How do you characterise proteins in complex biological matrices?

To characterise proteins in a complex biological matrix, whether serum, plasma, cell lysate, fermentation broth or another crude, unpurified sample, the most direct approach is to measure in free solution, with neither binding partner immobilised and without first purifying the protein. Flow Induced Dispersion Analysis (FIDA), the in-solution method developed by Fida Biosystems (Fidabio), does exactly this. It follows the diffusion of a fluorescent indicator as it is carried through a microcapillary in laminar flow, and derives binding affinity, hydrodynamic size, concentration and aggregation state from how that indicator behaves within the sample itself.

The matrix is not stripped away; it becomes the measurement environment. That makes it possible to study a protein close to its native biological state, in the presence of the endogenous partners, competitors and solution conditions that govern its behaviour, rather than in a buffer that only approximates them.

What makes complex matrices hard to measure?

A complex matrix is any sample in which the molecule of interest is embedded among many others: serum and plasma, cell lysates, cell-culture and fermentation supernatants, cerebrospinal fluid, and food or protein hydrolysates. Four properties of these samples confound most biophysical methods.

Molecular crowding and dynamic range. Serum is dominated by albumin and immunoglobulins; a lysate carries the entire host proteome. Any method that reads a bulk signal, such as total light scattering or total protein absorbance, integrates over all of these species, and the analyte of interest is lost against an overwhelming background.

Non-specific adsorption. Methods that immobilise one binding partner on a sensor surface expose that surface to every component of the sample. Matrix proteins and lipids adsorb non-specifically, contributing signal unrelated to the interaction under study.

Optical background. Biological fluids absorb and fluoresce. Lysates in particular are strongly autofluorescent, and the background scales with matrix concentration.

Altered physicochemical conditions. Matrices differ from buffer in viscosity, ionic strength and pH, and each of these perturbs any measurement that relies on molecular mobility.

The conventional remedy is purification. It removes the interferents, but it also removes the biological context: endogenous binding partners, competing ligands and the physiological milieu that shapes conformation, oligomeric state and binding.

How does FIDA characterise proteins without purifying them?

FIDA isolates one signal against everything else. The capillary is filled with the sample (the analyte), and a nanolitre plug of a fluorescent indicator is injected into it. As the plug is mobilised under laminar, pressure-driven flow, Taylor dispersion broadens it in a way set by the indicator's diffusion coefficient, and the detector records the resulting dispersion profile, the Taylorgram. The width of that profile yields the diffusion coefficient and, through the Stokes–Einstein relation, the hydrodynamic radius. When the indicator binds its target in the sample, its apparent diffusivity falls and its apparent size increases. The principle is set out in What is FIDA?

Five features of this design make complex matrices tractable.

Selectivity resides in the indicator. The detector registers only the fluorescent indicator. The matrix is present throughout, but unless it binds the indicator it leaves the measured size unchanged. The indicator can be a labelled ligand, antigen or antibody, a fluorescent peptide, or a protein expressed as a fusion with a fluorescent protein such as GFP.

The matrix contributes only a baseline. Because the capillary is filled with sample, the matrix's own fluorescence appears as a constant baseline beneath the indicator peak, and the size is extracted from the peak rather than the baseline. How to measure protein binding in serum or plasma? illustrates this.

No sensor surface to foul. Both partners remain free in solution, so there is no functionalised surface for matrix components to adsorb to. The capillary wall is the only interface, and adsorption to it is reported directly as peak asymmetry, the stickiness readout, rather than propagating as a hidden error.

Viscosity is measured, not assumed. A matrix more viscous than buffer slows every species within it. FIDA determines the viscosity of each sample in the same run and compensates for it, so that a change in size reflects binding rather than the solution. See What is sample viscosity?

Nanolitre consumption. Each measurement consumes nanolitres of sample, which matters when the matrix is a patient specimen or a single well of an expression screen.

Which properties can be determined in a complex matrix?

A single run returns several orthogonal readouts, all derived from the indicator:

Binding affinity (K_D), from the change in indicator size across a titration. See What is K_D?
Concentration of the binding partner, by interpolating an unknown sample on a standard curve, without a secondary detection reagent.
Binding kinetics (k_on and k_off), resolved in solution. See How to measure binding kinetics in solution?
Size, complex formation and oligomeric state, as hydrodynamic radius. See What is hydrodynamic radius?
Aggregation, detected as discrete spikes in the Taylorgram. See What is protein aggregation?
Sample quality: polydispersity, viscosity, stickiness and labelling quality, recorded with every run.

Which matrices have been measured with FIDA?

