Combinatorial Chemistry in Drug Design
Combinatorial chemistry is somewhat hard to define. Essentially, it is a collection of techniques which allow for the synthesis of multiple compounds at the same time. These techniques are now largely automated, but do not necessarily have to be. The collection of compounds made in this way is generally known as a library. This branch of chemistry is very young, but in this short time it has had profound effects. This can be seen by its impact on medicinal chemistry and, in particular, the drug design process. Traditionally, potential lead compounds were synthesized one at a time. The biological activity of this compound was assayed, and the results would be reflected in the next round of design. This traditional method was useful, but time consuming and expensive when compared to the emerging combinatorial chemistry. Computational chemistry led to more rational design of compounds to be tested, and high throughput screening led to quick in vitro assays. Synthesis of one compound at a time could no longer keep up, and thus became the rate limiting step in the process. Combinatorial chemistry was the solution to this problem. Instead of doing multiple A x B type reactions, one can cover many combinations An x Bn in one reaction. In combinatorial chemistry, large numbers of compounds are made at the same time in small amounts, forming libraries which can be assayed for desired properties all at once.
A Comparison of Tradition and Combinatorial Strategies in Drug Design
Combinatorial chemistry is a very young science, having only been around for approximately 20 years. It has been applied to drug design for an even shorter period of time. The roots of combinatorial chemistry can be dated back to the 1960’s, however. It was at this time that Bruce Merrifield developed a method for solid phase synthesis of peptides. He won the 1984 Nobel Prize in Chemistry for this work. In the 1980’s, the first combinatorial techniques were developed. H. Mario Geyson attached resins to a surface, an idea which resembled having individual peptide synthesis reactions on the head of a pin. These pins were reacted in individual wells, as seen below, and were a way to positionally map the products being synthesized.
Also contributing to this growing field in the mid 1980’s was Richard Houghton, who introduced a technique involving placing resins in individual, porous bags similar to tea bags. A big devleoped in this time period was by Furka, who took the ideas of Merrifield a step further. He applied the solid phase synthesis of peptides in a combinatorial way, developing the ‘split and mix’ technique which will be discussed in more detail below. Through the 80’s and into the early 1990’s, combinatorial chemistry was focused on peptide synthesis and later oligonucleotide synthesis. This all changed in 1992 with the publication of a paper by Bunin and Ellman. This paper described the application of combinatorial methods to the synthesis of a benzodiazepine library. This was the first example of small molecule synthesis by combinatorial chemistry. The possibilities of using these techniques in drug design were now evident, and this branch of science began to really take off with the adoption of its techniques by the pharmaceutical and medicinal chemistry industries.
Solid Phase Synthesis vs. Solution Phase Synthesis
Combinatorial chemistry employs both solution and solid phase synthesis. Solid phase synthesis has traditionally been the most popular, but new techniques for solution phase synthesis are now emerging. Solid phase and solution phase each offer their own distinct advantages. However, there are downsides to both as well. Some of these are summarized below.
In solid phase synthesis, the starting compound is attached to an insoluble resin bead. There are various types of linkers employed for this. The resin material most commonly used in combinatorial chemistry is polystyrene that is 1-2% divinylbenzene. Reagents are added to the solution, and the resulting products can be isolated by simple filtration, which traps the beads while the excess reagent is washed away. The ease of this isolation is one of the biggest advantages of solid phase synthesis. It makes it particularly useful for multi-step reactions, as the intermediates resulting in each step can be isolated quickly by this method. Another related advantage is that since removal of unreacted reagents is possible, large excesses can be used to drive the reaction to completion. However, optimal reaction conditions for solid phase synthesis can be difficult to determine, and developing these are far more time consuming than the actual reactions will be. The attachment to the resin puts limits on the chemistry that can be performed in this phase. Assessment of the purity of the resin attached intermediates is also difficult. The best options today are NMR or FTIR. Purifying the final product after cleavage from the resin also proves to be a challenge.
Plastic beads for solid phase combinatorial chemistry
An example of solid-phase
The use of solution phase synthesis is on the rise, with multiple techniques available. The biggest challenge with solution phase synthesis is isolation of the product, and ways to automate this. Some progress has been made in automating liquid-liquid extraction. However, several new techniques have been introduced to make the extraction process easier. Ion exhange resins are currently in use. These remove byproducts and therefore eliminate the need for an aqueous work up. Fluorous phase-chemistry is another method which eases extraction of the final product. The starting compound has a perflourinated group attached to it. After the reaction has been completed, the product can be isolated by using a fluorocarbon solvent, which the compound will preferentially by extracted by. Removal of the perfluorinated group gives the desired final product. An example of this is pictured below.
