Automated And Semiautomated Dna Synthesis

The primary advantage of solid-phase synthesis is the ease with which the solid-phase supports can be manipulated. Successful chemical strategies for oligonucleotide synthesis on solid-phase supports were a vast improvement over the tedious methods previously performed in solution. In the earliest work, solid-phase synthesis was manually performed using simple funnels or test tubes to hold the supports.104341 However, researchers sought easier and faster ways to perform the syntheses. One such method used a syringe to hold the insoluble support.342 343 This reduced exposure of the support to atmospheric moisture and made it easier to work on small synthesis scales. Alternately, manually operated reagent manifolds were used to deliver the necessary reagents and solutions.344345 Although, these improvements were more convenient when compared to previous methods, they were not automated and not suitable for use outside of an organic chemistry laboratory.

The first automated solid-phase synthesizers were developed to support peptide synthesis, since this chemistry matured more quickly than oligonucleotide synthesis. One of the first attempts at

Bypass Waste Valve

FIGURE 1.17 Schematic diagram of an early prototype DNA synthesizer used at McGill University in 1980.

Bypass Waste Valve

FIGURE 1.17 Schematic diagram of an early prototype DNA synthesizer used at McGill University in 1980.

automated solid-phase oligonucleotide synthesis (using phosphodiester coupling chemistry) was performed on a modified Beckman 990 peptide synthesizer.82 This instrument, which was very large and programmed using a punched paper tape, was considered a remarkable innovation and state of the art for peptide synthesis in the 1970s. However, this technology did not provide much advantage to nucleic acid chemists because of the significant differences between peptide and oligonucleotide synthesis in reagents, supports, and operating scales. Instead, successful automated oligonucleotide synthesis required the independent development of instrumentation specific to oligonucleotide chemistry.

Early automated DNA synthesizers with chlorophosphite and phosphoramidite coupling chemistry used solid-phase supports packed into high-pressure columns similar to HPLC columns.98 This was, not surprisingly, because the early silica gel supports were based on HPLC media. It was a relatively simple matter to assemble a linear manifold of three-way valves to deliver the appropriate synthesis reagents and solvents to a high-pressure HPLC pump under some type of programmable control. The pump then forced the reagents through the synthesis column and into a waste receptacle. Addition of another valve on the outlet end of the column allowed reagents to be recycled through the synthesis column. A schematic diagram of an early semi-automated DNA/RNA prototype synthesizer used by the author during his graduate studies (early 1980s) is shown in Figure 1.17. This instrument was programmed by shading appropriate lines on a paper card with a pencil.

More sophisticated instruments were soon offered commercially from vendors such as Vega, Bio Logicals, KabiGen AB, and Biosearch.100 103 346 Although these instruments were significant because they were the first generation of fully automated DNA synthesizers available to the general research population, they all shared certain shortcomings. The valves, fittings, and pumps were based on readily available HPLC components and were not particularly reliable when used with the organic reagents required for DNA synthesis. This problem was compounded by the relatively high operating pressures of the HPLC-like column configuration. These synthesizers were also limited to the synthesis of one oligonucleotide at a time. Consequently, none could be considered a market success.

General acceptance of automated DNA/RNA synthesizers as reliable, easy-to-use laboratory instruments traces its roots to the laboratory of Leroy Hood at the California Institute of Technology. In this laboratory a microchemical facility for the analysis and synthesis of genes and proteins was established.107 These researchers developed a collection of improved instrumentation that began with a protein microsequenator and eventually included an automated DNA synthesizer. The original DNA synthesizer used a conical Pyrex flask as a prereactor to mix the nucleoside phosphoramidite reagents with tetrazole before delivering the activated mixture to a second Pyrex reactor containing the support.108 However, mixing in a prereactor was unnecessary because of the speed of the phosphoramidite activation, and this feature was later eliminated. However, two key innovations from these early instruments were destined to greatly simplify and improve the reliability of future DNA synthesizers. First, an innovative use of positive argon gas pressure, instead of a mechanical pump, was used to move reagents and solvents. This eliminated the problems with leaky check valves, piston seals, and air bubbles that plagued earlier pump-based synthesizers. Second, new zero-dead volume valves with chemically inert Kalrez diaphragms, which were mounted side by side on a single Kel-F block, were developed. These valves were highly reliable and the availability of multiple valves (up to 12) in a single valve block significantly reduced line volumes and eliminated the number of connections required.

