Preliminary mechanistic insights of a brain-penetrant microtubule imaging PET ligand in a tau-knockout mouse model

MPC-6827 hydrochloride (Cat. # 5231) was purchased from Tocris (Minneapolis, MN, USA) and Epothilone d (EpoD) from Cayman Chemicals (Ann Arbor, MI, USA). Tau46 mouse antibody (Cat. # 4019), 4019), β-Tubulin (Cat. # 2128), α-Tubulin (Cat. # 3873), Acetyl α-Tubulin (Cat. # 5335T), β-Tubulin (9F3) Rabbit mAb HRP Conjugate (cat # 5346) and α-Tubulin (DM1A) Mouse mAb HRP Conjugate (cat #12351), and HRP-conjugated secondary antibody (Cat. # 7076) were obtained from Cell Signaling Technology (Danvers, MA, USA). The MT/tubulin assay kit (Cat. # BK038) was obtained from Cytoskeleton, Inc. (Denver, CO, USA.) RIPA lysis buffer (Cat. # 20-188), protease inhibitor (Cat. # P8340), eosin Y (Cat. # P8340), and all the anhydrous reagents and solvents were purchased from Sigma-Aldrich (St. Louis, MO, USA). PierceTM BCA protein kit was obtained from Thermoscientific (Cat. # 23225). Modified Hematoxylin Solution (ab220365) was obtained from Abcam (Cambridge, UK). Exposer cassettes (BAS IP SR 2025) were purchased from Cytiva (Marlborough, MA, USA). Desmethyl MPC6827 (precursor for [¹¹C]MPC-6827 radiochemistry) was purchased from ABX biochemical compounds (Radeberg, Germany).

Female Balb/C mice (6 months old) for EpoD treatment studies were purchased from Jackson Laboratory (Bar Harbor, ME, USA). Female Mapt^0/0 (referred to as tau KO), and wild type (WT) mice (BG.129X1-Maptt^m1Hnd) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). All animal experiments were conducted under eIACUC approved protocols in compliance with the guidelines for the care and use of research animals established by Wake Forest Medical School Animal Studies Committee.

[¹¹C]MPC-6827 was produced following our reported procedures [11]. Briefly, desmethyl MPC-6827 (ABX) was bubbled with [¹¹C]MeI in a NaOH/DMF solvent in the GE-FXC radiochemistry module at 80 °C for 5 min, followed by semi-prep HPLC column purification, and C18 SepPak elution with 10% ethanol in saline (n > 80 productions). Chemical, and radiochemical purity and specific activity of radioactive aliquots were determined by QC-HPLC testings. Radiochemical synthesis, including [¹¹C]CO₂/MeI transfer, reaction, HPLC purification and radiotracer formulation was completed in 50 min.

Cell-binding assays in vitro were performed in SH-SY5Y cells with [¹¹C]MPC-6827 following our published protocols [20, 21]. SH-SY5Y cells are patient-derived neuroblastoma cells routinely used to study neurotransmitter- and receptor-based effects in brain cells [22‐24]. They (0.5 × 10⁶ cells/well) were treated with 1.0 µM of (a) stabilizing agents that increase MT polymerizations (paclitaxel, EpoD) [1, 25] or destabilizing agents increasing free tubulin content (vinblastine, mertasine) [26, 27] and 3 h later, [¹¹C]MPC-6827 (0.074 MBq/well) was added and incubated for 30 min at RT (n = 6/group). To demonstrate tracer specificity [28], a subgroup of cells (n = 4) was pre-treated with the non-radioactive MPC-6827 (1.0 µM), radiotracer was added 60 min later, and the sample was incubated for 30 min. All the cells were washed with PBS and lysed with 1 N NaOH. Finally, the lysate from each well was γ-counted (PerkinElmer, Waltham, MA, USA) and counts-per-minute (cpm) values were normalized to the amount of radioactivity added to each well. Protein concentration was estimated in a subgroup of cells (n = 3) using the established Pierce BCA Protein Assay Kit from ThermoFisher [28, 29]. Cpm values were then matched to protein concentration per well, and the data expressed as %ID/mg of protein in each well.

