TECHNOLOGY EXPLAINER
Chromatography (the purification step in biopharmaceuticals)
— what picks the antibody out is the design of a porous bead
An antibody grown in culture sits in a liquid mixed with cellular proteins, DNA and viruses. What raises it to pharmaceutical purity is chromatography, the downstream purification step. The lead player is the porous bead (the resin, or media) a few tens of micrometres across packed into a column. Base matrix, particle size, pores, how the ligand is immobilised, resistance to caustic cleaning — almost all of the performance is set by the material.
- What chromatographic purification is (the short version)
- The purification sequence — the antibody platform
- Five separation modes — affinity, ion exchange, hydrophobic, size exclusion, mixed mode
- Designing the medium — base matrix, particle size, pores
- Our calculation: how long diffusion inside a bead takes
- A materials engineer's view (1): particle size, pores and pressure pull against each other
- Ligands — the molecular design of Protein A, immobilisation, dynamic binding capacity and caustic stability
- A materials engineer's view (2): a ligand is a functional material that wears out
- Membrane chromatography and monoliths — from diffusion to convection
- A materials engineer's view (3): surface area against mass transfer
- What is still hard, and what this article could not confirm
- Glossary / References / Claim-to-source audit
Sourced = stated in published material or a paper (link given)
Our calculation = a figure this article derived, with the assumptions spelled out
Not yet confirmed = a plan, an outlook or research-stage work with no confirmed practical track record
Structural summaries and readings from a materials-design standpoint are marked separately as Commentary.
Product names are given as fact where papers cite them as objects of comparison, and are not recommendations of any particular product.
1. What chromatographic purification is (the short version)
- What it does: liquid is passed through a column packed with porous beads, and species are separated by differences in how readily they stick, or in their size (our commentary)
- For antibodies: the usual sequence is a large purification step first, Protein A affinity, which captures only the antibody out of the culture fluid, followed by one or two chromatographic steps to take out what remainsSourced
- The materials side: base matrices fall broadly into glass or silica, agarose, and organic polymer. A different matrix or particle size changes how much can be captured at the same flow rate (the dynamic binding capacity)Sourced
Inside a bead, an antibody can only move inwards by diffusion. In a study comparing 15 Protein A media, dynamic binding capacity depended strongly on residence time: agarose-based media gave high capacity above three minutes of residence time, while the small-particle polymer and glass media depended far less on flow rateSourced. The productivity of purification is set by a dimension of the material: the diffusion distance in a porous body (our commentary).
2. The purification sequence — the antibody platform
Liu and colleagues describe the antibody purification sequence as follows: Protein A chromatography is placed after harvest and gives a product of relatively high purity. Only small amounts of process- and product-related impurities remain, and one or two chromatographic steps are used for polishing. A low-pH hold after Protein A and a virus filtration step are also generally includedSourced. Shukla and colleagues at Amgen report a flexible, generic purification platform exploiting the biochemical similarity of antibodies and Fc fusion proteins, with data from 14 products, and identify a highly selective capture step using Protein A as its coreSourced.