Serum and plasma

Serum and plasma are the most extensively published FIDA matrices. Studies range from quantifying human serum albumin in plasma to profiling autoantibody binding in plasma from patients with systemic lupus erythematosus, determining the affinity and concentration of donor-specific HLA antibodies in pre-transplant patient serum, and measuring anti-drug antibodies together with residual drug activity in serum from patients treated with adalimumab. FIDA has also been used to follow how the signalling lipid sphingosine 1-phosphate partitions between albumin and HDL in plasma from patients with sepsis, and to size the immune complexes that the autoantigen TRIM21 assembles with circulating antibodies in plasma from patients with Sjögren's disease. Binding kinetics have been resolved in human serum as well. These studies, and guidance on serum and plasma concentrations, are covered in How to measure protein binding in serum or plasma?

Cell lysates

Lysates answer whether a newly expressed protein is functional before it has been purified. In Screening de novo designed protein binders in unpurified lysate (Protein Science, 2025), computationally designed binders were screened against a fluorescent target directly in bacterial lysate, so that hits could be identified without purifying each design. The authors also characterise how lysates perturb the measurement, which is directly useful for assay design: autofluorescence at low dilution, peak splitting when lysate and running buffers are mismatched, and longer residence times from elevated viscosity, together with how the lysis method affects the false-positive rate.

In Intracellular TDP-43 amyloid nucleates from arrested nascent condensates (bioRxiv, 2026), a fluorescently tagged C-terminal domain of TDP-43, the protein implicated in ALS and frontotemporal dementia, was sized directly in clarified yeast lysate to establish whether it remained monomeric or assembled into small clusters at different expression levels. In FibrilPaint to determine the length of Tau amyloids in fluids (PNAS, 2025), a fluorescent fibril-binding peptide was used to measure Tau fibril size in buffer, in bacterial lysate and in human serum, allowing the same readout to be compared across matrices.

Fermentation broth and culture supernatant

In bioprocessing, the matrix is the harvest itself. In-Solution IgG Titer Determination in Fermentation Broth (ACS Omega, 2020) measured antibody titre in clarified CHO-cell broth, using a small fluorescently labelled anti-IgG affibody as the indicator, with no purification step and with results in agreement with protein A purification and UV quantification. The broth proved an asset rather than an obstacle: its proteins suppressed adsorption of the indicator to vials and capillary. The poster Clone Selection from unpurified samples and the application note Fidabio Clone Selection Assay rank nanobody-expressing clones on expression level and affinity simultaneously, directly in culture supernatant.

Protein hydrolysates and food matrices

In Degree of hydrolysis is a poor predictor of the sensitizing capacity of whey- and casein-based hydrolysates (Food Research International, 2024), whey and casein hydrolysates of the kind used in infant formula, each a heterogeneous mixture of peptides, were sized label-free and screened for aggregates on a Fida 1, as part of establishing which physicochemical properties relate to allergenic potential.

Cerebrospinal fluid and blood

Amyloid fibril size has been measured directly in patient blood and cerebrospinal fluid as a route to the early detection of Parkinson's disease. How can you measure amyloid fibril length directly in cerebrospinal fluid or blood? tells that story.

How do you set up a FIDA assay in a complex matrix?

Choose a selective indicator. The indicator should bind its target with high affinity and specificity, because any binding to other matrix components also changes its apparent size. Some small molecules bind serum albumin, fluorescein among them; that is how albumin itself was quantified in plasma in the Analyst 2015 study, whose authors note that such indicators may also bind other plasma proteins. A control run of the free dye or unconjugated indicator in the matrix shows whether this happens, as in the sepsis study of sphingosine 1-phosphate carriers.

Make the target the only fluorescent species of interest. Label-free detection reads the intrinsic fluorescence of tryptophan and tyrosine, so it reports every protein or peptide that contains them. That suits purified proteins, or mixtures that are themselves the object of study, such as the whey and casein hydrolysates above. The published serum, plasma and lysate studies all used a fluorescent indicator. In a protein-rich matrix, a fluorescently labelled indicator or a fluorescent fusion protein restores specificity. See How to label a protein for FIDA measurements?

Dilute as far as the background requires, and hold it constant. Matrices have been measured from a few percent to undiluted. Keep the matrix fraction identical across samples and standards, so that results remain directly comparable.

Match the buffers. A mismatch in ionic strength or pH between sample and running buffer can split the indicator peak. A matrix-matched running buffer avoids it.

Monitor the quality readouts. Viscosity, stickiness and aggregate spikes are recorded in every run and flag when the matrix is influencing the result. Where the indicator adsorbs to the capillary, a coated capillary or a low concentration of non-ionic surfactant usually resolves it; see How do you stop protein sticking to surfaces?

Frequently asked questions

Does a protein need to be purified before FIDA analysis?

No. FIDA measures the fluorescent indicator against the whole sample, so its binding partner can be characterised in serum, plasma, lysate or fermentation broth without purification. Selectivity comes from the indicator binding its target, not from removing everything else.

