Scalable Membrane Ion-Exchange Chromatography Purification of an Antisense Oligonucleotide

Rising drug costs have increased public pressure on the biopharmaceutical industry to find ways to identify and eliminate high-cost unit operations. Biopharmaceutical manufacturing groups now routinely evaluate both productivity as well as economic feasibility for every process step.

Significant strides in the development of more-efficient upstream processes have engendered research into developing downstream processing media that provide high throughputs (e.g., efficient capture of target molecules from large volumes of feedstock at fast flow rates with high dynamic binding capacities). Although conventional beaded chromatography media have served the biopharmaceutical industry admirably for several decades in various applications, these media have several limitations such as high pressure drops, slow diffusion limited binding of the target molecule into the sorbent pores, low dynamic binding capacity for large molecules such as plasmid DNA and viruses, and slow operational flow rates that result in long processing times for capture from large feedstocks (1). Biopharmaceutical process developers and manufacturers have accepted these drawbacks and have accounted for them in their processes for more than half a century. New, highly productive and cost-effective purification technologies, however, are now emerging as potential alternatives. Membrane chromatography is one such alternative that has been reviewed for various applications (2–5).

Examples of crude antisense oligonucleotide (AO) purification by sample self-displacement using anion exchange on beaded media column have been reported from the milligram to the 100-g scale (6, 7). Gram-scale oligonucleotide purification on strong anion-exchange membranes has been reported with product purity of 90%, similar to what has been reported for conventional beaded media chromatography (8).

Although membrane chromatography in a stacked-disk format provides additional high-resolution purification of a process stream derived from a reversed-phase liquid chromatography fraction containing a partially purified AO (9), this membrane format has limited scalability. Hence, the focus of this article is on developing an oligonucleotide purification method on a pleated membrane format that is scalable to a manufacturing level.

Oligonucleotides are synthesized from the 3'- to 5'- end of a nucleoside. The terminal nucleoside is capped at the 5'- end with 4,4'-dimethoxytrityl (DMT) protecting group. If the downstream purification involves anion-exchange chromatography, then the DMT group is cleaved at the end of the synthesis. This is called the DMT-off oligonucleotide (see Figure 1).

Figure 1: Structure of a phosphorothioate linkage. When X 5 DMT, the oligonucleotide is called a DMT-on oligonucleotide, and when X 5 H, it is called a DMT-off oligonucleotide. "B" represents adenine (A), cytosine (C), guanine (G), or thymine (T).

Data presented in this article indicate that the purification of a crude DMT-off AO on a strong anion-exchange membrane capsule results in product purity that is higher than that obtained from beaded chromatography media with comparable recovery (7). The membranes in a capsule are in pleated format that is scalable to manufacturing scale.

Figure 2: A schematic representation of the diffusive pores in beaded chroma-tography and convective pores in membrane chromatography.

Pleated ion-exchange membranes possess large convective pores, and the dynamic binding capacity is relatively independent of flow rates (see Figures 2 and 3). Thus, large volumes of feedstock containing large biomolecules such as plasmids and viruses can be captured efficiently with high dynamic binding capacity at high flow rates.

Figure 3: Membrane pore structure as captured by scanning electron micrograph.

In this study, the target molecule was a DMT-off 20-mer phosphorothioate deoxyoligonucleotides. A phosphorothioate linkage consists of a phosphodiester with the nonbridging oxygen replaced by sulfur (10). The feedstock was synthesized by solid-phase synthesis followed by cleavage of the DMT protecting group and cleavage of the oligonucleotide from the solid support (11). Following evaporation of ammonia, the oligonucleotide was lyophilized. High-performance liquid chromatography (HPLC) analysis of the lyophilized oligonucleotide showed 70% purity in the 20-mer phosphorothioate oligonucleotide by reversed-phase ion-pair (RP–IP) HPLC analysis (see Figure 4).

Table I: Recovery and purity of a 20-mer antisense oligonucleotide purification on a 10-mL membrane capsule.