- The low-pH elution doubles as a viral measure: Liu and colleagues note that elution from Protein A requires a low pH of 2.5 to 4, and that these conditions effectively inactivate enveloped viruses such as murine retrovirusesSourced
- Anion exchange is used in flow-through: operating pH is typically 8 to 8.2 and conductivity up to around 10 mS/cm, conditions chosen so that the product does not bind while acidic impurities such as nucleic acids and host cell proteins doSourced
3. Five separation modes — affinity, ion exchange, hydrophobic, size exclusion, mixed mode
| Mode | What it separates on | Use and characteristics as described in the literature |
|---|---|---|
| Affinity (Protein A) | A ligand specifically recognises the Fc part of the antibody | A cell-surface protein of Staphylococcus aureus. Each of its five repeated domains interacts with the Fc of IgG. Elution is at pH 2.5 to 4. Alternatives have been proposed because the resin is expensive and its tolerance of extreme conditions is somewhat limited |
| Cation exchange (CEX) | Positively charged molecules bind to a negatively charged surface | Has the resolution to reduce antibody variants and high-molecular-weight species (aggregates), such as deamidated, oxidised and N-terminally truncated forms |
| Anion exchange (AEX) | Negatively charged molecules bind to a positively charged surface | Powerful for removing host cell proteins, DNA, endotoxin, leached Protein A, dimers and aggregates, and endogenous retroviruses and parvoviruses. Used in flow-through |
| Hydrophobic interaction (HIC) | Binding of hydrophobic regions exposed as salt weakens hydration | Salt in the buffer interacts with water molecules, reducing protein hydration and exposing hydrophobic regions. Elution is by a descending salt gradient. Used in flow-through, it can remove most aggregates at high yield |
| Size exclusion (SEC) | Molecular size (whether a molecule can enter the pores) | Unlike the other modes, ideally there is no adsorption (ΔH = 0) and partitioning is governed by entropy alone. Widely used to quantify aggregates. Sample volume is ideally 5 to 10% of the column volume; beyond that, resolution falls |
| Mixed mode (multimodal) | A combination of ionic, hydrogen-bonding and hydrophobic interactions | Resins with multimodal ligands can be used in antibody polishing. Two or more physicochemical properties are used to raise the specificity of the protein–ligand interaction |
| Hydroxyapatite | Metal affinity for calcium plus cation exchange at the phosphate groups | Used for IgG purification since the 1950s, with ceramic media developed in the 1980s. An elution strategy has been reported that reduced antibody aggregates from over 60% to below 0.1% |
All Sourced (Protein A: Liu et al. [Reference 1], Linhult et al. [Reference 6]; CEX, AEX and HIC: Liu et al. [Reference 1]; SEC: Hong et al. [Reference 13]; mixed mode: Liu et al. [Reference 1], Kallberg et al. [Reference 12]; hydroxyapatite: Gagnon [Reference 14]).
4. Designing the medium — base matrix, particle size, pores
A chromatographic medium (resin, media) is a porous bead with ligands immobilised on its internal surface. Its performance is set by the type of base matrix, the particle size, the pore size, and the density and manner of ligand attachment (our commentary).
Three families of base matrix (plus inorganic ceramics)
| Base matrix | What the literature says |
|---|---|
| Agarose-based | Liu and colleagues divide Protein A resins into three by base matrix, one of them being agarose-based (Protein A Sepharose Fast Flow, MabSelect and others)Sourced. The use of agarose spheres in chromatography was reported by Hjertén in 1964Sourced. In the comparison by Hahn and colleagues, agarose media had high equilibrium binding capacity and high dynamic binding capacity above three minutes of residence timeSourced |
| Glass and silica | One of Liu and colleagues' categories (Prosep vA, Prosep vA Ultra and others)Sourced. Hahn and colleagues describe ProSep-vA Ultra as porous glass with a pore size of 70 nmSourced. In analytical SEC, modified porous silica — mechanically strong, non-swelling and inert over a wide range of conditions — became dominant in the 1970sSourced |
| Organic polymer | Another of their categories (polystyrene–divinylbenzene media such as Poros A and MabCapture)Sourced. In Hahn and colleagues' comparison, polymer and glass media had smaller particles and larger apparent rate constants (faster mass transfer), so they depended less on flow rate and residence timeSourced |
| Ceramic (hydroxyapatite) | Ceramic hydroxyapatite media were developed in the 1980s. Here the base matrix itself does the binding, through the metal affinity of calcium and cation exchange at the phosphate groupsSourced |
On SEC, Hong and colleagues write that soft polymeric resins are compressed by pressure and flow, which limits how far the particle size can be reducedSourced. In other words, the mechanical strength of the base matrix sets the lower limit on usable particle size (our commentary).
5. Our calculation: how long diffusion inside a bead takes
In 2003 Hahn and colleagues compared 15 Protein A media and showed that dynamic binding capacity depends strongly on residence timeSourced. In a 2005 follow-up on two improved agarose media and one porous glass medium, they found that dynamic binding capacity did not change once the residence time was at least four minutes, and that mass transfer was well described by a model of film and pore diffusionSourced. Below we check, to an order of magnitude, why the answer comes out in minutes.
