The compound assets developed over many
years of drug discovery effort are some of the most valuable assets of a
drug discovery organization. Through the process of biological screening,
these compounds comprise the future leads that eventually will become drug
candidates. Therefore, it is critical that these compounds be managed
effectively and are widely accessible by the scientific staff. This review
will describe the role of the compound management function within the drug
discovery environment and..
| Jun
1, 2005 |
| By:
Michael
J. Sofia, Jay
M. Stevenson, John
Houston |
| Pharmaceutical
Discovery |
|
In any business, possession of strategic assets is recognized as
essential to attaining sustainable competitive advantage, but it is also
true that competitive advantage can only be achieved if these same
strategic assets are aggressively leveraged to meet business objectives.
Within any pharmaceutical drug discovery organization, strategic assets
can be classified into three general categories: the collective knowledge
or know-how of the scientific staff, intellectual property and physical
infrastructure. The chemical compounds synthesized over many years of drug
discovery endeavors, as well as those selected from many external sources,
comprise the output of many man-years of collective knowledge and
experience in the field. These compounds are the physical embodiment of
the creativity and innovation within a pharmaceutical drug discovery
organization and are, to a large extent, unique to each particular
organization. Through the process of biological and physicochemical
screening, in all its forms, more and more data is associated with each
compound, adding to their novelty, value and potential utility. This
collection of relatively unique, highly annotated compounds then becomes
the source from which new leads and, ultimately, future drugs are derived.
Thomke stated that, "Access to chemically diverse libraries of
compounds (compound collections) has thus been a very important
competitive advantage to pharmaceutical drug companies, and thus libraries
have been considered to be one of the pharmaceutical company's 'most
carefully guarded assets'" (1).
It could be argued, that the impact of screening and the integrity of
the screening data are only as good as the quality of the compounds being
tested. Therefore, any future success of the drug discovery endeavor is
critically dependent on the quality, integrity and diversity of the
compound collection, and consequently the need to protect, preserve and
efficiently manage this highly valuable and unique compound resource has
become an obvious business-critical activity. Compound management
organizations have arisen to manage this critical resource, and as the
process of drug discovery has become more sophisticated, these
organizations have developed into technologically advanced operations,
servicing a vast array of compound needs for global discovery operations
(2–9).
The Role of Compound Management Over
the past decade, the role of most drug discovery compound management
organizations has evolved; what started out as a somewhat peripheral
enterprise consisting of simple storerooms containing samples in boxes on
shelves, with manual retrieval and human readable codes, has become a
fully automated, highly integrated and centralized operation at the core
of the drug discovery process. The content of the compound collection, the
integrity of the samples that comprise it and access to samples for
screening of targets all impact the success and rate of lead discovery
efforts (10). The advent of highly automated screening capabilities that
can evaluate large numbers of samples against a biological target, the
development of combinatorial chemistry methods that produce a larger
number of compounds for biological evaluation and the overall drug
industry objectives of reducing timelines and increasing productivity
contributed to the need to more effectively manage compound collections
and make them readily available to drive discovery efforts.
In the current drug discovery environment, compound management is the
conduit through which all compounds pass from chemists to biologists. Its
mission has expanded to include not only the archive of all compounds
prepared or purchased in support of drug discovery programs, but also to
support all drug discovery biological testing by distributing everything
from single compound requests to large high-throughput screening (HTS)
decks. In some cases (e.g., Bristol-Myers Squibb, Wallingford,
Connecticut, USA), compound management also has the responsibility of
managing the content of the compound collection and the make-up of
screening decks, from small thematic compound sets to the large HTS decks.
In our view, a number of key functional responsibilities are required
for compound management to optimally accomplish its drug discovery support
mission. These responsibilities include sample registration and inventory
tracking, sample archiving, sample distribution, sample integrity
assurance and process quality control. To support these areas of
responsibility, state-of-the-art compound management functions rely on
sophisticated informatics, automation and process environments. However,
it is the integration of these capabilities into the drug discovery
environment that leverages the full value of the compound collection.
