Al18F-NODA Benzothiazole Derivatives as Imaging Agents for Cerebrovascular Amyloid in Cerebral Amyloid Angiopathy.

In this study, we synthesized four novel Al18/19F-labeled 2-phenylbenzothiazole derivatives conjugated to 1,4,7-triazacyclononane-1,4-diacetic acid via alkyl linkers and evaluated them as imaging agent targets to amyloid-β (Aβ) plaques deposited in the blood vessels of cerebral amyloid angiopathy (CAA) brain. The four ligands exhibited moderate-to-high binding ability to Aβ1-42 aggregates, of which complex 17 possessing the most potent affinity (K i = 11.3 nM) was selected for further biological evaluations. In vitro fluorescent staining and in vitro autoradiography studies on brain sections from CAA patients proved that this ligand could label Aβ deposits in blood vessels selectively. In biodistribution study, [18F]17 can hardly penetrate the blood-brain barrier (brain2 min = 0.3% ID/g) and displayed a rapid blood washout rate (blood2 min/blood60 min = 25.2), which is favorable as CAA imaging agents. In conclusion, this Al18F-labeled 2-phenylbenzothiazole complex was developed and proved to be a promising CAA positron emission tomography agent.

Introduction

The high prevalence of cerebral amyloid angiopathy (CAA), which is characterized by cerebrovascular deposition of amyloid-β (Aβ), is associated with not only cognitive impairment but a major cause of lobar intracerebral hemorrhage in the elderly.[1−3] Pathological studies suggest that the main difference between CAA and Alzheimer’s disease (AD) is the location of Aβ deposits, as AD is recognized as Aβ deposits mainly in the brain parenchyma, whereas in CAA, the Aβ deposits in the cerebrovascular walls.[4] Distinct biological overlaps occur between CAA and AD; about a quarter of individuals with moderate-to-severe CAA are present in AD brains.[5−7] In addition, CAA is also present in about 30% of nondemented elderly. Besides conservative biopsy or autopsy, the clinical diagnosis of CAA almost relied on advanced manifestations such as hemorrhage or microbleeding and is obviously late and inadequate.[8] Due to the common pathogenesis shared by these two diseases and impending clinical requirements, it is highly significant to develop noninvasive imaging method for detecting Aβ deposits in the cerebrovascular walls.

Over the previous decades, considerable strides have been made in targeting Aβ plaques in AD. A large amount of effective Aβ-specific radiotracers have been reported for early diagnosis of AD using positron emission tomography (PET) or single-photon emission computed tomography (SPECT). For instance, three 18F-labeled radiotracers (Amyvid, Neuraceq, and Vizamyl) have been approved by the U.S. Food and Drug Administration in clinic.[9−11] However, none of them are able to distinguish Aβ deposits in two different regions. Among a number of imaging agents, [11C]PIB was reported to have the ability to diagnose CAA[12−14] but it could also penetrate the blood–brain barrier (BBB), thus making it hard to spot Aβ deposits in parenchyma and blood vessels apart.[12,15] According to previous studies, CAA could be specifically diagnosed through reducing the brain uptake of imaging agents.[16]99mTc-labeled probe is a good choice due to its relative high molecular weight and multiform of oxidation of 99mTc states to form charged molecules. However, the low spatial resolution and inaccuracy of quantification for SPECT should not be ignored. By contrast, PET with higher spatial resolution and sensitivity could overcome these disadvantages and indeed, some 18F and 68Ga-labeled stilbene derivatives were reported as CAA imaging agents; they displayed lower brain uptake and specific binding to Aβ deposits on blood vessel walls.[17,18]

Recently, aluminum fluoride (Al–F) complex has got considerable attention in radiofluorination of biomolecules. Compared with conventional radiofluorination strategies, this new Al–18F labeling method has many advantages, such as a one-pot procedure, which could be accomplished in water solution within 15 min.[19,20] In general, a classical bifunctional penta-dentate 1,4,7-triazacyclononane-1,4-diacetic acid (NODA) chelator with three nitrogen atoms and two carboxylates is ideal for aluminum to form an octahedral complex and leave an exactly binding site for fluoride ion.[21] In addition, the Al–F coordinate bond is stronger than that with most other metals[22,23] and some Al–18F complexes are not only highly stable in vivo but also possess biocompatibility in biosystem.[24,25]

In this study, to detect Aβ plaques in cerebrovascular system selectively, four CAA-targeted ligands were designed and synthesized with 2-phenylbenzothiazole as Aβ homing group. NODA was conjugated by different lengths of carbon linker and finally formed stable Al–F complexes with relatively high molecular weights around 600 Da, which reduces its ability to penetrate the BBB. Reported in this paper is the synthesis and biological evaluation of these Al–18F based complexes as specific CAA imaging agents.

