What is Dr. Zubair Khalid's research focus?

Dr. Zubair Khalid specializes in molecular virology, mRNA vaccine development, and computational biology, with a focus on avian pathogens like IBDV and Avian Reovirus.

Where is Dr. Zubair Khalid currently working?

Dr. Zubair Khalid is a Postdoctoral Research Associate at the University of Maryland (UMD), specifically within the Department of Animal and Avian Sciences.

Downstream and Upstream in Bioprocessing: A Practical Guide

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Introduction to Downstream and Upstream in Bioprocessing

In biopharmaceutical manufacturing, the terms upstream and downstream describe the two major phases of the production workflow. Upstream processing (USP) encompasses everything from cell line selection through cell culture and harvest—the steps that generate the biological product. Downstream processing (DSP) covers all operations following harvest: purification, viral inactivation, formulation, and final fill. This division is operational, not merely descriptive; it dictates facility design, staffing, equipment selection, and process economics.

The terminology can be confusing because molecular biologists use "upstream" and "downstream" to describe the direction of transcription relative to a gene. In bioprocessing, the terms refer to the physical flow of material through a manufacturing train. A molecule that is "upstream" in the process is simply earlier in the workflow. This guide uses the bioprocessing definitions throughout.

Defining Upstream Processing (USP)

Upstream processing begins with a master cell bank and ends with a harvested, clarified cell culture fluid containing the product of interest. The core activities are cell line development, media formulation, bioreactor operation, and harvest. The goal is to maximize product titer (concentration) while maintaining product quality attributes within a defined specification. Typical upstream operations run from 2 to 4 weeks for a fed-batch mammalian cell culture, with perfusion processes extending run durations to several months.

Key upstream parameters include temperature (typically 36–37°C for mammalian cells), pH (7.0–7.2), dissolved oxygen (30–50% of air saturation), and osmolality (280–320 mOsm/kg). These parameters are monitored and controlled in real time by the bioreactor control system.

Defining Downstream Processing (DSP)

Downstream processing begins with the harvested cell culture fluid and ends with a formulated, sterile drug substance suitable for fill-finish. The goal is to remove impurities—host cell proteins (HCPs), DNA, endotoxins, viruses, and product-related variants—while recovering the maximum possible yield of active product. Downstream operations are typically organized into three phases: capture, intermediate purification, and polishing.

The downstream train is often described as a series of unit operations, each with a specific function. A typical monoclonal antibody (mAb) process includes protein A affinity chromatography, low-pH viral inactivation, cation exchange chromatography, anion exchange chromatography, viral filtration, and ultrafiltration/diafiltration (UF/DF). Each step contributes to purity but also to product loss, so step yield is a critical metric.

The Bioprocess Workflow Overview

The complete bioprocess can be visualized as a linear sequence:

Cell line development and master cell bank creation
Media preparation and sterilization
Seed train expansion (from vial to production bioreactor)
Production bioreactor (fed-batch or perfusion)
Harvest and clarification (centrifugation, depth filtration)
Capture chromatography
Viral inactivation
Intermediate purification (polishing chromatography)
Viral filtration
Final formulation (UF/DF, excipient addition)
Sterile filtration and fill-finish

Steps 1–5 constitute upstream processing; steps 6–11 constitute downstream processing. The interface between the two is the harvest step, where the decision is made about what material enters the purification train.

Upstream Processing: From Cell Line to Harvest

Cell Line Development and Selection

The upstream process begins with the choice of expression system. For therapeutic proteins, the most common systems are Chinese hamster ovary (CHO) cells, Escherichia coli, and Saccharomyces cerevisiae. CHO cells are preferred for complex glycosylated proteins because they perform human-like post-translational modifications. E. coli is used for non-glycosylated proteins such as insulin and for antibody fragments. Yeast offers a middle ground with faster growth and simpler media.

Cell line development involves transfection of the host cell with the gene of interest, selection of stable clones, and screening for productivity and product quality. The glutamine synthetase (GS) system in CHO cells is a standard selection method: cells lacking endogenous GS are transfected with the GS gene and the gene of interest, then selected in media lacking glutamine. Methotrexate-based dihydrofolate reductase (DHFR) amplification is an older but still used alternative.