Drug Discovery Support
One of the fundamental activities in small-molecule drug discovery is the
identification of an appropriate ligand that binds to a therapeutically
relevant biological target. From this starting point, the characteristics
needed to make a compound 'drug like' are built in, and eventually a
clinical development candidate emerges. This lengthy and difficult
discovery process is divided into three distinct phases: target
validation, lead discovery and lead optimization. Each of these phases
places unique demands on the organization. Target validation processes can
involve the use of a compound tool to help elucidate a pathway or target
mechanism; some thematic or target class focused compound sets, or 'mini
decks', can greatly aid the discovery of such tools. Lead discovery
efforts require delivery of a large number of compounds over a short time
frame, while lead optimization primarily requires the delivery of small
numbers of compounds on an ongoing basis to meet program testing
timelines. It is the job of compound management to coordinate the diverse
sample needs of these three phases.
Strategies that support the lead
discovery process depend heavily on compound management's ability to
deliver compound sets of varying sizes in support of HTS, thematic
screening and virtual screening of compound collections. High-throughput
screening demands the preparation and delivery of large decks of hundreds
of thousands to millions of compounds. Typically, HTS decks are
pre-constructed in either 96-, 384- or 1536-well microtiter plates, with
the majority of users employing 384-well technology. Construction of these
large HTS compound decks requires systems that will allow rapid
"cherry picking" of individual solubilized compounds from an
archive storage environment, building master plates that can support
compound needs for many screens through the process of plate replication.
High-throughput plate replication systems have evolved to rapidly copy an
entire compound screening deck of hundreds of thousands to millions of
compounds in a period of days.
Focused screening and virtual
screening approaches rely on constructing and delivering a custom set of
compounds based on knowledge of the target or chemotypes known to bind to
the target of interest. Delivering custom compound sets to support these
more discerning or knowledge-based screening approaches again requires
high-speed sample cherry picking capabilities. Once the custom compound
set is chosen from the archive, downstream sample processing may be
required to transfer a portion of the sample to a microtiter plate or
another type of usable format. This process typically demands parallel
liquid handling capability to facilitate rapid sample processing of
multiple samples from the source to destination formats.
In any screening process, once the
initial screening actives (hits) are identified, retrieval of samples for
retest and concentration response curve (CRC) determination is needed.
Compound resupply for retest and CRC needs either can be achieved by
cherry picking samples from the original solubilized, archived sample or
by taking samples directly from a copy of the screening plate. Rapid
liquid handling is again required to process the samples into appropriate
screening formats.
Although it requires fewer numbers of
compounds, lead optimization projects tend to be more demanding on liquid
handling and information processing because of the complex array of
biological assay and critical path project processes that require samples.
These processes include primary and secondary screening to support
structure–activity relationship studies and in vivo efficacy,
metabolism and toxicology studies. To support these types of activities,
compound management functions can be expected to provide everything from
weighing dry samples to preparation of serial dilution plates and
replication into assay-ready plate formats. The capabilities needed to
support these processes include basic analytical balances coupled to
inventory management systems and liquid handling automation for supporting
serial dilution and replication needs.
Figure 1. The TekCel plate store and
plate server.
|
To provide the required level of support
for drug discovery, compound management functions have become fully
integrated into the process of drug discovery and are no longer
stand-alone isolated functions. Integration includes both the physical
sample flow as well as the flow of data associated with the sample. These
data can include sample concentration, compound molecular weight, compound
identification number, plate barcode information and compound well
location. Integration must occur from the point when a chemist enters
sample registration data and physically prepares the sample for submission
to the time biological data is captured in central data repositories. To
support timely and accurate decision making, all the information about a
sample — including the sample's history, available inventory and
accessible formats — needs to be available to the scientific staff. In
some companies, this compound data set is combined with all of the test
data to provide a comprehensive view of a compounds history. In addition,
to reduce experimental cycle times, sample data needs to be immediately
available and transferred seamlessly to various databases throughout the
discovery value chain. Processes must exist that link each sample from the
chemist's hands to every possible use of that sample within the drug
discovery environment, ensuring the achievement of maximal value from that
sample. This requires that the compound management staff understand
upstream and downstream processes and work closely with chemists and
biologists to integrate these processes across all discovery operational
seams. This effective, seamless integration is what ultimately drives
improved productivity and efficiency to all discovery efforts that rely on
samples for testing.