Results and Discussion

Chemistry

The synthetic strategies used for the preparation of Al–19F complexes 14–17 are shown in Scheme . tert-Butyl-protected NODA chelator 1 was prepared by 1,4,7-triazacyclononane and tert-butylbromoacetate in acetonitrile according to the methods described by Gai et al., with minor modifications.[26] Pure disubstituted compound 1 was obtained by a carefully pH-controlled extraction in 76.6% yield from the mono- and trisubstituted byproducts. The brominated intermediates 2–5 with different lengths of carbon chains (n = 3 and 5) and mono- or dimethylamino groups were prepared from 2-phenylbenzothiazole hydroxyl compounds according to our previously reported methods.[27] Compounds 6–9 were synthesized from protected NODA chelator and brominated compounds 2–5 in the presence of N,N-diisopropylethylamine (DIEA) as base (yield: 27.6–38.5%). Removal of the protecting groups using trifluoroacetic acid (TFA) led to the dicarboxylic acid precursors 10–13 in high yield (89.7–93.0%). The X-ray quality crystal of precursor 13 was obtained by slow evaporation from N,N-dimethylformamide-methanol mixed system at room temperature. The ORTEP plot (30%) structure of 13 and its crystallographic data are illustrated in Figure and Table . Compound 13 crystallized in the monoclinic space group P1211; the presence of carbon chains adjacent to a substituted nitrogen atom insured the binding of fluoride to aluminum with no other ring-forming interference. Finally, the Al–19F complex 14–17 was successfully obtained by complexation of 10–13 with AlF3·3H2O in ethanol and sodium acetate buffer (pH = 4.2) in yields of 29.6–43.9% and it was characterized by NMR and high-resolution mass spectra (HRMS) analysis.

Scheme 1

Reagents and Conditions: (a) DIEA, CH3CN, 70 °C; (b) TFA, Room Temperature; and (c) AlF3·3H2O, NaOAc, EtOH, 100 °C

Figure 1

Crystal structure of compound 13 with the thermal ellipsoids drawn at the 30% probability level.

Table 1

Summary of the X-ray Crystallographic Data for Compound 13

formula sum	C30H39N5O5S
formula wt (g/mol)	581.72
crystal system	monoclinic
space group	P1211
cell parameters	a = 15.1260(11) Å, b = 7.5829(4) Å, c = 26.1864(18) Å, β = 90.547(6)°
cell ratio
cell volume (Å3)	3003.42(30)
Z	4
calcd density (g/cm3)	1.28642
RAll	0.1951
Pearson code	mP352
formula type	NO5P5Q30R39
Wyckoff sequence	a176

Biological Evaluation

In Vitro Fluorescent Staining

To preliminarily confirm the specific binding of these Al–19F complexes to Aβ plaques, in vitro neuropathological fluorescent staining was implemented on brain sections from two CAA patients. As shown in Figures and S1, four ligands displayed varying degrees of affinity to Aβ deposits in the blood vessel walls. Ligands with N,N-dimethylamino group (15, 17) exhibited preferable staining compared with ligands with N-methylamino group (14, 16), and ligand 17 with longer length of carbon chain stood out (Figures and S1). In comparison, no apparent labeling was observed in the brain section of the normal human (Figure G). The Aβ deposits were further verified by DANIR-3b (a highly sensitive near-infrared probe for Aβ);[28] the fluorescence signals from two different channels merged perfectly on the same section (Figure C,F,I).

Figure 2

In vitro fluorescent staining of complex 17 on human brain sections. (A) CAA patient, 76 year old, male, temporal lobe, 10×. (D) CAA patient, 70 year old, male, temporal lobe, 4×. (G) Healthy control, 89 year old, male, temporal lobe, 10×. (B, E, and H) The same brain section was stained by DANIR-3b. (C, F, and I) Merged image of 17 and DANIR-3b staining. 4′,6-Diamidino-2-phenylindole (DAPI) filter was used for 17, and red fluorescent protein (RFP) filter was used for DANIR-3b.

In Vitro Binding Assay Using Aβ Aggregates

The binding affinity of four Al–19F complexes (14–17) for Aβ1–42 aggregates was assessed quantitatively using [125I]IMPY as a competing radio ligand by means of inhibition binding assay. As shown in Table , the Ki values of these Al–19F complexes ranged from 11.3 to 124.1 nM. The obtained data also showed that N,N-dimethylamino compounds (15, 17) tended to have higher affinity than that of their corresponding N-methylamino compounds (14, 16) similar to the trend observed from another Aβ probe reported previously.[29] In addition, similar to our previous results, with the length of carbon chain extended from n = 3 to 5, the affinity was significantly increased. Under the same assay conditions, IMPY with Ki values of 4.6 nM proved the experimental system is feasible. Among these complexes, 17 displayed highest binding affinity (Ki = 11.3 nM), which is comparable to that of IMPY. Thus, its dicarboxylic acid precursor 13 was selected for 18F-labeling and further evaluation.

