In order to offer an exhaustive report on the history of oligonucleotide synthesis, it would be necessary to examine the history of modern biotechnology, for the two are tightly interwoven. For example, early biotech innovations such as fermentation, pasteurization, and vaccine development, were key stepping stones to the birth of oligonucleotide synthesis. Another event crucial to establishing this novel chemistry was the passage of the Bayh-Dole act. This legislation allowed academic institutions to own and patent federally funded research, which in turn made it possible to secure the huge levels of capital investment necessary to finance the development of groundbreaking diagnostics and therapeutics. In an attempt to present a condensed chronicle of recent scientific contributions that led specifically to the launch of oligonucleotide synthesis, this article will discuss certain biotech milestones, beginning in the 1950′s.
First, for the uninitiated, an oligonucleotide, in the strictest sense, is a short piece of nucleic acid less than 50 nucleotides in length. In the past 20 years, the meaning has broadened to include all chemically synthesized nucleic acids, regardless of the length.
Next, it is necessary to have some understanding of why researchers initially became interested in synthesizing these molecules. Although the general make-up of nucleic acids and their biological function as carriers of the genetic code was known by the mid-1940′s, it was the landmark paper published by Watson and Crick in 1953 describing DNA′s double helix structure (Watson and Crick, 1953) that revealed the link between the chemistry of genetics and the biological result. Thus was born molecular biology, the science of investigating the interface between biology and chemistry. It was only natural that chemists would soon have an interest in trying to synthetically prepare some of the newly elucidated bio-macromolecules: proteins and nucleic acids. Although the biotech industry came into existence through the growing knowledge about biological systems at the molecular level, it was ultimately fueled by the development of precise tools. It was the ability to simulate and modify biological systems, through tools such as recombinant protein synthesis and cloning techniques that allowed science to create new biomolecules. Because the chemistry was somewhat simpler, the synthesis of peptides (or small proteins) developed faster than that of oligonucleotides. Consequently, peptide synthesizers were the first automated systems available. Vega Biotechnologies pioneered this field and also introduced the first DNA synthesizer, which can be viewed in the Smithsonian. Although it used chemistry that is now outdated, it marked the first attempt to make oligonucleotide chemistry easily available to research laboratories.
However, general accessibility to the chemistry would require many advancements before reliable, automated oligonucleotide synthesis was a reality. The beginning of the era of investigating the human genome can be dated to the collaboration between Professor Marvin Caruthers of the University of Colorado, and Professor Leroy Hood of CalTech, when they set out to automate Caruthers′ new phosphoramidite oligonucleotide synthesis chemistry. This collaboration, which formed Applied Biosystems, Incorporated (ABI), commercialized the first phosphoramidite DNA synthesizer in the early 1980′s. Many labs now had routine access to oligonucleotides, which was critical in advancing the overall understanding of biological systems.
The invention of Polymerase Chain Reaction (PCR) by Kary Mullis in the 1980s proved to be the catalyst for the rapid development of a myriad of applications, including the more sophisticated sequencing methods, which made it possible to sequence the human genome. The following sections of this article will examine in greater detail the contributions of some of the influential researchers who were the direct parents of oligonucleotide synthesis.
The first published account of the directed chemical synthesis of an oligonucleotide occurred in 1955 when Michelson and Todd reported the preparation of a dithymidinyl nucleotide (Michelson and Todd, 1955).
In their report, the phosphate link between two thymidine nucleosides was made by first preparing the 3′ phosphoryl chloride of a 5′ benzyl protected thymidine, using phenylphosphoryl dichloride. This was then reacted with the 5′ hydroxyl of a 3′ protected thymidine. The chemistry worked reasonably well, albeit slowly. Additionally, the phosphoryl chloride intermediate was not stable, being susceptible to hydrolysis. Figure 1 shows the basic scheme.
In the late 1950′s a creative and forward-thinking researcher at the University of Chicago by the name of H. Gobind Khorana became interested in the synthesis of oligonucleotides. He introduced two concepts to the field that made possible the convenient synthesis of oligonucleotides more than just a few bases long. One concept, the on-off protection scheme necessary for sequential oligonucleotide synthesis, is still widely used today by oligonucleotide chemists, virtually unmodified from Khorana′s initial publications (Schaller, et al., 1963; Smith, et. al., 1961). The other was the first use of a stable phosphorylated nucleoside that coupled to the desired nucleoside when activated (Khorana, et. al., 1956).
