Hydroxyethyl-Modified Cycloheptatriene-BODIPY Derivatives as Specific Tau Imaging Probes

Abstract

Near-infrared fluorescence (NIRF) imaging as an exquisite sensitive, high spatial-resolution, and real-time tool plays an important role in visualizing pathologies in the brain. In this study, we designed and synthesized a series of NIR probes of hydroxyethyl cycloheptatriene-BODIPY derivatives that have demonstrated strong binding specificity to native neurofibrillary tangles (NFTs) in Alzheimer’s disease (AD) brain sections. The improved hydrophilicity of TNIR7–9 and TNIR7–11 resulted in faster clearance rates from healthy brains (4.2 and 10.9, respectively) compared to previously reported compounds. Furthermore, TNIR7–13, which features a fluorinated modification, exhibited a high binding affinity to Tau aggregates (K_d = 11.8 nM) and held promise for future PET studies.

Neurofibrillary tangles (NFTs), composed of hyperphosphorylated Tau protein, are considered a crucial neuropathological hallmark of Alzheimer’s disease (AD) and related neurodegenerative disorders, such as primary age-related tauopathy (PART), chronic traumatic encephalopathy (CTE), and Down syndrome (DS).¹⁻⁵ As a result, there is growing interest in developing specific probes for noninvasive visualizing of NFTs in these diseases. Numerous simulation studies pointed out that the binding affinities are primarily dependent on the formation of hydrogen bonds between the probes and the β-sheet structure of NFTs.⁶⁻⁸ However, it is worth noting that Tau fibrils are not the only protein aggregates with β-sheet architecture in the brain, especially considering the coexistence of Aβ plaques in AD brains. Therefore, a prerequisite for Tau probes is the ability to discriminate Tau fibrils from Aβ and other β-sheet-structured protein aggregates.⁹⁻¹¹

Recent progress in developing Tau probes has enabled the in vitro staging of Tau pathology and improved our understanding of the pathogenesis of diseases in small-animal models by near-infrared (NIR) fluorescence imaging.¹²⁻¹⁷ BODIPY-derived fluorophores with excellent fluorescence characteristics and flexible chemical modifications,^18,19 such as BD-Tau,²⁰Tau 1, and Tau 2,⁶ have contributed to Tau detection. However, the poor applications of these probes were caused by unfavorable optical properties (short fluorescence wavelength and slight Stokes shift) for NIR imaging or the lack of selectivity over Aβ plaques for specific detection in AD. Encouragingly, we recently reported two novel Tau probes,²¹TNIR7–1A and TNIR7–1B, with a fused heptacyclotriene-BODIPY scaffold that overcomes these primary limitations. Previous report revealed that these two probes exhibited significant fluorescence properties (λ_em > 650 nm and Stokes shift >100 nm) and selectivity to NFTs proven by fluorescence staining. Moreover, TNIR7–1A can effectively distinguish tauopathy in mice with NFTs (rTg 4510) from control mice using NIR fluorescence imaging in vivo. However, we note that the modest solubility of these probes limits their further biological use for Tau mappings, such as fluorescent staining on the tissue or living cell level.

To achieve these goals, we attempted to change the N,N-dimethylamino donor group with the N-hydroxyethylamino group to decrease lipophilicity and improve solubility while maintaining high affinity and optical properties,²² resulting in the development of a new generation of TNIR7s (TNIR7 7–12, Figure 1) in this study. The fluorescent and biological properties of these compounds were evaluated, with a particular focus on their biological applications including fluorescence imaging in okadaic acid (OA)-induced cells and healthy mice. Additionally, inspired by previous strategy of introducing fluorine atom, we explored the potential of introducing the fluoroethyl group onto the cycloheptatriene-BODIPY scaffold (TNIR7–13) for specific PET imaging of native Tau fibrils in AD brains.

Figure 1.

Chemical structure of hydroxyethyl-modified cycloheptatriene-BODIPY derivatives (TNIR7 7–12) developed in this study.

The synthesis of the hydroxyethyl-modified cycloheptatriene-BODIPY derivatives described in this study is presented in Supporting Information (SI), Scheme S1. The synthesis of compounds TNIR7 7–12 involved a 1,4-conjugated addition between the activated methyl group of BODIPY 1 and the β-carbon atom of unsaturated aldehydes with yields ranging from 1.4% to 4.2%. The details of synthesis, ¹H/¹³C NMR, and HRMS spectra are provided in the Supporting Information.

