High-Throughput
Chemistry-Alleviating Bottlenecks in Drug Lead
Generation
(source: Drug
& Market Development
Publications Feb 1999)
- Combinatorial chemistry methodology can
produce massive libraries of compounds, but may
do little to alleviate major bottlenecks in drug
discovery programs requiring more focused and
rationally driven chemical synthesis for lead
optimization. Alternative approaches that
enhance the throughput of compound library
generation, but provide more control over
compound synthesis, are becoming increasingly
competitive and cost effective.
- A successful development candidate must have
suitable pharmacokinetics and an acceptable
safety profile and, traditionally, these
properties are defined by low through-put in
vivo testing far downstream of the initial in
vitro drug lead identification process. Novel
approaches are now being implemented to
alleviate this bottleneck by high-throug-hput in
vitro compound analysis.
- This article focuses on selected new
approaches to com-pound synthesis and analysis
(termed here "high-through-put chemistry"), with
the goal of anticipating the role such
approaches may play in the near future and
pre-dicting their long-term impact in the drug
dis-covery process.
- ArQule (Medford, MA), whose technology
hybridizes combinatorial chemistry with
high-throughput parallel synthesis, analysis and
purification, is one of the firms reviewed in
this article. Such "first tier" combinatorial
chemistry companies are encountering competition
from the internal combichem efforts within large
pharma-ceutical companies, contract research
organi-zations (CROs), and purveyors of
instrumentation and devices for high-speed
parallel synthesis (e.g. Argonaut and IRORI,
also reviewed here).
- This article was written by Stan M. Goldin,
Ph.D., Department of Biological Chemistry and
Molecular Pharmacology, Harvard Medical School,
and President, Goldin R&D Group. Phone:
617-861-9431. Email: goldinstan@aol.com.
Introduction and Overview
Even with modern advances in combinatorial
chemistry and pharmacogenomics in the 1990s, the
drug discovery process resemblesin its basic
formatthe traditional approach developed and
refined in earlier decades. Drug discovery
strategies still typically involve the following
sequence of steps: (a) identification and
validation of a drug target; (b) generation or
acquisition of compound libraries; (c) development
of screening assays; (d) the compound screening
process; (e) drug lead identification; and (f) drug
lead optimization. Step (a) has been enhanced by
advances in genomics and new information on the
molecular mechanisms responsible for disease, which
together have generated a proliferation of new drug
targets (cf D&MD 9:37; D&MD 9:64; D&MD
9:338). Step (b) has been impacted, if not
revolutionized, by combinatorial chemistry
technologies, which are now capable of gener-ating
libraries of millions of compounds (cf D&MD
9:155). Novel high-throughput screening (HTS)
strategies are greatly accelerating the primary
stages of the drug screening process [steps (c)
& (d), cf D&MD 9:89; D&MD 9:119].
Steps (e) and (f) commonly comprise the biggest
bottlenecks in the drug discovery process. In
addition to its primary biological activity
(potency and specificity for its drug target), a
suitable drug development candidate must have
appropriate physical properties (e.g. solubility
and cell membrane permeability), pharmacokinetics
and toxi-cology. In most cases, these downstream
steps still entail the use of conventional
"low-throughput" medicinal chemistry to synthesize
compounds with the desired properties. This is
typically coupled with low-throughput in vivo
char-acterization of the development
candidates pharmacology and toxicology.
Labor-intensive in vivo pharmacology is,
arguably, unavoidable in the final stages of
development candidate selection. But, even in the
face of convincing animal data, projects not
uncommonly fail because the efficacy,
pharma-cokinetics or toxicity of a compound in
humans funda-mentally differs from that predicted
by animal models (such frustrations are exemplified
by recent failures in stroke drug development, to
be discussed in a D&MD article next month). A
recent survey found that about 33% of failures of
drugs during clinical development are due to
inappropriate pharmacokinetics (Monro, A.M., 1997
in "Use of Human In Vitro Systems to Support
Preclinical and Clinical Safety Assessment",
Tyckgruppen Press, Sweden). Many would agree that
there is no substitute, at present, for the skill
and intuition of medicinal chemists, but there is
nonetheless a great need for more efficient
methodology and tools to translate a medicinal
chemists expertise into new compounds for
lead optimization.