A critical early decision is whether to use a clonal or pooled cell population. Clonal lines, derived from a single cell, offer greater consistency but take 6–9 months to develop. Pooled populations are faster but carry the risk of genetic instability and batch-to-batch variation. For commercial manufacturing, a clonally derived cell line with documented stability over at least 70 generations is the regulatory expectation.

Media and Feed Strategies

Cell culture media must supply amino acids, vitamins, sugars, salts, trace elements, and growth factors. Chemically defined media, in which every component is known, are now standard for regulatory filings. Serum-free media are mandatory for therapeutic production because serum introduces undefined proteins, viruses, and batch variability.

The two dominant feeding strategies are fed-batch and perfusion. In fed-batch, the culture begins with a low working volume and receives concentrated nutrient feeds at intervals. Glucose and glutamine are the primary energy sources; lactate and ammonia accumulate as byproducts and inhibit growth. A typical fed-batch CHO process runs 12–14 days, achieving titers of 3–8 g/L for mAbs.

Perfusion continuously removes spent media while adding fresh media, maintaining cells in exponential growth. This approach achieves much higher cell densities (50–100 × 10⁶ cells/mL vs. 10–20 × 10⁶ in fed-batch) and allows extended run times. Perfusion is essential for unstable products or when the product is secreted continuously and degrades in the bioreactor. The trade-off is increased complexity, higher media consumption, and the need for cell retention devices such as alternating tangential flow (ATF) filters.

Bioreactor Operation and Control

Production bioreactors range from 500 L to 25,000 L for mammalian cell culture. The seed train expands cells from a cryovial through progressively larger vessels—typically a 125 mL shake flask, then 1 L, 5 L, 50 L, 500 L, and finally the production vessel. Each step requires a 1:5 to 1:10 split ratio, and the total expansion takes 2–4 weeks.

Key process parameters are controlled by the bioreactor's distributed control system (DCS). Dissolved oxygen is maintained by sparging air or oxygen through a fritted sparger; pH is controlled by CO₂ addition (to lower pH) and base addition (typically sodium bicarbonate or sodium hydroxide, to raise pH). Temperature is controlled via a jacket or internal coil. Foam is controlled by antifoam agents, though these can interfere with downstream filtration and must be minimized.

Shear stress is a major concern. Stirred-tank bioreactors use marine or pitched-blade impellers; tip speed should be kept below 2 m/s for mammalian cells. Sparging can cause bubble-associated damage; the addition of Pluronic F-68 (a non-ionic surfactant) at 0.1–1 g/L protects cells by coating the cell membrane and reducing bubble-cell adhesion.

Harvest and Clarification

Harvest separates cells and debris from the product-containing supernatant. The standard sequence is centrifugation followed by depth filtration. Disc-stack centrifuges operate at 7,000–10,000 × g and remove the bulk of cells. Depth filters, typically using diatomaceous earth as the filter medium, remove remaining particulates and some colloids. A final 0.2 µm sterile filter precedes the downstream train.

For intracellular products (e.g., E. coli-produced inclusion bodies), harvest includes cell lysis. High-pressure homogenization at 800–1,000 bar is the standard method, followed by centrifugation to collect inclusion bodies. The inclusion bodies are then washed, solubilized with 6–8 M urea or guanidine hydrochloride, and refolded by controlled dilution or dialysis.

Downstream Processing: Purification and Formulation

Cell Lysis and Inclusion Body Processing

For E. coli expression systems, the product often accumulates as insoluble inclusion bodies. After harvest, cells are lysed by high-pressure homogenization (e.g., 2–3 passes at 800 bar). Inclusion bodies are recovered by centrifugation at 10,000–15,000 × g, washed with buffer containing 1–2% Triton X-100 to remove membrane contaminants, and then solubilized.

Solubilization typically uses 6 M guanidine hydrochloride or 8 M urea with a reducing agent such as dithiothreitol (DTT) or β-mercaptoethanol. Refolding is the critical and often yield-limiting step. The solubilized protein is diluted 20–100-fold into a refolding buffer containing an oxidizing agent (e.g., oxidized glutathione) to promote disulfide bond formation. Refolding yields of 10–40% are typical; the remainder aggregates and must be removed by subsequent chromatography.

