|Home | About | Journals | Submit | Contact Us | Français|
Alzheimer’s disease (AD) is a neurodegenerative disease characterized by Aβ plaques in the brain. The aim of this study was to evaluate the effectiveness of a novel radiotracer, 4-[11C]methylamino-4′-N,N-dimethylaminoazo-benzene ([11C]TAZA), for binding to Aβ plaques in postmortem human brain (AD and normal control (NC)).
Radiosyntheses of [11C]TAZA, related [11C]Dalene (11C-methylamino-4′-dimethylaminostyrylbenzene), and reference [11C]PIB were carried out using [11C]methyltriflate prepared from [11C]CO2 and purified using HPLC. In vitro binding affinities were carried out in human AD brain homogenate with Aβ plaques labeled with [3H]PIB. In vitro autoradiography studies with the three radiotracers were performed on hippocampus of AD and NC brains. PET/CT studies were carried out in normal rats to study brain and whole body distribution.
The three radiotracers were produced in high radiochemical yields (>40%) and had specific activities >37 GBq/μmol. TAZA had an affinity, Ki = 0.84 nM and was five times more potent than PIB. [11C]TAZA bound specifically to Aβ plaques present in AD brains with gray matter to white matter ratios >20. [11C]TAZA was displaced by PIB (>90%), suggesting similar binding site for [11C]TAZA and [11C]PIB. [11C]TAZA exhibited slow kinetics of uptake in the rat brain and whole body images showed uptake in interscapular brown adipose tissue (IBAT). Binding in brain and IBAT were affected by preinjection of atomoxetine, a norepinephrine transporter blocker.
[11C]TAZA exhibited high binding to Aβ plaques in human AD hippocampus. Rat brain kinetics was slow and peripheral binding to IBAT needs to be further evaluated.
Alzheimer’s disease (AD) is the most common type of dementia, accounting for 50%–80% of dementia cases characterized by the accumulation of amyloid β (Aβ) plaques and neurofibrillary tangles (NFT) in the brain (Braak and Braak, 1991). Benzothiazole derivatives, such as [11C]PIB (Fig. 1; 1), have been developed to bind amyloid β (Aβ) plaques with high affinity and are capable of crossing the brain–blood barrier to be valuable for the application of in vivo Aβ plaque imaging in PET studies (Mathis et al., 2003). Imaging with [11C]PIB plays an essential role in the clinical studies for evaluation of drug efficacy in relation to the development of the Aβ plaques in vivo (Klunk et al., 2004). As an Aβ plaque radiotracer, [11C]PIB characteristic changes in the brains of patients with AD has played a major role in the AD neuroimaging initiative (e.g., Weiner et al. 2015). Efficacy of bapineuzumab, a humanized N-terminal specific anti-Aβ monoclonal antibody, in AD was conducted in a large multicentre phase 3 clinical trial and used [11C]PIB for monitoring changes in Aβ plaques (Rinne et al., 2010;Salloway et al., 2014).
Following a large number of human research studies using [11C]PIB in mild cognitive impairment (MCI) and AD (Raniga et al., 2008; Weiner et al., 2015), fluorinated [18F]Florbetapir (Fig. 1; 2) was the first fluorine-18 agent approved for clinical use in AD (Stephenson et al., 2007; Zhang et al., 2005; Zhu et al., 2014). Although fluorine-18 offers advantages of the longer half-life, the wider use of [18F]Florbetapir has been slow (Camus et al., 2012). High white matter binding and the low standard uptake values (SUV) in the cortex of AD has caused caution. Thus, an agent that can provide a significantly higher SUV in the AD cortex may be an improvement toward clinical value. There has been continued interest in the development of Aβ plaque imaging agents that are labeled with fluorine-18 and which may provide a higher target to nontarget ratio (Cai et al., 2007; Mathis et al., 2012; Ni et al., 2013; Verhoeff et al. 2008; Zhu et al., 2014). Table I describes agents for imaging plaques some of which were recently approved by FDA, although there have been many more promising agents under development (Eckroat et al., 2013; Tu et al., 2015; Vandenberghe et al., 2010). With increasing efforts to find treatments and cure for AD, there is much research into imaging plaques and NFT essential to the diagnosis and clinical management of AD (Ariza et al., 2015; Barten and Albright, 2008; Schenk et al. 1999; Verhoeff, 2007; Zimmer et al., 2014).
Congo red (Fig. 1; 3) is a well-known dye which has also been used to assay amyloid aggregation (e.g., Jameson et al., 2012). The unique “diazo” structure present in Congo red makes it different from the other known structures such as thioflavin and stilbenes from which PIB and Florbetapir were derived (Fig. 1; 1,2). Iodinated phenyldiazenyl benzothiazole derivatives (Fig. 1; 4) have recently been reported for imaging NFT in AD brains (Matsumura et al., 2011). We have identified a unique compound, [11C]TAZA [4-methylamino-4′-N,N-dimethylaminoazobenzene(Mukherjee et al., 2012; Fig. 1; 5)] as a radioligand for the binding site of Aβ plaque which has the “azo” functionality found in Congo red. It was anticipated that the structure of TAZA, with the additional “azo” heteroatoms (Fig. 1b, shown in green ovals, in structures 3, 4, and 5) may render it less lipophilic toward white matter. [11C]TAZA has the “aminomethylphenyl” structure (Fig. 1a, shown in red boxes, in structures 1, 2, 5, and 6). For purposes of comparison, we also prepared [11C]Dalene [4-methylamino-4′-N,N-dimethylaminostyrylbenzene; (Fig. 1; 6; reported previously Kung et al., 2003)], which instead of the “azo” moiety found in TAZA, contains the olefin. This olefin structure in Dalene is similar to that found in Florbetapir and also in its related analogue, Florbetaben (Villemagne et al., 2011). The “dimethylaminophenyl” structure (Fig. 1c, shown in blue boxes, is in structures 4, 5, and 6).
