SPR Kinetic Artifacts: Mass Transport, Rebinding, Surface Heterogeneity, and Model Selection

An SPR sensorgram can be fitted precisely and still return the wrong kinetic constants. The fitted values are conditional on the physical experiment: analyte transport through the flow cell, ligand presentation on the surface, analyte homogeneity, reference subtraction, and the reaction model all contribute to the observed response. A low fitting error is therefore not sufficient evidence that the reported association and dissociation rates are intrinsic molecular properties.

When an SPR study produces unexpectedly slow dissociation, concentration-dependent rate constants, biphasic curves, or structured residuals, the first task is not to select a more elaborate model. It is to perturb the experiment in ways that distinguish mass transport, rebinding, surface heterogeneity, avidity, bulk refractive-index effects, and genuine multistep binding.

Assay-Configuration Dependence of SPR Kinetic Constants

For an ideal 1:1 interaction, the soluble analyte reaches a homogeneous immobilized ligand population rapidly enough that the concentration at the surface is effectively the injected bulk concentration. Association and dissociation can then be described by one association rate constant, one dissociation rate constant, and a shared maximum response. Real sensor surfaces often violate one or more of these assumptions.

The observable response can contain contributions from:

• diffusion and convection of analyte from bulk solution to the sensing layer;
• depletion of analyte close to a high-capacity surface;
• rebinding of dissociated analyte before it returns to bulk flow;
• orientation, crowding, partial activity, or chemical modification of immobilized ligand;
• multivalent analyte binding and avidity;
• analyte aggregates or mixed oligomeric states;
• nonspecific binding to the matrix or reference surface;
• buffer mismatch, injection dispersion, baseline drift, and incomplete regeneration.

These mechanisms can produce similar-looking sensorgrams. A two-state fit, for example, may absorb effects caused by ligand heterogeneity or transport without proving a conformational transition. Mechanistic interpretation should follow discriminating experiments, not the software menu.

Figure 1. An SPR response reflects the molecular reaction together with transport, surface presentation, and measurement effects.

Mass Transport Limitation in SPR Kinetic Analysis

Mass transport limitation occurs when analyte is captured at the surface faster than convection and diffusion replenish it. The concentration experienced by immobilized ligand then falls below the nominal injected concentration. Association becomes partly transport-controlled, so a fitted k_on can describe delivery to the surface rather than the intrinsic chemical association step.

The same condition affects dissociation. Molecules released from the ligand remain temporarily within the sensing layer and may bind again before being swept into bulk flow. This retention makes dissociation appear slower and can create non-exponential or biphasic decay. The resulting apparent residence time may be a property of the surface configuration rather than the complex in free solution.

Several experimental features increase transport risk: high ligand capacity, rapid association, large or slowly diffusing analytes, a thick hydrogel matrix, and insufficient flow. Analyte concentration alone is not a complete diagnostic because transport and binding remain coupled across the concentration series.

Flow-Rate and Ligand-Density Tests

Run the same analyte concentration over the same surface at multiple flow rates. A systematic change in the association profile or fitted rate constants is evidence that delivery contributes to the observed kinetics. Flow-rate independence supports—but does not by itself prove—reaction-limited behavior, because a study may not have perturbed flow sufficiently to expose the effect.

A second test uses surfaces with different ligand densities. If the fitted k_on, k_off, or apparent affinity changes with immobilization level, crowding, transport, rebinding, or surface heterogeneity is likely. Kinetic measurements generally benefit from a lower-capacity surface than concentration or qualitative binding assays. The appropriate response level must be determined from analyte mass, expected stoichiometry, instrument sensitivity, and the study objective rather than a universal RU target.

If higher flow and lower surface capacity reduce the artifact while preserving adequate signal, redesigning the experiment is preferable to relying solely on a mass-transport model. A transport term can be useful when limitation is moderate and the model is identifiable, but strongly transport-controlled data may not contain enough information to recover intrinsic rates reliably.

Rebinding and Apparent Slow Dissociation

Rebinding is often discussed as part of mass transport, but it deserves separate attention because it directly distorts k_off. A dissociated analyte may encounter the original ligand, a neighboring ligand, or a second binding site before leaving the sensor matrix. High ligand density and multivalent analytes increase this probability.