Split and Mix
In theory, if one wanted to make a library of the compounds A-B1, A-B2, A-B3, and A-B4, it could be accomplished by adding the reagents B1, B2, B3, and B4 to a solution of resin-bound A. All four of the desired compounds should be produced. However, there is a problem here. The reaction kinetics will not be the same for each reagent added. Therefore, the resulting mixture will not be equimolar, and will likely contain some dominating products. As well, individual beads will contain a mixture of products, not just a single one. In combinatorial chemistry, equimolar mixtures of products and one product per bead are desired. The Split and Pool technique, also known as the Split and Mix technique, is the solution to this.
In this method, resins are divided equally into wells. The number of pools needed is dictated by the number of reagents the will be used during the reaction. As well, the overall number of beads used must be at least equal to the total number of compounds which will be generated by the process. Each individual well is reacted with a different reagent. The products from these reactions are isolated via filtration, and the pooled together to give an equimolar mixture. This mixture is then equally divided amongst the wells, and reacted again with reagents. The process is repeated as many times as necessary. The end result is equimolar library of compounds representing all possible combinations of reagents. Only a very small amount of each product is produced (often a single bead represents one product), but only a small amount is needed for biological testing. When an active compound is discovered, it would need to be produced on a larger scale for further assays regardless. However, since the majority of compounds formed will not have any biological activity, there is no point than making any more of it than is needed to determine this.
Split and Pool synthesis is a great saver of time and money. Say, for example, there are three different reagents X, Y, and Z which are to be added to a resin-bound organic molecule in a three step process, as pictured below. Using split and pool synthesis, the 27 member library these compounds can represent is generated in three steps, using 3 vessels. To generate these individually, 27 reaction vessels would be required, and each of these would have to undergo 3 steps. However, this reaction is on a very small scale compared to those which would actually be employed in industry. As you increase the number of reagents, these numbers become even more significant. For example, for the three step addition just discussed, let’s have 10 reagents to choose from instead of just 3. Using the split and pool method, only 10 reaction vessels would be required, and 3 reactions would occur in each, with pooling and division of the resins between each. Thus the 1000 member library made up of all the possible products could be synthesized with only 30 reactions! Synthesizing each of these compounds separately would require 3000 reactions. This is the power of split and pool synthesis. It is especially useful for building large libraries, like those containing thousands of compounds which are required at the initial stages of drug design while searching for lead compounds. It would take an excruciatingly long time to synthesize mass amounts of compounds individually, and would also require larger amounts of reagents. Thus the split and pool method is both time and cost effective.
Bioassays must be done on the resulting products from split and pool synthesis. If only a small library was formed, all of the pools may be combined and the entire library tested in one mixture. However, this method is typically used to form larger libraries, so it is more likely that each of the individual pools from the final round of reactions will be tested separately. Testing should not be done on too many compounds at once, or biologically active ones may be missed due to dilution by the large number of inactive ones present.
Once an active mixture has been discovered, the next task is discovering which individual compound(s) in that mixture are active.
Determining this is known as deconvolution. One way to do this would be to synthesize each of the compounds in the mixture individually. But for larger mixtures this is very impractical. Therefore, a more common method is to synthesize these compounds in increasingly smaller mixtures, as pictured below, in a process known as interative resynthesis and rescreening. Active pools have been identified, and last step leading to these active pools is known. One now must work backwards to determine the order of reagent addition leading to the active compound(s). To do this, the reaction is begun anew, but when the last step is reached, the beads are not pooled and split, but separately reacted with the reagent that led to the active pool. These pools are now tested for activity. From the active pool found here, the second-last reagent added to give the active compound is now determined. This process is repeated again, until the order of reagents, and thus the active compound, has been identified. This process is easiest when only a small number of compounds are expected to show any activity, as in the early stages of drug designs. Deconvolution is a time consuming method, taking longer than the original synthesis of the library. It also uses up reagents, thus it can also be an expensive process. Thus methods of tagging the resins, which allow for even simpler determination of the active compound, are generally employed now. These will be discussed in detail below.
The ‘Good and Bad’ of Split and Mix Synthesis
The advantage of split and pool synthesis is that you can create a large library of compounds very quickly when compared to traditional methods. The disadvantages are that when compared to traditional solution phase synthesis, the amount of product made is very small, often in such small numbers that one compound is attached to one bead. Biological assays performed on mixtures are sometimes deceiving as well, as one product which is extremely biologically active could be masked by many inactive compounds in the same pool. Similarly, a chemist could be fooled by identifying a pool containing a large number of moderately active molecules as a pool containing a few largely active molecules.
Parallel synthesis involves running multiple reactions, each in a separate vessel, at once instead of in series. Like the split and pool method, it results in the production of multiple compounds at the same time. However, unlike split an pool, parallel synthesis gives individual compounds, not a mixture. Thus deconvolution is not an issue in this method.