- Assumption 1: take the hydrodynamic diameter of IgG as 10.7 nm (the value in the review by Li and MooneySourced)
- Assumption 2: obtain the free diffusion coefficient from the Stokes–Einstein equation. At 25 °C with a water viscosity of 0.89 mPa·s, D0 = kT / (6πηr) is about 4.6 × 10-11 m²/sOur calculation
- Assumption 3: inside the pores, walls and tortuosity slow diffusion down. Here we assume an effective diffusion coefficient one tenth of the free value (4.6 × 10-12 m²/s) (this ratio varies greatly with pore structure)
- Assumption 4: take the time to reach the centre as t ≈ R² / De, with R the particle radius
- Particle size 30 µm (R = 15 µm) gives about 49 seconds
- Particle size 50 µm (R = 25 µm) gives about 136 seconds (about 2.3 minutes)
- Particle size 70 µm (R = 35 µm) gives about 267 seconds (about 4.5 minutes)
- Particle size 90 µm (R = 45 µm) gives about 442 seconds (about 7.4 minutes)
Assumptions and limits: the particle sizes are values set here for illustration, not those of any particular product. The ratio for the effective diffusion coefficient (one tenth) is also an assumption. Because an antibody adsorbs to the pore walls as it travels, real breakthrough behaviour is more complex. The calculation is there to show the order of magnitude and the proportionality: at these particle sizes, diffusion takes minutes, and halving the particle size cuts the time to a quarter.
The calculated diffusion times (a few minutes) come out at the same order as the residence times at which capacity stabilised in the papers (3 to 4 minutes)Our calculation. The direction also agrees with Hahn and colleagues' explanation that mass transfer is faster in media with smaller particlesSourced.
6. A materials engineer's view (1): particle size, pores and pressure pull against each other
The calculation behind Fig. 5 brings the trade-offs in medium design into view (our commentary).
- Make the particle smaller: the diffusion distance shrinks and capacity holds up at higher flow rates. But the pressure drop across the packed bed rises with the inverse square of particle size (the general Kozeny–Carman relation)
- To withstand that pressure: the base matrix has to be hard. As Hong and colleagues write, soft polymeric resins are compressed, which limits how far the particle size can be reducedSourced
- Make the pores larger: the antibody enters more easily and diffuses faster. But the internal surface area falls and capacity with it (our commentary)
A medium is therefore a porous body in which particle size distribution, pore size distribution, mechanical strength and specific surface area all have to be optimised together. That is precisely the problem familiar from designing catalyst supports, adsorbents and separation membranes.
In Hahn and colleagues' comparison the contrast was clear: agarose media had high equilibrium binding capacity but needed a long residence time, while polymer and glass media had smaller particles and held their capacity at higher flow ratesSourced. "Holds a lot but is slow" or "fast but holds less" — choosing the base matrix is also choosing the productivity of the step, in throughput per hour (our commentary).
And the technology for making porous spheres of uniform size, with controlled pores, strong enough not to break, in quantity is an extension of things chemical manufacturers already do: suspension polymerisation, emulsification, sol–gel processing and the manufacture of porous glass.
7. Ligands — the molecular design of Protein A, immobilisation, dynamic binding capacity and caustic stability
(1) Molecular design of the ligand
According to Linhult and colleagues, Protein A is a cell-surface protein of Staphylococcus aureus, and each of its five repeated domains interacts with the Fc fragment of IgG (and with some Fab fragments). One problem in large-scale purification is that protein ligands are sensitive to alkaline conditions. Protein A is comparatively alkali-tolerant, but a protein engineering strategy of substituting asparagine residues was used so that it would last longer under the harsh conditions of cleaning in place (CIP)Sourced. The synthetic Z domain that this work built on was designed by Nilsson and colleagues in 1987. Because it contains no asparagine–glycine sequence and no methionine, unlike native Protein A it withstands treatment with hydroxylamine and cyanogen bromideSourced.
(2) Resistance to caustic cleaning
Liu and colleagues record that agarose and polymer resins can be cleaned with 0.05 to 0.2 N sodium hydroxide for a 30-minute contact after every cycle, and that an alkali-stabilised Protein A-derived ligand tolerating harsh cleaning solutions of 0.1 to 0.5 N sodium hydroxide was developed in the agarose-based MabSelect SuReSourced. The 2005 paper by Hahn and colleagues likewise treats MabSelect SuRe as a medium functionalised with alkali-stabilised Protein ASourced.
