Biomedical Engineering Reference
In-Depth Information
means of affinity-based selection strategies offers enormous technical benefits over
conventional screening approaches. First, the cost, time, robotics, and storage needs
are radically minimized. Second, the simple affinity capture of small molecule on an
immobilized target protein does not require the development of complicated biolog-
ical assays. As a result, artifacts commonly observed in traditional high-throughput-
screening campaigns are circumvented (e.g., compound aggregation, insolubility,
abnormal fluorescence absorption/quenching), and the deploying of target proteins,
such as components involved in protein-protein interactions, particularly challenging
to tackle with conventional screening approaches, is allowed, [56,65]. Third, selec-
tion procedures compel low amounts of target protein (i.e., a few milligrams), and
multiple panning experiments can be performed rapidly in parallel, applying various
experimental conditions (e.g., different protein supports, coating density, number of
washing steps, re-panning) or adopting alternative selection strategies (e.g., pres-
ence of related proteins, competition with ligands previously discovered, addition or
absence of cofactors and substrates).
In principle, other naturally (i.e., RNA) and nonnaturally (i.e., PNA) occurring
nucleic acid residues have been considered for the encoding of chemical libraries.
However, ribonucleic acids are very susceptible to RNAse cleavage, thus making the
handling and storage of RNA-encoded libraries very inconvenient. Conversely, more
stable and chemically versatile macromolecules, such as peptide nucleic acids [122],
cannot be amplified by PCR, restraining their use to libraries of smaller size (up to
about 10
4
compounds), in combination with low-throughput decoding strategies such
as microarray-based technologies [123-126].
DNA-encoded chemical library technology has benefited profoundly from the lat-
est high-throughput-sequencing developments. Cutting-edge deep-sequencing plat-
forms, capable of gathering millions of DNA-sequence information per sequenc-
ing run in just a few days [107,111,127,128], allowed simultaneous evaluation of
thousands to millions of structurally related compounds (including stereoisomers
and enantiomers). This approach straightforwardly provided instant SAR databases
after each selection experiment, invaluable sets of information for the design and
improvement of lead structures by medicinal chemistry optimization and/or further
DNA-encoded affinity maturation cycles (see Section 11.2.2.1) [54,55].
To date, sequencing throughput defines itself the natural limit for the largest
library, which can be consistently interrogated. As shown in Figure 11.22d (Sec-
tion 11.3.2), rigorous statistical analysis is indispensable for establishing an accurate
correlation between sequencing counts and the relative abundance of the individual
library compounds before and after imposing the selection pressure. Therefore, it is
crucial to obtain a high degree of sequencing coverage after decoding with respect
to library size. For the appropriate assessment of a selection experiment using a 1
million compound library, approximately 10 million raw sequencing “reads” are typ-
ically required. Considering that sequencing power is currently settled to about 50
million DNA tags per run (at a cost of about 5000 euros), it is not yet conceivable
to screen DNA-encoded libraries comprising more than a few million member com-
pounds. Although several strategies have been implemented for the construction of