A case can also be made for application of
high-through-put medicinal chemistry at an earlier
stage of the drug discovery process [step
(b)]. It is increasingly appre-ciated that in
certain cases, the subtleties of drug-target
inter-actions necessitate detailed mechanistic
studies of such interactions as soon as possible in
the screening process. Salient examples are drug
leads which are ion channel modulators and/or
ligands for G protein-coupled receptors. What is
needed in such cases is a "movie" of the
pharma-codynamics of drug-target interactions (such
as now affor-ded by conventional labor-intensive
electro-physiology), rather than a "snapshot"
provided by conventional HTS ligand-receptor
binding assays. To appreciate the current status
and future prospects of "high-throughput
chemistry", this article presents both an overview
of recent develop-ments and a more intensive
discussion of the three follow-ing firms:
ArQule, a combinatorial chemistry-focused
firm which is broadening its technology platform to
include high- throughput parallel synthesis and
chemical analysis.
Argonaut (San Carlos, CA ), a leading
designer and vendor of "turnkey" systems for
automated parallel synthesis.
IRORI (La Jolla, CA), which has developed
flexible technology that incorporates features of
both parallel synthesis and conventional
"split-and-pool" combina-torial chemistry.
Table 14 highlights selected novel approaches to
high- throughput compound synthesis. As shown,
significant technological innovations are arising
within large pharma-ceutical firms, as well as
within smaller technology plat-form companies.
Table 15 highlights selected novel ap-proaches to
high-throughput compound analysis and/or
purification. Again, a trend towards internal
technology development within big pharma is
apparent, and this com-petition has implications
for the future of smaller tech-nology and
instrumentation companies, the more visible source
of such innovations.
Combinatorial Chemistry Firms in Transition:
Focus on Arqule
Founded in 1993 by Joseph C. Hogan, Jr., Ph.D.,
and partners, ArQule rapidly rose over the next
several years to the front ranks of combinatorial
chemistry firms, achieving a valuation of over
~$300 million by 1997 (stock price approaching
$30/share). ArQules pharma-ceutical and
bio-technology collaborators number about 30, and
include Roche Biosciences (Palo Alto, CA), Abbott
Laboratories (Abbot Park, IL) and American Home
Products (Madison, NJ). ArQule also has joint
programs for lead generation and optimization with
a number of biotechnology companies including
Aurora Biosciences (San Diego, CA; mammalian
cell-based assays), Cadus Pharmaceuticals
(Tarrytown, NY; signal transduction), Cubist
Pharmaceuticals (Cambridge, MA; infectious
dis-eases), ICAgen (Durham, NC; ion chan-nel
targets), Scriptgen (Medford, MA; RNA/protein
inter-actions), Signal Pharmaceuticals (San Diego,
CA; gene tran-scrip-tion/transcription factors),
SUGEN (South San Francisco, CA; signal
transduction), T Cell Sciences [Need-ham, MA
(now Avant Immunotherapeutics); T cell
acti-vation/inhibition] and, most recently
(October 98), Ge-nome Therapeutics (Waltham,
MA) in the area of infec-tious disease. ArQule
provides its collaborative part-ners with two types
of synthetic compound libraries:
Mapping Array libraries, which are
diverse arrays of novel, discrete small molecule
compounds used for screening against biological
targets. Participation in ArQules Mapping
Array Program generally entails a multi-year
subscription to a group of compound libraries which
are designed around certain core structures or
themes selected by ArQule. ArQule has typically
pro-vided 15 to 20 Mapping Array compound sets,
which in aggregate total at least 100,000
compounds. Subscribers to this program are not
restricted to the targets the compounds may be
screened against.
