perspective
From Levinthal to pathways to funnels
Ken A. Dill and Hue Sun Chan
‘Anew view of protein folding kinetics replaces the idea of ‘folding pathways’ with the broader notions of
energy landscapes and folding funnels. New experiments are needed to explore them.
A soxalled "New View? of the kinetics of protein folding has
emerged during the pas few years resulting from a combination
ofadvancsin experimental methods that are more informative t
the atomic lve, and from anew conceptual framework de to
simpli statistical mechanical model In the new view, dings
seen asa paral low proces of an ensemble of chain molecules.
In te metaphor of energy landscape, ding is sen x mor ke
the tickle of water down mountainsides of complex shapes, and
ies ike the ow though sing uly Ina nutshell the new view
puts more emphasis on ensemble and multiple folding routes and
less emphasis on specific tractres and pathways.
‘The ‘New View": what's new?
“The groundwork for protein folding kinetics was laid 35 years ago
‘The famous experiments of Christian B. Anfinsen and co-workers
beginning in the early 19606 showed that proteins can fold
reversibly, implying thatthe ative structures of sone small lobular
proteins are thermodynamicaly stable states, and therefore ae con-
formations atthe global minima oftheir acessile fre energies. But
(Gyrus Levinthal made the argument inthe ate 1960, ltr called
“Levinthals paradox", hat there are too many possible conforma-
tions for proteins to find the ‘needle’ (the native structure) in the
“haystack (conformational space) by random searching. Levithal
concluded that proteins must fold by specific folding pathway.
Levinthal framed the puzzle as i the two goalsachieving the
lobal minimum and doing so quickly—were mutualyexchsive
‘These two mutually exclusive options came to be called thermo-
Oy
then all the other bond angles inthe chain will also have paticulat
characteristic values, But if proteins fold by idealized perfect fun-
nels then the fuctuation of Gp around its mean value ata given
time during folding will not be correlated with the corresponding
fluctuation of #4; or any other bond angle. In reality, we expect
correlations hetwoen fluctuations of variables suchas 8 and 8, in
real proteins tobe intermediate between these two extreme cases,
and to change with time as folding proceeds,
“Measuring these correlations requires measuring distributions
of populations, not commonly done, not ust measuring the aver-
ages observed in most experiments. That is, since folding is coop-
crati, i means that the ensemble average value (@,) wil
naturally proceed toward is native value oer the same time course
that (0) proceeds toward its native value. Thisis not the correla-
tion we sek, Rather we ask whether the fluctuations in 8 corre-
late with the uctuations in Gy, that is, whenever there is a
transient increase in @ theres, sya tansient decrease in @ In
mathematical terms, we seck the ensemble averages of corelation
quantities such as (@@) (Fig. 8). The aim isto observe how folding
fecuces the dimensionality ofthe conformational space and how
the diffrent degrees of freedom become increasingly correlated
until no feedom remains. The funnel perspective expects few cor-
relations early in the folding proces, and increasing amounts or
correlation as folding process,
Our description above applies to some property of each
ronomer like @ dihedral angle. Other examples of monomer
properties include the solent envionment or degree of burial
which can be measured by orescence or hydrogen exchange pro-
tection, OF equal interest are pairwise quantities lke contacts or
distances between monomers J andj. In this case, isa contac, say
(G43) correlated with some other contact, say (jy j#3)2 That i
B
and
ature stu bly «valine 4 number aruary 1857
Fig. 6 Meszuring