To further investigate the tubulin selectivity of [¹¹C]MPC-6827 in SH-SY5Y cells, tubulin assay was performed using the commercially available kit-based assay (Cytoskeleton Inc, USA). Briefly, SH-SY5Y cells (7 × 10⁶ cells/well) were pretreated with 1.0 µM of same MT stabilizing (EpoD, paclitaxel) and destabilizing agents (mertasine, vinblastine) for 3 h (n = 3/group). Radiotracer (0.074 MBq/well) was then added and incubated for 30 min at RT. Samples were loaded on to ultra-centrifuge following the kit-based assay parameters to separate stabilized/bound and destabilized/free tubulins. Stabilized tubulins from low-speed centrifugation (1000×g for 5 min at 37 °C) and destabilized tubulins from high speed centrifugation (100,000×g for 20 min at 37 °C) were collected separately and their associated radioactivity was measured using a γ counter (Perkin Elmer). Cpm values were then matched to protein concentration per well, and the data expressed as %ID/mg of protein in each well.

Small animal microPET/CT imaging [11] was performed in anesthetized tau-KO and WT mice (n = 8/group) at 6 months of age. All mice received a bolus injection of [¹¹C]MPC-6827 (18.1 ± 0.01 MBq) via the tail vein. For the microPET/CT scans, the head/brain area was centered in the field of view, and the list-mode data from the emission scans were reframed into a 0-60 min dynamic binning sequence [30]. Time-activity curves (TACs) were obtained, and radioactive uptake was expressed as standardized uptake values (SUVs) using the PMOD analysis software. Regions of interest (ROIs) were drawn for the whole brain and commonly AD-affected regions including cortex, striatum, hippocampus, and cerebellum, using Π-pmod Atlas software.

To validate the in vitro MT-stabilization cell uptake results, normal Balb/C mice (n = 6, 6 mo old) were treated with epothilone D (EpoD), a brain-penetrant, MT-stabilizing drug commonly used in tauopathies. [¹¹C]MPC-6827 brain PET/CT imaging was performed at baseline and after 3 days of acute treatment of EpoD treatment (3.0 mg/kg intraperitoneal injection). Whole-brain SUV_max was calculated and compared between the baseline and post-drug treatment scans.

Post-PET radiotracer tissue biodistribution studies were performed with [¹¹C]MPC-6827 in the KO and WT mice (n = 4/group). After dynamic brain imaging from 0 to 60 min, mice were euthanized, and samples of brain, blood, heart, lung, liver, spleen, pancreas, kidney, and muscle were harvested, weighed, and γ-counted using a Wallac Gamma Counter (PerkinElmer) [11, 30] with a standard dilution of the injectate. Percentages of the injected dose per gram of tissue (%I.D/g; mean ± SD) were calculated and decay-corrected and were summarized in Table 1.

Autoradiography studies in vitro were performed on postmortem frozen brain sections from tau KO and control mice (n = 6/group) following published protocols [30]. Briefly, the sagittal sections were mount on a glass slides (Super frost plus slides, Fisher Scientific, Waltham, MA), air-dried for 30 min and incubated in PBS (pH 7.4) for 10 min to remove any endogenous binding. Additionally, a sub-section of tau KO brains (n = 4) were self-blocked with non-radioactive MPC-6827 (10 µM) 1 h prior to radiotracer treatment. The slides were air-dried and ~ 0.5 MBq of [¹¹C]MPC-6827 in PBS was added to each slide and incubated for 30 min. The slides were then washed with PBS (3×) and water (1×) at 4 °C and quickly air-dried. Slides containing the brain tissues were exposed to radioluminographic imaging cassette BAS IP SR 2025 (GE Healthcare) for 12 h at − 20 °C and scanned with a GE Amersham Typhoon scanner (25 µm pixel size). Autoradiographs were analyzed using ImageQuant TL 8.2 and ROI were manually drawn (see Additional file 1: 3), and specific binding in each ROI was calculated and expressed as photo-stimulated luminescence signals per square millimeter (PSL/mm²).

To investigate differences in MT stability between tau KO and control WT mouse brains (n = 3/group), we performed two different assays: (a) a commercially available MT-based assay (Cytoskeleton, Inc.,) [31‐33] and (b) capillary electrophoresis immunoblotting experiments.