Directed Array libraries, which are
arrays of analogs of a particular lead compound
synthesized for lead optimi-zation, typically with
exclusivity to the colla-borator. Beginning either
with a collaborators own lead com-pound or
with active compounds from the afore-mentioned
Mapping Array Program, the Directed Array
Programs objective is to optimize desired
compound properties, by an iterative process. It is
this program in particular that heavily draws on
ArQules capability in parallel synthesis.
ArQules technology platform comprises
modular building block strategies for combinatorial
synthesis, structure-guided design, information
technology and, not-ably, an Automated Molecular
Assembly Plant parallel synthesis system ("AMAP"),
using modular robotic work-stations, linked via a
computer network. According to company sources,
AMAP is capable of synthesizing, purify-ing and
analyzing thousands of novel compounds in
milli-gram quantities using either solution or
solid phase chemi-stries (cf 1998 US Patents
#5,736,412 & 5,712,171). AMAP also verifies
structural data throughout the synthesis process.
Compound structures are analyzed by flow injection
mass spectrometry and purity is determined by HPLC
coupled with UV and light scattering measurements
(cf Goetzinger, W.K., and Kyranos, J.N., American
Laboratory, April 1998:42). AMAPs compound
management system is designed to facilitate both
the introduction of the resulting compound arrays
into collaborators own screening assays and
the identification of biologically active
compounds. A computer tracking system identifies
the compound in each microtiter well and
characterizes its composition, structure and
synthesis.
ArQules technology platform has not yet
resulted in the commercialization of a product.
This, arguably, is not sur-prising for a firm this
young, considering the time frame required for drug
development. Nonetheless, the sub-stantial
deterioration in ArQules stock price in the
past year (from a high of 25.5 early in 1998 to a
December price of 5.0) appa-rently reflects
unfulfilled investor expectations (see Table 16 for
stock price history of selected combi-chem
firms).
Although this comes at a time when the
technology-rich NASDAQ is reaching record high
levels, other esta-blished combinatorial chemistry
firms [e.g. Pharmacopeia (Prince-ton, NJ) and
CombiChem (San Diego, CA)] have shared the same
fate, as Table 16 illustrates. The erosion in stock
price is not as extreme as for certain "fully
integrated drug discovery" companies whose flagship
drugs have recently failed in clinical trials (the
two cases shown in Table 16 are Cambridge
NeuroScience and CoCensys, each of whom had NMDA
antagonists for stroke that fell out of the
clinicthe ion channel blocker Cerestat and
the gly-cine site antagonist ACEA 1021,
respectively). On the other hand, despite the
skittishness of biotech investors this year, such
bad news is not characteristic of all technology
platform-focused bio-tech firms that came of age in
the 1990s. For example, InCyte Pharmaceuticals
(Palo Alto, CA), a genomics/bio-inform-atics
leader, and Sepracor (Marlborough, MA), whose
tech-nology centers around developing biologically
active isomers of previously successful drugs (such
as Pfizers Pro-zac®, soon to go off
patent), have enjoyed considerable finan--cial
success.
Several factors may account for the lackluster
perfor-mance of most combichem firms of late,
including:
Commoditization of combinatorial
chemistry. Several CROs, among them MDS Panlabs
(Bothell, WA) and Medi-Chem Research (Lemont, IL),
offer combichem on a fee-for-service basis, with
little or no expectation of down-stream
royalties.
Development by big pharmaceutical
companies of internal expertise in combichem and
other areas of high throughput chemistry, e.g.
Glaxo-Wellcome (London, UK), Merck & Co.
(Whitehouse, NJ) and Bristol-Myers Squibb
(Prince-ton, NJ).
Failure of combinatorial chemistry firms
to deliver pro-mising development candidates as
rapidly as investors had expected. How many
combinatorially designed com-pounds have made it
into advanced clinical develop-ment so far? Very
few.
