Imaging of Cerebral Amyloid Angiopathy with Bivalent 99mTc-Hydroxamamide Complexes

Cerebral amyloid angiopathy (CAA), characterized by the deposition of amyloid aggregates in the walls of cerebral vasculature, is a major factor in intracerebral hemorrhage and vascular cognitive impairment and is also associated closely with Alzheimer’s disease (AD). We previously reported 99mTc-hydroxamamide (99mTc-Ham) complexes with a bivalent amyloid ligand showing high binding affinity for β-amyloid peptide (Aβ(1–42)) aggregates present frequently in the form in AD. In this article, we applied them to CAA-specific imaging probes, and evaluated their utility for CAA-specific imaging. In vitro inhibition assay using Aβ(1–40) aggregates deposited mainly in CAA and a brain uptake study were performed for 99mTc-Ham complexes, and all 99mTc-Ham complexes with an amyloid ligand showed binding affinity for Aβ(1–40) aggregates and very low brain uptake. In vitro autoradiography of human CAA brain sections and ex vivo autoradiography of Tg2576 mice were carried out for bivalent 99mTc-Ham complexes ([99mTc]SB2A and [99mTc]BT2B), and they displayed excellent labeling of Aβ depositions in human CAA brain sections and high affinity and selectivity to CAA in transgenic mice. These results may offer new possibilities for the development of clinically useful CAA-specific imaging probes based on the 99mTc-Ham complex.

use of ICH as a surrogate marker for CAA. Accordingly, the development of a noninvasive technique to diagnose CAA-associated diseases specifically by the detection of amyloid using a probe is strongly needed.
Positron emission tomography (PET) and single photon emission computed tomography (SPECT) have generally been utilized as major in vivo imaging techniques to carry out the noninvasive diagnosis of amyloidoses. PET/SPECT can provide the information on localization of amyloid aggregates, while CT and MRI render the anatomical information. To date, many attempts to image Aβ aggregates constituting SP in AD brains using PET and SPECT tracers have been made. Several clinical studies using [ 11 C]PIB, a neutral thioflavin-T analogue, have proved this utility for AD diagnosis [16][17][18][19] . More recently, [ 18 F]florbetapir (Amyvid) 17,20,21 , [ 18 F] flutemetamol (Vizamyl) 16,22,23 , and [ 18 F]florbetaben (Neuraceq) 24,25 have been approved by the US Food and Drug Administration for clinical AD diagnosis.
Similarly to SP in AD brains, there are several reports regarding the detection of cerebrovascular amyloid depositions using [ 11 C]PIB [26][27][28] . However, since [ 11 C]PIB is designed to penetrate the blood-brain barrier (BBB), it is considered to bind to not only vascular amyloid aggregates but also parenchymal amyloid aggregates, indicating that it detects amyloid depositions in the whole brain; therefore, [ 11 C]PIB cannot help detecting SP as background signal in case of the diagnosis of CAA. Several efforts toward the development of imaging probes targeting Aβ deposition in CAA have been made. These probes, designed as fluorescent dye 29 , MRI contrast [30][31][32] , or PET/ SPECT imaging [31][32][33][34] agents, showed a potential use for CAA; however, in vivo specificity for Aβ aggregates in CAA was not demonstrated. Further research into the development of imaging probes for selective binding to Aβ deposited in the walls of the cerebral vasculature and to differentiate CAA from AD is desired.
To detect CAA but not SP, low brain uptake of an imaging probe targeting Aβ aggregates may be favorable 33,34 . We previously reported a series of 99m Tc-hydroxamamide ( 99m Tc-Ham) complexes with a multivalent amyloid ligand 35 , and utilized stilbene (SB) and benzothiazole (BT) as ligands for amyloid aggregates. These compounds are believed to hardly cross the BBB in vivo, and their high binding affinity for Aβ aggregates is feasible for imaging CAA. However, in that report, the binding affinity of 99m Tc-Ham complexes was evaluated using Aβ  aggregates present mainly in SP and less often CAA. It is generally accepted that compounds with high binding affinity for Aβ  aggregates except for antibodies can bind to other amyloid aggregates such as tau and α -synuclein [36][37][38] . Therefore, 99m Tc-Ham complexes are considered to bind to Aβ  aggregates, the predominant amyloid found in CAA, similarly to Aβ  aggregates.
