolecules play in cell physiology and pathology
remains problematic, especiallyfor proteins ofunknown
biochemical or cellular function.
The functions of certain proteins, such as adaptor or
scaffolding proteins, can be gleaned from large-scale
protein-interaction maps generated by technologies
like yeast two-hybrid [6, 7], protein microarrays , and
MS analysis of immunoprecipitated protein complexes
cesses principally through catalysis. Thus, elucidation
of the activities of the many thousands of enzymes en-
coded by eukaryotic and prokaryotic genomes requires
knowledge of their endogenous substrates and prod-
ucts. The functional annotation of enzymes in prokary-
otic systems has been facilitated by the clever analysis
of gene clusters or operons [11, 12], which correspond
to sets of genes adjacently located in the genome that
encode for enzymes participating in the same metabolic
cascade. The assembly of eukaryotic enzymes into met-
abolic pathways is more problematic, however, as their
nized into operons, but rather are scattered randomly
throughout the genome.
Given the absence of a functional architecture con-
necting eukaryotic genomes and proteomes, the activi-
ties of their enzyme constituents are typically assessed
in an empirical manner in vitro by using candidate sub-
strates and purified preparations of protein. The out-
come of these ‘‘test-tube’’ biochemistry studies can be
difficult to translate into a clear understanding of the
roles that enzymes play in living systems, where these
proteins are subjected to posttranslational regulation
 and typically operate as parts of larger metabolic
networks . We hypothesized that the determination
of endogenous catalytic activities for uncharacterized
enzymes could be accomplished directly in living sys-
tems by the integrated application of global profiling
technologies that survey both the enzymatic proteome
and its primary biochemical output (i.e., the metabo-
lome). Here, we have tested this premise by utilizing
multidimensional profiling to characterize an integral
membrane enzyme of unknown function that is highly
elevated in human cancer.
Development of a Selective Inhibitor for the
Uncharacterized Enzyme KIAA1363
Previous studies using the chemical proteomic technol-
ogy activity-based protein profiling (ABPP) [15–17] have
identified enzyme activity signatures that distinguish
human cancer cells based on their biological properties,
including tumor of origin and state of invasiveness .
A primary component of these signatures was the pro-
tein KIAA1363, an uncharacterized integral membrane
hydrolase found to be upregulated in aggressive cancer
cells from multiple tissues of origin. Since that time, the
mouse ortholog of KIAA1363 has been found to repre-
agents in brain tissue ; however, the endogenous
metabolic function(s) of this enzyme in mammalian
physiology and pathology remains unknown. To investi-
gate the role that KIAA1363 plays in cancer cell metabo-
lism and signaling, a selective inhibitor of this enzyme
was generated by competitive ABPP [20, 21].
Key advantages of competitive ABPP include that
it can be performed in native proteomes and used to
identify inhibitors for enzymes that, like KIAA1363, lack
known substrates. Moreover, because inhibitors are
screened against many enzymes in parallel, both
2These authors contributed equally to this work.
potency and selectivity factors are simultaneously
assigned. Previous competitive ABPP screens with a
library of candidate inhibitors and a fluorophosphonate
(FP) activity-based probe that targets the serine hydro-
lase superfamily identified a set of trifluoromethyl ke-
tone (TFMK) inhibitors that showed activity against the
mouse ortholog of KIAA1363 in brain extracts .
These TFMK inhibitors also inhibited human KIAA1363
in vitro, but they showed only limited activity in living
cells (data not shown). We postulated that the in situ
activity of KIAA1363 inhibitors could be enhanced by
replacing the reversibly binding TFMK group with
a carbamate, which inactivates serine hydrolases via
a covalent mechanism (Figure S1; see the Supplemental
Data available with this article online). Carbamate
AS115 (Figure 1A) was synthesized and tested for its
effects on the invasive ovarian cancer cell line SKOV-3
by competitive ABPP (Figure 1B). AS115 was found to
potently and selectively inactivate KIAA1363, dis-
playing an IC50value of 150 nM, while other serine hy-
drolase activities were not affected by this agent (IC50
values > 10 mM) (Figures 1B and 1C). AS115 also
selectively inhibited KIAA1363 in other aggressive can-
cer cell lines that possess high levels of this enzyme,
including the melanoma lines C8161 and MUM-2B
Profiling the Metabolic Effects of KIAA1363
Inactivation in Cancer Cells
We next compared the global metabolite profiles of
SKOV-3 cells treated with AS115 or vehicle (DMSO)
to identify endogenous small molecules regulated by
KIAA1363. These experiments were performed by using
a recently described, untargeted liquid chromatogra-
phy-mass spectrometry (LC-MS) platform for compara-
tive metabolomics . AS115 (10 mM, 4 hr) was found
to cause a dramatic reduction in the levels of a specific
set of lipophilic metabolites (m/z 317, 343, and 345) in
SKOV-3 cells (Figure 2A). These KIAA1363-regulated
metabolites did notcorrespond toanyofthetypicallipid
species found in cells, including free fatty acids, phos-
pholipids, ceramides, and monoacylglycerides, none
of which were significantly altered by AS115 treatment
(TableS1). High-resolution MSofthe m/z317metabolite
provided a molecular formula of C19H40O3(Figure 2B),
which suggests that this compound might represent
a monoalkylglycerol ether bearing a C16:0 alkyl chain
(C16:0 MAGE). This structure assignment was corrobo-
rated by tandem MS and LC analysis, in which the en-
dogenous m/z 317 product and synthetic C16:0 MAGE
displayed equivalent fragmentation and migration pat-
terns, respectively (Figure S3). By extension, the m/z
343 and 345 metabolites were interpreted to represent
Figure 1. Characterization of AS115, a Selec-
tive Inhibitor of the Cancer-Related Enzyme
(A) Structure of AS115.