This technology was commercialized with the formation of Applied Biosystems, Inc. (Foster City, CA) in the early 1980s. The model 380A DNA synthesizer, introduced by Applied Biosystems in 1982, was an immediate success because of its low reagent consumption, reliable design, built-in computer control, and ability to simultaneously prepare up to three different oligonucleotides at the same time (on scales of either 0.2, 1, or 10 |imol). Eventually, the model 380A synthesizer was succeeded by the 380B synthesizer, which added a touch-screen CRT for programming and a floppy disk drive for data storage. A more economical single column 381A synthesizer was also introduced. These synthesizers quickly dominated the market for automated DNA/RNA synthesizers and became very widely used. Later, the synthesizer design was further refined with the introduction of the model 391 (single-column), 392 (two-column), and 394 (four-column) synthesizers, which introduced optional online trityl conductivity measurement177 and external control from a Macintosh computer. Operating costs were also significantly reduced with the implementation of low-volume (LV) columns347 and smaller (0.04 | mol) synthesis scales. The 390 series synthesizers have proven to be highly reliable, and many core DNA synthesis laboratories, including the author's, have relied upon them. For example, at Amgen, Inc., a high-throughput, highly automated oligonucleotide production facility was developed with over 20 394 DNA synthesizers and a variety of robotic workstations.348 With each synthesizer capable of synthesizing four different oligonucleotides at a rate of one synthesis cycle every 7 min, this facility could synthesize more than 300 20-base-long oligonucleotides each day.

Other automated DNA synthesizers, using the same phosphoramidite chemistry, were also introduced by vendors such as Pharmacia, Beckman, PerSeptive Biosystems (now part of Applied Biosystems), and Eppendorf. However, these instruments never gained the market acceptance of the Applied Biosystems synthesizers.

The rapidly increasing demand for oligonucleotides has been a constant incentive for the development of techniques and instrumentation that can produce more oligonucleotides faster and cheaper. One approach to the synthesis of more oligonucleotides per instrument is to simply increase the number of column positions. In addition to the above Applied Biosystems instruments, which could perform simultaneous synthesis on up to four synthesis columns,334 other DNA synthesizers have had up to 10 positions available in a row, or up to 24 positions available in a cylindrical cartridge. However, the 10-position instrument was never commercialized, and the 24-position instrument was a proprietary design for the exclusive use of Genset Oligos, a commercial DNA synthesis company.

Another high-capacity DNA synthesizer was the Applied Biosystems 3948 instrument, introduced in 1996,334 which was described as a high-throughput automated oligonucleotide production system. In this instrument, three synthesis columns were used in parallel for the actual solid-phase synthesis. However, a revolving carousel containing 16 sets of three columns allowed continuous production of up to 48 different primer-length oligonucleotides in each 24-h period. This instrument completely automated the ammonium hydroxide cleavage and deprotection steps. Each synthesis column contained both a derivatized polystyrene solid-phase support for synthesis and an underiva-tized polystyrene support. Solid-phase synthesis occurred on the former material, while the underivatized polystyrene support acted as a solid-phase extraction medium to automate the purification of the final 5'-tritylated product. Thus, one instrument produced deprotected and purified oligonucleotides suitable for immediate use. However, this instrumentation was expensive and best suited for the synthesis of small amounts of oligonucleotide primers. A less expensive alternative was the Expedite 8909 DNA synthesizer (which replaced the 391/392/394 models in 1998) with the multiple oligonucleotide synthesis (MOSS) option. This instrument allowed up to 12 oligonu-cleotide columns to be configured at one time. However, synthesis was only performed on two columns at a time and automatic cleavage and deprotection were not performed.