Chromatography Techniques

Chromatography is the workhorse of downstream purification. The three main modes are affinity, ion exchange, and hydrophobic interaction chromatography (HIC).

Affinity chromatography exploits a specific biological interaction. Protein A chromatography is the standard capture step for mAbs: protein A binds the Fc region of IgG with high specificity and affinity (Kd ≈ 10⁻⁸ M). The column is loaded with clarified harvest, washed to remove unbound impurities, and eluted with low pH (3.0–3.5) using 100 mM glycine or citrate buffer. Protein A yields >95% purity in a single step, but the resin is expensive ($10,000–$15,000 per liter) and requires 100–200 cycles of use to be economical.

Ion exchange chromatography separates based on surface charge. Cation exchange (CEX) uses negatively charged resins (e.g., sulfopropyl groups) to bind positively charged proteins; anion exchange (AEX) uses positively charged resins (e.g., quaternary ammonium groups) to bind negatively charged proteins. CEX is commonly used in intermediate purification after protein A; AEX is often used in flow-through mode as a polishing step to remove DNA, endotoxins, and residual HCPs.

Hydrophobic interaction chromatography separates based on surface hydrophobicity. Proteins bind to the resin at high salt concentration (1–2 M ammonium sulfate) and elute as salt is decreased. HIC is useful for removing aggregates and product-related variants.

A typical mAb purification train is:

Protein A affinity capture
Low-pH hold (viral inactivation, 60 min at pH 3.5)
Cation exchange chromatography (bind-and-elute)
Anion exchange chromatography (flow-through)
Viral filtration (20 nm)
UF/DF into formulation buffer

Viral Inactivation and Filtration

Viral safety is a regulatory requirement for all biologics. The standard approach is a three-pronged strategy: (1) screening of raw materials, (2) viral inactivation, and (3) viral removal. Low-pH incubation (pH 3.0–3.8 for 30–60 min at room temperature) inactivates enveloped viruses by disrupting the lipid membrane. Solvent/detergent treatment (e.g., 0.3% tri-n-butyl phosphate and 1% Triton X-100) achieves the same effect.

Viral filtration uses 20 nm or 15 nm pore-size filters that retain viruses by size exclusion. These filters are placed late in the process, after most impurities have been removed, to avoid fouling. The filters are single-use and can process 100–500 L/m² before clogging. A validated process must demonstrate a log reduction factor (LRF) of ≥4 for relevant viruses.

Final Formulation and Fill-Finish

The final step is buffer exchange and concentration by ultrafiltration/diafiltration (UF/DF). The product is concentrated to the target protein concentration (typically 10–100 mg/mL for mAbs) and the buffer is exchanged to the final formulation buffer. Common formulation buffers include 10–20 mM histidine or citrate at pH 5.5–6.5, with 150 mM sodium chloride and 0.01–0.05% polysorbate 80 as a stabilizer.

The formulated drug substance is then sterile-filtered through a 0.2 µm filter and filled into vials or pre-filled syringes. Fill-finish is performed under aseptic conditions in a classified environment (ISO 5). The entire process, from harvest to fill, typically takes 5–10 days for a mAb.

Key Differences Between Upstream and Downstream

Scale and Throughput

Upstream operates at the largest scale in the facility. A 10,000 L bioreactor produces 10,000 L of harvest, but after protein A capture, the product is concentrated to perhaps 100 L. Downstream operations therefore handle progressively smaller volumes but with exponentially higher product concentrations. This difference dictates equipment design: upstream uses large stainless steel or single-use vessels; downstream uses chromatography columns, filters, and smaller tanks.

Biological vs. Physical/Chemical Processes

Upstream is fundamentally a biological process. The cells are living organisms that respond to their environment; variability in cell growth, metabolism, and productivity is inherent. Downstream is primarily a physical and chemical process. Chromatography, filtration, and viral inactivation follow predictable physico-chemical principles, and the operations are more reproducible.

This distinction has practical consequences. Upstream processes require careful control of biological parameters (pH, temperature, dissolved oxygen) and are subject to biological variability. Downstream processes are more deterministic but are limited by physical constraints such as column pressure limits, filter capacity, and mass transfer.