This commercially available kit separates large complexes of polymerized MTs attached to nuclei and Golgi bodies into bound or non-polymerized/free tubulins. Briefly, prefrontal brain sections from euthanized mice were immediately placed in MT stabilization buffer (kit-based preparation) using a battery operated handheld homogenizer at 37 °C. Tissue lysates were centrifuged at 1000 × g for 5 min to separate them to nuclei and Golgi bodies (low-speed pellet), then supernatant samples were re-centrifuged at high speed 100,000 × g for 1 h at 37 °C to separate bound and free tubulins. The high-speed pellet was diluted in a microtubule depolymerization buffer (kit-based preparation). After the bound and free tubulin fragments were isolated, they were all slowly loaded onto a 12% SDS-polyacrylamide gel (PAGE) as recommended by the kit. The samples were then transferred onto a nitrocellulose membrane and allowed to settle for an hour at room temperature (RT) in TBST (tris-buffered saline and polysorbate 20). The membranes were incubated in primary tubulin antibody (1:2000) at 4 °C overnight, then washed with TBST (3 × 15 min) at RT with agitation. Next, they were incubated with HRP-conjugated anti-sheep secondary antibody (1: 10,000) for an additional hour and washed again in TBST (3 × 15 min) at RT. Finally, the tubulin bands were visualized using an Enhanced Chemiluminescence kit, and images were captured using a Amersham Imager 600—IA600 instrument, using the Molecular ImagJ analysis software to quantify tubulin concentrations.

These assays were performed using Jess™ western analyses according to the manufacturer's protocol (ProteinSimple, BioTechne, Santa Clara, CA). This technology is more sensitive and quantitative than traditional western immunoblotting, generating highly reproducible electropherograms [34‐36]. In brief, lysates were diluted to 0.4 μg/µL in sample buffer, added to a master mix containing dithiothreitol (DTT) and a fluorescent molecular weight marker, then heated at 95 °C for 5 min. The chemiluminescent substrate, primary antibody, HRP-conjugated secondary antibody, NIR-conjugated secondary antibody, protein normalization reagent, blocking reagent, samples, and separation and stacking matrices were dispensed into 384-well total protein plates. Subsequent separation electrophoresis and immunodetection steps in the capillary system were fully automated. Simple western analysis was carried out at RT using instrument default settings. Primary antibodies used included rabbit anti-β-tubulin (rabbit monoclonal 9F3 [Cell Signaling Technology]; 1:50); mouse anti-α-tubulin (mouse monoclonal DM1A [Invitrogen, Rockford, IL, USA]; 1:50); rabbit anti-acetylated alpha tubulin (rabbit monoclonal [Cell Signaling Technology]; 1:50). All antibodies were diluted with milk-free antibody diluent (ProteinSimple). HRP-conjugated antibodies were used for mouse anti-α-tubulin and rabbit anti-β-tubulin, while NIR-conjugated antibody was used for acetyl-α-tubulin. Total α tubulin and acetylated-α-tubulin were superplexed in the same well. The digital image was analyzed with Compass software (ProteinSimple) and the protein densitometry was calculated by dividing the area under the curve of each protein of interest by the area under the curve of total protein within the capillary.

All statistical analyses for cell uptake, microPET imaging, biodistribution, and autoradiography data were performed using GraphPad prism (version 7.0) and reported as the average value ± standard deviation. Statistical analysis was performed using a two-tailed student’s t test, with *p ≤ 0.05 and α = 0.05 being considered significant.

[¹¹C]MPC-6827 radiochemistry was fully automated and optimized in GE FXC module (see Additional file 1: 1) with a ~ 40% yield (n > 80 productions), and high radiochemical purity (> 98%), and specific activity (~ 114 ± 10 GBq/µmol); decay was corrected to the end of synthesis (EOS).

Initial [¹¹C]MPC-6827 radioactive uptake experiments were conducted in SH-SY5Y cells, a human neuroblastoma cell line. Cell uptake assays demonstrated differential uptake dependent upon the class of MT agents used (Fig. 1a). The MT stabilizing agents, EpoD and paclitaxel decreased radioactive uptake by ~ 44 (± 2)% and ~ 49 (± 3)%, respectively (*p = 0.036). In contrast, MT destabilizing agents mertasine and vinblastine increased uptake by 63 (± 2)% (*p = 0.046) and ~ 78 (± 2)% (*p = 0.006), respectively. In self-blocking assays, radioactive uptake was ~ 70 (± 1)% (**p = 0.0011) lower after non-radioactive MPC-6827 was added. Cytoskeleton kit-based tubulin assay with [¹¹C]MPC-6827 showed > 90 (± 4)% (***p = 0.00014) increased radioactivity in destabilized/free tubulin fractions compared to the bound fractions (Fig. 1b). Also, MT stabilizers, EpoD and paclitaxel treatments decreased radioactivity by ~ 44 (± 1)% and ~ 36.8 (± 3)% (*p = 0.04) in free tubulin fractions and by ~ 62 (± 6)% and ~ 60 (± 2)% (*p = 0.05) in bound tubulin fractions, respectively, compared to the untreated fractions. While, MT destabilizers mertasine and vinblastine increased radioactivity by ~ 39 (± 3)% and ~ 43 (± 1)% (**p = 0.004) in free tubulin fractions and by ~ 66 (± 2)% and 84 (± 4)% in bound fractions, respectively, compared to untreated fractions (Fig. 1a, b). Collectively, the in vitro experiments indicated higher radiotracer uptake in conditions of MT instability.