The impurities to be removed consisted of shorter failure sequences of the target phosphorothioate, namely, (n – x) in which x = 1, 2, ... 19, also known as (n – 1), (n – 2), ... failure sequences that are formed because of the incomplete addition of each nucleotide during synthesis. In addition, phosphodiester by-products of the shorter failure sequences (P = O)_x are formed as a result of oxidation of phosphorothioate bond during synthesis. The target molecule is known as the full-length oligonucleotide or n-mer, which in this case is a 20-mer oligonucleotide. Oligonucleotides are polyanionic with the number of charges on the molecule depending on the number of nucleotides. Therefore, oligonucleotides can be purified by anion-exchange chromatography.

Figure 4: Analytical RP–IP HPLC chromatogram of a crude DMT-off 20-mer AO.

Sample self-displacement chromatography is a useful method that has been previously reviewed (12, 13). In sample self-displacement chromatography of phosphorothioate oligonucleotides, the full-length, fully thioated oligonucleotide that has the highest affinity for the stationary phase acts as the displacer. This article describes an example of oligonucleotide capture and purification on anion-exchange membrane chromatography media under sample self-displacement conditions that provide high purity (>99.0% by RP–IP HPLC) and yields (83%) that are comparable to those cited in the literature using a displacer (14, 15). Nonetheless, the advantage of more than 10-fold faster flow rates (≥10 membrane volume [MV] per minute) that is ≥100 mL/min at high dynamic binding capacity (19 mg/mL) considerably improves the throughput over beaded anion-exchange chromatography media.

Application protocol

Process development on a 10-mL membrane capsule.

Figure 5 shows purification on a 10 mL membrane volume Mustang Q strong anion-exchange membrane capsule (Pall Life Sciences). The 190-mg sample was dissolved in 19 mL of 20 mM NaOH and filtered through a 0.2-μm Supor membrane syringe filter.

Figure 5: Purification of 190 mg DMT-off 20-mer antisense oligonucleotide on a 10-mL bed volume membrane capsule.

The capsule was connected to a peristaltic pump that could deliver at least 10-MV/min (100 mL/min) flow rates using sanitary fittings and platinum-cured silicone tubings. The data collection system involved a computer-controlled UV–vis detector to monitor absorbance at a specific optical wavelength.

The sample was crude 20-mer DMT-off phosphorothioate oligonucleotide (a gift from ISIS Pharmaceuticals, Carlsbad, CA). The sample load was 190 mg in 19 mL of mobile phase A. The stationary phase was 10 mL of strong anion–exchange membrane capsule. Mobile phase A was 20 mM NaOH, and mobile phase B was 2.5 M NaCl in A. The flow rate was 95 mL/min (9.5 MV/min), and the step gradient was 40 MV A, 45 MV of 25% B, 48 MV of 38% B, 20 MV of 55% B, and 6 MV of 100% B.

Experiment

Oligonucleotide purification was monitored by measuring at the 260-nm wavelength. Sample load and oligonucleotide recovery was calculated from absorbance of diluted samples at 260 nm using the conversion factor 25 A

₂₆₀

units = 1 mg of oligonucleotide. This is a common standard conversion factor for phosphorothioate oligonucleotides. The loading buffer and all sample dilutions were prepared in 20 mM NaOH to prevent oligonucleotide aggregation.

Purity was determined by analytical RP–IP HPLC. The percent purity was calculated as follows: (peak area of full-length product ÷ total peak area) × 100.

Recovery or yield expressed in milligrams of full-length product was determined from absorbance readings at 260 nm and was calculated as A₂₆₀ × 0.04. The percent yield or recovery was calculated as a weight-by-weight (w/w) ratio of milligrams of oligonucleotide in elution to the amount of oligonucleotide in the sample load.

Results and discussion

The conditions shown in Figure 5 were derived from a series of experiments that were performed at several sample loads using step elution. The basic strategy involved increasing the sample load until a sample self-displacement effect was observed to maximize purity and lowering the salt concentration in the low salt wash to maximize yield. The product purity in the 55% B elution pool was determined at each sample load using RP–IP HPLC. Figure 5 shows the chromatographic conditions that provided maximum product purity and yield at a given sample load.

