As a final point on selegiline, it has long been thought that its efficacy in Parkinson's is due to the inhibition of the metabolism of dopamine.  Which is true, but there may be more to it than that.

A summary of this fascinating article: 

Apoptosis  is different from necrosis in a fundamental way: it is signaled, rather than directly caused.  In necrosis, the cell rapidly dies, the plasma membrane ruptures (with resultant irreversible ion shifts),  but DNA stays intact.  In apoptosis, the plasma membrane stays intact, but the cell shrinks, chromatin condenses and the DNA fragments.

Signaling is important: the genes p53, bad and bax induce apoptosis, while the Bcl family of genes promotes survival.  It is now thought that apoptosis mediates substantia nigra neuronal death.

So anything that delays or stops apoptosis could be neuroprotective.

Selegiline seems to be such a drug.  Since it inhibits the metabolism of dopamine, it will also prevent the formation of free radicals associated with this metabolism.  But, through a mechanism totally independent of MAO-B inhibition, it  protects dopamine neurons from MPTP and its metabolite MPP+.  (This is a double effect: MAO-B inhibition prevents the metabolism of MPTP to the toxic MPP+; and then selegiline's other unexplained mechanism protects neurons from MPP+.) Its metabolite desmethylselegiline is actually a more potent neuroprotector; and  P450 inhibitors which block the metabolism of selegiline to desmethylselegiline, also inhibit the overall neuroprotection.

The anti-apoptotic mechanism of selegiline (and the even more powerful irreversible MAO-B inhibitor rasagiline) is via glyceraldehyde-3-phosphate dehydrogenase (GAPDH).  GAPDH is usually in a dimer with a stem-loop of RNA in the cytoplasm.  In mitochondrial oxidation, NAD+ levels rise and then knock off the GAPDH, which then floats to the nucleus.  There, GAPDH inhibits the formation of of anti-apoptotic molecules, and thus causes apoptosis.  Propargylamines insert themselves into the RNA dimer and obstruct GAPDH from dislocating-- thus it cannot go to the nucleus and cause apoptosis.

Additionally, rasagiline upregulates Bcl-2 and Bcl-xl, among other anti-apoptotic molecules.

The study goes on to describe some clincial trials.  Indeed, the entire May 2006 supplement in Neurology is about neuroprotection in Parkinson's-- definitely worth the read.  We'll have more on this topic after some research.