Quality Control Focus

Upstream quality control focuses on cell culture parameters: viable cell density, viability, metabolite concentrations (glucose, lactate, glutamine, ammonia), and product titer. Downstream quality control focuses on purity: HCP levels (measured by ELISA, target <100 ppm), residual DNA (<10 ng/dose), endotoxin (<5 EU/kg/hour), aggregates (<5%), and product variants.

The regulatory framework treats the two phases differently. Upstream processes are validated for consistency and genetic stability; downstream processes are validated for impurity clearance and viral safety. Both are subject to Good Manufacturing Practice (GMP), but the specific validation studies differ.

Integration of Upstream and Downstream Processes

Impact of Cell Culture Conditions on Purification

Upstream decisions directly affect downstream performance. High cell densities produce more HCPs and DNA in the harvest, which can overload the capture column. Cell viability at harvest matters: if viability drops below 70%, intracellular proteases are released, degrading the product and generating fragments that are difficult to remove downstream.

The choice of media components also matters. Antifoam agents can foul depth filters and chromatography resins. Pluronic F-68, while protective to cells, can bind to hydrophobic interaction chromatography resins and reduce capacity. Some media supplements, such as yeastolate, introduce HCPs that must be cleared downstream.

The most significant upstream-to-downstream link is product titer. A doubling of titer from 2 to 4 g/L does not double the downstream cost; the cost of downstream processing per gram actually decreases because the same chromatography columns and filters process more product per cycle. However, higher titers also mean more impurities to remove, and the impurity-to-product ratio may change.

Continuous Bioprocessing and Perfusion

Traditional bioprocessing is batch-wise: each unit operation is completed before the next begins. Continuous bioprocessing integrates upstream and downstream into a single, uninterrupted flow. Perfusion cell culture continuously produces harvest, which is fed directly into a continuous capture step (e.g., periodic counter-current chromatography). This approach reduces equipment size, increases productivity, and improves product quality by reducing residence time.

The main challenge is process control. A continuous process requires real-time monitoring of product titer, impurity levels, and column capacity. It also requires robust automation and a higher level of process understanding. The regulatory framework for continuous manufacturing is still evolving, but the FDA has expressed support for the approach.

Process Analytical Technology (PAT)

Process Analytical Technology is a framework for designing, analyzing, and controlling manufacturing processes through the measurement of critical process parameters (CPPs) and critical quality attributes (CQAs). PAT tools include in-line sensors (pH, dissolved oxygen, Raman spectroscopy), at-line analyzers (HPLC, cell counters), and multivariate data analysis.

In upstream processing, Raman spectroscopy can monitor glucose, lactate, and cell density in real time, enabling automated feeding. In downstream processing, in-line UV absorbance at 280 nm monitors protein concentration during chromatography, and in-line pH and conductivity sensors track buffer transitions. PAT is not just about sensors; it is a philosophy of process understanding that enables real-time release testing and reduces the reliance on end-product testing.

Applications in Biotech Industry

Monoclonal Antibody Production

Monoclonal antibodies are the dominant product class in biopharmaceuticals, with global sales exceeding $150 billion annually. The standard platform is CHO cell culture in fed-batch mode, followed by protein A capture and two polishing chromatography steps. The process is highly standardized, and the main differentiation between manufacturers is in titer, yield, and cost per gram.

The upstream process for a typical mAb uses a CHO cell line with the GS expression system. The production bioreactor runs for 12–14 days at 36.5°C, pH 7.1, and 40% dissolved oxygen. Glucose is fed to maintain 2–4 g/L, and lactate is controlled by temperature shift (to 33°C) when it exceeds 2 g/L. Harvest occurs at viability >80%, and the clarified harvest is loaded onto protein A at 20–40 mg of mAb per mL of resin.

Downstream, the protein A eluate is adjusted to pH 3.5 and held for 60 minutes for viral inactivation. After neutralization, the product is purified by CEX (bind-and-elute) and AEX (flow-through). The final UF/DF step concentrates the product to 50 mg/mL in 20 mM histidine, pH 6.0, with 150 mM NaCl and 0.02% polysorbate 80. Overall process yield is typically 60–80%.