The above instruments can be considered as employing a closed reagent delivery system, in which all reagents flow through Teflon tubing, common manifolds, and sealed reaction columns. Although reliable, this type of delivery system has two problems. First, increasing the number of synthesis columns significantly increases the number of valves required and the complexity of the tubing network. Second, such a design leads to excess reagent consumption because of the necessity of rinsing common pathways and expelling dead volume during synthesis. A more efficient open reagent delivery system was developed at the Lawrence Berkeley Laboratory at the University of California, Berkeley in 1995.333 In this design, a stationary dispensing manifold drops reagents into an open, multichannel reaction chamber mounted on a movable linear table inside an argon atmosphere. Each reaction chamber contained 12 individual wells holding insoluble support. Up to eight reaction chambers could be used for a total synthesis capacity of 96 oligonucleotides. This design significantly reduced the plumbing complexity, increased throughput by a factor of 24, and reduced reagent consumption and hazardous waste generation by approximately 70%.

A similar but improved high-throughput DNA synthesizer was also independently developed at the Stanford DNA Sequencing and Technology Center at Stanford University.332 This instrument used readily available 96-well microtiter plates as the reaction chambers, 11 banks of valves (each mounted in a set of eight), and a linear movable stage to position each well under the appropriate reagent dispenser. This automated multiplex oligonucleotide synthesizer (AMOS) instrument had an 11-min cycle time (96 couplings) and could produce a plate of 96 different 20 base-long oligonucleotides in <4.5 h (not including deprotection and lyophilization times). Installation of the correct prederivatized support (i.e., A, C, G, or T) in the correct well of multiwell synthesis plates was a new difficulty arising from this type of technology. However, this problem can be solved by either robotic support dispensing or the use of universal solid-phase supports or the alternative in situ nucleoside derivatization methods described in the previous section.

Later, 96-well plate synthesis technology was further improved by including two-dimensional movement. In this modification, the reagent dispensers moved in the y-axis, and up to 64 different reagents could be rapidly delivered into any of the 96 wells for automated ribozyme or combinatorial solid-phase organic synthesis.349 Although these new designs represented a dramatic improvement, the inventors had initial difficulties finding commercial backing, and eventually exclusive rights to this design were awarded to a single company, ProtoGene Laboratories (purchased later by Life Technologies, Inc.). However, once available the cost savings and advantages of this instrumentation were quickly realized. Indeed within 1 year (1994-1995) of producing the first oligonucleotide Life Technologies became the largest custom oligonucleotide-synthesis company in the world.

This new competition also started a major price war among commercial oligonucleotide synthesis services, and oligonucleotide users benefited from a several-fold decrease in cost. Although this further increased the availability of synthetic oligonucleotides, it reversed the trend toward oligonucleotide synthesis on local, laboratory-scale instrumentation in favor of synthesis in highly specialized, high-throughput commercial facilities. For example, by 1997 Life Technologies had a daily oligonucleotide synthesis capacity in excess of 7500 oligonucleotides per day on approximately 25 96-well DNA synthesizers. Other large commercial DNA synthesis companies also expanded rapidly (using different and often proprietary instrumentation), and a synthesis capacity of thousands or tens of thousands of oligonucleotides per day soon became common. For example, at Illumina (San Diego, CA), a proprietary DNA synthesizer called the Oligator 768™ uses a centrifugal rotor to produce eight 96-well plates of oligonucleotides per run. Twelve of these instruments, operating within an in-house synthesis facility known as the Oligator Farm™, provide Illumina with an annual oligonucleotide synthesis capacity of greater than 5 million oligonucleotides per year. Such incredible progress in speed and capacity must surely be astonishing to the researchers who toiled so hard in Khorana's laboratory making the first synthetic genes.