Figure 2. The Automation
Partnership's Haystack store.
|
Sample Archive
The sample archive is the center piece of the compound management
function. The basic philosophy for this is to store compounds in formats
that maximize sample access while supporting sample integrity over long
periods of time. The scale of archive systems range from encapsulated
refrigerators that can support academic and small biotech needs (Figure 1)
to large enterprise-wide facilities requiring significant infrastructure
to support global R&D operations (Figure 2). Most modern archive
systems now are designed on a modular format that allows scalable
capabilities. While the smaller systems typically support archiving
samples in only one format, the enterprise-wide systems can archive in
multiple formats with multiple process threading.
Archive systems are available to
support individual sample storage in any combination of dry powder
formats, "wet" solution formats and microtiter plate (96-, 384,
1536-well) formats. Dry powders usually are stored in pre-tared glass
vials that are barcoded to support sample tracking. Wet solutions can be
stored in tubes (1.4 mL to 0.3 mL) to allow removal of liquid solutions
using various liquid handling devices. These tubes typically are composed
of various grades of polypropylene. Storage of compounds in tubes composed
of polymeric materials has raised concerns regarding maintenance of sample
integrity because of air permeability and leaching of polymeric materials.
However, they still remain the preferred storage medium. (11, 12)
Figure 3. The REMP multi-use
microtube 96-well format with capping system.
|
There are two general approaches for
storing compound solutions in tubes: single-use or multi-use (Figure 3).
Multi-use tubes (0.3–1.4 mL range) allow multisipping from a single
stock solution of compound and are returned to the archive after an
aliquot is removed. Multi-use tubes are sealed by either a removable cap
or pierceable septum. Single-use tubes (75–300 nL range) store small
volumes and are disposed after the sample is accessed. Sample tracking of
wet solution samples in tubes is done in two ways: through informatic
location designation, which relies on a grid positioning approach, or
through the use of two-dimensional (2-D) barcoded tubes. The ability to
array tubes in 384-well formats has lead to the development of
high-density storage systems.
Microtiter plates are another mode
for storing compound solutions. These types of archives rely on barcoded
plates and associated plate maps to identify compound locations. A
plate-based storage strategy does afford the ability to store solutions at
high density but limits the ability to rapidly cherry pick random compound
sets from the larger collection. In addition, accessing a single sample
that was stored in this manner necessitates exposing all other compounds
contained in that plate to conditions such as freeze-thaw cycles and air
exposure, which could unnecessarily compromise their integrity.
To maximize sample stability, the
storage environment usually is temperature- and humidity-controlled.
Temperature control options can range from 25 °C to –80 °C.
Low-temperature storage is intended to reduce thermal routes of compound
decomposition (13). Humidity control minimizes water uptake by hygroscopic
compounds and reduces water condensation into "wet" compound
solutions usually stored as DMSO solutions. Water content in DMSO
solutions is believed to contribute to sample decomposition, by increasing
the acidity of the solution, as well as to precipitation of the organic
compound (12–17). DMSO concentrations range from 1 mM to 20 mM stock
solutions; however, storage at the higher concentrations usually risks
compound precipitation. The single-use tube for storing solutions is
intended to improve solution sample integrity by eliminating freeze-thaw
events and minimizing water exposure (12, 16, 18).
To support sample access, an archive
system needs to retrieve samples from the archive and return them for
subsequent storage. In the modern compound management environment, sample
retrieval and return (i.e., pick-and-place) operations are accomplished by
informatically driven automated systems. It is the pick-and-place capacity
of these systems that determines overall sample access throughput and,
therefore, the efficiency of the system. State-of-the-art compound
management pick-and-place robots can support up to a rate of 50,000 picks
and places per day. These robots span the scope of technologies that
include finger picking, vacuum-mediated picking, push-pin based systems or
systems that combine these technologies.
The compound archive supports the
majority of downstream discovery efforts, from the single compound
resupply for in vivo testing to the construction of large compound
sets in microtiter plates for screening. Therefore, internal compound
management processes that address sample processing and reformatting must
exist to ensure that archive samples are available in the appropriate
formats for the scientific staff. These internal processes are all part of
the archive function and require automated systems to support sample
transfer among archived formats and careful inventory tracking and
scheduling of internal sample rebuilding efforts. Examples of internal
processes include preparation of DMSO stock solutions from dry powder
samples and preparation of HTS master plates for screening deck creation
from DMSO stock solutions. These internal processes can be unique to a
particular organization's compound management function and are dependent
on the type of downstream sample presentation formats dictated by a
discovery organization.
