Peptide Adsorption to Labware and How It Changes Concentration

A peptide solution has two concentrations: the one calculated from mass and volume, and the one actually present in the liquid when a pipette enters it. Between the two sits every surface the solution has touched. Peptides adsorb to glass, polypropylene, polystyrene, pipette tips and autosampler vials, and the proportional losses reported in the analytical literature are worst exactly where they matter most — in dilute solutions, in small volumes, in protein-free diluents. This guide covers the surface chemistry behind the loss, which sequences are worst affected, what the measured losses actually are, and which mitigations have evidence behind them.

Adsorption is a partition, not a leak

Nothing escapes a closed tube. What happens is that the peptide distributes between two phases — the bulk solution and the solid–liquid interface — and the interface has a finite capacity. That capacity is set by wetted area rather than by the amount dissolved, and the bound is what makes concentration matter: a wall that can hold at most some quantity of peptide takes an unnoticeable share of a stock at milligrams per milliliter and can take a large share of a standard at nanograms per milliliter. How much it actually takes at a given working concentration is not fixed. Uptake falls as the free concentration falls, and how steeply the fraction lost changes with concentration depends on how near the surface is to saturation, which varies with the peptide and the material. The direction is general; the size of the proportional error is a measurement, not a calculation.

The surfaces involved are not exotic. A review of strategies for reducing aspecific adsorption in liquid chromatography–mass spectrometry bioanalysis identifies nonspecific binding at pipette tips, sample tubes and LC vials, as well as at autosampler, column and mass spectrometer components, where it also produces carryover that biases both identification and quantification ^[1]. That review makes a point worth repeating before any recommendation below: because peptides and proteins span such a wide range of physicochemical properties, no single general approach to minimizing adsorption is available, and the review's response is to set out a range of strategies that can be generically applied rather than a single rule ^[1]. Which of them helps a given peptide is a question this guide treats as one for testing, not one that the review settles.

Two surfaces, two mechanisms

Bench labware presents two chemically distinct kinds of surface, and a peptide that behaves well on one may behave badly on the other.

Hydrophobic polymers

Polystyrene and polypropylene present largely nonpolar surfaces. An in situ study of LK14 — a 14-residue amphiphilic peptide built from leucine and lysine — compared adsorption onto hydrophobic polystyrene and hydrophilic silica from a pH 7.4 buffer using quartz crystal microbalance, atomic force microscopy and sum frequency generation spectroscopy. On polystyrene, adsorption was fast, complete in under two minutes, and the sum frequency spectra showed CH resonances from the leucine side chains at the interface; on silica the process was slower and multistep, taking more than thirty minutes, and only NH modes were visible, indicating a different average orientation ^[2]. The authors confined these conclusions to LK14 and to their two model surfaces, but the shape of the result is instructive: on a hydrophobic polymer, the nonpolar face of an amphipathic peptide turns toward the wall, and it does so quickly.

Glass and silica

Glass is not inert. Its surface carries silanol groups that deprotonate to negatively charged siloxide, so a basic peptide is drawn to it electrostatically. Work on amorphous silica nanoparticles between 15 and 450 nm found that between 5% and 20% of silanol groups were ionized at pH 7, with acidity increasing with particle size, and that positively charged peptides were attracted through ion pairing of protonated N-termini and lysine and arginine side chains with siloxide groups; hydrogen bonding, ion–dipole and van der Waals contributions were also implicated, and noncationic peptides containing serine, histidine and aspartate adsorbed as well ^[3]. The same study reported an inverse correlation between the peptide concentration required for measurable adsorption and the peptide's isoelectric point ^[3]. These are nanoparticle surfaces, not vial walls, and the numbers should not be transplanted onto a borosilicate vial without checking — but the mechanism explains why cationic sequences are the ones that vanish into glass.

Electrostatics alone does not close the books. An ellipsometry and Monte Carlo study of histatin 5, an intrinsically disordered cationic peptide, on hydrophilic silica found that although attraction between the positive peptide and the negative surface was important and grew in importance with pH, the simulations required an additional non-electrostatic short-ranged attraction of roughly 2.9 kBT per amino acid to bring the simulated surface coverages close to the experimental values ^[4]. In other words, a purely charge-based prediction underestimates how much sticks.

Which sequences are worst affected

Three properties recur in the measurements.

Hydrophobicity. In a multiple-reaction-monitoring study of peptides from a standard protein digest, many peptides did not show a linear relationship between signal intensity and amount under standard conditions, and the degree of non-linearity was related to observed retention time and predicted hydrophobicity ^[5]. Raising the organic content of the sample solvent improved the accuracy and precision of quantification, which the authors took to suggest that the non-linearity was concentration-dependent surface adsorption ^[5].