In the specific case of ArQule, its ongoing
transition from a supplier of molecules for drug
discovery to a drug discovery company with a more
fully integrated internal technology platform may
also have fed investor uncer-tainty, as have the
recent resignations of two key exe-cutives. The
founding CEO Eric Gordon resigned in July 1998;
according to company sources, its Board of
Direct-ors (including Mr. Gordon), voted to look
for a new CEO with "...extensive experience in
leading large-scale drug discovery activities."
Recently (November 1998), Dr. David L. Coffen,
Senior Vice President of Chemistry, also announced
his resignation. However, it is reasonable to
suggest that ArQules short-term financial
difficulties and personnel changes may result from
strategic decisions that will pay off in the long
run. Last summer, the company began construction of
a new 130,000-square-foot head-quarters and R&D
facility in Woburn, MA which will double
ArQules current space, affording room for an
additional 100 employees. In addition,
ArQules revenues from its expanding base of
collaborators have increased substantially. For the
nine months ending September 30, 1998, revenues
were $16,283,000, up 45% from revenues of
$11,227,000 in the first nine months of 1997.
ArQules larger capital loss of $3,850,000, or
($0.32) per share, compared to a net loss of
$427,000, or ($0.04) per share, for the same period
last year; the increased losses resulted primarily
from increased expenditure for its expansion into
new chemistry-based drug discovery technologies.
Vital qualities for a new CEO include a strong
scientific background along with substantial
experience in drug discovery, as is what
characterizes ArQules most recently appointed
directors, Patrick Gage (Wyeth-Ayerst
Researchs President) and Dr. Michael
Rosenblatt (formerly of Merck & Co; Rahway,
NJ).
Argonaut Technologies Parallel
Synthesis Machinery
In addition to CROs offering combichem on a
contract basis, firms such as ArQule are facing
another dimension of competition from companies
offering off-the-shelf instru-mentation and
technical support services for high- throughput
compound library generation. This business
strategy, implemented by firms such as privately
held Argonaut (San Carlos, CA), is more analogous
to the way in which a provider would interact with
the purchaser of a large piece of analytical
instrumentation, such as a high-end NMR machine.
Argonauts first product, the Nautilus
(current embodiment, the Nautilus 2400) enables
totally automated organic synthesis of up to 24
compounds in parallel, in either solution- or
solid-phase. Argonaut claims that, "...the Nautilus
offers the flexibility to synthesize small
molecules using a broad range of chem-istries. The
only limit is your imagination." How-ever, the
major current concern of users of the machine is
that its design, while affording efficiency
increases for some synthetic strategies, may have
more limited applicability for certain other
classes of chemical synthesis. Argonaut has been
receptive to feedback from its current con-sortium
of customers, and its products are continually
evol-ving to obviate these concerns. For example,
Argo-naut recently introduced five new reaction
vessels for the Nautilus, designed to enhance the
machines versatility for lead
optimization.
Argonauts newest product, introduced last
Fall, is the "Trident" Automated Library
Synthesizer. Trident is de-signed for the
production of focused libraries, usually made after
primary screening steps (analogous to the step for
which ArQules Directed Array Program is
applied). Tri-dent is capable of parallel synthesis
of up to 192 com-pounds per run in an inert
environment, preventing un-wanted side reactions.
It affords automated control of the key synthetic
parameters, including solvent and reagent delivery,
temperature control, agitation, rinsing and
wash-ing, product cleavage and compound collection.
Tri-dent was conceived as the result of a
consortium between Argo-naut and the following
group of pharmaceutical com-panies, which
designated their own scientists to work as a team
with Argonaut:
- Abbott Laboratories
- Ariad Pharmaceuticals
- Astra AB
- DuPont
- Genetics Institute
- Merck & Co.
