Coarse-Grained Modeling and Simulations of Thermoresponsive …
41
(i.e., besides DNA or RNA), and development of new force fields is time consuming.
Thus, CG models, where the degrees of freedom are reduced as compared to an allatom description, are more appropriate for studies of novel synthetic DNA mimics
with varying backbone chemistries at experimentally relevant length and time scales.
CG modeling of DNA has received significant attention in the past few decades;
a review article from Theodorakis et al. [59] summarizes the CG modeling of DNA,
especially aimed at studying DNA-functionalized nanoparticles. We highlight a few
models here, in particular, the three-site-per-nucleotide (3SPN) [10–12] and OxDNA
[13, 14] models. Both these models are known to reproduce experimentally observable mechanical and structural properties of DNA and RNA. However, the extensive
list of bonded and non-bonded potentials used in these models makes it difficult
to re-parameterize these CG models for other ONAs. Some coarser CG models
use two-site-per-nucleotide, where one site represents the backbone bead and the
other represents the nucleobase. While such models can successfully reproduce the
hybridization thermodynamics observed in experiments, they do not capture the
details of the secondary structure (e.g., helical turns, major and minor grooves). Starr
and Sciortino [15] developed a two-site-per-nucleotide model to study the assembly
and gelation of DNA dendrimers. Later, this model has also been used to study the
stability of DNA-linked nanoparticle crystal [17] and DNA-functionalized nanoparticles [60]. In their model, the directional h-bonding interactions are incorporated
using small sticky beads. The model, however, does not distinguish A, T, G and C
and it does not include varying base–base stacking interactions depending on the
identity of the bases in the DNA strand. Knorowski et al. [18, 61] used a similar
two-site-per-nucleotide model extended from the Starr and Sciortino [15] model, to
study dynamics of DNA-programmable nanoparticle self-assembly. Instead of sticky
beads, they used a smaller bead with attractive interactions to model the h-bonding
interactions and a flanking bead to ensure the directionality of the h-bonds. Li et al.
[62] studied the crystallization of nucleic acid–nanoparticle conjugates using a modified version of the model developed by Knorowski et al. [61] where they introduced
the effect of chain stiffness. Kenward and Dorfman [63, 64] considered base–base
stacking interactions in addition to identifying complementary bases in their CG
model. Ding and Mittal [65] introduced electrostatic interactions between backbone–backbone beads using a short-range Yukawa potential to the model developed
by Kenward and Dorfman [63].
Inspired by the early work of Starr and Sciortino [15], Ghobadi and Jayaraman [19]
developed a similar CG model for ONAs in general. This model ensures specificity
in the h-bonding interactions, distinguishes the strength of base–base stacking interactions for various pairs of adjacent nucleobases in a strand and includes backbone
electrostatics, which put together replicate experimental DNA melting temperature
trends. They conducted a systematic study to understand the effect of ONA backbone
design, in particular, its flexibility and presence/absence of charged groups, on its
melting temperature. They observed that for 8 base pairs long/8-mer ONAs, electrostatically neutral backbone ONAs have higher melting temperatures as compared
to ONAs with negatively charged backbone, and increasing the flexibility of ONAs
reduces the ONA melting temperature. They then extended the model to ONAs
41
(i.e., besides DNA or RNA), and development of new force fields is time consuming.
Thus, CG models, where the degrees of freedom are reduced as compared to an allatom description, are more appropriate for studies of novel synthetic DNA mimics
with varying backbone chemistries at experimentally relevant length and time scales.
CG modeling of DNA has received significant attention in the past few decades;
a review article from Theodorakis et al. [59] summarizes the CG modeling of DNA,
especially aimed at studying DNA-functionalized nanoparticles. We highlight a few
models here, in particular, the three-site-per-nucleotide (3SPN) [10–12] and OxDNA
[13, 14] models. Both these models are known to reproduce experimentally observable mechanical and structural properties of DNA and RNA. However, the extensive
list of bonded and non-bonded potentials used in these models makes it difficult
to re-parameterize these CG models for other ONAs. Some coarser CG models
use two-site-per-nucleotide, where one site represents the backbone bead and the
other represents the nucleobase. While such models can successfully reproduce the
hybridization thermodynamics observed in experiments, they do not capture the
details of the secondary structure (e.g., helical turns, major and minor grooves). Starr
and Sciortino [15] developed a two-site-per-nucleotide model to study the assembly
and gelation of DNA dendrimers. Later, this model has also been used to study the
stability of DNA-linked nanoparticle crystal [17] and DNA-functionalized nanoparticles [60]. In their model, the directional h-bonding interactions are incorporated
using small sticky beads. The model, however, does not distinguish A, T, G and C
and it does not include varying base–base stacking interactions depending on the
identity of the bases in the DNA strand. Knorowski et al. [18, 61] used a similar
two-site-per-nucleotide model extended from the Starr and Sciortino [15] model, to
study dynamics of DNA-programmable nanoparticle self-assembly. Instead of sticky
beads, they used a smaller bead with attractive interactions to model the h-bonding
interactions and a flanking bead to ensure the directionality of the h-bonds. Li et al.
[62] studied the crystallization of nucleic acid–nanoparticle conjugates using a modified version of the model developed by Knorowski et al. [61] where they introduced
the effect of chain stiffness. Kenward and Dorfman [63, 64] considered base–base
stacking interactions in addition to identifying complementary bases in their CG
model. Ding and Mittal [65] introduced electrostatic interactions between backbone–backbone beads using a short-range Yukawa potential to the model developed
by Kenward and Dorfman [63].
Inspired by the early work of Starr and Sciortino [15], Ghobadi and Jayaraman [19]
developed a similar CG model for ONAs in general. This model ensures specificity
in the h-bonding interactions, distinguishes the strength of base–base stacking interactions for various pairs of adjacent nucleobases in a strand and includes backbone
electrostatics, which put together replicate experimental DNA melting temperature
trends. They conducted a systematic study to understand the effect of ONA backbone
design, in particular, its flexibility and presence/absence of charged groups, on its
melting temperature. They observed that for 8 base pairs long/8-mer ONAs, electrostatically neutral backbone ONAs have higher melting temperatures as compared
to ONAs with negatively charged backbone, and increasing the flexibility of ONAs
reduces the ONA melting temperature. They then extended the model to ONAs