Other 48- or 96-oligonucleotide-capacity and commercially available DNA synthesizers have also been described. In one, a pipetting robot — originally designed for peptide synthesis350 — was used to deliver reagents to a rack of 48 disposable pipet tips containing a derivatized membrane support. In others, either a single 96-well synthesis plate (PolyPlex DNA Synthesizer, GeneMa-chines) or a dual 96-well plate oligonucleotide synthesizer (the "MerMade," BioAutomation) were developed335 and made commercially available. A totally different approach based on the concept of "tea-bag" synthesis (originally developed for peptide synthesis) on movable pins has appeared. In the PrimerStation 960 DNA synthesizer from Intelligent Automation Systems, Inc. (Cambridge, MA), the solid-phase support is enclosed in a porous pouch and attached to the end of a pistonlike rod. The 96 independent rods move up and down to dip the pouch in and out of the reagent-filled tray. However, this tray must be rinsed and refilled with new reagent for every step, and this leads to high reagent consumption. Also, the method is only suitable for small-scale synthesis due to the small size of the pouches (~2 to 3 mg), and automated cleavage or deprotection is not performed. Finally, in 2000 Applied Biosystems introduced its model 3900 DNA synthesizer. This instrument was a hybrid synthesizer using individual prepacked columns instead of multiwell synthesis plates. However, the columns were mounted in an open-reagent system using a circular 48-column rotor. Thus, this instrument combined the speed and economy of the above 96- and 384-well synthesizers with the flexibility and convenience of individual prepacked synthesis columns. Up to 288 primer-length oligonucleotides (six runs of 48) could be prepared in a 10-h period using any combination of 40-, 200-, and 1000-nmol scales.

Another method for speeding simultaneous synthesis of large numbers of different oligonucle-otides is known as segmental solid-phase synthesis. In this approach, the synthesis is interrupted after each coupling cycle. Each solid-phase support is sorted into one of four groups (A, C, G, or T), depending on which base is required for the next addition. The supports in each group are then stacked together, and the next coupling cycle is performed on the entire stack. Although each coupling must be performed on four different stacks, up to 100 or more oligonucleotides can be processed at a time. When first introduced, this method used paper disks as the insoluble support345,351-354 (cellulose has many surface hydroxyl groups that can anchor growing oligonucle-otide chains). These paper disks were inexpensive, easy to handle, and easy to label with a pencil. However, they were not suitable for automation. Subsequently, simultaneous synthesis of multiple oligonucleotides using segmental units composed of various stackable cartridges or wafers has been reported.344,355-357 This has led to the development of the Abacus commercial high-throughput oligonucleotide synthesis employed by Sigma-Genosys (Woodlands, TX). However, fully automated sorting is still very difficult. Consequently, the systems developed have only been semimech-anized, and the actual sorting steps have been left to human operators. However, even with this manual intervention, one instrument and one operator can produce 144 25-base-long oligonucle-otides in only 4 h. After solid-phase synthesis, the wafers are moved to a separate automated cleavage, and deprotection instrument and a laboratory with only ten instruments can produce >2500 oligonucleotides per day. Segmental synthesis is also not limited to macroscale syntheses, as in the above examples. Use of fluorescent dye encoding can allow the size of a single support to be reduced to a single 8.8-|jm microsphere. Sorting of these individual microspheres can be performed at sort rates of approximately 25,000 events/sec by flow cytometry. This concept has recently been demonstrated,358 but it remains to be fully developed.