Table 2

Inhibition Constants of Al–19F Complexes (14–17) for the Binding of [125I]IMPY to Aβ1–42 Aggregatesa

ligands	14	15	16	17	IMPY
n	3	3	5	5
R	H	CH3	H	CH3
Ki (nM)a	124.1 ± 23.9	25.2 ± 1.9	40.9 ± 1.8	11.3 ± 0.74	4.6 ± 1.6

Measured in triplicate with results given as the mean ± standard deviation (SD).

Radiochemistry

The whole labeling process could be accomplished in a one-pot reaction in aqueous solution, which is distinct from the conventional nucleophilic or electrophilic radiofluorination strategies. As shown in Scheme , [18F]17 was obtained by mixing precursor with (18F–Al)2+, which was prepared by adding 18F– saline solution to AlCl3 stock solution in sodium acetate buffer (pH = 4) at room temperature. In this procedure, the formation of metal complex depended largely on pH-control and pH value of 4 was the most suitable condition as reported.[21] The decay-corrected radiochemical yield (RCY) based on 18F at the end of synthesis (40 min) was 17.8%; after reverse-phase high-performance liquid chromatography (HPLC) purification, [18F]17 was obtained with radiochemical purity (RCP) higher than 98%. The identification of [18F]17 was further confirmed by coinjection HPLC analysis of retention time between radiofluorinated tracer and corresponding nonradioactive complex 17 (Figure ).

Scheme 2

18F-Labeling Method of [18F]17

Figure 3

HPLC coinjection profiles of [18F]17 and 17. Venusil MP C18 column (Bonna-Agela Technologies, 5 μm, 4.6 mm × 250 mm); CH3CN/H2O = 40:60, 1 mL/min.

Ex Vivo Biodistribution

In general, optimized lipophilicity (1.5–4.0) of probe is thought to be a major factor affecting the BBB penetrability.[30] The measured log D value of [18F]17 (0.74 ± 0.004) by shake-flask method is lower than the optimal coefficient, which means it will be hard to penetrate BBB. In addition, compared with [18F]GE-067 used as Aβ imaging probe in brain parenchyma, [18F]17 has more than doubled molecular weight; this further ensures the decreased brain uptake. Then, ex vivo biodistribution study of [18F]17 was performed in normal ICR mice to evaluate its BBB penetrability and pharmacokinetic properties. The detailed time-radioactivity data obtained after intravenous administration are displayed in Table . Consistent with our prediction, the initial brain uptake was significantly decreased (0.30 ± 0.04% ID/g at 2 min) compared with the FDA approved Aβ tracers (> 4% ID/g at 2 min). High uptake in liver (ranging from 37.71 to 77.17% ID/g) and gradual rising uptake in intestine (from 5.60 to 37.64% ID/g) during the whole study indicate that [18F]17 was mainly metabolized by the hepatobiliary system and excreted by intestines. Furthermore, continuingly low levels of bone uptake observed during the entire investigation (from 1.38 to 3.14% ID/g) indicated the coordination bond of Al–18F was biologically stable against in vivo defluorination.

Table 3

Ex Vivo Biodistribution in Normal Mice (ICR, 18–22 g, Male) after Intravenous Injection of [18F]17a

organ	2 min	10 min	30 min	60 min
blood	7.81 ± 0.85	3.58 ± 1.31	0.54 ± 0.12	0.31 ± 0.06
brain	0.30 ± 0.04	0.16 ± 0.05	0.12 ± 0.10	0.09 ± 0.05
heart	8.65 ± 1.34	4.12 ± 1.46	0.83 ± 0.14	0.62 ± 0.26
liver	37.71 ± 3.47	74.71 ± 5.23	77.17 ± 7.73	57.18 ± 3.70
spleen	8.38 ± 1.84	6.94 ± 1.83	1.76 ± 0.14	1.01 ± 0.40
lung	15.59 ± 3.18	9.21 ± 2.33	1.83 ± 0.48	1.12 ± 0.26
kidney	14.06 ± 1.82	8.06 ± 2.28	3.46 ± 0.47	3.83 ± 0.89
pancreas	4.96 ± 0.38	3.16 ± 0.66	1.20 ± 0.20	1.14 ± 0.70
muscle	3.28 ± 0.47	2.96 ± 0.69	0.80 ± 0.09	0.81 ± 0.76
bone	3.14 ± 0.57	1.97 ± 0.79	1.38 ± 0.45	1.44 ± 0.53
stomachb	1.00 ± 0.18	1.05 ± 0.26	2.81 ± 2.13	3.83 ± 1.39
intestineb	5.60 ± 1.41	8.59 ± 2.00	14.22 ± 4.03	37.64 ± 6.80

Expressed as % injected dose per gram. Each value represents the mean ± SD for 5 mice at each interval.

