Protein Production & Purification

Michael Jeltsch

Practical course in protein purification (DPDR-305) &�Introduction to Cell and Molecular Biology�Methods (PROV-004)

Autumn term 2025/Spring term 2026

Protein purification

Slide 1 of 51

Most recent version of this presentation: mjlab.fi/aappp

Creative Commons (attribution not to Michael Jeltsch)

Public domain

The idea is to make all material in this presentation publicly and freely available using a creative commons license (CC BY-NC-SA 4.0). However, this goal has not been reached yet.

©

Reproduced with permission from the copyright holder

Copyrighted material (scheduled for replacement)

Jeltsch Lab & UH

Why to produce your own protein?

Slide 2 of 51

• Because nobody produces it.
• Because it’s too expensive to buy it.
• Because the commercial protein is bad quality.
• Because you want to sell it.

Example VEGFR-3/Fc fusion protein (VGX-300/OPT-302/sozinibercept)

Commercial: For 50 µg (R&D Systems) $421

Selfmade: For 50 mg total costs excl. equipment ~$500 (vs. $421,000)

Jeltsch Lab & UH

The big picture

Slide 3 of 51

Jeltsch Lab & UH

(Incomplete) decision tree for protein production & purification

Slide 4 of 51

Jeltsch Lab & UH

Classification of expression systems

Slide 5 of 51

Different types of expression

• Stable expression

• Nonviral�delivery

• Inducible expression

• Constitutive expression

• Viral�delivery

• Transient expression

Jeltsch Lab & UH

Different expression hosts

Slide 6 of 51

• Bacteria�E. coli, Bacillus subtilis,�Lactococcus, Pseudomonas

• Yeast�Saccharomyces cerevisiae, Pichia pastoris

• Insect cells�Drosophila (Schneider S2), Sf9, High5

• Chemical Protein Synthesis

• Cell-free expression

• Mammalian cells

HEK293, CHO

Jeltsch Lab & UH

Bacterial expression

Slide 7 of 51

• Cheap & fast!
• Some mammalian proteins can be expressed�easily in E. coli (no fancy post-translational modifications).
• Sometimes 5% activity might be enough (IL-3)
• Several commercial systems, e.g. pET, pBAD, GST
• Secreted expression into periplasm is possible if the�protein does not feature too many disulfide bonds.
• The first biopharmaceutical made with recombinant DNA tech-�nology was E. coli-produced human insulin (Humulin), still available.
• Distribution of all biopharmaceuticals by expression system (2011):

Jeltsch Lab & UH

The lac operon

Slide 8 of 51

• Mostly relatively short promoters are used (Amp 105bp, EM7 48bp, T7 19bp)
• No polyA, but other termination signals (GC-rich palindromes + polyT stretch)

Jeltsch Lab & UH

The lac operon

Slide 9 of 51

Jeltsch Lab & UH

The lac operon

Slide 10 of 51

• Mostly relatively short promoters are used (Amp 105bp, EM7 48bp, T7 19bp)
• No polyA, but other termination signals (GC-rich palindromes + polyT stretch)

Jeltsch Lab & UH

More stringent control of expression: λDE3 strains

Slide 11 of 51

E. coli DE3 strains are infected with a mutant λ virus (which cannot lyse the cell, but is integrated in the host genome).

Jeltsch Lab & UH

More stringent control of expression: Uninduced

Slide 12 of 51

Jeltsch Lab & UH

More stringent control of expression: Induced

Slide 13 of 51

Jeltsch Lab & UH

Bacterial expression

Slide 14 of 51

• Special E. coli strains for expression of proteins (protease-deficient): BL21, BL21(DE3) or BL21(DE3)pLysS�DE3 = T7 RNA polymerase under lacUV5/lacO control (genomic)�pLysS = T7 lysozyme, which inhibits T7 RNA polymerase
• Many other systems with tighter regulation
• Constitutive expression: EM7 promoter for non-toxic proteins
• GOI up to 50% of total cell protein
• Solubility: inclusion bodies (partially predictable,�main factor: charge)
• Increasing the solubility: MBP (maltose binding protein)- or�GST (gluthathion-S-transferase)-tag; low temperature growth
• Inclusion bodies can be useful for purification if they can be solubilized and refolded!

