[125I]INFT: Synthesis and Evaluation of a New Imaging Agent for Tau Protein in Post-Mortem Human Alzheimer’s Disease Brain

Aggregation of Tau protein into paired helical filaments causing neurofibrillary tangles (NFT) is a neuropathological feature in Alzheimer’s disease (AD). This study aimed to develop and evaluate the effectiveness of a novel radioiodinated tracer, 4-[125I]iodo-3-(1H-pyrrolo[2,3-c]pyridine-1-yl)pyridine ([125I]INFT), for binding to Tau protein in postmortem human AD brain. Radiosynthesis of [125I]INFT was carried out using electrophilic destannylation by iodine-125 and purified chromatographically. Computational modeling of INFT binding on Tau fibril was compared with IPPI. In vitro, autoradiography studies were conducted with [125I]INFT for Tau in AD and cognitively normal (CN) brains. [125I]INFT was produced in >95% purity. Molecular modeling of INFT revealed comparable binding energies to IPPI at site-1 of the Tau fibril with an affinity of IC50 = 7.3 × 10−8 M. Binding of [125I]INFT correlated with the presence of Tau in the AD brain, confirmed by anti-Tau immunohistochemistry. The ratio of average grey matter (GM) [125I]INFT in AD versus CN was found to be 5.9, and AD GM/white matter (WM) = 2.5. Specifically bound [125I]INFT to Tau in AD brains was displaced by IPPI (>90%). Monoamine oxidase inhibitor deprenyl had no effect and clorgyline had little effect on [125I]INFT binding. [125I]INFT is a less lipophilic imaging agent for Tau in AD.


