Synthesis and In Vitro Evaluation of Novel Nortropane Derivatives as Potential Radiotracers for Muscarinic M2 Receptors

Disturbances of the cerebral cholinergic neurotransmitter system are present in neurodegenerative disorders. SPECT or PET imaging, using radiotracers that selectively target muscarinic receptor subtypes, may be of value for in vivo evaluation of such conditions. 6β-acetoxynortropane, a potent muscarinic M2 receptor agonist, has previously demonstrated nanomolar affinity and high selectivity for this receptor. Based on this compound we synthesized four nortropane derivatives that are potentially suitable for SPECT imaging of the M2 receptor. 6β-acetoxynortropane and the novel derivatives were tested in vitro for affinity to the muscarinic M1−3 receptors. The original 6β-acetoxynortropane displayed high affinity (K i = 70–90 nM) to M2 receptors and showed good selectivity ratios to the M1 (65-fold ratio) and the M3 (70-fold ratio) receptors. All new derivatives showed reduced affinity to the M2 subtype and loss of subtype selectivity. It is therefore concluded that the newly synthesized derivatives are not suitable for human SPECT imaging of M2 receptors.


Introduction
Central cholinergic disturbances are present in many neuropsychiatric and neurodegenerative diseases.In various forms of dementia, such as Alzheimer's dementia (AD) or Lewy body dementia, cholinergic deficits in the brain [1,2] are associated with cognitive decline [3][4][5] and are thought to precede clinical symptoms.
The majority of the cholinergic deficits in these diseases arise from degenerative events in basal forebrain regions such as the nucleus basalis of Meynert [6,7], which provides the cholinergic input of the cerebral cortex.In degenerative diseases such as AD, disruption of basal cholinergic forebrain projections leads to a presynaptic cholinergic defect in cortical brain areas [1,8].Being part of a family of five muscarinic receptor subtypes, the muscarinic M 2 receptor is located predominantly presynaptically [9] and is consequently a potential target for the evaluation of the integrity of the cholinergic neurotransmitter system by molecular imaging.
In vivo assessment of the central cholinergic system in patients suffering from dementia by means of positron emission tomography (PET) or single photon emission computed tomography (SPECT) may be of value for early diagnosis or monitoring of such diseases, to predict response to cholinergic therapies (such as acetylcholinesterase inhibitors) or to evaluate effects of experimental drugs.Molecular imaging of the cholinergic system of the brain requires radiotracers that ideally selectively target specific neuroreceptors of this neurotransmitter system, such as the muscarinic M 2 receptor.
Iodine-123 has favorable properties for SPECT imaging of neuroreceptors [19][20][21][22][23].The abundant 159 keV γ photons of 123 I (83% abundancy) are suitable for high-resolution brain SPECT imaging using LEHR (Low Energy High Resolution) or fanbeam collimators.Furthermore, unlike 18 Flabeled radiopharmaceuticals (T 1/2 = 109.8min), the halflife of 13.2 hours permits transportation over long distances of centrally produced 123 I-labeled radiopharmaceuticals, for instance, to the majority of European nuclear medicine centers from a single manufacturing site.Moreover, suitable methods have been developed for convenient radiosynthesis of radiopharmaceuticals labeled with 123 I selectively at one appropriate position with high radiochemical yields, for instance, by oxidative iododestannylation [24].However, radioiodinated SPECT analogues of TZTP so far have demonstrated disappointing results in vivo [25].Radioiodinated Z-IQNP is another compound that has recently been evaluated for imaging of the muscarinic M 2 receptor [12,26], although the muscarinic subtype selectivity of this compound for the M 2 receptor subtype is limited.