(3) Immobilisation and spacers
As a classic method of chemically coupling a ligand to a polysaccharide support, Axén, Porath and Ernback reported immobilisation with cyanogen halides (cyanogen bromide) in 1967Sourced. In 2016, Müller and Vajda noted that the binding capacity of affinity resins is about 50% lower than that of modern ion exchange media, and that the high capacity of ion exchangers rests on spacer technology. On that basis they present oligomerisation (multimerisation) of Protein A domains as an effective way to raise capacity, analogous to the "tentacle" technology of ion exchange. They report that the dynamic binding capacity of a hexameric ligand was about twice that of a tetramer, and that raising the sodium phosphate buffer from 20 mM to 100 mM took it up to 130 mg/mL at mostSourced.
(4) Dynamic binding capacity (DBC)
Liu and colleagues put the dynamic binding capacity of Protein A at 15 to 50 g of antibody per litre of resin, varying with flow rate, the antibody being purified and the resin. Lowering the flow rate lengthens the residence time and raises the binding capacitySourced. Ghose and colleagues investigated why capacity differs between antibodies and Fc fusion proteins, showing that the trend in capacity follows the apparent size of the molecule and the static binding capacity rather than the pore diffusion coefficient, and that steric hindrance between immobilised Protein A molecules means capacity plateaus as ligand density is increasedSourced.
- Assumption: 2,000 L of culture fluid at an antibody concentration of 5 g/L, giving 10 kg of antibody (an assumed figure, not that of any particular process)
- At a DBC of 15 g/L: 10,000 g ÷ 15 g/L = about 667 L of resin (if processed in one cycle)Our calculation
- At a DBC of 50 g/L: 10,000 g ÷ 50 g/L = 200 LOur calculation
- Fix the resin at 50 L and the number of cycles needed is about 14 at 15 g/L and 4 at 50 g/LOur calculation (10,000 / (15 × 50) = 13.3, so 14 cycles; 10,000 / (50 × 50) = 4 cycles)
Assumptions and limits: in practice a column is not loaded to 100% of its DBC; a safety margin is kept. The culture volume and antibody concentration are our assumptions. The calculation is there to show the relation that a threefold difference in DBC means about a threefold difference in either resin volume or cycle count. Given the high cost of Protein A resin that Liu and colleagues point toSourced, a difference in capacity translates directly into a difference in cost (our commentary).
(5) Leaching of Protein A
Liu and colleagues cite work by Carter-Franklin and colleagues showing that intact Protein A leaches whether purified antibody or unpurified culture fluid is loaded, and that with unpurified culture fluid fragments of Protein A leach as well. Adding EDTA to the culture fluid to inhibit proteases reduces the amount that leachesSourced. Leached Protein A then becomes something for downstream steps such as anion exchange to removeSourced.
8. A materials engineer's view (2): a ligand is a functional material that wears out
Protein A resin is not a material used once and thrown away. Loading, washing, elution and caustic cleaning in place are repeated over many cycles. Each time, the ligand is exposed both to low pH (elution at pH 2.5 to 4) and to high pH (the NaOH clean)Sourced.
What makes this interesting as a materials story is that the route to caustic stability was to reduce the asparagine content — eliminating a weak point at the molecular levelSourced. It is the same thinking as raising the durability of a polymer by reducing the number of readily hydrolysed bonds, such as esters (our commentary).
Ghose and colleagues' result then shows that packing ligands ever more densely is not simply better. Steric hindrance between immobilised ligands means capacity plateaus as density risesSourced. Müller and Vajda, by contrast, report roughly doubling the capacity by linking domains so that they extend from the surface (multimerisation)Sourced.
In other words, how the ligands are arranged and how far they reach matters more than how many are attached to the surface. Polymer brushes, graft polymerisation, spacer-length design — for a materials maker with surface-modification technology, this is a direct point of contact (our commentary).