In the present study, we evaluated the binding affinity for Aβ    (Fig. 1), and their utility for the in vivo specific detection of vascular amyloid aggregates but not parenchymal amyloid aggregates. Results Synthesis and 99m Tc labeling. The 99m Tc labeling reaction was performed by the complexation reaction using the Ham precursor, 99m Tc-pertechnetate, and tin (II) tartrate hydrate as a reducing agent 35 . The 99m Tc complexation reaction with Ham derivatives provided two isomers of 99m Tc-Ham complexes, as described in previous reports 35,39 . We defined the specific isomers with shorter retention times on reversed-phase high-performance liquid chromatography (RP-HPLC) as A-form ( Assessment of BBB permeability. To evaluate brain uptake of 99m Tc-Ham complexes, biodistribution experiments of 99m Tc-Ham complexes were performed in normal mice. We selected 18 F-florbetapir as a control and compared the results of 99m Tc-Ham complexes with that of 18 F-florbetapir (Fig. 2)  in brain sections from a CAA patient was evaluated by in vitro autoradiography. In CAA brain sections, [ 99m Tc] SB2A intensively labeled Aβ depositions (Fig. 3A), while almost no accumulation of radioactivity was observed in the control brain sections (Fig. 3D). Furthermore, the labeling pattern was consistent with the immunohistochemical staining pattern observed in the same brain sections with anti-Aβ (1-40) antibody (Fig. 3B). In addition, the labeling of Aβ depositions with [ 99m Tc]SB2A was blocked to a large extent with an excess of nonradioactive PIB (Fig. 3C).

Discussion
We previously reported 99m Tc-Ham complexes with a bivalent amyloid ligand showing high binding affinity for Aβ (1-42) aggregates 35 . In the present study, we evaluated their utility as CAA-specific imaging probes. Recently, several new 99m Tc-labeled CAA imaging agents were reported [42][43][44]    aggregates (Table 1). However, specific isomers of 99m Tc-Ham complexes with SB derivatives displayed similar binding affinity for Aβ  aggregates to those of the other isomers, while significant differences between IC 50 values with two isomers of SB derivatives were shown in the inhibition assay using Aβ (1-42) aggregates 35 . All 99m Tc-Ham complexes showed blocked binding to amyloid aggregates with a very high concentration (μ M order) of PIB, although the Aβ imaging probes reported previously have a binding affinity equal to or lower than that of PIB, suggesting that they have a much higher binding affinity than any other tracers targeting Aβ including CAA-specific imaging Blocking study with nonradioactive PIB was also performed using the adjacent brain section (C). In vitro autoradiogram of a brain section from a healthy control (male, 73 years old) labeled with [ 99m Tc]SB2A (D). Tc-Ham complexes also showed a lower initial brain entry than even other CAA imaging probes reported previously (0.61-1.21%ID/g at that time) 44 . These results suggest that 99m Tc-Ham complexes could hardly cross the BBB. Therefore, they may be incapable of binding to Aβ aggregates deposited within the brain parenchyma. According to the results of binding affinity for Aβ  and Aβ  aggregates in vitro and brain uptake in normal mice ex vivo, further studies were conducted using [ 99m Tc]SB2A and [ 99m Tc]BT2B with high binding affinity for Aβ aggregates and very low brain uptake.