Introduction
The aggregation of Tau protein into paired helical filaments causing neurofibrillary tangles (NFT) is a neuropathological feature in Alzheimer's disease (AD) [1]. Efforts have been underway on developing and using Tau PET imaging agents since they can play an essential role in clinical studies to evaluate disease progression [2,3]. Pyrrole derivatives such as [ 18 [6] are being used for human Tau PET imaging in AD. Although off-target MAO-B binding has been raised as a concern for the pyrrole derivatives, in vivo, PET data are assumed to be free of this off-target binding [7]. The next generation of Tau PET radiotracer, [ 18 F]MK-6240 (Figure 1, 4) [8,9], based on azaindole structure without the MAO-B off-target binding concerns, is now being used in PET studies [10]. Recent PET studies show a significant amount of off-target binding of [ 18 F]MK-6240 in the meninges of AD subjects which may confound the measurement of cortical Tau [11].
Progress in the development of radioiodinated imaging agents for Tau has been relatively slow. Recently, a series of radioiodinated imidazo derivatives (e.g., Figure 2, 5, 6) have been developed, and in vitro, studies in postmortem AD brains have been reported [12,13]. The binding of these agents to Tau in the AD brain was prominent, and in vivo brain uptake in normal mice appeared suitable. Off-target binding to MAO-B of these radioiodinated derivatives has not been reported. We have previously reported azaindole derivatives [ 125 I]IPPI [14] and [ 124 I]IPPI [15] as radioiodinated analogs of [ 18   Progress in the development of radioiodinated imaging agents for Tau has been relatively slow. Recently, a series of radioiodinated imidazo derivatives (e.g., Figure 2, 5, 6) have been developed, and in vitro, studies in postmortem AD brains have been reported [12,13]. The binding of these agents to Tau in the AD brain was prominent, and in vivo brain uptake in normal mice appeared suitable. Off-target binding to MAO-B of these radioiodinated derivatives has not been reported. We have previously reported azaindole derivatives [ 125 I]IPPI [14] and [ 124 I]IPPI [15] as radioiodinated analogs of [ 18 F]MK-6240 for Tau imaging. Selective binding of [ 124/125 I]IPPI to Tau was observed in the anterior cingulate of postmortem human AD brains. The binding of [ 124/125 I]IPPI was quantitatively correlated with the Tau load measured by anti-Tau immunohistochemistry of the same subjects [15]. Off-target binding of [ 124/125 I]IPPI to MAO-A or MAO-B would not be a concern because of a lack of effect of clorgyline (MAO-A) or deprenyl (MAO-B) on [ 125 I]IPPI binding in the AD brains [14].
Our previous work with [ 124/125 I]IPPI confirmed the delineation of Tau in the human AD postmortem brain, and they appear to be sensitive to different levels of Tau. They may suggest their ability to differentiate the stage of the disease [14,15]. For in vivo preclinical evaluation, transgenic AD mice models play an essential role in evaluating proteinopathy imaging agents. Although differences may occur between the protein aggregates found in transgenic AD mice and AD humans, when found to be similar, they can play a significant role in imaging agent development and potential therapeutics evaluation. We have recently reported the successful evaluation of [ 124 I]IBETA binding to Aβ plaques in 5xFAD transgenic AD mice, both in vitro and by in vivo PET/CT [16]. This is a good example of using transgenic AD mice models for imaging agent development headed for translational studies to humans. Similar efforts for evaluating Tau imaging agents in AD transgenic mice have been pursued [17]. Although transgenic mice expressing Tau are available, [18]   For optimal in vivo imaging, a lipophilicity log P of approximately 2 is preferred. It enables optimal brain uptake across the blood-brain barrier and minimizes nonspecific binding by relatively faster clearance from non-target brain regions. These attributes can   Our previous work with [ 124/125 I]IPPI confirmed the delineation of Tau in the human AD postmortem brain, and they appear to be sensitive to different levels of Tau. They may suggest their ability to differentiate the stage of the disease [14,15]. For in vivo preclinical evaluation, transgenic AD mice models play an essential role in evaluating proteinopathy imaging agents. Although differences may occur between the protein aggregates found in transgenic AD mice and AD humans, when found to be similar, they can play a significant role in imaging agent development and potential therapeutics evaluation. We have recently reported the successful evaluation of [ 124 I]IBETA binding to Aβ plaques in 5xFAD transgenic AD mice, both in vitro and by in vivo PET/CT [16]. This is a good example of using transgenic AD mice models for imaging agent development headed for translational studies to humans. Similar efforts for evaluating Tau imaging agents in AD transgenic mice have been pursued [17]. Although transgenic mice expressing Tau are available, [18] the ability to use them for imaging agent evaluation of Tau agents currently in human use continues to evolve.
For optimal in vivo imaging, a lipophilicity log P of approximately 2 is preferred. It enables optimal brain uptake across the blood-brain barrier and minimizes nonspecific binding by relatively faster clearance from non-target brain regions. These attributes can provide higher target-to-nontarget ratios, thus improving the properties of the imaging agent. The reported Log D of [ 18 F]MK-6240, the clinically used PET imaging agent for Tau, is 3.32 (Table 1) [8,10]. This is higher than the optimal lipophilicity for in vivo imaging agents. The calculated lipophilicity of [ 125 I]IPPI is significantly higher (log P = 4.34), compared to [ 18 F]MK-6240 and may lead to high levels of nonspecific binding in vivo.  [14]. To what extent the isoquinoline ring in [ 18 F]MK-6240 and [ 125 I]IPPI helps avoid MAO binding remains to be seen. Changing the iodoisoquinoline ring to the iodopyridine ring would remove the benzene ring and thus reduce the log P significantly. However, the effect of this change on both the binding affinity to Tau and the lack of binding to MAO-A and MAO-B remains to be determined. Thus

Results
Synthesis of INFT 11 was carried out in a single step by reacting azaindole 9 with 2chloro-4-iodopyidine 10 ( Figure 3). The nucleophilic displacement reaction resulted in displacing either the iodine or chlorine, thus providing INFT 11 as well as ClNFT 12 in the product mixture. Additionally, some product as a result of azaindole displacing both iodine-and chlorine-in the same molecule was also observed (structure not shown). In modest yields, INFT and ClNFT were isolated as pure products by preparative chromatography. For preparing the fluorinated derivative, FNFT 13, nucleophilic displacement of iodine in INFT was carried out using tetrabutylammonium fluoride ( Figure 3).
We used our previously reported procedures of Chimera-AutoDock to assess the binding of INFT, ClNFT and FNFT ( Figure 4A) to the cryo-EM three-dimension structure of Tau fibril [14,15]. Energy-minimized molecular models of INFT, ClNFT and FNFT were made using Chem Draw 3D ( Figure 4B-D). In our previous findings with IPPI binding to the Tau fibril in AD, four binding sites were identified [14]. Comparing the binding energy values (Kcal/mol) at the four sites for IPPI, the lowest energies for INFT were found for Site 1. Docking studies with INFT revealed preferential binding at Site 1 ( Figure 4E) and was similar to the binding of IPPI to Site 1 shown in Figure 4F. The binding energies of INFT to Sites 2-4 were weaker when compared with Site 1 and weaker than the binding energies of IPPI for these sites. The lack of the second phenyl ring in INFT (compared to isoquinoline in IPPI) potentially reduces the hydrophobic interactions, thus weakening the binding energies. Compared to INFT, binding energies of ClNFT and FNFT were

