DEER spectroscopy (double electron-electron resonance) is a pulsed EPR technique that measures the distance between two unpaired electrons, usually spin labels attached to chosen sites on a protein, across a range of roughly 1.5 to 8 nanometers. Because every molecule in the frozen sample contributes to the signal, DEER EPR returns a full distribution of distances instead of a single average, which shows how many conformations a protein adopts and how flexible it is. This guide is for protein scientists and drug discovery teams deciding whether DEER fits a project. It covers how the experiment works, how DEER relates to PELDOR, the distance limits, and where DEER is used in drug discovery.
What is DEER spectroscopy?
DEER spectroscopy is a pulsed electron paramagnetic resonance (EPR) experiment that measures the magnetic dipolar coupling between two electron spins and converts it into a distribution of distances between them. EPR only detects unpaired electrons, and most proteins have none, so the signal comes almost entirely from the probes you put there. That selectivity is what makes the measurement clean: there is no background from the rest of the protein, the buffer or the membrane. If electron paramagnetic resonance is new to you, start with the basics of continuous-wave and pulsed EPR, then come back here for the distance experiment.
DEER belongs to a family of methods called pulsed dipolar spectroscopy. It is the most widely used member of that family for distance measurements on proteins.
How does DEER EPR measure distances?
DEER EPR measures distance through the dipolar coupling between two spins, a magnetic interaction whose frequency falls off with the cube of the distance between them. For two nitroxide spin labels the dipolar frequency is about 52 MHz divided by the distance in nanometers cubed (the relationship is given as equation 1 in the 2021 community benchmark and guidelines paper by Schiemann et al. in JACS). Short distances produce fast oscillations in the signal and long distances produce slow ones.
The dipolar frequency and thetime one oscillation takes at common distances:
• 2nm: 6.5 MHz, one oscillation every 0.15 µs.
• 3nm: 1.9 MHz, one oscillation every 0.52 µs.
• 4nm: 0.81 MHz, one oscillation every 1.2 µs.
• 5nm: 0.42 MHz, one oscillation every 2.4 µs.
• 6nm: 0.24 MHz, one oscillation every 4.2 µs.
These numbers show why longdistances are hard. Doubling the distance from 3 nm to 6 nm makes theoscillation eight times slower, so the experiment has to record a much longersignal to see it.
What is site-directed spin labeling?
Site-directed spin labelingattaches a stable radical, most often a nitroxide, to a cysteine introduced bymutagenesis at the position you want to measure. The most common label is MTSL,which reacts with the cysteine thiol to form the side chain called R1. Nativecysteines that are not part of the design are usually mutated away first so thelabel goes only where you want it. Two labels give one distance. A homodimerlabeled at one site per subunit also gives one distance, measured across theinterface.
Other labels exist for caseswhere nitroxides fall short. Gd(III) labels stay intact in the reducingenvironment inside cells, Cu(II) bound to a double-histidine motif gives a morerigid probe, and trityl radicals give strong, narrow signals and can be measuredat higher temperatures than nitroxides. Each label has its own linker lengthand flexibility, which affects how a measured distance maps back onto theprotein backbone.
How does the four-pulse DEER sequence work?
The four-pulse DEER sequenceuses two microwave frequencies: one to observe a set of spins and one to fliptheir partners. The dead-time-free version used in nearly all protein worktoday was introduced by Pannier,Veit, Godt, Jeschke and Spiess in the Journal of Magnetic Resonance in 2000.The sequence runs like this:
1. An observer pulse at the first frequency tips the observed spins (the A spins)into the plane where they can produce an echo.
2. A second observer pulse after a fixed delay refocuses them into a spin echo.
3. A pump pulse at a second frequency flips the partner spins (the B spins) at a variable time t. Flipping B changes the local magnetic field felt by A by an amount set by the dipolar coupling.
4. A third observer pulse refocuses the echo a second time. This refocused echo is the signal that gets recorded.
5. The pump pulse is stepped through a range of times t, and the refocused echo intensity at each step builds up the DEER time trace.
Measurements are made on flash-frozen samples, typically near 50 K, so the trace captures every conformation present at the moment of freezing. Most protein DEER is run at Q-band (about 34 GHz), which gives better sensitivity than the older X-band(about 9.5 GHz) setups.
How is a DEER time trace turned into a distance distribution?
A DEER time trace becomes a distance distribution in two stages: the decay caused by spins on neighboring molecules (the intermolecular background) is separated out, and the remaining dipolar signal is converted into distances. The conversion is a mathematically ill-posed problem, so small amounts of noise can change the answer unless the analysis is constrained. The common approaches are:
• Two-stepanalysis. Fit and divide out the background, then apply Tikhonovregularization, as implemented in DeerAnalysis. A regularization parametertrades smoothness of the distribution against fit to the data.
• One-stepanalysis. Fit the background and the distribution together and estimateuncertainty from the noise level, as in DeerLab.
• Neuralnetwork analysis. DEERNet converts traces to distributions using a networktrained on simulated data, with no user-adjusted parameters.