In order to assess the value of [11C]TAZA in Aβ plaque imaging, we report here the following: 1. Measurement of in vitro binding affinity of TAZA to human Aβ plaques; 2. Radiosynthesis of [11C]TAZA, as well as the radiosynthesis of [11C]Dalene for purposes of comparison using in vitro and in vivo studies with [11C]PIB. 3. Comparison of the binding of the 3 radiotracers, [11C]TAZA, [11C]Dalene and [11C]PIB to postmortem AD human brain slices. 4. Evaluation the three radiotracers, [11C]TAZA, [11C]Dalene and [11C]PIB in normal rats in order to assess brain and whole body distribution of the radiotracers.
All compressed gases including helium, hydrogen, and nitrogen with 0.25% oxygen were supplied by Airgas. A GE Scanditronix MC17 cyclotron (17 MeV, 45 μA) was used to generate high specific activity [11C]CO2 using a nitrogen gas target ([14N] to [11C] using p, α reaction). Carbon-11 radioactivity were counted in a Capintec CRC-15R dose calibrator while low level counting was carried out in a Capintec Caprac-R well-counter. Automated synthesizer, GE TRACERlab FXcPRO (TRACERlab; from GE Healthcare, Milwaukee, WI) was used for all radiosyntheses. Shimalite-Ni by Shimadzu was applied as a catalyst for [11C]CO2 reduction. Molecular Sieves 4A, 80/100 mesh was from Alltech Ascarite, 20–30 mesh, silver triflate, iodine, and other chemicals were supplied by Aldrich-Sigma Corporation. Porapak Q 50–80 mesh was used for the [11C]CH3I trap and 60–80 mesh. Carbosphere was prepared for the silver triflate column. All solvents used were provided by Fisher Scientific. For quality control (QC) residual solvent gas chromatography with FID, a HP 6890 with 30 m Rtx 624 silica column was used. For QC chemical purity, a Waters or Gilson HPLC system with UV detector set at 350 nm was used with 4.6 × 250 mm C18 Econosil reverse-phase analytical column was used. A semipreparative HPLC column 100 × 250 mm 10 μm Econosil C18 reverse-phase was supplied by Grace Discovery Corp. for purification within the TRACER-lab. Millex-FG sterile pyrogen free 0.20 μm filters and Corning sterile filters were used as vent cartridges. Syringes were supplied by BD. The precursor, 2-(4′-aminophenyl)-6-hydroxybenzothiazole, and reference standard 2-(4′-methylaminophenyl)-6-hydroxybenzothiazole (PIB) were obtained from ABX, Germany and precursors, 4-amino-4′-(N,N-dimethylamino)stilbene (TCI America, Portland, OR) and 4-amino-4′-N,N-dimethylaminoazobenzene (Aldrich Chem Co., St. Louis, MO) were obtained commercially. Empty sterilized vials (10 mL), sterile water, saline USP were from Hospira Pharmaceuticals and ethanol 200 proof, USP (Pharmco-AAPER) were used for formulation. C-18 Sep Pak light cartridges were supplied by Waters Corp. for extraction.
Analytical thin-layer chromatography (TLC) was used to monitor reactions (Baker-flex, Phillipsburg, NJ). Electrospray mass spectra were obtained from a Model 7250 mass spectrometer (Micromass LCT). Proton NMR spectra were recorded on a Bruker OM EGA 500-MHz spectrometer. Tritium was assayed using a Packard Tri-Carb Liquid scintillation counter with 65% efficiency. Human postmortem brain tissue samples were obtained from the UCI Alzheimer’s Disease Research Center (ADRC) brain tissue repository for in vitro experiments. Age and gender matched AD brain and normal control brain tissue (hippocampus; n =4) were used. AD brain samples selected for end-stage pathology (Braak & Braak stage of VI (Braak and Braak, 1991). Chunks of frozen tissue were dissected for immunohistochemical and autoradiographic techniques, as well as for biochemical experiments. Human postmortem brain slices were obtained on a Leica 1850 cryotome. Carbon-11 autoradiographic studies were carried out by exposing tissue samples on storage phosphor screens (Perkin Elmer Multisensitive, Medium MS). The apposed phosphor screens were read and analyzed by Opti-Quant acquisition and analysis program of the Cyclone Storage Phosphor System (Packard Instruments Co., Boston, MA). A preclinical Inveon dedicated PET scanner (Siemens Medical Solutions, Knoxville, TN) with a transaxial FWHM of 1.46 mm, and axial FWHM of 1.15 mm (Constantinescu and Mukherjee, 2009) was used for the PET studies. PET/CT images of rats were obtained and analyzed using ASIProVM, IRW, and PMOD softwares. All animal studies were approved by the Institutional Animal Care and Use Committee of University of California-Irvine. Postmortem human brain studies were approved by the Institutional Biosafety Committee of University of California, Irvine, CA.