Results
Synthesis of INFT 11 was carried out in a single step by reacting azaindole 9 with 2-chloro-4-iodopyidine 10 ( Figure 3). The nucleophilic displacement reaction resulted in displacing either the iodine or chlorine, thus providing INFT 11 as well as ClNFT 12 in the product mixture. Additionally, some product as a result of azaindole displacing both iodineand chlorine-in the same molecule was also observed (structure not shown). In modest yields, INFT and ClNFT were isolated as pure products by preparative chromatography. For preparing the fluorinated derivative, FNFT 13, nucleophilic displacement of iodine in INFT was carried out using tetrabutylammonium fluoride ( Figure 3).
We used our previously reported procedures of Chimera-AutoDock to assess the binding of INFT, ClNFT and FNFT ( Figure 4A) to the cryo-EM three-dimension structure of Tau fibril [14,15]. Energy-minimized molecular models of INFT, ClNFT and FNFT were made using Chem Draw 3D ( Figure 4B-D). In our previous findings with IPPI binding to the Tau fibril in AD, four binding sites were identified [14]. Comparing the binding energy values (Kcal/mol) at the four sites for IPPI, the lowest energies for INFT were found for Site 1. Docking studies with INFT revealed preferential binding at Site 1 ( Figure 4E) and was similar to the binding of IPPI to Site 1 shown in Figure 4F. The binding energies of INFT to Sites 2-4 were weaker when compared with Site 1 and weaker than the binding energies of IPPI for these sites. The lack of the second phenyl ring in INFT (compared to isoquinoline in IPPI) potentially reduces the hydrophobic interactions, thus weakening the binding energies. Compared to INFT, binding energies of ClNFT and FNFT were weaker at Site 1. The larger iodine atom compared to the smaller chlorine and fluorine ( Figure 4B-D) significantly reduces hydrophobicity and weakens the binding. Interestingly, the isoquinoline analog of FNFT, with the additional aromatic ring, has a significant affinity for Tau [9,14].
Sodium iodide [ 125 I]NaI (ARC Inc., St. Louis, MO, USA) was used to prepare the electrophilic substitution of the tributyltin derivative using our previously reported radioiodination methods. [16,19]. The same reaction was used to synthesize [ 125 I]INFT from the tributyltin derivative 14 (0.1 mg in 0.1 mL ethanol) and 3.4 MBq [ 125 I]NaI. The reaction was allowed to proceed at room temperature for 60 min before it was terminated by adding sodium bisulfite. weaker at Site 1. The larger iodine atom compared to the smaller chlorine and fluorine ( Figure 4B-D) significantly reduces hydrophobicity and weakens the binding. Interestingly, the isoquinoline analog of FNFT, with the additional aromatic ring, has a significant affinity for Tau [9,14]. Sodium iodide [ 125 I]NaI (ARC Inc., St. Louis, MO, USA) was used to prepare the electrophilic substitution of the tributyltin derivative using our previously reported radioiodination methods. [16,19]. The same reaction was used to synthesize [ 125 I]INFT from the tributyltin derivative 14 (0.1 mg in 0.1 mL ethanol) and 3.4 MBq [ 125 I]NaI. The reaction was allowed to proceed at room temperature for 60 min before it was terminated by adding sodium bisulfite.
The purification and isolation of [ 125 I]INFT were conducted on preparative TLC. Two rounds of extraction were performed using dichloromethane. The extract was then dried using anhydrous MgSO4. Radio TLC confirmed a radiochemical purity of >95% [ 125 I]INFT ( Figure 5). Using the molar activity of no-carrier added [ 125 I]sodium iodide, the molar activity of [ 125 I]INFT was estimated to be approximately 90 TBq/mmole under the no-carrier added conditions. No other major radiolabeled organic side products were observed successfully substituted with tributyltin substituent in 25% yield, sufficient for use in radiolabeling procedures.
Lipophilicity    Figure 4E,F) based on the Tau model study. Unlike INFT, IPPI also has a significant binding ability to additional Tau Sites (Sites 2,3,4; [14]), which may account for the difference in affinity when a [ 3 H]isopquinoline derivative was used for measuring the binding affinity of IPPI [9]. In vitro binding affinity of the unlabeled compounds was evaluated in AD brain slices labeled with [ 125 I]INFT. The anterior cingulate of the subjects were first evaluated for the presence of Tau using [ 125 I]IPPI, and as expected, all AD subjects showed the presence of Tau, as reported previously [14]. Assay conditions using [ 125 I]INFT were similar to our reported procedures using [ 125 I]IPPI. Different compounds (Table 1) Figure 4E,F) based on the Tau model study. Unlike INFT, IPPI also has a significant binding ability to additional Tau Sites (Sites 2,3,4; Binding of [ 125 I]INFT was evaluated in six AD subjects, 6 cognitively normal (CN) and 6 PD subjects. Figure 6A shows the binding of [ 125 I]INFT to the GM regions of one of the AD subjects (AD 11-78). Low levels of nonspecific binding were observed in the WM regions. The adjacent brain section of AD 11-78 was Immunostained for total Tau, shown in Figure 6B. Areas of Tau IHC in Figure 6B corresponded to [ 125 I]INFT binding in Figure 6A. A closer view of IHC in AD 11-78 in Figure 6C confirmed the presence of NFT. Brain slices from CN subjects (CN 12-21) are shown in Figure 6D and AD GM/PD GM = 5.14. The same AD subjects have been previously shown to exhibit significant amounts of [ 18F ]Flotaza binding to Aβ plaques [15,20].
dases have been reported [14], we performed competition experiments of [ I]INFT with (R)-deprenyl for potential binding to MAO-B and clorgyline for potential binding to MAO-A [21]. There was no decrease in the binding of [ 125 I]INFT in the presence of 10 μM (R)-deprenyl (GM/WM = 2.68; Figure 7C), suggesting a lack of MAO-B binding by [ 125 I]INFT. This is similar to our previous observations on the lack of any effect of (R)deprenyl on the binding of [ 125 I]IPPI to Tau [14]. In the presence of 10 μM clorgyline, the [ 125 I]INFT ratio was reduced, GM/WM = 1.88 ( Figure 7C) [14,15]. The availability of radioiodinated Tau imaging agents will enable in vitro evaluation of drugs [23,24].

