Circular Dichroism for Peptides: Reading Secondary Structure

What circular dichroism measures

Circular dichroism is the difference between how much left- and right-circularly polarised light a sample absorbs. Only chiral, asymmetric structures give a signal, and a chain of L-amino acids folded into a regular shape gives a characteristic one. Instruments usually report the difference as an ellipticity in millidegrees (mdeg).

Two wavelength regions are used. In the far ultraviolet, about 190 to 250 nm, the signal comes mainly from the amide bonds of the backbone, so the spectrum reflects secondary structure. In the near ultraviolet, about 250 to 320 nm, aromatic side chains and disulfide bonds contribute, which reports on their surroundings in a folded molecule.^[1] The most common uses are to check whether a molecule is folded, whether a change in sequence or conditions alters its conformation, and how stable that conformation is; measurements can be made on samples of 20 µg or less in a few hours.^[2]

The spectra to recognise

Each kind of secondary structure has a typical far-UV shape:^[2]

• Alpha-helix. Two negative bands near 222 and 208 nm and a strong positive band near 192 nm. The 222 nm band is the one most often used to follow helix content.
• Beta-sheet. A single negative band near 216 to 218 nm and a positive band near 195 nm. The exact shape varies with how the strands are arranged and twisted.
• Disordered and polyproline II. A strong negative band near 195 to 200 nm and little signal above 210 nm, or a weak positive band near 220 nm. Many short linear peptides in water give this spectrum, and much of it has been attributed to the extended polyproline II conformation rather than to a truly random chain.^[3]

Real spectra are usually mixtures. A peptide that is partly helical in water gives a spectrum between the helix and disordered shapes, and the 222 nm signal grows as helix content rises.

Units: from mdeg to mean residue ellipticity

Raw ellipticity depends on how much peptide is in the light path, so spectra are converted to mean residue ellipticity, which normalises per amide bond and allows molecules of different lengths to be compared:^[1]

[θ] = (θ in mdeg × MRW) / (10 × path length in cm × concentration in mg/mL), in units of deg cm² dmol⁻¹, where the mean residue weight MRW is the molecular weight divided by the number of peptide bonds (residues minus one).

A worked example: a 20-residue peptide of 2200 Da has an MRW of 2200 / 19 ≈ 115.8. At 0.2 mg/mL in a 0.1 cm cell, a reading of -10 mdeg at 222 nm gives (-10 × 115.8) / (10 × 0.1 × 0.2) = -1158 / 0.2 ≈ -5,790 deg cm² dmol⁻¹. The peptide mass calculator gives the molecular weight and residue count.

The concentration enters the result directly, so a 20% error in concentration is a 20% error in [θ] and in any helix estimate built on it. A concentration calculated from the weighed powder includes water and counter-ions (see purity versus net peptide content); a value measured by absorbance with an extinction coefficient or by amino acid analysis is more reliable.^[1]

Preparing the sample and buffer

Most problems in CD come from the sample rather than the instrument. Published guidelines stress calibrating the instrument, knowing the concentration accurately and keeping the total absorbance of the sample, cell and solvent low enough that the detector still receives enough light.^[1] In practice:

• Choose a transparent buffer. Low-concentration phosphate works well; chloride, imidazole, high Tris concentrations and many organic additives absorb strongly below about 200 nm. Sodium fluoride can replace sodium chloride when salt is needed. The buffer and pH guide covers choosing the pH itself.
• Match path length to concentration. Far-UV spectra are typically recorded in cells of 0.1 cm or less at a few tenths of a mg/mL; near-UV spectra need longer cells and higher concentrations.
• Record a blank. Measure the buffer alone in the same cell and subtract it, and watch the detector voltage so that data where too little light reaches the detector are discarded.
• Check the sample is clear. Particles and aggregates scatter light and distort the spectrum. Centrifuging or filtering before measurement helps; see the solubility and aggregation guide.

Estimating secondary structure

Deconvolution programs fit a measured spectrum as a combination of reference spectra from molecules of known structure. A comparison of three widely used methods, CONTIN, SELCON3 and CDSSTR, found that they performed comparably despite different algorithms, that the best choice depended on the reference set and wavelength range, and that analysing a spectrum with all three improves reliability.^[4] Reference databases and web servers such as DICHROWEB have made these analyses routine, and data extending to lower wavelengths increase the information in a spectrum.^[5]

Beta-rich structures are harder. Their spectra vary with the parallel or antiparallel arrangement and the twist of the sheets, and the BeStSel method was built to take this into account; it can distinguish parallel from antiparallel sheets and was applied to amyloid fibrils of disease-related proteins and peptides.^[6]

Most reference sets are made of folded proteins, so fractions reported for a short, flexible peptide carry more uncertainty than those for a globular protein. For many peptide questions the shape of the spectrum, the change in the 222 nm signal, or the presence of an isodichroic point in a titration is more informative than a percentage.

Solvents, temperature and aggregation

Many peptides that are mostly disordered in water become helical when a cosolvent such as 2,2,2-trifluoroethanol (TFE) is added. In a study of repeating-sequence peptides of 7 to 22 residues, thermal unfolding curves measured by CD between 0 and 50% TFE fitted a single helix-coil model, suggesting that helix formation is basically the same in TFE-water mixtures as in water; helix propensity rose steadily up to about 25% TFE and then levelled off.^[7] A review of TFE and related alcohols concluded that no single mechanism yet accounts for all of their effects on conformation.^[8] A TFE spectrum therefore shows what a sequence can adopt, not what it adopts in buffer, and the two should be reported separately.

Following the 222 nm signal while the temperature is raised gives a thermal unfolding curve. The same study found that the ellipticity of a complete helix itself depends strongly on temperature, which has to be considered when interpreting such curves.^[7] Spectra recorded over hours or days can also show aggregation: a shift from a disordered spectrum towards a single minimum near 218 nm is a common sign that sheet-rich assemblies are forming.^[6]

What CD does not tell you, and what to record

CD reports on the average conformation of everything in the cell. It does not give residue-by-residue information of the kind obtained from crystallography or NMR,^[2] and it does not establish identity or purity: an impurity or a degraded form contributes to the spectrum without being visible as a separate signal. Identity and purity are questions for mass spectrometry and HPLC. For a CD result to be interpretable, record:

• Sample. Sequence, how the concentration was determined, and the buffer, pH and any cosolvent.
• Measurement. Cell path length, temperature, wavelength range, bandwidth, number of scans and the blank subtracted.
• Units and analysis. Whether data are in mdeg or mean residue ellipticity, the MRW used, and the deconvolution method and reference set.

This guide describes general laboratory methods and published literature only. It is not a statement about the structure or behaviour of any particular material. Terms used here are defined in the glossary, and questions can be sent to support.