The diagnostic question is whether dissociation changes when opportunities for recapture are altered. Compare low- and high-density surfaces, increase flow during dissociation where the platform permits, and consider adding a soluble competitor that captures released analyte without perturbing the surface complex. A competitor experiment requires controls: the competitor must not change refractive index excessively, bind the immobilized ligand independently, or destabilize the complex by another mechanism.

Multivalent systems require additional restraint. An antibody engaging repeated epitopes can show an avidity-dominated dissociation phase that does not correspond to the monovalent affinity of one paratope. A bivalent-analyte model may describe the curve, but the parameters are meaningful only if analyte valency, ligand spacing, and surface mobility support that mechanism. When the required claim is intrinsic site affinity, a monovalent Fab, a monomeric target construct, or a solution-phase method may be more informative than fitting intact multivalent species.

An orthogonal solution assay can help determine whether the unusually slow SPR dissociation is surface-dependent. Microscale thermophoresis does not provide k_on and k_off, but it can test whether a solution-phase affinity estimate is compatible with the surface-derived equilibrium constant.

Surface Heterogeneity from Ligand Immobilization

Immobilization can create ligand populations with different activities. Random amine coupling may attach a protein through several accessible residues, place some molecules close to the matrix, modify residues near a binding site, or generate orientation-dependent steric constraints. A carboxymethylated dextran layer also presents a three-dimensional microenvironment rather than a perfectly uniform plane.

Surface heterogeneity commonly appears as systematic residuals: one part of the association phase is underfit while another is overfit, and the dissociation phase cannot be represented by a single exponential. Reproducible complexity does not prove two biological binding sites. It may reflect a distribution of surface states created during immobilization.

Test the surface hypothesis by changing how the ligand is presented. Capture-based or site-specific immobilization can improve orientation and preserve activity. Comparing direct coupling with capture, or preparing independent surfaces at lower density, is more discriminating than applying a heterogeneous-ligand model to one dataset. The active fraction also matters: total immobilized RU is not automatically equal to binding-competent ligand.

Regeneration can introduce time-dependent heterogeneity. Harsh conditions may progressively reduce activity, while incomplete regeneration leaves residual analyte and changes the available capacity in later cycles. Inspect pre-injection baselines, binding capacity, and control injections across the run. If surface performance drifts, a global fit that assumes one stable R_max across all cycles is physically inconsistent.

Analyte Quality, Avidity, and Nonspecific Binding

The soluble sample can be the source of complex kinetics. Aggregates create disproportionately large SPR responses and may engage multiple surface ligands. Mixed oligomeric states can mimic heterogeneous binding, and active concentration may differ from concentration determined by absorbance or a bulk protein assay.

Before kinetic interpretation, evaluate the analyte under the running-buffer conditions. Relevant checks include size-exclusion chromatography, light scattering, centrifugation or filtration where compatible, and a concentration series that remains free of precipitation or concentration-dependent oligomerization. The analysis should use independently prepared biological material when reproducibility of the molecular interaction is important; repeated injections from one unstable vial do not address reagent heterogeneity.

Nonspecific binding should be tested on a reference surface that resembles the active surface in matrix, immobilization chemistry, and ligand density. An empty or activated-and-blocked flow cell may not reproduce nonspecific interaction with an unrelated immobilized protein. Salt, detergent, carrier protein, pH, and surface chemistry can be optimized, but each change can also affect the intended interaction and must be documented.

For difficult systems, biolayer interferometry provides a related surface-based measurement with different fluidics and sensor architecture. Agreement across SPR and BLI is useful, but it is not fully orthogonal if both measurements retain the same valency or immobilization artifact.

Reference Subtraction and Bulk Refractive-Index Artifacts

SPR measures refractive-index change near the sensor surface. Buffer mismatch, solvent differences, glycerol, salt, DMSO, and sample matrix can therefore generate responses unrelated to binding. A reference channel corrects shared surface and bulk effects, while blank injections help remove systematic injection artifacts. Both are needed when the objective is high-quality kinetic fitting.