Radiolabeled
Based on the structural features of the limited series of compounds, it may be surmised that the iodopyridine ring in INFT and the iodoisoquinoline ring of IPPI bind in a hydrophobic pocket. The larger iodine atom may be preferred because of the size of this pocket and enhanced hydrophobic interactions. Replacing the iodine with fluorine marginally lowered the affinity in the case of INFT, whereas replacing the iodine with fluorine   [14,15]. The availability of radioiodinated Tau imaging agents will enable in vitro evaluation of drugs [23,24].

Radiolabeled
Based on the structural features of the limited series of compounds, it may be surmised that the iodopyridine ring in INFT and the iodoisoquinoline ring of IPPI bind in a hydrophobic pocket. The larger iodine atom may be preferred because of the size of this pocket and enhanced hydrophobic interactions. Replacing the iodine with fluorine marginally lowered the affinity in the case of INFT, whereas replacing the iodine with fluorine in the case of IPPI had little effect. [14] Tau molecular models suggest that INFT may be more selective in binding to one particular site (Site 1, Figure 4) Radiolabeled fluorine-18 analog of FNFT may be prepared for evaluation using our previously reported methods of [ 18 F]-nucleophilic substitution on pyridine rings [25]. However, although FNFT has the lowest lipophicity in the series (clogP 2.05) and is similar to some of the optimal fluorinated imaging agents developed [26], it may not be a suitable in vivo Tau PET imaging agent due to its lower affinity for Tau. Other potential fluorinated analogs of INFT would be to replace the iodine with a trifluoromethyl group or a fluoroalkyl group. Fluorine-18 trifluoromethyl derivatives have been previously developed as PET imaging agents [27]. Fluorine-18 labeled fluoropropyl substitutions in pyridine derivatives have been used as successful PET radiotracers [28].