This protocol, called the phosphodiester method of oligonucleotide synthesis (Figure 2), is the same cyclic scheme used today with the exception of the addition of one step, oxidation. In place of the hydrolysable phosphoryl chloride, he prepared 3′ phosphates of the 5′ protected nucleoside using phosphorochloridates that then hydrolyzed to the phosphomonoester. These 5′ protected nucleoside 3′ phosphates were subsequently activated using a condensation reagent, such as dicyclohexyl carbodiimide (DCC), to couple to the 5′ hydroxyl of another 3′-protected nucleoside. This method was revolutionary at the time and produced a truly remarkable feat: the synthesis of an active 72-mer tRNA molecule, which was published in Nature (Khorana, 1970).
Like most archetypes, the method did have shortcomings. Because the phosphate itself was not protected, branching at the internucleotide phosphate linkages of the previous couplings was a major problem. As a result, it was necessary to follow a very arduous multi-step purification process in which the branched contaminants were removed. However, as the oligonucleotide length increased, the percentage of branching also increased, making purification even more challenging. The solution phase chemistry made the process very slow, because the oligonucleotide had to be purified or precipitated between steps to remove excess reagents. When one considers the magnitude of the task, the accomplishment of preparing an active tRNA molecule becomes even more remarkable.
Khorana's most lasting contribution, however, was in the area of nucleoside protecting groups. The key to developing an efficient, cyclic, step-wise synthesis is a good protecting group scheme that allows the selective removal of a specific protecting group at the desired time. To make matters more challenging, the protecting groups must be removable almost quantitatively. Otherwise, the yield of desired product will be low and the product itself may be irresolvable from contaminants. To raise the bar even further, purines are susceptible to depurination under mildly to moderately acidic conditions (pH 4-5 for extended periods, pH 1-3 for fairly short periods), so strong acids should be avoided.
The solution Khorana offered for 5′ hydroxyl protection, the dimethoxytrityl (DMT) protecting group (Smith, et. al., 1961), is ubiquitous in oligonucleotide chemistry today. The combination of good general stability and easy removal with mild acid has been unbeatable. Several options are available, such as leuvenyl and FMOC, but none are as popular as the unique trityl family.
The reason this triphenyl methyl ether cleaves so readily under acidic conditions lies in the fact it is one of the few molecules that actually likes to form a carbocation. The back bonding of the pi electron cloud system formed by the three phenyl groups is sufficient to allow the methyl carbon to remain stable as a positively charged species under very mildly acidic conditions. Like many carbocations, the trityls have a distinctive color when ionized, which has turned into an extremely useful diagnostic tool. The dimethoxytrityl (DMT) carbocation has a very strong orange color in mild acid that has a high extinction coefficient, which means that even at very low concentrations it can still be accurately measured optically. The monomethoxytrityl (MMT) carbocation has a yellow color, while the parent trityl (Tr) itself is deep red (Figure 3). The efficiency of each cycle of nucleoside addition can be followed by measuring the absorbance of the released DMT and comparing it to the previous step.
Khorana also introduced the protecting groups for the nucleosidyl exocyclic amines that are today known as the standard protecting groups; isobutyryl for guanosine and benzoyl for adenosine and cytidine (Schaller, et. al., 1963; Brown, et. al., 1979) (Figure 4). Although others exist, these are the most commonly used groups today, with the possible exception of acetyl-protected cytidine, which is more readily removed compared to benzoyl.
Professor Khorana influenced many with his work, both through his publications and through his labs, where many of the great names in oligonucleotide chemistry passed as graduate students, post-docs, or visiting scholars. Those names include Marvin Caruthers, who will be discussed later in this article, and Robert Letsinger, who worked nearby at Northwestern University and developed two important steps in the field: solid phase synthesis and phosphite-triester chemistry.
Professor Letsinger began his career at Northwestern University in the late 1940's as a boron chemist. He was a significant player in that field, but in the early 1960's he turned his sights onto a newly emerging field, biomacromolecule synthesis. At that time, the target was peptide synthesis. However, a twist of fate moved Letsinger from peptide to oligonucleotide chemistry in the mid-1960′s.
