Highly Selective Tau-SPECT Imaging Probes for Detection of Neurofibrillary Tangles in Alzheimer’s Disease

Neurofibrillary tangles composed of aggregates of hyperphosphorylated tau proteins are one of the neuropathological hallmarks of Alzheimer’s disease (AD) in addition to the deposition of β-amyloid plaques. Since the deposition of tau aggregates is closely associated with the severity of AD, the in vivo detection of tau aggregates may be useful as a biomarker for the diagnosis and progression of AD. In this study, we designed and synthesized a new series of radioiodinated benzoimidazopyridine (BIP) derivatives, and evaluated their utility as single photon emission computed tomography (SPECT) imaging agents targeting tau aggregates in AD brains. Five radioiodinated BIP derivatives were successfully prepared in high radiochemical yields and purities. In in vitro autoradiographic studies using postmortem AD brains, all BIP derivatives displayed high accumulation of radioactivity in the brain sections with abundant neurofibrillary tangles, while no marked radioactivity accumulation was observed in the brain sections with only β-amyloid aggregates, indicating that the BIP derivatives exhibited selective binding to tau aggregates. Biodistribution studies in normal mice showed high brain uptake at 2 min postinjection (3.5–4.7% ID/g) and rapid clearance at 60 min postinjection (0.04–0.23% ID/g), which is highly desirable for tau imaging agents. The results of the present study suggest that [123I]BIP derivatives may be useful SPECT agents for the in vivo imaging of tau aggregates in AD.