Molecular Modeling
Using ChemDraw (ChemOffice version 21.0), energy-minimized molecular structures of INFT and IPPI were saved as mol files. To assess the binding of INFT to tau, we used the UCSF Chimera molecular modeling program as described previously for IPPI studies [14]. The reported cryo-EM three-dimension (3D) structure of tau fibril was used to perform molecular prediction on the tau fibril. The 3D model of tau fibril consists of the paired helical filament, which is the principal component of neurofibrillary tangles in AD with the characteristic appearance generated by a "double-helical stack of morphological units, each with a C-shaped cross-section displaying three domains". This approach was previously used for the assessment of binding sites of IPPI. A radioiodination hood (CBS Scientific, Inc., Escondido, California, USA) placed inside a fume hood designated to handle radioactive materials was used to carry out iodine-125 radiolabeleling of 14 using our previously reported methods [14,16]. The crude mixture was purified on preparative TLC (CH 2 Cl 2 :CH 3 OH 9:1) and separated from the unreacted starting material.

Human Tissue
All postmortem human brain studies were approved by the Institutional Biosafety Committee of the University of California, Irvine. Human postmortem brain tissue samples were obtained from Banner Sun Health Research Institute, Sun City, AZ, brain tissue repository for in vitro experiments. All AD brain, Parkinson's disease and cognitively normal (CN) brain tissue samples were selected for end-stage pathology [1,15]. Human postmortem brain slices were obtained from chunks of frozen tissue on a Leica 1850 cryotome cooled to −20 • C.

In Vitro Postmortem Human Brain Autoradiography
Human anterior cingulate sections containing corpus callosum were sectioned from the subjects (AD, PD and CN). These sections were used to evaluate the effect of drugs on the binding of [ 125 I]INFT to Tau. Unlabeled IPPI (10 µM) was used to measure nonspecific binding. The slides containing the sections (10 µm thick) were preincubated in PBS buffer for 15 min in eight slide chambers (one total binding and seven with the different drugs). The preincubation PBS buffer was discarded, and the appropriate amount of each drug (dissolved in ethanol) was added to the chambers with the slides. Each chamber was added [ 125 I]INFT and 60 mL of 10% ethanolic PBS buffer for a final concentration of 3.7 kBq/mL of [ 125 I]INFT. The chambers were incubated at 25 • C for 1.25 h. The slides were then washed with cold PBS buffer, 50% ethanolic PBS buffer twice, PBS buffer and cold water for 5, 5, 5, 5, 3 min, respectively. The slides with the brain sections were air dried, exposed overnight on a phosphor film, and then placed on the Phosphor Autoradiographic Imaging System/Cyclone Storage Phosphor System (Packard Instruments Co., Waltham, MA, USA). Regions of interest (ROIs) were drawn on the slices and the extent of binding of [ 125 I]IPPI was measured in DLU/mm 2 using the OptiQuant acquisition and analysis program (Packard Instruments Co.).

Immunohistochemistry
University of California-Irvine, Pathology Services used Ventana BenchMark Ultra protocols for immunostaining of brain sections. To determine the localization of Tau in the human AD brain sections, neighboring slices were immunostained with the DAKO polyclonal antibody, which binds to total Tau which detects all 6 six isoforms of Tau (dilution 1: 3000, A0024; Agilent, CA, USA). Immunostained sections were scanned using the Ventana Roche instrumentation and the images were analyzed using QuPath software version 0.4.3.

Image Analysis
Statistical differences between groups (AD, CN and PD) were determined using Microsoft Excel 16. Statistical power was determined with Student's t test and p < 0.05 was considered to be significant.

Conclusions
In summary, a less lipophilic Tau imaging agent, [ 125 I]INFT has been developed, which is suitable for autoradiographic studies of postmortem human AD brains. Further studies are planned to evaluate this new agent's potential in vivo imaging value when labeled with iodine-124 for PET studies and iodine-123 for SPECT studies in transgenic mice expressing Tau. Possibility of using this less lipophilic azaindole backbone structure for potential fluorine-18 analogs for PET imaging of AD mice models of Tau will be investigated.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.