What Is a Recombinant Protein and How Is It Made?

A recombinant protein is any protein produced by a cell whose genetic material has been deliberately modified to include a gene from another source. The host cell reads the inserted gene and builds the corresponding protein, which can then be harvested and purified. This is the technology behind recombinant insulin, many cancer drugs, the enzymes in your laundry detergent, and even the rennet used to make most cheese sold today. The process sounds straightforward, but each step involves choices that determine whether you end up with a functional, safe, usable product or a clump of misfolded waste.

Getting the Gene Into a Host Cell

Every recombinant protein starts with a gene. Researchers identify the stretch of DNA that codes for the protein they want, then insert it into a small, circular piece of DNA called an expression vector. The vector acts as a delivery vehicle and instruction manual rolled into one: it carries the gene into the host cell and contains the regulatory signals that tell the cell when and how aggressively to produce the protein.

One underappreciated step in this process is codon optimization. Different organisms have preferences for which three-letter DNA codes they use to specify the same amino acid. A human gene dropped into a bacterium may be read slowly or inefficiently because the bacterium rarely encounters certain codons. By swapping in synonymous codons that the host prefers, researchers can dramatically boost output. In one example, codon-optimizing a malaria vaccine candidate gene for production in E. coli increased protein yield at least threefold.¹ Modern deep-learning tools are now being developed to automate this process, using neural networks to predict which codon arrangements will maximize expression in a given host.²

Choosing an Expression System

The host cell you grow the protein in is arguably the most consequential decision in the entire process. Each type of cell offers a different balance of speed, cost, yield, and ability to produce a protein that actually works. The main options are bacteria, yeast, mammalian cells, insect cells, and plants.

Bacteria

Escherichia coli is the workhorse of the field and the most popular expression platform, period. It grows fast, is cheap to culture, and benefits from decades of genetic tool development: a huge catalog of expression vectors, engineered strains, and well-established production protocols.³ A significant share of approved therapeutic proteins have been produced in E. coli.⁴

The catch is that bacteria are simple cells. They lack the machinery that more complex cells use to add sugar chains and other chemical decorations to proteins after they are assembled. Many human proteins need those modifications to fold correctly, stay stable in the bloodstream, or avoid triggering an immune response. Without them, bacterial production can yield misfolded, insoluble, or nonfunctional proteins.⁴ Some proteins are also outright toxic to the bacterial host, killing the cells before enough product accumulates.

Yeast

Yeast cells, especially the species Pichia pastoris, offer a middle ground. They grow nearly as fast and cheaply as bacteria but are complex enough to fold proteins properly and secrete them outside the cell, which simplifies collection. The Pichia system is particularly valued for appropriate folding inside the cell’s internal compartments and efficient secretion of the finished protein into the surrounding growth medium.⁵ Yeast can add some sugar modifications, though the patterns differ from human cells, which can be a problem for certain therapeutic proteins.

Mammalian Cells

When a protein absolutely must look and behave like the human version, mammalian cells are the standard. Chinese hamster ovary (CHO) cells dominate this space. They are the most common production platform for glycosylated biopharmaceuticals thanks to their efficient secretion systems and their ability to add the post-translational modifications that human-targeted drugs require.⁶ CHO cells are used to produce monoclonal antibodies, vaccines, and hormones.⁷

The downside is cost and speed. Mammalian cells grow much more slowly than bacteria or yeast, require expensive growth media, and need carefully controlled conditions. Building a large-scale mammalian cell production facility can require capital investment approaching hundreds of millions of dollars.⁸

Insect Cells

Insect cell systems, usually driven by a baculovirus vector, occupy a niche between yeast and mammalian cells. They can perform many of the protein modifications that bacteria cannot and are widely used for recombinant protein production, particularly for research-grade proteins and some vaccines.⁹ Their glycosylation patterns are simpler than mammalian cells, which can limit their use for certain human therapeutics, but they handle complex multi-subunit proteins reasonably well.

Plants

Plant molecular farming is the newest entrant. Researchers engineer tobacco, rice, or other crop plants to produce pharmaceutical proteins in their leaves or seeds. The appeal is scalability: unlike bioreactors, where expansion requires building larger tanks while maintaining the right mixing and oxygen conditions, plant systems scale linearly by simply growing more plants. A plant-based facility can be established at less than one-tenth the capital cost of a comparable mammalian cell facility.⁸ The environmental footprint is also roughly 80% lower than stainless-steel fermenters.⁸

The trade-off is yield. Plant systems typically produce far less protein per volume than bacterial or CHO cultures, and extracting a pure product from plant tissue adds processing complexity.¹⁰ Plants are best positioned as a complementary platform, especially for products where cost, safety from human pathogens, and sustainability matter more than raw volumetric output.

