Figure 3.

The misfolding hypothesis. A, Outcomes of translation. Most proteins exit the ribosome (left) with no errors (bottom), but a substantial proportion contain at least one error (top). The probability of misfolding after correct translation is lower than after erroneous translation (center). Some proteins attain native state but then improperly unfold (right). Natural selection can act at four points: at 1), to reduce the frequency of translation errors in certain proteins; at 2), to reduce the proportion of error-containing proteins which misfold; at 3), to reduce the number of error-free proteins which misfold; and at 4), to reduce the number of proteins (with or without errors) which improperly unfold. B, Adaptations to higher misfolding costs constrain sequence evolution because adapted sequences are rare. So long as evolutionarily viable gene sequences are substantially more adapted than random sequences and adaptation levels are roughly bell-shaped in distribution, the number of alternative sequences (possible alleles) compatible with higher levels of adaptation (accuracy, robustness, etc.) declines rapidly. Adaptation to increasing misfolding costs therefore leads to increasing evolutionary constraint and slower sequence evolution.