Dr. Letsinger was developing a peptide synthesis scheme using solid phase chemistry that had originated mainly for the support of catalysts. Letsinger utilized flow-through technology with a cyclic chemistry scheme of adding units sequentially. When applied to peptide synthesis, it added an internal filtering system that proved to be an incredibly important step forward. However, he wasn′t the only researcher following this lead. Another scientist, Bob Merrifield, was also investigating the synthesis of peptides using solid phase technology. At the time they were neck and neck in the process of discovery and struggling to publish their findings as soon as possible. Bob Merrifield submitted his seminal paper describing the solid phase synthesis of peptides first and eventually won the Nobel Prize for his work. This unexpected scoop prompted Letsinger to regroup and focus his attention on another nascent chemistry: oligonucleotide synthesis. He rapidly converted his method using solid phase synthesis for peptides to the improvement of the oligonucleotide synthesis procedures taught by Khorana, thus converting a stroke of hard luck into scientific advancements that benefited an entire industry.
Letsinger made three major contributions to the field. First, he introduced solid phase chemistry as stated above. Secondly, he introduced the phosphotriester method of synthesis, an important improvement on Khorana's phosphodiester method. Finally, he introduced a radical departure, the P(III) based phosphite-triester method, which is the root of Marvin Caruthers′ phosphoramidite method.
Letsinger's first support for peptide synthesis was described in papers published in 1963 and 1964 (Letsinger and Kornet, 1963; Letsinger, et. al., 1964). The support consisted of what was called a "popcorn" polymer, a styrene-divinylbenzene polymer that had the unfortunate property of swelling in some solvents. In 1965 he published the first paper describing the solid-phase synthesis of dimer and trimer oligonucleotides using the same support (Letsinger and Mahadevan, 1965). In the initial report, 2′ deoxycytidine (dC) was attached through the amine at the 4 position of the base itself to acid chloride modified support and forming an amide bond that was cleaved with ammonium hydroxide. The 3′ hydroxyl of the dC was protected with a benzoyl group and the 5′ position with a DMT group. The DMT group was then removed with mild acid to prepare the support bound nucleoside for oligonucleotide synthesis. The attachment was made to the support, which was activated to an acid chloride, thus forming an amide bond that was cleavable with base (Figure 5).
Through the 1960′s he continued to explore the solid phase synthesis technique. He quickly determined that the best approach to solid phase synthesis was to attach the 3′ hydroxyl to the support, as is done today. In fact, the graduate student instrumental in that work was Marvin Caruthers. Letsinger explored a number of polymer formulations, but never found a solution to the problem of swelling that was so detrimental to the chemistry. That role fell to his former student, Caruthers, who will shortly have his own section in this history.
In the late 1960's, Letsinger published the first paper on the phosphotriester method of oligonucleotide synthesis (Letsinger, et. al., 1969) (Figure 6). The key advance of this method was the protection of the phosphate group to prevent the branching that plagued the phosphodiester approach. The protecting group most commonly used was the β-cyanoethyl group that is easily removed with ammonium hydroxide (Letsinger and Ogilvie, 1969). An o-chlorophenyl was also used but it required a more complicated deprotection mixture. It turned out, however, that the key to pushing the efficiency of the reaction, which reached levels in excess of 95% per step, was the selection of a proper activator. Mesityl sulfonyl chloride (MSCl) and mesityl sulfonyl nitrotriazole (MSNT) were by far the most popular (Devine and Reese, 1986; Letsinger and Ogilvie, 1969).
This was the first chemistry that was simple enough to reproduce successfully in many labs. The combination of a chemistry that worked relatively easily with solid phase methodology led to the creation of the first viable automated and semi-automated DNA synthesizers, exemplified by the early instruments developed by Vega Biotechnologies. Another early entrant was Ron Cook and his company Biosearch. He introduced the SAM I in the late 1970's, which was based on phosphotriester chemistry and was the most popular instrument of its era. These instruments allowed non-chemists to prepare simple oligonucleotides, and created the ability to probe genes sequence desired. Thus equipped, the industry was primed for the emerging techniques of gene mapping, PCR, and target validation.