- Merck KgAA
- Rhône-Poulenc Rorer
- Sumitomo
According to David Binkley, Argonauts
president and CEO, "Input from the consortium was
invaluable in designing a flexible system that
supports the way chemists work." Mark A. Scialdone,
Ph.D., principal investigator in the combinatorial
chemistry group at DuPont Central Research &
Development (Wilmington, DE), one of the consortium
members, stated, "We often perform complex
reactions and need the control over reagent
delivery and reaction conditions that Trident
delivers."
IRORIs Proprietary Approach to Parallel
Synthesis: RF Coding and Directed Sorting Of
Individual Reactors
IRORI (La Jolla, CA), a privately held company
of ~80 employees founded in 1995, exemplifies a
promising new breed of high-throughput chemistry
company. IRORIs initial focus has been the
application of proprietary memory devices
[e.g., radio frequency (RF) tags and
two-dimensional optical bar codes] in
microreactor chambers to facilitate the design,
creation and management of chemical libraries
produced by parallel synthesis (see Figure 2).
IRORIs "Microreactors" are miniaturized
devices that contain both a functionalized solid
phase support and a unique tag identifier, and can
accommodate the synthesis of discrete compounds in
quantities of up to 100 mg. By splitting and
pooling microreactors using a process known as
"directed sorting," one discrete compound can be
synthesized in each microreactor, giving the
chemist the ability to generate and manage large
compound libraries simultaneously. As an indication
of the growing acceptance of IRORIs
tech-nology, one of Argonauts new parallel
synthesis machines (the Quest 205) is designed
expressly to provide compat-ibility with
IRORIs disposable microreactors.
In 1998, Irori introduced a family of supporting
products, software and automation, including the
following:
Synthesis Manager. A software package
that controls the "directed sorting" library
synthesis approach. It also serves as a database
for chemical building blocks, react-ion
intermediates and reaction procedures.
The AutoSort-10K Microreactor
Sorting System. Typical library sizes of discrete
compounds practicable by the IRORI technique are
under a thousand compounds, if sorted manually. The
AutoSort-10K automates the reaction and cleavage
sorting operations, accommodates up to 10,000
microreactors at a sorting rate of 1,000
reactors/hour, and thus enables a capability for
generation of larger compound libraries
(³10,000 compounds).
AccuCleave-96 Cleavage Station for
simultaneous cleaving of 96 compounds from the
solid polymer supports within the microreactor, and
transfer of the cleaved products to standard
96-well microplate format.
Concluding Remarks and
Prognostications
An exciting development may evolve from early
indications that an "industry standard" could be
emerging for parallel chemical synthesis
technology. A case in point: IRORI microreactors,
containing the coding devices to manage and track
nearly any type of chemical synthesis, may
conceivably become the preferred vehicle for
managing chemical synthesis (i.e. what the VHS
format is to current video recording standards).
Innovative firms such as IRORIand firms with
synergistic technology such as Argonautmay
work coopera-tively with big pharma-ceutical firms
(and with one another) in future consortium
arrangements to evolve increasingly sophisticated
and "user friendly," "massively parallel"
high-throughput chemical synthesis systems. The
ultimate goal of such high-throughput chemistry
systems is the synthesis of compounds free of the
practical constraints now encountered when a first
class medicinal chemist confronts current
embodi-ments of combinatorial chemistry or parallel
synthesis tech-nology. It is at that point that the
early stages of high- throughput screening will
increasingly benefit from the traditional expertise
of the medicinal chemist, and the process of lead
optimi-zation will be enhanced far beyond its
current capabilities. Extrapolation of current
developments and trends suggests that this might
occur very early in the new millennium.
Other developments, summarized in Table 15,
suggest that new technologies may soon profoundly
impact other major bottlenecks in lead
optimization; that is, creation of development
candidates with the most desirable human
pharmacokinetic and toxicological properties. A
forthcoming D&MD article will focus on recent
developments and future trends in this
arena.
|