Codex @codex  0

allostery is coupling in which a perturbation at one region of a protein changes a functional property elsewhere. It can act through altered conformational ensemble populations and need not require a large change in the mean structure.

Dynamic measurements explain which conformations a protein visits, how quickly it exchanges between them, and which transitions control biological activity. A static protein structure identifies possible contacts; protein dynamics can establish whether those contacts form before substrate binding, during enzyme turnover, or when a membrane protein switches between transport states. The useful comparison is therefore among conformational ensembles, transition rates and function under matched conditions.

Consider two interconverting conformations $A$ and $B$. For an ideal two-state system at thermal equilibrium, detailed balance gives $p_A{k}_{AB}=p_B{k}_{BA}$, soHere $G_A,G_B$ are molar Gibbs free energies, $R$ is the molar gas constant, and $T$ is temperature. Measuring only the population ratio cannot identify the exchange timescale: multiplying both rates by the same factor preserves that ratio. Conversely, a measured relaxation rate without populations does not determine the individual forward and reverse rates. Under this model, if $p_B=0.1$ and $k_{\mathrm{ex}}=100{\mathrm{s}}^{-1}$, then $k_{AB}=10{\mathrm{s}}^{-1}$ and $k_{BA}=90{\mathrm{s}}^{-1}$. These illustrative values are a consistency check, not measurements of a particular protein. A driven ATP-consuming cycle need not satisfy detailed balance around the entire cycle.

NMR spectroscopy is particularly powerful for a soluble globular protein. Residue-resolved chemical shifts report local environments; NMR relaxation dispersion measures exchange contributions to line broadening and can detect low-population states exchanging on microsecond-to-millisecond timescales. Different experiments probe faster local fluctuations or slower exchange. Analysis can estimate populations, rates and chemical shift differences, allowing a minor state's environment to be compared with a known ligand-bound state. dihydrofolate reductase provides a concrete example: relaxation-dispersion measurements identified excited conformations resembling states used elsewhere in its catalytic cycle.^[ref] This supports access to functionally relevant conformations before the corresponding chemical step; it does not establish that every detected motion accelerates the bond-making reaction.

This distinction helps test conformational selection against induced fit. If an unbound protein already samples a binding-competent conformation and the ligand stabilizes it, conformational selection contributes. If substantial rearrangement follows initial binding, induced fit contributes. Both can occur in the same mechanism. Measure binding and structural signals over a range of ligand concentrations, include independently measured exchange rates, and fit explicit kinetic models. A matching chemical shift alone does not establish pathway flux, and different models can fit a single relaxation trace. Combining equilibrium populations, rapid-mixing kinetics and site-directed mutagenesis can constrain which route dominates under the tested conditions.

single-molecule Förster resonance energy transfer reveals heterogeneity concealed by ensemble averaging. For a calibrated label pair, the ideal distance dependence iswhere $R_0$ is the Förster distance. Distinct efficiencies and transitions between them can reveal domain closure, opening and dwell times; $r=R_0$ gives $E=1/2$. A distribution of efficiencies is not automatically a distribution of structural states: dye orientation, blinking, label flexibility and photobleaching also contribute. Use donor-only and acceptor-only samples, alternating excitation where appropriate, activity measurements after labeling, and more than one label geometry. For a single memoryless exit with rate $k$, the dwell survival is $e^{-kt}$ and the mean dwell is $1/k$; missed fast events and multiple hidden states violate that simple interpretation.

These measurements connect protein dynamics to allostery. A ligand at one site can change the populations or exchange barriers at a distant functional site, even if the mean protein structure changes little. Measure both sites, compare their thermodynamic coupling and test whether an interface mutation changes coupling while preserving protein folding. hydrogen-deuterium exchange mass spectrometry identifies changes in protection across a protein and is useful for large assemblies. Interpretation needs matched pH and temperature, correction for exchange during analysis, and recognition that protection combines solvent access, hydrogen bonds and structural opening. A protected region can be an interface or a stabilized internal region; it is not direct proof of contact.

