Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Two approaches have emerged for creating libraries of compounds for use in biological screening assays for drug discovery — fragment-based ligand design and diversity-oriented synthesis. Advocates of each approach discuss their favoured strategy.
It's interesting to read both sides of the discussion of what I refer to as the "kitchen sink" approach, where one can throw every thing at the target, including the kitchen sink with varying results. I agree, to a degree, with both sides here and believe that each approach has its own strengths and limitations. The problem with both methodologies lies in the fact that at the end of the screen, the lead compounds may be rather poor drugs.
The overarching weakness of both methodologies is that neither guarantees lead scaffolds will have desirable ADME properties, in particular solubility, stability and permeability. In addition, it is possible that both DOS and FBS can lead to weak effectors that are susceptible to the aforementioned shortcomings as well as not being amenable to modification and further optimization.
To address this weakness, screening libraries should be designed to contain compounds that have desirable molecular attributes, e.g. drug-likeness and primed for modification, even if that, initially, means covering less chemical space. Creation of such a compound collection will necessitate the collaboration of medicinal chemists and pharmacologists and will undoubtedly have incredible utility for future high throughput screens.
dougal kensington
Great article. Nice to see some debate. CP Snow talked about the breakdown of communication between the two cultures of humanities and science. There is a similar situation between the cultures of academic drug discovery and the pharmaceutical industry – the principle reason being that pharma has to confront ugly reality and make money from science and academia doesn't.
Two important ideas have emerged in the last few years in drug discovery:
1. There is a 'physiciochemical space sweetspot', where the probability of failure of an oral clinical candidate is at a minimum.
2. The super-exponential growth of "chemical space" with increasing molecule size means that it is much more efficient to sample fragment-like molecular space rather than drug-like molecular space.
This means that diversity-oriented synthesis, while often a beautiful outlet for clever synthetic chemistry, will never produce a drug. It may produce molecular tools for chemical biology, and some nice papers, but the molecules it finds will all be dead-end ones.
Fragments on the other hand, provide a clever way of navigating through ever-expanding chemical space to this physicochemical sweetspot.
If only the world's great synthetic chemists could do DOS methodology in fragment-sized space. Difficult of course, because it's small and polar, but the molecules would actually be useful.
Perhaps Hajduk and Spring should swap labs and cultures for six months.
Andrew Mitchinson
Readers might be interested in seeing Derek Lowe's thoughts about this Forum article, and some ensuing discussion, on his In The Pipeline blog:
Any thoughts about the Forum article will, of course, also be most welcome here on Nature's online comments!
Andy Mitchinson, News & Views editor
John Warren
The recent forum on drug discovery (Nature 470, 42; 2011) describes how the pharmaceutical industry has screened millions of compounds for activity as potential drugs. This followed the National Cancer Institute's lead from 1957 to test naturally occurring compounds for anti-cancer activity. The diversity-orientated synthesis (DOS) of huge libraries over the last 15 years required high output screening and has met with limited success. Over this period drug discovery has waned in productivity. In an attempt to reduce the workload of DOS, fragment-based screening (FPS) has been used to build potential therapeutic molecules from lead chemical fragments. But both DOS and FPS rely on blind screening. Perhaps it is time to revisit hypothesis driven chemistry, given its proven track record. For example, hypothesis driven chemistry led to the synthesis of the dopamine analogue, methyl dopa, to interfere with dopaminergic neurotransmission and treat high blood pressure. Similarly George Hutchins believed that analogues of nucleic acids might interfere with cell division. From this concept Gertrude Elion manipulated purine chemistry to achieve the first remission of childhood leukaemia, many anti-cancers, an anti-bacterial, azathioprine for immunosuppression, an anti-malarial, allopurinol for gout and the first anti-viral (J. Warren BJCP 71, in press; 2011). James Black synthesised the first H2 antagonist and the first beta blocker by targeted chemistry, synthesising few compounds by today's standards. Akiro Endo screened only 6,000 compounds for the first statin and subsequent statin blockbusters were minor modifications of the lead molecule. Though random screening has produced some treatments for cancer, it has not been cost effective. Making a specific target molecule is a better approach – some 21 monoclonal antibodies have been approved as therapies since 1994. Past success in drug development suggests that making molecules to test a hypothesis, purine analogues are the prime example, is more productive than random screening of whole molecules or fragments.
John Warren
Both DOS and FPS rely on blind screening and perhaps it is time to revisit hypothesis driven chemistry, given its proven track record. For example, hypothesis driven chemistry led to the synthesis of the dopamine analogue, methyl dopa, to interfere with dopaminergic neurotransmission and treat high blood pressure. Similarly George Hitchins believed that analogues of nucleic acids might interfere with cell division. From this concept Gertrude Elion manipulated purine chemistry to achieve the first remission of childhood leukaemia, many anti-cancers, an anti-bacterial, azathioprine for immunosuppression, an anti-malarial, allopurinol for gout and the first anti-viral (J. Warren BJCP 71, in press; 2011). James Black synthesised the first H2 antagonist and the first beta blocker by targeted chemistry, synthesising few compounds by today's standards. Akiro Endo screened only 6,000 compounds for the first statin and subsequent statin blockbusters were minor modifications of the lead molecule. Though random screening has produced some treatments for cancer, it has not been cost effective. Making a specific target molecule is a better approach – some 21 monoclonal antibodies have been approved as therapies since 1994. Past success in drug development suggests that making molecules to test a hypothesis, purine analogues are the prime example, is more productive than random screening of whole molecules or fragments.