Binding affinities of TAZA (prepared by reacting 4-amino-4′-dimethylaminoazobenzene with methyl triflate and purified by preparative TLC to provide TAZA (4-methylamino-4′-N,N-dimethylaminoazobenzene mass spectra (m/z, %), 255, [M + H]+, 100%) and PIB were measured using human AD brain homogenate. Human brain homogenates were labeled with 3H-PIB which is known to bind to human Aβ-amyloid. The brain tissue (hippocampal tissue from AD subject; 0.16 g) was homogenized in 10 mL of assay buffer for 30 s (10 mM PBS, pH 7.4) using Tekmar tissumizer (15 s, half maxima speed). A fixed concentration of [3H]PIB (2 nM) was incubated with the brain homogenate in the presence of various concentrations of TAZA and PIB (10−11–10 −5M) in the assay buffer (10% ethanol, 90% 10 mM PBS, pH 7.4). Nonspecific binding was determined by including 10 μM of PIB. Total assay volume was 0.5 mL. To start, 0.2 mL of the brain homogenate was added to each test tube containing [3H]PIB, different concentrations of drugs and 10 μM of PIB for nonspecific binding. The assay was done in duplicate and all test tube samples were incubated for 1 h in a 37°C water bath. After incubation a rapid vacuum filtration was implemented through Whatman GF/C filter article (pre-soaked in 0.1% polyethylamine in 10 mL of Millipore water) using Brandel tissue harvester. The filter was washed three times with 5 mL of cold buffer and was transferred to vials with 5 mL Bio-Safe II scintillation cocktail and counted for 10 min in a scintillation counter. Data were analyzed using following procedure: (a) the nonspecific binding of [3H]PIB was subtracted for all samples; (b) the specific binding was normalized to 100% (no competitive ligand), and (c) the binding isotherms were fit to the Hill equation (KELL BioSoft software (v 6), Cambridge, U.K.). The Ki was calculated by the Cheng-Prussof equation using the reported kda value, for 3H-PIB.
The [11C]CO2 (>37GBq) was produced in the MC-17 cyclotron and transferred to the TRACERlab automated radiosynthesis unit where the [11C]CO2 was trapped on the molecular sieve (0.3 g) in the packed column at T =25°C. Conversion to [11C]methane ([11C]CH4) was achieved by reduction of [11C]CO2 with H2 using 0.2 g Shimalite-Ni (Shimadzu) loaded on the molecular sieves (350°C for 30 s). Yields of [11C]CH4 averaged of 16–18 GBq. Gas-phase iodination of [11C]CH4 to [11C]methyl iodide ([11C]CH3I) occurred at 760°C and iodine vessel at 90°C in a closed loop. The iodinated product, [11C]CH3I was passed over a MeI trap (Porapak Q, Alltech Corp.) with an average yield of [11C]CH3I of 16 GBq starting from >37GBq of [11C]CO2. The [11C]CH3I passed through the triflate column, which had been preheated to 195°C, and bubbled into the reaction vessel. The total automated reaction from [11C]CO2 to [11C]methyl triflate ([11C]CH3OTf) was approximately 11 min and provided approximately 11 GBq of [11C]CH3OTf.
Radiosynthesis of [11C]TAZA was carried out by reacting 4-amino-4′-dimethylaminoazobenzene (1 mg/0.5 mL acetone) with [11C]CH3OTf prepared in the TRACERlab. 11C-Methyltriflate was trapped at −20°C in acetone containing the precursor and subsequently heated for 5 min at 80°C under helium gas atmosphere in a closed reaction vessel. This mixture was very bright yellow-red in color (due to the color of the precursor). Reverse phase HPLC purification (40:60 of 0.1% aqueous trimethylamine/acetonitrile, flow rate of 2.5 mL/min) provided [11C]TAZA as the major radioactive product eluting at approximately 20 min. Purified [11C]TAZA was taken up in 10% ethanol-90% sterile saline and filtered through a sterilizing filter into a sterile pyrogen-free single dose vial to provide approximately 500 MBq of [11C]TAZA for in vitro and in vivo studies. Radiochemical purity was >95% and specific activity of [11C]TAZA was >37 TBq/mmol.
Similar to the procedure outlined above in [11C]TAZA section, radiosynthesis of [11C]Dalene was carried out by reacting 4-amino-4′-dimethylaminostyrylbenzene (1 mg/0.5 mL acetone) with [11C]CH3OTf prepared in the TRACERlab. This mixture was very yellow in colour (due to the colour of the precursor, albeit less intense than TAZA). Reverse phase HPLC purification (40:60 of 0.1% aqueous trimethylamine/acetonitrile, flow rate of 1.5 mL/min) provided [11C]Dalene as the major radioactive product eluting at approximately 40 min. Purified [11C]Dalene was taken up in 10% ethanol-90% sterile saline and filtered through a sterilizing filter into a sterile pyrogen-free single dose vial to provide approximately 370 MBq of [11C]Dalene for in vitro and in vivo studies. Radiochemical purity was >95% and specific activity of [11C]Dalene was >37 TBq/mmol.
