Nanobody and Antibody Fragment Production: VHH, Fab, and scFv Expression Systems, Purification, and Bioprocess Development

Nanobody production has emerged as one of the fastest-growing areas of biopharmaceutical manufacturing. With over 200 VHH-based candidates in clinical pipelines and the commercial success of caplacizumab (the first approved nanobody drug), process development scientists face a practical question: which expression system and purification strategy best fits their antibody fragment format? This guide covers the bioprocess engineering of VHH nanobodies, Fab fragments, and scFv constructs, from host selection through downstream processing to manufacturing scale-up.

Unlike conventional monoclonal antibodies (~150 kDa) that require mammalian expression systems for proper folding and glycosylation, antibody fragments are small enough (12–50 kDa) to fold correctly in microbial hosts. This fundamental advantage makes nanobody production significantly cheaper and faster than full-length antibody manufacturing. However, each fragment format presents its own bioprocess challenges, particularly around disulfide bond formation and aggregation.

Antibody Fragment Formats Compared

Five antibody fragment formats dominate current pipelines, each with distinct size, disulfide bond requirements, and expression complexity. The choice of format directly determines which expression system and purification strategy is feasible.

VHH nanobodies are the simplest fragment to produce because they contain only a single immunoglobulin variable domain with one intradomain disulfide bond. Their small size (2.5 nm diameter, 4 nm height) enables tissue penetration that full-length antibodies cannot achieve, making them particularly valuable for solid tumor imaging and intracellular targeting applications.

Fab fragments retain both the variable and first constant domains of heavy and light chains, giving them higher thermal stability and longer serum half-life than VHH. However, the interchain disulfide bond between heavy and light chain constant domains adds expression complexity: both chains must be co-expressed and correctly paired.

Expression Systems for Nanobody Production

The choice of expression system for nanobody production is primarily driven by yield requirements and cost targets. E. coli remains the workhorse for early-stage discovery and research-grade material, while Pichia pastoris is increasingly adopted for clinical and commercial manufacturing where gram-per-liter yields are essential.

E. coli Expression: Periplasm vs Cytoplasm

E. coli is the default starting point for nanobody production because of its fast doubling time (20–30 min), well-established molecular biology tools, and low media cost. The critical decision is whether to target the periplasm or the cytoplasm, and this choice hinges on how the single intradomain disulfide bond in the VHH will form.

Periplasmic expression

Periplasmic expression uses an N-terminal signal peptide (pelB, OmpA, or DsbA) to translocate the nascent VHH across the inner membrane via the Sec or SRP pathway. The periplasm provides an oxidizing environment where the DsbA/DsbC foldase system catalyzes disulfide bond formation. Typical yields range from 10–100 mg/L, limited by the small volume of the periplasmic space (~8–16% of total cell volume) and Sec translocon capacity.

• Signal peptide choice matters: pelB (Sec-dependent) is the most common, but DsbA signal peptide (SRP-dependent, co-translational) often improves yield for aggregation-prone VHH because the protein enters the periplasm before completing translation.
• Induction conditions: Low IPTG (0.1–0.5 mM) and reduced temperature (20–25 °C) slow translation rate, giving each VHH molecule more time to fold correctly before the next one is translocated.
• Extraction: Osmotic shock releases periplasmic contents selectively, avoiding the bulk of cytoplasmic contaminants and simplifying downstream purification.

Cytoplasmic expression in engineered strains

E. coli SHuffle T7 Express is a strain engineered with mutations in the thioredoxin reductase (trxB) and glutathione reductase (gor) genes, making the cytoplasm sufficiently oxidizing for disulfide bond formation. It also constitutively expresses DsbC in the cytoplasm, a disulfide bond isomerase that corrects mispaired disulfides. This eliminates the translocation bottleneck, and yields of 200–300 mg/L soluble VHH are routinely achieved.

Pichia pastoris: The High-Yield Secretion Platform

Pichia pastoris (Komagataella phaffii) achieves 6–14 times higher nanobody titers than E. coli periplasmic expression by secreting functional VHH directly into the culture medium, bypassing the periplasmic bottleneck entirely. The methanol-inducible AOX1 promoter combined with the alpha-mating factor (αMF) secretion signal is the standard configuration.

Secretion into the culture medium offers three process advantages over intracellular E. coli expression:

1. Simplified harvest: Cell removal by centrifugation yields a clarified supernatant containing the VHH as the dominant protein (70–90% of total secreted protein), eliminating the need for cell lysis.
2. Correct disulfide bonds: VHH passes through the ER where protein disulfide isomerase (PDI) catalyzes disulfide bond formation under oxidizing conditions.
3. Product homogeneity: Pichia-derived VHH elutes as a single peak in SEC, compared to the pre-peak and post-peak shoulders often seen with E. coli material (aggregates and degradation products).