Does the background fluorescence of serum or lysate interfere with FIDA?

Matrix fluorescence appears as a constant baseline beneath the indicator peak, and the size is extracted from the peak. Very high background, as in concentrated lysate, lowers the signal-to-noise ratio, so the matrix is diluted to a level at which the indicator peak remains well resolved.

Can FIDA measure label-free in serum or plasma?

Label-free FIDA reads the intrinsic tryptophan and tyrosine fluorescence of every protein in the sample, so in serum or plasma it would report the matrix proteins alongside the target. For a target-specific measurement in a protein-rich matrix, a fluorescently labelled indicator or a fluorescent fusion protein is used.

Does the viscosity of the matrix affect the size measurement?

In general, viscosity slows diffusion, so an uncorrected viscous sample would report an inflated size. This is why the Fidabio Software determines the viscosity of each sample within the same run and compensates for it, so the reported hydrodynamic radius reflects the molecule. This viscosity measurement and compensation ensures precise size measurement.

How much sample does a measurement in a complex matrix need?

A single FIDA measurement consumes nanolitres of sample, leaving the remainder of the vial or well for replicates. See How much sample do you need to measure binding affinity and size?

References

Aragonès Pedrola, J., et al. (2025). FibrilPaint to determine the length of Tau amyloids in fluids. Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.2502847122

Bøgh, K. L., et al. (2024). Degree of hydrolysis is a poor predictor of the sensitizing capacity of whey- and casein-based hydrolysates in a Brown Norway rat model of cow's milk allergy. Food Research International, 181, 114063. https://doi.org/10.1016/j.foodres.2024.114063

Hug, M. N., et al. (2023). HLA antibody affinity determination: From HLA-specific monoclonal antibodies to donor HLA specific antibodies (DSA) in patient serum. HLA. https://doi.org/10.1111/tan.15047

Jones, E. L., et al. (2026). The autoantigen TRIM21 assembles proinflammatory immune complexes after lytic cell death. Science Immunology. https://doi.org/10.1126/sciimmunol.ads9680

Otzen, D. E., Buell, A. K., & Jensen, H. (2021). Microfluidics and the quantification of biomolecular interactions. Current Opinion in Structural Biology, 70, 8–15. https://doi.org/10.1016/j.sbi.2021.02.006

Pedersen, M. E., Østergaard, J., & Jensen, H. (2020). In-solution IgG titer determination in fermentation broth using affibodies and flow-induced dispersion analysis. ACS Omega. https://doi.org/10.1021/acsomega.0c00791

Pedersen, M. E., Østergaard, J., Glintborg, B., Hetland, M. L., & Jensen, H. (2022). Assessment of immunogenicity and drug activity in patient sera by flow-induced dispersion analysis. Scientific Reports. https://doi.org/10.1038/s41598-022-08682-3

Pinheiro, F., et al. (2025). Screening de novo designed protein binders in unpurified lysate using flow induced dispersion analysis. Protein Science, 34, e70286. https://doi.org/10.1002/pro.70286

Poulsen, N. N., et al. (2015). Flow induced dispersion analysis rapidly quantifies proteins in human plasma samples. Analyst, 140, 4365–4369. https://doi.org/10.1039/c5an00697j

Poulsen, N. N., et al. (2016). Flow-induced dispersion analysis for probing anti-dsDNA antibody binding heterogeneity in systemic lupus erythematosus patients. Analytical Chemistry, 88, 9056–9061. https://doi.org/10.1021/acs.analchem.6b01741

Seidita, A., et al. (2024). Rapid determination of sphingosine 1-phosphate association with carrier molecules by flow induced dispersion analysis to predict sepsis outcome. iScience. https://doi.org/10.1016/j.isci.2024.111168

Willmer, P., et al. (2025). In-solution characterization of biomolecular interaction kinetics under native conditions. Analytical Chemistry. https://doi.org/10.1021/acs.analchem.5c02164

Wu, J., et al. (2026). Intracellular TDP-43 amyloid nucleates from arrested nascent condensates. bioRxiv. https://doi.org/10.64898/2026.02.10.705006

Related resources

The review Microfluidics and the quantification of biomolecular interactions (Current Opinion in Structural Biology, 2021) sets out why measuring size and binding in solution suits plasma, serum and fermentation broth. For patient samples, the publications Assessment of immunogenicity and drug activity in patient sera and Rapid determination of sphingosine 1-phosphate association with carrier molecules show complete clinical workflows. Prof. Markus Gräler describes the sepsis work, carried out in unpurified patient plasma, in the blog Using FIDA for sepsis and lipid-signalling research: a conversation with Prof. Markus Gräler.

Want to know whether FIDA can measure your protein in its own matrix? Book a discovery call.