Artificial Distinctions: Protein A Mimetic Ligands for Bioprocess Separations

Protein A is a 47-kDa defensive weapon displayed on the cell membrane of the Gram-positive pathogen, Staphylococcus aureus. Protein A selectively binds human immunoglobulin IgG by its Fc region (the descending tail of the Y-shaped antibody), effectively holding the antibody "upside down," preventing recognition by phagocytes. Joined to molecular spacers and bound to chromatographic supports, a milliliter of Protein A sorbent can bind 30 mg or more of human polyclonal IgG, making it the standard tool for separating antibodies (and other high-value proteins) from recombinant cell-culture broths, blood plasma, milk, and other complex liquids.

But at $8000–$10,000 per liter, Protein A is expensive, and it imposes operational requirements that chafe process developers charged with handling higher and higher volume processes at lower and lower costs. The appeal of a workable alternative is clear.

"The largest columns I know of are 300 L," notes one separation scientist. "At $10,000 per liter, that would be $3 million for one fill, enough to purify 15 or 20 fermentor runs. This is big money and a big part of your cost of goods."

In addition, Protein A is toxic, and it can leach into the process stream. So process developers must follow the Protein A separation with a second process step—typically cationic-exchange chromatography—to remove any fugitive ligand. And then, they must add an analytical step to confirm that no Protein A remains in the process stream.

Though recent developments have produced recombinant Protein A columns that can tolerate washing with up to 0.1 M sodium hydroxide, the columns cannot withstand repeated cleaning with industry-preferred 0.5–1 M sodium hydroxide. They require, instead, longer cleaning cycles with milder, and sometimes more expensive, cleaning agents.

And finally, Protein A binds some targets so avidly that they can be eluted only under highly acidic conditions, which can degrade antibody activity or promote aggregation.

These factors all add up to significant costs—for direct expenses, prolonged run times, and lost product—amounting to 50% or more of the total production cost (1).

Counterbalancing these limitations, Protein A is widely available from several sources. It is well understood. And it is well accepted by world regulatory agencies.

"Protein A is the benchmark, and I don't see it changing in the short run," the separation scientist observes.

Jerold M. Martin, senior vide-president and global technical director of Pall Life Sciences (East Hills, NY, www.pall.com), sees strong motivation for change, though. "As you increase the market demand and as competitive monoclonals start to come into place—we're already seeing development of biogeneric or biosimilar monoclonal antibodies—the whole market is going to shift and the cost of manufacturing is going to become quite critical."

Synthetic immunoselective ligands

Today's Protein A mimetics are relatively new products with roots in the 1980s, at the dawn of rational drug design when many in the field blithely assumed that by 2005, molecular engineers would be specifying and building custom ligands as casually as design engineers specify and stamp out plastic gears. In fact, though, the market now supports only a handful of synthetic alternatives to Protein A. These include:

• 4-mercaptoethyl pyridine (MEP, marketed as MEP HyperCel), with a molecular weight of 139 and a dynamic binding capacity of ≥20 mg human polyclonal IgG/mL sorbent. MEP operates by hydrophobic charge induction chromatography (Pall);

• 2-mercapto-5-benzymidazole sulfonic acid (MBI, marketed as MBI HyperCel) with a molecular weight of 194 and rated at 20–40 mg human polyclonal IgG/mL of sorbent (Pall);

• The triazine derivatives (marketed as MAbsorbent A1P, A2P, and A3P) were developed to mimic the Phe-132, Tyr-133 dipeptide binding site in the hydrophobic core structure of Protein A, from a library of IgG binding ligands built with aromatic amines on a triazine scaffold. These sorbents bind up to 50 mg IgG/mL sorbent (ProMetic Life Sciences Inc., Montreal, Quebec, Canada, www.prometic.com).

The synthetic sorbents cost $2700–$3000 per liter, about a third of Protein A's cost. But proponents say that even greater savings come from other factors: The synthetics have rated service lives of at least 200 cycles of binding, eluting, and cleaning. They tolerate cleaning with 1 M sodium hydroxide. And they operate over a wide range of pH (pH 1.5–14 for the triazine, pH 3–12 for 4-MEP). Finally, synthetic ligands do not leach into the process stream, obviating both the secondary ion-exchange step and the confirming assay.

The drawback of synthetic Protein A ligands is that they are, in pharmaceutical years, very young, and have yet to be part of an approved manufacturing process.

Learning from dyes

ProMetic Life Sciences grew out of Affinity Chromatography Ltd, founded by a group of Cambridge University scientists in 1988 to build on their knowledge of dye chemistry. The company's first Drug Master File number (for Mimetic Blue, used to separate mammalian serum albumin from plasma) came in 1992. Other products, many for blood products separations, followed, based on the same library of functional moieties and a combination of rational design techniques and sequential optimization that ProMetic uses to produce custom mimetic ligands, according to the company's president, Pierre Laurin. All together, these products make up what Laurin calls ProMetic's cascade concept—a range of off-the-shelf synthetic ligands that provide highly specific functions, designed to be applied in series for product capture and impurity removal.

