Short answer: Circular dichroism (CD) is a spectroscopic measurement of the difference in absorbance of left- versus right-circularly polarized light as a function of wavelength [1][3]. In peptides, the peptide-bond chromophore absorbs in the far-UV; when those amides are aligned in a regular secondary structure, exciton coupling splits and shifts the transitions so that an alpha helix, an antiparallel beta sheet and a disordered or polyproline-II-like chain each produce a characteristic spectrum [2][4]. Labs report far-UV data as mean residue ellipticity, estimate secondary-structure content with reference-protein algorithms, and use the near-UV aromatic region as a fingerprint of tertiary environment — always remembering that CD does not assign structure residue by residue the way NMR or a crystal structure can [2][3][5].
This article is educational analytical and organic chemistry for laboratory readers. It does not describe effects in people or animals and is not guidance for human or animal use.
What circular dichroism actually measures
IUPAC defines circular dichroism as a spectroscopic method that measures the difference in absorbance of left- and right-handed circularly polarized light by a material, as a function of wavelength [1]. Most biological molecules, including proteins and nucleic acids, are chiral and show CD in their ultraviolet absorption bands, which can indicate secondary structure [1].
Plane-polarized light can be treated as two circularly polarized components of equal magnitude, one left-handed (counter-clockwise) and one right-handed (clockwise) [3]. If the sample absorbs the two components equally, they recombine as plane-polarized light; if it absorbs them differently, the transmitted beam is elliptically polarized [3]. Spectropolarimeters report that difference as ΔA = A_L − A_R, and they commonly also report it as an ellipticity θ (in degrees) whose tangent is the ratio of the minor axis to the major axis of that ellipse [2][3]. The two scales are related by a fixed numerical factor: θ (degrees) = 32.98 ΔA [3]. A CD signal appears only where the sample absorbs, and only if the chromophore is chiral — intrinsically, because it is covalently attached to a chiral centre, or because the three-dimensional fold places it in an asymmetric environment [3][4].
That last point is why peptides are CD-active at all. The backbone amides of an L-residue chain sit in a chiral molecular framework (see D-amino acids for the configurational background), and when those amides line up in a helix or sheet their optical transitions couple [2]. CD is therefore a solution-phase reporter of average conformation, not a stand-in for peptide purity by HPLC area or for chemical identity by mass spectrometry.
Far-UV versus near-UV: two different chromophore windows
Peptide and protein CD is conventionally split by wavelength into two analytical windows [3][4]:
- Far-UV (about 240 nm and below, typically down to about 180–190 nm on a laboratory instrument). Absorption here is due principally to the peptide bond: a weak, broad n→π transition centred around 220 nm and a more intense π→π transition around 190 nm [3][4]. Aromatic side chains can also contribute in this range in some sequences [3]. Conventional xenon-arc instruments struggle much below about 180 nm because source intensity falls off and because nitrogen purge gas and water itself absorb; synchrotron radiation CD (SRCD) can push into the vacuum-UV (about 160–190 nm), where the extra wavelength range increases the information content of the spectrum [3][5][6].
- Near-UV (about 260–320 nm). This window reports aromatic side chains — tryptophan (a peak near 290 nm with fine structure toward 305 nm), tyrosine (a peak between about 275 and 282 nm) and phenylalanine (weaker, sharper bands between about 255 and 270 nm) — plus weak, broad disulfide absorption centred around 260 nm [3][4]. The shape and magnitude depend on how many of each aromatic residue are present, how mobile they are, and what hydrogen-bonding and polarity they experience, so the near-UV CD is a tertiary-environment fingerprint rather than a secondary-structure library [3][4].
Characteristic far-UV signatures of peptide secondary structure
When backbone amides are aligned, exciton coupling of their transitions produces secondary-structure-specific band shapes [2]. Greenfield's widely used summary of those shapes is [2]:
- α-helix: negative bands at 222 nm and 208 nm and a positive band at 193 nm.
- Well-defined antiparallel β-sheet: a negative band at 218 nm and a positive band at 195 nm.
- Disordered chains: very low ellipticity above 210 nm and negative bands near 195 nm.
The phrase "random coil" needs care. Charged polypeptides such as poly-L-lysine or poly-L-glutamate at neutral pH were originally described as random coils, but their far-UV CD resembles the extended left-handed polyproline II (PPII) helix of trans poly-L-proline [2]. Rucker and Creamer showed that even a seven-residue lysine peptide adopts a PPII conformation, and that this short peptide keeps PPII character when side-chain charges are screened or neutralized — evidence that the backbone itself favors PPII in water, not only electrostatic repulsion along a long lysine homopolymer [8]. The PPII CD spectrum is remarkably similar to that of unfolded proteins, which is why an "unordered" peptide spectrum should not be read as a structureless statistical coil [2][8].
How labs report a peptide CD spectrum
Raw instrument output is usually in millidegrees of ellipticity or in ΔA. To compare samples measured at different path lengths and concentrations, data are converted to a molar or mean-residue scale [2][3].
