Circular Dichroism Most protein secondary structure studies use CD - PowerPoint PPT Presentation

Circular Dichroism • Most protein secondary structure studies use CD • Method is bandshape dependent. Need a different analysis • Transitions fully overlap, peptide models are similar but not quantitative • Length effects left out, also solvent shifts • Comparison revert to libraries of proteins • None are pure, all mixed

UV-vis Circular Dichroism Spectrometer Sample Slits PMT PEM quartz Xe arc source This is shown to provide a Double prism comparison to Monochromator (inc. dispersion, VCD and ROA instruments dec. scatter, important in uv) JASCO –quartz prisms disperse and linearly polarize light

Polypeptide Circular Dichroism ordered secondary structure types α -helix Δε β -sheet turn Brahms et al. PNAS, 1977 λ poly-L-glu( α , ____ ), poly-L-(lys-leu)( β, - − - −), L-ala 2 -gly 2 (turn, . . . . . ) Critical issue in CD structure studies is SHAPE of the Δε pattern

Protein Circular Dichroism Δ A Myoglobin-high helix ( _______ ) , Immunoglobin high sheet ( _______ ) Lysozyme, a+b ( _______ ) , Casein, “unordered” ( _______ ) ,

260 UIC Basis set - 22 proteins ECD 240 Wavelength [nm] α 220 200 180 10 0 Δε

2D CORRELATION SPECTRA - - ECD ECD 2D CORRELATION SPECTRA 1 . 0 0 . 8 2 correlation coefficient r 0 . 6 0 . 4 0 . 2 0 . 0 1 8 0 2 0 0 2 2 0 2 4 0 2 6 0 W a v e l e n g t h [ n m ] 3D surface obtained by fitting the set of ECD spectra with polynomial Correlation coefficients of the polynomial fit of the ECD spectral intensity as the function of α -helical FC .

2D CORRELATION SPECTRA - - ECD ECD 2D CORRELATION SPECTRA Synchronous correlation map of the protein ECD spectra with respect to α -helix FC perturbation. Positive contours : blue/cyan, negative contours: red/pink.

Simplest Analyses – Single Frequency Response Basis in analytical chemistry � Beer’s law response if isolated Protein treated as a solution � % helix, etc. is the unknown Standard in IR and Raman , Method : deconvolve to get components Problem – must assign component transitions, overlap -secondary structure components disperse freq. Alternate: uv CD - helix correlate to negative intensity at 222 nm, CD spectra in far-UV dominated by helical contribution Problem - limited to one factor, -interference by chromophores]

Single frequency correlation of Δε with FC helix θ (222 nm) vs FC helix θ (193 nm) vs FC helix Δε at 222nm/193 nm 10 0 0 20 40 60 80 FC helix [%]

BETA-LACTOGLOBULIN • M w 18,400 Da, 162 residues • Primarily β -sheet (42% sheet, 16% helix) • High propensity for helical conformation • Structural homolgy to retinol binding protein

Far-UVCD spectra of BLG titrated with SDS (0-50 mM) 8 6 4 Δε (M-1cm-1) 2 0 0 mM -2 3 mM } 5 - 50 mM -4 -6 180 200 220 240 260 wavelength (nm)

Near-UVCD spectra of BLG titrated with SDS 0.0005 3, 5 mM 0.0000 Δε (M -1 cm -1 ) -0.0005 1.0 mM -0.0010 0.1 mM -0.0015 0 mM -0.0020 260 280 300 320 wavelength (nm)

PC/FA determined secondary structure change 40 helix Secondary Structure (%) 35 30 25 20 sheet 15 Critical Micelle Concentration (8.2 mM) 10 0 10 20 30 40 50 [SDS] (mM)

Problem of Secondary Structure Definition • where do segments begin and end • what are turns, bends, etc. • what is basis for helix or sheet - φ,ψ or H-bond pattern ? • sources: X-ray report - non-uniform (visual) Levitt-Greer - C α relationships dominate Kabsch-Sander - H-bond patterns dominate (DSSP) Frishman-Argos - “knowledge-based” (STRIDE) King-Johnson - CD oriented

Problem of secondary structure definition No pure states for calibration purposes ? ? ? helix sheet ? Need definition: Where do segments begin and end?

Comparison of secondary structure definitions: 80 Turn Helix 40 60 KJ or AF KJ or AF 40 20 20 0 0 0 20 40 60 80 0 5 10 15 20 25 30 Kabsh-Sander (DSSP) Kabsh-Sander (DSSP) Sheet Other 40 60 KJ or AF KJ or AF 40 20 20 0 0 0 20 40 0 20 40 60 Kabsh-Sander (DSSP) Kabsh-Sander (DSSP) King-Johnson Comparison with DSSP (Kabsh-Sander): Frishman-Argos (STRIDE)

Next step - project onto model spectra –Band shape analysis Peptides as models - fine for α -helix, -problematic for β -sheet or turns - solubility and stability -old method:Greenfield - Fasman --poly-L-lysine, vary pH θ i = a i φ α +b i φ β + c i φ c -- Modelled on multivariate analyses Proteins as models - need to decompose spectra - structures reflect environment of protein - spectra reflect proteins used as models Basis set (protein spectra) size and form - major issue

Freedom from model spectra Series of methods developed assuming : • spectral response was (fully) related to the secondary structure • sampling structures with sufficient proteins creates a spectral basis Milestones: • Provencher - Glockner --(CONTIN) - ridge regression, no intermediate • Hennessey - Johnson -- Single value decomposition (SVD) initial step is same as principle component or Factor analysis simplifies spectral variation - monitor component loadings 5 factors (independent component spectra) Fractional structure from (total)inversion of SVD result A = USV T F = XA X = F(VS’U T ) Modifications: Project out model spectra (Compton -Johnson) Variable selection - optimize basis (Manavalan-Johnson) permits analysis of why proteins are outliers.

