the modular format of the charges in charmm will allow you to directly
apply the hydroxyl charges (and adjacent methylene group) from
threonine to hydroxy proline. the majority of the internal parameters
should also be available. however, i strongly suggest that QM calculations
be performed on selected conformations of the ring and hydroxyl in, say,
the HYP dipeptide to validate the accuracy of the dihedral parameters.
i suspect you will need to do some optimization of those parameters
to get good relative conformational energies.