Ian Doxsee

ENGW3307
3/17/2015
ACS Style
2749 Words

Bioisosterism in Drug Design: Paths Forward Toward the Next Generation of Drugs
Abstract:
Bioisosteric replacements of chemical moieties have long been a hot research area, particularly
in the field of medicinal chemistry. Recent developments in the area have allowed early-stage hits to
be efficiently translated into market-ready drugs. The discovery of new bioisosteres can be performed
via a high-throughput shotgun approach of testing huge compound libraries, either in biochemical
assays or in silico with the aid of computational modeling software designed to probe ligand-receptor
interactions. However, bioisostere discovery can also be performed via more methodical knowledgebased approaches, including divergent synthesis and computationally-predicted functional group
modification.
Introduction:
Medicinal chemistry is the design of new drug candidates through a process of logical
modification of drug properties. This process is called analoging and involves the synthesis of many
different slight variants of an existing compound by medicinal chemists, in order to probe the ideal
space-filling of the compound. Once each portion of the molecule has been fully explored, the best
functional group for that area is used in a final drug candidate that is taken on for further testing,
including analysis of drug absorbance, distribution, metabolism, excretion (colloquially known as
ADME). Sometimes a drug that had the best bioactivity based on space-filling has undesirable
properties (toxicity, low bioavailability) when examined at this stage. What, then, do scientists do? Do
they scrap the existing compound and move on to a worse compound? No, instead they will attempt to
find compounds with similar bioactivity and space-filling but different
N N
composition. These compounds are known as bioisosteres, bio
O
N
referring to biological activity, iso meaning same, and ster referring
N
OH
H
to sterics or the actual size of the molecule. By switching one group for
another of similar size and electronics, medicinal chemists hope to get
rid of these adverse effects by changing how the molecule acts in the Illustration 1: A Classical
body. Usually this bioisosterism is to avoid toxicity effects as a result of Example of a Bioisostere: The
how the body metabolizes the compound. For example, an amine
Tetrazole as Carboxylic Acid
(NH2) might be substituted for an alcohol (OH) because the amine is Bioisostere
metabolized into a toxic amine-oxide (N-O) which is a free-radical
promoter in the body a process implicated in cancer and other aging-related diseases.
Bioisosterism is very interesting on paper, but how do scientists find bioisosteres for
problematic functional groups? In fact, there are many different methods of finding replacements,
ranging from shotgun approaches (throw compounds at an enzyme and see what is effective) to
extremely planned approaches derived from complex knowledge of the intricacies of functional group
electronics. The first method we'll examine is the most unplanned approach random screening of a
compound library.
High-Throughput Screening:
High-throughput screening (HTS) involves the analysis of many compounds against an existing
enzyme. Many times in industry this is done as a preliminary step in a new project in order to obtain