The above strategy represents a simple combinatorial approach to solid-phase synthesis. However, more elaborate combinatorial strategies can also be employed. The most powerful of these is the approach developed by Stephen Fodor and others at Affymetrix to prepare high-density oligo-nucleotide arrays on glass wafers (Genechips).359-364 In this approach, photolithography, lightsensitive protecting groups, and a series of combinatorial masking steps are employed to produce two-dimensional arrays of oligonucleotides permanently immobilized on a flat glass surface. Very large numbers of different oligonucleotides with defined sequence and defined location can be rapidly produced because the combinatorial process allows 4N different sequences to be prepared using only 4N cycles. The precision of modern photolithography also allows extremely small cell dimensions (10 |im x 10 |im) in the arrays. This technology has made it possible to mass-produce oligonucleotide arrays containing hundreds of thousands to more than a million different oligonu-cleotide sequences on small (1.28 cm2) glass chips. The high cost of fabricating individual photolithographic masks makes this technology somewhat inflexible and quite expensive. However, new micro-mirror technology may make maskless photolithography possible. These micro-mirrors were developed for video projection systems and employ one electronically controlled mirror for each pixel in the image. A maskless array synthesizer using this technology has recently been used to prepare oligonucleotide arrays containing more than 76,000 16 |im2 features.365 The diagnostic potential for these arrays holds enormous promise for applications in DNA sequencing, gene expression studies, and health care, and a great deal of investment in and development of DNA array technology is under way.10,11,136,366-378

An alternative in situ method to the synthesis of oligonucleotide arrays on glass chips uses inkjet printing technology. In one approach piezoelectric capillary jets delivered phosphoramidite monomers in propylene carbonate microdroplets to exact positions on a hydrophobic glass wafer.379-381 Conventional phosphoramidite synthesis conditions, i.e., no photolithography, were used to produce arrays of ~25,000 different oligonucleotides. In another approach photolithography is used to create spatially addressable, circular features containing amino-terminated organosilanes on glass.382 Piezoelectric ink-jet reagent delivery of conventional, dimethoxytrityl protected phos-phoramidite reagents is used to synthesize the oligonucleotide features. This method, uses the differential surface tension between the modified and unmodified glass sites to define each synthesis site and has been termed a surface tension array.

The amount of synthetic oligonucleotide produced in a single cell of a high-density DNA array is very small (at an average loading of 20 pmol/cm,2 a 10 |im2 cell contains only 20 attomoles of DNA) and represents the smallest end of the scale for automated oligonucleotide synthesis. However, the various applications for synthetic oligonucleotides span a very large range. On the high end of the scale are the synthetic requirements for nucleic acid medicines, which are expected to exceed 1000 kg/year in the near future. In order to produce such large quantities, very high capacity automated DNA synthesizers operating at up to a 2-mol scale are being developed.383 Thus, chemical oligonucleotide synthesis spans a scale of 17 orders of magnitude.

However, despite the therapeutic promise of antisense oligonucleotides, development of synthetic methods and instrumentation capable of a 1-million- to 10-million-fold increase in synthesis scale has been difficult. These large-scale syntheses must be performed under the strictly controlled and validated conditions required for pharmaceutical products, and, most importantly, the final products have to be produced as cheaply as possible in order to be affordable to the patient. In particular, there are three factors that make very large scale oligonucleotide synthesis very challenging.

First of all, synthetic oligonucleotides are much larger and more complex than any other type of pharmaceutical prepared by organic synthesis. More than 80 individual steps and over 24 raw materials are required, and any single failure can result in the loss of an entire synthetic run. Second, the cost for materials and reagents is very high, and certain materials (such as the deoxyribonucle-osides, which are commercially obtained from salmon sperm) have limited availability. Consequently, oligonucleotide manufacturing methods that have high yields, few and simple steps, high atom efficiency, and use materials that are readily accessible, inexpensive, and environmentally acceptable have had to be developed.149 Finally, scale-up of reactor (i.e., column) and instrument design over several orders of magnitude is not straightforward, and significant problems must be addressed to maintain synthesis efficiency, reliability, and economy.