and 10). After the reaction for the formation of the BIP scaffold, the tributyltin derivatives (11, 12, 13, 14, and 15) were prepared from the corresponding bromo compounds using a conventional bromo to tributyltin exchange reaction catalyzed by Pd(0). The tributyltin derivatives were reacted with I 2 in chloroform at room temperature to yield the iodo compounds (16, 17, 18, 19, and 20). The tributyltin derivatives were also used as a precursor for labeling with 125 I. In the present study, we used 125 I as radioiodine instead of 123 I because of its availability and longer half-life. Radioiodination of BIP derivatives was performed by the iododestannyaltion reaction from the corresponding tributyltin derivatives using hydrogen peroxide as the oxidant (Fig. 1C). Then, we obtained radioiodinated BIP derivatives in 30-65% radiochemical yields and over 99% radiochemical purities after HPLC purification.
In vitro autoradiography using AD brain sections. In recent papers regarding the development of tau imaging probes, several methods to evaluate the selective binding affinity for tau aggregates have been reported 17,19,21 . Among them, we selected an in vitro autoradiographic (ARG) study using brain sections from AD patients, because an in vitro ARG study enables more direct evaluation than an in vitro binding assay using aggregates of recombinant tau proteins. When we performed immunohistochemical staining of two kinds of AD brain sections (frontal and temporal lobes) with anti-Aβ antibody and anti-phosphorylated tau antibody, we could observe the extensive accumulation of Aβ plaques in the gray matter of the frontal lobe, but not tau aggregates ( Figure S1A and B). In contrast, the gray matter of the temporal lobe showed marked accumulation of both Aβ plaques and tau aggregates in a different pattern of immunohistochemical staining ( Figure S1C and D). Therefore, in the case that we use these two kinds of AD brain sections, ideal tau imaging probes should specifically accumulate only in the temporal lobe positive for both tau and Aβ and not accumulate in the frontal lobe negative for tau. Also, in vitro autoradiography of each probe should correspond to the immunohistochemical staining of tau. On the basis of these criteria, we carried out in vitro autoradiography from the perspective of these evaluations. Figure 2 shows the results of the in vitro autoradiography of [ 125 I]BIP derivatives and [ 125 I]IMPY. When we performed optimal adjustment of the autoradiograms for each probe, all BIP derivatives showed a similar pattern on in vitro autoradiography, and the marked radioactivity accumulation was specifically observed in the gray matter of the temporal lobe. In contrast, no marked radioactivity accumulation was found in the gray matter of the frontal lobe, which does not contain a tau pathology. As we mentioned above, the results of immunohistochemical staining suggested that there are both Aβ plaques and neurofibrillary tangles in the gray matter of the temporal lobe, while only Aβ plaques mainly deposit in the gray matter of the frontal lobe. This pattern was completely different from that of [ 125 I]IMPY, which is well known as a SPECT imaging probe for Aβ plaques 30,31 .
[ 125 I]IMPY showed marked binding to Aβ plaques in the gray matter of both the frontal and temporal lobes. Figure S2 shows comparative enlarged autoradiograms of [ 125 I]IMPY and [ 125 I]BIP-NMe 2 . The radioactivity of the autoradiogram of [ 125 I]IMPY was seen as numerous spots reflecting the binding to Aβ plaques, while that of [ 125 I]BIP-NMe 2 was observed as the laminar accumulation characteristic of neurofibrillary tangles. These results strongly suggest that all BIP derivatives may bind to tau aggregates selectively. Figure 3 shows a representative enlarged image of an in vitro ARG study using BIP-NMe 2 together with that of immunohistochemical staining with anti-phosphorylated tau antibody. BIP-NMe 2 displayed laminar radioactivity accumulation along the gray matter characteristic of tau accumulation, suggesting that BIP derivatives could clearly detect the tau pathology.  Furthermore, to determine whether the radioactivity accumulation of [ 125 I]BIP-NMe 2 in the gray matter of the temporal lobe is specific to tau aggregates, we carried out in vitro autoradiographic studies in the presence of excess nonradioactive BIP-NMe 2 at 1 and 10 μ M (Fig. 4). As a result, the high radioactivity accumulation of [ 125 I] BIP-NMe 2 in the brain section was markedly reduced by adding excess nonradioactive BIP-NMe 2 , indicating that [ 125 I]BIP-NMe 2 can specifically bind to tau aggregates in the brain (Fig. 4).
Quantitative analysis of autoradiography (ARG). Next, we conducted quantitative analyses of in vitro autoradiographic studies as described above. The regions of interest (ROIs) were set in the gray matter of the frontal lobe, white matter of the frontal lobe, gray matter of the temporal lobe, and white matter of the temporal lobe, respectively, in the ARG images (n = 3-7), and the radioactivity accumulation (cpm/mm 2 ) in each region was determined for each probe (Fig. 5). As a result, all BIP derivatives showed higher radioactivity accumulation in the gray matter of the temporal lobe, in which abundant tau aggregates exist, than any other regions. This tendency was marked for BIP-Ph, BIP-Me, BIP-OMe, and BIP-NMe 2 other than BIP-H. When we performed statistical analyses to compare the radioactivity accumulation in the gray matter between the frontal and temporal lobes, BIP-Ph, BIP-Me, BIP-OMe, and BIP-NMe 2 showed significantly higher radioactivity levels than BIP-H. This also suggests that the substituted group in the BIP scaffold may play an important role in the selective binding affinity to tau aggregates. To further evaluate the selective binding between tau and Aβ , we compared the radioactivity accumulation observed in the gray matter of the frontal lobe with only Aβ aggregates with that observed in the gray matter of the temporal lobe with both Aβ and tau aggregates. When we calculated the ratio of the radioactivity accumulation in the gray matter of the temporal lobe against the gray matter of the frontal lobe, the values varied with the kind of the substituted group introduced into the BIP scaffold (Table 1). All BIP derivatives showed higher radioactivity accumulation in the gray matter of the temporal lobe than that in the gray matter of the frontal lobe. In particular, the BIP derivatives (BIP-Me, BIP-OMe, and BIP-NMe 2 ) with a compact substituted group showed a higher ratio than BIP-H which has no substituted group. Although BIP-Ph showed marked accumulation in the tau pathology, the selective binding of BIP-Ph in the τ pathology (3.3) was lower than that of BIP-Me (23.9), BIP-OMe (12.2), and BIP-NMe 2 (12.9), because BIP-Ph also showed relatively high-level binding to Aβ plaques. Although IMPY also showed a ratio of 1.8, this value may reflect the ratio of Aβ aggregates deposited in the frontal and temporal lobes, and not tau aggregates. These findings in the in vitro autoradiographic studies revealed that the selective binding of BIP derivatives to tau aggregates can be markedly increased by introducing the compact substituted group into the BIP scaffold.
Brain uptake and clearance. Next, we evaluated uptake into and clearance from the brain after the injection of BIP derivatives into normal mice ( Fig. 6 and Table 2). In order to obtain highly specific signals of tau aggregates, favorable tau imaging probes need to show not only high brain uptake early after administration but Figure 5. Quantitative analysis of in vitro autoradiography with AD brain sections. Data are presented as mean ± SEM (n = 3-7). Statistical significance was analyzed between temporal lobe and frontal groups in gray matter using unpaired t-test, with Welch's correction as necessary (*p < 0.05, **p < 0.01, ***p < 0.001).