Expressed as % injected dose per organ.

In Vitro Autoradiography

To further demonstrate the specific binding ability of radiofluorinated version of complex, in vitro autoradiography study of [18F]17 was implemented on brain sections from CAA patient and healthy control. As shown in Figure , autoradiographic images suggested that [18F]17 possessed highly potent binding with Aβ concentrated at blood vessels of CAA patient; due to the low resolution of autoradiography, the Aβ occupied blood vessels were displayed as hot spots. In contrast, the distinct labeling on brain sections of healthy control was not observed (Figure S3). The same sections were further stained by DANIR-3b, and the radioactive spots corresponded well to the fluorescent staining result. These results further validated that [18F]17 has excellent binding ability for Aβ deposits in cerebral vessels.

Figure 4

In vitro autoradiography of [18F]17 on brain section from CAA patient (70 year old, male, temporal lobe). The presence and distribution of plaques were confirmed by fluorescence staining using DANIR-3b.

Conclusions

In summary, for the first time, four Al18/19F-labeled 2-phenylbenzothiazole complexes with NODA chelator were successfully designed, synthesized, and evaluated as PET tracers for the detection of Aβ deposits in CAA brain. The results of competitive binding assay suggested that complex 17 with dimethylamino group and longer carbon linker possessed highest binding affinity to Aβ aggregates. Radiofluorinated tracer [18F]17 was obtained by a one-pot reaction in aqueous solution with high RCY and RCP. In vitro fluorescent staining and autoradiography studies showed that [18F]17 could selectively label Aβ deposits in blood vessels of CAA brain sections. In biodistribution study, the decreased initial brain uptake and rapid blood washout rate further confirmed the selectivity of [18F]17 for Aβ deposits in blood vessels. All of these results demonstrated that [18F]17 could be a promising PET agent for diagnosis of CAA. However, the design of CAA specific probe is very complicated, since there are very few reports on this and the criteria was not fully established; thus, subsequent experiments will be complemented to verify its further application.

Experimental Section

General Remarks

All chemicals used in synthesis were commercial products without further purification. 18F– was provided by Chinese PLA General Hospital. 1H NMR and 13C NMR spectra were recorded in CDCl3, dimethyl sulfoxide (DMSO)-d6, or TFA-d solutions on a Bruker Avance III NMR spectrometer (400/100 MHz) at room temperature. Chemical shifts are recorded as δ values relative to the internal trimethylsilyl. Mass spectra (MS) were acquired from a Surveyor MSQ Plus (ESI) (Waltham, MA) instrument. High-resolution mass spectra (HRMS) were acquired by Thermo scientific Q-Exactive (ESI) mass spectrometer. Chemical reactions were monitored by TLC (Silica gel 60 F254 aluminum sheets, Merck), and compounds were visualized using a hand-held UV lamps with short wavelengths of 254 and 365 nm. Silica gel (45–75 μm) used for column chromatography purification was obtained from Qingdao Haiyang Chemical Co, Ltd. Radio-HPLC purification and analysis was performed on a Shimadzu system CL-20AVP, with a SPD-20A UV detector (λ = 254 nm) and a Bioscan flow count 3200 NaI/PMT γ-radiation scintillation detector. Radioactivity was measured by automatic γ-counter (WALLAC/Wizard 1470). Fluorescent images were observed by EVOS FL imaging system (Life Technologies) equipped with DAPI and RFP filter sets. Synthetic Aβ1–42 was purchased from Peptide Institute (Osaka, Japan). Postmortem brain tissues from autopsy-confirmed CAA patients (70 year old, male, temporal lobe; 76 year old, male, temporal lobe) and a healthy control subject (male, 89 year old, temporal lobe) were obtained from the Chinese Brain Bank Center. Normal ICR mice (male, 18–22 g) used for biodistribution study were purchased from Vital River Co. Ltd. All guidelines involving the use of mice were followed in accordance with the Animal Care Committee of Beijing Normal University.