9. Membrane chromatography and monoliths — from diffusion to convection
If the weak point of a bead is diffusion inside it, then build a structure that does not need diffusion. That is the idea behind membrane chromatography and monoliths.
| Format | Advantages (as described in the literature) | Weaknesses (as described in the literature) |
|---|---|---|
| Packed beads | Agarose media have high equilibrium binding capacity. In improved products, capacity is stable above four minutes of residence timeSourced | Dynamic binding capacity depends strongly on residence timeSourced |
| Membrane chromatography | Dispenses with pores for diffusion and places the binding sites along through-pores. Because mass transfer depends on convection rather than diffusion, binding capacity is almost independent of flow rate (Liu et al.)Sourced. Being disposable, low in buffer consumption and light on equipment cost, it can substantially reduce purification costs (Orr et al.)Sourced | Uneven flow distribution, non-uniform membrane pore size and thickness, the availability of suitable scale-down models, and low binding capacity. The low capacity comes from small specific surface area and uneven flow (Liu et al.)Sourced. Compared with packed beds it is a relatively new and immature technology (Orr et al.)Sourced |
| Monoliths | Described as "the fourth generation of chromatographic materials". Plate height and dynamic binding capacity do not change with flow rate, because transport is convective through channels over 1000 nm in diameter. Columns of up to 8 L are commercially available, used to separate proteins, aggregates, plasmid DNA and viruses. Mostly moulded as blocks of polymethacrylate or polyacrylamide (Jungbauer and Hahn)Sourced | The absolute specific surface area is low. But the large channels give ample area for adsorbing large biomolecules that cannot enter the pores of a conventional medium (Jungbauer and Hahn)Sourced |
10. A materials engineer's view (3): surface area against mass transfer
Boiled down, the difference between beads, membranes and monoliths is a contest between pore size and surface area.
- The porous glass medium has a pore size of 70 nm (ProSep-vA Ultra, as described by Hahn and colleagues)Sourced. Against an IgG diameter of 10.7 nmSourced that is about 6.5 timesOur calculation
- Monolith channels are over 1000 nmSourced — more than about 90 times the size of IgG and more than about 14 times the pores of the glass mediumOur calculation
The larger the pore, the faster convection brings molecules in, but the smaller the surface area per unit volume, so for a mid-sized molecule such as an antibody the capacity falls short. Conversely, as Jungbauer and Hahn write, large molecules such as viruses and plasmid DNA cannot enter the pores of a conventional medium, so the large channels of a monolith actually offer more usable surface areaSourced.
The design principle, then, is not which format is better but matching the pore to the size of the molecule to be carried (our commentary). Beads for antibodies, monoliths or membranes for very large molecules such as viral vectors and plasmids — that division follows naturally from the same principle.
And the weaknesses listed for membrane chromatography — uneven flow, and variation in membrane thickness and pore sizeSourced — are precisely questions of membrane manufacturing quality. The casting technology for a narrow pore size distribution and uniform thickness has been honed in water treatment membranes and battery separators. For a membrane maker, bioprocess purification could be an outlet for the same technology at higher valueNot yet confirmed.
11. What is still hard, and what this article could not confirm
(1) The specifications of commercial media
The particle size, pore size, dynamic binding capacity and caustic stability (how many cycles they last) of individual commercial media are given in each supplier's technical literature, but this article has not checked company documents one by one, so only values stated in papers (such as the 70 nm pores of the glass medium) are given here. The particle sizes in Section 5 (30 to 90 µm) are assumptions for illustration.
(2) The price of Protein A resin
Liu and colleagues write that Protein A resin is expensiveSourced, but no specific price is given here because it could not be confirmed in the primary sources used. No market-research estimates have been used.
(3) The use of size exclusion chromatography in processing
SEC is widely used to analyse aggregatesSourced, but no primary source showing how far it is used in commercial manufacturing purification could be confirmed within the scope of this article.
(4) Newer formats such as continuous chromatography
Developments such as running several columns in rotation for continuous operation are covered in other parts of this series, for example our explainer on batch and continuous culture. They are not covered here.
- Antibody purification usually captures broadly with Protein A, then removes what remains in one or two polishing stepsSourced
- Base matrices fall broadly into agarose, glass/silica and organic polymerSourced
- Dynamic binding capacity depends strongly on residence time, with about four minutes a guide even for improved products. Calculating the diffusion time inside a bead gives the same order, in minutesOur calculation
- Protein A was made to withstand repeated caustic cleaning by molecular design, substituting asparagine residuesSourced
- Packing ligands more densely plateaus; multimerisation roughly doubled capacity in one reportSourced
- Membranes and monoliths deliver molecules by convection, so capacity does not depend on flow rate. The price is less surface areaSourced
- Particle size, pores, strength, surface modification — the performance of a medium is almost entirely the material design of a porous body (our commentary)
12. Glossary
- Downstream
- The separation, purification and formulation steps that follow culture (the upstream).