Jeltsch Lab & UH

Inducible expression in eukaryotic cell culture: Tet-On® system

Slide 15 of 51

Dox (Tet-On, Clontech; T-Rex, LifeTechnologies), IPTG (LacSwitch, Agilent), Edcysone (Complete Control, Agilent), Mifepristone (GeneSwitch, LifeTechnologies), RSL1 (RheoSwitch, NEB), Cumate (Cumate gene switch), Me²⁺ (methallothionine promoter)...

Jeltsch Lab & UH

Yeast

Slide 16 of 51

Yeast

• Can be sometimes an alternative...
• VEGF-C produced in yeast is active, while its closest homolog VEGF-D is inactive (glycosylation issue)

Jeltsch Lab & UH

Eukaryotic gene organization

Slide 17 of 51

Jeltsch Lab & UH

The Kozak sequence

Slide 18 of 51

Efficient eukaryotic initiation requires a good Kozak sequence!

Marilyn Kozak used mammalian sequences! Insect cells, yeast, etc. have divergent initiation preferences (although the purin preference at -3 seems universal)

Jeltsch Lab & UH

To intron or not to intron

Slide 19 of 51

• Introns can increase expression levels and are included in some vectors (typically before the MCS)
• Splice donor and splice acceptor sites are weakly defined, risk to activate „cryptic“ splice sites or to generate them de-novo
• Can be checked, e.g. at http://spliceport.cbcb.umd.edu/

Jeltsch Lab & UH

To intron or not to intron

Slide 20 of 51

Jeltsch Lab & UH

Promoter choice

Slide 21 of 51

• Constitutive vs. inducible promoters
• Promoter are species and cell-line specific (due to specific transcription factor binding sites)
• To bypass this restriction, mostly viral promoters (which use general and virally encoded transcription factors) are used that function in most cell lines

Qin et al. (2010) Systematic Comparison of Constitutive Promoters and the Doxycycline-Inducible Promoter. PLoS ONE 5(5): e10611.

Jeltsch Lab & UH

PolyA signal

Slide 22 of 51

polyA signal

• Important for mRNA export from nucleus, mRNA stability and recruitment of ribosomes
• Not as species-specific like promoters
• In bacteria, polyadenylation promotes mRNA degradation.
• No absolute requirement

Jeltsch Lab & UH

Insect cells

Slide 23 of 51

Comparatively few insect cell lines:

• Schneider S2 (Drosophila melanogaster)
• Sf9, Sf21 (Spodoptera frugiperda)
• Hi5 (BTI-TN-5B1-4, Trichoplusia ni)

Trichoplusia ni

(cabbage looper) https://commons.wikimedia.org/wiki/File:Trichoplusia_ni_larva.jpg

Spodoptera frugiperda (fall armyworm)

Drosophila melanogaster

(fruitfly)

https://commons.wikimedia.org/wiki/File:Drosophila_melanogaster_-_side_(aka).jpg

Jeltsch Lab & UH

Commonly used insect cell expression systems

Slide 24 of 51

Insect cell expression

Baculovirus

Stable transfection

Transient transfection

Drosophila S2 cells

Drosophila S2 cells

Sf9 cells

Hi5 cells

BmNPV:

Bombyx mori (both cells and larvae)

AcNPV:

Sf9,Sf21

Hi5

Bac-to-Bac®*

DES*

InsectSelect™*

​

*Invitrogen/LifeTechnologies/ThermoFisher

Jeltsch Lab & UH

Good to remember...

Slide 25 of 51

• Mammalian promoters (CMV, SV40, EF1α) do not work in insect cells! Special insect cell plasmids are needed. Baculoviral promoters do not work in S2 cells and S2 cell promoters do not work in Sf or Hi5 cells!
• For secreted proteins: Mammalian signal peptides may or may not work!
• Baculoviruses require class 1 biosafety level (they do efficiently infect mammalian cells, hence they are considered as gene therapy vectors)
• Typically 3-4 weeks from ready construct to protein expression
• Transfection of insect cells is not as straightforward as transfection of mammalian cells; fewer well-performing transfection reagents on the market than for mammalian cells (e.g. Qiagen's Effecten)
• Insect cells don't need CO2-incubators and can be grown at RT!
• Selection of stable cell lines is usually done without clonal isolation.