Recently, 6β-acetoxynortropane, a tropane alkaloid, was reported to be a potent and highly selective agonist for muscarinic M 2 receptors [27], and radiolabeled derivatives of this compound may thus be of value for in vivo imaging of these receptors.Moreover, as compared with the mentioned TZTP analog these nortropane analogs have favorable physicochemical properties.Notably, the lipophilicity of iodinated nortropane compounds remains in the optimal range 1-3 of log D (pH 7.4).The log D (pH 7.4) of iodinated TZTP compounds, on the other hand, is too high (i.e., >4), likely resulting in a high degree of nonspecific binding in the brain.In order to obtain good in vivo stability, the iodine label should be bound to an sp2-carbon, where iodoallyland iodophenyl compounds are the most suitable candidates.Of these, iodophenyl compounds are synthetically the most accessible and more stable.
Based on 6β-acetoxynortropane, we synthesized four derivatives as potential radiotracers for use in SPECT imaging.The four synthesized nortropane derivatives were tested for affinity to cloned human muscarinic M 1 -M 3 receptor subtypes on membrane fractions of Chinese hamster ovary (CHO) cells by in vitro competitive binding assays.

Material and Methods
Two nortropane analogues with an iodine containing moiety on the 6β-position have been synthesized.The tropane skeleton was formed in a single-step multicomponent reaction in analogy to the classical Robinson tropinone synthesis [28,29], as displayed in Figure 1.The resulting tropinone was reduced under Wolff-Kishner conditions to give 6-hydroxy-N-benzylnortropane (3).Alkylation or acylation of the hydroxyl function of 3 resulted in 4a-c, which were debenzylated in two steps [29,30] to provide the previously described 6β-acetoxynortropane (5a), and its iodinated analogues, the 6β-4 -iodobenzyl ether (5b) or the 6β-4 -iodobenzoate ester (5c) of 6β-nortropinol, respectively.
Accordingly, a bromophenyl ring was introduced at the 3β-position.First, the 6-hydroxyfunction of tropinone (2) was protected as tert-butyldimethylsilyl ether.With a Grignard reaction 4-bromophenyl was introduced at the C3 of the tropane.According to the signal of the C6α-H in the proton NMR spectrum, only the isomer with the 4-bromophenyl in the equatorial position had been formed.Desilylation followed by acetylation yielded 3β-(4-bromophenyl)-6βacetoxy-N-benzyl nortropane (9).
The benzyl group was removed by hydrogenation, but simultaneously also the bromo substituent was removed to result in 3α-hydroxy-3β-phenyl-6β-acetoxy-nortropane (10b).In one occasion also the 3β-phenyl-6β-acetoxy-nortropane (10a) was isolated, presumably due to the presence of a small amount of acid.
Relative binding affinity and selectivity ratios of the various nortropane derivatives for the M 1 -M 3 were determined by competitive binding assays against [ 3 H]N-methylscopolamine ([ 3 H]NMS, Perkin Elmer, Waltham, USA; specific activity 78 Ci (2886 GBq)/mmol).Assays were performed on membrane suspensions from CHO cells expressing either the recombinant human muscarinic M 1 , M 2, or the M 3 receptor subtype (Perkin Elmer, Waltham, USA) [25].
In the competitive binding assays, incubation buffer contained 50 mM TRIS-HCl, 10 mM MgCl 2 , and 1 mM EDTA (pH 7.4 at 4 • C).The assays were incubated during 60 min (M 1 or M 2 receptor subtypes) or 120 min (M 3 receptor subtypes) at 27 • C. Nonspecific binding was determined using atropine as a competitor in a concentration of 1 μM.
