When microbial expression is the right choice

Microbial expression wins whenever the protein does not need mammalian post-translational modification: a microbial fermentation cycle is measured in days, media are chemically defined and inexpensive, and cell densities of 100 g/L wet weight or more are achievable in fed-batch E. coli. The decision rests on four questions.

1. Does the protein need glycosylation for activity or half-life? If yes, go mammalian. If glycans are absent or dispensable, as for insulin, growth hormone, interferons, many cytokines and enzymes, microbial fermentation is the default.
2. How many disulfide bonds? Proteins with few disulfides fold in the E. coli cytoplasm or periplasm; proteins with many often form inclusion bodies, shifting the work to refolding. Pichia secretes disulfide-rich proteins through its secretory pathway and avoids refolding.
3. Is the product a protein at all? Plasmid DNA is made only by E. coli fermentation.
4. What is the dose and the market? High-dose or high-volume products, and biosimilars competing on cost, benefit most from the economics of microbial fermentation and bulk manufacturing.

ICH Q5B and CTD section 3.2.S.2.3 require the host and expression construct to be described and justified, so these answers become regulatory text.

E. coli versus Pichia pastoris

E. coli is the faster, simpler host for intracellular expression; Pichia pastoris is the better host when secretion, disulfide folding or the absence of endotoxin matters more than a short cycle. The table sets out the differences that shape microbial fermentation design.

Both hosts start from a two-tier cell bank. ICH Q5D covers microbial as well as mammalian cell substrates, so master and working cell banks are tested for identity, purity, plasmid retention and viability before cGMP use.

Fermentation scale-up: 20 L to 1,000 L

Microbial fermentation scales on oxygen transfer and heat removal, not on volume: a 50-fold step from 20 L to 1,000 L multiplies metabolic heat and oxygen demand 50-fold while cooling surface and impeller performance do not follow. Four parameters are managed deliberately.

• Oxygen transfer (kLa). High-density E. coli cultures consume oxygen at rates far above mammalian cells. At scale, kLa is held by raising power input per volume, gas flow and back pressure, and by oxygen enrichment when air cannot hold dissolved oxygen above the 20–30% set point.
• Mixing and gradients. Mixing time grows with vessel size, so fed substrate and base arrive in bursts. Feed points, feed concentration and agitation are set to prevent local glucose excess, which triggers acetate overflow metabolism in E. coli and lowers yield.
• Heat removal. Jacket area per volume falls as vessels grow, so the specific growth rate during the feed phase is often reduced at scale to keep metabolic heat within cooling capacity, notably during Pichia methanol induction.
• Feed strategy. Exponential feeding at a controlled specific growth rate, or DO-stat feeding, builds cell density before induction. The feed profile optimized in 2 L fermentors is reproduced per volume and verified at 20 L before the 1,000 L run.

Induction timing, post-induction temperature and harvest time are fixed from the 20 L runs; lowering post-induction temperature to 25–30 °C often raises soluble expression in E. coli at the cost of a longer run.

Cell disruption, inclusion body recovery and refolding

For an intracellular E. coli product, recovery after microbial fermentation is a three-step sequence, and the refolding step, when needed, sets the overall process yield.

Cell recovery and disruption. Cells are concentrated by disc-stack centrifugation, resuspended in lysis buffer, and passed through a high-pressure homogenizer, typically two passes at 600–1,000 bar. Homogenizer throughput in liters per hour is the sizing parameter: a 100 L/hr unit processes the cell mass from a 1,000 L fermentation in a few hours. Released DNA raises viscosity, so nuclease or dilution precedes clarification.

Inclusion body recovery. Inclusion bodies are denser than cell debris and are collected by centrifugation, then washed with detergent, dilute urea or salt to strip membrane fragments and host proteins. Washed inclusion bodies are often 60–90% product.

Solubilization and refolding. Inclusion bodies are dissolved in 6–8 M guanidine hydrochloride or urea with a reducing agent, then refolded by rapid dilution, stepwise dialysis or on-column exchange at a protein concentration usually below 1 mg/mL. A glutathione redox pair drives disulfide formation; arginine suppresses aggregation. Yield, measured by RP-HPLC and a functional assay, is optimized over concentration, denaturant endpoint, redox ratio, pH, temperature and time.

Soluble products go from clarified lysate directly to capture. Secreted Pichia products skip disruption entirely; the clarified supernatant is loaded directly.

Purification and endotoxin control

Purification after microbial fermentation is built around the absence of a Protein A step and the presence of endotoxin, so capture is usually ion exchange and at least one step is positioned for lipopolysaccharide removal.

A representative E. coli train runs capture on ion exchange, an intermediate hydrophobic interaction or mixed-mode step for misfolded species and host proteins, polishing anion exchange in flow-through mode, and UF/DF into formulation buffer. Pichia trains start from a cleaner feed and add a step for protease-clipped variants.

Endotoxin. Lipopolysaccharide is negatively charged and forms large aggregates, so it binds anion exchange resins at neutral pH and is retained by ultrafiltration membranes when the product is small enough to pass. Where the product is itself acidic and co-binds, endotoxin-affinity resins are used. Endotoxin is measured by the LAL or recombinant Factor C test at every pool during development, so removal is assigned to a step and validated there.

Other impurities. Host cell protein by ELISA, residual DNA by qPCR, residual refolding reagents by HPLC, and, for plasmid DNA, supercoiled content and residual RNA. Specifications follow ICH Q6B and are confirmed by the same orthogonal purity panel used for mammalian products.

Bulk filling and cGMP status

Bulk filling closes the microbial fermentation campaign by converting the formulated UF/DF pool into frozen drug substance containers under the same batch record, so that fermentation, recovery, purification and fill are released as one bulk manufacturing lot.

The pool is filtered through a 0.2 µm filter, with pre-filtration bioburden and a post-use integrity test, into single-use bags or bottles sized to the drug product batch. Containers are frozen under a validated program and stored at −40 °C or below, with shelf life supported by real-time stability data. Fill weight, filter integrity and temperature history join the release package.

GMP status is defined by scope and by inspection. ICH Q7 applies increasing GMP stringency from maintenance of the working cell bank through fermentation, recovery, purification and bulk filling, so the entire chain sits inside the GMP boundary. A site's certifications show which authorities have inspected the microbial operation and for what scope; for bulk manufacturing aimed at several markets, inspection history from more than one authority shortens pre-approval preparation.