Jeltsch Lab & UH

Mammalian expression systems

• Mostly stable cell lines are used:�CHO (Chinese Hamster Ovary) cells, ~90% of all biopharmaceuticals�293 HEK (Human Embryonic Kidney)
• Transient transfection works as well (and is faster, but costs for transfection reagents become prohibitive)

Production of recombinant proteins in mammalian hosts

• Typical product titre: ~50mg/l (1986) vs. 4.7g/l (2004)
• Sources of improvement:�a) Cell density is 4-5x higher due to medium & fermentation optimization�(e.g. roller-bottles vs. fermenter), host cell engineering�b) Production phase is 2-3x longer due to host cell engineering and�fermentation optimization�c) Specific productivity is 5-10x higher due to HT screening, host-cell�engineering (site-specific integration, protein folding and posttranslational�modification enhancements)
• Original CHO system^1,2 has no IP issues, therefore it is still frequently�used!

1. Kaufman, R et al. (1985) Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase in CHO cells. Mol Cell Biol 5, 1750-59.

2. Kingston RE et al. (2002) Amplification using CHO cell expression vectors. Curr Protoc Mol Biol Chapter 16: Unit 16.23.

Enemies of physical stability

Slide 28 of 51

Heat

pH

→ Denaturation/Aggregation

Jeltsch Lab & UH

Enemies of physical stability

​

Slide 29 of 51

Why does the egg white denaturate despite no pH and temperature change?

https://foto.wuestenigel.com/woman-beats-egg-whites-with-a-mixer-top-view-of-the-process-of-making-a-pie/

Jeltsch Lab & UH

Enemies of physical stability: phase transitions

​

Slide 30 of 51

Jeltsch Lab & UH

Enemies of physical stability

Slide 31 of 51

How to treat antibody solutions�(and generally proteins) in the lab

• Temperature: Keep on ice!
• pH: Never change the pH* (e.g. by diluting into a buffer with a different pH)!
• Phase transitions: Avoid freezing and thawing! Avoid making bubbles when pipetting!

* and if you MUST do it, test it out on a small aliquot!

Jeltsch Lab & UH

Tricks to increase protein expression levels

Slide 32 of 51

Possible protein concentrations in conditioned cell culture medium have been constantly rising: ~50mg/l (1986), 4.7g/l (2004), ~10g/l (2019)*

• Higher cell density (medium and fermentation optimization, host cell engineering)
• Longer production phase (host cell engineering, fermentation optimization)
• Specific productivity (HT screening, host cell engineering, site-specific integration)

​

1. Gene amplification systems (MTX, GST)
2. Targeted integration
3. Improved vectors
4. Genetically engineering host cells (e.g. antiapoptotic, capacity for post-translational modification)
5. Better screening methods
6. Media optimization
7. Better culture conditions (“process development innovations”)

Jeltsch Lab & UH

Cell line development

Slide 33 of 51

Jeltsch Lab & UH

The MTX gene amplification system

Slide 34 of 51

Jeltsch Lab & UH

The MTX gene amplification system

Slide 35 of 51

10.1016/0022-2836(82)90103-6

Jeltsch Lab & UH

From inoculation to harvest

Slide 36 of 51

Cells don’t like to grow alone!

Jeltsch Lab & UH

Bioreactor types and operation modes

Slide 37 of 51

Different operation

modes:

• Batch: Inoculate final culture with cells. Done.
• Fed-batch: Inoculate final culture with cells, but keep adding nutrients/medium during production (mostly glucose)
• Continuous feed: Inoculate final culture with cells, add continuously medium during production and withdraw continuously medium & cells for purification.

Jeltsch Lab & UH

Bioreactors

Slide 38 of 51

MINIFOR laboratory fermentor-bioreactor advanced kit 1L vessel-right" by LambdaCZ

Large scale bioreactor by Sanofi Pasteur

High-tech bioreactors do not necessarily result in higher yields compared to shaker flasks!

Jeltsch Lab & UH

cGMP facilities

Slide 39 of 51

Retail pricing of recombinant proteins. rFVIIa—recombinant activated factor VII; rFVIII—recombinant factor VIII (https://doi.org/10.3390/pr7080476).