In vitro autoradiography of human CAA brain sections with [ 99m Tc]SB2A showed intensive labeling of Aβ depositions (Fig. 3A), confirmed by immunostaining of the same brain sections with anti-Aβ (1-40) antibody (Fig. 3B). Many 99m Tc-labeled Aβ imaging probes with preferable binding affinity have exhibited no marked labeling of Aβ depositions in human brain sections; however, Jia et al. recently reported a 99m Tc-labeled tracer showing positive autoradiography results for brain sections from AD patients 44 . As well as those results, 99m Tc-Ham complexes showed excellent labeling of Aβ depositions in human brain sections. Additionally, a blocking study with nonradioactive PIB confirmed the specific binding of [ 99m Tc]SB2A to Aβ depositions in CAA brain sections (Fig. 3C). In vitro autoradiography with [ 99m Tc]BT2B showed specific binding to Aβ depositions in CAA brain sections as well as [ 99m Tc]SB2A (Fig. S1 in Supplementary information). In addition, two bivalent 99m Tc-Ham complexes, [ 99m Tc]SB2A and [ 99m Tc]BT2B, also showed intensive labeling of Aβ depositions in brain sections from another patient with CAA ( Fig. S2 in Supplementary information).
Ex vivo autoradiography with [ 99m Tc]SB2A displayed specific binding to Aβ aggregates in the living Tg2576 mouse brain (Fig. 4A,B) but not wild-type mouse brain (Fig. 4C). Since Tg2576 mice are known to overproduce Aβ aggregates in the brain, they have been commonly used to evaluate the specific binding of Aβ aggregates in experiments in vitro and in vivo 45,46 . The accumulation of radioactivity in Tg2576 mouse brain sections was observed only at the presence of both amyloid aggregates (Fig. 4D,E) and endothelial cells (Fig. 4G,H), suggesting that [ 99m Tc]SB2A selectively bound to amyloid aggregates deposited along vessels but not within parenchyma. In addition, [ 99m Tc]BT2B displayed specific detection of CAA in Tg2576 mice (Fig. S3 in Supplementary information). These results are consistent with the biodistribution study showing the low brain entry of 99m Tc-Ham complexes. These findings in the present study suggest that our bivalent 99m Tc-Ham complexes, [ 99m Tc]SB2A and [ 99m Tc]BT2B, can specifically detect CAA in vivo. However, these tracers seemed to label areas in the cortex that are not apparent in the thioflavin-S staining. Thioflavin-S has much lower affinity (K d : μ M order) than useful Aβ imaging probes reported previously such as PIB, florbetapir (K d : nM order) 19,47 . An in vitro inhibition assay showed that our 99m Tc-labeled compounds blocked binding to Aβ aggregates due to a much higher concentration of unlabeled-PIB, suggesting that our compounds have a much higher binding affinity than other Aβ and CAA imaging agents reported previously. Accordingly, it is considered that thioflavin-S can detect fewer depositions of amyloid than our compounds. In addition, we also carried out ex vivo autoradiography using perfused mouse brains, and obtained results showing differences with Tg2576 and wild-type mice (data not shown), suggesting that radioactivity was derived from tracers binding to depositions of amyloid, and not from tracers contained in the blood. Although the possibility that the BBB is leaky in the brains of AD patients has been suggested in several reports 48,49 , it has remained controversial whether or not the BBB dysfunction depends on the stage of AD. Not all studies have indentified an index of BBB disruption in AD brains 50-52 , but Zipser et al. recently reported that dysfunction of the BBB could increase stepwisely with the degree of pathology in AD 53 , indicating that the BBB should be intact in an early stage of preclinical AD. Moreover, CAA imaging probes should be used in the preclinical stage of the disease when the BBB functions normally. Therefore, 99m Tc-Ham complexes developed in the present study, which may be incapable of penetrating the BBB, can serve as CAA-specific imaging probes for the early diagnosis of AD.