Similar to the procedure outlined above in [11C]TAZA section, radiosynthesis of [11C]PIB was carried out by reacting 5-OH-BTA-0 (1 mg/0.5 mL acetone) with [11C]CH3OTf prepared in the TRACER-lab. Reverse phase HPLC purification (40:60 of aceto-nitrile/0.1 M ammonium formate, flow rate of 1 mL/min) provided [11C]PIB as the major radioactive product eluting at approximately 8 min. The HPLC fraction was collected and diluted with 70 mL of MilliQ water, and isolated by solid phase extraction to a <10% ethanolic isotonic saline solution which is filtered through a sterilizing filter into a sterile vial provided approximately 740 MBq of [11C]PIB for in vitro and in vivo studies. Radiochemical purity was >95% and specific activity of [11C]PIB was >37 TBq/mmol.
[11C]TAZA, [11C]PIB and [11C]Dalene were used for autoradiographic studies. Human hippocampus sections (7 μm thick) were preincubated in buffer (40% EtOH) for 10 min. The brain sections were placed in a glass chamber and incubated with [11C]TAZA, [11C]PIB, and [11C]Dalene (approximately 740 kBq/cc) in 40% EtOH at 37°C for 1 h. The slices were then washed with cold Millipore water, 70%–90%–70% EtOH, water for 2,1,1,1,1 min, respectively. Nonspecific binding was measured in the presence of 10 μM PIB. The brain sections were air-dried, exposed overnight on a phosphor film, and then placed on the Phosphor Autoradiographic Imaging System (Packard Instruments Co). Regions of interest (ROIs) were drawn on the slices and the extent of binding of 11C-PIB was measured with DLU/mm2 using the Opti-Quant acquisition and analysis program (Packard Instruments Co). Neighboring slices were immunostained with 4G8 antibody using modifications of reported methods (Braak et al., 2011). Slides were warmed to room temperature and washed in TBS (Tris-buffered saline, pH 7.5), followed by antigen ret-rival in 70% formic acid for 15 min. After endogenous peroxidase quenching, sections were stained with bio-tinylated anti-Aβ antibody 4G8 (Covance, Princeton, NJ) at 1:800 dilution followed by incubation according to manufacturer instructions (Vector Labs, Burlingame, CA). Peroxidase reaction was developed with 3,3′-diaminobenzidine reagent (Vector Labs, Burlingame, CA). Pictures were taken on Olympus BX61 microscope.
Male Sprague-Dawley rats (308–468 g) were single housed in a climate controlled room and had full access to food and water. The rats were fasted 24 h prior to time of scan. On the day of the study, rats were anesthetized using 4.0% isoflurane. The rat was then positioned on the scanner bed by placing it on a warm-water circulating heating pad and kept anesthetized with 2.5% isoflurane anesthesia applied using a nose-cone. Preparation of dose injection was as follows: approximately 100 ± 25 MBq of [11C]TAZA, [11C]Dalene, or [11C]PIB was drawn into a 0.5 mL syringe with a 29 gauge needle and diluted with sterile saline to a final volume of 0.3 mL. Rats received two 90 min scans on two separate days (1–4 weeks apart) with an Inveon dedicated PET scanner (Siemens Medical Solutions). On first day each rat received a baseline scan, one with [11C]Dalene and the other with [11C]TAZA. The second day each animal was preinjected iv with a 50 μL bolus of atomoxetine (ATX [1–2 mg/kg], 17 min before [11C]Dalene and 2 min before [11C]TAZA. Following PET each animal received a CT scan with an Inveon MM CT, which was used for attenuation and scatter correction. The CT scan was performed at 2 overlapping bed positions with detector-source rotating 220 degrees around the animal with 120 projections acquired. CT images were reconstructed with a cone beam algorithm into 480 × 480 × 632 image arrays with a 206 μm pixel size. All PET images were corrected for scatter, attenuation, and radioactive decay. Acquired list-mode were sorted dynamically into multiple frames. Image reconstruction was performed with an OSEM3D/fast MAP algorithm (16 OSEM 3D subsets, 2 iterations, 18 MAP iterations) resulting in 128 × 128 × 159 image array with a 0.79 mm pixel size. Images were visualized and analyzed with ASIPro (CTI Concorde Microsystems, LLC.) and PMOD (PMOD Technologies) software.
Analysis was performed with PMOD software package (PMOD Technologies). All images were normalized to Paxinos & Watson space via coregistration with an MR rat template (Schweinhardt et al. 2003). Volumes of interest (VOIs) were drawn on the MR template and placed on brainstem, frontal cortex, cerebellum and other brain regions. VOIs of the inter-scapular BAT (IBAT) were delineated visually by contouring the uptake that was clearly above normal background uptake. VOIs from all regions were used to derive time-activity curves (TACs) and were normalized by dividing them the injected dose.
In vitro binding affinity of TAZA (Fig. 2) in human AD brain homogenates using [3H]PIB yielded an IC50 of 1.28 nM for TAZA compared with 10 nM for PIB. These IC50 values for PIB are lower compared with previously reported values for PIB in AD brain homogenates of frontal cortex, IC50 of 3.84 nM (Hellstrom-Lindahl et al., 2014). Inhibition constants were calculated using the Cheng-Prusoff equation (Ki = IC50/(1+(conc of [3H]PIB)/kDa)) and a kda= 3.77 nM for [3H]PIB (Fodero-Tavoletti et al., 2007). The inhibition constant, Ki for TAZA was 0.84 nM and that for PIB was 6.67 nM. The measured Ki value for PIB is in the range of previously reported Ki values for PIB (Klunk et al., 2004).
