Solid-Phase Synthesis of Phosphorothioate Oligonucleotides Using

Publication Date (Web): September 4, 2018 ... in which a four-reaction cycle consisting of detritylation, coupling, sulfurization, and failure sequenc...
0 downloads 0 Views 2MB Size
Article Cite This: J. Org. Chem. 2018, 83, 11577−11585

pubs.acs.org/joc

Solid-Phase Synthesis of Phosphorothioate Oligonucleotides Using Sulfurization Byproducts for in Situ Capping Jimin Yang, Jessica A. Stolee, Hong Jiang, Li Xiao, William F. Kiesman, Firoz D. Antia, Yannick A. Fillon, Austen Ng, and Xianglin Shi* Antisense Oligonucleotide Development and Manufacturing; Analytical Development, Biogen, 115 Broadway, Cambridge, Massachusetts 02142, United States

J. Org. Chem. 2018.83:11577-11585. Downloaded from pubs.acs.org by UNIV OF SUNDERLAND on 10/05/18. For personal use only.

S Supporting Information *

ABSTRACT: Oligonucleotides containing phosphorothioate (PS) linkages have recently demonstrated significant clinical utility. PS oligonucleotides are manufactured via a solid-phase chain elongation process in which a four-reaction cycle consisting of detritylation, coupling, sulfurization, and failure sequence capping with Ac2O is repeated. In the capping step, uncoupled sequences are acetylated at the 5′−OH to stop the chain growth and control the levels of deletion, or (n−1), impurities. Herein, we report that the byproducts of commonly used sulfurization reagents react with the 5′−OH and cap the failure sequences. The standard Ac2O capping step can therefore be eliminated, and this 3-reaction cycle process affords a higher yield and higher or comparable overall purity compared to the conventional 4-reaction synthesis. This improvement results in reducing the number of reactions from ∼80 to ∼60 for the synthesis of a typical length 20-mer oligonucleotide. For every kilogram of an oligonucleotide intermediate synthesized, > 500 L of reagents and organic solvents is saved, and the E-factor is decreased to 200 equiv relative to the failure sequences). While

othioate 4. The last step in this process, step 4, is the capping of the unreacted 2 with Ac2O. Repeating these four reactions affords full-length oligonucleotide products in a variety of lengths. The capping reaction is used to control deletion impurities. It does not contribute to chain elongation and would be unnecessary if the coupling reaction was complete. However, in practice, the coupling reactions do not proceed to 11578

DOI: 10.1021/acs.joc.8b01553 J. Org. Chem. 2018, 83, 11577−11585

Article

The Journal of Organic Chemistry Scheme 3. Structures of Oligonucleotides 9−14

effective in controlling (n−1) impurities, side reactions caused by the capping step lead to impurities15,16 and decrease product yield.17 The mechanism for the yield decrease is still unclear. The impurities identified include nucleobase guanine-15 and adenine-modified15,16 structures. Therefore, skipping the capping step in some cycles to improve the yield and/or purity of ASOs has been reported in the patent literature.18 However, to our knowledge, a systematic study examining the effects of eliminating the Ac2O capping step during PS oligonucleotide manufacturing has not been reported. Herein, we report the results from our investigation, propose a new failure sequence capping role played by the byproducts of the sulfurization agents, and summarize the environmental impact of the process change.



Figure 1. Reactions to attach each nucleotide.

sulfurization reagents in the manufacturing of PS oligonucleotides, phenylacetyl disulfide (6)13,20 and 3-amino-1,2,4dithiazole-5-thione (7),21 were used because they afford products in high yield and purity. After completion of the syntheses, the product was cleaved from the resin, and the base protecting groups were deprotected by ammonolysis. The 5′ DMT-on product was analyzed via reversed-phase (RP) UHPLC-UV-MS to determine the yield and purity of the product and the level of (n−1) impurities. Figure 2 shows the UHPLC-UV chromatograms of crude DMT-on 9 obtained from the syntheses with and without the Ac2O capping step, using 6 as the sulfurization agent. The small peaks eluting before the main product peaks include the deprotected capped failure sequences. Identities of the failure sequences were confirmed by high-resolution mass spectrometric (HRMS) analysis. As the figure shows, the two