To determine translatability of our in vitro findings, we performed complementary experiments in vivo. We focused on the blood brain barrier (BBB)-penetrant MT stabilizing agent, EpoD, that is well-published in the tauopathy literature [37‐39]. Consistent with the in vitro data, in vivo microPET imaging in female Balb/C (n = 6) also demonstrated ~ 23 (± 1) % lower brain uptake with EpoD treatment compared to baseline scans (Fig. 1c, d, *p = 0.036). Next to determine whether our radiotracer provided uptake differences with a genetic model of aberrant neuronal microtubule stability in the absence of tau pathology, including dysregulated axons, dendrites, and network functions [40], we performed microPET scans in tau KO mice and compared them to age-matched WT mice. Western blot assay confirmed tau protein expression between genotypes (see Additional file 1: 2). MicroPET/CT imaging in tau KO and WT mice [11] showed brain penetration and high retention of [¹¹C]MPC-6827 (Fig. 2a). However, SUV_max for whole-brain in tau KO was 33 (± 2)% higher than that measured in WT controls (Fig. 2b, 1.74 vs. 2.33; *p = 0.032). TACs also demonstrated higher initial uptake in tau KO than in age-matched controls (Fig. 2c). Both genotypes demonstrated similar radioactivity profiles, peaking at around 4 min and gradually washing out by 60 min.

Post-PET biodistribution results showed ~ 41% higher brain uptake in tau KO mice (%ID/g = 1.76 ± 0.31) compared to the WT controls (%ID/g = 1.03 ± 0.29, Table 1). More specifically, uptake in the cortical regions of tau KO brains was 37% higher than in controls (%ID/g = 1.11 ± 0.03 vs. 0.69 ± 0.07) with **p = 0.004.

Additionally autoradiography experiments in vitro showed significantly higher radioactivity in the whole brain (~ 46[± 4]%, **p = 0.0028), cortex (~ 38[± 2]%, *p = 0.04), hippocampus (~ 33[± 3]%, *p = 0.035), and midbrain regions (30[± 3]%, *p = 0.015) of tau KO brains compared to age-matched controls (Fig. 3). Blocking with MPC-6827 showed ~ 63–70% (**p = 0.0014) lower radiotracer uptake in whole brain, cortex, hippocampus, and mid-brain regions compared to the untreated tau KO samples.

To investigate the role of tubulin stability and subunit expression on differential brain uptake of [¹¹C]MPC-6827 between the two genotypes, brain tissues from both 6 month-old tau KO and WT mice (n = 3/group) were analyzed for total, bound or free tubulin using a cytoskeleton-based MT kit assay where protein expression was analyzed using traditional Western blotting (Fig. 4) and tubulin subunit expression was further evaluated using capillary electrophoresis (Fig. 5). An antibody that detects global tubulin protein expression (i.e., both α- and β-tubulin) indicated a modest, non-significant decrease in tau KO mice compared to WT mice; however, free and bound tubulin expression were the same between genotypes. Capillary electrophoresis of the total tubulin fraction using antibodies specific to α- and β-tubulin subunits revealed a significant decrease in β tubulin expression in the tau KO mice compared to controls (26% reduction, p = 0.0106, Fig. 5a–c). The α and acetylated-α-tubulin expression were statistically similar between the two genotypes, though both were modestly lower in tau KO mice (18% and 36.7%, respectively, Fig. 5d, e). These results demonstrate that in the absence of tau protein associated MT stability (i.e., tau KO), tubulin expression patterns significantly differ than in the presence of physiological tau expression.

Collectively all in vitro cell uptake, in vivo microPET/CT imaging, and ex vivo biodistribution studies, and autoradiography (n = 6/group) detected a power of > 80%, a standard deviation of 0.54 units, and α = 0.05, with *p ≤ 0.05 indicating statistical significance.