In addition, we performed an in vivo SPECT/CT study with [ 99m Tc]SB2A using Tg2576 and wild-type mice at 30 min postinjection (Fig. S4 in Supplementary information). Although ex vivo autoradiography demonstrated the specificity of [ 99m Tc]SB2A for CAA, [ 99m Tc]SB2A was not differentially distributed in the brains of Tg2576 and wild-type mice in vivo. The radioactivity accumulation was observed mostly in the limbic region of the brain, which seemed to be derived from the blood. This inference is supported by the observations that blood vessels were rich in this region of the brain confirmed by immunostaining of CD31 (Fig. 4G,I), and a biodistribution study which suggested that [ 99m Tc]SB2A has a high retention rate in the blood (5.13%ID/g at 30 min postinjection, Table S1 in Supplementary information). Although the influence of radioactivity in the cerebral blood could be removed by perfusion in ex vivo autoradiography examination, it was inevitable to detect radioactivity in the blood as a background signal in the in vivo SPECT study. Furthermore, we carried out an in vivo SPECT imaging study at a later time point (120 min postinjection) of [ 99m Tc]SB2A. However, we obtained a similar result to that of at 30 min postinjection, suggesting that the radioactivity in the blood still remained at 120 min postinjection (Fig. S5 in Supplementary information). Therefore, further acceleration of the clearance of 99m Tc-labeled probes from the blood pool is essential for the development of in vivo imaging probes targeting CAA. The introduction of a hydrophilic substituted group including hydroxyl and carboxyl groups may constitute one of the strategies to enhance the clearance of probes from the blood. For instance, the replacement of the dimethylamino group in [ 99m Tc]SB2A with a hydroxyl group reduces its lipophilicity, contributing to lower binding to plasma proteins. This modification of probes should facilitate the more rapid clearance of [ 99m Tc]SB2A from the blood, leading to a lower background signal that can bring about an increase in the specific signal of the probes on binding to CAA.
In the current study, we applied bivalent 99m Tc-Ham complexes that we reported previously to imaging probes targeting CAA. All 99m Tc-Ham complexes including a monovalent or bivalent amyloid ligand showed binding affinity for Aβ  aggregates in vitro and very low brain uptake in normal mice ex vivo. In vitro autoradiography showed specific binding of 99m Tc-Ham complexes including a bivalent amyloid ligand ([ 99m Tc]SB2A and [ 99m Tc]BT2B) with high binding affinity in the inhibition assay to Aβ depositions in brain sections from a CAA patient. Additionally, [ 99m Tc]SB2A and [ 99m Tc]BT2B displayed excellent and selective labeling of Aβ depositions in vessels but not parenchyma in mouse brains. The results suggest that [ 99m Tc]SB2A and [ 99m Tc]BT2B have potential as CAA-specific imaging probes. Although the in vivo SPECT/CT study with [ 99m Tc]SB2A showed no marked difference in radioactivity accumulation in the brain between Tg2576 and wild-type mice, the findings in the present study reveal new possibilities of developing clinically useful CAA imaging probes based on the 99m Tc-Ham complex. Further optimization to improve the clearance of 99m Tc-Ham complexes from the blood is underway.

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). Female Tg2576 mice and wild-type mice were purchased from Taconic Farms, Inc. (New York, USA). Animals were fed standard chow and had free access to water. All efforts were made 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 and National Cerebral and Cardiovascular Center. Informed consent was secured from all subjects in this study. Postmortem brain tissues from autopsy-confirmed cases of CAA (female, 67 years old, and female, 85 years old) and a control (male, 73 years old) were obtained from the Graduate School of Medicine, Kyoto University, National Cerebral and Cardiovascular Center, and BioChain Institute, Inc. (California, USA), respectively. BT2) were prepared as we reported previously 35

Ex vivo biodistribution in normal mice.
A saline solution (100 μ L) of 99m Tc-Ham complexes (20 kBq) containing EtOH (10 μ L) was injected directly into the tail vein of ddY mice (male, 5 weeks old). The mice were sacrificed at 2, 10, 30, and 60 min postinjection. The organs of interest were removed and weighed, and radioactivity was measured using a γ counter (PerkinElmer). The %ID/g of samples was calculated by comparing the sample counts with the count of the diluted initial dose.