The Inclusion Body Problem

When E. coli produces a foreign protein at high levels, the protein molecules often pile up inside the cell faster than they can fold into the correct three-dimensional shape. The result is dense clumps called inclusion bodies, essentially aggregates of mostly a single type of protein that are insoluble and inactive.¹¹ For anyone making a recombinant protein in bacteria, inclusion bodies are one of the most common headaches.

Recovering useful protein from these aggregates is a multi-step rescue operation. First, the inclusion bodies are isolated and then dissolved using chemical agents called denaturants, which unfold the tangled proteins. Then the dissolved proteins must be carefully refolded into their correct shape, and finally purified.¹² The process is labor-intensive and yields can be low.

An encouraging finding is that proteins inside inclusion bodies often retain native-like secondary structures rather than being completely scrambled. This means that gentler solubilization methods, using mild conditions like moderate urea concentrations at alkaline pH rather than harsh denaturants, can preserve that partial structure and make refolding much more successful. Using mild solubilization, researchers have recovered more than 40% of inclusion body proteins back to their bioactive form.¹¹ That may not sound impressive, but in a field where harsh methods often yielded single-digit recoveries, it represents a significant improvement.¹³

Growing Cells at Scale

Producing a recombinant protein in a test tube is one thing. Making grams or kilograms of it requires fermentation at industrial scale, and the transition from bench to factory is far from automatic. Cells behave differently in a 50-liter tank than in a 1.5-liter flask. Oxygen transfer, nutrient mixing, heat removal, and waste buildup all change with volume, and these physical parameters directly affect how much protein the cells produce.

Scaling up typically involves identifying a physical parameter that can be held constant across different volumes. In one study scaling E. coli production of a meningitis vaccine protein from bench to 50-liter culture, matching a specific oxygen-transfer parameter nearly doubled total protein output compared to the smaller scale, reaching about 546 milligrams per liter versus 284 milligrams per liter at bench scale.¹⁴ The improvement came not from changing the biology but from finding the right engineering conditions to keep the cells happy in a bigger vessel.

Purifying the Product

Once the cells have produced the protein, it needs to be separated from everything else: the host cell’s own proteins, DNA, lipids, growth-medium components, and any debris from broken cells. This downstream processing can account for a large share of total production costs, especially for therapeutic proteins that require extreme purity.

One of the most widely used purification tricks relies on a small molecular tag. Researchers attach a short stretch of histidine amino acids to one end of the recombinant protein during the cloning step. This “His-tag” has a strong affinity for nickel or cobalt ions immobilized on a chromatography column. When the crude cell extract is poured over the column, the tagged protein sticks while nearly everything else washes through.¹⁵ The protein is then released by adding a competing molecule. This single step can achieve purities above 95%.¹⁶ In some cases, a single pass can purify a protein a thousandfold or more from a crude extract.¹⁷

For therapeutic antibodies produced in CHO cells, the workhorse purification method is Protein A chromatography, which exploits a natural binding interaction between Protein A and antibodies. Clearing out residual host cell proteins after this step remains an active area of optimization. Research has shown that many of the host cell protein contaminants that survive Protein A capture are stuck to the antibody product itself rather than to the column, which means improved wash steps need to target product-contaminant interactions specifically.¹⁸ Additional polishing steps using multimodal chromatography resins can reduce remaining host cell proteins by about 80%.¹⁹

Why Sugar Chains Matter So Much

Many proteins that work in the human body carry chains of sugar molecules attached after the protein itself has been assembled. These glycans are not decoration. They influence how long a protein survives in the bloodstream, whether the immune system attacks it, how stable it is during storage, and sometimes whether it works at all.²⁰

Heavily sialylated sugar chains, for example, help proteins circulate longer by shielding them from receptors in the liver that would otherwise grab them and send them for degradation. On the other hand, nonhuman sugar patterns can trigger pre-existing antibodies in patients, leading to rapid clearance or even dangerous allergic reactions.²¹ This is exactly why the choice of expression system matters so much for therapeutic proteins: bacteria add no sugars, yeast adds sugars but in patterns foreign to humans, and only mammalian cells produce patterns close enough to human for most drug applications.