- Medium (resin, media)
- The porous beads packed into a column, with ligands immobilised on their internal surface.
- Ligand
- The molecule or functional group fixed to the surface of the medium that binds the target.
- Protein A
- A surface protein of Staphylococcus aureus that binds the Fc of antibodies and is used to capture them.
- Dynamic binding capacity (DBC)
- How much the medium can capture, with liquid actually flowing, before the target breaks through. Given in g/L.
- Residence time
- How long the liquid stays in the column. Lowering the flow rate lengthens it.
- Cleaning in place (CIP)
- Cleaning and sanitising equipment without dismantling it. Sodium hydroxide is commonly used on media.
- Host cell protein (HCP)
- Protein from the cells that made the antibody. An impurity to be removed.
- Flow-through
- An operating mode in which the product does not bind and only impurities are captured by the medium.
- Polishing
- The step after capture that removes the small amounts of impurity remaining.
- Membrane chromatography
- A format with ligands on the walls of a membrane's through-pores, separating as liquid passes through the membrane.
- Monolith
- A single moulded porous block, with liquid flowing through large channels.
- Mixed mode
- A ligand combining several interactions, such as ionic, hydrophobic and hydrogen bonding.
- Hydroxyapatite
- A form of calcium phosphate, used as a ceramic medium to remove antibody aggregates.
13. References (primary sources)
- Liu HF, Ma J, Winter C, Bayer R. "Recovery and purification process development for monoclonal antibody production", mAbs 2:480 (2010) (PMC) — pmc.ncbi.nlm.nih.gov
- Shukla AA et al. (Amgen) "Downstream processing of monoclonal antibodies — application of platform approaches", J Chromatogr B 848:28 (2007) — doi.org
- Hober S, Nord K, Linhult M. "Protein A chromatography for antibody purification", J Chromatogr B 848:40 (2007) — doi.org
- Hahn R, Schlegel R, Jungbauer A. "Comparison of protein A affinity sorbents", J Chromatogr B 790:35 (2003) — doi.org
- Hahn R et al. "Comparison of protein A affinity sorbents II. Mass transfer properties", J Chromatogr A 1093:98 (2005) — doi.org
- Linhult M et al. "Improving the tolerance of a protein A analogue to repeated alkaline exposures using a bypass mutagenesis approach", Proteins 55:407 (2004) — doi.org
- Nilsson B et al. "A synthetic IgG-binding domain based on staphylococcal protein A", Protein Eng 1:107 (1987) — doi.org
- Axén R, Porath J, Ernback S. "Chemical coupling of peptides and proteins to polysaccharides by means of cyanogen halides", Nature 214:1302 (1967) — doi.org
- Müller E, Vajda J. "Routes to improve binding capacities of affinity resins demonstrated for Protein A chromatography", J Chromatogr B 1021:159 (2016) — doi.org
- Jungbauer A, Hahn R. "Polymethacrylate monoliths for preparative and industrial separation of biomolecular assemblies", J Chromatogr A 1184:62 (2008) — doi.org
- Orr V et al. "Recent advances in bioprocessing application of membrane chromatography", Biotechnol Adv 31:450 (2013) — doi.org
- Kallberg K, Johansson HO, Bulow L. "Multimodal chromatography: an efficient tool in downstream processing of proteins", Biotechnol J 7:1485 (2012) — doi.org
- Hong P, Koza S, Bouvier ES. "Size-exclusion chromatography for the analysis of protein biotherapeutics and their aggregates", J Liq Chromatogr Relat Technol 35:2923 (2012) (PMC) — pmc.ncbi.nlm.nih.gov
- Gagnon P. "Monoclonal antibody purification with hydroxyapatite", N Biotechnol 25:287 (2009) — doi.org
- Li J, Mooney DJ. "Designing hydrogels for controlled drug delivery", Nat Rev Mater 1:16071 (2016) (PMC) — pmc.ncbi.nlm.nih.gov
- Ghose S, Hubbard B, Cramer SM. "Binding capacity differences for antibodies and Fc-fusion proteins on protein A chromatographic materials", Biotechnol Bioeng 96:768 (2007) — doi.org
- Hjertén S. "The preparation of agarose spheres for chromatography of molecules and particles", Biochim Biophys Acta 79:393 (1964) (PubMed) — pubmed.ncbi.nlm.nih.gov
14. Claim-to-source audit
| Claim in the text | Basis | Label |
|---|---|---|