In the first series of competitive binding assays, the 6β-4 -iodobenzyl ether (5b) and the 6β-4 -iodobenzoate ester (5c) of 6β-nortropinol, as well as the lead compound 6βacetoxynortropane (5a), were tested.Protein concentrations for undiluted receptor subtype suspensions were 1.2 mg/mL (M 1 ), 4.3 mg/mL (M 2 ), and 2.0 mg/mL (M 3 ).Aliquots (n = 3) of diluted membranes (factor 1 : 100) containing the M 1 -M 3 receptor subtypes were incubated in a total volume of 540 μL containing 500 μL diluted membranes, 20 μL [ 3 H]NMS, and 20 μL of the nortropanes in increasing concentrations.The [ 3 H]NMS was used in a final concentration of 0.2 nM for the M 1 and M 2 assays and 0.09 nM for the M 3 assays.The equilibrium dissociation constants in nM of [ 3 H]NMS for the three receptor subtypes, provided by the manufacturer, were 0.15 (M 1 ), 0.19 (M 2 ), and 0.08 (M 3 ).Final competitor concentrations ranged from 1.0 • 10 −10 M to 1.0 • 10 −4 M.After incubation, the reaction was rapidly terminated by vacuum filtration over GF/C glass fiber filters, presoaked in 0.3% polyethylenimine (Sigma-Aldrich, Munich, Germany), and washed 5 times with 1 mL of ice-cold buffer.Filters were placed in vials with 10 mL of scintillation fluid (Ultima Gold, Perkin Elmer, Waltham, USA) and counted in a liquid scintillation counter (Tri-Carb 2900 TR Liquid Scintillation Analyzer, Packard.Software version: 3100).
In the second series of competitive binding assays, derivative 10a and 10b and the lead compound 6β-acetoxynortropane (5a) were tested.In this series, the protein concentrations for undiluted receptor subtype suspensions were 0.6 mg/mL (M 1 ), 7.5 mg/mL (M 2 ), and 1.5 mg/mL (M 3 ).Aliquots (n = 4) of diluted membranes (factor 1 : 30) containing the M 1 , M 2 , or M 3 receptor subtype were incubated on a microplate in volumes of 190 μL containing 150 μL diluted membranes, 20 μL [ 3 H]NMS, and 20 μL of the nortropanes in increasing concentrations.In these assays, the [ 3 H]NMS was used in a final concentration of 0.13 nM for the M 1 and M 2 assays and 0.065 nM for the M 3 assays.The K d in nM of [ 3 H]NMS in these experiments were as stated above.Final competitor concentrations ranged from 1.0•10 −10 M to 1.0•10 −5 M.After incubation, the assays were filtrated over UniFilter 96 GF/C filter plates, presoaked in 0.3% polyethylenimine (Sigma-Aldrich, Munich, Germany), and washed 9 times with 200 μL of ice-cold buffer.30 μL of scintillation fluid (MicroScint, Perkin Elmer, Waltham, USA) was added, and the filter plates were counted in a liquid scintillation counter (TopCount 5.0 Liquid Scintillation Analyzer, Perkin Elmer, Waltham, USA).For each competitor, the inhibition constant (K i ) was calculated from the EC 50 for the muscarinic M 1 , M 2 , and M 3 subtypes with nonlinear regression curve fitting using Graphpad Prism (version 3.02), relative to the K d of [ 3 H]NMS as provided by the manufacturer.

Results
In Figure 2, the results of the competitive binding experiments are displayed.The affinity of 6β-acetoxynortropane, relative to [ 3 H]NMS, for the muscarinic M 2 receptor subtype proved to be high in both experiments.In the first experiment the K i of 6β-acetoxynortropane was determined as 88.1 ± 23.8 nM (average ± SD; n = 3) and in the second experiment as 71.6 ± 4.8 nM (average ± SD; n = 4).In our experiments, selectivity ratios of the compound for the M 2 over M 1 or M 3 receptor subtype proved to be approximately 65 and 70, respectively.
The 6β-4 -iodobenzyl ether of 6β-nortropinol (5b) performed substantially less than 6β-acetoxynortropane and displayed a K i of only 3.0 ± 0.7 μM, while selectivity for the M 2 receptor was lost.The selectivity ratios of this derivative for the M 2 over the M 1 and M 3 receptors of the compound were determined as 0.1 and 0.2, respectively.
The 6β-4 -iodobenzoate ester of 6β-nortropinol (5c) also performed less than 6β-acetoxynortropane, and a K i of 6.8 ± 1.5 μM was estimated for the M 2 receptor, while selectivity ratios over the M 1 and M 3 receptor, were determined as 0.6 and 2.0, respectively.