However, the phosphotriester chemistry still suffered from critical drawbacks. Among them was the fact that despite the years of work by a number of research groups, the average step-wise efficiency could never reproducibly be raised above 97%, and often failed to reach 95%. This limited the method to the routine synthesis of oligonucleotides less than 20 bases in length. Another problem was the extensive coupling time, which resulted in cycle times that commonly ran longer than an hour and a half.
In the mid-1970's, Letsinger published the first papers describing the phosphite-triester method of oligonucleotide chemistry (Letsinger, et. al., 1975; Letsinger and Lunsford, 1976) (Figure 7). This chemistry is based on the use of reactive phosphorus in the P(III) state, instead of the classic P(V) phosphoryl chemistry. The scheme required an additional step in the synthesis cycle, oxidation, in order to prepare the natural P(V) backbone. The major advantage of this chemistry was the significant reduction in time required for coupling due to the highly reactive nature of the nucleoside phosphomonochloridite intermediate.
Oxidation of the phosphite-triester intermediate into a phosphotriester was needed in order to stabilize the backbone. This oxidation was required at each step of the cycle because of the instability of the phosphite-triester intermediate to the acid required to remove the DMT group. Fortunately, a very simple mixture of iodine, water and some base very efficiently and quantitatively oxidizes phosphorus within seconds.
The research community was quick to accept this new chemistry as a significant step forward. Not only could standard DNA be prepared faster, but the door was opened for the investigation into a variety of backbone modified oligonucleotides. Biologics, a company partially comprised of former Letsinger students, marketed an automated synthesizer based on this chemistry and another was in development by Vega Biotechnologies. However, the early form of the phosphite-triester chemistry did indeed have major drawbacks.
The most significant problem was the highly reactive nature of the nucleoside phosphomonochloridite intermediate. It was very susceptible to hydrolysis. The intermediate was not easy to store and therefore was best made just prior to each coupling. Another issue was that the formation of active intermediate was very tricky. The phosphodichloridite activating reagent had to be added to the 5′ protected nucleoside in such a manner as to maximize the formation of desired intermediate while reducing the formation of 3′-3′ dimer (Figure 9). The formation of this side-product did double damage in that it reduced the amount of desired material and increased the amount of unused phosphodichloridite that remained in solution. This unused reagent would very efficiently cap off the growing chain before the desired intermediate had time to couple. That was the reason that an excess of the reagent could not be used to reduce formation of the 3′-3′ adduct. Using too few equivalents of the phosphodichloridite had a like-wise harmful effect in that too much 3′-3′ adduct would be formed, reducing the concentration of active nucleoside reagent below a critical threshold. Increasing the concentration of the reagents to combat that only led to the opposite effect and an even less controllable reaction.
The protocols designed to optimize this reaction called for the slow addition of a very slight excess of solubilized 5′ protected nucleoside to a solution of RO-PCl2 at extremely cold temperatures (-78° C). As it turned out, the combination of the requirement for preparing the active reagent just prior to each coupling, and the need for arduous conditions during this activation, removed nearly all of the advantages brought about by the faster coupling time.
This problem was not solved until the early 1980′s. A serendipitous discovery was made by a graduate student that showed if there was a rapid introduction of the phosphodichloridite reagent to the nucleoside at room temperature, it formed a useful active reagent without making too much of the 3′-3′ adduct or leaving too much phosphodichloridite in the mix (Hogrefe, 1987). This improved method was later coupled with a scavenger system involving trityl alcohol that selectively removed any excess RO-PCl2 from the reaction mixture. It was this new protocol, which finally allowed the development of a practical automated DNA synthesizer with coupling times of 15 minutes or less. This instrument was also developed by Vega Biotechnologies in collaboration with Letsinger. Although this particular instrument was a significant improvement over the phosphotriester instruments described earlier in this section, it was never sold. The phosphodichloridite method was soon eclipsed by a new chemistry discovered by Marvin Caruthers, the phosphoramidite method. It solved many of the problems that clouded the entry of the phosphodichloridite method into the market.