Table 1. Ratio of radioactivity accumulation in the gray matter of the temporal lobe (a) against the gray matter of the frontal lobe (b).
also rapid clearance from the brain because no tau aggregates exist in the brains of normal mice. We determined the log P-values of the 125 I-labeled BIP derivatives, and the values ranged from 2.64 to 3.37. A previous report suggested that the low-molecular-weight compounds, which have moderate log P-values ranging from 1 to 3, sufficiently penetrate the blood-brain barrier 32 . Since the BIP derivatives showed moderate lipophilicity close to the appropriate log P-values reported in the previous paper, they were expected to penetrate the blood-brain barrier. All BIP derivatives displayed high brain uptake, ranging from 3.5 to 4.7% ID/g, at 2 min postinjection. Thereafter, the radioactivity in the brain cleared with time, and it decreased to 0.12-0.65% ID/g at 30 min postinjection and 0.04-0.23% ID/g at 60 min postinjection. When we calculated the ratio of the radioactivity accumulation of the BIP derivatives, the values ranged from 5 to 40 for 2 min/30 min and from 15 to 104 for 2 min/60 min, indicating that the BIP derivatives showed rapid clearance from the brain. Initial brain uptake of the BIP derivatives at 2 min postinjection were well correlated with their molecular weight, and not their log P-values. In other words, as the molecular weight of the BIP derivatives decreased, they showed higher brain uptakes. In particular, BIP-H and BIP-Me with log P-values with over 3 showed faster clearance from the brain than other BIP derivatives with log P-values less than 3. In addition, the BIP derivatives with a higher molecular weight tended to be cleared from the brain faster in comparison with the BIP derivatives with a lower molecular weight. Previous studies reported that the maximum brain uptake of [ 11  have superior radioactivity pharmacokinetics in the brain. The findings suggest that the BIP derivatives may be applicable to clinical use from the viewpoint of radioactivity pharmacokinetics in the brain, in addition to highly selective binding to tau aggregates. The biodistribution of BIP derivatives in the whole mouse body displayed typical pharmacokinetics of lipophilic compounds, which means high uptake into the liver and the subsequent excretion into the intestine (Table S1). In addition, the low accumulation of BIP derivatives in the thyroid also demonstrated high stability against deiodination in vivo.
When we determined the stability of [ 125 I]BIP-NMe 2 in the blood by HPLC analyses, the HPLC profile after incubation for 60 min was similar to that of the intact form, suggesting that they showed high stability in the blood ( Figure S3). Furthermore, we analyzed the radiometabolites in the blood and brain after the injection of [ 125 I]BIP-NMe 2 into normal mice. When the radioactivity level was analyzed in plasma by HPLC after the injection of [ 125 I]BIP-NMe 2 into normal mice, we found that [ 125 I]BIP-NMe 2 converted to some radiometabolites with different chemical forms (Fig. 7A). The percent of the intact form decreased with time and reached 63.6 and 16.8% at 2 and 10 min after the injection, respectively. This result regarding in vivo metabolism in the blood was   BIP-NMe 2 in organs including the liver and kidney. When we analyzed the radioactivity in the brain by HPLC at 2 and 10 min postinjection of [ 125 I]BIP-NMe 2 , 84.7 and 62.0% of [ 125 I]BIP-NMe 2 at 2 and 10 min postinjection, respectively, existed in the brain as an intact form (Fig. 7B), indicating that [ 125 I]BIP-NMe 2 may be sufficiently applicable for the in vivo imaging of tau aggregates. It is generally accepted that the in vivo metabolism of radiotracers is different between mice and human subjects. Therefore, further investigation of the characteristics of radiometabolies derived from BIP derivatives in human subjects is essential for clinical application.

Conclusions
In the present study, we newly designed and synthesized radioiodinated BIP derivatives with various substituted groups, and evaluated their utility as tau-SPECT imaging probes. In in vitro autoradiographic studies using AD brain sections, the BIP derivatives depicted tau clearly. In addition, the in vitro autoradiographic studies also revealed highly selective binding of the BIP derivatives to the tau pathology compared with Aβ pathology. In the biodistribution study in normal mice, all BIP derivatives showed high uptake into and fast clearance from the brain comparable to the tau-PET imaging probes in clinical use. The results of the present study suggest that 123 I-labeled BIP derivatives with highly selective affinity for tau aggregates may be new candidates for in vivo detection of the tau pathology.

Materials and Methods
General remarks. All reagents were commercial products and used without further purification unless indicated otherwise. All compounds were purified by Smart Flash EPCLC W-Prep 2XY (Yamazen corporation) unless indicated otherwise. 1 H NMR spectra were recorded on a JNM-ECS400 (JEOL) with tetramethyl silane (TMS) as an internal standard. Coupling constants are reported in Hertz. Multiplicity was defined as singlet (s), doublet (d), or multiplet (m). ESI mass spectrometry was conducted with a SHIMADZU LCMS-2020. High-resolution mass spectrometry (HRMS) was carried out with a GCMate II (JEOL). HPLC was performed with a Shimadzu system (an LC-20AT pump with an SPD-20A UV detector, λ = 254 nm) using a Cosmosil C 18 column (Nacalai Tesque, COSMOSIL 5C 18 -AR-II 4.6 mm I.D. × 150 mm) and acetonitrile/H 2 O (with or without TFA) as the mobile phase at a flow rate of 1.0 mL/min. All key compounds were proven by this method to show > 95% purity.

