I lost a full week of method development once because a peptide standard kept “disappearing”. Recovery sat at 60%, then 55%, then lower as I diluted further. The mobile phase was fine, the column was fine, and the culprit was the nylon syringe filter I had been using out of habit. Nylon binds peptides. This article exists so you can skip that particular week.
The Short Answer
For protein and peptide samples, choose a low protein-binding membrane: PES (polyethersulfone) is the default for aqueous buffers and cell culture, hydrophilic PVDF is the versatile option for mixed aqueous-organic samples, and regenerated cellulose suits trace work needing very low extractables. Avoid nylon and mixed cellulose ester (MCE), which bind proteins strongly. Use 0.22 µm when the filtrate must be sterile (our sterile vs non-sterile filters guide covers the packaging and handling differences), 0.45 µm for routine clarification, and the smallest filter diameter your volume allows, because hold-up volume and membrane area both eat precious sample.
Why the Membrane Matters More Than the Pore Size
Pore size decides what particles get through. Membrane chemistry decides what gets left behind. For dilute proteins and peptides, adsorption is usually the bigger loss.
A membrane’s internal surface area is enormous relative to its footprint, so a small difference in surface affinity becomes a large difference in recovery. Charged residues and hydrophobic patches on your peptide interact with the membrane the same way they interact with any other surface. The sterile filtration of peptide solutions guide from Peptide Research Reviews makes the point sharply: two filters with identical 0.22 µm ratings can deliver very different amounts of peptide to the receiving vial.
Membrane-by-Membrane Breakdown
PES (polyethersulfone)
hydrophilic, fast flow, very low protein binding. The default for aqueous buffers, media, and most bioanalytical work. Its weakness is solvent tolerance; strong organics are not its territory.
Hydrophilic PVDF
low binding with better solvent range than PES, which is why it dominates mixed aqueous-organic protein work and western-blot heritage applications. The Millex syringe filter family from MilliporeSigma positions Durapore PVDF explicitly as their lowest-adsorption membrane for protein-rich solutions.
Regenerated cellulose (RC)
extremely low adsorption and low extractables across both aqueous and organic samples. The quiet specialist for trace and formulation analysis.
Cellulose acetate (CA)
very low protein binding but aqueous-only; good for protein solutions that never leave water.
Nylon
mechanically robust and broad-solvent, and precisely why it is everywhere in organic analysis. Its polyamide surface binds proteins through hydrogen bonding and charge. For dilute peptides, it is disqualified. (Yes, I learned this the expensive way.)
MCE (mixed cellulose ester)
binds proteins avidly. Nitrocellulose, its cousin, is literally the standard blotting membrane because it captures protein so well. Using a capture medium as a passive filter is asking for 40% recovery.
PTFE
hydrophobic and solvent-proof, meant for organic solvents and vents. It will not even wet with aqueous sample without alcohol prewetting, and it has no place in aqueous protein work.
The Hawach syringe filter selection guide summarizes the same hierarchy in a selection table if you want a printable version for the bench.
Pore Size and Diameter: The Second Decision
Pore size: 0.45 µm is enough for routine clarification ahead of analytical columns packed with 3-5 µm media. Step down to 0.22 µm for UHPLC, sub-2 µm columns, or when the filtrate itself must be sterile-grade. The finer pore costs you flow speed and clogs sooner on dirty matrices.
Diameter: this is where protein samples punish defaults. A 25 mm filter has more membrane area (more binding sites) and more housing dead volume (more retained sample) than a 13 mm. For small, precious samples, the smallest filter that passes your volume without stalling is the correct one. On a 500 µL plasma extract, hold-up volume is not a rounding error; it is a meaningful fraction of your sample.
Three Techniques That Protect Recovery
- Discard the first fraction. The earliest filtrate meets an unsaturated membrane and gives up the most analyte. Running the first 0.2-0.5 mL to waste means the filtrate you keep meets binding sites that are already occupied. If you cannot spare the volume, prewet with buffer instead, which fixes flow and bubbles but does not saturate binding sites.
