Exploring the Implications of Vitamin B12 Conjugation to Insulin on Insulin Receptor Binding and Cellular Uptake

In press, in the journal ChemMedChem (and, because I think it's hip, I note that the current "obligatory" image for the wikipedia article for ChemMedChem features the image I made for the review article on the topic addressed in this new study). As with many theory papers (there's some experiment in there, too), this very brief article summarizes several months of cyanocobalamin (B12) parameterization and molecular dynamics (MD) simulations. The purpose of the theory was to address all of the major structural snapshots in the uptake process associated with the insulin-B12 bioconjugate being developed as part of the much heralded oral insulin project in Robert Doyle's group here at Syracuse. These structures include:

1. The structure and dynamic properties of the insulin-B12 bioconjugate
2. The binding of B12 to Transcobalamin II (TCII) (for B12 parameterization)
3. The binding of the insulin-B12 bioconjugate to TCII (and the steric demands therein)
4. The interaction of the insulin-B12 bioconjugate, bound to TCII, with the insulin Receptor (IR)

The quantum chemical (for the B12 geometry and missing force constants) and molecular dynamics (GROMACS with the GROMOS96 (53a6)) simulation work is going to serve as the basis for several posts here (eventually) about parameterization, topology generation, and force field development.

As an example of some of the insights modeling provides, the figure above shows the insulin-B12 bioconjugate (the insulin is divided into A and B chains, the A chain in blue and the important division of the insulin B chain in the front half of the rainbow). Insulin is a rather large-scale example of many of the same molecular issues that arise in the analysis of solid-state molecular crystals by either terahertz or inelastic neutron scattering spectroscopy. The packing of molecules in their crystal lattices can lead to significant changes in molecular geometry, be these changes in the stabilization of higher-energy molecular conformations or even deformations in the covalent framework. In the case of insulin, it is found that the crystal geometry (also the geometry of stored insulin in the body) is quite different from the solution-phase form. It's even worse! The B chain end (B20-B30) in the solid-state geometry covers (protects?) the business-end of the insulin binding region to the Insulin Receptor. One can imagine the difficulty in proposing the original binding model for insulin to its receptor from the original crystal data given that the actual binding region is blocked off in the solid-state form! The "Extended" form in the figure is representative of "multiple other" conformations of the B20-B30 region (which mimics the characterized T-state of insulin), those geometries for which the insulin binding region (blue and green) is completely exposed. This extended geometry is also the one that separates the bulk of the insulin structure from the covalently-linked B12 (at Lys29) and, it is argued from the MD simulations in the paper, enables the B12 to still tightly bind to TCII despite the presence of all this steric bulk.

Amanda K. Petrus1, Damian G. Allis1, Robert P. Smith2, Timothy J. Fairchild3 and Robert P. Doyle1

1. Department of Chemistry, Syracuse University, Syracuse, NY 13244, USA
2. Department of Construction Management and Wood Products Engineering, SUNY, College of Environmental Science and Forestry, Syracuse, NY 13210, USA
3. Department of Exercise Science, Syracuse University, Syracuse, NY 13244, USA

Extract: We recently reported a vitamin B12 (B12) based insulin conjugate that produced significantly decreased blood glucose levels in diabetic STZ-rat models. The results of this study posed a fundamental question, namely what implications does B12 conjugation have on insulin's interaction with its receptor? To explore this question we used a combination of molecular dynamics (MD) simulations and immuno-electron microscopy (IEM).


The Solid-State Terahertz Spectrum of MDMA (Ecstasy) – A Unique Test for Molecular Modeling Assignments

In press, in Chemical Physics Letters (CPL).  Yes, the blog has taken a bit of a turn, from high explosives to illicit drugs.  I expect my google rating to rise sharply with this post.  The protonated form of the 3,4-methylene-dioxymethamphetamine (Ecstasy, but I'll keep the post legit, so it is herein referred to as MDMA) molecule (herein referred to as MDMA:H+) and MDMA:H+ in its crystal cell with a chloride ion (Cl, the crystal form herein referred to as MDMA:HCl) is shown below in yet another fantastic NanoEngineer-1 rendering (if I do say so myself).