Greenfield's working conversions are [2]:
- Mean residue ellipticity [θ], in deg·cm²·dmol⁻¹ = (millidegrees × mean residue weight) / (path length in mm × concentration in mg/ml).
- Equivalently, [θ] = millidegrees / (path length in mm × protein concentration × number of residues).
- Δε = [θ] / 3298, where Δε is the differential molar extinction coefficient.
If the peptide is monomeric and does not aggregate under the measurement conditions, spectra collected at different concentrations and path lengths should overlay once they are converted to mean residue ellipticity [2]. Failure to overlay is a warning that path length, concentration or scattering is not under control [2][3].
What a reliable peptide CD experiment needs
The value of a CD trace is only as good as the sample and the instrument setup. Reviews of laboratory practice converge on a short list of controls [2][3]:
- Purity. Greenfield specifies that samples should be at least 95% pure by HPLC, mass spectrometry or gel electrophoresis [2]. A strong CD band cannot rescue a mixture: the spectrum is a population average. Complementary HPLC and mass-spectrometry identity checks still come first.
- Concentration. Correct concentration is the hardest part of high-quality CD [2]. Lowry and Bradford dye-binding assays give protein-dependent answers and must not be used to normalize CD data [2]. Quantitative amino-acid analysis (using stable residues such as alanine or lysine) is the most accurate route and can be run on aliquots of the actual CD sample; published molar extinction coefficients are acceptable if the peptide is buffer-exchanged into the CD solvent immediately before the scan and filtered (0.1–0.2 μm) to cut scattering [2]. Secondary-structure work may use concentrations from about 0.005 to 5 mg/ml, chosen to match the cell path length [2].
- Path length and cells. High-transparency quartz cells with path lengths from 0.01 to 1 cm are used; 0.1 cm rectangular or most cylindrical cells typically give usable baselines, but every cell is different and a matched solvent blank must always be subtracted [2]. Longer rectangular cells (path greater than 0.2 cm) can show strain birefringence [2].
- Transparent buffers. Chloride salts absorb in the far-UV and raise the practical short-wavelength cutoff. Greenfield's examples for roughly 0.1 mg/ml peptide in a 0.1 cm cell include about 185 nm for 10 mM potassium phosphate plus 100 mM potassium fluoride, versus about 195 nm when 100 mM potassium chloride or 100 mM sodium chloride replaces the fluoride [2]. Kelly, Jess and Price likewise flag chloride (for example NaCl used to elute ion-exchange fractions) as a far-UV absorber and prefer sulfate or fluoride salts that do not absorb significantly in this region [3]. DMSO and formamides are too absorbing to use; trifluoroethanol, hexafluoroisopropanol and hexane are optically transparent to 185 nm and below but change peptide and protein structure [2]. Glycerol can be present at up to 20%, but then measurements typically stop near 200 nm [2].
- Detector voltage as a quality check. Below the solvent cutoff the photomultiplier (dynode, HT or PMT high-voltage, depending on the manufacturer) rises sharply, signal-to-noise collapses, and ellipticity is no longer a linear function of path length [2][3]. Blank spectra are the practical way to confirm that the buffer itself is not overloading the detector [3].
- Calibration. A common amplitude and wavelength standard is (1S)-(+)-camphor-10-sulfonic acid (CSA) [2][3]. Greenfield records that 1 mg/ml CSA in a 1 cm cell has an absorbance of 0.1486 at 285 nm and an ellipticity peak of 335 millidegrees at 291 nm, and that the ratio of CSA ellipticities at 192.5 nm to 290.5 nm should lie between 2.05 and 2.08 (corresponding to Δε = 2.36 at 290.5 nm and 4.90 at 192.5 nm) [2]. Kelly, Jess and Price note the same pair of CSA peaks (290.5 nm and 192.5 nm) as a convenient near- and far-UV check [3].
- Scattering. Insoluble aggregates produce artefacts from differential light scattering and must be absent; solutions should be clear [3].
Estimating secondary-structure content — and why short peptides are harder
Because each regular structure has a distinctive far-UV shape, a measured spectrum can be treated as a linear combination of those component spectra plus a noise term [2][5]. Empirical algorithms compare the experimental trace with a reference database of proteins whose crystal structures (and therefore secondary-structure fractions) are known [5][6]. Widely used methods include SELCON (self-consistent), CONTIN / CONTINLL, CDSSTR, VARSLC and the neural-network method K2d [3][5][7]. Sreerama and Woody compared CONTIN, SELCON3 and CDSSTR on expanded reference sets (29 to 48 proteins, depending on wavelength range and whether denatured proteins were included) [7]. The DICHROWEB server packages several of those algorithms and reference databases, accepts laboratory and synchrotron file formats, and returns secondary-structure fractions together with a reconstructed spectrum and a goodness-of-fit parameter so that different method/database combinations can be compared [5][6].
Two caveats matter especially for synthetic peptides.