- Centrifuge first for dirty matrices. Precipitated protein and particulates eat filter capacity. A quick spin removes the bulk load so the membrane only handles the fines, which extends throughput and reduces the pressure that forces unwanted adsorption. If clogging keeps recurring anyway, our article on why syringe filters clog walks the upstream causes.
- Match the filter to the volume. The instinct that bigger filters are safer is backwards for small samples. More area binds more peptide; more housing keeps more liquid. Small sample, small filter.
Peptide-Specific Notes
Peptides add one wrinkle that bulk proteins do not: short, hydrophobic sequences can bind low-binding membranes too, just more slowly. If you work near the detection limit with amphipathic peptides, two adjustments help.
Prewet and equilibrate the membrane with your actual sample matrix (buffer plus the organic modifier you run), not just water or buffer. The wetting liquid occupies hydrophobic sites that would otherwise grab your peptide. And run a simple recovery check on every new filter lot: one sample filtered and unfiltered, same dilution, same run. A five-minute side-by-side on a fresh lot catches the membrane changes that spec sheets never mention.
Heavily hydrophobic peptides sometimes fare better with a small organic prewash step or a filter pre-rinsed with the mobile phase. There is no universal answer here; there is only your peptide, your membrane, and the recovery number between them. Measure it once per lot and you will never wonder.
Method Validation Notes
If the method is regulated, filtration belongs in your validation the same as any other sample-prep step: run recovery comparisons (filtered vs unfiltered or filtered vs centrifuged) across your concentration range, and verify the membrane does not contribute extractables at your detection wavelength. The first-millilitre discard rule interacts with low-UV detection (below 220 nm) more than people expect, since membrane wetting agents show up as ghost peaks there.
Conclusion
Protein and peptide filtration is one of the few areas where the cheapest filter in the drawer is systematically the wrong one. Lock the membrane chemistry first (PES or hydrophilic PVDF as defaults, nylon and MCE off the table for dilute proteins), then right-size the pore for your column and the diameter for your volume, then use the discard-first-fraction trick when recovery matters. A three-minute check of the membrane code on the filter package protects weeks of downstream quantitation, which is the best time-to-benefit ratio in sample prep. For the broader filter-choice picture, our guide to nylon vs PTFE vs PVDF vs PES membranes covers the solvent-compatibility side, and our syringe filter pore size explainer goes deeper on when 0.22 versus 0.45 µm actually matters.
Frequently Asked Questions
Which syringe filter is best for protein samples?
Low protein-binding membranes: PES for aqueous buffers and media, hydrophilic PVDF for mixed aqueous-organic samples, cellulose acetate or regenerated cellulose for aqueous or trace work respectively. Avoid nylon and MCE entirely for dilute proteins, since both bind proteins strongly enough to cut recovery dramatically.
Why is my peptide recovery low after filtration?
The most likely cause is membrane adsorption, especially with nylon filters or oversized filter diameters on small volumes. Switch to a low-binding membrane, use the smallest diameter your volume allows, and discard the first 0.2-0.5 mL of filtrate so the filtrate you keep passes over binding sites that are already saturated.
What pore size should I use for protein samples before HPLC?
0.45 µm is adequate for routine clarification ahead of 3-5 µm analytical columns. Use 0.22 µm when running UHPLC or sub-2 µm columns, or when the filtrate needs to be sterile-grade. The finer pore flows slower and clogs faster, so do not pay that cost without a reason.
Can I prewet a PTFE filter for aqueous protein samples?
You can prewet hydrophobic PTFE with methanol and flush with water, but there is no good reason to: PTFE’s solvent resistance is wasted on aqueous protein work, and hydrophilic low-binding membranes (PES, PVDF) do the job without the prewetting ritual. Choose the membrane for the sample, not for habit.
Does filtration remove endotoxin from protein solutions?
No. Endotoxin molecules are far smaller than 0.22 µm and pass freely through standard syringe filters; sterile filtration removes particles and organisms, not dissolved contaminants. Endotoxin control requires dedicated removal resins or ultrafiltration steps designed for that purpose.