General Procedures for the Synthesis of Compounds 6–9

A solution of compounds 1 (102.6 mg, 0.6 mmol), 2–5 (1.5 mmol), and DIEA (1 mL) in acetonitrile (40 mL) was stirred at 70 °C for 12 h. After removal of solvent, the residue was purified by silica gel chromatography (CH2Cl2/MeOH = 20:1, v/v) to afford 6–9 as yellow solid.

Di-tert-butyl 2,2′-(7-(3-((2-(4-(methylamino)phenyl)benzo[d]thiazol-6-yl)oxy)propyl)-1,4,7-triazonane-1,4-diyl)diacetate (6)

Yield: 36.9%. 1H NMR (400 MHz, CDCl3) δ 7.86 (d, J = 8.3 Hz, 2H), 7.84 (s, 1H), 7.29 (d, J = 2.4 Hz, 1H), 6.98 (dd, J = 8.9, 2.4 Hz, 1H), 6.68 (d, J = 7.9 Hz, 1H), 4.14 (t, J = 5.5 Hz, 2H), 3.86–3.83 (m, 2H), 3.56–3.49 (m, 4H), 3.48–3.36 (m, 4H), 3.20–3.04 (m, 4H), 2.91 (s, 3H), 2.91–2.79 (m, 2H), 2.41–2.37 (m, 2H), 1.44 (s, 18H). MS (ESI): m/z calcd for C35H51N5O5S 653.3611, found 654.3593 [M + H]+.

Di-tert-butyl 2,2′-(7-(3-((2-(4-(dimethylamino)phenyl)benzo[d]thiazol-6-yl)oxy)propyl)-1,4,7-triazonane-1,4-diyl)diacetate (7)

Yield: 38.5%. 1H NMR (400 MHz, CDCl3) δ 7.89 (d, J = 8.7 Hz, 2H), 7.84 (d, J = 8.9 Hz, 1H), 7.31 (d, J = 2.4 Hz, 1H), 6.98 (dd, J = 8.9, 2.4 Hz, 1H), 6.73 (d, J = 8.8 Hz, 2H), 4.16 (t, J = 5.5 Hz, 2H), 3.86–3.84 (m, 2H), 3.54–3.47 (m, 4H), 3.39–3.35 (m, 4H), 3.33–3.08 (m, 4H), 3.04 (s, 6H), 2.90–2.68 (m, 2H), 2.41–3.38 (m, 2H), 1.44 (s, 18H). MS (ESI): m/z calcd for C36H53N5O5S 667.3767, found 668.3998 [M + H]+.

Di-tert-butyl 2,2′-(7-(5-((2-(4-(methylamino)phenyl)benzo[d]thiazol-6-yl)oxy)pentyl)-1,4,7-triazonane-1,4-diyl)diacetate (8)

Yield: 29.4%. 1H NMR (400 MHz, CDCl3) δ 7.93 (d, J = 8.5 Hz, 2H), 7.89 (d, J = 8.3 Hz, 1H), 7.31 (d, J = 2.4 Hz, 1H), 7.00 (dd, J = 8.9, 2.4 Hz, 1H), 6.75 (d, J = 8.7 Hz, 2H), 4.04 (t, J = 6.0 Hz, 2H), 3.87–3.83 (m, 2H), 3.61–3.10 (m, 12H), 3.05 (s, 3H), 2.76–2.41 (m, 4H), 1.90–1.83 (m, 4H), 1.62–1.56 (m, 2H), 1.44 (s, 18H). MS (ESI): m/z calcd for C37H55N5O5S 681.3924, found 682.4121 [M + H]+.

Di-tert-butyl 2,2′-(7-(5-((2-(4-(dimethylamino)phenyl)benzo[d]thiazol-6-yl)oxy)pentyl)-1,4,7-triazonane-1,4-diyl)diacetate (9)

Yield: 27.6%. 1H NMR (400 MHz, CDCl3) δ 7.91 (d, J = 8.7 Hz, 2H), 7.86 (d, J = 8.8 Hz, 1H), 7.30 (d, J = 2.0 Hz, 1H), 7.00 (dd, J = 8.9, 2.0 Hz, 1H), 6.74 (d, J = 8.7 Hz, 2H), 4.04 (t, J = 6.0 Hz, 2H), 3.86–3.82 (m, 2H), 3.57–3.09 (m, 12H), 3.05 (s, 6H), 2.76–2.40 (m, 4H), 1.98–1.83 (m, 4H), 1.66–1.55 (m, 2H), 1.44 (s, 18H). MS (ESI): m/z calcd for C38H57N5O5S 695.4080, found 696.4042 [M + H]+.

General Procedures for the Synthesis of Compounds 10–13

A solution of compounds 6–9 (1 mmol) in TFA (3–5 mL) was stirred at room temperature for 5 h. After the removal of TFA by vacuum, the yellow oil was treated with diethyl ether and filtered to afford 10–13 as yellow powder.

