Computer-assisted

design of nanomedicine
for cancer therapeutics and diagnostics

Prof. Sabrina Pricl, Molecular Biology and

Nanotechnology Laboratory
(MolBNL@UniTS), DEA, University of Trieste,
Trieste, Italy

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 1

 Chemistry by experiment
•  Chemists synthesize
molecules
–  Typically, they do this by
seeking connections
between molecular
structure and property/
activity

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 2

 Chemistry by computer
•  Computer scientists
–  Same goal is exactly and
same procedure, yet…
–  The difficult step
(materials preparation)
is removed.

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 3

 Why should we do molecular
simulations ?
•  For effective design and synthesis we need to:
–  Learn to think like a molecule
•  To do this we need to:
–  see what a molecule sees and
–  feel what a molecule feels
•  Molecular models give us the best and most direct
view of the molecular world
•  Molecular models are easy to use, inexpensive and
safe

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 4

 Should molecular simulations replace
experimental chemistry?
•  Of course not!
–  The goals of chemistry are not changed by molecular
simulations
•  On a practical level we want to learn how:
–  to make things (synthesis)
–  and how to figure out what things are made of
(analysis)
•  On an intellectual level we want to:
–  understand the rules that describe chemical and
physical behaviors

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 5

 Should molecular simulations replace
experimental chemistry?
•  Like e.g., IR or NMR, molecular simulations are a tool
for achieving the synthesis and analysis goals
•  Since these two goals are experimental, they cannot -
(and should not - be done away with
•  However, molecular simulations do change the way
we do syntheses and analysis
•  And they speak directly to the intellectual goals of
chemistry
•  A modern chemical education requires practical
training BOTH in experimentation and in molecular
simulations

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 6

 Keywords: length and time scale
•  In today’s parlance, molecular simulations are
classified according to the length and time scale of
the phenomena they can explore
•  As rule of thumb:
–  quantum mechanics
•  electron distributions
•  ≈ 10-10 m and 10-12 s
–  molecular simulations
•  atom and molecule distributions
•  ≈ 10-9 m and 10-9 - 10-6 s
–  mesoscale simulations
•  coarse-grained beads or super-atoms
•  ≈ 10-6 m and 10-6 – 10-3 s
–  macroscopic simulations
•  continuum models
•  ≥ 10-3 m and ≥ 10-3 s

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 7

 Keywords: length and time scale

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 8

 KEY ASPECTS OF QM SIMULATIONS

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 9

Quantum Chemistry – electron density
•  Electron densities (ED)
shows the locations of
electrons
–  Larger ED values reveal
atomic positions and
chemical bonds (X-ray
diffraction experiments)
–  Smaller ED values
indicate overall
molecular size

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 10

 Quantum Chemistry – bonding
•  EDs assume no prior knowledge about
bonding
à can be used to elucidate bonding
•  Example: QM calculation for diborane (B2H6)

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 11

Quantum Chemistry – transition states
•  EDs allow description of bonding in transition
states (no experimental info available)
•  Example: pyrolysis of ethyl formate

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 12

Quantum Chemistry – electrostatic potential

•  Electrostatic potential (EP)

–  energy of interaction of a point
positive charge (electrophile) with
the nuclei and electrons of a
molecule
–  Negative electrostatic potentials
indicate areas that are prone to
electrophilic attack
•  Example: benzene and pyridine
10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 13
Quantum Chemistry – electrostatic potential

•  EPs can be mapped onto EDs to

visualize both molecular size and
shape and molecular
electrostatic potential (EP maps)
•  Red colors à negative EP values
•  Blue colors à positive EP values

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 14

Quantum Chemistry – electrostatic potential

•  EP maps convey information molecular charge

distributions
•  Example: β-alanine (zwitterionic form)