Length. A method-development study on two neuropeptides set out the difficulties faced when quantifying larger neuropeptides of 23 to 36 residues compared with smaller ones of fewer than 15 residues, and described reaching the required quantification limits for these compounds in biological samples as an extremely challenging task, especially because of their nonspecific binding ^[6].

Net charge. Basic sequences are the ones drawn to silica, through the ion-pairing mechanism described above ^[3], and glass presents the same silanol chemistry. A high isoelectric point is therefore a warning flag for glass and other silica-containing labware, and a high predicted hydrophobicity is a warning flag for polymer labware.

What does not exist, so far as the analytical literature goes, is a validated general predictor. The review cited above is explicit that no generic approach covers the diversity of these molecules ^[1], and the hydrophobicity correlation reported for peptides from a standard protein digest was a correlation, not a calibrated model ^[5]. Treat sequence properties as a reason to test, not as a substitute for testing.

What the measured losses actually are

The most complete single dataset on labware comes from a study of colistin, a cationic lipopeptide, in a simulated time-kill setup with no bacteria and no plasma proteins present, over a tested range of 0.125 to 8 mg/L. Across soda-lime glass, polypropylene and polystyrene test tubes, the measured starting concentrations ranged from 44% to 102% of expected, and after 24 hours the range was 8% to 90%; half-lives of loss ran from 0.9 to 12 hours ^[7]. Fitting a Langmuir model gave maximum binding capacities between 0.4 and 1.1 µg/cm² across the three materials ^[7]. In one of two polystyrene microplates tested, starting concentrations fell below the lower limit of quantification at worst ^[7]. For measuring colistin concentrations in experimental settings, the authors concluded that low-protein-binding polypropylene should be used when possible; their results discouraged the use of polystyrene, and they recommended minimizing the number of dilution steps when diluting colistin in protein-free media ^[7].

Two figures anchor the low end. Adsorption of amyloid-β(1–42) during preanalytical handling of cerebrospinal fluid was measured at approximately 5% loss per tube transfer, rising to about 10% in small volumes; loss was detectable after 30 seconds of contact and after contact with a pipette tip alone ^[8]. And a recombinant His-tagged human glycoprotein was found to lose roughly 25% to 35% of the material in solution to polypropylene microcentrifuge tubes even after a short incubation, with a measurable effect on downstream functional assays ^[9].

Losses of a few percent per step compound. Three transfers at 5% each leave 0.95³ = 0.857 of the original, a total loss of 14.3% rather than 15%; three transfers at 10% each leave 0.729, a loss of 27.1% rather than 30%.

Why dilute standards drift

Two things change together when a working standard is made small and dilute: the amount of peptide falls, and the wetted area per unit volume rises. Both push in the same direction.

Consider an idealized cylindrical tube of internal radius 0.40 cm — a geometric simplification, not a measurement of any particular product. At a 1.00 mL fill, the liquid column is 1.00/(π × 0.40²) = 1.989 cm tall, and the wetted area is the base plus the wall, π × 0.40² + 2π × 0.40 × 1.989 = 0.503 + 5.000 = 5.503 cm², or 5.50 cm² per mL. At a 0.100 mL fill, the column is 0.199 cm tall and the wetted area is 0.503 + 0.500 = 1.003 cm², or 10.03 cm² per mL. The smaller fill therefore presents about 1.8 times as much surface per unit volume.

Now suppose — again as an illustration, not a measured value — that the wall takes up 0.05 µg/cm² from a nominal 1.00 µg/mL solution. The 1.00 mL fill contains 1.000 µg and loses 5.503 × 0.05 = 0.275 µg, leaving 0.725 µg/mL: 27.5% low. The 0.100 mL fill contains 0.100 µg and loses 1.003 × 0.05 = 0.050 µg, leaving 0.0499 µg in 0.100 mL, or 0.499 µg/mL: 50.1% low. Same tube, same peptide, same nominal concentration; the reported number differs by a factor of about 1.45 purely because of fill volume. This fixed-uptake model overstates the loss at the small volume, because uptake in reality falls as the free concentration falls — it is an upper bound, not a prediction.

The direction is confirmed experimentally. Aliquot volume changed measured concentrations of amyloid-β(1–42) in cerebrospinal fluid: 0.5 mL aliquots read higher than 0.1 mL aliquots by two independent immunoassays, and a mass spectrometry assay confirmed that recovery was linked to sample volume ^[10]. In the cerebrospinal fluid work cited earlier, small volumes roughly doubled the per-transfer loss ^[8].

The practical consequence is that a calibration curve built by serial dilution in a protein-free diluent can bend at its low end, and the bend is a property of the labware, not of the detector.