Advances in automation have made parallel synthesis possible. Many different apparatuses have been developed to either fully or partially automate this process. Synthesizers are machines in which the process is completely automated. The complete reaction, including all addition and separation steps as well as cleavage of the final product from the resin are done within the machine, allowing the reaction to occur unattended. Workstations allow for an assembly line-like set up. There are individual stations for reagent addition, product purification, ect. The group of reaction vessels, most commonly set up as a 96 well plate, are moved manually from station to station. However, as one vessel is moved to the next station, another can replace it.
Parallel synthesis does allow for the production of a large number of individual compounds, but other methods (ie-split and pool) are capable of higher numbers in less time, and thus are better for the earlier stages of drug discovery. Parallel synthesis is better suited to stages of lead optimization.
Split and Pool (a) vs. Parallel (b)
The ‘Good and Bad’ of Parallel Synthesis
The benefit of parallel synthesis is that it creates the compounds individually and in their own ivessel. Thus the identity of the product is already known, meaning the time consuming and expensive deconvolution step is skipped. However, the amount of vessels required for this process is large, and the number of reactions performed is even greater. This means this process is better suited, as stated above, to lead optimization rather than library generation.
Chemically Encoded Tags
A bead with an active compound is found, but how can the structure of this active compound be determined? Amounts of it on the bead are very small. One solution is to use bifunctional polymers, which are tagged. In this case, in every cycle of the split and pool method two chemical reactions occur: one at the substrate, and one at the tag. The success of this method depends on having a method to analyze the tag. Four types of tags are used: oligonucleotides, peptides, haloaromatic tags, and secondary amine tags attached to the polymer via amide linkages.
However, there are problems with using this type of tagging. A high-throughput analysis of these codes is necessary for this method to be successful, or else product determination would still be very time consuming. The biggest issue, though, is that this having tags of this sort further limits the types of chemistry which can be performed in the reaction on th substrate. Thus this type of tagging has mostly given way to mechanical tagging methods, and the process of directed sorting.
Mechanically Encoded Tags and Directed Sorting
Resins are mechanically encoded by enclosing them within small, porous packages. The number of ‘packages’ of resin is equivalent to the number of compounds which one will form. The packages themselves are labelled. This can be done by visual labels, electonic tags, chemical tags, or even radiofrequency tags. Mechanical tagging of the resins allows for the process of directed sorting to occur.
Using tagged microreactors, which are the small, porous packages mentioned above, and resins, the directed sorting method makes use of the split and pool technique, but with a twist. The tag is read during the split methodology and places the microreactors containing the specific beads into specific vessels for reaction. It “directs” the beads into vessels when “sorting” the resin and compounds, thus the term “directed sorting”. IRORI accutag technology and SynLanterns uses radio frequency tags to tag their “Kans” while IRORI XKans use 2D bar codes and a specially designed reader to sort the microreactors automatically.
Directed sorting uses the best of both parallel and split and pool techniques. The use of tagged microreactors allow for easy identification of the product within. Also like parallel, it can be used to make larger amounts of product if so desired. However, directed sorting still maintains the speed of parallel synthesis, as well as using a smaller number of reaction vessels.
Role of Technology
As technology advances, so do combinatorial techniques. The advancement in data storage, automation of solid phase and solution phase synthesis, advancement of biological assay and testing technologies all mean that larger libraries can be created and tested using combinatorial chemistry.
Advances in Automation
IRORI has developed a system to make its directed sorting methodology more automated by using coded plates as well as coded microreactors. This means that not only are the microreactors sorted but 96 well plates are as well, allowing the automation of larger and larger synthesis systems, and therefore the creation of larger libraries. These coded plates are sorted into a large sorting station, each pictured below.
A coded plate and a sorting station
Also in use is an automated washing system that accomodates the company’s microreactors, called Kans, and the plates they are contained on. It facilitates the washing of 10 000 cans in most standard laboratory solvents, making the process of washing less tedious and fully automated in the combinatorial process.
Once the reactions are complete and the compounds are washed, the product must be cleaved from the resin. This can be done in several automated ways, two of which are described. The Clevap station can fit the IRORIs XKans and microkans, as well as Mimotype lanterns. It performs the cleavage step along with several complimentary steps, and can be used to wash the microreactors if desired. It is important because it also sorts compounds onto different cleavage plates based on the cleavage type, so this does not have to be done by the operator.
Another method of automated cleavage is by the Accucleave series of cleavage stations. These stations allow for the cleavage of several plates, but do not allow for transfer to vials and drydown, which must be done by the operator/chemist.
Other automation advances are available with advances in robotics to allow for automated sorting, filling and capping of the microreactors. As well, advances in software allows for faster synthetic planning, sorting, and larger storage of modelled libraries created by combinatorial chemistry.