Double referencing is not a cosmetic processing step. The reference surface should experience the same injection sequence as the active surface, and blank cycles should be distributed through the run so that drift or changing carryover is visible. Large corrections relative to the binding response indicate that the experimental contrast is weak even if subtraction yields a smooth curve.

Solvent correction is particularly important for small-molecule work because small differences in DMSO concentration can exceed the binding response. Excluding the injection start from a fit may hide a bulk shift without resolving its effect on the association phase. Match samples and running buffer as closely as practical and use a validated correction series.

Figure 2. Experimental perturbations distinguish transport, rebinding, surface, analyte, and bulk-signal artifacts more reliably than model fit alone.

SPR Kinetic Model Selection and Residual Analysis

Model selection should begin with the simplest physically plausible description. Fit a globally linked 1:1 model across a concentration series, then inspect the raw traces, residual structure, parameter stability, and consistency with equilibrium response. A random residual distribution near the acquisition noise is more informative than one summary fit statistic.

Use this order of questions:

1. Is the processed signal trustworthy? Check blank and reference responses, bulk shifts, spikes, drift, carryover, and regeneration.
2. Are the reagents homogeneous and active? Review analyte aggregation, ligand activity, valency, and concentration assignment.
3. Do physical perturbations change the kinetics? Compare flow rates, ligand densities, immobilization formats, and replicate surfaces.
4. Does the 1:1 model remain internally consistent? Examine concentration-dependent rate estimates, R_max, residuals, and equilibrium affinity.
5. Is there independent evidence for a more complex mechanism? Use known valency, conformational data, competition, mutations, or an orthogonal assay before assigning a two-state or heterogeneous model.

Adding parameters will usually improve the numerical fit. It does not establish the biological reality of those parameters. A two-state model should not be labeled “conformational change” without evidence for an induced transition. A heterogeneous-ligand model may summarize surface behavior, but its two affinities should not automatically be reported as two solution binding sites. A mass-transport model cannot rescue data collected under conditions so transport-limited that intrinsic association is no longer identifiable.

SPR Sensorgram Artifact Diagnosis

SPR binding kinetics should therefore be designed as a series of falsifiable controls rather than a single concentration ladder followed by software fitting.

Orthogonal Confirmation of SPR Kinetic Results

An orthogonal method should address the specific uncertainty left by SPR. Equilibrium methods can test whether surface-derived affinity agrees with solution behavior. Isothermal titration calorimetry can provide solution-phase affinity and stoichiometry when concentration, heat, and material requirements are feasible. Competition experiments can test binding-site overlap. Structural or footprinting methods can evaluate a proposed conformational step.

Disagreement is not automatically a failure. SPR may use an immobilized construct, one orientation, a defined flow environment, and short observation windows, whereas a solution assay observes a different physical context. The useful task is to explain which assumption changes between methods and which result best supports the intended claim.

Figure 3. Intrinsic kinetic claims require signal quality, parameter stability under assay perturbation, and mechanism-appropriate confirmation.

Reporting Intrinsic, Apparent, and Model-Dependent SPR Kinetic Parameters

An SPR kinetic report should preserve enough detail to evaluate whether the rate constants are intrinsic, apparent, or model-dependent. Include:

• ligand and analyte constructs, oligomeric state, purity, and concentration method;
• immobilization or capture chemistry, surface capacity, reference-surface design, and active fraction where assessed;
• running buffer, solvent content, temperature, flow rate, contact time, dissociation time, and regeneration conditions;
• raw and processed sensorgrams, blank/reference corrections, excluded regions, and replicate behavior;
• concentration series and whether concentrations were globally fitted;
• model equations, linked and fixed parameters, fitting window, residual plots, and parameter uncertainty;
• flow-rate, ligand-density, regeneration, and nonspecific-binding diagnostics;
• equilibrium analysis and orthogonal measurements where used;
• explicit designation of parameters as intrinsic estimates, apparent rates, avidity-dominated values, or descriptive model components.

The strongest SPR result is not the curve with the smallest residual after the most complex fit. It is the kinetic interpretation that remains stable when the relevant physical assumptions are challenged experimentally.