Carrier proteins, surfactants and competitive blocking

The oldest fix is to give the surface something else to bind. In the neuropeptide method above, adding 0.05% plasma as an adsorption competitor was described as essential to avoiding peptide loss to nonspecific binding ^[6]. Surfactants are the other route. Two mechanisms are generally described for surfactant-mediated stabilization of proteins: preferential location of the surfactant at nearby interfaces, which precludes protein adsorption, and association of surfactant with protein into complexes that keep the protein away from surfaces ^[11].

Neither fix is complete, and the literature says so. Tween-20 supplementation increased amyloid-β(1–42) recovery, but it did not effectively resolve the concentration differences associated with aliquot size ^[10]. A carrier protein or a surfactant is a competitor for surface sites, not a barrier.

Both carry costs. A carrier protein adds a large excess of extraneous material to a mass spectrometry sample and can interfere with binding assays; a surfactant can suppress ionization and contaminate a chromatographic system. Whether an additive is acceptable is a property of the assay, not of the peptide — a matter of convention decided case by case rather than a finding from any specific study.

Low-bind plastics: what the label does and does not promise

"Low-binding" and "low-retention" are vendor descriptors rather than standardized specifications, and we are not aware of a common test method behind them. Vendors rarely publish the surface treatment. The honest framing is that the label indicates a product intended to adsorb less, and the evidence for how much less is analyte-specific.

Where it has been measured, the direction is favorable but partial. Low-protein-binding microtubes held colistin at 63% to 99% of expected at the start and 59% to 90% at 24 hours, against 44% to 102% and 8% to 90% for the large test tubes of the same study. The floor rose sharply while the ceiling barely moved: the worst case is what improved, and the tubes are still lossy ^[7]. Among a panel of low-retention polypropylene tubes, 1.5 mL Eppendorf LoBind tubes were the least absorptive for amyloid-β(1–42) ^[10]. In a comparison of eight polypropylene tubes across three concentration pools, three tubes gave significantly higher recovery than the other five at all three concentrations, and infrared spectra taken from the cut-open tubes distinguished the materials ^[12]. A separate biospecimen study reported that a non-irradiated homopolymer type of polypropylene had the lowest adsorption properties in their comparison ^[13].

There is a real tension here worth flagging rather than smoothing over. The cerebrospinal fluid study that measured 5% loss per transfer found that tube brand, vortexing and continuous tube movement did not influence adsorption in their system ^[8], while the tube-comparison studies found brand and polypropylene grade to make a significant difference to recovery ^{[10, 12, 13]}. The studies used different tube panels, matrices and assays. The safe reading is that tube identity can matter a great deal and sometimes does not, which is an argument for fixing the part number and lot in your protocol rather than for assuming any particular tube is safe.

Handling steps that add loss

Contact time is short. Adsorption of amyloid-β(1–42) was detectable after 30 seconds, and contact with a pipette tip alone was enough to produce it ^[8]. Every transfer is a fresh surface. The colistin study's recommendation to minimize the number of dilution steps in protein-free media follows directly ^[7], as does the observation that raising the organic content of a sample solvent improved quantitative linearity ^[5].

Several further habits are conventional rather than evidence-based, and should be labeled as such: pre-rinsing a tip or tube with the sample solution before the transfer that counts; choosing the largest fill volume the protocol tolerates; making the working standard fresh rather than storing it dilute; and preparing dilutions in the container that will be measured from, rather than transferring afterward. These follow logically from the mechanism, but we are not aware of primary studies that quantify each of them for peptides.

What is not known

There is no validated sequence-based model that predicts, for an arbitrary peptide and an arbitrary labware material, what fraction will be lost. Predicted hydrophobicity correlated with non-linear response in one digest dataset ^[5], length was identified as a source of difficulty in one two-peptide method development ^[6], and isoelectric point correlated with adsorption in the silica nanoparticle work ^[3], but none of these has been turned into a calibrated tool. Published measurements are dominated by a handful of analytes — amyloid-β peptides, colistin, a few neuropeptides — and each was characterized in a single matrix, so how far any of these numbers travel to a different sample type is untested. Kinetics vary by more than an order of magnitude even within one study: half-lives of colistin loss spanned 0.9 to 12 hours across three materials ^[7]. Whether adsorption is reversible on a useful timescale, and how much of the adsorbed layer can be recovered by rinsing with an organic or acidified solvent, is not established generally. And the manufacturing variables that make one polypropylene grade better than another — irradiation, and homopolymer versus copolymer — have been reported to matter ^[13] but are not disclosed on most product labels.

Measure it, then, rather than assume it: a recovery check in your own tubes, your own diluent and your own concentration range is the only thing that turns any of the above into a number you can defend.