Variations on a Theme • Self-consistent methods - Sreerama - Woody - (SELCON) – probably the most widely used now, Web site connect • Restricted multiple regression (RMR) of Factor Analysis loadings Pancoska - Keiderling (et al.) applied to many spectral types • Factor analysis is general - same as SVD build correlation matrix of all experimental spectra, diagonalize to get eigenvalues, eigenvectors yielding weights (singular values), loadings and components Useful for analysis of spectral variation with structural variation • Quantitative Secondary Structure application: Spectral shape and intensity is influenced by many factors eg. solvent, pH, sequence, secondary structure, chromophore RMR idea is to find spectral components sensitive to structure

Factor Analysis Method Decomposition of an experimental spectrum θ ( λ ) into linear combination of independent component spectra φ j ( λ ): p p ∑ ∑ θ λ = φ λ = φ λ ( ) ( ) ( ) C A c i ij j i ij j = = 1 1 j j where λ ∫ = θ λ λ 2 2 ( ) “norm” A d i i λ 1 C / “loadings (expansion coefficients)” ij c ij φ j ( λ ) “component spectra”

Factor Analysis Method 1. Construct Correlation Matrix [R]: 1 p ∑ R = λ λ λ = θ λ = φ λ [ ] [ ( )] [ ( )] T ( ) ( ) ( ) w w , where w c i i i i ij j A = 1 j i (normalized spectral data) 2. Diagonalize [R] to obtain Principal Components : = Λ δ [ ] [ ][ ] [ ] T q R q ij ij 3. Calculate component spectra and corresponding loadings (coefficients): φ λ = λ [ ( )] [ ( )][ ] = [ ] [ ] T w j q and c q j ij

FA component spectra - 22 proteins ECD φ 1 Δε normalized φ 2 φ 3 φ 4 φ 5 180 200 220 240 260 Wavelength [nm]

Factor (Principle Component) Analysis • Approach is functionally equivalent to Principle Component Analysis - Singular Value Decomposition – No curve fitting is necessary – Band assignments are not necessary – Method is general - any technique • Method: – treat set of protein spectra as basis set of functions, [ φ ] – Diagonalize the co-variance matrix to • find most common elements- ψ 1 • find most common deviation - ψ 2 • continue – Reconstruct Spectra: [ φ ] = [ ψ ][ α ], where [ α ] is a matrix of coefficients, c ij for i th protein and j th subspectrum – Use vector of c ij for protein i to characterize protein. Note ψ i depends on training set, construct to be orthogonal

Tyr92 Ribonuclease A Tyr115 Tyr97 Tyr73 combined uv-CD H1 and FTIR study H2 H3 Tyr76 Tyr25 • 124 amino acid residues, 1 domain, MW= 13.7 KDa • 3 α -helices � • 6 β -strands in an AP β -sheet 6 β β sheet � • 6 Tyr residues (no Trp), 4 Pro residues (2 cis, 2 trans) , 2 )

0 . 0 6 F T I R RibonucleaseA 0 . 0 5 0 . 0 4 Absorbance 0 . 0 3 0 . 0 2 FTIR —amide I 0 . 0 1 Loss of β -sheet 0 . 0 0 1 7 2 0 1 7 0 0 1 6 8 0 1 6 6 0 1 6 4 0 1 6 2 0 1 6 0 0 - 1 ) W a v e n u m b e r ( c m 0 - 2 - 4 Ellipticity (mdeg) Near –uv CD - 6 Loss of tertiary - 8 - 1 0 structure - 1 2 - 1 4 N e a r - U V C D - 1 6 2 6 0 2 8 0 3 0 0 3 2 0 W a v e l e n g t h ( n m ) Far-uv CD 5 Ellipticity (mdeg) Loss of α -helix 0 - 5 Spectral Change - 1 0 F a r - U V C D Temperature 10-70 o C - 1 5 1 9 0 2 0 0 2 1 0 2 2 0 2 3 0 2 4 0 2 5 0 W a v e l e n g t h ( n m ) Stelea, et al. Prot. Sci. 2001

- 6 . 4 1 . 0 Ribonuclease A F T I R PC/FA loadings - 6 . 8 0 . 5 2 ) Temp. variation i1 (x10 - 7 . 2 0 . 0 C - 7 . 6 - 0 . 5 FTIR ( α,β ) - 8 . 0 - 1 . 0 - 5 1 0 - 7 5 - 9 0 Near-uv CD i2 i1 N e a r - U V C D - 1 1 C C - 5 - 1 3 (tertiary) - 1 0 - 1 5 - 1 7 - 1 5 - 1 0 5 0 Far-uv CD - 1 1 - 5 ( α -helix) F a r - U V C D i1 i2 - 1 0 C C - 1 2 - 1 5 - 2 0 - 1 3 - 2 5 Temperature - 3 0 0 2 0 4 0 6 0 8 0 1 0 0 Stelea, et al. Pre-transition - far-uv CD and FTIR, not near-uv Prot. Sci. 2001