Jeltsch Lab & UH

cGMP facilities

Slide 40 of 51

• Recouping the development costs�(17% of revenue versus 2% average for a S&P500 company)
• Clinical phase 3 trial costs: ~$41k per patient (https://aspe.hhs.gov/system/files/pdf/77166/rpt_erg.pdf)
• Safety issues: keeping the culture contamination-free�(disposable bioreactors for up to 4000l, example of severe drug shortage due to contamination: Cerezyme for Gaucher disease)
• Purity requirements for the end product and the starting material�(also for trivial chemicals such as water)
• Quality control & regulatory oversight

Jeltsch Lab & UH

Purification of antibodies

Slide 41 of 51

F_C region

Bacterial

surface

proteins:

• Protein A
• Protein G
• Protein L

Jeltsch Lab & UH

Purification of antibodies

Slide 42 of 51

*Protein L binds only antibodies that contain the a subset of kappa light chains: human VκI, VκIII and VκIV (but not to VκII), mouse VκI.

Jeltsch Lab & UH

Purification of antibodies

Slide 43 of 51

Polishing

• Size-exclusion chromatography�(ideal for research but does not scale well since it requires a low V_sample/V_column)
• Ion exchange
• “Mixed-mode resins” (e.g. hydroxyapatite:�ion exchange & hydrophobic�interaction)

Capture

Jeltsch Lab & UH

Intellectual property

Slide 44 of 51

• Many advances are kept proprietary (trade-secrets) and are not patented.
• Other systems have been commercialized (i.e. are not available for academic research due to budget limitations).
• Startup companies sometimes prefer old-fashioned systems, which are free from intellectual property rights.

Jeltsch Lab & UH

Expression of two or more genes from one plasmid

Slide 45 of 51

GOI 1

GOI 2

ATG

CTG

Promoter

internal ribosome entry site (IRES)

Ratio of expression between GOI1 and GOI2 can be regulated: IRES < cap-dependent translation, CTG << ATG

GOI 2

Promoter

2A element*

GOI 1

G↔A

non-AUG initiation

GOI 2

Promoter

intronic expression

intron

GOI 1

Prom.

intron

intron

intron

GOI 2

Promoter

„protein intron**“

intron

GOI 1

modified intein

Jeltsch Lab & UH

Comparison of expression systems

Slide 46 of 51

1 except for modified strains, 2 under development, 3 except for Sf9 mimic™, 4 with exceptions, 5 yeast-NatB-transgenic E. coli

Jeltsch Lab & UH

Producing antibodies in the lab and in industry

Slide 47 of 51

Why are hybridomas not used to produce antibodies in large scale?

1. They are all initially inherently unstable (because they have a duplicate set of chromosomes); most of them stabilize after prolonged culture.
2. Due to a), hybridomas are all a bit different from each other.

​

Antibody production workhorse: CHO (Chinese Hamster Ovary) cells

1. Fast growth
2. High protein production
3. Can be grown as adherent and suspension cultures
4. Mutant lines for cell line selection and amplification systems to increase protein production: Dhfr-negative CHO cells (e.g. CHO-DG44, evolution of CHO cells role in cell line development)

​

Why not to use transgenic animals to produce antibodies (e.g. sheep and goats who produce it in the milk)?

Making a CHO cell line takes about 6 weeks. Making a transgenic goat takes about 2 years (and establishing a trip of transgenic goats takes several years).

​

Why are animal cells used and not human cells?

Human cells have been/are used (e.g. HT-1080 for making Epo). However, there is concern about human viruses. Most human viruses do not propagate in rodent cells (see e.g. the vesivirus case).

Jeltsch Lab & UH

Methods & Tags

Slide 48 of 51

Tags: https://mjlab.fi/aappt

Methods: https://mjlab.fi/aappm

Jeltsch Lab & UH

Questions, contact

Slide 49 of 51

• Laboratory: mjlab.fi (https://www.helsinki.fi/en/researchgroups/lymphangiogenesis-research-and-antibody-development)
• Core facility for protein production and purification: b3p.it.helsinki.fi
• jeltsch.org (private rumblings)
• jeltsch.org/science (private rumblings without the non-scientific stuff)
• Questions to: michael@jeltsch.org or via Skype: jeltsch
• This presentation: mjlab.fi/pp, (jeltsch.org/teaching)
• All material is CC-licensed by the lecturer with the following exceptions:� images marked as public domain� images marked explicitly as Creative Commons-licensed by others� © copyright-protected images are reproduced with permission of the right holder� copyright-protected, need to be replaced if this is made publicly available

Jeltsch Lab & UH