Multi-column countercurrent solvent gradient purification (MCSGP) — turning idle batch resin into a continuously producing capture train
Protein A affinity chromatography is the workhorse capture step for essentially every monoclonal antibody and Fc-fusion process — and its resin is also the single most expensive consumable in the downstream train, often costing several thousand dollars per liter. In conventional batch operation, that resin only binds product during the load phase. For the rest of the cycle — wash, elution, strip, clean-in-place, re-equilibration — the column produces nothing while the clock and the capital keep running.
A single-column Protein A capture run moves through a fixed sequence of phases, and only one of them actually captures product:
• Equilibration: column flushed with binding buffer to set pH and ionic strength • Load: feed passed through the bed; antibody binds Protein A ligand until the bed approaches its dynamic binding capacity (DBC) • Wash 1 / Wash 2: unbound host-cell protein, DNA and media components are rinsed away • Elution: low-pH buffer disrupts the Protein A–Fc interaction; purified antibody is collected as a single pulse • Strip / regeneration: residual aggregates and tightly bound species removed • Clean-in-place (CIP) and re-equilibration: bed reset for the next load
Only the load and elution windows generate value. Wash, strip, CIP and re-equilibration are necessary but purely time-consuming — the resin sits fully loaded (during wash) or fully empty (during CIP) without adding or releasing product.
Because a batch column must load, elute and fully regenerate before it can load again, roughly 60–70% of every cycle is spent on steps that consume time and buffer but produce zero purified antibody — the fundamental inefficiency that continuous multi-column capture is designed to eliminate.
Batch loading is also self-limiting: as the resin bed approaches saturation, the breakthrough curve rises and unbound antibody starts to appear in the column effluent. To avoid losing product, operators stop loading well before the bed reaches its static binding capacity — typically at 1–5% breakthrough — leaving usable capacity on the table.
The discarded flowthrough at the tail of the load, and the leading/trailing edges of the elution peak that fall outside the purity cut, represent product that was made, captured, and then thrown away purely because a single column has nowhere else to send it. This lost fraction is exactly what countercurrent multi-column operation recovers (Stage 3).
The core idea of continuous capture is deceptively simple: replace one large column with several smaller columns and stagger their cycles. While column 1 loads, column 2 washes, column 3 elutes and column 4 regenerates — then the roles rotate. At every instant, at least one column is doing each job, so the system as a whole is never idle even though any individual column still passes through the same wash/elute/regen phases.
Continuous multi-column capture is generally implemented as Periodic Countercurrent Chromatography (PCC) for simple bind-and-elute steps, or the more elaborate Multi-Column Countercurrent Solvent Gradient Purification (MCSGP) when overlapping impurity fractions must be recycled for polishing separations. In both schemes:
• N columns (commonly 3–6) are connected through a valve manifold to a single automated skid • A master timer divides the total cycle into N equal "shift" intervals • At each shift, every column advances one phase and the feed/buffer routing rotates to the next column in sequence • A smaller column means a shorter individual load, so shifts happen frequently and the aggregate output smooths into a near-continuous stream
Because each column is much smaller than the equivalent batch column, its own binding capacity is reached faster, and because there are always other columns finishing wash or elution, the shared skid hardware, pumps and buffer tanks stay in constant use — the same total resin volume now processes far more feed per day.
A second, less obvious benefit compounds the scheduling gain: smaller columns loaded at higher linear velocity exhibit sharper breakthrough curves relative to their bed length, so operators can safely load closer to full dynamic binding capacity without excessive product loss to breakthrough. Combined with the elimination of idle phases, this is why reported resin utilization in multi-column systems commonly reaches 80–95%, versus 30–40% for a single batch column of equivalent total resin volume.
MCSGP's defining innovation over simpler periodic countercurrent schemes is what it does with the "grey zone" fractions — the overlapping regions at the start and end of elution, and the late breakthrough at the end of loading, that are neither pure product nor pure waste. Instead of discarding them, the system routes these impure but product-containing streams countercurrently onto the next column in the rotation, where they are re-adsorbed and re-purified in the following cycle.
Two overlap streams are recycled countercurrently:
• Breakthrough overlap: as a column approaches saturation near the end of its load phase, low concentrations of unbound (but perfectly good) product begin to appear in the effluent. Rather than sending this to waste, MCSGP diverts it directly onto the next column in the sequence, which is just beginning its own load phase and has fresh capacity to capture it.
• Elution overlap: the leading and trailing edges of the elution peak are often below the purity specification for direct pooling — early eluting weakly-bound impurities and late-eluting aggregates flank a pure product-rich core. Instead of narrowing the pooled fraction (and discarding the flanks), MCSGP re-loads these flanking fractions onto the next column for another pass through binding, washing and elution, progressively enriching them each cycle.
Because the columns are interconnected in a loop, this reprocessing happens automatically as part of the normal rotation — no separate recycle tank or manual re-injection step is required.