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 15

Quantum Chemistry – electrostatic potential

•  EP maps also inform about charge

delocalization
•  Example: benzyl planar vs. perpedicular cation

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 16

Quantum Chemistry – electrostatic potential

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 17

 Quantum Chemistry – Molecular orbitals

•  Molecular orbitals
–  solutions of the approximate quantum mechanical
equations of electron motion
–  sums and differences of atomic solutions (atomic
orbitals)
–  for very simple molecules may interpreted in
terms of familiar chemical bonds
•  Example: acetilene

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 18

 Quantum Chemistry – Molecular orbitals

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 19

 Quantum Chemistry – Molecular orbitals

•  Molecular orbitals
–  solutions of the approximate quantum mechanical
equations of electron motion
–  sums and differences of atomic solutions (atomic
orbitals)
–  Or nonbonded lone pairs
–  Example: sulfur tetrafluoride

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 20

 Quantum Chemistry – Molecular orbitals

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 21

 Quantum Chemistry – Molecular orbitals

•  Unoccupied molecular orbitals may also

provide useful information
•  Example: LUMO in planar and perpendicular
benzyl cation

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 22

 Quantum Chemistry – Molecular orbitals

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 23

 Quantum Chemistry – Molecular orbitals

•  Molecular orbitals may also be mapped onto

EP maps
•  Example: LUMO of cyclohexenone

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 24

 CHIP vs. WET LAB
•  Chemistry with computer is a
multi-step process
–  not so different from experimental
chemistry
•  Main advantage of this approach
–  requires the fewest resources
•  hardware and software support
•  student training

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 25

 Potential energy surface
•  A potential energy surface is
a plot of energy vs. reaction
coordinate
•  It connects reactants to
products via a transition
state
•  Energy minima correspond
to equilibrium structures
•  Energy maxima correspond
to transition state structures

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 26

 Potential energy surface
•  Relative energies of
equilibrium structures
–  relative reactants/
products stabilities
•  Reaction
thermodynamics
•  The energy of the
transition state relative
to the equilibrium
structures
–  relative difficulties in
passing from reactants
to products
•  Reaction kinetics

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 27

 Potential energy surface
•  Computational chemitry is primarily a tool for
calculating the energy of a given molecular
sysyem
•  Thus, the first step in designing an in silico
investigation is to define the problem as one
involving a structure-energy relationship
•  There are two conceptually different ways of
thinking about energy
–  Quntum mechanics (QM)
–  Molecular mechanics (MM)

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 28

 QM – A physicist’s model
•  The energy of a molecule is
defined in terms of
interactions among nuclei
and electrons
–  the (infamous) Schrödinger
equation
•  The square of the
wavefunction
–  probability of finding an
electron
•  This is the electron density
–  Experimentally obtained
from X-ray diffraction
experiments

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 29
 QM – A physicist’s model

•  Schroedinger’s equation
exact solutions
(wavefunctions) for the H
atom are the familiar s, p,
d . . . atomic orbitals

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 30

 QM – A physicist’s model
•  While the Schrödinger equation is easy to
write down for many-electron atoms and
molecules, it is impossible to solve
–  Approximations are needed

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 31

 QM – A physicist’s model

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 32

 QM – Practical methods
•  Semiempirical methods •  QM Software
–  MNDO –  CASTEP
–  MINDO/3 –  CP2K
–  AM1 –  Quantum Espresso
–  PM3 –  Siesta
–  SAM1… –  Yambo
•  Ab initio methods –  DMOL3…
–  Self-consistent field methods •  For a complete list of free
–  Valence bond theories and commercial available
–  Density functional theory QM software
–  https://en.wikipedia.org/
wiki/
List_of_quantum_chemistry_
and_solid-
state_physics_software

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 33

 MM – A chemist’s model
•  Molecular mechanics
describes the energy of a
molecule in terms of a
simple function which
accounts for
–  distortion from ideal bond
distances and angles
–  for nonbonded van der Waals
interactions
–  and for electrostatic (aka
Coulomb) interactions

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 34

 MM – A chemist’s model
•  The forcefield concept

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 35

 MM – A chemist’s model
Ebond is linked to bond length between atoms

1 2
Eijbond (
= Kij bij − bij0
2 )
Eangle considers the contribution to energy due to the deformation of the angle