Sample Processing Whether
it is for archiving needs or to support the many discovery compound
evaluation processes, sample processing is a major component of compound
management effort; it includes weighing of solid samples, preparation of
arrays of solution samples in microtiter plates and high-throughput
microtiter plate replication of large screening decks.
Weighing of solid samples is a
manually intensive process. Although automated systems exist for weighing
solid samples, these systems have severe limitations on the nature of the
samples they are able to handle. Most samples prepared by chemists in a
drug discovery environment are not easily handled, free flowing solids.
Many are oils and gums that do not lend themselves to the currently
available solid sample transfer automation. In addition, the amount of
compound usually prepared by discovery chemists is not sufficient to
utilize solid sample transfer technology. To circumvent handling difficult
to manipulate samples and small sample quantities, sample dissolution and
automated liquid handling methods have become key.
Figure 4. TekCel Hummingbird
technology.
|
Systems that support preparation of
small compound arrays in plates utilize medium-throughput liquid handling
systems. These systems typically are syringe-based multichannel pipetters
containing one to eight probes in either a fixed-tip or disposable-tip
configuration. They support solution delivery from vials or tubes to 96-
and 384-well plate formats. The typical volume range is 1 mL to 10 μL
of solution. These systems can deliver solutions in DMSO solvent or other
volatile solvents but are limited by their liquid dispensing speed and
integration readiness. It usually takes an eight-probe system at least 45
minutes to replicate a 384-well plate of compounds in solution.
Figure 5. The Caliper Sciclone
replicator
|
High-throughput plate replication is
critically required to support high-throughput screening obligations.
Without such systems, the timelines to support compound orders for
high-throughput screening would be unacceptably long. Generally,
high-throughput replication systems rely on using 96- or 384- tip fixed
volume dispensing heads (Figure 4). These systems typically have
integrated plate sealing, bar coding, centrifugation and plate stacking
capabilities (Figure 5). They can use high-speed robotic arms to traffic
plates among the various integrated workstations. The usual working volume
range for these systems is 5 mL to 500 nL, though new capabilities based
on acoustic dispensing technology has the potential to support sample
plating in the 5–50 nL range with high accuracy and precision (19, 20).
Acoustic dispensing technology provides a potentially viable approach to
supporting low voume sample needs for high-density 1536-well plate
screening and significantly reduces the waste solvent usage resulting from
intermediate tip washing steps or the cost of using disposable tips. The
technology currently is being evaluated by several companies for its
potential to meet stringent sample dispensing demands.
Figure 6. A compound management data
integration scheme.
|
The Informatics Environment
Three facets of the informatics environment are critical for supporting a
compound management function: sample registration, inventory management
and sample ordering. The data that resides in these information systems
are necessary for supporting the drug discovery process. Therefore, it is
essential that these systems provide enterprise-wide access and seamless
integration with all databases and tools that utilize this information
(Figure 6).
Sample registration. The
sample registration system is the repository of non-biological information
on a compound. Input of critical compound information into the
registration system first begins with the chemist. This information
includes compound structure, proprietary registration number, molecular
weight, chemist preparer, notebook origin information, amount of sample
prepared and other information on compound physical properties. Subsequent
translation of key pieces of this information into compound management
informatic inventory systems is necessary to support sample processing
activities, such as preparation of defined solution concentrations of
compounds for microtiter-based plate screening. Plate-based screening also
requires creation of plate map information such as compound-well location,
sample molecular weight and sample concentration for supporting screening
data analysis. The ideal system directly captures the relevant information
from one application and populates the necessary database fields in
another.
Inventory management. The
inventory management system is a dynamic informatics environment where the
quantity of every sample in the archive is tracked and updated as samples
are integrated and portions of samples are dispensed from the archive.
