1.6 Radiotracers for β-amyloid
1 Ligand Development
1.6 Radiotracers for β-amyloid
The degree of morbidity and mortality inflicted on individuals and society by Alzheimer’s disease (AD) and other dementias is well known, as are the hugely detrimental economic consequences in the absence of effective therapies for arresting or reversing disease progression. The β-amyloid cascade hypothesis75 has been central to the development of treatment strategies over recent years. Essentially, under this hypothesis over-production of the peptides Aβ40 and, especially Aβ42, from amyloid precursor protein with eventual deposition of amyloid (Aβ) plaques is the mechanism of pathogenesis. The aim of developing PET radiotracers useful for determining Aβ burden in living human subjects has received intense attention from several groups.76 Radiotracers are sought for imaging Aβ preceding clinical signs of dementia, although we have argued that such radiotracers will eventually have more utility for investigating pathophysiology and the validity of the β-amyloid cascade hypothesis.77 We began our work to discover such radiotracers preceding disclosure by others of promising radiotracers derived from histological dyes (e.g. thioflavin-T) for Aβ plaques. Most prominent is [11C]PIB (9), which in AD patients gives a moderate ratio (~ 2) for radioactivity in Aβ plaque-rich cortex to that in plaque-poor cerebellum.78 Despite such global progress, more sensitive PET radiotracers are still needed and we are working towards this goal. Longer-lived 18F-labeled tracers are also sought, since these might be produced in very high radioactivities for widespread distribution to clinical imaging centers.
For establishing SAR on ligands to be synthesized, it was necessary to establish an in vitro binding assay. Initially, we followed a literature protocol79 using synthetic Aβ aggregates as protein source and [125I]TZDM as reference radioligand. However, we observed that only about 15% of total binding was specific. We also found only a low concentration of binding sites. We changed the reference radioligand to [3H]PIB. Greater than 95% of radioactivity could be displaced with non-radioactive PIB.80 For this radioligand, the concentration of binding sites to Aβ42 monomer was again low. We derived a simple mathematical model for the calculation of Bmax based on the displacement curve ([B] = Bmax[F*]/(Kd + [F]), where F* is the concentration of reference radioligand and F the combined concentration of radioactive and non-radioactive ligand), without need to vary the concentration of added [3H]PIB. The equation was verified experimentally.81 We observed that Ki values for PIB are lower in AD brain homogenates than in synthetic Aβ aggregates. We modified an existing procedure82,83 to isolate intact Aβ plaques from human AD brain homogenates free of other proteins. Ki values for PIB and test compounds were similar for homogenates and isolated plaques.
We modified an ELISA assay to measure Aβ42 and Aβ40, which in turn allowed us to measure the number of binding sites per Aβ42 monomer. The number for synthetic Aβ aggregates is 10-4-10-3, while that for human AD brain homogenates or isolated Aβ plaques is between 1 and 10. These results and Ki values point to some of the major limitations of synthetic Aβ aggregates as a platform for predicting the binding characteristics of test ligands for human Aβ plaques.
Based on a computer modeling analysis of already described high affinity Aβ ligands, we initiated the synthesis of three types of compound, namely derivatives of IMPY (52)84, strapped IMPY (53) and a strapped pyrrole (54). We modified and developed three types of reactions to access over 60 such compounds, as follows.
For IMPY derivatives, we developed a microwave-promoted method for the rapid, selective and efficient substitution of bromo or iodo substituents in aryl halides by thiolates.85 This procedure is applicable to substrates with an easily reducible iodo group, in either the presence or absence of a bromo group. For strapped IMPYs, we developed a new method for the synthesis of imidazo[1,2-a]pyridines from the reaction of 2-aminopyridines with α-halo ketones.86 The critical reagent is titanium(IV) chloride, which appears to serve as a strong dehydrating agent to promote formation of a putative Schiff base intermediate that may subsequently cyclize to the desired product. The reactions were performed rapidly under microwave conditions. Various substitution patterns in both the α-halo ketone and 2-aminopyridine substrates were examined to evaluate the scope of the reaction. For strapped pyrroles, the classical Fisher indole synthesis was adapted for the synthesis of tetracyclic pyrroles, also under microwave conditions.
We initially prepared and evaluated two tertiary amine [18F]N-fluoroalkyl derivatives of IMPY, namely [18F]FEM-IMPY (Ki, 27 nM; cLogP, 4.48; 55) and [18F]FPM-IMPY (Ki, 40 nM; cLogP, 4.42 ; 56), as candidate Aβ radioligands.87 After administration of either radioligand to rodent or monkey there is a rapid and adequate uptake of radioactivity into brain followed by biphasic clearance. Metabolism is rapid via dealkylation of the tertiary aromatic amino group, culminating in rapid defluorination and high uptake of radioactivity in bone. With the goal of avoiding defluorination and residual brain radioactivity, reducing lipophilicity and increasing binding affinity, we decided to make use of ‘isosteric’ and ‘isoelectronic’ effects in the design of further analogs of IMPY.88 One set of analogs consist of secondary amines in which a methyl group is ‘shifted’ from the tertiary amine nitrogen to the nearest ortho ring position.89 The first new radiotracer of this type, [11C]MeBrMPY (Kd, 4.1 nM; cLogP, 3.66; 57), was examined with PET in mouse, rat and monkey and showed acceptable brain entry and wash-out of radioactivity.90 The second set has a thiol ether instead of iodine in the 6-position. The third set combines both features. SAR of these compounds shows that the PIB binding site is small and tolerates structural change in only the 6- position of IMPY derivatives.91
Tg mice (Tg2576 and CNRD8) and Tg rat became available to us for radiotracer evaluation with PET. With [11C]PIB in Tg2576 mice younger than 24 mo we observed that uptake of radioactivity in cortex was no higher than in cerebellum.92 However, we obtained a ratio of 1.8 in a Tg 2576 mouse greater than 28 mo old. This exceeds the highest such signal (1.32), which had been reported for another type of tracer, [18F]FDDNP, in Tg rat.93