52
A. Jayaraman et al.
CLP triple helix as evident by larger root mean square deviations (RMSD), larger
fluctuations in Ramachandran angles, and changes in the solvent environment as
compared to CLPs that have aspartic acid located at the X position (i.e., GDO).
In a follow-up study, Raman et al. used all-atom MD simulations to study (GPO)based CLPs with varying lengths from a single to ten (GPO) repeat units [98]. They
concluded that a minimum of five repeat units is required to form a stable triple
helix as shown by their RMSD, water structure and inter-molecular h-bonding analyses. Similar computational studies have also shown that the stereochemistry of
the amino acids in the X and Y positions can also affect the stability of the triple
helix. For example, the work of Punitha et al. showed that L to D substitutions of
aspartic acid, proline and alanine residues produced a kink at the site of substitution leading to a large local disruption to the triple helical structure as shown by
an absence of h-bonds at the D substituted positions [100]. Although many studies
have focused on the stability of CLP triple helices in solution, there have also been
several studies that have investigated the molecular level interactions between CLPs
and gold nanosurfaces and nanoparticles using all-atom MD simulations [101, 102].
Tang et al. showed that the CLP triple helix unfolds upon adsorption to gold nanosurfaces in which the peptide backbone adopts a flat conformation where the N–H and
carbonyl oxygens are no longer capable of h-bonding to each other, thus, inducing
significant unfolding of the triple helical structure [101]. Similar work by Gopalakrishnan et al. showed that CLPs adsorb to gold nanosurfaces with the help of OH-Au
interactions involving the OH group on the pyrrolidine ring of hydroxyproline [102].
Clearly, there are numerous atomistic simulation studies that have extensively examined the interactions of CLPs both in solution and their interactions with surfaces as
a function of CLP composition (i.e., sequence and length).
Although atomistic simulations are good at examining structural features at an
atomic level, they are unable to reproduce CLP melting due to the intractable simulation times required to observe CLP triple helix melting. The above studies show
that most computational CLP studies have been focused on atomistic rather than
coarse-grained systems. To the best of our knowledge, the only two coarse-grained
(CG) models of CLP that have been published are the CG CLP models developed by
Buehler and coworkers [103–105] and Condon and Jayaraman [106]. Therefore, there
are lots of opportunities for the development of CG models which can reproduce the
h-bonding involved in the stabilization of the CLP triple helix. Earlier versions of the
CG CLP model of Buehler and coworkers [103, 104] grouped hundreds of atoms into
particles or beads. Therefore, these models were unable to capture salient biochemical features such as amino acid sequence. Thus, inspired by their previous CG model,
Buehler and coworkers then developed a modified version of the MARTINI force
field for proteins [105] that accounts for sequence-specific biochemical features of
collagen. Furthermore, their CG model can reproduce the structural and mechanical properties (i.e., persistence length and elastic modulus) of single trospocollagen
molecules; however, these models did not focus on recapitulating the canonical N–
H–C = O h-bonding pattern in collagen even though their model did maintain the
correct triple helical structure. Given that inter-strand h-bonding is a quintessential
feature of CLP melting, our work on the development of a CG model of CLP that can
A. Jayaraman et al.
CLP triple helix as evident by larger root mean square deviations (RMSD), larger
fluctuations in Ramachandran angles, and changes in the solvent environment as
compared to CLPs that have aspartic acid located at the X position (i.e., GDO).
In a follow-up study, Raman et al. used all-atom MD simulations to study (GPO)based CLPs with varying lengths from a single to ten (GPO) repeat units [98]. They
concluded that a minimum of five repeat units is required to form a stable triple
helix as shown by their RMSD, water structure and inter-molecular h-bonding analyses. Similar computational studies have also shown that the stereochemistry of
the amino acids in the X and Y positions can also affect the stability of the triple
helix. For example, the work of Punitha et al. showed that L to D substitutions of
aspartic acid, proline and alanine residues produced a kink at the site of substitution leading to a large local disruption to the triple helical structure as shown by
an absence of h-bonds at the D substituted positions [100]. Although many studies
have focused on the stability of CLP triple helices in solution, there have also been
several studies that have investigated the molecular level interactions between CLPs
and gold nanosurfaces and nanoparticles using all-atom MD simulations [101, 102].
Tang et al. showed that the CLP triple helix unfolds upon adsorption to gold nanosurfaces in which the peptide backbone adopts a flat conformation where the N–H and
carbonyl oxygens are no longer capable of h-bonding to each other, thus, inducing
significant unfolding of the triple helical structure [101]. Similar work by Gopalakrishnan et al. showed that CLPs adsorb to gold nanosurfaces with the help of OH-Au
interactions involving the OH group on the pyrrolidine ring of hydroxyproline [102].
Clearly, there are numerous atomistic simulation studies that have extensively examined the interactions of CLPs both in solution and their interactions with surfaces as
a function of CLP composition (i.e., sequence and length).
Although atomistic simulations are good at examining structural features at an
atomic level, they are unable to reproduce CLP melting due to the intractable simulation times required to observe CLP triple helix melting. The above studies show
that most computational CLP studies have been focused on atomistic rather than
coarse-grained systems. To the best of our knowledge, the only two coarse-grained
(CG) models of CLP that have been published are the CG CLP models developed by
Buehler and coworkers [103–105] and Condon and Jayaraman [106]. Therefore, there
are lots of opportunities for the development of CG models which can reproduce the
h-bonding involved in the stabilization of the CLP triple helix. Earlier versions of the
CG CLP model of Buehler and coworkers [103, 104] grouped hundreds of atoms into
particles or beads. Therefore, these models were unable to capture salient biochemical features such as amino acid sequence. Thus, inspired by their previous CG model,
Buehler and coworkers then developed a modified version of the MARTINI force
field for proteins [105] that accounts for sequence-specific biochemical features of
collagen. Furthermore, their CG model can reproduce the structural and mechanical properties (i.e., persistence length and elastic modulus) of single trospocollagen
molecules; however, these models did not focus on recapitulating the canonical N–
H–C = O h-bonding pattern in collagen even though their model did maintain the
correct triple helical structure. Given that inter-strand h-bonding is a quintessential
feature of CLP melting, our work on the development of a CG model of CLP that can