This system also must track the various sample formats stored within the
archive because it is possible that a sample could be found in both a
solid and a solution form, as both forms may be needed to support various
types of sample requests. To facilitate inventory updates, the inventory
tracking system can be linked directly with sample dispensing automation.
Sample ordering. Informatics
systems that support sample ordering can be complex, depending on the
level of detailed access and number and breadth of order types offered to
an organization's scientific community. Sample ordering has migrated from
a simple paper, phone or e-mail request to detailed, web-based ordering
systems. These web-based systems have the potential to provide the
requester with critical information that can enhance their sample
selection process. By linking to the inventory database, modern ordering
systems can allow the requester to view real-time inventory information,
such as quantity and format availability. It also allows them to easily
designate which format they wish their request delivered in. These systems
easily allow ordering large numbers of compounds by utilizing list
generation methods; reducing the time and effort needed to access samples
from the sample archive.
Quality Control and Performance
Measurement The many sample
dependent processes within drug discovery require that quality control be
instituted in many areas of compound management. Accurate delivery of
sample amounts, whether they be as a dry solid or a solution, is critical
to ensuring that the results of biological assays are accurately
interpreted. In addition, assessing the integrity of the samples within
the compound collection also has been recognized as an area for quality
control focus. As in many production type environments, quality control
begins with effective training and system maintenance. Establishing
standard operating procedures (SOPs) for all processes ensures consistent
operational practices. Evaluating instrument performance on a regular
basis reduces the possibility of instrument-related sample handling
errors. Thus, calibration of instruments that determine sample weights and
assessment of liquid handling instrument accuracy and precision is
essential for ensuring delivery of consistent and true sample amounts.
Ensuring the quality and integrity of
the compounds within a corporate compound collection has been found to
improve data quality from biological testing and reduce cycle times for
downstream lead evaluation (12). Compounds that exhibit low levels of
purity contribute to both false positive and false negative biological
data; this errant data can result in a potential valuable active compound
being missed or pursuing a compound that is not the active constituent in
the sample tested. The latter event can result in costly efforts to
identify what the active constituent is within the impure sample.
Sample integrity can be impacted by
the storage conditions, the inherent chemical stability of the compounds
or the sample handling practices. As mentioned above,
environmentally-controlled storage systems have been developed to minimize
the impact of chemical instability. Standard, validated sample handling
best practices are used to reduce the chances of sample degradation by
reducing exposure to unfavorable environmental conditions. For example, it
is important to minimize the number of freeze-thaw events for a particular
compound. To understand the sample integrity issue more completely, many
pharmaceutical organizations have undertaken a systematic analysis of
their compound collections. Typically, high-throughput liquid
chromatography–mass spectrometry (LC–MS) methods are used to assess
the purity and identity of a sample. Based on the results of these
analyses and the accepted criteria established for sample integrity,
decisions need to be made on whether to retain or dispose a compound. The
cost of the global analysis of a corporate compound collection can be
significant when one considers analyzing hundreds of thousand to millions
of samples stored in different formats. However, several pharmaceutical
companies have decided to make this investment to improve downstream
processes.
The production character of many
aspects of the compound management function — and its very demanding
scientific customer base — requires continual attention to operational
effectiveness. To accomplish this, institution of a performance
measurement culture becomes invaluable. Developing an overall
understanding of potential gaps in a critical process or infrastructure
capacity can be extremely valuable in forecasting trends that will inform
how to redirect resources to meet the shifting demand. Measures that have
been used as indicators of performance include processing time for
particular order types, the number of samples processed per FTE over time
and the cost of processing various sample requests. Measures used as
indicators of demand include the number of orders placed per order type,
the number of compounds ordered per order type, the ratio of samples
requested to support various discovery activities and the number of
samples submitted by chemists.
As the discovery scientific staffs
are expected to increasingly rely on compound management functions to
support their sample needs, it is important that compound management
functions provide reliable service such that delivery of a compound is not
the rate-limiting step in the execution of an experiment. The
time-critical nature of the experimental process requires sample delivery
within a specified time period, and the business need demands reduced
cycle times throughout all areas of drug discovery and development. In
some organizations, to ensure that organizational needs are met, strict
service-level agreements (SLAs) have been established and monitored. These
SLAs guarantee that for a particular type of request, a compound or set of
compounds will be delivered to the requester within a pre-established
period of time. This guarantee ensures that experiments can be scheduled
to time with compound receipt. Detailed understanding of every aspect of
sample processing and continuous monitoring is fundamental to functioning
under a SLA environment.