(B) Effects of AS115 on the membrane (left
image) and soluble (right image) serine
hydrolase activity profiles of SKOV-3 cells,
as judged by competitive ABPP with a rhoda-
mine-tagged FP probe . In-gel fluores-
cence scanning of FP-labeled proteomes
derived from SKOV-3 cells treated in culture
with AS115 (0–10 mM) revealed selective
inactivation of KIAA1363 (red box). Note
that KIAA1363 migrates by SDS-PAGE as
a 43 and 45 kDa glycosylated doublet,
which, upon treatment with PNGaseF, is
converted into a single 40 kDa protein. This
protein is predominantly found in the mem-
brane fraction of cancer cells (left image).
(C) AS115 inhibited the FP labeling of
KIAA1363 with an IC50 value of 150 nM
(110–200 nM, 95% confidence limits; red
curve), while other serine hydrolases were
not affected by this reagent (IC50values >
10 mM, representative hydrolases shown in
black curves corresponding to the double-
arrowed proteins in [B]). Results represent
the average values 6 standard error (SE)
for three independent experiments.
Chemistry & Biology
3-kinase inhibitor, also potently inhibits mammalian polo-like
kinase. Chem. Biol. 12, 99–107.
43. Yee, M.C., Fas, S.C., Stohlmeyer, M.M., Wandless, T.J., and
Cimprich, K.A. (2005). A cell-permeable activity-based probe
for protein and lipid kinases. J. Biol. Chem. 280, 425–437.
44. Barglow, K.T., and Cravatt, B.F. (2004). Discovering disease-
associated enzymes by proteome reactivity profiling. Chem.
Biol. 11, 1523–1531.
45. Adam, G.C., Sorensen, E.J., and Cravatt, B.F. (2002). Proteomic
mon chemotype. Nat. Biotechnol. 20, 805–809.
46. Speers, A.E., and Cravatt, B.F. (2005). A tandem orthogonal
proteolysis strategy for high-content chemical proteomics. J.
Am. Chem. Soc. 127, 10018–10019.
47. Kidd, D., Liu, Y., and Cravatt, B.F. (2001). Profiling serine hydro-
lase activities in complex proteomes. Biochemistry 40, 4005–
48. Patricelli, M.P., Giang, D.K., Stamp, L.M., and Burbaum, J.J.
(2001). Direct visualization of serine hydrolase activities in com-
plex proteome using fluorescent active site-directed probes.
Proteomics 1, 1067–1071.
49. Fuji, M., Watanabe, F., Fujii, Y., Hashizume, H., Okuno, T., Shir-
ahase, K., Teshirogi, I., and Ohtani, M. (1997). A stereoselective
and highly practical synthesis of cytosolic phospholipase A2
3-O-phosphocholine. J. Org. Chem. 62, 6804–6809.
(RS)- and (S)-1-hexadecyl-2-acetylglycero-thiophosphocholine.
J. Chem. Soc. Chem. Commun. 21, 1608–1609.
51. Smith, C.A., Want, E.J., O’Maille, G., Abagyan, R., and Siuzdak,
G. (2006). XCMS: processing mass spectrometry data for me-
tabolite profiling using nonlinear peak alignment, matching,
and identification. Anal. Chem. 78, 779–787.
52. Ritter, K. (1991). Affinity purification of antibodies from sera us-
ingpolyvinylidenedifluoride (PVDF) membranesascoupling ma-
trices for antigens presented by autoantibodies to triosephos-
phate isomerase. J. Immunol. Methods 137, 209–215.
53. Jessani, N., Humphrey, M., McDonald, W.H., Niessen, S., Ma-
suda, K., Gangadharan, B., Yates, J.R., III, Mueller, B.M., and
Cravatt, B.F. (2004). Carcinoma and stromal enzyme activity
profiles associated with breast tumor growth in vivo. Proc.
Natl. Acad. Sci. USA 101, 13756–13761.
54. Tanyi, J.L., Morris, A.J., Wolf, J.K., Fang, X., Hasegawa, Y.,
Lapushin, R., Auersperg, N., Sigal, Y.J., Newman, R.A., Felix,
E.A., et al. (2003). The human lipid phosphate phosphatase-3
decreases the growth, survival, and tumorigenesis of ovarian
cancer cells: validation of the lysophosphatidic acid signaling
cascade as a target for therapy in ovarian cancer. Cancer Res.
Chemistry & Biology