Advantages and Disadvantages of Combinatorial Chemistry
The creation of large libraries of molecules in a short amount of time is the main advantage of combinatorial chemistry over traditional. The cost of combinatorial chemistrys library generation and analysis of said library is very high, but when considered on a per compound basis the price is significantly lower when compared to the cost of individual synthesis.
There is a limit to the chemistry you can do when using solid phase synthesis. The resin you use is often affected by the reaction types available and care must be taken so that the attachment of the reagent to the substrate and bead are unaffected. Each reaction step has to be carefully planned, and often a reaction isn’t available because the chemistry affects the resin.
While a large number of compounds are created, the libraries created are often not focused enough to generate a sufficient number of hits during an assay for biological activity. There is a great deal of diversity created, but not often a central synthetic idea in the libraries. One can argue that there should be a focus on the type of molecule developed in order to maximize hits.
Other Applications of Combinatorial Chemistry
Once it was realized that combinatorial chemistry could be applied to the synthesis of small organic molecules, it was quickly adopted by the pharmaceutical and medicinal chemistry industries. These remain the areas where combinatorial techniques are the most prevalent. Peptide and oligonucleotide synthesis also still take advantage of this branch of chemistry. However, combinatorial libraries are now being designed for other purposes in a variety of areas. Many groups are now using combinatorial chemistry in search for new catalysts. Another use is in the field of materials science. With electronics getting smaller and smaller, the insulators needed within them also need to get smaller. Combinatorial chemistry is being employed in the developement of thinner insulators. The two examples below also illustrate the diverse uses of this science. The wide array of applications of combinatorial chemistry even include the agricultural industry and food science.
The Artificial Nose
An artifical nose is composed of optical fibres which have various chemicals at the end of them. These chemicals change color when other chemicals bind to them. However, rational design of compounds which change color in this way is a very difficult thing to do. Thus the majority of them are found through random screening. Combinatorial chemistry is used to created the large libraries which are screened. The photo below is a time-sequence of an artificial nose’s reaction to benzene. The total color change that occurs over all of the optical fibres over time is what distinguishes one compound from another. This artificial nose has been shown to be capable of detecting the presence of benzene in small amounts. Artificial noses are generally used to detect landmines or toxins.
Phosphors are inorganic molecules which emit a phosphorescence when excited by light or electrons. They are used in lighting and computer monitors. Combinatorial chemistry has been used to create large libraries of phosphors of different colors for use in these applications. One such library, called the “Christmas Library” by Symyx technologies, consists of 25,000 different phosphors created using combinatorial chemistry techniques and is pictured below.
Because of the potential advantages of combinatorial chemistry, and considering the shortcomings outlined in the disadvantages section, more work is needed in the field and much focus is on a new field of chemistry that uses combinatorial chemistry techniques while focusing the libraries created towards natural product like molecules which have a higher potential for biological activity.
Diversity oriented synthesis
DOS is this new field. It focuses on things like Isomer generation reactions, divergent reaction pathways, and folding pathways to generate a large library of molecules all using the same privileged structure from a biologically active natural product to make changes to the structure of the natural product to enhance a desired property, and identify which changes cause the enhancement. Essentially, it focuses on traditional combinatorial chemistry ideas while increasing the hit rates within the libraries created.
Diversity Oriented synthesis is still technically combinatorial chemistry, with focused reaction types and starting reagent.
Who is using DOS?
Diversity Oriented synthesis is being used on the industry scale by several different companies. The National research Council of Canada(NRC) has funded much research, under the direction of Prabhat Arya, has developed libraries of indoline and tetrahydroquinalone derivatives by using stereocontrolled DOS methods. Merck Frost has a spinoff company in France called Edelris which creates libraries for lead compound sources using DOS methods. A company called VivoQuest screens such libraries against targets for drug design and has identified preclinical candidates for the treatment of hepatitis-C, as well as other drug target preclinical candidates. Pfizer and GlaxoSmithKline have academic partnerships with universities to develop DOS methods and examine the potential of this emerging field of chemistry.
Lazo, J, and Wipf, P. 2000. Combinatorial Chemistry and Contemporary Pharmacology. JPET 293 : 705 – 709
Miertus, S., Fassina, G., and Seneci, P. 2000. Concepts of Combinatorial Chemistry and Combinatorial Technologies. Chem. Listy 94 : 1104 – 1110
Mitscher, L, and Dutta, A. (2003). Combinatorial Chemistry and Multiple Parallel Synthesis. In D.J. Abraham(Ed.), Burger’s Medicinal Chemistry and Drug Development (pp. 1-36) John Wiley & Sons, Inc.
Weller, H. (2000). An Introduction to Combinatorial Chemistry. In M. E. Swartz (Ed.), Analytical Techniques in Combinatorial Chemistry (pp. 1-28) New York: Marcel Dekker, Inc.