The JACS benchmark tested how reproducible this is. In that study, seven laboratories, blinded to sample identity, measured and analyzed aliquots of the same four doubly labeled protein samples, and the resulting distributions were similar overall. The main source of disagreement between labs was background separation, driven by the different trace lengths each lab recorded. For that reason the guidelines recommend automated analysis workflows that leave no room for subjective choices.
How do you read a DEER distance distribution?
You read a DEER distancedistribution by looking at where the peaks sit, how wide they are, and how manythere are, then checking whether any signal at long distances is real.
• Peak position gives the most probable distance between the two spin labels. Labels sit on flexible linkers, so the spin-spin distance is not the Cα to Cαdistance. Model the label rotamers before comparing with a structure.
• Peak width reflects backbone flexibility plus label flexibility. A narrow peak means a well-defined conformation. Trust the width only if the trace is long enough (see the distance-range section).
• Number of peaks shows distinct conformations and their relative populations. Extra peaks can also come from label rotamer populations, from three or more spins on one molecule, or from noise, so confirm them with a second label pair or a ligand titration.
• Signalat long distances can mean long-range states or aggregation. Do notinterpret features beyond the reliable distance for the trace length.
Tessmer and Stoll's review describes why label modeling matters for this step: the tether between the spinand the backbone makes the measured distribution differ from the backbonedistribution, and rotamer-ensemble tools such as MMM and mtsslWizard are usedto bridge the two. Our cheatsheet on how to read a DEER distance distribution puts these checks onone page, with annotated examples of single-peak, multi-peak and broaddistributions.
What is the difference between DEER and PELDOR?
There is no difference: DEER(double electron-electron resonance) and PELDOR (pulsed electron-electrondouble resonance) are two names for the same pulsed EPR experiment. The technique was first described by Milov and colleagues in Novosibirsk, includinga 1984 Chemical PhysicsLetters paper by Milov, Ponomarev and Tsvetkov that used the originalthree-pulse version. Which name a paper uses mostly reflects the tradition oft he group that wrote it. The 2021 JACS benchmark, written by authors from both traditions, uses "DEER/PELDOR" throughout.
The names you will see in theliterature:
• DEER(double electron-electron resonance) is the most common name in protein andstructural biology papers.
• PELDOR(pulsed electron-electron double resonance) is the original name and is commonin European chemistry and physics groups.
• Four-pulse DEER is the dead-time-free sequence published in 2000. It is the standard experiment today, and "DEER" usually means this version.
• Pulsed dipolar spectroscopy (PDS) is the family of distance experiments, which includes DEER plus related methods such as RIDME, DQC and SIFTER.
When you search the literature, search both names. A method section that says PELDOR and one that says DEER describe the same measurement and produce the same kind of data.
What distance range can DEER measure?
DEER measures distances of about1.5 to 8 nanometers (15 to 80 Å) in most protein samples, and up to about 16nanometers (160 Å) in fully deuterated proteins, according to Tessmer andStoll's 2025 Annual Review of Biophysics review, which cites a GroEL study bySchmidt and colleagues for the long-distance result. Those limits come from thephysics of the experiment, and where a given sample falls depends on how long atrace you can record.
What sets the upper distance limit in DEER?
The upper limit is set by how long the DEER trace can be, and that is limited by the phase memory time, thetime it takes the echo to decay. The JACS benchmark guidelines state that to quantify the width of a distance peak, the trace must cover at least two periods of the slowest oscillation, which gives a maximum distance of about 3nm multiplied by the cube root of the trace length in microseconds. To quantify only the mean distance, one period is enough.
Applied to common trace lengths,the rule gives these limits (the mean-only figure uses the same equation withone period instead of two):
• 2µs trace: widths reliable to 3.8 nm, mean distances to 4.8 nm.
• 4µs trace: widths reliable to 4.8 nm, mean distances to 6.0 nm.
• 6µs trace: widths reliable to 5.5 nm, mean distances to 6.9 nm.
• 8µs trace: widths reliable to 6.0 nm, mean distances to 7.6 nm.
Longer traces come at the costof signal-to-noise, because the echo keeps decaying. Deuterated solvents andcryoprotectants slow that decay, and deuterating the protein itself slows itfurther, which is how the 16 nm measurements were reached.
What sets the lower distance limit in DEER?
The lower limit, around 1.5 to 2nm, comes from the dipolar coupling becoming so strong that the pulses can nolonger excite the coupled spins cleanly. Below that range other methods workbetter, including continuous-wave EPR line broadening and single-frequencypulsed experiments.
How does DEER compare with FRET for measuring distances?
DEER and single-molecule FRETcover similar distances, but they differ in labels, sample conditions and whatthe result describes. A 2022Nature Communications study by Peter and colleagues measured the sameproteins with both methods and compared the practical requirements.
The main differences reported inthat study:
• Distancerange. DEER covers 1.5 to 8 nm, and up to 16 nm with deuterated protein.smFRET covers 3 to 8 nm, with 10 to 15 nm possible using multiple acceptordyes.