Animals.
Animal experiments were conducted in accordance with our institutional guidelines and were approved by the Kyoto University Animal Care Committee. Male ddY mice were purchased from Japan SLC, Inc. (Shizuoka, Japan), and were fed standard chow and had free access to water. We made all efforts to minimize suffering. Human brain tissues. Experiments involving human subjects were performed in accordance with relevant guidelines and regulations and were approved by the ethics committee of Kyoto University. Informed consent was secured from all subjects in this study. Postmortem brain tissues from an autopsy-confirmed case of AD (male, 76 years old) were obtained from the Graduate School of Medicine, Kyoto University.
Chemistry. Compounds 1 and 5 were synthesized according to a method reported previously 27,28 . We used compounds 2, 3, and 4 of commercially available products.

7-Bromo-3-dimethylaminopyrido[1,2-a]benzimidazole (10). 2-Bromo-N,N-dimethylpyridin-
4-amine (5) was synthesized according to the method reported previously 28 . Using 5, the same reaction described above to prepare 6 was used, and 94.9 mg of 10 was obtained in a yield of 6.00%. 1 (14). The same reaction described above to prepare 11 was used, and 47.9 mg of 14 was obtained in a yield of 23.9%. 1  In vitro autoradiography. The presence and location of tau and Aβ deposits in the AD brain sections were confirmed with immunohistochemical staining using an anti-phosphorylated tau antibody (AT8) and Aβ 1-42 antibody (BC05), respectively (See the Supporting Information). [ 125 I]IMPY was synthesized according to the method reported previously 32 . Six-micrometer-thick serial sections of paraffin-embedded blocks were used for staining. The sections were subjected to two 15-min incubations in xylene, two 1-min incubations in 100% EtOH, one 1-min incubation in 90% EtOH, and one 1-min incubation in 70% EtOH to completely deparaffinize them, followed by two 2.5-min washes in water. The sections were incubated with radioiodinated ligands (370 kBq/mL in 10% or 50% EtOH) for 2 h at room temperature. They were then dipped in 50% EtOH for 1 h and washed with H 2 O for 1 min. After drying, the 125 I-labeled sections were exposed to a BAS imaging plate (Fuji Film) overnight. Autoradiographic images were obtained using a BAS5000 scanner system (Fuji Film). A two-tailed, unpaired t-test, with Welch's correction as necessary, was performed to determine the significance of the difference in radioactivity accumulation in the gray matter between the frontal and temporal lobes. Statistical analyses were performed with GraphPad Prism 5 (GraphPad Software, Inc.).

7-Tributylstannyl-3-methoxypyrido[1,2-a]benzimidazole
In vitro blocking study. We used the same postmortem brain tissues as employed for in vitro autoradiography. The sections were incubated with [ 125 I]BIP-NMe 2 (370 kBq/mL in 10% EtOH) for 2 h at room temperature in the presence or absence of nonradioactive BIP-NMe 2 (1 or 10 μ M). After incubation, we dipped the sections and exposed them to a BAS imaging plate with the same methods as in vitro autoradiographic studies. The autoradiographic images were obtained using a BAS5000 scanner system. Analysis of radiometabolites in blood and brain. We performed the HPLC analyses of radiometabolites in the blood and brain according to a previously reported method 26 . [ 125 I]BIP-NMe 2 (1.11-1.48 MBq in 150 μ L) was injected into the tail vein of ddY mice (n = 4 for each time point for the blood and brain). The mice were sacrificed at 2 and 10 min postinjection. Bood samples were obtained and centrifuged at 4,000 × g for 5 min at 4 °C. The plasma (200 μ L) was separated and transferred to a tube containing acetonitrile (200 μ L). The mixture was stirred in precipitate from the aqueous phase. The brain was removed from the mice at 2 and 10 min postinjection of [ 125 I]BIP-NMe 2 , and homogenized in test tubes containing 500 μ L of tris-buffered saline with a homogenizer (POLYTRON PT 10-35, KINEMATICA). Acetonitrile (500 μ L) was added to a tube containing homogenized brain tissue and centrifuged at 4,000 × g for 5 min at 4 °C. The supernatants of the plasma and brain homogenates were filtrated using a 0.45-μ m filter (Millipore). Then, the filtrate was analyzed using HPLC on a COSMOSIL 5C 18 -AR-II column with an isocratic solvent of acetonitrile/H 2 O/TFA (25/75/0.1) at a flow rate of 1.0 mL/min. The eluent was collected with a fraction collector (Frac-920, GE Healthcare) at 30-s intervals, and the radioactivity in each fraction (500 μ L) was measured with an automatic γ counter (Wallac WIZARD 1470).