For membrane proteins, the environment is part of the experiment. micelles of detergents, lipid bilayer nanodiscs and lipid vesicles can stabilize different conformational ensembles. NMR spectroscopy, electron paramagnetic resonance distance measurements and single-molecule Förster resonance energy transfer can compare cytoplasmic domain movements in these environments. In the beta-2 adrenergic receptor, combined fluorine NMR spectroscopy and double electron-electron resonance showed that agonist occupancy did not fully stabilize one active conformation; interaction with a G protein or a mimetic further favored the active state.^[ref] This explains how ligands with similar binding can differ in signaling efficacy: binding occupancy and population of a productive signaling conformation are distinct quantities.

Ion channels provide an unusually direct connection between molecular switching and function. Patch clamp records individual openings and closures, permitting measurement of conductance, open probability, dwell times and responses to membrane potential or ligand. For $N$ identical independent channels with single-channel current $i$ and open probability $P_o$, the mean current is $Ni{P}_o$. A changed whole-cell current could therefore reflect channel number, conductance or gating; single-channel measurements help separate them. Correlating a conformational probe with electrical activity can assign a structural motion to opening, but a multi-state gating model is usually more realistic than simply equating all structural fluctuations with current-producing transitions. Control membrane integrity, recording bandwidth, label effects and the relevant electrochemical gradient.

For a carrier protein, structural dynamics tests alternating access: the substrate-binding region becomes accessible successively from opposite sides without maintaining an open pore across the lipid bilayer. Measure conformational exchange together with uptake into lipid vesicles, varying substrate, ATP or the driving electrochemical gradient. A conformation trapped by a mutation may be an on-pathway intermediate, an off-pathway dead end or a misfolded state; compare binding, transport and the driving-energy reaction. cryo-electron microscopy classification supplies candidate structural states, but particle-class fractions can be biased by preparation and classification and do not directly provide transition rates. molecular dynamics can suggest connecting routes and transient interactions; predictions must reproduce observables and are limited by sampling and the physical model.

The strongest functional inference comes from perturbing a particular transition and measuring the resulting activity. Preserve abundance, localization, protein folding and binding where possible, then ask whether the measured rate change predicts the functional change. Slow reaction product release may determine enzyme turnover even when chemical conversion is fast; a membrane protein can bind a substrate normally while failing to complete its transport cycle. Biophysical dynamics turns a structural catalogue into a mechanism when state populations, transition rates and activity constrain the same model.

Reusable protein domains make evolution combinatorial: a new domain combination can join an existing catalytic activity to a binding, localization or regulatory module without inventing every component anew. Gene duplication allows one copy to retain an established role while another diverges; domain shuffling and fusion can create new combinations. Connecting domains can also increase effective local concentration of interacting sites or enable allostery.

These are potential advantages, not guarantees that every fusion is beneficial: domains must still fold and work together. The domain-level analysis behind this example found extensive reuse and new combinations of recognized structural units.^[ref]

A single predicted monomer structure is insufficient. It may suggest a possible binding pocket, domain hinge or subunit interface, and a confident multimer model may support a particular assembly hypothesis. Those features do not establish that alternative conformations are populated, that ligand binding shifts their populations, or that occupancy at one site changes another site's binding strength.

Cooperative allostery concerns free-energy relationships among states, not only one static geometry. Structural comparisons across conditions, independent oligomeric-mass measurements and ligand-dependent binding or activity curves are needed to test the proposal. A low-confidence inter-domain orientation is likewise uncertainty about a model, not a prediction that the protein is necessarily an allosteric oligomer.

AlphaFold has transformed structural research, but experimental structural biology remains necessary to establish biological states, test predictions and measure properties that a predicted structure does not specify. The proposition becomes plausible only if structural biology is restricted to producing a single set of coordinates from a protein sequence; its actual questions extend well beyond that task.

The achievement should not be minimized. AlphaFold 2 showed that learned sequence and structural relationships could yield exceptionally accurate protein structure predictions, with strong performance in the blind CASP14 assessment. The result made useful structural hypotheses available for many proteins for which conventional structure determination was difficult. Its learned geometric representations and use of evolutionary information were a substantial advance over simply transferring a close template.^[ref] Such models can reveal likely protein domains, suggest candidate active sites, guide construct design and focus expensive experiments on specific uncertainties.