hits or compounds with a small amount of activity. This screen is typically performed by running the
compound in assays against an enzyme wet work or actual chemical-in-flask science. However, in
Tuyishime's 2014 paper the initial step was a virtual screen of a huge number of commerciallyavailable compounds. These compounds are typically screened on a computer against a 3-dimensional
structure of the enzyme the compounds bind to, but in this case a 3-dimensional structure was not
known. Thus, a hypothesized binding conformation was calculated using known binders to the enzyme,
and compounds were screened against their virtual binding to this theoretical site. Using this method, a
number of hits were discovered. One of these compounds contained a piperidine motif, a sixmembered ring with two nitrogens that is exceedingly common in pharmaceutical drugs. In order to
achieve distinction from an existing Bristol-Meyers Squibb compound under
patent, it was decided that the group would follow an old Pfizer paper that
H
N
reported a novel bioisostere for piperidine the dipyrrolidine strucuture.
H
N
This structure, containing two fused five-member rings with two nitrogens,
achieved a reasonable similarity score in the group's computational model.
N
This structure has tremendous applications to other drugs, since a
N
dipyrrolidine has different acidity and polarity than a piperidine. Granted,
these are minor differences but in the final stages of drug optimization these Illustration 2: A
minor properties can have tremendous impacts.
Piperidine for
One major complicating factor in HTS is in weeding out false
Dipyrrolidine Bioisosteric
positives, inappropriate compounds, and covalently-binding fragments
Replacement
collectively known as pan-assay interference compounds (PAINS). These
compounds seem to crop up in every HTS and must be carefully examined to determine whether their
activity is real, or is a result of complex phenomena like covalent binding (irreversible binding gives
false positives and is also ill-suited for drug development) and aggregation (some compounds will stick
together and form fibrils that will cause false positives). PAINS is absolutely crucial for researchers to
understand in order to avoid months or years of wasted development. Activity in a cell-based assay
does not correspond to a potential drug candidate. Another concern is when the compounds being
assayed are impure. Incredibly small quantities of metals can poison the results, showing a compound
to have binding where there is none. For this reason the first step after determining hits off a screening
is to repurify every compound and also resynthesize them. Sometimes the resynthesis is the most
important part. Many hits have been discovered to be from degradation products of the compound
researchers thought they were testing, and sometimes even when a real hit is discovered the structure of
the compound is determined to be incorrect. Skipping this important structural verification step can
lead to absolutely disastrous consequences down the line. One infamous instance of a misassigned
structure even led to an approved drug's approval being pulled an enormous loss for the company
who developed it. The FDA does not look kindly on these sorts of mistakes and it is therefore in
everyone's best interests to thoroughly vet every hit that comes from an HTS.
Divergent Synthesis:
An extremely common method of discovering new drug candidates based on an existing
structure is through divergent synthesis. In this method, a synthetic scheme is designed such that the
last step (or very nearly last step) can tolerate many different functional groups being added in order to
create a large amount of diversity at a single position in the molecule. This allows for all the structureactivity relationships (SAR) at that one position to be determined in a time-sensitive manner, perfect
for the fast pace of pharmaceutical industry. In Junker's 2014 paper Synthesis, binding affinity, and
structure-activity relationships of novel, selective and dual targeting CCR2 and CCR5 receptor
antagonists, an existing compound that non-selectively binds to both CCR2 and CCR5 receptors was
modified in a rational way in order to probe the overall molecular SAR. The middle-stage hit
compound Junker's group based their studies on was a bicyclic compound with an aryl ring tethered to

one end and a complex substituted amide at the other

R
end. These dual functional groups represented the vast
OR
H
n
R
majority of SAR work on this compound, through
R
N
synthesis of aryl derivatives and new substituted amides.
n
O
The synthesis of these compounds was undertaken in the
Illustration
4:
Two
Examples
of
Divergent
exact same way up until a bicyclic aryl bromo and
Synthesis Scaffolds
protected-ether compound. At this point, the order of
synthesis of the aryl and amide compounds was split. To make various aryl compounds, simple Suzuki
reactions were run to couple the aryl
bromide to aryl boronic acids, before the
O
standard amide group was put on. In order to
O
R
R vary
make different amide replacements, the
aryl
O
Aryl
O
Br
amide portion was substituted first, followed
vary
by addition of the standard terminal aryl
amine
group. The fact that this synthesis only
diverges between different compounds over
H
H
N
the last two steps means that many grams of
N
R
R
precursor can be taken through the initial
O
O
Aryl
Br
steps to form the top left compound in
Illustration 3: Junker's Divergent Synthesis of CCR
Illustration 3, before individual reactions
have to be run on each separate compound. Receptor Analogs
This is remarkably more efficient than taking each different compound through multiple reactions in
order to reach the final product.
Junker's group discovered some important things about the ability of various functional groups
to act as bioisosteres in this class of compounds. A terminal biphenyl ring was well-tolerated but
suffered from various pharmacokinetic liabilities. Junker chose to substitute a single phenyl ring at the
para (far) position. He quickly discovered that a lipophilic isopropoxy group had very similar steric size
to a biphenyl ring but suffered from none of the metabolic flaws of the biphenyl. Other phenyl
substitutions were not nearly as successful at maintaining potency. This isopropoxy switch should be
used by every medicinal chemistry department who has a biphenyl compound. Biphenyls show up in a
surprising number of drugs but typically suffer from their lack of decoration and metabolic issues.
Isopropoxyphenyl compounds are smaller, more activated, and easier to make than biphenyls and also
are deactivated against traditional CP450 liver metabolism.
Another example of a traditional medicinal chemistry approach to drug discovery is in
Capparelli's 2014 paper SAR Studies on Tetrahydroisoquinoline Derivatives: The Role of Flexibility
and Bioisosterism To Raise Potency and Selectivity toward P-glycoprotein. In this paper, Capparelli's
group focuses less on modifying a single functional group and instead focuses on modifying the
internal structure of the molecule in such a way that optimal spacing is achieved between each end of
the molecule. Performing chemical modifications on the center of a molecule is less reaction-efficient
than changing the ends of the molecule for the simple reason that every variation must be taken through
multiple reaction steps to achieve the final product, while terminal modifications can be divergently
synthesized, with the variation built-in on the ultimate or
penultimate reaction. Capparelli's group focused on the A
O
Linker N
A
ring of their hit molecule, a decently potent and poorly
selective molecule toward p-Glycoprotein (pGp), as well as
O
the linker between the A ring and the bicyclic heterocycle
on the right side of the molecule. Switching the A-phenyl R
ring for a five-membered furan (an oxygen-containing ring) Illustration 5: Capparelli's SAR
led to compounds with decreased activity, while a change Exploration of p-Gp Ligands