Purification on the membrane capsule results in a target AO with a purity level >99%, as determined by analytical RP–IP HPLC (see Figure 6) with 58% yield and 83% overall yield from recycling low-salt wash fractions (see Table I). The purity was calculated from the area under the 20-mer oligonucleotide peak in the analytical RP–IP HPLC chromatogram, whereas the yield was calculated from UV absorbance at 260 nm using the conversion factor of 1 AU₂₆₀ (absorbance unit at 260 nm) = 25 mg of oligonucleotide.

Figure 6: RP-IP HPLC analysis of the 55% B pool of fractions from elution.

The flow through contains <2% of the product, whereas each of the two low-salt washes predominantly contain failure sequences and phosphodiester impurities as well as an additional 10–15% of the product. The 100% B fractions contained 6% of product. The 55% B elution pool contained 58% of the purified product in >99.0% purity by RP–IP HPLC (see Figure 6).

Recycling the flow-through fractions, 25% B and 38% B pools through the membrane chromatography capsule resulted in recovering an additional 25% of product and hence 83% overall yield with >99.0% purity by RP–IP HPLC (see Table I).

As the sample load increased, the product purity increased, suggesting a sample self-displacement effect (see Figure 7). This effect emerged at a sample load of 190 mg. Further increasing the sample load to 240 mg AO did not provide additional product purity although it may result in slightly higher throughput.

Figure 7: Effect of sample load on product purity.

In this mixed-mode chromatography in which the sample self-displacement effect is coupled with elution chromatography, as the sample load increased the product yield in elution decreased because of early displacement of product in the low-salt wash (see Figure 8). Therefore, to recover more product from the salt-wash step and yet maintain high purity in the 55% B elution at a higher sample load, the concentration of NaCl in the salt-wash step would have to be optimized to lower than 38% B.

Figure 8: Effect of sample load on product yield.

A typical oligonucleotide capture and purification operation using an optimized step elution protocol takes 20 min at a flow rate of 95 mL/min (9.5 MV/min) on a 10-mL ion-exchange membrane chromatography capsule (see Figure 5). Because the capture and purification properties of membranes are independent of the flow rates, a flow rate as fast as 400 mL/min (40 MV/min) could be used to reduce the cycle time and significantly increase throughput. Thus, the capture and elute protocol is nine-fold faster than a three-hour purification run on a conventional anion-exchange beaded media chromatography column (7), with the potential for gaining further advantage by increasing the flow rate on the membrane chromatography capsule. These data indicate that, in addition to providing faster oligonucleotide processing speeds, ion-exchange membranes provide comparable or slightly higher purity (99% by RP-IP HPLC) than beaded anion-exchange chromatography media (98% by anion-exchange chromatography) as well as higher yield (83% versus 75%)(7).

Demand for driving down the membrane cost and the need for manufacturing-scale membrane chromatography units to justify their use in commercial capture applications has spurred further ongoing improvements that address these process issues. Membrane chromatography products are typically used in disposable formats in polishing and clearance applications. Reusing disposable membrane chromatography capsules can reduce the process cost for a capture application. The protocol described in this article can be scaled-up by making minor changes to a reusable membrane chromatography unit that can capture 0.5–1.0 kg of a DMT-off oligonucleotide from a solid-phase oligonucleotide synthesis process.

As oligonucleotides hydrolyze under weakly acidic conditions, a mild clean-in-place method can be used (7). This involves running a 4-mM phosphoric acid solution at a low flow rate (3 or 4 MV/min) for 30 min followed by pumping 20 mM NaOH in 2.5 M NaCl for 30 min at the same flow rate, and finally equilibrating with the loading buffer before starting the next purification cycle. The membrane capsule was stored in 1 M NaCl at 4 °C for as long as one year with several uses within that period.

The purified product can be concentrated (ultrafiltration) and desalted (diafiltration) using tangential-flow filtration systems fitted with an ultrafiltration membrane cassette of 650 Da molecular weight cut-off. The principles of concentration and the desalting of biomolecules by tangential-flow filtration have been previously presented by Schwartz (16). Process development of ultrafiltration and diafiltration and its scale-up on a 100-L scale that was performed on a 20-mer phosphorothioate oligonucleotide also has been described in the literature (9).

Conclusion

This article demonstrates that when the sample loading and elution conditions are carefully optimized, a strong anion-exchange membrane chromatography unit in a scalable, pleated format can provide good resolution and yields within a short processing time.