1 0 2
Eijbond (
= Kijk θ ijk − θ ijk
2
)
Etors is related to torsion around a chemical bond

1
tors
Eiklj =
2
[
En 1 − cos(nφijkl ) ]
⎡ ⎛ 0 ⎞9 0 ⎞6 ⎤
⎢ ⎜ rij ⎟ ⎛
qi q j
EvdW = ∑ ε ij ⎢2⎜ 3⎜ rij ⎟ ⎥
E elec = ∑ ⎟⎟ −
⎜ rij ⎟ ⎥
i> j rij i> j ⎜ r
⎢ ⎝ ij ⎠ ⎝ ⎠ ⎥
⎣ ⎦

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 36

 Molecular dynamics (MD) – A chemist’s model

•  Mimics what atoms do in real life, assuming a given

potential energy function (forcefield)
–  The energy function allows us to calculate the force
experienced by any atom given the positions of the other
atoms
–  Newton’s laws tell us how those forces will affect the
motions of the atoms

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 37

 MD basic algorithm

•  Divide time into discrete time steps, no more than a

few femtoseconds (10–15 s) each
•  At each time step:
–  Compute the forces acting on each atom, using a
molecular mechanics forcefield
–  Move the atoms a little bit: update position and velocity of
each atom using Newton’s laws of motion F = ma
•  Recall that F(x) = −∇U (x)
•  Velocity v is the derivative of position, and
accelertion a is the derivative of velocity
10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 38
 MD basic algorithm
•  This is a system of ordinary
differential equations
–  For n atoms we have 3n
position coordinates and 3n
velocity coordinates
•  Analytical (algebraic) solution
is impossible
•  Numerical solution is
straightforward

1st CA17140 TS, University of Trieste, April

10/04/19 39
8-11 2019
 KEY ASPECTS OF MD SIMULATIONS

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 40

 Atoms never stop jiggling
•  In real life – and in a MD
simulation – atoms are in
constant motion
•  Given enough time, the
simulation samples the
Boltzmann distribution
–  i.e., the probability of
observing a particular
arrangement of atoms is a
function of the potential energy
•  This is not the only way to
explore the energy surface but
it is a well efficient method to
do so

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 41

 Energy conservation
•  Total energy (potential + kinetic) must be
conserved
–  In atomic arrangements with lower potential
energy atoms move faster
–  A mechanisms (thermostat) is added to keep
the temperature constant

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 42

 Water is important
•  Ignoring the solvent leads to major artifacts
–  Water, salt ions (e.g., sodium, chloride), lipids of the cell
membrane
•  Two options for taking solvent into account
–  Explicitly represent solvent molecules
•  Highly computational expensive but accurate
•  Usually assume periodic boundary conditions (a water molecule
that goes off one side of the simulation box will come back in the
opposite side, like in PacMan)
–  Implicit solvent
•  Matematical model to approximate average effects of solvent
•  Less accurte but faster

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 43

 MD main properties calculation
•  Structural properties
–  Geometrical features
–  Mass distributions
•  Physical properties
–  Stabiliy
–  Diffusion
–  Phase equilibria
•  Mechanical properties
–  Stress-strain behavior
–  Rheological behavior
•  Biomolecular interactions
–  Binding energies
10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 44
 MD limitations - Timescales
•  Simulations require short time steps for numerical stability
–  1 time step ≈ 2 fs (2×10–15 s)
•  Biological processes can take nanoseconds (10–9 s), microseconds
(10–6 s), milliseconds (10–3 s), or longer
–  Millions to trillions of sequential time steps for nanosecond to
millisecond events (and even more for slower ones)
•  Until recently, simulations of 1 microsecond were rare
•  Advances in computer power have enabled microsecond
simulations, but simulation timescales remain a challenge
•  Enabling longer-timescale simulations is an active research area,
involving:
–  Algorithmic improvements
–  Parallel computing
–  Advanced MD techniques
–  Hardware: GPUs, specialized hardware