Radiolabeling reactions used [11C]CH3OTf produced in the TRACERlab. Both [11C]TAZA and [11C]Dalene were the single major radiochemical product. Both the radiotracers were produced in high radiochemical yields and specific activity. The slower HPLC flow rate in the case of [11C]Dalene increased the retention time but minimized the potential contamination from the starting material due to tailing of the starting material. It should be noted that the HPLC purification of [11C]TAZA and [11C]Dalene were carried out on a separate HPLC outside the TRACERlab unlike that of [11C]PIB which was done using the in-built HPLC in TRACERlab. The three radiotracers were prepared in amounts of 370–740 MBq in specific activities generally >37 TBq/mmol and were found to be stable in 10% ethanolic saline solution for in vitro and in vivo studies.
Extensive binding of [11C]TAZA was seen in the gray matter regions of the two AD subjects as seen in Figures 3B and 3F while white matter had significantly lower binding. This gray matter binding was significantly reduced when the brain sections were treated with PIB (Fig. 3C). The binding in the gray matter was confirmed by immunostaining for the presence of Aβ plaques as can be seen in Figures 3D and 3G. Normal control subjects had very little gray matter binding compared with the AD subjects (Figs. 3H and 3I). As can be seen in Figure 3J, AD2 subject had the greatest amount of [11C]TAZA binding followed by AD1 subject. The NC1 subject had little [11C]TAZA binding while NC2 exhibited some localized hot spots. Over 90% of the binding of [11C]TAZA was displaced by PIB from the AD1 and AD2 subjects.
Shown in Figure 4 is a comparison of the degree of binding of [11C]TAZA, [11C]Dalene and [11C]PIB in AD2 and NC1 subjects. The highest amount of gray matter binding was observed with [11C]TAZA and the highest amount of white matter binding was observed with [11C]Dalene. The three radiotracers exhibited little binding in the normal control subjects, although [11C]Dalene exhibited higher nonspecific binding (Fig. 4D) compared with [11C]TAZA and [11C]PIB. The gray matter to white matter ratio (GM/WM) was the greatest for [11C]TAZA as shown in Table II. The ratio of 19.6 for AD1 and 30.5 for AD2 was due to both the high binding in GM and low binding in WM for [11C]TAZA as can be seen in Figure 4G. Lowest ratios (approximately 5-fold lower than [11C]TAZA) were measured for [11C]Dalene due to the higher WM binding. The GM/WM ratios of [11C]PIB were higher than [11C]Dalene, but about 4-fold lower than [11C]TAZA (Table II).
Rat brain uptake of [11C]TAZA after intravenous administration occurred slowly over the 90 min PET scan (Figs. 5A–5C). Regional brain localization of [11C]TAZA was confirmed by coregistering the PET scan with a rat brain MRI template (Fig. 5A). Greatest uptake of [11C]TAZA occurred in the brainstem followed by cerebellum and the frontal cortex (Fig. 5B). Uptake of [11C]TAZA plateaued at approximately 60 min. The maximum activity levels in the various brain regions reached approximately 0.3% injected dose/cc. Ratios of brainstem to frontal cortex was 1.78 at 85 min while cerebellum to frontal cortex was 1.20, respectively.
A similar slow brain uptake over the 90 min scan was observed for [11C]Dalene (Figs. 5D–5F). Regional brain localization of [11C]Dalene was similar to [11C]TAZA, which was confirmed by coregistering the PET scan with the rat brain MRI template (Fig. 5D). [11C]Dalene distribution was observed in the brainstem which was a little more discrete compared with [11C]TAZA, followed by cerebellum and the frontal cortex (Fig. 5E). Uptake of [11C]Dalene appeared to plateau at approximately 60 min. The peak levels in the various brain regions were lower than those measured for [11C]TAZA reaching approximately 0.1% injected dose/cc. Ratios of brainstem and cerebellum to frontal cortex at 85 min were 1.80 and 1.04, respectively.
Whole body PET/CT scans with [11C]TAZA and [11C]Dalene revealed distinct distribution features and similarities between the two radiotracers (Fig. 6). Localization of [11C]Dalene was visualized in interscapular brown adipose tissue (IBAT) and confirmed with the coregistered PET/CT scan (Fig. 6A). Uptake of [11C]Dalene in IBAT was gradual and similar to that observed in the brainstem. Activity in IBAT leveled off sooner (approximately 20 min) than was observed in the brain to a similar level of 0.12% injected dose/cc (Fig. 6C). Compared with [11C]Dalene, higher levels of [11C]TAZA binding was observed in IBAT (Fig. 6B). In addition to IBAT, other regions of brown adipose tissue (BAT) were also observed (Fig. 6B) consistent with our previous studies on activated BAT in the rat (Mirbolooki et al., 2011). Uptake of [11C]TAZA in IBAT was gradual and continued to increase beyond the scan time of 90 min with 0.43% injected dose/cc at the end of the scan. This uptake was greater than [11C]TAZA measured in the brain regions.