• Labels.DEER needs one type of label. smFRET needs two, a donor and an acceptor.
• Sampleamount. Standard Q-band DEER uses about 80 µL of 10 to 30 µM labeledprotein. smFRET uses 100 to 400 µL of 15 to 100 pM labeled protein.
• Samplestate. DEER runs on frozen solution at about 50 K. smFRET runs in liquidsolution at room or physiological temperature.
• Timeper measurement. DEER takes several hours at Q-band. smFRET takes 30 to 60minutes for freely diffusing molecules.
The study found similar overallaccuracy for the two methods across four proteins with and without ligands, andtraced the remaining inconsistencies to cryoprotectants in the DEER samples andlabel-protein interactions in the smFRET samples. DEER reports the fullensemble frozen in place, while smFRET follows individual molecules in solutionwith time resolution down to microseconds.
What is DEER used for in drug discovery?
In drug discovery, DEER is usedto measure how ligands, binding partners and mutations shift a protein'sconformational ensembles, and to test structural models before more expensivestructural work. Tessmer and Stoll's review lists ligands, light, pH andpressure among the stimuli DEER has been used to study, alongside proteincomplexes, domain orientation and disordered regions. The common uses are:
• Ligand-induced conformational change. Comparing the distance distribution with and withouta compound shows whether it shifts the population between states, for example toward an active or inactive conformation of a receptor or transporter.
• Model validation. DEER distances can confirm or challenge an AlphaFold prediction or a cryo-EM model. The AlphaFold protein dynamics article covers how a single predicted structure compares with a measured ensemble.
• Complexes and oligomers. Labels on two different proteins, or on two subunits of one protein, give intermolecular distances. This applies to protein dimerization and to ternary complexes formed by PROTACs and molecular glues.
• Flexible and disordered proteins. Intrinsically disordered proteins and flexible loops give broad distributions that other structural methods tend to miss or blur.
A hypothetical example shows howthis reads in practice. A kinase is labeled at two sites on either side of itsactivation loop. Without inhibitor, the distribution has a main peak at 3.2 nmand a smaller one at 4.1 nm. With a candidate inhibitor bound, the 4.1 nm peakgrows to become the main population. That result says the compound stabilizesthe open state, which is information a single crystal structure of the complexwould not give.
What does a DEER experiment need from your sample?
A standard Q-band DEERmeasurement needs about 80 µL of doubly labeled protein at around 10 to 30 µM,frozen in a glass-forming buffer with deuterated solvent, and takes severalhours per sample, according to the comparison table in Peter and colleagues'2022 Nature Communications study. The same table notes that measurements atsub-micromolar spin concentrations have been published. Before measurement youalso need labeling efficiency, since unlabeled or singly labeled proteinreduces the modulation depth and the signal.
Sample volume, measurement timeand the need for an EPR specialist have kept DEER out of many drug discoverylabs. High Q Technologies built FATHOM®, aquantum-enabled pulsed EPR spectrometer with a superconducting resonator, toresolve distance distributions on micromolar samples in under a few hours, withmulti-sample cartridges and load-and-go operation that do not require an EPRexpert on staff.
How do I get DEER data on my protein?
You can get DEER data on yourprotein by running it on an in-house pulsed EPR spectrometer or by sendinglabeled samples to a lab that measures them for you. If you are deciding whichsites to label, start with the cheat sheet on how to read a DEER distance distribution, whichshows what good and problematic data look like before you design constructs.For the wider set of techniques, from continuous-wave EPR to relaxation anddistance methods, see the hubguide to EPR methods for protein scientists.
If you have samples ready, the SampleMeasurement Service measures them on FATHOM in High Q's lab and returnsdistance distributions without you buying an instrument. If you want to discusswhether DEER fits your target, or see FATHOM run on data like yours, book acall through Talk toan Application Scientist.
Key takeaways
• DEER spectroscopy, also called PELDOR, is a pulsed EPR experiment that measures distances between two spin labels on a protein.
• DEER returns a distribution of distances, so it reports flexibility and the number of conformations, not only an average structure.
• Typicalprotein DEER covers about 1.5 to 8 nm, and up to about 16 nm in fullydeuterated proteins.
• Thereliable maximum distance depends on trace length: quantifying a peak widthneeds at least two oscillation periods in the recorded trace.
• Aseven-lab blinded benchmark published in JACS in 2021 found DEER distributionsreproducible across labs, with background separation as the main source ofvariation.
• Indrug discovery, DEER is used to see ligand-induced conformational shifts,validate AlphaFold and cryo-EM models, and characterize complexes anddisordered proteins.
Next Steps for Your Research
Ready to bring solution-state dynamics to your biophysics or drug discovery pipeline?
- Send a Sample: Test your protein dimers, GPCRs, or complexes through our Sample Measurement Service.
- Talk to an Expert: Schedule a consultation with an application scientist to discuss integrating FATHOM into your lab.