After Caruthers left Letsinger′s lab he did a post-doc with Khorana, who was now at MIT. In 1973, he joined the faculty at the University of Colorado, where he continued his research in the synthesis of oligonucleotides. Over the next decade Caruthers worked out the solution to two major problems, the swelling of the organic polymer supports and the instability of the phosphitylated active nucleoside intermediate. In 1981, he published the use of inorganic matrices as supports for oligonucleotide synthesis (Matteucci and Caruthers, 1981). Originally, chromatography grade silica was used. It was much later supplanted by controlled pore glass (CPG). This solved the swelling problem experienced with the polymer supports being used at the time, thus increasing efficiency by allowing freer flow of reagents over the support. It was easier to rinse the support of old reagents prior to the next step and easier to access all portions of the support. This reduced washing time and improved coupling efficiency, as any contamination left on the support invariably interfered with the desired reaction through hydrolysis, premature oxidation, or undesired capping.
As far as modifying Letsinger′s overall phosphite-triester method, Caruthers′ contribution appears almost trivial: simply, the exchange of one leaving group, a chloride, for another, an amine (Beaucage and Caruthers, 1981; Mc Bride and Caruthers, 1983). However, this subtle change was pivotal to the development of routine oligonucleotide synthesis because it resulted in very significant changes to the properties of the molecule. Now the phosphitylated nucleoside intermediate or phosphoramidite could be made in advance, isolated as a stable solid and stored until needed. The phosphoramidite was then activated just prior to coupling by simply adding a weak acid, tetrazole. This allowed the commercial scale manufacture and distribution of DNA synthesis reagents. As shown in Figure 10, the only difference between the phosphoramidite synthesis cycle and the existing method offered by Letsinger was the activiation step. The cycle starts with the deprotection of the 5′ hydroxyl of the nucleoside bound to the support. The removal of the DMT group is normally done with a mild solution (2-3%) of either dicholoracetic acid or trichloroacetic acid in dichloromethane. Next, the support is washed well with acetonitrile to remove all traces of acid and reduce adventitious water. Now come the most important steps, activation of the phosphoramidite and coupling to the 5′ hydroxyl of the bound nucleoside. The activation is done with tetrazole, an unusual secondary amine that actually acts as a mild acid. The donation of the proton allows the formation of a thermodynamically favored anionic aromatic ring. The pKa of this acid is high enough that it does not remove the DMT group from the reagent, yet it is still acidic enough to activate the phosphoramidite.
Besides the delicate balance in activator pKa needed to achieve efficient activation without loss of the 5′ hydroxyl protecting group (Figure 11), the actual mechanism is interesting, and controversial. The currently accepted mechanism is that the amine becomes protonated (slow step), followed by rapid replacement of the protonated amine with the nearby recently generated tetrazolide. It is actually this intermediate that reacts with the 5′ hydroxyl. The evidence for this was found through 31P NMR experiments (Berner et al, 1989). The chemical shift of the phosphorus after activation suggests that an aromatic amine is bound to it, similar to what would be obtained if excess tetrazole was added to a nucleoside phosphochloridite.
After coupling, the support is oxidized using essentially the same iodine/water/base reagent originally described by Letsinger. Then the few remaining unreacted strands ending with free hydroxyls are capped using a mixture of acetic anhydride and N-methylimidazole.
This cycle, elegant in its simplicity, has remained virtually unchanged for almost two decades.
With a viable chemistry in hand, Caruthers was introduced to Hood: an instrument was designed and ABI was born. Prior to the introduction of the phosphoramidite method, the synthesis of oligonucleotides was largely in the hands of experienced chemists. The instruments were temperamental and when combined with the chemistries of the time, nearly impossible to run successfully. At its most advanced state in the mid-1980′s, the phosphodichloridite method was tricky and required a high degree of experience to reproduce successfully. Coupled with the phosphoramidite method, the automated synthesizer became useful to those who could best make use of it, the biologists.
Without a doubt, the oligonucleotide synthesis methods used today are the fruits of decades of research by progressive scientists such as Khorana, Letsinger, and Caruthers. As the field advances, TriLink follows the tradition of seeking better ways of making oligonucleotides. Today the challenges lie in the need for increasingly larger quantities of modified oligonucleotides for therapeutic applications; better high-throughput methods for the screening and PCR markets; and improved synthesis quality of dye modified oligonucleotides for the diagnostic industry. Chemists are indeed still needed for this work, but to use an old cliché, we've come a long way.