Peter Kenny
I was underwhelmed by this forum discussion although my main criticism is of the way is was set up rather than of what the participants had to say. I wrote a blog post on the forum discussion and will point you towards that rather than trying to summarise it here.
victor kenyon
It's interesting to read both sides of the discussion of what I refer to as the "kitchen sink" approach, where one can throw every thing at the target, including the kitchen sink with varying results. I agree, to a degree, with both sides here and believe that each approach has its own strengths and limitations. The problem with both methodologies lies in the fact that at the end of the screen, the lead compounds may be rather poor drugs.
The overarching weakness of both methodologies is that neither guarantees lead scaffolds will have desirable ADME properties, in particular solubility, stability and permeability. In addition, it is possible that both DOS and FBS can lead to weak effectors that are susceptible to the aforementioned shortcomings as well as not being amenable to modification and further optimization.
To address this weakness, screening libraries should be designed to contain compounds that have desirable molecular attributes, e.g. drug-likeness and primed for modification, even if that, initially, means covering less chemical space. Creation of such a compound collection will necessitate the collaboration of medicinal chemists and pharmacologists and will undoubtedly have incredible utility for future high throughput screens.
dougal kensington
Great article. Nice to see some debate. CP Snow talked about the breakdown of communication between the two cultures of humanities and science. There is a similar situation between the cultures of academic drug discovery and the pharmaceutical industry – the principle reason being that pharma has to confront ugly reality and make money from science and academia doesn't.
Two important ideas have emerged in the last few years in drug discovery:
1. There is a 'physiciochemical space sweetspot', where the probability of failure of an oral clinical candidate is at a minimum.
2. The super-exponential growth of "chemical space" with increasing molecule size means that it is much more efficient to sample fragment-like molecular space rather than drug-like molecular space.
This means that diversity-oriented synthesis, while often a beautiful outlet for clever synthetic chemistry, will never produce a drug. It may produce molecular tools for chemical biology, and some nice papers, but the molecules it finds will all be dead-end ones.
Fragments on the other hand, provide a clever way of navigating through ever-expanding chemical space to this physicochemical sweetspot.
If only the world's great synthetic chemists could do DOS methodology in fragment-sized space. Difficult of course, because it's small and polar, but the molecules would actually be useful.
Perhaps Hajduk and Spring should swap labs and cultures for six months.
Andrew Mitchinson
Readers might be interested in seeing Derek Lowe's thoughts about this Forum article, and some ensuing discussion, on his In The Pipeline blog:
http://pipeline.corante.com...
Any thoughts about the Forum article will, of course, also be most welcome here on Nature's online comments!
Andy Mitchinson, News & Views editor
John Warren
The recent forum on drug discovery (Nature 470, 42; 2011) describes how the pharmaceutical industry has screened millions of compounds for activity as potential drugs. This followed the National Cancer Institute's lead from 1957 to test naturally occurring compounds for anti-cancer activity. The diversity-orientated synthesis (DOS) of huge libraries over the last 15 years required high output screening and has met with limited success. Over this period drug discovery has waned in productivity. In an attempt to reduce the workload of DOS, fragment-based screening (FPS) has been used to build potential therapeutic molecules from lead chemical fragments. But both DOS and FPS rely on blind screening. Perhaps it is time to revisit hypothesis driven chemistry, given its proven track record. For example, hypothesis driven chemistry led to the synthesis of the dopamine analogue, methyl dopa, to interfere with dopaminergic neurotransmission and treat high blood pressure. Similarly George Hutchins believed that analogues of nucleic acids might interfere with cell division. From this concept Gertrude Elion manipulated purine chemistry to achieve the first remission of childhood leukaemia, many anti-cancers, an anti-bacterial, azathioprine for immunosuppression, an anti-malarial, allopurinol for gout and the first anti-viral (J. Warren BJCP 71, in press; 2011). James Black synthesised the first H2 antagonist and the first beta blocker by targeted chemistry, synthesising few compounds by today's standards. Akiro Endo screened only 6,000 compounds for the first statin and subsequent statin blockbusters were minor modifications of the lead molecule. Though random screening has produced some treatments for cancer, it has not been cost effective. Making a specific target molecule is a better approach – some 21 monoclonal antibodies have been approved as therapies since 1994. Past success in drug development suggests that making molecules to test a hypothesis, purine analogues are the prime example, is more productive than random screening of whole molecules or fragments.
John Warren
Both DOS and FPS rely on blind screening and perhaps it is time to revisit hypothesis driven chemistry, given its proven track record. For example, hypothesis driven chemistry led to the synthesis of the dopamine analogue, methyl dopa, to interfere with dopaminergic neurotransmission and treat high blood pressure. Similarly George Hitchins believed that analogues of nucleic acids might interfere with cell division. From this concept Gertrude Elion manipulated purine chemistry to achieve the first remission of childhood leukaemia, many anti-cancers, an anti-bacterial, azathioprine for immunosuppression, an anti-malarial, allopurinol for gout and the first anti-viral (J. Warren BJCP 71, in press; 2011). James Black synthesised the first H2 antagonist and the first beta blocker by targeted chemistry, synthesising few compounds by today's standards. Akiro Endo screened only 6,000 compounds for the first statin and subsequent statin blockbusters were minor modifications of the lead molecule. Though random screening has produced some treatments for cancer, it has not been cost effective. Making a specific target molecule is a better approach – some 21 monoclonal antibodies have been approved as therapies since 1994. Past success in drug development suggests that making molecules to test a hypothesis, purine analogues are the prime example, is more productive than random screening of whole molecules or fragments.
Peter Kenny
I was underwhelmed by this forum discussion although my main criticism is of the way is was set up rather than of what the participants had to say. I wrote a blog post on the forum discussion and will point you towards that rather than trying to summarise it here.
http://fbdd-lit.blogspot.co...