Biology Reference
In-Depth Information
identified strains pass the first two criteria). The final step
ensures that the small molecule acts with a high degree of
specificity. For example, the HIP profile of the chemical
probe tunicamycin clearly shows that the probe exhibits
a high degree of specificity, and identifies the heterozy-
gous deletion strain alg7
assays consistent with specific inhibition of protein activity
[38]
. Thus far we have successfully validated 26% of the
novel HIP targets, demonstrating HIP as a robust platform
for identifying both known and novel chemical probe and
their cognate protein targets.
/ALG7 as having the greatest
fitness defect, thereby successfully identifying the known
yeast target of tunicamycin, ALG7 (
Figure 8.2
A). As
a result of inhibiting ALG7, which encodes a protein
required in the first step of N-linked glycosylation,
protein folding in the endoplasmic reticulum is disrupted,
which in turn induces the unfolded protein response
(UPR). The UPR pathway has been extensively charac-
terized and the primary cellular response is transmitted
via the translational upregulation of the HAC1 transcrip-
tion factor through an unconventional splicing mecha-
nism by IRE1 (e.g.,
[63]
). Both genes are 'classic'
reporters of the UPR. As is readily apparent in the HOP
profile of tunicamycin (
Figure 8.2A
), these genes
and others exhibit highly significant chemical
D
HIP Predicts the Yeast Druggable Genome is
Double that of Current Estimates
From a global perspective, analysis of our HIP targets
revealed a significant enrichment for 'druggable'
proteins, i.e., those with precedence for inhibition by
'drug-like' small molecules, relative to the essential
yeast genome. While the definition of 'druggable' is
contentious, we observed similar enrichments using five
different metrics that estimate druggability (
2
c
test, p
<
0.001). Since the beginning of our ~3500 compound
screen, the number of HIP targets (defined by our
criteria) identified has increased approximately linearly
with the number of compounds screened, and the rate of
discovery of druggable compared to undruggable HIP
targets is nearly identical. Assuming both discovery rates
remain constant as more small molecules are screened
has allowed an empirical prediction of the size of the
yeast druggable genome by simple extrapolation. This
analysis results in a prediction that 44% of the essential
yeast genome is druggable by our HIP target criteria,
nearly doubling current estimates
[31]
.Byextension,
using yeast
genetic
interactions with the probe. Moreover, GO-based
functional enrichment analysis of all genes exhibiting
chemical
e
genetic interactions with tunicamycin high-
lights the biological processes that are both directly
(e.g., 'glycosylation') and indirectly related to the target
of the probe (e.g., 'vacuolar transport';
Figure 8.2
B).
Taken together, the most significant interaction in the HIP
profile
e
identifies
the drug target, while
the HOP
assay identifies the chemical
genetic interactions that
buffer the target pathway, and when combined, the
enrichment profile provides support for the MOA of
tunicamycin, clearly characterizing the cellular response
to the probe.
In addition to identifying the known targets of
commonly used chemical probes (e.g., tunicamycin,
rapamycin), we also rediscovered the targets of two
compounds that were originally identified in our labora-
tory and published in separate studies (erodoxin and
cantharidin
[27,64]
). To successfully identify targets with
the above HIP target method, the small molecule must
inhibit growth and the targets must have yeast homologs
that are essential. Overall, we successfully identified the
known targets of the screened molecules that meet these
criteria, with few exceptions
[65,66]
.
To date, our efforts to identify novel chemical probes,
using the HIP assay and the criteria described above, have
resulted in the identification of over 100 specific inhibitors
that target 54 unique essential gene products. From the
novel small molecule
e
human orthology, a prediction of the human
druggable genome is also twice that of current estimates.
This result confirms the suspicion that the portion of the
genome currently defined as undruggable is not likely to
be due to inherent 3D chemical properties characteristic
of the encoded proteins, but is more likely a designation
assigned by default due to the focus of the pharmaceu-
tical industry on proven druggable targets and a corre-
sponding lack of exploration of other proteins in HTS
campaigns
[67]
.Insum,ourlargeHIPHOPdatase ,
collected in a systematic and unbiased manner, resulted
in the first empirically based estimate of the yeast
druggable genome, and by extension, the human drug-
gable genome. We have identified lead chemical probes
for54proteinssofa ,overha fofwhichtarget
undruggable proteins, representing a valuable resource to
the yeast and chemical genetics communities, as well as
to the systems biology community at large.
e
HIP target interactions, six were
validated by direct in vitro enzymatic (Ole1, Pma1, Sec14,
Sec7) or polymerization/binding (Tub2, Arc19) assays and
eight using in vivo measurements (Erg7, Sec13, Ssl2, Fas1,
Erg1, Cdc12, Tor2, Erg11), including metabolomics,
metabolite complementation, and fluorescence microscopy
e
Co-Inhibition Reflects Structure
and Therapeutic Class
A HIPHOP profile is a genome-wide in vivo snapshot of the
cellular response to a small molecule, and as clearly illus-
trated in the tunicamycin profiles (
Figure 8.2
A), highlights