In vitro autoradiography of human CAA brain sections. Six micrometer thick serial human brain sections of paraffin-embedded blocks were used for autoradiography. To completely deparaffinize the sections, they were incubated in xylene for 30 min two times and in 100% EtOH for 1 min two times. Subsequently, they were subjected to 1-min incubation in 90% EtOH and 1-min incubation in 70% EtOH, followed by a 5-min wash in water. Each slide was incubated with a 50% EtOH solution of [ 99m Tc]SB2A or [ 99m Tc]BT2B (370 kBq/mL) at room temperature for 1 h. For blocking experiments, the adjacent sections were incubated with a 50% EtOH solution of [ 99m Tc]SB2A or [ 99m Tc]BT2B (370 kBq/mL) in the presence of nonradioactive PIB (1.0 mM). The sections were washed in 50% EtOH for 3 min two times and exposed to a BAS imaging plate (Fuji Film, Tokyo, Japan) for 2 h. Autoradiographic images were obtained using a BAS5000 scanner system (Fuji Film). After autoradiographic examination, the same sections were immunostained by an antibody against Aβ  to confirm the presence of Aβ depositions. For immunohistochemical staining of Aβ , the sections were autoclaved for 15 min in 0.01 M citric acid buffer (pH 6.0) to activate the antigen. After three 5-min incubations in PBS-Tween 20 (PBST), they were incubated with anti-Aβ (1-40) primary antibody (BA27; Wako, Osaka, Japan) at room temperature overnight. Subsequently, they were incubated in PBST for 5 min three times, and incubated with biotinylated goat anti-mouse IgG (Wako) at room temperature for 3 h. After three 5-min incubations in PBST, the sections were incubated with Streptavidin-Peroxidase complex at room temperature for 30 min. After three 5-min incubations in PBST, they were incubated with diaminobenzidine (Merck, Hesse, Germany) as a chromogen for 5 min. After washing with water, the sections were observed under a microscope (BIOREVO BZ-9000; Keyence Corp., Osaka, Japan).
Ex vivo autoradiography using Tg2576 and wild-type mice. Tg2576 transgenic mice (female, 29 months old) and wild-type mice (female, 29 months old) were used as the AD model and age-matched control, respectively. A saline solution (150 μ L) of [ 99m Tc]SB2A or [ 99m Tc]BT2B (18.5 MBq) containing EtOH (30 μ L) was injected through the tail vein. The mice were sacrificed at 30 min postinjection. The brains were immediately removed, embedded in carboxymethylcellulose solution and then frozen in a dry ice/hexane bath. Sections of 30 μ m were cut and exposed to a BAS imaging plate (Fuji Film) overnight. Autoradiographic images were obtained using a BAS5000 scanner system (Fuji Film). After autoradiographic examination, the same sections were stained by thioflavin-S to confirm the presence of Aβ depositions. For thioflavin-S fluorescent staining, the sections were immersed in a 100 μ M thioflavin-S solution containing 50% EtOH for 3 min, washed in 50% EtOH for 1 min two times, and examined using a microscope (Keyence Corp.) equipped with a GFP-BP filter set. Additionally, the same sections were immunostained by anti-CD31 antibody to confirm the presence of endothelial cells. For immunohistochemical staining of CD31, the sections were incubated in PBST for 5 min three times, and incubated with anti-CD31 primary antibody (SZ31; Abcam, Cambridgeshire, U.K., dilution 1:50) at room temperature overnight. After three 5-min incubations in PBST, anti-rabbit secondary antibody (Dako, California, USA) incubation was carried out at room temperature for 3 h. Subsequently, the sections were incubated in PBST for 5 min three times, and incubated with diaminobenzidine (Merck) as a chromogen for 5 min. After washing with water, the sections were observed under a microscope (Keyence Corp.).