RESULTS AND DISCUSSION

The impact of the capping step on the synthesis of a DMT-on 18-mer 2′-MOE ribose phosphorothioate oligonucleotide 9 (Scheme 3) was examined using a NittoPhase HL Unylinker resin and 5′-DMT phosphoramidites as the starting materials following a typical solid-phase synthesis protocol.13,19 The syntheses were carried out under similar conditions, except, in the 4-reaction cycle synthesis, an Ac2O capping step was included; but in the 3-reaction cycle synthesis, the capping step was excluded and the sulfurization reagent was recirculated through the column (Figure 1). The two most widely used 11579

DOI: 10.1021/acs.joc.8b01553 J. Org. Chem. 2018, 83, 11577−11585

Article

The Journal of Organic Chemistry

(see values in Table 1) in the 3-reaction process might be attributed to small process variation, including the detritylation Table 1. DMT-on 9 Crude Product Yield, Purity, and (n−1) Impurities 4-reaction process

3-reaction process

percentage (%)

6

6

7

n-p(MOE G) n-P(MOE A) n-p(MOE MeC) + p(MOE T) MS purity UV purity yield

0.49 0.26 0.70 92 92 71

0.45 0.31 0.80 92 94 75

0.48 0.31 0.74 94 95 77

step as well as analytical variation.23 Similar results were obtained with 7 as with the sulfurization agent, as summarized in Table 1 (see the SI, Figures S5 and S6). The 3-reaction cycle process also afforded ∼5% higher yield and the same or better overall purity. Five more DMT-on 20-mer PS oligonucleotides 10−14 were synthesized using 20 different phosphoramidites, including those with 2′-MOE24 and 2′,4′-constrained 2′-Oethyl25 modifications, common oligonucleotide building blocks (Scheme 3). Compounds 11−14 consist of 2′- modified and 2′-deoxy ribose nucleotides, resembling the structures of gapmer oligonucleotide compounds. Each of the nucleotides was synthesized four times using the same 3- and 4-reaction synthesis conditions with 6, 7, or 8 as the sulfurization agents, respectively. UHPLC-UV-MS analysis data showed that the 3-reaction and 4-reaction processes afforded similar results for the failure sequences and (n−1) impurities for oligonucleotide compounds 10−14 (see SI Figures S7−16). Representative examples, the overlays of the zoomed-in mass spectra of the (n−1) region of 10 synthesized by the 3- and 4-reaction processes using 6, 7, and 8 as the sulfurization reagents, are shown in Figure 4. UV absorbance (or optical density (OD)) per μmol of compounds 10−14 obtained from the 3-reaction process is similar to or better than that obtained from the 4reaction process. All of these results demonstrate that the conventional Ac2O capping step adds no value to the control of (n−1) impurities in the synthesis of phosphorothioate oligonucleotides. Mechanistic Studies. We suspected that the sulfurization reagent byproducts performed an in situ capping reaction of failure sequences (Scheme 4). Hanusek and co-workers in studies of sulfurization mechanisms found that the byproducts of 7 and 8, thiocarbamoyl isothiocyanate or derived compounds, react with EtOH efficiently.26,27 This result led us to question whether a byproduct of 6 could react in a similar manner with a free 5′−OH. Careful review of data from the synthesis of a PO/PS mixed backbone oligonucleotide (15) (Scheme 5), consisting of 4 PO and 15 PS linkages, lent some support to the hypothesis that capping occurs in situ in the sulfurization step with 6 as the sulfurization agent. In a 3-reaction synthesis, in situ capping during sulfurization would take place when installing a PS linkage, but no capping would occur when installing a PO linkage. As a result, (n−1) species associated with PS linkages would be expected to be similar in the 3-reaction and 4-reaction syntheses, but the (n− 1) species associated with PO linkages would be higher in the 3-reaction process because no capping species would be

Figure 2. Stacked UHPLC UV chromatograms of crude DMT-on 9 synthesized with (red trace) and without (black trace) the Ac2O capping step with 6 as the sulfurization agent.

syntheses resulted in crude product with the same failure sequences and in similar quantity. The overlaid mass spectra in Figure 3 shows the impact of removing the capping step on the level of (n−1) impurities.