In 2002, the company entered the Protein A market with MAbsorbent A2P. Though the product "worked perfectly well" in the core blood products and related markets, separating human immunoglobulin from blood products, for example, or purifying high-value proteins from milk, Laurin says the company erred in the monoclonal antibody market by positioning the product as a unit-for-unit replacement for Protein A. As it turns out, A2P also binds strongly to pluronic acid, a surfactant added to protect suspended cultures from shear damage. To prevent competitive binding, process developers would have to remove pluronic acid with an ion-exchange step before the affinity column. Though the bioprocessors were accustomed to inserting an ion-exchange step after the affinity column (to remove any leached Protein A), they proved reluctant to shuffle steps around, Laurin says.

Applying its design strategy, however, ProMetic is following up with A3P, a true Protein A mimetic—cured of its pluronic acid habit—due later this year, says Laurin, which will offer unit-for-unit replacement replacement.

Nitrogen and sulfur

The MEP ligand arose from Genencor research licensed to BioSepra (acquired by Pall in 2004). The studies correlated immunoglobulin selectivity with a nitrogen heterocycle and a proximal thioether linkage.

BioSepra introduced MEP early in 2000 and received a US Food and Drug Administration Drug Master File number in May 2001. (A Drug Master File submission is in preparation for the MBI sorbent.) MEP is currently used in ten processes to produce clinical trial materials for Phase I and Phase II studies, half of which are for antibody products, and half are for other types of proteins, using the sorbent as an alternative to conventional hydrophobic interaction chromatography (HIC).

HIC probably ranks second to ion-exchange chromatography in general protein purification, says Warren E. Schwartz, senior technical director at Pall Life Sciences. "But the speed bump in traditional HIC is that you've got to use lots of salt to achieve binding, and when it's time to prompt desorption, you ramp down the salt concentration. The salt is expensive to buy, and it can be expensive to dispose of."

In classical HIC, Schwartz says, the target protein is typically eluted in a 0.2–0.5-M salt solution. "So now you're stuck: before you go on to an ion-exchange step, you have to do a diafiltration or a heavy dilution to get rid of the residual salt. That's expensive."

When used in HIC mode, MEP requires much lower salt concentrations than do classical HIC sorbents, Schwartz emphasizes. And in antibody purifications, binding is typically achieved without any added salt. Elution is achieved by dropping pH from neutral to acidic (about pH 4–5.5), producing positive charges on both the ligand and the target protein, charges strong enough to drive bound proteins off of the sorbent (Figure 1) in buffer containing little or no residual salt, he says.

Like Protein A, the synthetic sorbents bind preferentially to the antibody's Fc fragment. Unlike Protein A, though, MEP and MBI bind much more strongly to aggregated or misfolded antibody than to the properly folded monomer. This differential binding lets process designers capture antibody and remove aggregates in a single chromatographic step.

Figure 1: 4-mercaptoethyl pyridine (4-MEP) binds to human IgG through a combination of molecular recognition and hydrophobic interaction. Lowering the pH induces positive charges in the ligand and the antibody, thereby driving it off of the sorbent.

Researchers at Hematech (Sioux Falls, SD, www.hematech.com—acquired by Kirin in July 2005), for example, had difficulty separating monomeric and aggregated polyclonal antibodies from bovine plasma. Using MEP sorbents, they found that they could elute properly folded monomeric antibodies at a relatively mild pH 4.4 and then flush out aggregates at pH 3.0 (3, 4).

The day of protein mimetic separation ligands has not quite broken, but it may not be far off.

"As far as market penetration goes, it's too soon to make definitive statements," says Schwartz. "We know we're in evaluation pretty broadly, and we know that this is a conservative industry. My personal prediction is that the first time MEP is identified as being in a commercial process for an approved pharmaceutical, there is going to be a sea-change fueled by the people who have learned to use the technique in the laboratory and then studied it further in process development."

Figure 2: With InstrActions "polymer instruction" method, a sample of the target molecule serves as the template for assembling binding moieties to the sides of a polymer mesh.

References

1. D. McCormick, "Bioseparations Look Ahead to the Past," Pharm. Technol. 28 (7), 36–44, (2005).

2. G.T. Weatherly et al., "Initial Purification of Recombinant Botulinum Neurotoxin Fragments for Pharmaceutical Production Using Hydrophobic Charge Induction Chromatography," J. Chromatogr. A, 952, 99–110 (2002).

3. W. Schwartz et al., "Isolation and Purification of Antibodies on Immunoglobulin-Selective Sorbents," paper presented at the IBC Antibody Production and Downstream Processing Conference, San Diego, CA, Feb. 24–27, 2004.

4. W Schwartz et al., "Application of Chemically Stable Immunoglobulin-Selective Sorbents: Capture and Purification of Antibodies with Resolution of Aggregate," BioProcessing J. 3 (6), 53–62 (2004). PT