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 45

 MD limitations – Forcefield accuracy
•  Molecular mechanics forcefields
inherently are approximations
•  They have improved substantially
over the last decade, but many
limitations remain
•  In practice, one needs a strong
experience to know how to perform
and what to trust in a simulation

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 46

MD limitations – no chemical reactions
•  Once a molecule is created, most of its covalent
bonds do not break or form during typical
physiological function
•  However, a few covalent bonds do form and
break more frequently (in real life):
–  Disulfide bonds between cysteines
–  Acidic or basic amino acid residues can lose or gain a
hydrogen (i.e., a proton)
–  ….
•  And some chemical reactions do occur!
•  In these cases, reactive forcefields and or QM/
MD mixed techniques must be selected
10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 47
 MD practical methods
•  Main MD SW •  Main Forcefields for MD
–  AMBER –  AMBER
–  LAMMPS –  CHARMM
–  NAMD –  GROMOS
–  CHARMM –  UNIVERSAL
–  GROMACS… –  MMFF….
•  For a complete list of free •  For more details on FF:
and commercial available –  https://en.wikipedia.org/
QM software: wiki/
–  https://en.wikipedia.org/ Force_field_(chemistry)
wiki/
Comparison_of_software_f
or_molecular_mechanics_
modeling

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 48

 KEY ASPECTS OF MESOSCALE
SIMULATIONS

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 49

 Mesoscale simulation techniques
•  Introduce simplifications to atomistic methods to remove
the faster degrees of freedom and/or treat groups of atoms
(blobs of matter) as individual entities interacting through
effective potentials
•  Pros:
–  Can be used to study structural features of complex systems
with 108-109 atoms
–  Can study dynamical processes on timescales inaccessible to
classical methods, even up to 1s
•  Cons:
–  Can often describe only qualitative tendencies, the quality of
quantitative results may be difficult to ascertain
–  In many cases, the approximations introduced limit the ability
to physically interpret the results

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 50

 Mesoscale simulation techniques
•  Dissipative particle dynamics (DPD)
–  Direct simulation Monte Carlo (DSMC) method
–  Lattice Boltzmann method….
•  In DPD
–  The fundamental particles are beads that
comprise a number of atoms
–  All forces between particles smaller than the bead
are averaged out
•  only the resulting forces between the beads are
modeled

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 51

 The DPD particle concept
What is a DPD particle?

q  molecule

q  part of a molecule

q  fluid element = many molecules

The colored area symbolizes the interaction range of all

forces acting on a particle

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 52

 The mesoscale molecule

Atomistic and mesoscopic model of the same

complex macromolecule
10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 53
 DPD basic math
Instantaneous force acting on particle i:
S A
Fi (t ) = ∑[ FijC + FijD + FijR ]+ ∑ f + ∑ f ij ijz
j ≠i j jz

between no bonded particles between bonded particles

S
F C
ij conservative force f ij
spring force
D
F
A
ij
R
dissipative force f ijz angle force
F ij random force

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 54

 DPD basic math
In DPD the force between particles i and j is given by the sum:

F DPD
ij
C
=F +F +F
ij
D
ij
R
ij
FijC conservative force

FijD dissipative force

R
Fij random force

Instantaneous force acting on particle i:

[
Fi (t ) = ∑ FijC + FijD + FijR ]
j ≠i

We assume pairwise additivity for all three types of force

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 55

 DPD equation of motion
dri dv i
= v i (t ) = f i (t )
dt dt
where
q  Beads follow Newton’s equation of motion

q  fi is the total force acting on particle i

q  All the masses are normalized to 1 for simplicity. All beads have the

same volume.
q  Integration of the equation of motion as in MD
q  Conservative, Dissipative and Random force obey Newton’s third law
Fij = - Fji :
- mass, momentum and energy are conserved
- hydrodynamic behavior correctly treated

DPD looks like MD

10/04/19 1st CA17140 TS, University of Trieste, April 8-11 2019 56