| That Protein A is placed after harvest, with one or two polishing steps, a low-pH hold and virus filtration generally included. That elution at pH 2.5 to 4 inactivates enveloped viruses. That AEX flow-through runs at pH 8 to 8.2 and up to 10 mS/cm, binding acidic impurities. That CEX reduces variants and high-molecular-weight species. That AEX is powerful for removing HCP, DNA, endotoxin, leached Protein A, aggregates and viruses. The principle of HIC, elution by descending salt, and removal of aggregates in flow-through. That mixed-mode resins can be used in polishing. The three base matrix categories for Protein A resins and their examples. A DBC of 15 to 50 g/L, varying with flow rate, antibody and resin, and rising as flow rate falls. Cleaning with 0.05 to 0.2 N NaOH for 30 minutes, and MabSelect SuRe tolerating 0.1 to 0.5 N. Leaching of Protein A and its reduction with EDTA. The advantages of membrane chromatography (no diffusion pores, convective transport, capacity near-independent of flow) and its weaknesses. That Protein A resin is expensive | Reference 1 https://pmc.ncbi.nlm.nih.gov/articles/PMC2958570/ | Sourced |
| That Amgen's flexible, generic purification platform was shown with data from 14 products, with a highly selective Protein A capture step at its core | Reference 2 https://doi.org/10.1016/j.jchromb.2006.09.026 | Sourced |
| That Protein A was among the earliest immunoglobulin-binding molecules discovered, became one of the most used affinity purification systems, and was improved to withstand harsh industrial conditions | Reference 3 https://doi.org/10.1016/j.jchromb.2006.09.030 | Sourced |
| That in a comparison of 15 Protein A media, DBC depended strongly on residence time, with agarose media high above three minutes and polymer and glass media less dependent on flow rate. That agarose media have high equilibrium binding capacity. That the other media have smaller particles and faster mass transfer | Reference 4 https://doi.org/10.1016/s1570-0232(03)00092-8 | Sourced |
| MabSelect Xtra, MabSelect SuRe (alkali-stabilised Protein A) and ProSep-vA Ultra (porous glass with 70 nm pores). That the behaviour was described by a film and pore diffusion model. That DBC did not change above four minutes of residence time | Reference 5 https://doi.org/10.1016/j.chroma.2005.07.050 | Sourced |
| That Protein A is a surface protein of Staphylococcus aureus whose five repeated domains interact with Fc (and some Fab). That the alkali sensitivity of protein ligands is a problem. That substituting asparagine residues raised tolerance of CIP | Reference 6 https://doi.org/10.1002/prot.10616 | Sourced |
| That the synthetic IgG-binding domain of 1987 contains no Asn-Gly sequence and no methionine, and withstands hydroxylamine and cyanogen bromide | Reference 7 https://doi.org/10.1093/protein/1.2.107 | Sourced |
| The 1967 report of chemically coupling peptides and proteins to polysaccharides with cyanogen halides | Reference 8 https://doi.org/10.1038/2141302a0 | Sourced |
| That affinity resin capacity is about 50% lower than ion exchange, which rests on spacer technology. That multimerisation is analogous to tentacle technology. That a hexamer gives about twice a tetramer, up to 130 mg/mL in 100 mM phosphate buffer | Reference 9 https://doi.org/10.1016/j.jchromb.2016.01.036 | Sourced |
| That monoliths are called a fourth generation of materials; that plate height and DBC do not vary with flow rate; channels over 1000 nm; columns up to 8 L; polymethacrylate and polyacrylamide; and low specific surface area that is nonetheless ample for large biomolecules | Reference 10 https://doi.org/10.1016/j.chroma.2007.12.087 | Sourced |
| That membrane chromatography is a relatively new and immature technology, and that being disposable, low in buffer consumption and light on equipment cost it can reduce purification costs | Reference 11 https://doi.org/10.1016/j.biotechadv.2013.01.007 | Sourced |
| That mixed mode raises the specificity of the interaction using two or more physicochemical properties | Reference 12 https://doi.org/10.1002/biot.201200074 | Sourced |