This CPL article is, to some extent, a response to those in the terahertz community who continue to attempt spectral assignments of crystalline and poly-crystalline samples using isolated-molecule quantum chemical calculations.  The MDMA:HCl sample and MDMA molecule, as a pair, are a very interesting case study of theory and experiment for reasons detailed below.  The spectra, shown below from a previous version of the paper (but the same spectra), show quite a bit of detail that will make sense shortly.

Panel A shows the isolated-molecule calculation for the neutral MDMA molecule at a B3LYP/6-31G(d) level of theory (in red).  You will note that this simulated spectrum is in very good agreement with experiment (in black), reproducing all of the major features and showing a number of smaller features that account for shoulders.  This agreement was the basis for the assignment of the MDMA:HCl spectrum reported in: G Wang, J Shen, Y Jia. "Vibrational spectra of ketamine hydrochloride and 3,4-methylenedioxymethamphetamine in terahertz range." Journal of Applied Physics 102 (2007) 013106/1-06/4.

The new theoretical analysis reported in the CPL article was instigated by this assignment in this previous publication.  Relevant to the measured sample and the previously reported assignment, two points arise that require address.

1. The previous calculation, as reported, was of the neutral MDMA molecule and is reasonably close to the MDMA spectrum shown above (in red.  This calculation was redone for the CPL article for comparative purposes).  As the experimental THz sample was of solid-state MDMA:HCl, the appropriate form of the molecule to run is not the neutral MDMA molecule, but the protonated form, MDMA:H+.  The protonated form has a different vibrational spectrum (shown in green in Panel A) than the isolated molecule form.  At the very least, the isolated-molecule to consider for the MDMA:HCl sample must be the protonated form.  Interestingly, the re-calculation at B3LYP/6-31G(d) reported for the CPL article predicts a fifth vibrational mode at 48.0 cm-1 that was not reported in the previous study.  We do not know if the previous group missed that peak in the write-up, decided that (since it is a low-intensity mode) it was not worth reporting, or if their starting molecular geometry was somehow different so that the other four modes were predicted to be in the same region and this mode was somehow turned off.

2. The solid-state spectrum shown in Panel B at a BP/DNP level of theory does not agree as well as the isolated-molecule MDMA B3LYP/6-31G(d) calculation. That being said, THAT IS NOT THE POINT.  The goal of a simulated spectrum IS NOT to obtain the closest spectral agreement with experiment.  The goal IS to explain the solid-state spectrum with the best theoretical model possible that, hopefully, is as close to the experimental result as possible.  In this case, the solid-state BP/DNP spectrum contains a finite number of vibrational modes that do group according to features in the THz spectrum, making the assignment reasonably straightforward.  Interestingly, the two most intense modes in the solid-state BP/DNP calculations involve the motions of the Cl…H+-N chains, which CANNOT be accounted for in an isolated-molecule calculation of either the neutral MDMA molecule or the protonated MDMA:H+.

In summary, as taken from the CPL paper:

With all of these considerations taken into account in this re-examination of the MDMA.HCl THz spectrum, it is found that this system serves as a fortuitous example of one whose THz spectrum is predicted quite precisely by two very different approaches, but is only described accurately by one that considers the crystal environment and the actual state of the molecule in its solid-state form.

Damian G. Allis1,2, Patrick M. Hakey1, and Timothy M. Korter1

1. Department of Chemistry, Syracuse University, Syracuse NY 13244-4100 USA
2. Nanorex, Inc., P.O. Box 7188, Bloomfield Hills, MI 48302-7188 USA

Abstract: The terahertz (THz, far-infrared) spectrum of 3,4-methylene-dioxymethamphetamine hydrochloride (Ecstasy) is simulated using solid-state density functional theory.  While a previously reported isolated-molecule calculation is noteworthy for the precision of its solid-state THz reproduction, the solid-state calculation predicts that the isolated-molecule modes account for only half of the spectral features in the THz region, with the remaining structure arising from lattice vibrations that cannot be predicted without solid-state molecular modeling.  The molecular origins of the internal mode contributions to the solid-state THz spectrum, as well as the proper consideration of the protonation state of the molecule, are also considered.