to a thiophene ring (a five-membered aromatic ring containing sulfur) led to compounds with increased
potency and good selectivity toward pGp. This thiophene for phenyl replacement is a relatively
standard bioisosterism despite the change from a six-membered ring to a five-membered ring, the size
of the sulfur atom means that bond angles are quite similar, and the greasiness of sulfur leads to very
similar electronic properties.
Capparelli's group also changed the alkyl linker from the A ring to the bicyclic heterocycle in an
attempt to probe the proper length and electronic properties of this linker. Their first compound was a
simple elongated alkyl chain containing an additional carbon atom. This compound showed improved
activity and selectivity toward pGp, spurring the scientists on to future modifications. The next
modification made was to change from a simple alkyl to an -O- linked alkyl chain. This compound led
to a strong increase in activity for all compounds but a surprising unchanged selectivity. This result
indicates that the linker portion has little effect on binding in terms of how it fills space but has a large
effect in terms of how it affects the electronics of the phenyl ring in the molecule. More compounds
based on this O-alkyl substitution with longer alkyl chains continued to improve potency, indicating
that a larger degree of conformational flexibility is important in allowing the terminal groups to interact
in the proper orientations.
De Novo Thinking:
A completely different way of discovering what bioisosteric functional groups can work for a
given molecule is to think on an academic level about the electronic properties of the functional group
you wish to replace and using de novo thinking develop hypotheses for why a wildly different
functional group could work as well. Piemontese's group worked on this problem in his 2015 paper
Design, synthesis and biological evaluation of a class of bioisosteric oximes of the novel dual
peroxisome proliferator-activated receptor a/y ligand LT175. His group focused on the concept of
phenyl rings as privileged structures in drug candidates. These simple aromatic rings seem to exist in
every drug, but are fairly plain in terms of electronic properties (no H-bond acceptors/donors) and serve
primarily as a flat space-filling electron-rich ring. Additionally, their metabolism can lead to
carcinogenic by-products in cases where the phenyl ring is activated by electron-rich functional groups.
For this reason, bioisosteric replacements for a phenyl ring are highly desired by medicinal chemists
and academic chemists alike.
Piemontese's group explored the possibility of using an oxime to mimic the sp2
N
character of a phenyl ring while also approximating the size, but changing the
O
R
electronic properties and also providing many novel ways to functionalize this portion
Illustration 6:
of the molecule. There are many known ways of functionalizing a phenyl ring, but
Phenyl for
most of these methods rely on the presence of labile groups that can be easily
replaced. Modifying an unsubstituted phenyl ring is considerably more difficult and Oxime
Bioisostere
reactions that can effect this change are often not tolerated by other sensitive
functional groups within a molecule, making these processes completely unsuitable for pharmaceutical
drug development. However, oximes can easily be installed from simple starting materials indeed, a
ketone and an amine are all that are required and these are exceedingly common functional groups. The
reaction chemistry is also well-established and is well-tolerated by most functional groups. An
additional benefit to using oximes instead of phenyl rings is that there is inherent stereochemistry in
these molecules substituted oximes have two conformations, cis and trans, and most reactions will
produce some of both. As a final reaction for scale-up in drug production, we obviously don't want to
produce multiple products but when we're simply screening as many compounds as we can get our
hands on, sometimes these un-clean reactions are preferable.
Conclusions:
As we've seen, there are a multitude of ways to find replacements for functional groups in a