| That SEC ideally has no adsorption and partitions on entropy alone; its use in quantifying aggregates; that sample volume is ideally 5 to 10% of column volume; that soft polymeric resins compress, limiting particle size reduction; and that modified porous silica became dominant in the 1970s | Reference 13 https://pmc.ncbi.nlm.nih.gov/articles/PMC3556795/ | Sourced |
| That hydroxyapatite has been used for IgG purification since the 1950s, with ceramic media developed in the 1980s; calcium metal affinity and phosphate cation exchange; and an elution strategy that reduced aggregates from over 60% to below 0.1% | Reference 14 https://doi.org/10.1016/j.nbt.2009.03.017 | Sourced |
| That the hydrodynamic diameter of IgG is 10.7 nm | Reference 15 https://pmc.ncbi.nlm.nih.gov/articles/PMC5898614/ | Sourced |
| That the trend in capacity follows the apparent size of the molecule and static binding capacity rather than the pore diffusion coefficient, and that steric hindrance between immobilised Protein A molecules makes capacity plateau as ligand density rises | Reference 16 https://doi.org/10.1002/bit.21044 | Sourced |
| The 1964 report of preparing agarose spheres for chromatography | Reference 17 https://pubmed.ncbi.nlm.nih.gov/14163524/ | Sourced |
| The free diffusion coefficient of IgG of about 4.6 × 10-11 m²/s and effective coefficient of 4.6 × 10-12 m²/s; the diffusion times at particle sizes of 30, 50, 70 and 90 µm (about 0.8, 2.3, 4.5 and 7.4 minutes); and the quarter-time relation on halving the particle size. Resin volumes of about 667 L and 200 L and cycle counts of about 14 and 4. That 70 nm pores are about 6.5 times the size of IgG, and 1000 nm more than about 90 times IgG and more than about 14 times the glass pores | Our calculation. The 25 °C and 0.89 mPa·s viscosity, the effective diffusion coefficient at one tenth of free diffusion, t ≈ R²/De, the particle sizes, and 2,000 L of culture at 5 g/L with 50 L of resin are all assumptions set by this article, not values for any particular product or process | Our calculation |
| That pressure drop scales with the inverse square of particle size (about fourfold on halving it) | Added by this article as the general relation given by the Kozeny–Carman equation. No individual primary source is cited for it here | Commentary |
| That bioprocess purification could be a high-value outlet for membrane makers | An outlook of this article; no track record has been confirmed | Not yet confirmed |
| The specifications of commercial media, the price of Protein A resin, and how far SEC is used in commercial manufacturing | Not stated, because no primary source could be confirmed within the scope of this article | Commentary |
| The six-step summary of the sequence, the diagram of separation modes, the three levels of the medium, the three-way trade-off, reading asparagine substitution alongside measures against hydrolysis in polymers, the points of contact with surface-modification and membrane-casting technology, the principle of matching pores to molecular size, and the wording "ultrafiltration to formulate" and "SEC mainly used in analysis" | Our summary and commentary based on published content. Not views expressed by the authors | Commentary |
| That Figs. 1, 2, 3, 6 and 7 are explanatory drawings, that Fig. 5 is a drawing including our calculation, and that the hero image and Fig. 4 are AI-generated images | Our note | Commentary |
Last updated 23 September 2026. Sources are limited to primary material (peer-reviewed original papers and reviews). No market-size, share or price estimates have been used. Because the article includes structural summaries and readings from a materials-design standpoint, those are marked as Commentary and kept separate from sourced fact. The specifications of commercial media, the price of Protein A resin, and how far SEC is used in commercial manufacturing are not stated here because they could not be confirmed in published primary sources. Product names are those cited in papers as objects of comparison and are not recommendations of any particular product. All figures are explanatory. Figs. 1, 2, 3, 6 and 7 are conceptual diagrams, Fig. 5 is a drawing including our calculation, and the hero image and Fig. 4 are AI-generated images; none of them pictures a real product, electron micrograph or process.