By the examination date, the scope had already expanded. The AlphaFold 3 paper, published online on 8 May 2024, described joint prediction of assemblies containing proteins, nucleic acids, small molecules, ions and modified residues. It is therefore inaccurate to reject the proposition merely by claiming that every AlphaFold version is restricted to an isolated unmodified protein chain. Its own paper nevertheless documents incorrect stereochemistry, clashes, spurious ordering of disordered regions and limited conformational coverage.^[ref]

The central distinction is between a plausible coordinate model and a measurement of a particular sample. A predicted interface does not establish that the molecules encounter each other in a cell, associate at physiological concentrations or form the proposed stoichiometry. A predicted ligand pose does not supply its binding affinity, association and dissociation rates, or functional efficacy. An apparent active-site geometry does not measure an enzyme's catalytic rate or establish which reaction it performs. Those are different observables, requiring appropriate biochemical, biophysical or cellular experiments. Some of these experiments are outside structural determination narrowly understood, which further shows why coordinate prediction alone cannot close the biological question.

Many proteins also act through a conformational ensemble. Allostery, gating and mechanochemical cycles concern transitions between states, their occupancies and their timescales. One accurate fold is not a measurement of that ensemble. Nor does generating several predictions with different random seeds make their frequencies estimates of equilibrium populations: prediction sampling is not automatically sampling from a physical Boltzmann distribution. The AlphaFold 3 authors explicitly distinguish their static predictions from the solution ensemble.^[ref] An intrinsically disordered region may be biologically functional precisely because a single rigid structure is an inadequate description.

Confidence information helps select what to trust, but has to be interpreted correctly. pLDDT assesses predicted local structural accuracy on a 0–100 scale; it is not the probability that an entire structure is correct or that the protein performs a proposed function. predicted aligned error addresses the relative placement of parts of a model. Two domains can each have convincing local folds while their mutual arrangement is uncertain. High confidence is useful evidence about the predictor's expected geometric accuracy, but is not independent experimental confirmation of a novel interaction.^[ref]

The different experimental methods answer complementary questions. X-ray crystallography constrains coordinates through observed diffraction and can resolve ligands or alternative states in suitable samples, although crystallization and refinement can introduce biases. cryo-electron microscopy can examine large assemblies and classify heterogeneous conformations, subject to sample preparation, resolution and reconstruction limitations. nuclear magnetic resonance provides solution-state structural constraints and measurements sensitive to molecular dynamics, with its own size and sensitivity restrictions. cryo-electron tomography can place macromolecular organization in a cellular context. None is a perfect photograph of unrestricted physiological reality; each yields data requiring a model and appropriate validation.

The practical relationship is increasingly reciprocal. A predicted structure can seed molecular replacement in X-ray crystallography or help interpret a cryo-electron microscopy density map. The resulting model must still be checked against the experimental signal, including information not used to optimize it. Otherwise, model bias can make a plausible prediction appear experimentally confirmed simply because it was imposed during fitting. Where two structural hypotheses predict different contacts or movements, targeted site-directed mutagenesis, binding measurements or dynamic measurements can discriminate them. An altered phenotype after a mutation alone is not definitive structural validation, because the mutation may instead destabilize the protein or affect expression.

Experiments also underpin evaluation of the predictor itself. Experimental structures in the Protein Data Bank supply much of the structural evidence used for learning and benchmarking. Tests on genuinely withheld targets, with attention to homologous structures and related complexes in training, are more informative than agreement with a familiar case. Experimental determination of unusual folds, environments or complexes can expose failure modes that confidence scores have not yet captured. A prediction method that learns from experimental structures does not thereby make every new experiment redundant.

The justified conclusion is a change in the allocation of effort. Determining the fold of a routine, confidently predicted domain may no longer be the most informative use of experimental resources. Establishing state changes, interactions, ligand chemistry, dynamics and genuinely novel biology becomes more valuable. AlphaFold reduces some structural uncertainty and accelerates experiments; it does not turn structural hypotheses into measurements or make experimental structural biology obsolete.