Every gram of "impure" product that a batch column would flush to waste, MCSGP instead pipes onto the next column's inlet — converting a purity-vs-yield trade-off that batch chromatography accepts as unavoidable into a scheduling problem the multi-column system solves automatically, cycle after cycle.
Increasing the overlap fraction recovers more product, but every recycled milliliter also consumes column capacity that could otherwise process fresh feed, and repeated re-adsorption of near-boundary material can gradually let a small amount of impurity through with it. Process development therefore optimizes overlap against three coupled objectives:
• Yield: rises steeply with overlap at first, then plateaus once nearly all boundary product is being recovered • Purity: falls slowly and predictably as overlap grows past ~25–30%, requiring re-validated pooling criteria • Productivity: recycled volume competes with fresh feed for column time, so very high overlap settings can modestly reduce throughput even as yield improves
Most validated MCSGP platforms for monoclonal antibody capture settle on 15–25% overlap as the sweet spot balancing all three.
After the first one or two full rotations of the carousel — the startup transient, during which columns are still filling their first cycles and no recycle streams yet exist to reprocess — the system settles into a repeating steady state. From that point on, every column sees an identical sequence of events each rotation, and the aggregate output at the skid outlet becomes a continuous, near-constant stream of purified product rather than the sharp pulses characteristic of batch elution.
In batch mode, the entire purified antibody titer for a cycle appears as a single sharp elution peak every 2–4 hours — a large pulse followed by a long silent stretch while wash, strip and CIP run. Downstream unit operations (viral inactivation, polishing chromatography, UF/DF) must be sized to absorb that peak, or a surge tank is needed to smooth it out.
In a staggered multi-column carousel at steady state, a different column reaches its elution phase every "shift" interval (roughly cycle time ÷ number of columns), so purified product exits the skid in a train of closely spaced pulses that — viewed on the timescale of the overall process — behaves like a quasi-continuous stream. This is what allows continuous capture to be directly coupled to continuous or semi-continuous downstream steps without large intermediate hold tanks.
Continuous multi-column capture is most powerful when paired with perfusion cell culture upstream, which itself continuously harvests cell-free broth at roughly constant titer rather than delivering one large batch harvest at the end of a fed-batch run. The two continuous unit operations connect directly:
• Perfusion bioreactor: constant-density culture with continuous cell-free harvest via ATF/TFF, feeding a surge or hold vessel sized in hours, not days • Continuous multi-column capture (MCSGP/PCC): consumes that harvest stream directly, producing a quasi-continuous purified eluate • Continuous or semi-continuous polishing (e.g., single-pass tangential flow, connected ion-exchange/CEX-AEX flow-through, viral filtration): finishes the process without ever pooling a full-batch intermediate
The result is described in the literature as an "end-to-end continuous bioprocessing train": smaller equipment footprint throughout, reduced intermediate hold volumes, tighter process control (steady-state operating points instead of batch-to-batch drift), and the ability to run a single campaign for days to weeks rather than restarting a fresh batch cycle every few hours.
Pulling the four stages together: eliminating idle time, loading closer to full dynamic binding capacity, and recycling countercurrent overlap fractions compound into substantial, repeatedly demonstrated gains in resin utilization, volumetric productivity and product yield relative to conventional single-column batch Protein A capture — without changing the chemistry of the resin itself.
The overall productivity improvement is the product of several distinct, additive effects rather than one single mechanism:
• Eliminated idle time (Stage 2): raising the fraction of time resin spends actively binding or eluting product from ~35% to ~90% alone contributes roughly a 2–2.5× productivity gain • Higher achievable loading (Stage 2): smaller, higher-velocity columns support loading closer to full dynamic binding capacity, adding a further 10–20% • Overlap recycling (Stage 3): recovering product that batch mode discards as waste or narrow-cut elution flanks adds 15–30% yield without any change in resin volume or feed titer
Multiplied together, these effects explain the 2–4× volumetric productivity figures (grams of purified product per liter of resin per day) consistently reported for MCSGP and PCC platforms across CHO-derived monoclonal antibody processes.
Because the same resin now produces 2–4× more purified antibody per liter per day, a continuous capture skid can match the output of a much larger batch system using 50–70% less resin volume and a proportionally smaller footprint — directly reducing both capital cost (smaller columns, skids, buffer tanks) and the single largest recurring consumable cost in downstream purification.
These gains come with added complexity: multi-column skids require more valves, more sensors, and tighter automation and process-control validation than a single batch column, and technology transfer / regulatory filing packages must document the countercurrent recycle logic explicitly. For high-titer, high-volume commercial monoclonal antibody manufacturing — where resin cost and facility footprint dominate the economics — these trade-offs are increasingly accepted, and continuous or semi-continuous Protein A capture is now offered as a standard configuration by major chromatography skid vendors and is in use at commercial biomanufacturing scale.