Glycoengineering takes this a step further by deliberately modifying sugar chains to improve a protein drug’s performance. The most famous example is darbepoetin alfa, a modified version of erythropoietin (a hormone that stimulates red blood cell production). By adding two extra sugar chain attachment sites, researchers created a version with roughly three times the circulating half-life and greater biological activity than standard recombinant erythropoietin.²²

Medical Applications

Recombinant proteins have transformed medicine in areas where patients lack a functional version of a critical protein. Enzyme replacement therapy is one of the clearest examples: patients with inherited conditions like Gaucher disease, Fabry disease, or Pompe disease receive regular infusions of the enzyme their bodies cannot make. Twenty different recombinant enzymes have been approved for human therapy, covering not just inherited storage diseases but also conditions like myocardial infarction, cystic fibrosis, chronic gout, and certain leukemias.²³

Enzyme replacement therapy is not a cure-all, though. For the mucopolysaccharidoses, a family of diseases where sugar molecules accumulate in cells, infused enzymes effectively reduce certain biomarkers and shrink enlarged livers and spleens. But they struggle to reach cartilage, bone, and eye tissue, so patients continue to experience problems in those areas.²⁴ The proteins are large molecules that do not cross certain tissue barriers well, a limitation that gene therapy and next-generation delivery systems are trying to address.²⁵

Recombinant Proteins in the Food Industry

Not all recombinant proteins go into medicine. One of the earliest and most commercially successful applications is cheese-making. Traditional cheese production depends on rennet, an enzyme mixture harvested from the stomachs of young calves. Due to cost, limited supply, and ethical concerns about using animal-derived enzymes, the industry largely shifted to fermentation-produced chymosin, which is the key enzyme in rennet made by microorganisms carrying the calf gene.²⁶

Skeptics might wonder whether recombinant chymosin makes different cheese. In head-to-head trials comparing Cheddar cheese made with recombinant chymosin versus standard calf rennet, the cheesemaking characteristics, composition, and maturation rates were similar. A taste panel could not tell the cheeses apart at 3, 6, or 12 months of aging.²⁷ Today, fermentation-produced chymosin dominates commercial cheese production in many countries.

Cell-Free Production

All the expression systems described so far rely on living cells. Cell-free protein synthesis takes a different approach: researchers extract the molecular machinery needed for protein production from cells, then use it in a test tube, adding the gene template and raw materials directly. Without a living cell to maintain, the system sidesteps several problems.

The most compelling advantage is for proteins that are toxic to their production host. If a protein kills bacteria or shuts down mammalian cell growth, you simply cannot make much of it in a living system. Cell-free platforms allow toxic protein production at high yield because there are no cell viability constraints to worry about.²⁸ Cell-free systems are also valued for their flexibility: you can modify the chemical environment, add unnatural amino acids, or screen many protein variants in parallel without the overhead of cloning into cells and growing cultures.²⁹ The current limitation is scale. Cell-free reactions are typically small and expensive per milligram of product, so the technology is used more for research and rapid prototyping than for bulk manufacturing.

Quality Control and What Regulators Look For

For any recombinant protein destined for human use, manufacturers must demonstrate that the final product is free of harmful impurities. Host cell proteins are among the most scrutinized contaminants. These are the thousands of different proteins the production cell naturally makes alongside the desired product. Even trace amounts can trigger immune reactions in patients or affect the drug’s stability over time.³⁰

Regulatory agencies like the FDA and EMA evaluate biosimilar products, copies of existing biologic drugs, by requiring extensive comparability testing across physicochemical properties, biological activity, pharmacokinetics, and clinical outcomes.³¹ The analytical comparability assessment is the foundation: if the molecular fingerprint of a biosimilar does not closely match the reference product, the clinical data cannot compensate.³² Because recombinant proteins are made by living systems, no two production runs are perfectly identical, which makes establishing acceptable ranges for these quality attributes one of the central challenges in biopharmaceutical manufacturing.

Engineering Better Host Cells with CRISPR

Rather than accepting the natural strengths and limitations of a given cell line, researchers are increasingly rewriting the host cell’s own genome to make it a better protein factory. CRISPR gene-editing tools have opened up a wide range of modifications in CHO cells specifically. Recent work has used CRISPR to modulate glycosylation patterns so that sugar chains on the product more closely match therapeutic requirements, to boost productivity by removing genes that limit how much protein the cell can secrete, to eliminate host cell proteins that are particularly difficult to remove during purification, and to knock out genes that make cells susceptible to viral contamination.³³

One practical goal is developing antibiotic-free selection systems. Traditionally, researchers identify which cells successfully took up the gene of interest by including an antibiotic-resistance gene alongside it, then dosing the culture with antibiotics. Only cells that incorporated the new DNA survive. CRISPR-based approaches can replace this with cleaner alternatives, reducing concerns about antibiotic resistance genes ending up in pharmaceutical products. Site-specific transgene integration, where the gene lands in a predetermined safe spot in the genome rather than a random location, also helps ensure consistent protein output from batch to batch.³³ These engineering strategies are gradually reshaping what production cell lines can do, blurring the line between “choosing” an expression system and “designing” one from scratch.