Chapter 7. Conclusions.

7.1. Summary

One goal of our research is determine what are the conformations and configurations of the tetrahedral intermediates in carnitine acyl transfers. So far, the approach has been experimental, by designing conformationally constrained analogues of the tetrahedral intermediates. To make the approach cost-effective, we need reliable modeling methods to guide the choice of the next generation of inhibitors. The modeling methods developed in this dissertation fulfill that need for the most part.

This dissertation addressed two additional questions:

• Why does acetylcarnitine have such a high free energy of hydrolysis?
• What are the conformations of our heterocyclic moieties that could best fit the tetrahedral intermediate for acetylcarnitine-CoA exchange?

From the preliminary results for the modeling of the tetrahedral intermediate, [shown in the table in chapter 3,] we see the RMS fits of the dioxazaphosphacinium rings to the intermediate. The most promising candidates are both boat-boat conformers that correspond to the R,S diastereomer of the tetrahedral intermediate. The cis-substituted ring and the dimethyl-substituted ring yield the closest fit to the R,S tetrahedral intermediate. These fits should be considered as one parameter in a quantitative structure-activity relationship for the inhibitory potency of the inhibitors already developed by our group. This will help in the prediction of the next generation of inhibitors. The morpholinium rings should receive a similar treatment.

• Upon hydrolysis, the zwitterion of acetylcarnitine keeps a nearly constant dipole moment thanks to the influence of conformational flexibility on the positioning of the charged centers. This keeps the solvation energy nearly constant, leaving more chemical energy available to be released.
• Conformational change can act as a mechanism for the storage of chemical energy.
• In morpholinium rings, the conformational preference in aqueous media obeys more the increased steric interactions due to the shorter C-N and C-O bonds than the differences in dipole moment.
• Gem substitution in morpholinium rings can help control the relative stabilities of the chair and the twist-boat.
• Addition of substituents in a 1,1,3,3-digem pattern on dioxazaphosphacinium rings can favor a twist-boat global minimum over a boat-chair conformation.
• Differences in polarity in dioxazaphosphacinium rings override steric effects, when methyl groups are involved, in determining the preferred conformation in aqueous solution.
• We have developed parameters for accurate modeling of neutral dialkyl alkylphosphonates by molecular mechanics (AMBER*). We used three measures of the goodness of fit when testing against ab initio results and crystal structures, and found our parameters more reliable than other commonly used force fields (MM2*, MMX and UFF)

7.2. Future work.

Based on the results obtained from the acetylcarnitine hydrolysis, the aqueous modeling of the heterocycles including a methylenecarboxylate should pay attention to the influence of the dipole moments. In the dioxazaphosphacinium rings, there will be two sources of polarity: the dipoles due to the C-O-P=O torsions, and the dipole due to the zwitterion. How will the conformational preference of the rings be affected by the presence of the zwitterion?. Answering such question will require careful attention to the charge set used in the calculation. It may be desirable to determine a new charge set for the methylenecarboxylate-substituted ring, at the HF/6-31+G* level of theory using the CHelpG scheme, to ensure that the charge set found in this dissertation is still applicable (see the file BondPC.txt for the output from the CHelpG calculation). After all, changing from a cation to a neutral species will likely affect the charge distribution in the molecule. In addition, unless newer versions of MacroModel provide utilities for the calculation of the dipole moment, it will be necessary to supplement the MacroModel results with some other modeling package. Single-point calculations on the AMBER*-optimized geometries, using the AM1 Hamiltonian in MOPAC93, would provide an appropriate starting point.

• Improved modeling of the transition structure for the acetyl exchange between acetylcarnitine and CoA by:
  1. Modeling by semiempirical (or ab initio) methods, including a solvent model.
  2. Determination of molecular mechanics parameters specific to the sulfur found in the molecule (from ab initio).
  3. Molecular mechanics modeling including longer chains in place of the acetyl group.
• Modeling of the morpholinium rings including a methylenecarboxylate, both in the gas phase and solution.
• Modeling of the phosphonate rings including a methylenecarboxylate, both in the gas phase and solution.