molecule. Certainly the most interesting from a purely academic standpoint are those from research
groups like Piemontese's, given that the entire project was de novo determined from an intellectual
argument. Synthetic chemists gravitate toward these papers much more than to simple HTS
communications, since the knowledge base leveraged by these groups is important to many projects
and the types of thinking that Piemontese and others use have tremendous applications to all chemists.
Approaching a problem by thinking about the steric and electronic properties of the functional group is
important in every situation, since it will inevitably lead to novel ideas about what replacements can be
made. Sticking to the handbook of bioisosterism (C for O, N, etc) is boring and ignores important
steric/electronic effects that are specific to the molecules being designed. In order to probe the
boundaries of bioisosterism and find functional group replacements that are genuinely useful to your
specific project it's absolutely necessary to focus on the specific features of the molecule at hand and
exploit these properties in order to improve the potency, selectivity, metabolism, and other properties of
the drug candidate.

References:
Capparelli, E.; Zinzi, L.; Cantore, M.; Contino, M.; Perrone, M.; Luurtsema, G.; Berardi, F.; Perrone,
R.; Colabufo, N. SAR Studies on Tetrahydroisoquinoline Derivatives: The Role of Flexibility and
Bioisosterism to Raise Potency and Selectivity toward P-glycoprotein. J. Med. Chem. 2014. 57, 99839994.
Junker, A.; Kokornacyzk, A.; Zweemer, A.; Frehland, B.; Schepmann, D.; Yamaguchi, J.; Itami, K.;
Faust, A.; Hermann, S.; Wagner, S.; Schafers, M.; Koch, M.; Weiss, C.; Heitman, L.; Kopka, K.;
Wunsch, B. Synthesis, binding affinity and structure-activity relationships of novel, selective and dual
targeting CCR2 and CCR5 receptor antagonists. Org. & Biomolecular Chem. 2015. 13, 2407-2422.
Lu, H.; Yang, T.; Xu, Z.; Wren, P.; Zhang, Y.; Cai, X.; Patel, M.; Dong, K.; Zhang, Q.; Zhang, W.;
Guan, X.; Xiang, J.; Elliott, J.; Lin, X.; Ren, F. 2-Aminopyrimidin-4(1H)-one as the novel bioisostere
of urea: Discovery and novel and potent CXCR2 antagonists. Bioorganic & Med.. Chem. Letters. 2014.
5493-5496.
Piemontese, L.; Fracchiolla, G.; Carrieri, A.; Parente, M.; Laghezza, A.; Carbonara, G.; Sblano, S.;
Tauro, M.; Gilardi, F.; Tortorella, P.; Lavecchia, A.; Crestani, M.; Desvergne, B.; Loiodice, F. Design,
synthesis and biological evaluation of a class of bioisosteric oximes of the novel dual peroxisome
proliferator-activated receptor a/y ligand LT175. Euro. Journal of Med. Chem. 2015. 583-594.
Tuyishime, M.; Danish, M.; Princiotto, A.; Mankowski, M.; Lawrence, R.; Lombart, H.; Esikov, K.;
Berniac, J.; Liang, K.; Jingjing, J.; Ptak, R.; Madani, N.; Cocklin, S. Discovery and optimization of
novel small-molecule HIV-1 entry inhibitors using field-based virtual screening and bioisosteric
replacement. Bioorganic & Med. Chem. Letters. 2014. 5439-5445.