minimum highest temperature to employ. In the case of our p53
example, a good scheme could have eight replicas (at 296,
298, 300, 308, 320, 332, 345, and 358 K) where the width of
the energy distribution of all the replicas was increased except for a
“neutral” 298 K replica [62].
All replicas are further subjected to an additional biasing force
through metadynamics in which a Gaussian of width 0.1 nm in all
the CV dimensions (i.e., the four distances in our example) is
deposited in the collective variable space every 4 ps with an initial
height of 0.12 kcal/mol and a bias factor of 6. All these parameters,
i.e., the deposition time, the initial height of the Gaussian and the
bias factor could be tuned according to the process of interest. We
used quite mild biases since the changes in free energy among the
minor and major states were not expected to be high, and we aimed
to reproduce them accurately.
In our example (Fig. 4), the simulations were run for at least
300 ns per replica, checking the evolution of the monodimensional
free energy surface (FES) along each collective variable and also the
evolution of two-dimensional FES over the simulation time using
the sumhills preprocessing tool of Plumed. In an ideal scenario, we
could interrupt the simulation when we are confident that we are
sampling the changes in the collective variable space multiple times
(i.e., we observe multiple events of opening/closing of the loop).
This is required to achieve the sufficient statistical power. Another
criterion is to verify that the FES does not change remarkably over
time, and the relevant minima have been explored.
3.7 Identification
of Biological Partners
Recruited
at the Long-Range
DNA- Modulated Sites
Once the structural mechanism behind long-range communication
or allostery has been unveiled, in a case as p53, it becomes crucial to
give a biological rationale to it. Does this conformational change
have any meaning from the biological point of view? Alternatively, it
is just an unrelated event to protein function?
In the p53 case, which needs to interact with multiple partners,
an obvious working hypothesis could be that the distal region acts
as an interface for recruitment of other biological partners and that
the conformational change can either “activate” or “inactivate” this
function.
Thus, we can first retrieve the available information on p53
partners using databases where experimental (or predicted but to
be taken with caution) protein-protein interactions are annotated.
For example, the I2D database can be used since it acts as a
“metaserver” integrating annotations from different sources,
included the literature. The pool can always be enriched by manual
annotation from recent literature or other databases. The target list
can also be pruned according to the CRAPome definition from hits
that are likely to be artifacts in proteomics [125]. Once the target
proteins have been identified, it becomes crucial to retain only
those for which at least one experimental structure is available in
Dynamics of p53
237
example, a good scheme could have eight replicas (at 296,
298, 300, 308, 320, 332, 345, and 358 K) where the width of
the energy distribution of all the replicas was increased except for a
“neutral” 298 K replica [62].
All replicas are further subjected to an additional biasing force
through metadynamics in which a Gaussian of width 0.1 nm in all
the CV dimensions (i.e., the four distances in our example) is
deposited in the collective variable space every 4 ps with an initial
height of 0.12 kcal/mol and a bias factor of 6. All these parameters,
i.e., the deposition time, the initial height of the Gaussian and the
bias factor could be tuned according to the process of interest. We
used quite mild biases since the changes in free energy among the
minor and major states were not expected to be high, and we aimed
to reproduce them accurately.
In our example (Fig. 4), the simulations were run for at least
300 ns per replica, checking the evolution of the monodimensional
free energy surface (FES) along each collective variable and also the
evolution of two-dimensional FES over the simulation time using
the sumhills preprocessing tool of Plumed. In an ideal scenario, we
could interrupt the simulation when we are confident that we are
sampling the changes in the collective variable space multiple times
(i.e., we observe multiple events of opening/closing of the loop).
This is required to achieve the sufficient statistical power. Another
criterion is to verify that the FES does not change remarkably over
time, and the relevant minima have been explored.
3.7 Identification
of Biological Partners
Recruited
at the Long-Range
DNA- Modulated Sites
Once the structural mechanism behind long-range communication
or allostery has been unveiled, in a case as p53, it becomes crucial to
give a biological rationale to it. Does this conformational change
have any meaning from the biological point of view? Alternatively, it
is just an unrelated event to protein function?
In the p53 case, which needs to interact with multiple partners,
an obvious working hypothesis could be that the distal region acts
as an interface for recruitment of other biological partners and that
the conformational change can either “activate” or “inactivate” this
function.
Thus, we can first retrieve the available information on p53
partners using databases where experimental (or predicted but to
be taken with caution) protein-protein interactions are annotated.
For example, the I2D database can be used since it acts as a
“metaserver” integrating annotations from different sources,
included the literature. The pool can always be enriched by manual
annotation from recent literature or other databases. The target list
can also be pruned according to the CRAPome definition from hits
that are likely to be artifacts in proteomics [125]. Once the target
proteins have been identified, it becomes crucial to retain only
those for which at least one experimental structure is available in
Dynamics of p53
237