TECHNICAL BRIEF – Synthesis of Long Oligonucleotides
With the advent of gene synthesis and the requirement of longer oligonucleotides for mutagenesis, we thought it would be useful to our customers to review the chemistry of synthesizing long oligonucleotides (>75 nucleotides) and to provide some suggestions for their successful production.
All of the DNA synthesis reagents have an impact on good quality DNA synthesis, as are all the steps in the synthesis cycle – coupling, capping, oxidation and detritylation. However, for the successful synthesis of long oligos, some factors are absolutely critical, as noted below.
Table 1: Moisture Control while Dissolving Phosphoramidites
Fundamental to the successful synthesis of longer oligos is the need to maintain as high a coupling efficiency as possible. While an average coupling efficiency of 98.0% would theoretically yield 68% full-length product for a 20mer, it would only yield 13% full-length product for a 100mer. So, it is imperative to maintain high coupling efficiency throughout the synthesis. One of the main obstacles to this is the presence of moisture, which lowers coupling efficiency in two ways:
- Water reacts with the activated tetrazolide of the incoming monomer instead of the 5’-hydroxyl of the support, thereby scavenging excess monomer and lowering the coupling efficiency.
- Water catalyzes the conversion of the phosphoramidite to the phosphonate as the phosphoramidite sits on the synthesizer, thereby lowering the concentration of phosphoramidite itself.
It is of little surprise, then, that during the humid summer months low coupling efficiency can be problematical. As such, steps must be taken to reduce the water content of all the reagents involved in the coupling step – the acetonitrile (ACN) on the synthesizer, the ACN used to dilute the phosphoramidites, the activator solution (1H-tetrazole, DCI, ETT or others), and even the argon or helium used on the synthesizer, which should be dried with an in-line drying filter before reaching the synthesizer. And of course, the phosphoramidites themselves should be dry.
The most simple and effective first measures to increase coupling efficiency are to use truly anhydrous ACN on the synthesizer (10-15 ppm water content or lower), use fresh phosphoramidites during synthesis, and to dissolve the phosphoramidites while maintaining an anhydrous atmosphere. We recommend purchasing septum-sealed bottles of ACN (e.g., 40-4050-45) and to use a fresh bottle when putting new monomers on the synthesizer. The technique we suggest to dissolve a phosphoramidite under anhydrous conditions is described in Table 1.
Some synthesizers are more susceptible to humidity than others. In extreme situations, customers have gone so far as to make a ‘tent’ of non-static plastic sheeting around the synthesizer and placed a dehumidifer inside. The increase in coupling efficiency was dramatic.
A final note on humidity and synthesizers: if the synthesizer has been sitting idle for a while, it generally takes some time before it is fully dried out. Generally, the first oligos synthesized are not of stellar quality. However, as the lines in the machine dry and become truly anhydrous, the coupling efficiency will rise - especially after a fresh set of phosphoramidites is installed.
While it may not be readily apparent, it is critical to maintain a high capping efficiency when synthesizing long oligos. The reasons are two-fold:
- There is the practical concern of post-synthesis purification. When the capping efficiency is low, deletion mutants begin to build up to very significant levels. These give rise to the ‘n-1’ deletion mutant, which is actually a population of n-1mers, with the missing base scattered throughout the sequence. This impurity is very difficult to remove as it has a DMT group just as the full-length oligo.
- It seems that having more efficient capping helps dry the support for the next coupling step. This is quite apparent with the Expedite 8909 synthesizers, which use a CAP/OX/CAP cycle.
Certain synthesizers are less efficient at capping than others. Our in-house work indicates that an ABI 394 caps total failures (generated by injecting ACN instead of monomer solution) with about a 97% efficiency, while an Expedite 8909 only caps at 90% efficiency in these tests. Part of this disparity is due to the concentration of N-methylimidazole in their respective Cap B mixes. For an ABI, it is normally a16% solution, whereas it is 10% for an Expedite. Indeed, we found that when 10% N-methylimidazole mix is used on the ABI 394, the capping efficiency drops to 89%. As a result, we recommend increasing the delivery of the Cap A/B mix on Expedites by 50% (going from 8 to 12 pulses) as well as increasing the time interval by 50% (going from 15 to 22 seconds).
The most efficient capping reagent though is 6.5% DMAP solution used for Cap B. When this is used on an ABI 394, the capping efficiency jumps to >99%. While early work suggested that capping with DMAP could lead to a fluorescent adduct (ABI Nucleic Acid Research News, 7, Oct 20, 1988), we have never observed this side-reaction.
The supports themselves are critical to maintaining a high coupling efficiency throughout the sequence. It is for that reason that we report the recommended synthesis length of every batch of support. This number reflects the drop-off point of the CPG and indicates when the pores of the suport are beginning to become essentially ‘clogged’ with nascent DNA strands. This leads to a drop in the coupling efficiency because the reagents are unable to diffuse quickly enough to pass through the nascent DNA strands before the next step in the synthesis cycle. For very long oligos (>100), we generally recommend a 2000 Å support. The difficulty of using 2000 Å CPGs is that they are quite friable, which can lead to 3’ base deletions, and have very low loadings (10 - 20 µmoles/g), which means it is generally not possible to synthesize oligos at a 1 umole scale. However, polystyrene (PS) supports are generally also good for long oligonucleotide synthesis and can be a worthy alternative. Indeed, it has been argued that it is easier to make the hydrophobic PS anhydrous prior to the coupling step.