Conclusion
Drug discovery compound management organizations have evolved into
sophisticated technology environments capable of supporting a vast array
of drug discovery sample needs. Compound management functions not only
archive the physical compound assets of a discovery organization but also
provide samples in many formats to all aspects of the discovery process
from target validation through lead discovery to lead optimization.
Integrated automation and informatics systems are essential for managing
and distributing the millions of compounds in any given collection with
the quality of process and sample and data integrity that is necessary to
ensure timely and reliable experimental results.
Acknowledgements
We would like to thank the following vendors for providing images that are
included in this manuscript: The Automation Partnership, Velocity11, REMP,
RTS Life Science, Matrix Technologies Corp., Caliper Life Sciences, TekCel
Inc., Tecan and Labcyte Inc., have been very supportive in the process of
preparing this review.
References
1. S. Thomke and W. Kuemmerle, Strat. Mgnt. J. 23, 619–635 (2002)
2. G.P. Jourdan, D.H. Krause and C.T.
Mosburg, Lab. Robotics and Auto. 9, 119–128 (1997).
3. W. Harrison, J. Biomol. Screening
2, 203–206 (1997).
4. M. Beggs, H. Blok and A. Diels. J.
Biomol. Screening 4, 143–149 (1999).
5. P.A. Gosnell, A.D. Hilton, L.M.
Anderson et al., J. Biomol. Screening 2, 99–102 (1997).
6. R. Boutin and W. Zinsser, Cur.
Drug Disc. 29–31 (2001).
7. K. Holden, Cur. Drug Disc. 9–10
(2003).
8. I. Yates, Drug Disc. World 4,
35–42 (2003).
9. M. McCoy, Chem. Eng. News 81,
12–13 (2003).
10. G. Milne, Jr., in Annual Reports
in Medicinal Chemistry Volume 38, A.M. Doherty, Ed. (Elsevier Academic
Press, New York, 2004), pp. 383–396.
11. X. Dheng, J. Hochlowski, H. Tang
et al., J. Biomol. Screening 8, 292–304 (2003).
12. K. Morand and X. Cheng, in
Analysis and Purification Methods in Combinatorial Chemistry, B. Yan, Ed.
(Wiley-Interscience, Hoboken, New Jersey, USA, 2004), pp. 323-350.
13. B. Kazikowski, T.M. Burt, D.A.
Tirey et al., J. Biomol. Screening 8, 205–209 (2003).
14. P. Talaga, Drug Disc Today 9,
51–53 (2004).
15. K.V. Balakin, Cur. Drug Disc.
27–30 (2003).
16. C. Lipinski, in Analysis and
Purification Methods in Combinatorial Chemistry, B. Yan, Ed. (Wiley-Interscience,
Hoboken, New Jersey, USA, 2004), pp. 407–434.
17. J. Hochlowski, X. Cheng, D. Sauer
and S. Djuric, J. Comb. Chem. 5, 345–349 (2003).
18. B.A. Kozikowski, T.M. Burt, D.A.
Tirey et al., J. Biomol. Screening 8, 210–215 (2003).
19. R. Ellson, M. Mutz, B. Browning
et al., J. Assoc. Lab. Auto. 8, 29–34 (2003).
20. T. Spicer, Impact of Acoustic
Non-contact Transfer Upon Compound Management and Ultra High Throughput
Screening, Presented at Laboratory Robotics Interest Group Mid Atlantic
Chapter, January 13, 2005.
Michael J. Sofia* is group
director of new leads chemistry in applied biotechnology, Jay M.
Stevenson is group leader of new leads chemistry in applied
biotechnology and John Houston is vice president of applied
biotechnology and discovery biology at Bristol-Myers Squibb. Michael J.
Sofia can be reached at Bristol-Myers Squibb, New Leads Chemistry, Applied
Biotechnology, 5 Research Parkway, Wallingford, Connecticut 06492 USA.
E-mail michael.sofia@bms.com
.
* To whom all correspondence should
be addressed.
|