2,2′-(7-(3-((2-(4-(Methylamino)phenyl)benzo[d]thiazol-6-yl)oxy)propyl)-1,4,7-triazonane-1,4-diyl)diacetic Acid (10)

Yield: 92.4%. 1H NMR (400 MHz, DMSO-d6) δ 7.79 (d, J = 8.9 Hz, 1H), 7.74 (d, J = 8.4 Hz, 2H), 7.62 (s, 1H), 7.04 (d, J = 8.7 Hz, 1H), 6.61 (d, J = 8.4 Hz, 2H), 4.11 (t, J = 6.0 Hz, 2H), 3.57–3.46 (m, 4H), 3.32–3.14 (m, 4H), 3.10–3.04 (m, 2H), 2.91–2.77 (m, 4H), 2.72 (s, 3H), 2.65–2.59 (m, 2H), 2.19 (s, 2H), 1.23 (t, J = 6.1 Hz, 2H). MS (ESI): m/z calcd for C27H35N5O5S 541.2359, found 542.2703 [M + H]+.

2,2′-(7-(3-((2-(4-(Dimethylamino)phenyl)benzo[d]thiazol-6-yl)oxy)propyl)-1,4,7-triazonane-1,4-diyl)diacetic Acid (11)

Yield: 89.7%. 1H NMR (400 MHz, DMSO-d6) δ 7.82 (d, J = 8.9 Hz, 1H), 7.79 (d, J = 8.4 Hz, 2H), 7.63 (s, 1H), 7.06 (d, J = 9.2 Hz, 1H), 6.79 (d, J = 8.4 Hz, 2H), 4.08 (t, J = 6.0 Hz, 2H), 3.52–3.49 (m, 4H), 3.35–3.32 (m, 4H), 3.17–3.04 (m, 2H), 2.99 (s, 6H), 2.91–2.77 (m, 4H), 2.65–2.53 (m, 2H), 2.19 (s, 2H), 1.23 (t, J = 6.1 Hz, 2H). MS (ESI): m/z calcd for C28H37N5O5S 555.2515, found 556.2876 [M + H]+.

2,2′-(7-(5-((2-(4-(Methylamino)phenyl)benzo[d]thiazol-6-yl)oxy)pentyl)-1,4,7-triazonane-1,4-diyl)diacetic Acid (12)

Yield: 93.0%. 1H NMR (400 MHz, DMSO-d6) δ 7.76 (d, J = 8.9 Hz, 1H), 7.74 (d, J = 8.4 Hz, 2H), 7.59 (s, 1H), 7.02 (d, J = 8.9 Hz, 1H), 6.61 (d, J = 8.4 Hz, 2H), 4.05 (t, J = 6.2 Hz, 2H), 3.55–3.34 (m, 4H), 3.23–3.18 (m, 4H), 3.04–2.97 (m, 2H), 2.81–2.76 (m, 4H), 2.72 (s, 3H), 2.65–2.57 (m, 2H), 1.77–1.74 (m, 4H), 1.47–1.45 (m, 2H), 1.23 (t, J = 6.3 Hz, 2H). MS (ESI): m/z calcd for C29H39N5O5S 569.2672, found 570.3038 [M + H]+.

2,2′-(7-(5-((2-(4-(Dimethylamino)phenyl)benzo[d]thiazol-6-yl)oxy)pentyl)-1,4,7-triazonane-1,4-diyl)diacetic Acid (13)

Yield: 92.6%. 1H NMR (400 MHz, DMSO-d6) δ 7.80 (d, J = 8.9 Hz, 1H), 7.76 (d, J = 8.4 Hz, 2H),7.60 (s, 1H), 7.04 (d, J = 8.8 Hz, 1H), 6.79 (d, J = 8.7 Hz, 2H), 4.05 (t, J = 6.1 Hz, 2H), 3.61–3.41 (m, 4H), 3.27–3.02 (m, 8H), 2.99 (s, 6H), 2.65–2.48 (m, 4H), 1.77 (dd, J = 13.9, 6.8 Hz, 4H), 1.45 (d, J = 8.9 Hz, 2H), 1.31–1.16 (m, 2H). MS (ESI): m/z calcd for C30H41N5O5S 583.2828, found 584.3211 [M + H]+.

General Procedures for the Synthesis of Complexes 14–17

Precursors 10–13 (0.2 mmol) were dissolved in a solution of NaOAc (2 mL,10 mM) and ethanol (1 mL) and treated with AlF3·3H2O (0.6 mmol). After adjusting pH to 4.5–5.0, the mixture was refluxed for 15 min and cooled down to room temperature to adjust pH again. The pH was maintained between 4.5 and 5.0 during the entire process and refluxed overnight. The solvent was removed by vacuum, and the residue was purified by silica gel chromatography using methanol and dichloromethane (1:1, v/v) to afford complexes 14–17.

