L9 / C4: 3D Structure Determination

Biochemistry I

In 1958, what did John Kendrew create?

the 3D model of sperm whale myoglobin from X-ray crystallography

What are the three main methods for Protein Experimental Determination?

• X-ray crystallography

• Nuclear Magnetic Resonance (NMR)

• Cryogenic electron microscopy (cryo-EM)

How does X-ray Crystallography?

• X-rays are bounced off of the protein and deflected by electrons in the various atoms/bonds

• The diffraction pattern of the X-rays is measured and an electron density map is created

• Amino acids structures are fit into the electron density

Given the crystal consistency, what can be said?

• the density maps are not as precise as they could be

• crystal is said to have a “resolution limit”

What are some advantages of X-ray crystallography?

• crystalline proteins assume conformations that are very similar to that of the protein in solution (near-native structure)

• most independent x-ray crystallography experiments describe the same conformation for the same structure (consistency/reproducibility)

• many enzymes are catalytic active in the crystalline state

  • since activity is highly dependent on structure, this is strong evidence that the crystalline conformations must indeed be near-native

What are some limitations in X-ray Crystallography?

• need for a protein crystal (not always possible, very specific conditions required for each different protein)

• types of molecules that are harder to crystalize or to analyze

  • ex: transmembrane proteins, carbohydrates, IDPs

• resolution limit of obtained crystals (additional data helps)

• protein structures are determined in a static state (conformation)

• parts of protein structure might be distorted by crystal packing effects

NMR Spectroscopy

• paramagnetic nuclei have interactions with external magnetic field

• nuclei can absorb energy at particular frequencies (resonance frequencies)

• resonance frequencies are sensitive to chemical environment and nearby nuclei

• correlation spectroscopy (COSY)

• Nuclear Overhauser Spectroscopy (NOSEY)

Correlation spectroscopy (COSY)

provides interatomic distances between protons that are covalently connected through one or two atoms

Nuclear Overhauser spectroscopy (NOSEY)

provides interatomic distances for protons that are close in space, but not necessarily connected

NMR Spec advantages

• no need for crystallization

• can be used to determine the protein structure in solution

• provides not a single conformation, but an ensemble of conformations

• can probe motions over time scales spanning 10 orders of magnitude

• resolution can be comparable with x-ray crystallography (usually consistent with crystallographic data)

NMR spec limitations

• protein size limited to ~100 kD

• raw data can be difficult to interpret, depending on protein size and flexibility (most manual annotation of chemical shifts)

• higher computational cost for model building (fit structure to data)

• requires relatively large amounts of pure samples (on the order of several mg) to achieve a reasonable signal to noise level

Cryo-EM advantages

• no need for a crystal

• no need for large amounts of sample ( about 0.1 mg)

• no limit of size or weight (bigger is easier than smaller)

• protein conformations in native state

• can capture very different conformational states (if they exist in solution)

Cryo-EM limitations

• molecules are detected in unknown orientations

• raw data can be very noisy and difficult to analyze

• different “shapes” might reflect the same conformation (diff. orientation)

• similar “shapes” might reflect different conformations

• resolution was inferior to both NMR and X-ray crystallography

Imaging processing and machine learning methods can be used to cluster conformations, improving model fitting and improving the resolution of predicted structures - fast improvement in methods!

Three main methods for computational protein structure prediction

• Ab Initio (from sequence to structure)

• Fold recognition (Threading)

• Homology modeling (same function = same folding)

but also

• AI modeling using MSAs (correlated mutations = structural constraints)