Even if the coupling efficiency is high and maintained throughout the length of the sequence, there are a number of side reactions that can lead to poor quality oligos. The most prominent of these is depurination. Trichloroacetic acid (TCA), which is the standard acid used in deblock solutions, is quite strong, with a pKa of approximately 0.7. Detritylation using TCA is quite fast and, for this reason, it is the standard deblocking reagent on most DNA synthesizers. However, TCA is strong enough to protonate the N7 nitrogen of adenosine and guanosine, which can lead to depurination and the formation of abasic sites. Upon deprotection, the abasic sites cleave, leading to the production of DMT-ON species truncated at the 3'-terminus, greatly complicating down-stream purification.
To limit depurination, there are two strategies:
- Use monomers that are resistant to depurination. The dimethylformamidine protecting group, commonly known as dmf, is electron donating. As such, it quite effectively protects the guanosine from depurination. The difficulty is that the dmf group is not stable enough on adenosine to be used. While more stable alternatives, such as di-n-butylformamidine, have been described, they are rather expensive and difficult to synthesize.
- Rather than use a more expensive depurination-resistant dA, the alternative is to use a deblocking agent with a higher pKa. The best choice is dichloroacetic acid (DCA). DCA has a pKa of 1.5. When using 3% DCA as the deblocking solution, we have not been able to induce depurination in a standard column synthesizer. (Note: When synthesizing on other platforms, such as on a chip, this may not be the case.) The drawback of using DCA is that the rate of detritylation is much slower. To some extent, this is compensated by the use of 3% v/v DCA/DCM, which is actually 4.5% w/v and 1.5 X as concentrated as 3% w/v TCA/DCM. Regardless, we recommend at least doubling the delivery of deblock when going from TCA to DCA since incomplete DMT removal will also lead to deletion mutations. With the increased delivery of DCA, we have not observed any appreciable depurination.
GG Dimer addition
The activators used in DNA synthesis are mild organic acids. They protonate the nitrogen of the phosphoramidite, leading to the highly reactive tetrazolide intermediate. However, because these activators are acidic, they can remove a small percentage of the 5’-DMT from the dG phosphoramidite during the coupling step, which, in turn, can react with activated dG phosphoramidite. This leads to the formation of a GG dimer and its subsequent incorporation in the sequence. Over the large number of couplings in the synthesis of long oligonucleotides, this can lead to a significant n+1 peak. As with the n-1 impurity, this too is DMT-ON and difficult to separate from the full-length oligo. The reason that GG dimer addition is seen rather than AA or TT is that guanosine detritylates faster than the other bases and hence leads to more dimer formation. To minimize this side reaction, strongly acidic activators such as BTT which has a pKa of 4.1, and ETT (pKa 4.3) should be avoided. Probably the best activator is DCI. While a strong activator, its pKa is 5.2 – even less acidic than tetrazole. DCI is a much better nucleophile than the tetrazole derivatives and this compensates for its lower acidity.
Alkylation of the N-3 position on thymidine can occur from the reaction of acrylonitrile (which is produced in situ when the cyanoethyl protecting group is eliminated) with thymidine during ammonia deprotection. It is especially noticeable when synthesizing very long oligonucleotides. The result is in an impurity that runs on reverse phase HPLC like an n+1 peak. It is easily recognized by mass spectrometry as a +53 Da species. This side reaction can be minimized by using a larger volume of ammonia when cleaving the oligo or using AMA, since methylamine is better at scavenging acrylonitrile. However, this side reaction can be eliminated completely by treating the column after the synthesis is completed with a solution of 10% diethylamine (DEA) in acetonitrile prior to cleavage in ammonia. Typically, the column if fitted with a syringe and a few milliliters of the DEA solution are pushed through slowly over a 5 minute period. Alternatively, the synthesizer can be set up with the DEA solution on one of the additional ports. In this way, the DEA rinse can be automated by writing a custom end procedure.
Purification difficulties can not be considered a side reaction per se, but we would be remiss not to mention that the purification of long oligos can pose unique problems. Traditional "OPC-type" purification cartridges do not purify long oligos very well since the DMT group is swamped by the large number of charged phosphate groups in the backbone. Recent improvements have included high affinity fluorous purification of long oligos and specially designed packing with a high affinity for DMT such as Glen-Pak™ DNA cartridges.
However, the longer the oligo, the greater is the probability of the formation of unusual secondary structures. We have found, even for cartridges like Glen-Pak cartridges, that these sequences are problematic and give poor yields. However, by heating the crude DMT-ON oligo in the loading buffer to >65 °C just prior to loading on to the Glen-Pak cartridge, the yield of purified oligo increased dramatically, as did the purity of the oligonucleotide.