Vaccine Manufacturing

Vaccines use a different upstream/downstream split because the product is often a particle (virus, virus-like particle, or inactivated pathogen) rather than a soluble protein. Upstream for viral vaccines uses either egg-based (influenza), mammalian cell (Vero, MDCK), or insect cell (baculovirus) systems. The product is intracellular or cell-associated, requiring cell lysis and multiple clarification steps.

Downstream for vaccines relies heavily on ultracentrifugation, size-exclusion chromatography, and density gradient centrifugation. The product is often formulated with adjuvants (e.g., aluminum hydroxide) in the final step. Viral inactivation for vaccines uses formalin or β-propiolactone, which cross-link viral proteins and destroy infectivity.

The key challenge in vaccine DSP is that the product is large and fragile. Shear stress, pH extremes, and freeze-thaw cycles can damage virus particles. Process yields are often lower than for mAbs, and the cost of goods is dominated by downstream processing.

Recombinant Enzymes and Hormones

Recombinant enzymes (e.g., DNase, tissue plasminogen activator) and hormones (insulin, growth hormone) are produced in E. coli, yeast, or mammalian cells. Insulin is produced in E. coli as inclusion bodies, refolded, and purified by reverse-phase HPLC. The process involves multiple chromatography steps because the product is small and the impurities (including misfolded forms) are similar in size and charge.

Growth hormone is produced in E. coli and purified by a combination of ion exchange and size-exclusion chromatography. The product is formulated as a lyophilized powder for stability. Enzymes such as DNase (Pulmozyme) are produced in CHO cells and purified by a combination of affinity and ion exchange chromatography.

Methods Used to Study and Optimize Bioprocesses

Scale-Down Models and High-Throughput Screening

Scale-down models are small-scale systems that mimic the performance of production-scale equipment. For upstream, the standard scale-down model is the 250 mL–2 L benchtop bioreactor, which reproduces the pH, dissolved oxygen, and mixing characteristics of larger vessels. The key is to match the volumetric mass transfer coefficient (kLa) and the impeller tip speed to the production scale.

For downstream, scale-down models use small chromatography columns (1–5 mL resin volume) and small filter discs. These models are used to screen resins, optimize loading conditions, and determine binding capacities. High-throughput screening uses 96-well plates with miniature columns or filter plates to test many conditions in parallel. A typical screening experiment might test 10 resins × 5 buffer pH values × 3 salt concentrations = 150 conditions in a single day.

Design of Experiments (DoE)

Design of Experiments is a statistical methodology for planning experiments that efficiently map the relationship between process parameters and product quality. Instead of varying one factor at a time, DoE varies multiple factors simultaneously. A central composite design for a chromatography step might vary pH (5.0–7.0), conductivity (10–30 mS/cm), and load density (20–60 mg/mL resin) in 20 runs, generating a response surface model that predicts yield and purity as a function of all three parameters.

DoE is now expected by regulators for process characterization. The output is a design space—a multidimensional region where the process is robust to parameter variation. Operating within the design space does not require regulatory approval for changes; operating outside it does.

Process Modeling and Modeling and Simulation

Process modeling uses mathematical equations to describe unit operations. For chromatography, the lumped kinetic model describes the breakthrough curve and predicts column performance as a function of flow rate, resin properties, and binding kinetics. For filtration, the pore-blocking model predicts flux decline and filter lifetime.

Mechanistic models are increasingly used for process development. A model of a protein A column can predict the effect of load volume, flow rate, and elution pH on yield and purity, reducing the number of experiments needed. These models are validated against experimental data and then used for in-silico process optimization.

Common Pitfalls and Practical Considerations

Shear Sensitivity in Upstream

A common mistake is assuming that all cells respond identically to shear. CHO cells are more shear-sensitive than E. coli or yeast. High impeller speeds or aggressive sparging can reduce viability and increase HCP levels in the harvest. The solution is to measure the actual shear rate in the bioreactor and match it to the cell line's tolerance. Using Pluronic F-68 at 0.5–1 g/L is standard, but the concentration should be optimized for each cell line.