Figure 3. Mass spectra shows the (n−1) impurities of crude DMT-on 9 synthesized with (red) and without (black) the Ac2O capping step. Full mass spectra are provided in the SI (Figure S4). The letter p indicates that the deletion includes the corresponding nucleotide.

The three (n−1) impurity peaks represent a total of 1422 impurities that resulted from omitting one of the four different nucleotide building blocks during the 18 synthesis cycles. For example, the n-p(MOE G) peak consists of two different structures with the same mass. Compared with the parent compound (Scheme 3), one structure lacks the fifth and the other lacks the first or second nucleotide from the 3′-end, since deletion of the first or second MOE G results in the same structure. Due to one mass unit difference, the n-p(MOE MeC) and n-p(MOE T) peaks are not resolved, and their sum is reported. These results show that two syntheses result in products with little difference in the levels of (n−1) impurities arising from the deletion of MOE G and MOE A, although the capping step was designed to reduce these impurities. The peak for n-p(MOE MeC) + p(MOE T) is the sum of nine different (n−1) species due to the deletion of one of the seven MOE T and four MOE MeC nucleotides. The slight increase 11580

DOI: 10.1021/acs.joc.8b01553 J. Org. Chem. 2018, 83, 11577−11585

Article

The Journal of Organic Chemistry

Figure 4. Overlays of the (n−1) impurities of 10 synthesized using the 3- and 4-reaction process, in solid and dotted traces, respectively, using 6 (purple), 7 (navy blue), and 8 (black).

Scheme 4. Potential in Situ Capping Reactions of Failure Sequences in the Sulfurization Step

Scheme 5. Structure of PS/PO Mixed Backbone Oligonucleotide 15

synthesis of PO/PS mixed backbone oligonucleotides with four or less PO linkages. Due to reduction of the solvent usage by roughly 25% and the increase of the yield, the E-factor is improved significantly. For example, the synthesis of 1 kg of a typical 20-mer oligonucleotide drug candidate consumes ∼2000 L of CH3CN and PhCH3 and generates the same amount of organic waste. This process improvement reduces the E-factor to ∼1300 from ∼2000 (Table 2). To provide further support of the role of the PADS byproduct in capping, the waste stream of the PADS sulfurization step generated during the solid-phase synthesis of 15 was collected and mixed with n-BuOH at 22 °C. The esterification product PhCH2CO2Bu was isolated, and its structure was confirmed by NMR (Scheme 6). This result indicates that a reaction byproduct with a structure PhCH2CO-Y (the structure of Y is unclear) is generated in PADS sulfurization and can convert the 5′−OH of the failure sequences into phenyl acetate. In addition, PhCH2CONH2, a

present. In accord with these expectations, the results (Figure 5) show that (n−1) impurities arising from PS linkages are nearly the same for both the 3- and 4-reaction synthesis processes; however, (n−1) species, n-p(MOE A)PO, np(MOE G)PO, and n-p(MOE MeC)PO, arising from PO linkages are higher in the 3-reaction synthesis (see the SI Figures S17−S18 for full UHPLC-UV-MS chromatograms and MS spectra). Multiple runs of the synthesis of 15 confirmed this trend. These results are consistent with the high level of total (n−1) impurities (>10%) obtained from a synthesis of a full PO 20-mer carried out without the Ac2O capping step (results not shown) and the role of the capping step in full PO oligonucleotide synthesis. From lab scale results, the 3-reaction cycle process on average affords ∼9% higher yield and the same or slightly higher overall purity of 15 than the 4-reaction cycle process. Due to the yield increase and improvement of overall purity, this 3-reaction cycle process is also attractive at least for the 11581

DOI: 10.1021/acs.joc.8b01553 J. Org. Chem. 2018, 83, 11577−11585

Article

The Journal of Organic Chemistry

NMR (SI) from a larger scale reaction carried out under the same conditions. The sulfurization went to completion in