First, single-wavelength shortcuts exist — helical content is sometimes estimated from the CD intensity at 208 nm and 222 nm alone — but those estimates should be treated with caution; data collected over a wider far-UV range are preferred [3]. Lower-wavelength data carry more information and can be decomposed into more structural classes [5][6].
Second, the reference databases are built from folded proteins, not from short oligopeptides [3]. Kelly, Jess and Price note that CD spectra of oligopeptides therefore cannot be analysed reliably by those protein-trained methods except under conditions where a particular secondary structure predominates [3]. A visually "helical" 222/208 nm pair in a short peptide is still chemically informative; a CONTIN or SELCON percentage read out as if the sample were a globular protein is not.
CD also does not give the residue-specific assignments available from X-ray crystallography or NMR [2]. It reports the population-weighted average of whatever conformations are present in solution, on a modest sample amount (Greenfield notes that measurements can be made on multiple samples containing 20 µg or less of protein, in physiological buffers, in a few hours) [2].
Cosolvents such as TFE change the question being asked
Alcohol cosolvents, especially 2,2,2-trifluoroethanol (TFE), have been used for decades both to perturb proteins and to stabilize structure in peptides [9]. Kelly, Jess and Price describe TFE as a useful solvent for peptides and protein fragments because it usually promotes helix formation and can therefore be used to assess structure-forming potential [3]. Greenfield makes the complementary practical point that TFE, hexafluoroisopropanol and hexane are optically fine down to 185 nm but will change the structure of proteins and polypeptides [2]. A helical CD spectrum recorded in TFE or HFIP therefore answers a different question from a spectrum in aqueous buffer: it reports a structure the sequence can be induced to form, not necessarily the ensemble it populates in water [3][9]. Blank spectra of the chosen solvent system are still required to confirm that absorbance (and the high-tension voltage) remain acceptable [3].
FAQ
Is circular dichroism the same as optical rotation?
No. CD is the wavelength-dependent difference in absorbance of left- and right-circularly polarized light [1][3]. Optical rotation (a polarimetric quantity) is related but distinct; modern spectropolarimeters measure ΔA and commonly convert it to ellipticity with θ (degrees) = 32.98 ΔA [3].
Why do people quote 222 nm for helices?
An α-helix has characteristic negative bands at 222 nm and 208 nm and a positive band at 193 nm [2]. Intensity at 222 nm (or the 222/208 pair) is therefore a convenient helical marker, but a single-wavelength estimate of percent helix should be treated cautiously [3].
What is the difference between far-UV and near-UV CD?
Far-UV CD (peptide-bond n→π near 220 nm and π→π near 190 nm) reports secondary structure; near-UV CD (roughly 260–320 nm) reports the asymmetric environments of aromatic side chains and, more weakly, disulfides [3][4].
Can I use chloride-containing PBS for a far-UV scan?
Chloride raises the short-wavelength cutoff. In Greenfield's examples, phosphate buffers with 100 mM chloride typically limit useful data to about 195 nm in a 0.1 cm cell, whereas fluoride or sulfate analogues reach about 185 nm [2][3]. If the analysis needs data toward 190 nm, swap chloride for fluoride or sulfate and watch the detector voltage [2][3].
Why might a deconvolution of a short synthetic peptide look unreliable?
Reference algorithms such as CONTIN, SELCON and CDSSTR are trained on folded proteins whose crystal structures are known [5][7]. Those databases do not include oligopeptides, so protein-style percentage breakdowns of short peptides are not reliable unless a single secondary structure clearly predominates [3].
Does a helical spectrum in TFE mean the peptide is helical in water?
Not necessarily. TFE generally promotes helix formation and is used to probe structure-forming potential; it also changes peptide conformation relative to water [3][9]. Compare aqueous and cosolvent spectra rather than substituting one for the other.
Does CD replace HPLC purity or mass-spec identity?
No. CD reports average conformation of whatever is in the cuvette. Greenfield asks for at least 95% purity by HPLC, mass spectrometry or electrophoresis before the spectrum is interpreted [2].
Bottom line
Circular dichroism measures how differently a chiral sample absorbs left- and right-circularly polarized light [1][3]. For peptides, the far-UV peptide-bond region encodes secondary structure — negative 222 and 208 nm bands plus a positive 193 nm band for an α-helix; a negative 218 nm band and a positive 195 nm band for a well-defined antiparallel β-sheet; and a strong negative band near 195 nm with little intensity above 210 nm for disordered or PPII-like chains [2][8]. Mean residue ellipticity puts those traces on a comparable scale, provided concentration, path length and a transparent buffer are known [2][3]. Reference-protein algorithms (SELCON, CONTIN, CDSSTR and the DICHROWEB implementations) can estimate component fractions for folded proteins, but they are a poor default for short oligopeptides, and CD never replaces residue-level NMR or a crystal structure [2][3][5][7]. Used with a clean, well-quantitated sample, though, a peptide CD spectrum is one of the fastest solution experiments a laboratory can run to ask what the backbone is doing.
All products are for research use only — not for human or animal consumption.