The earliest commercial DNA synthesizers had a maximum synthesis capacity of about 10 to 15 |mol or 0.01 to 0.015 mmol (~400 to 500 mg of support). However, by the early 1990s commercial synthesizers for larger scales were introduced. The Biosearch (later Millipore) model 8800 DNA synthesizer was the most widely used of these early instruments for production of oligonucleotides in gram quantities.384 This instrument used a "mixed-bed" (fluidized) glass reactor that agitated the CPG support with argon gas bubbles. With this type of reactor only a fivefold excess of phosphoramidite reagent (instead of the tenfold excess commonly used for smaller synthesis scales) was necessary for satisfactory coupling efficiency. Initially, this instrument had a synthesis capacity of 0.03 to 0.4 mmol (1 to 10 g of support) using CPG with a loading of 30 to 40 |imol/g. Later, this capacity was increased to 0.6 mmol, and the amount of excess phosphora-midite was reduced to only 4 to 5 equivalents. Researchers at Hybridon, Inc. further increased the 8800's synthesis capacity by modifying it with larger reaction vessels (up to 250 ml, capable of holding 65 g of CPG) and using higher loading (~80 |imol/g) CPG.150 With these modifications, oligonucleotides were synthesized on either 1-, 2-, or 5-mmol scales with respective yields of 2.4, 4.8, or 12 g of purified product.

The Applied Biosystems 390Z synthesizer,385 introduced in 1991, also used a mixed-bed plastic reactor, but this was attached to a vortex mixer to keep the support agitated. This instrument had an initial synthesis capacity of only 0.025 to 0.2 mmol. Using this instrument and multiple 0.2 mmol scale syntheses, 3 g of an anti-rev (HIV III) phosphorothioate oligonucleotide was reported.148 The synthesis capacity was eventually increased to 1 mmol through the use of high-loaded (150 to 190 |imol/g), polyethylene glycol-polystyrene (TentaGel) supports.236,386

Other proprietary synthesizers with stirred-bed reactors have also been recently developed. Lynx Therapeutics has reported an instrument for phosphorothioate oligonucleotides synthesis on 1 to 10 mmol scales using high-loaded TentaGel (150 to 170 |imol/g) supports.151 Synthesis was performed using only 2.5 to 3.5 equiv. of phosphoramidite. Another large-scale stirred-bed reactor, for use with the CPG supports, was also developed at NexStar Pharmaceuticals (now part of Gilead Pharmaceuticals), but details are not available.

A different approach to reactor design was taken by the Large Scale Biology Corporation, which developed a production centrifugal oligonucleotide synthesizer (PCOS) with a zonal centrifuge rotor as the reactor.387 In this design a combination of centrifugal force and density differences between solutions is used to displace one solution by another quantitatively. The solid-phase support is packed in an annular bed (68 to 400 ml). This type of design was advantageous because the CPG solid-phase support could be exposed to exactly the same conditions for the same time intervals, regardless of the synthesis scale. In collaboration with Isis Pharmaceuticals a PCOS-2 large-scale synthesizer was satisfactorily evaluated.149 These tests showed that the solid-phase support was rugged enough for packed-bed use and that a packed-bed reactor was more efficient in both solvent and phosphoramidite consumption. However, the centrifugal reactor design was too complex (and the idea of large, industrial-scale reactors spinning at high speeds was unsettling from a safety perspective), and their development was halted. Instead, packed-bed reactor design using a flow-through, fixed-bed, low-aspect chromatography column was investigated instead.

A series of fixed-bed, flow-through DNA synthesizers was developed by Pharmacia (later Amersham Pharmacia Biotech) beginning in 1993. These synthesizers used packed-bed, stainless-steel columns to hold a proprietary high-loading (~80 | mol/g), polystyrene-based, solid-phase support known as Primer Support 30, HL. The synthesis capacity of these instruments was successively increased from 0.4 mmol (OligoPilot I, 1993), 5 mmol (OligoPilot II, 1996), 100 to 200 mmol (OligoProcess I, 1996), to 500 mmol (OligoProcess II, 1999), as both pilot- and process-scale instrumentation was developed. An even larger production-scale synthesizer, the OligoMax, which can produce over 6 kg of pure oligonucleotide per day on a 2000-mmol (2 mol) synthesis scale, has also been designed but not yet produced.383 Significantly, this series of synthesizers was designed so that method development could be linearly scaled up between instruments. Proprietary DNA synthesizers, such as the Hybridon 601, with fixed-bed reactors have also been developed by Hybridon for large-scale synthesis of oligonucleotides on a contract basis. The synthesizers have a capacity of up to ~100 mmol and use either CPG or proprietary high-loading solid-phase supports.

