AlphaFold2 as a replacement for solution NMR structure determination of small proteins: Not so fast!

Area of Science:

• Biophysics
• Structural Biology
• Computational Biology

Background:

• Protein structure determination is key to understanding protein function.
• AlphaFold2 (AF2) is a leading computational tool for protein structure prediction.
• Accurate prediction of protein structures is essential in biological research.

Purpose of the Study:

• To evaluate the accuracy of AlphaFold2 (AF2) in predicting the structure of small, folded proteins.
• To highlight potential limitations of AF2 in specific cases.
• To emphasize the continued importance of experimental methods in structural biology.

Main Methods:

• Computational protein structure prediction using AlphaFold2 (AF2).
• Comparison of AF2-predicted structure with experimentally determined structures.
• Focus on the specific case of pro-interleukin-18 (pro-IL-18).

Main Results:

• AF2 incorrectly predicted the structure of pro-interleukin-18 (pro-IL-18) with high confidence.
• The predicted structure by AF2 corresponded to the mature cytokine IL-18, not the pro-form.
• This demonstrates a failure of AF2 in a specific instance of a small, compact, folded protein.

Conclusions:

• While AF2 is a powerful tool, it is not infallible.
• Experimental structure determination is still necessary, even for small, well-folded proteins.
• Limitations of computational prediction necessitate validation through experimental methods.