The second series of experiments (Figure 2), using 3β-phenyl-6β-acetoxynortropane (10a) and 3α-hydroxy-3βphenyl-6β-acetoxynortropane (10b) as competitors, likewise showed weak affinity for muscarinic receptors, and small competitive effects to the binding of [ 3 H]NMS were only detected at the highest concentration of the tested range.The affinity for the muscarinic receptors could therefore not be assessed for these two derivatives.

Discussion
In the present study, we have synthesized derivatives that are based on 6β-acetoxynortropane, a tropane alkaloid described by Pei and coworkers, which was shown to be a muscarinic agonist with high affinity to muscarinic M 2 receptor subtypes, but lower affinity to other muscarinic receptor subtypes [27].Due to the apparent selectivity of 6β-acetoxynortropane for the M 2 receptor, the compound may be of interest for use as a muscarinic receptor radiotracer.
Two analogues of the tracer were synthesized in which the acetyl ester moiety on the 6β-position was replaced by either 4 -iodobenzyl ether (5b) or a 4 -iodobenzoate ester (5c).The competitive binding assays demonstrated that the substitution on the 6β-position of the tropane skeleton had shifted the affinity from the nanomolar range to the micromolar range and that the selectivity of the alkaloid for the M 2 receptor subtype was lost.Therefore, two other analogues were synthesized retaining the 6βacetoxy function, with substitution of a phenyl moiety on the 3β-position of the tropane skeleton: 3β-phenyl-6β-acetoxynortropane (10a) and 3α-hydroxy-3β-phenyl-6βacetoxynortropane (10b).Unfortunately, these derivatives demonstrated even less favorable affinity for the three tested muscarinic receptor subtypes.
The challenge of the present study was to create a derivative of 6β-acetoxynortropane that is suitable for (radio)iodination, while preserving the affinity for the M 2 receptor, optimizing lipophilicity to allow optimal bloodbrain-barrier (BBB) penetration and to limit nonspecific uptake, maintaining the size of the molecule as small as possible, while not compromising metabolic stability.
In an earlier study, our group evaluated the potential M 2 receptor tracer E-iodopentenyl-thio-TZTP, which showed moderate selectivity for the muscarinic M 2 receptor over the M 1 and M 3 receptors in vitro [25], although selectivity for M 2 receptors was less than the original FP-TZTP [10,14].However, in vivo experiments using the TZTP derivative proved to be unsuccessful due to high lipophilicity of the tracer and very rapid metabolism of the parent compound [10,14].The 6β-acetoxynortropane derivatives that were synthesized and evaluated in the present study have several advantages over the earlier tested TZTP derivative(s).The lipophilicity of derivative 5b and 5c or iodinated analogues of 10a and 10b is less than that of the earlier synthesized TZTPs, being within the estimated log P (P = partition coefficient in octanol-buffer at pH 7.4) range between 1 and 2 (data not shown), which is considered to be optimal for penetration of the BBB.Incorporation of an ester function such as in the 6β-4 -iodobenzoate ester of 6β-nortropinol (5c), 3β-phenyl-6β-acetoxynortropane (10a) or an additional hydroxyl group in 3α-hydroxy-3β-phenyl-6β-acetoxynortropane (10b) contributes to the reduction in lipophilicity as compared to the earlier reported TZTPs, which should theoretically limit nonspecific uptake of these potential tracers in the brain.Another advantage over the earlier tested TZTP derivatives is the position of the iodine atom.Although the previous TZTP derivatives also contained a sp2 carbon-bound iodine, ) and the 6β-4 -iodobenzyl ether (5b, K i for M 2 3.0 μM) and 6β-4 -iodobenzoate ester (5c, K i for M 2 6.8 μM) of 6β-nortropinol, 3β-phenyl-6βacetoxynortropane (10a, K i for M 2 > 1 μM), and 3α-hydroxy-3β-phenyl-6β-acetoxynortropane(10b, K i for M 2 > 1 μM) to the binding of [ 3 H]NMS to the muscarinic receptor subtypes M 1 -M 3 .In the curves of (a, c, e) (derivatives 5a, 5b, 5c), data are expressed as means ± SEM from 3 samples in a range of 1.0 • 10 −10 M to 1.0 • 10 −4 M of competitor concentrations.In the curves of (b, d, f) (derivatives 5a, 10a, 10b), data are expressed as means ± SEM from 4 samples in a range of 1.0 • 10 −10 M to 1.0 • 10 −5 M of competitor concentrations.Atropine curves are also displayed as a reference.