Al–19F Complex 14

Yield: 29.6%. 1H NMR (400 MHz, DMSO-d6) δ 7.74 (d, J = 8.3 Hz, 2H), 7.72 (d, J = 8.6 Hz, 1H), 7.65 (s, 1H), 7.05 (d, J = 8.8 Hz, 1H), 6.61 (d, J = 8.4 Hz, 2H), 6.38 (s, 1H), 4.07 (t, J = 6.0 Hz, 2H), 3.35–3.32 (m, 4H), 2.95 (s, 3H), 2.91–2.73 (m, 12H), 1.92 (t, J = 6.1 Hz, 2H), 1.21 (s, 2H). 13C NMR (101 MHz, TFA-d) δ 179.39, 169.07, 160.64, 141.22, 135.31, 132.72, 130.68, 128.02, 124.67, 121.98, 105.74, 66.62, 63.32, 63.18, 55.94, 54.63, 53.59, 53.45, 53.07, 50.90, 49.73, 38.70, 23.05. HRMS (ESI): m/z calcd for C27H33AlFN5O5S 585.2002, found 586.2075 [M + H]+.

Al–19F Complex 15

Yield: 37.4%. 1H NMR (400 MHz, DMSO-d6) δ 7.81 (d, J = 8.3 Hz, 2H), 7.76 (d, J = 8.6 Hz, 1H), 7.68 (s, 1H), 7.06 (d, J = 9.2 Hz, 1H), 6.79 (d, J = 8.4 Hz, 2H), 4.08 (t, J = 6.0 Hz, 2H), 3.34–3.32 (m, 4H), 2.99 (s, 6H), 2.96–2.69 (m, 12H), 1.94 (t, J = 5.5 Hz, 2H), 1.21 (s, 2H). 13C NMR (101 MHz, TFA-d) δ 178.97, 168.22, 160.27, 146.38, 134.98, 132.38, 130.61, 128.00, 122.61, 121.61, 105.31, 66.23, 62.95, 62.78, 55.56, 54.25, 53.23, 53.08, 52.72, 50.51, 49.33, 46.96, 22.67. HRMS (ESI): m/z calcd for C28H35AlFN5O5S 599.2158, found 600.2234 [M + H]+.

Al–19F Complex 16

Yield: 43.9%. 1H NMR (400 MHz, DMSO-d6) δ 7.75 (d, J = 8.4 Hz, 2H), 7.73 (d, J = 8.3 Hz, 1H), 7.59 (s, 1H), 7.02 (d, J = 8.9 Hz, 1H), 6.61 (d, J = 8.4 Hz, 2H), 6.39 (d, J = 4.3 Hz, 1H), 4.02 (t, J = 6.0 Hz, 2H), 3.15–2.96 (m, 6H), 2.88 (s, 3H), 2.72–2.65 (s, 12H), 1.73 (t, J = 6.3 Hz, 2H), 3.15–1.56 (m, 2H), 1.40 (t, J = 7.2 Hz, 2H). 13C NMR (101 MHz, TFA-d) δ 179.46, 168.51, 161.23, 141.12, 135.00, 132.78, 130.64, 128.10, 124.66, 122.36, 105.38, 69.19, 63.34, 63.13, 58.93, 54.66, 53.69, 53.44, 53.13, 50.80, 49.65, 38.73, 28.27, 23.43, 22.84. HRMS (ESI): m/z calcd for C29H37AlFN5O5S 613.2315, found 614.2394 [M + H]+.

Al–19F Complex 17

Yield: 38.2%. 1H NMR (400 MHz, DMSO-d6) δ 7.79 (d, J = 8.4 Hz, 2H), 7.75 (d, J = 8.5 Hz, 1H), 7.52 (s, 1H), 7.02 (d, J = 9.2 Hz, 1H), 6.76 (d, J = 8.8 Hz, 2H), 4.02 (t, J = 5.5 Hz, 2H), 3.37–3.34 (m, 6H), 2.99 (s, 6H), 2.92–2.65 (s, 12H), 1.78 (t, J = 6.0 Hz, 2H), 1.65–1.62 (m, 2H), 1.41 (t, J = 6.8 Hz, 2H). 13C NMR (101 MHz, TFA-d) δ 163.64, 154.42, 149.01, 137.29, 128.04, 126.28, 124.74, 122.81, 118.33, 117.94, 104.26, 75.34, 70.66, 70.49, 67.11, 63.72, 62.94, 62.74, 62.49, 60.62, 59.68, 57.94, 42.62, 38.72, 38.25. HRMS (ESI): m/z calcd for C30H39AlFN5O5S 627.2471, found 628.2554 [M + H]+.