The key process parameters for Pichia nanobody production are methanol feed rate (typically 3–6 g/L/h), dissolved oxygen (>20% air saturation to support methanol oxidation), and pH (5.5–6.0 for protease minimization). Fed-batch fermentation over 72–120 hours with a glycerol growth phase (24–36 h to 100–150 g/L wet cell weight) followed by a methanol induction phase (48–84 h) typically achieves 0.5–1.5 g/L VHH in research-scale bioreactors, with optimized processes reaching 3 g/L.

Methanol-free expression systems using the GAP promoter or engineered AOX1 variants are gaining traction, particularly for GMP manufacturing where flammable methanol handling adds facility complexity and cost. GAP-driven constitutive expression simplifies the process to a single growth phase but typically yields 30–50% less than methanol-induced systems.

The Disulfide Bond Problem

Every antibody fragment requires at least one intradomain disulfide bond for structural stability, and this single requirement drives most of the complexity in fragment expression. A VHH nanobody has one disulfide bond connecting the B and F beta-strands of the immunoglobulin fold (Cys22–Cys92 in Kabat numbering). Without this bond, the VHH unfolds and aggregates within minutes at 37 °C.

An alternative to engineered strains is the CyDisCo system, which co-expresses Erv1p (a sulfhydryl oxidase from yeast mitochondria) and human PDI in the E. coli cytoplasm. CyDisCo can be used in any E. coli strain, including high-density fermentation workhorses like BL21(DE3), and has achieved 100–400 mg/L soluble VHH depending on the construct.

Why Do scFv Constructs Aggregate?

scFv aggregation is the single largest barrier to scalable antibody fragment production. Unlike VHH nanobodies (which are inherently stable as single domains), scFv constructs link two variable domains (VH and VL) with a flexible peptide linker, creating multiple opportunities for misfolding.

Three mechanisms drive scFv aggregation in E. coli:

1. Intermolecular domain swapping. The VH domain of one scFv molecule pairs with the VL domain of a neighboring molecule instead of its own, forming dimers, trimers, and higher-order oligomers. A (Gly₄Ser)₃ linker (15 residues) minimizes this by keeping the two domains in close proximity, but shorter linkers (5–10 residues) deliberately promote dimerization to create diabodies.
2. Hydrophobic exposure during folding. The VH–VL interface is normally buried, but during folding in the crowded periplasm (protein concentration 200–400 mg/mL), partially folded intermediates expose hydrophobic patches that nucleate aggregation.
3. Disulfide scrambling. Each scFv has two intradomain disulfide bonds (one in VH, one in VL). If one bond forms before the domain is fully folded, it can lock the domain in a non-native conformation that exposes aggregation-prone regions.

Practical mitigation strategies include lowering induction temperature to 16–20 °C (slows folding rate, allowing each domain to reach native conformation before the next intermediate appears), co-expressing periplasmic chaperones (Skp, FkpA, SurA), and engineering more soluble frameworks (e.g., the Herceptin scFv framework has been optimized for E. coli solubility across hundreds of CDR grafts).

When periplasmic expression fails, deliberate inclusion body (IB) formation followed by in vitro refolding can recover 30–60% of functional scFv. The refolding protocol typically uses 6–8 M urea solubilization, rapid dilution into a redox-controlled refolding buffer (0.5 mM oxidized / 5 mM reduced glutathione), and 0.4 M L-arginine to suppress aggregation during the folding process.

Purification Without Protein A

The absence of an Fc region in all antibody fragments means that Protein A chromatography, the universal mAb capture step, is not available. This fundamentally changes the downstream processing strategy, requiring format-specific alternatives that are typically less selective and more expensive per gram of product.

IMAC (immobilized metal affinity chromatography)

IMAC capture of His₆-tagged fragments is the fastest path from cell lysate to purified product. Ni-NTA or Co-TALON resins bind the polyhistidine tag with high affinity (K_d ~10^-13 M), and step elution with 250–300 mM imidazole typically yields 80–90% recovery at 75–85% purity. A subsequent SEC polishing step removes aggregates and co-purifying E. coli proteins that contain natural histidine clusters.

For therapeutic applications, the His-tag must be removed. A TEV or HRV 3C protease cleavage site between the tag and the VHH adds a processing step but leaves a native N-terminus. Subtractive IMAC (passing the cleaved pool over Ni-NTA to capture uncleaved material and the His-tagged protease) typically achieves >99% tag removal.

Protein L for kappa chain fragments

Protein L from Peptostreptococcus magnus binds the framework region of kappa (κ) light chains without interfering with antigen binding. This makes it suitable for Fab and scFv fragments containing a kappa variable domain, but not for VHH nanobodies (which have no light chain) or lambda-bearing fragments. Typical step yield is 75–85% with 85–95% purity in a single step.

