Property Comparison
IgG Antibody (Target)
Endotoxin / LPS (Impurity)
Why This Separation Works
Endotoxin (LPS) removal exploits two key differences from the protein: charge and hydrophobicity. LPS carries multiple negative phosphate groups from its lipid A backbone, while IgG near its pI is nearly neutral or positive at pH 7–7.5. AEX in flow-through mode acts as a “negative selection” — the protein is the species that does NOT bind:
| Component | Charge at pH 7 | AEX Q Resin | Log Reduction |
|---|---|---|---|
| IgG antibody | Near-neutral to + | Passes through (product) | — |
| LPS / Endotoxin | Highly negative | Binds quantitatively | 2–3 log / step |
| DNA / Nucleic acids | Negative (phosphate backbone) | Binds | 2–4 log |
| Host cell proteins (acidic) | Negative | Binds (partial) | 1–2 log |
Each independent AEX step provides 2–3 log endotoxin reduction. A 3-step strategy (depth filtration + AEX + affinity) can achieve the overall >5 log reduction needed to go from typical bacterial lysate levels (10,000–100,000 EU/mL) to <0.1 EU/mg specification.
Recommended Process Route
Depth Filtration — Charged Media
Pass clarified broth or cell lysate through a positively charged depth filter (Millipore DOHC, 3M Zeta Plus, or Sartorius X0HC). The positively charged cellulose/diatomite matrix binds negatively charged LPS aggregates and cell debris by both size exclusion and electrostatic adsorption. Typical LPS reduction: 1–2 logs. Protein recovery: >95%. Process at pH 7–8 to maximize charge difference.
Clarification + endotoxin reductionAnion Exchange — Flow-Through Mode (Q Resin)
Load the depth-filtered protein solution onto Q Sepharose, Mustang Q membrane, or Q Ceramic HyperD resin equilibrated in 50 mM Tris, 50–100 mM NaCl, pH 7.0–7.5. IgG (pI ~8.5) passes through in the flow-through fraction; LPS, residual DNA, and acidic host cell proteins bind to the quaternary ammonium groups. Load capacity: 10–40 mg IgG/mL resin in flow-through mode (much higher than bind-and-elute). LPS reduction: 2–3 logs per step. Protein recovery: >97%.
Key separation stepAffinity Polishing — Polymyxin B or EndoTrap
For final polishing below 0.1 EU/mg, use a specific endotoxin-binding affinity resin. Polymyxin B agarose (e.g., Detoxi-Gel, Pierce) binds the lipid A component of LPS via electrostatic and hydrophobic interactions. Load protein at pH 7, endotoxin binds, protein passes through. Alternatively, use EndoTrap Red (recombinant peptide-based) or Acticlean Etox (charge-based synthetic ligand). Capacity: 50,000–500,000 EU/mL resin. LPS reduction: 2–4 logs.
Polishing0.22 μm Sterile Filtration + LAL Testing
Sterile filter the final protein solution through a 0.22 μm PES or PVDF membrane. This removes any large LPS aggregates/micelles and bioburden. Test endotoxin by Limulus Amebocyte Lysate (LAL) kinetic turbidimetric or recombinant Factor C (rFC) assay. Release specification for IV drugs: <0.1 EU/mg protein or <0.5 EU/mL solution. Record all endotoxin test results for batch record and regulatory submission.
Final productExpected Results
Regulatory limits: FDA guidance requires <0.2 EU/mL for non-intrathecal IV products, <0.1 EU/mg for most protein drugs. EMA Annex 1 mandates validated endotoxin removal with documented log reduction values (LRV) in the regulatory submission.