Between 1990 and 1995, Isis Pharmaceuticals (among others) was able to obtain a 20,000-fold scale-up using OligoPilot and OligoProcess instrumentation. Furthermore, the excess phosphora-midite required was reduced to only 1.5 equiv. without any decrease in coupling efficiency. This has allowed Isis Pharmaceuticals to produce kilogram quantities of phosphorothioate oligonucle-otides for its various clinical trials. A large-scale contract oligonucleotide manufacturing facility was established in 1999 using OligoProcess II instrumentation in a new £3.5 million oligonucleotide manufacturing suite at Avecia (formerly Zeneca Specialties, Grangemouth, Scotland). Operating on a 500-mmol scale, this suite can produce 1.5 kg of oligonucleotide every 10 h for annual capacity of 750 kg. This capacity is expected to exceed 3000 kg per year when the 2-mol scale OligoMax is available.

The success of the above fixed-bed reactor synthesizers has shown that even complicated, high-molecular-weight oligonucleotide compounds can be satisfactorily manufactured on pharmaceutical scales. However, the cost of producing these materials is still quite high. Prior to 1990, synthesis of gram quantities of oligonucleotides was unthinkable because of the extremely high cost (>$100,000/g). However, with the introduction of the Applied Biosystems 390Z and Millipore 8800 DNA synthesizers it became possible to prepare the first gram-scale quantities. However, the cost was still very high (~$10,000 to 40,000/g). Since then, refinements in virtually every aspect of large-scale synthesis have led to substantial cost reductions. For example, in the years 1993, 1994, and 1995 the cost of raw materials required to synthesize 1 g of oligonucleotide decreased to only $2000, $1600, and $500, respectively. Actual costs for present synthesis are not available, but when synthesis reaches tons per year, an ultimate raw materials cost of ~$50/g or less is hoped for.

A major obstacle to reducing the cost of synthesis is the relatively high cost of the insoluble support. This is the most expensive single material required, and in 1998 it accounted for ~40% of the total raw materials cost. The most commonly used solid-phase support (Pharmacia HL-30) costs around $25,000/kg and can be used only once before being discarded. On a 500-mmol scale, the cost of support is more than $150,000 per run. Since manufacturing operations will perform two runs per day, the annual cost for support will be very high. Therefore, there is great incentive to substantially reduce the cost of this material.

One way to do this would be to abandon solid-phase synthesis altogether and return to solution-phase methods for industrial production. Recently, impressive improvements in coupling efficiency, sulfurization efficiency, and workup conditions using an H-phosphonate synthetic strategy have been achieved.115 This new solution-phase approach, combined with a block-coupling strategy, has been successfully used on industrial scales to produce short oligonucleotides (<10 bases long). However, yields drop off rapidly as chain length increases, and so far this method has not been competitive with solid-phase synthesis for the longer (~20-base-long) oligonucleotides required for antisense therapeutics.

An alternative method to reduce the relative cost of solid-phase supports is development of supports that can be used more than once. This objective has been a goal of the author's laboratory for the last several years. During this time a new linker arm, the Q-Linker, was developed that can be rapidly removed under mild conditions.268 In addition, very fast methods for attaching the first nucleoside to a solid-phase support were developed that could be performed automatically by a DNA synthesizer.317,338 When combined with novel hydroxyl-derivatized solid-phase supports (instead of the amino-derivatized supports commonly used) these developments allow a strategy for multiple oligonucleotide syntheses on reusable solid-phase supports302-304 (Figure 1.18). The

+1 0

Post a comment