this was substituted on the alkenyl side chain, whereas in the present 6β-4 -iodobenzyl ether (5b) and 6β-iodobenzoate ester of 6β-nortropinol (5c), the iodine is bound to an aromatic sp2 carbon atom, which favors the in vivo stability and prevents rapid deiodination.Although we did not test iodinated versions of 3β-phenyl-6β-acetoxynortropane (10a) and 3α-hydroxy-3β-phenyl-6β-acetoxynortropane (10b), the phenyl ring would also be the appropriate location for coupling of the iodine atom in these two compounds.Moreover, unlike the earlier described and apparently metabolically unstable TZTP derivatives [25], the 3β-phenyl-6β-acetoxynortropane (10a) and 3α-hydroxy-3β-phenyl-6β-International Journal of Molecular Imaging acetoxynortropane (10b) should be more metabolically stable due to the direct substitution of the phenyl ring to the tropane skeleton, which is known to have favorable effects on the in vivo stability of tropane-derived radiotracers such as [ 123 I]FP-CIT [31,32] or [ 123 I]β-CIT [33].
In our competitive binding experiments, the inhibition constant of the original compound 6β-acetoxynortropane (5a) was substantially higher than the K i that was described by Pei and coworkers [27].Reasons for this may include differences in the reference tracer, which was [ 3 H]NMS in the present experiments, whereas Pei et al. used [ 3 H]quinuclidinyl benzilate (QNB), as well as a difference between rat and human muscarinic receptors.Pei et al. demonstrated a K i of 2.6 nM for 6β-acetoxynortropane at the muscarinic M 2 receptor and very high selectivity ratios over either M 1 or M 3 receptors using the tritiated antagonist.In the same study, an even lower inhibition constant was reported by Pei et al. when using the muscarinic agonist [ 3 H]oxotremorine as a reference.In our study, using only [ 3 H]NMS but not [ 3 H]oxotremorine as a reference, we calculated K i values to the M 2 receptor of 71 and 88 nM, in two separate series of competitive binding experiments using different protocols, and lower selectivity ratios to M 1 and M 3 receptors of 65 and 70, respectively.Nevertheless, such selectivity ratios would be very adequate for imaging of muscarinic M 2 receptors in vivo.However, the iodinated analogues that were tested (5b and 5c) only showed weak affinity for all three tested muscarinic receptor subtypes, whereas the K i of either 3β-phenyl-6β-acetoxynortropane (10a) or 3αhydroxy-3β-phenyl-6β-acetoxynortropane (10b) could not be assessed, but proved to be above the micromolar range.We tested the derivatives at a maximal concentration of 10 −5 M, which may be a limitation of the present study, but K i values in the micromolar range or higher were not considered of interest for our purposes.However, it cannot be excluded that iodination of derivative 10a and 10b would result in improved affinities for muscarinic receptors.Also, substitution of the phenyl group at the 3α-positition, which is known to have bulk tolerance in tropane-derived radiotracers (and muscarinic receptor antagonists such as atropine, NMS, and benztropine), may improve the in vitro binding characteristics.
In conclusion, we synthesized a series of analogues to 6β-acetoxynortropane, potentially being of interest for use as radiotracers for in vivo imaging of the muscarinic M 2 receptor subtype in neurodegenerative or neuropsychiatric diseases.However, changing the original molecule on the 6β-or the 3α/β-position by substitution of a iodophenyl or phenyl ring severely reduced both the affinity and selectivity of the nortropane for the muscarinic M 2 receptor subtype, and therefore, the synthesized analogues are not suitable for use in human SPECT imaging.