Alternative Techniques
| Technique | Feasibility | Notes |
|---|---|---|
| Ultrafiltration (UF) | Poor | LPS monomers (1–10 kDa) are smaller than IgG (150 kDa), but LPS forms large micelles (>100 kDa aggregates) at concentrations above its CMC (~1 ng/mL). UF gives highly variable LPS rejection depending on aggregation state, ionic strength, and temperature. Not reliable for regulatory purposes. |
| Two-Phase Aqueous Extraction | Moderate | PEG/dextran or PEG/salt systems can partition LPS (amphiphilic) away from hydrophilic proteins. Useful for early-stage crude material but difficult to scale and validate for GMP applications. |
| Activated Carbon Treatment | Moderate | Granular or powdered activated carbon adsorbs LPS and pyrogens non-specifically. Useful for small molecules and simple proteins. Not recommended for large proteins (risk of protein adsorption and denaturation). |
| NaOH CIP (Depyrogenation) | Excellent | 0.5–1 M NaOH destroys LPS by hydrolysis. Used for equipment and column depyrogenation, not for protein-containing solutions. Essential part of GMP cleaning validation to eliminate endotoxin from process equipment between batches. |
| Protein A Affinity | Good | For IgG specifically, Protein A captures antibody selectively from the load while LPS and most impurities pass through. Combined with the low pH elution (pH 2.5–3.5), provides both endotoxin removal and high antibody purity in one step. Standard first capture step in mAb platforms. |
Frequently Asked Questions
Why are endotoxins so dangerous and what is the regulatory limit?
Endotoxins (lipopolysaccharides, LPS) are fragments of the outer membrane of Gram-negative bacteria. When injected intravenously, as little as 1 ng/kg body weight triggers a cascade of immune activation via TLR4 receptors, causing fever, systemic inflammation, septic shock, and potentially death. The FDA requires <0.2 EU/mL for most IV products and <0.1 EU/mg for protein drugs. One Endotoxin Unit (EU) corresponds to approximately 0.1 ng of LPS from E. coli O55:B5 (the LAL assay reference standard). A typical bacterial fermentation harvest may contain 10,000–1,000,000 EU/mL, requiring 5–7 log reduction.
Why use AEX in flow-through mode rather than bind-and-elute for endotoxin removal?
In bind-and-elute mode, the protein binds to the resin and endotoxin would need to be separated in the elution profile. This is difficult because LPS and many protein contaminants elute in overlapping salt gradient fractions. Flow-through mode is more elegant: the protein is the flow-through species (doesn’t bind at operating pH) while LPS (highly negative) binds quantitatively. This provides very high protein recovery (>97%) and very high LPS rejection (>99.9%) simultaneously. The AEX resin is then stripped of endotoxin with high salt + NaOH and reused.
How do I choose between Q Sepharose resin versus Mustang Q membrane for flow-through AEX?
Q Sepharose bead-based resin offers high binding capacity (20–40 mg protein/mL at low flow rates) and is cost-effective at scale. However, diffusion limitation in beads means lower productivity at high flow rates. Mustang Q membrane adsorbers operate at much higher flow rates (100–1,000 cm/h vs. 5–30 cm/h for beads) with comparable LPS binding, making them suitable for large-scale or time-sensitive operations. Membranes are typically single-use, eliminating cleaning validation burden. Choose membranes for high-throughput polishing; choose beads for capture steps where residence time is not limiting.
How is endotoxin measured and validated in a GMP process?
The Limulus Amebocyte Lysate (LAL) assay is the compendial method (USP <85>, EP 2.6.14). Three formats exist: gel-clot (qualitative, limit test), turbidimetric kinetic (quantitative), and chromogenic (quantitative). Modern alternatives include recombinant Factor C (rFC) assays that avoid horseshoe crab harvesting. For regulatory submissions, all endotoxin removal steps must have validated log reduction values (LRVs) with a minimum number of worst-case spiking experiments. The cumulative LRV across all steps must exceed the calculated required LRV with at least 1-log safety margin.
Related Separation Guides
Simulate This Process Yourself
Build this endotoxin removal process in untangle.bio with drag-and-drop, model the log reduction at each step, and optimize for your specific protein and regulatory target.
Open untangle.bio