In Vitro Binding Assay Using Aggregated Aβ1–42 Peptides

The Aβ1–42 aggregation was prepared according to methods reported previously.[31] Competitive binding assays were carried out as the procedures below. Borosilicate glass tubes (12 mm × 75 mm) contained aggregated Aβ1–42 aggregation (100 μL, 0.76 μM), [125I]IMPY (100 μL, approximately 100 000 cpm), bovine serum albumin solution (700 μL, 0.1% in water), and complex 14–17 (100 μL, 10–4 to 10–9.5 M in ethanol) was incubated at 37 °C for 2 h. Then, the mixture was filtered by Whatman GF/B filters on a Brandel Mp-48T cell harvester for separating the bound and free radioactivity. The filter sections with bound [125I]IMPY were counted using a γ-counter. After repeating three times, the half maximal inhibitory concentrations (IC50) were calculated using GraphPad Prism 4.0, and the inhibition constant (Ki) values were calculated with by the Cheng–Prusoff equation: Ki = IC50/(1 + [L]/Kd).[32]

Partition Coefficient Determination

The log D value of [18F]17 was measured as follows. After the plastic centrifuge tube containing presaturated n-octanol (3.6 mL) and phosphate-buffered saline (PBS) (0.05 M, 3.0 mL) was added with 18F-labeled tracer (60 μCi), the mixture was vortexed for 3 min, followed by 5 min of centrifugation at 3000 rpm. Then, 50 μL of the n-octanol and 500 μL of the buffer were pipetted in two test tubes for measuring the amount of radioactivity, respectively, and each measurement was repeated three times. The partition coefficient was obtained by calculating the logarithm of the ratio from n-octanol versus PBS by the count. Then, the n-octanol layer was repartitioned with vortexing, centrifuging, and counting protocols until consistent values were obtained.

In Vitro Fluorescence Staining

Brain slices from CAA patients were deparaffinized in xylene and washed with ethanol according to the procedures reported previously. Then, the slices were incubated in solution of 10% ethanol–water of Al–19F complexes (1.0 μM) for 5 min and washed with 50% ethanol–water. Fluorescent images were observed on an EVOS FL imaging system with a DAPI filter set (excitation, 405 nm). And the same brain slices were stained by DANIR-3b (1.0 μM) to confirm the localization of Aβ plaques using the same method with the RFP filter.

Radiochemistry

18F– loaded on a QMA cartridge was eluted with 0.5 mL of saline. The pH of 18F– solution was adjusted with glacial acetic acid to 4. (Al–18F)2+ was obtained by adding AlCl3 solutions (2 mM, 22.5 μL) to 18F– saline solution and incubating for 10 min at ambient temperature. The stock aluminum chloride (2 mM) above was prepared by dissolving ultra pure AlCl3·6H2O in sodium acetate buffer (0.1 M, pH = 4). Then, the solution of precursors 13 (500 nmol) in sodium acetate buffer (0.1 M, 1 mL) was added and the mixture was kept at 110 °C for 10 min. The final product was purified by HPLC (Venusil MP C18 column, 5 μm, 4.6 mm × 250 mm, acetonitrile/water = 40:60%, flow rate = 1 mL/min). The total labeling process took about 40 min with radiochemical yield up to 17.8%. After HPLC purification, the purity of labeled tracers was higher than 98%.

Biodistribution in Normal Mice

A saline solution of HPLC-purified [18F]17 (100 μL, 10 μCi, 10% EtOH) was injected to ICR mice (18–22 g, male) via the tail vein. The mice were decapitated to sacrifice exactly at 2, 10, 30, and 60 min post injection. After dissection, organs of interest and samples of blood were weighed and measured by γ-counter. The results were expressed as the percentage of the injected dose per gram (% ID/g) of blood or organs.

In Vitro Autoradiography Studies

Brain slices from CAA patients and normal control were deparaffinized with xylene and washed by ethanol according to the procedures reported previously. Thereafter, the slices were incubated with purified [18F]17 (100 μL, 20 μCi) at room temperature for an hour. Thereafter, they were washed with 40% ethanol–water for 3 min and then exposed to a phosphorus plate (PerkinElmer) for 40 min. The autoradiographic images were obtained using a storage phosphor system (PerkinElmer) under 600 dpi resolution. Fluorescent staining by DANIR-3b was performed on the same slices to further confirm specific binding afterward.