Ion exchange chromatography for tag-free nanobodies

CEX (cation exchange) at pH 4.5–5.5 is the primary capture option for tag-free nanobody production at manufacturing scale. Most VHH nanobodies have a pI of 6–9, so they bind CEX resins at mildly acidic pH while most E. coli host cell proteins (average pI ~5.5) flow through. A salt gradient elution (0–500 mM NaCl) resolves the VHH from remaining contaminants, typically yielding 70–80% recovery at 65–75% purity. A polishing AEX (anion exchange) step in flow-through mode at pH 7–8 removes endotoxin, DNA, and acidic HCPs to achieve >95% purity.

Bioprocess Scale-Up Considerations

Scaling nanobody production from shake flasks to bioreactors introduces oxygen transfer, heat dissipation, and feed strategy challenges that differ between E. coli and Pichia platforms. The key parameters to control during scale-up depend on the host organism.

E. coli scale-up

High-cell-density E. coli fermentation targeting 50–100 g/L dry cell weight requires careful glucose feeding to avoid acetate overflow (critical threshold: specific glucose uptake rate <1.0 g/g/h). For SHuffle strains, the additional metabolic burden of maintaining an oxidizing cytoplasm makes them more sensitive to oxygen limitation. Maintain DO >30% air saturation during the induction phase, and consider reducing the induction temperature to 16–20 °C in a controlled ramp rather than a step change to avoid cold shock that reduces the proportion of soluble VHH.

• Oxygen demand: SHuffle at 30 °C post-induction requires approximately 15–25 mmol O₂/L/h, comparable to standard BL21 but lower than the 30–50 mmol/L/h needed during exponential growth at 37 °C.
• IPTG timing: Induce at OD₆₀₀ = 30–50 (fed-batch mid-log phase) rather than the shake-flask convention of OD₆₀₀ = 0.6–0.8. The higher cell density at induction increases volumetric yield proportionally.
• Harvest window: VHH accumulates linearly for 12–20 h post-induction, then plateaus as proteolytic degradation begins to match synthesis. Extend harvest beyond 20 h only if SEC confirms monomer fraction remains >90%.

Pichia scale-up

Methanol induction at bioreactor scale requires a calibrated methanol sensor (or off-gas CO₂ tracking) to maintain residual methanol at 0.5–2 g/L. Methanol concentrations above 5 g/L are toxic, while concentrations below 0.1 g/L starve the AOX1-driven expression system. The oxygen demand during methanol oxidation is exceptionally high (1.5 mol O₂ per mol methanol consumed), requiring kLa values of 200–400 h^-1 at production scale.

Proteolytic degradation of secreted VHH by Pichia vacuolar proteases (PEP4, PRB1) is the primary scale-up yield loss. Strategies include using protease-deficient strains (GS115 Δpep4 Δprb1), maintaining pH below 6.0, adding casamino acids (0.5–1% w/v) as competitive protease substrates, and minimizing culture time by increasing methanol feed rate.

Related tools

• Scale-Up Calculator — Compare scale-up criteria (constant P/V, tip speed, kLa) for moving your nanobody fermentation from bench to pilot scale.
• OTR & kLa Estimator — Check whether your bioreactor can supply the oxygen demand for high-cell-density E. coli or methanol-induced Pichia cultures.
• Refolding Buffer Generator — Design refolding conditions for scFv inclusion bodies with optimized redox pairs, L-arginine, and dilution ratios.

Frequently Asked Questions

References

1. de Marco A. (2020). Recombinant expression of nanobodies and nanobody-derived immunoreagents. Protein Expression and Purification, 172, 105645. doi:10.1016/j.pep.2020.105645
2. Zheng Y., Li B., Zhao S., Liu J., Li D. (2024). A Universal Strategy for the Efficient Expression of Nanobodies in Pichia pastoris. Fermentation, 10(1), 37. doi:10.3390/fermentation10010037
3. Li X., Liu W., Li H., Wang X., Zhao Y. (2022). Capture and purification of an untagged nanobody by mixed weak cation chromatography and cation exchange chromatography. Protein Expression and Purification, 191, 106030. doi:10.1016/j.pep.2021.106030
4. Sarker A., Rathore A.S., Gupta R.D. (2019). Evaluation of scFv protein recovery from E. coli by in vitro refolding and mild solubilization process. Microbial Cell Factories, 18(1), 5. doi:10.1186/s12934-019-1053-9
5. Duggan S. (2018). Caplacizumab: First Global Approval. Drugs, 78(15), 1639–1642. doi:10.1007/s40265-018-0989-0