Our previous [18F]FDG studies with atomoxetine, a norepinephrine transporter (NET) inhibitor have shown a high level of activation of IBAT as well as other BAT areas in the rodent model (Mirbolooki et al., 2013), similar to the accumulation of [11C]TAZA seen in Figure 6B. The activation occurred by elevated levels of norepinephrine due to the blocking of NET by atomoxetine in the sympathetic innervation in BAT regions. In order to assess the potential interaction of [11C]TAZA at the NET sites in BAT, competition studies with atomoxetine were carried out. Preinjection of atomoxetine prior to administration of [11C]TAZA had a major effect on BAT binding of [11C]TAZA (Figs. 7A and 7B). Binding of [11C]TAZA was reduced in all areas of BAT as seen in Figure 7B and time-activity curves shown in Figure 7C show a very different uptake and clearance in the presence of atomoxetine. Atomoxetine caused a rapid high initial uptake (0.15% injected dose/cc) followed by a clearance of [11C]TAZA from IBAT to 0.02% injected dose/cc. This suggests a competitive inhibition of [11C]TAZA by atomoxetine in BAT regions. The time-activity curve of [11C]TAZA in the presence of atomoxetine resembled the uptake and clearance curve of [11C]PIB in IBAT (Fig. 7C).
Brain uptake of [11C]TAZA in the presence of atomoxetine was dramatically different compared with the baseline scans (Fig. 7D). Initial brain uptake was very high (up to 1% injected dose/cc in the brainstem) which cleared rapidly over the duration of the 90 min from all brain regions. Like the baseline study, greater uptake of [11C]TAZA occurred in the brainstem followed by cerebellum and the frontal cortex (Fig. 7D). The minimal activity levels in the brainstem went down to approximately 0.24% injected dose/cc while the activity levels in the frontal cortex were similar to the maximal levels in the baseline study (Table III). Ratio of brainstem to frontal cortex was 1.58 at 85 min while cerebellum to frontal cortex was 1.10.
Preinjection of atomoxetine prior to administration of [11C]Dalene had a similar effect on BAT binding of [11C]Dalene (Fig. 7E). [11C]Dalene was reduced in IBAT and time-activity curves shown in Figure 7E show a very different uptake and clearance in the presence of atomoxetine. Atomoxetine caused a rapid high initial uptake (0.05% injected dose/cc) followed by a slow clearance of [11C]Dalene from IBAT to 0.05% injected dose/cc rather than a rise as seen in the baseline scan. This suggests a competitive inhibition of [11C]Dalene by atomoxetine in BAT regions. The clearance in the time-activity curve of [11C]Dalene in the presence of atomoxetine was not as rapid as seen in the case of [11C]TAZA (Fig. 7C).
Like [11C]TAZA, brain uptake of [11C]Dalene in the presence of atomoxetine was dramatically different compared with the baseline scans (Fig. 7F). Initial brain uptake was very high (up to 1.1% injected dose/cc in the brainstem) which cleared rapidly over the duration of the 90 min from all brain regions. Like the baseline study, greater uptake of [11C]Dalene occurred in the brainstem followed by cerebellum and the frontal cortex (Fig. 7F). The minimal activity levels in the brainstem went down to approximately 0.53% injected dose/cc while the activity levels in the frontal cortex were a little lower (0.30% injected dose/cc) (Table III). Ratio of brainstem to frontal cortex was 1.75 at 85 min while cerebellum to frontal cortex was 1.0.
Several amyloid plaque imaging agents continue to be developed in an effort to obtain new agents with high affinity for the amyloid plaques and lower non-specific binding (Eckroat et al., 2013). The outcome of such efforts is to translate to higher in vivo SUV values, thus making measurement of changes between NC, mild cognitive impairment (MCI) and AD more precise. Currently, there are at least four imaging agents in use for human Aβ amyloid plaques, in addition to [11C]PIB which continues to be the standard of reference. Table I summarizes some of these agents and their relative binding affinities with respect to PIB. The four fluorinated agents florbetaben, florbetapir, flutemeatmol, and NAV4,694 had relative binding of 0.4, 0.3, 1.1, and 0.4, respectively, with respect to PIB, suggesting that only flutemetamol has a slightly higher affinity than PIB (0.74 nM vs. 0.80 nM, Choi et al., 2009). The affinity of TAZA was found to be approximately 5 times higher than PIB, suggesting a significant effect of the “azo” functionality on the binding to Aβ plaques. Dalene, the analogue of TAZA without the “azo” linker had a weaker affinity (15 nM measured using 125I-TZDM, Kung et al., 2003) and was found to be about 5 times weaker compared to the related compound SB-13 (Kung et al., 2003). The significant effect of the “azo” linker was also confirmed by phenyldiazenyl benzothiazoles, which were found to be markedly more potent than SB-13 (Matsumura et al., 2011). Thus, it appeared that TAZA has a higher affinity for Aβ amyloid plaques and worthy of further in vitro and in vivo evaluation. It must be noted that phenyldiazenyl benzothiazoles (PDB derivatives) were also found to have a high affinity for tau aggregates (Matsumura et al., 2011; Tau =0.48 nM and Aβ1–42 = 8.24 nM for PDB-3, Fig. 1).
