Biology Reference
In-Depth Information
of the relationship between specific biological processes and
chemical stresses by examining the relevant chemical
structures. For example, we identified ring assemblies that
are significantly enriched among small molecules for which
ergosterol biosynthesis is important for resistance (Figure
8.4). Ultimately, establishing such relationships between
biological processes and chemical classes would permit the
inference of biological effects based on chemical structure
alone, and thus greatly simplify the characterization of the
effects of specific small molecules and guide the develop-
ment of drugs. To further exploit our dataset, we identified
relationships between cellular processes in the context of
chemical stress (
Figure 8.4
B). Specifically, we identified
processes that are frequently enriched alone or together in
the HIPHOP profiles of tested small molecules. The
processes define a high-level signature of the cellular
response to a subset of chemical stresses. As described
before, it would be interesting to investigate whether small
molecules inducing the same signatures have common
structural features. For example, these signatures may reveal
that a specific chemical class targets multiple processes.
Similarly, processes that co-occur may be indicative of
a functional relationship. For example, although several
profiles are enriched for vesicle-mediated transport genes,
a subset of these compounds are associated with both
vesicle-mediated transport and protein localization
(
Figure 8.4
B). As vesicle-mediated transport plays an
important role in protein trafficking, a functional relation-
ship between these main processes is expected. Additional
study is required to characterize the identified relationships
between cellular processes, and also their relationship back
to specific chemical stresses; however, efforts in this
direction are sorely needed for understanding the broad
effects and/or polypharmacology of compounds, particu-
larly for the development of therapeutics for complex
diseases.
non-genetic assays: for example, sodium fluoride,
cantharidin, and chemical probes targeting Tub2, Sec7,
Pma1, Sec14, Sec13, Ero1, Ole1 and Erg7
[38]
. In support
of the validity of this 'theme', we previously demonstrated
that a 'perfect' drug behaves 'perfectly' in the HIPHOP
assay by using an engineered CDC28 kinase inhibitor that
specifically acts only on a CDC28 allele designed to 'fit'
the inhibitor. Ligand analogs, such as the engineered
CDC28 inhibitor, have been well established as highly
potent and highly specific inhibitors of analog-sensitive
alleles
[99,100]
. The resulting HIP chemogenomic profile
of the CDC28 inhibitor revealed an extraordinary degree of
specificity for the Analog-sensitive allele, no sensitivity for
the wild-type heterozygous CDC28 deletion strain or any
other strain
[64]
. In sum, while modern drug and chemical
probe discovery aims to identify small molecules that
inhibit a single protein with high specificity, and the
HIPHOP platform can clearly identify such molecules,
<
1% of the screened molecules to date demonstrate this
property, suggesting that single-hit specificity rarely occurs
in vivo.
Overall,
from our
large-scale
discovery
effort
involving some 20
genetic interac-
tion measurements, we have so far discovered ~100 novel
inhibitors that hold promise for development into vali-
dated chemical probes. Over the course of the screening
campaign the rate of HIP target discovery has been
constant, that is, the number of HIP targets identified
increased linearly with the number of small molecules
screened. Moreover, the discovery rates are nearly
equivalent for druggable compared to undruggable HIP
targets. Significantly, nearly half of the HIP targets are
considered 'undruggable'. This confirms the suspicions
that the classification of a protein as undruggable is not
due to an inherent inability of the protein to bind or be
inhibited by a small molecule, but rather more likely due
to the protein never having been explored in HTS
campaigns and therefore, a lack of empirical evidence to
support the definition of 'druggable'.
Based on our extensive experience, supported by the
data provided in this chapter and the accumulating data in
the literature, the in vitro MDD paradigm (with important
exceptions) is unlikely to succeed as one that is viable, and
many describe it as having been 'played out'. This also
applies to the chemical probe initiative by the Molecular
Libraries Screening Centers Network and others, because,
while the focus is to develop molecular tools rather than
drugs, the strategies in large part parallel those defined by
MDD. Moreover, in order to qualify as a chemical probe,
stringent criteria must be met. Many have expressed
a view that these standards have been set too high, e.g.,
most approved drugs do not meet these criteria
[101]
.
Indeed, a recent analysis found that oral drugs seldom
possess nanomolar potency that is predicted by in vitro
million chemical
รพ
e
CONCLUSIONS AND FUTURE
CHALLENGES
Over the last decade the HIPHOP platform has been
developed, rigorously validated and applied in over 10 000
chemogenomic screens. From the beginning, the data
generated in these screens have reported several themes in
a consistent and recurring fashion. First, most small
molecules do not act by inhibiting a single protein in the
cell; however, there are exceptions. Defined by their highly
specific HIP chemogenomic profiles,
these exceptions,
representing
30 of the ~3500 small molecules screened,
include: (1) FDA-approved drugs: methotrexate, statins,
rapamycin, and the antifungal terbinafine; (2) established
chemical probes: tunicamycin, cerulinin and aureobasidin;
and (3) novel chemical probe, counting only those that we
identified and verified in vitro or validated using other
<