Buffer and Column Mismanagement

Downstream failures are often traceable to buffer preparation errors. A 0.1 pH unit error in a binding buffer can reduce column capacity by 30–50%. Conductivity errors of 1–2 mS/cm can cause premature elution or poor binding. The solution is to verify pH and conductivity of every buffer before use and to use in-line sensors for critical steps.

Column packing is another common failure point. Poorly packed columns show channeling, leading to early breakthrough and reduced capacity. The standard test is the height equivalent to a theoretical plate (HETP), which should be <0.03 cm for a well-packed column. Asymmetry factor should be between 0.8 and 1.2.

Regulatory and Quality Compliance

The most costly pitfall is inadequate process validation. Regulators expect a thorough understanding of how each process parameter affects product quality. This requires a formal risk assessment (e.g., Failure Mode and Effects Analysis, FMEA) and a DoE-based characterization study. Skipping these steps leads to regulatory rejection or, worse, a recall.

Another common issue is inadequate viral clearance documentation. Each viral inactivation and removal step must be validated with relevant viruses, and the log reduction factors must be documented. The total viral clearance must exceed the estimated viral load in the starting material by a comfortable margin (typically >6 logs).

Frequently Asked Questions

What is the difference between upstream and downstream in bioprocessing?

Upstream processing covers all steps from cell line development through cell culture and harvest—the operations that produce the biological product. Downstream processing covers all steps from harvest through purification, formulation, and fill—the operations that isolate and prepare the product for clinical use. The boundary is the harvest step, where cells are separated from the product-containing fluid.

What are examples of upstream and downstream processes?

Upstream examples include cell line transfection and clone selection, media preparation, seed train expansion, bioreactor operation, and harvest by centrifugation and depth filtration. Downstream examples include protein A affinity chromatography, low-pH viral inactivation, ion exchange chromatography, viral filtration, and ultrafiltration/diafiltration.

How do upstream and downstream processes relate in biology?

In molecular biology, "upstream" and "downstream" refer to positions relative to a gene's transcription start site. Upstream sequences (promoters, enhancers) regulate gene expression; downstream sequences are transcribed later. In bioprocessing, the terms refer to the physical flow of material through a manufacturing process. The two usages are unrelated.

What are the main challenges in downstream processing?

The main challenges are impurity clearance (HCPs, DNA, endotoxins, viruses, aggregates), product yield (each step loses 5–20% of product), and cost (chromatography resins and filters are expensive). Process robustness is also a challenge: columns must perform consistently across hundreds of cycles, and filters must not clog prematurely.

What are the main challenges in upstream processing?

The main challenges are cell line stability (productivity can decline over generations), metabolic byproduct accumulation (lactate and ammonia inhibit growth), and scale-up (conditions that work in a 2 L bioreactor may not translate to 10,000 L). Contamination is a constant risk, and a single contamination event can cost months of production time.

Why is integration of upstream and downstream important?

Upstream decisions determine the quantity and quality of the material entering the downstream train. High titers reduce downstream cost per gram but may increase impurity load. Cell viability at harvest affects HCP levels and product degradation. Continuous processing requires tight integration of both phases, with real-time monitoring and control.

What is the role of chromatography in downstream processing?

Chromatography is the primary purification method in bioprocessing. Affinity chromatography (e.g., protein A) provides high-purity capture in a single step. Ion exchange chromatography removes charge-based impurities. Hydrophobic interaction chromatography removes aggregates and variants. Size-exclusion chromatography is used for final polishing and buffer exchange. Each mode exploits a different physicochemical property of the product.

Key Takeaways

Upstream processing generates the product; downstream processing purifies it. The two phases have different goals, scales, and challenges.
The harvest step is the interface between upstream and downstream; its quality directly determines downstream performance.
Protein A chromatography is the standard capture step for monoclonal antibodies, achieving >95% purity in one step.
Viral safety requires a three-pronged strategy: raw material screening, viral inactivation (low pH or solvent/detergent), and viral removal (filtration or chromatography).
Upstream decisions—titer, viability, media composition—have direct consequences for downstream cost and yield.
Continuous bioprocessing and Process Analytical Technology are transforming the field, enabling real-time control and reduced cost.
Design of Experiments and scale-down models are essential tools for process development and regulatory submission.
The most common failures are buffer errors, column packing issues, and inadequate viral clearance documentation.