Acknowledgments

The author thanks ISIS Pharmaceuticals for providing the DMT-off oligonucleotide and the analytical RP–IP HPLC method, and Dr. Yogesh Sanghvi of Rasayan, Inc., for helpful discussions. Mustang Q and Supor are a registered trademarks of Pall Corporation.

Ajay R. Lajmi, PhD, is a senior research scientist at the Biopurification Laboratory of Pall Life Sciences, 8780 Ely Road, Pensacola, FL 32514, tel. 850.316.3504, fax 850.316.3601, ajay_lajmi@pall.com.

References

1. T.N. Warner and S. Nochumson, "Rethinking the Economics of Chromatography,"

BioPharm Int.

1,

58–60 (2003).

2. R. Ghose, "Protein Separation Using Membrane Chromatography: Opportunities and Challenges," J. Chromatogr. A. 952, 13–27 (2002).

3. S. Zhang, A. Krivosheyeva, and S. Nochumson, "Large-Scale Capture and Partial Purification of Plasmid DNA Using Anion-Exchange Membrane Capsules," Biotechnol. Appl. Biochem. 37, 245–249 (2003).

4. A.M. Davidoff et al., "Purification of Recombinant Adeno-Associated Virus Type 8 Vectors by Ion Exchange Chromatography Generates Clinical Grade Vector Stock," J. Virol. Methods 121, 209–215 (2004).

5. K.H. Gebauer, J. Thommes, and M.R. Kula, "Plasma Protein Fractionation with Advanced Membrane Adsorbers," Biotech. Bioeng. 53, 1–13 (1997).

6. R.R. Deshmukh, W.E. Leitch II, and D.L. Cole, "Application of Sample Displacement Techniques to the Purification of Synthetic Oligonucleotides and Nucleic Acids: A Mini Review with Experimental Results," J. Chromatogr.A 806, 77–92 (1998).

7. R.R. Deshmukh et al., "A Case Study: Oligonucleotide Purification from Gram to Hundred Gram Scale," Nucleos. Nucleot. Nucl. 20, 567–576 (2001).

8. R.R. Deshmukh et al., "Large Scale Purification of Antisense Oligonucleotides by High Performance Membrane Adsorber Chromatography," J. Chromatogr. A. 890, 179–192 (2000).

9. A. Lajmi, L. Schwartz, and Y.S. Sanghvi, "Membrane Purification of an Antisense Oligonucleotide," Org. Process Res. Dev. 8, 651–657 (2004).

10. Y. S. Sanghvi, "DNA with Altered Backbones in Antisense Applications in Comprehensive Natural Products Chemistry," D.H.R Barton and K. Nakanishi, Eds., in DNA and Aspects of Molecular Biology, E.T. Kool, Ed. (Pergamon Press, New York, NY, vol. 7, 1999), pp. 258–311.

11. Y.S. Sanghvi et al., "Chemical Synthesis and Purification of Phosphorothioate Antisense Oligonucleotides," in Manual of Antisense Methodology, G. Hartman and S. Enders, Eds.( Kluwer Academic Publishers, Norwell, MA, 1999), pp. 3-23.

12. S.M. Cramer and C. Horvath, "Displacement Chromatography in Peptide Purification," Prep. Chromatogr. 1, 29–49 (1988).

13. R.S. Hodges, T.W.L. Burke, and C.T. Mant, "Preparative Purification of Peptides by Reversed-Phase Chromatography: Sample Displacement Mode versus Gradient Elution Mode," J. Chromatogr. 444, 349–368 (1988).

14 A.A. Shukla and S.M. Cramer, "Bioseparations by Displacement Chromatography," in Handbook of Bioseparations, S. Ahuja, Ed. (Academic Press, New York, NY, 2000), pp. 379–415.

15. A.A. Shukla et al., "Purification of Oligonucleotides by High Affinity, Low Molecular Weight Displacers," Biotechnol. Prog. 16, 1064–1070 (2000).

16. L. Schwartz, "Diafiltration for Desalting or Buffer Exchange," Bioprocess Int., 1 (5), 43–49 (2003).