Since TAZA contained three N-methyl groups, radiolabeling with carbon-11 methyl group of the precursor in a single step without the need of any protecting group was the easiest option. Carbon-11 methylation has been carried out resulting in high radiochemical purity and specific activity of various radiotracers (e.g., Mukherjee et al., 2004; Shao et al., 2011; Wilson et al., 2004). The use of [11C]methyl triflate as a more reactive methylating agent was used for radiolabeling of [11C]SB-13 (Ono et al., 2003). The TRACERlab was designed to use either [11C]CH3I or [11C]CH3OTf for carrying out [11C]methylations in a routine, automated manner for multiple [11C]radiolabeled compounds, including [11C]PIB (Shao et al., 2011; Wilson et al., 2004). Since both [11C]TAZA and [11C]Dalene are chemically related structures to [11C]SB-13 and [11C]PIB, radiolabeling was efficiently carried out using [11C]CH3OTf in the TRACERlab. Product purification using an inbuilt HPLC system, a solid-liquid phase extraction through C18-filter cartridge, and continuing to dose formulation through a 0.2 μm sterile pyrogen-free filter was used in the automated radiosynthesis of [11C]PIB in the TRAC-ERlab. For [11C]TAZA and [11C]Dalene, we used the reaction vial for the radiolabeling reaction, but purification and dose formulation was carried out in a separate HPLC outside the TRACERlab. The “HPLC loop” method (Ono et al., 2003) may serve well in reducing the amount of precursor material needed (and help reduce the amount of highly chromophoric precursors in the HPLC purification).
High levels of [11C]TAZA binding were found in the AD hippocampus gray matter regions. The levels of binding were greater for AD subject #2 and correlated with the immunostained Aβ amyloid plaques. White matter had very low, background levels and were similar to those found in the normal control subjects. [11C]TAZA This binding was dramatically reduced (>94%) in the presence of PIB suggesting that binding of [11C]TAZA occurred predominantly at the similar site as PIB. Since [11C]PIB binds to Aβ amyloid plaques and not to NFT (Klunk et al., 2004), our findings suggests that [11C]TAZA binding in the AD hippocampus is to Aβ amyloid plaques. These observations also suggest that the “benzothiazole moiety” present in the PDB derivatives may be contributing to their affinity to NFT (Matsumura et al., 2011). Further studies with [11C]TAZA are required to establish the selectivity to Aβ amyloid plaques.
Comparing the binding of the three radiotracers, [11C]TAZA had the highest gray matter to white matter ratios in both AD subjects. The extent of binding in AD2 was greater than AD1, although both were at a similar Aβ amyloid plaque stage (Table II). The differences may be due to the size and location of the hippocampal tissue between the two AD subjects. The normal controls had little binding, although local hot spots were visible in normal control 1. Binding of [11C]PIB was about 4-fold lower than [11C]TAZA in both the AD subjects when the gray matter to white matter ratios were compared. Lowest ratios were found for [11C]Dalene with [11C]PIB showing a 1.4-fold greater ratio compared with [11C]Dalene. The pattern of binding in the two AD subjects was similar for the three radiotracers. Although in vitro ratios do not directly translate to in vivo measures, these findings suggest that [11C]TAZA may be expected to give a significantly higher uptake compared with [11C]PIB in PET studies of AD subjects.
In vivo PET studies were carried out in healthy rats with the three radiotracers. The uptake of the radiotracers was highest in the brainstem region followed by frontal cortex and cerebellum. Midbrain also had some scattered binding. Brain uptake of both, [11C]TAZA and [11C]Dalene was low as can be seen in Figure 5C,F in all brain regions. This is in contrast to [11C]PIB which rapidly enters the brain. The initial uptake of the three radiotracers, [11C]TAZA, [11C]Dalene and [11C]PIB at 4.5 min post-intravenous injection in the frontal cortex were 0.007, 0.015, and 0.714% injected dose/cc, respectively while at 85 min postinjection they were 0.16, 0.069, and 0.045% injected dose/cc (Table III). The early-to-later PET ratio for [11C]PIB in the frontal cortex was >15, while [11C]TAZA and [11C]Dalene were below unity. This high ratio for [11C]PIB is consistent with previously reported kinetics in mice brains (Mathis et al., 2012). Ratio of brainstem to frontal cortex at 85 min postinjection was 1.78 for [11C]TAZA, 1.80 for [11C]Dalene, and 1.24 for [11C]PIB, suggesting greater brainstem binding of [11C]TAZA and [11C]Dalene compared with [11C]PIB. Delivery of the two radiotracers, [11C]TAZA and [11C]Dalene to the brain appears to be limited in the various brain regions, but at later time points resembles the levels found with [11C]PIB. Brain levels appear to plateau at approximately 60 min postinjection (Figs. 5C and 5F). Similar slow brain uptake has been reported for the related “azo” derivative 4, 125I-PDB in mice with levels in the mice brain going from 0.94%injected dose/organ at 2 min to 2.89%injected dose/organ at 60 min postinjection (Matsumura et al., 2011). In a recent study, 13N-labeled azo benzenesulfonic acid derivatives were evaluated in transgenic mice (Tg 2576) and the derivative containing the synthon b and c in [11C]TAZA (Fig. 1) showed promise in binding to Aβ plaques, both in vitro and in vivo (Gaja et al., 2014).
Whole body PET/CT studies of the rats revealed significant binding of [11C]TAZA and [11C]Dalene in regions consistent with BAT (Figs. 6A and 6B). The uptake curves of the two radiotracers in the IBAT followed a similar, slow uptake, with [11C]TAZA showing higher levels compared with [11C]Dalene. For [11C]TAZA, the levels were higher than the brainstem while [11C]Dalene IBAT levels were similar to brainstem (Table III). Our previous findings with adrenergic pathway activation of BAT has revealed the innervation in interscapular BAT (IBAT), cervical BAT, periaortic BAT and intercostal BAT using β3 adrenoceptor agonist (Mirbolooki et al., 2011) and NET inhibitor atomoxetine (Mirbolooki et al., 2013). Localization of [11C]TAZA was clearly visualized in IBAT, cervical BAT and periaortic BAT as seen in Figure 6B. Such consistent localization of [11C]TAZA in BAT regions may be suggestive of a specific interaction with a molecular target in BAT. It should be noted that [11C]PIB exhibited rapid uptake and clearance from IBAT with levels similar to the brainstem (Table III).
Previous reports have revealed micromolar affinities for NET uptake of stilbene and N,N-dimethylaminophenyldiazenyl derivatives containing the guanidino group (Hadrich et al., 1999) and stilbene dimers (Smith et al., 2012). NET existence in BAT has been reported in both rodents (Okuyama et al. 2002) and humans (Hadi et al. 2007). NET inhibitors such as atomoxetine play an important role in BAT activity by increasing norepinephrine levels verified by 18F-FDG PET (Mirbolooki et al., 2013). Based on these observations, we anticipated that the IBAT uptake of both [11C]TAZA and [11C]Dalene may be mediated by NET. Thus, we carried out preinjection experiments with atomoxetine on both [11C]TAZA and [11C]Dalene with the expectation of decreasing the binding of both [11C]TAZA and [11C]Dalene. Figure 7B shows significant decrease of [11C]TAZA in BAT regions and the time-activity curve shown in Figure 7C exhibited a major change and rapid clearance which mimicked the curve for [11C]PIB. Similar change in time-activity curve for [11C]Dalene was seen with atomoxetine as seen in Figure 7F.
Brain NET are responsible for norepinephrine reuptake. NET has been shown to be involved in various psychiatric and behavioral disorders, such as attention deficit hyperactivity disorder (ADHD) (Bymaster et al., 2002, Spencer et al., 2002), substance abuse and depression (Klimek et al., 1997), and AD (Herrmann et al., 2004). NET imaging with PET is challenged due to the widespread distribution of NET in the brain and the lower contrast in density between NET-poor and NET-rich regions (Smith et al., 2006). Of the tracers developed so far [11C]MRB has shown improved specific binding (Gallezot et al. 2011; Logan et al., 2007). [11C]Dalene and [11C]TAZA binding pattern (brainstem. thalamus, midbrain) appeared consistent with the expected NET distribution in the rat brain. [11C]Dalene binding values were higher than those of [11C]TAZA. Atomoxetine preinjection promoted a large and fast increase in [11C]Dalene and [11C]TAZA brain uptake. This effect could be attributed to systemic effects, such as changes in tracer free fraction and effects on blood flow. For [11C]TAZA, interestingly, the ratio between the initial time (4.5 min postinjection) and later time (85 min postinjection) approached 5 for brain regions and in the case of IBAT it was 9 (Table III).
Although, our findings cannot be considered evidence that the secondary binding sites of these two tracers are on NET, changes in their accumulation in different regions in the presence of atomoxetine needs further evaluation, such as possible involvement of energy dependent Uptake-1 process or by diffusion, energy independent Uptake-2 process (Streby et al., 2015). Brain levels of [11C]TAZA in the presence of atomoxetine were significantly higher initially compared with baseline scans (without atomoxetine; Table III). At the end of the scan, the brainstem to frontal cortex ratio was only reduced by approximately 10% (from 1.78 to 1.58). This may suggest low “specific” binding to target NET regions in the normal brain. Similarly, autoradiographic studies in the normal human control hippocampus did not exhibit significant binding of [11C]TAZA. However, further studies are needed to fully characterize the large effect of atomoxetine on the brain uptake of [11C]TAZA in the rodent brain and identify species differences in the binding of [11C]TAZA to NET. These studies will include deciphering the possible uptake (active transport via NET) of [11C]TAZA into the NET presynaptic terminal.
In summary [11C]TAZA presents the following features: (1) Increased signal to noise ratio. [11C]TAZA has higher binding to the human Aβ-plaques compared with [11C]PIB. The unique azo structure of TAZA results in a higher binding affinity for the human Aβ-plaques compared with PIB thus contributing to the higher binding of [11C]TAZA. The increased signal to noise ratio may enhance sensitivity of detection of change in PET studies. (2) Decreased white matter binding. A lower amount of white matter binding was seen with [11C]TAZA due to the heteroatoms in the “azo” functionality. This difference between gray matter and white matter may improve earlier delineation of AD, MCI and normal controls. (3) Enable transgenic animal imaging. Transgenic mouse models of AD have been shown to present strong, distinct pathologies of specific aspects of the human disease. Feasibility of imaging animal models with [11C]TAZA will be useful for the development of therapeutic drugs and understanding AD (e.g., Gaja et al., 2014; Snellman et al., 2013).
We like to thank Alisha Bajwa for technical assistance with the binding affinity studies. We like to thank Dr. Vitaly Vasilevko for technical assistance with immunostaining and the UCI ADRC for the human brain tissue. Conflict of interest: the authors have no conflict of interest in the work reported here. Role of authors: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: JM. Acquisition of data: MLP, MTM, HHP, BP, RM, CL. Analysis and interpretation of data: MLP, MTM, HHP, CC, RM. Drafting of the manuscript: MLP, JM, HHP, CC. Statistical analysis: JM, HHP, CC, RM. Obtained funding: JM. Study supervision: JM.
Contract grant sponsor: NIH/NIA AG 029479 (JM).