Based on previous studies,²¹TNIR7 compounds maintained their fluorescence properties resulting from intramolecular charge transfer (ICT) transitions. As shown in SI, Table S1 and Figures S3–S4, the emission maxima displayed the bathochromic shift with the extension of the polyene chains and an increase in solvent polarity. Notably, incorporating hydroxyethyl groups into the electron donor moiety had little effect on the fluorescence properties, as seen in TNIR7–1A and TNIR7–1B. The emission wavelengths of these compounds remained in the “NIR window” (706–820 nm in a PBS solution) and held a large Stokes shift with values ranging from 160 to 183 nm. The emission maxima of TNIR7–8 and TNIR7–10 were found to be lower than those with only one hydroxyethyl group, particularly in a polar solvent, indicating that the additional hydroxyl group reduced the electron delocalization effects. Furthermore, the TNIR7 series developed in this study also exhibited high quantum yields (QYs), ranging from 10.0% to 52.8% in dichloromethane (with values as 0.4% and 52.8%, respectively), suggesting that these compounds could achieve significant fluorescence enhancement when interacting with the β-sheet in Tau aggregates. Additionally, as depicted in SI, Figure S5, compounds TNIR7–9 and TNIR7–11 exhibited good photostability over 30 min of excitation light irradiation.

We next evaluated the affinity and selectivity of TNIR7s toward Tau-K18Δ280 and Aβ_1–42 aggregates in vitro by fluorescence measurements. As a fluorescence turn-on sensor based on the solvent dependency of ICT molecules, these compounds exhibited a marked fluorescence response, including the emission blue-shift and intensity enhancement in the presence of Tau-K18Δ280 aggregates (Table 1, and SI, Figures S6–S7). However, no significant responses were observed after incubation with bovine serum albumin (BSA). Additionally, the fluorescence of TNIR7s increased upon interacting with Aβ_1–42 aggregates, likely due to the aggregated Aβ assembling into a hydrophobic β-sheet structure, as observed in synthetic Tau aggregates. Encouragingly, in agreement with our previous study,²¹TNIR7–9 and TNIR7–11 with one trans double bond exhibited good selectivity for Tau aggregates over Aβ aggregates, with the SI values as 2.5 and 1.4, respectively.

Table 1. Affinity Data of Binding of TNIR7s with Tau-K18Δ280 and Aβ_1-42 Aggregates in Vitro.

	with Tau-K18Δ280 aggregates			with Aβ1–42 aggregates
probes	λem (nm)	folda	Kd (nM)	λem (nm)	folda	SIb
TNIR7–7	628	42.0	620 ± 177	604	119.8	0.4
TNIR7–8	624	16.3	>1000	609	96.2	0.2
TNIR7–9	716	333.3	105 ± 26	718	83.2	2.5
TNIR7–10	718	71.7	289 ± 121	705	184.7	0.4
TNIR7–11	644	278.4	92.6 ± 27.8	660	23.6	1.4
TNIR7–12	771	99.9	23.2 ± 5.9	752	542.4	0.2

^aFold = FI (probe with proteins)/FI (probe).

^bSelectivity index = FI (with Tau)/FI (with Aβ_1–42).

Having confirmed the selective fluorescent turn-on response of these probes to Tau aggregates, we quantitatively determined their binding affinities to Tau-K18Δ280 aggregates through the saturation binding assay (SI, Figure S8). As expected, TNIR7 9–12 with one or two C=C units maintained low K_d values ranging from 23.2 to 289 nM, with affinities increasing with the extension of the double bond. Notably, the introduction of the hydroxyethyl group reduced the binding affinities for Tau aggregates to a certain degree. For instance, TNIR7–8 (>1000 nM) < TNIR7–7 (620 nM) < TNIR7–0A (125.8 nM), TNIR7–10 (289 nM) < TNIR7–9 (105 nM) < TNIR7–1A (16.8 nM), and TNIR7–11 (92.6 nM) < TNIR7–1B (12.7 nM), suggesting that the additional hydroxyl might impede the hydrogen-bond formation between the hydrophobic pockets of protein aggregates and the probes.

With the satisfactory affinities and selectivity demonstrated for synthetic heparin-induced Tau polymers, we evaluated the specificity of TNIR7s for native Tau depositions using in vitro fluorescence staining on AD brain sections. Consistent with our previous study,²¹TNIR7–9, TNIR7–11, and TNIR7–12 revealed significant flame-shaped fluorescent spots that highlighted Tau tangles (Figure 2). These specific signals were colocalized with the AT8-positive and Gallays–Braak positive regions (SI Figures S9–S10). In contrast, we did not observe labeling of Aβ plaques by the three probes, and costaining with Aβ dyes DANIR 3b(23) and Amylo-Glo (a commercial dye which could label Aβ plaques and NFTs on the brain sections: ex, 334 nm; em, 438 nm) confirmed the presence of Aβ on the sections (Figure 2, and SI, Figure S11). These results provided additional evidence that the binding interaction with the synthetic polymers may lead to unreliable results. The fluorescence intensity profiles from TNIR7–9/TNIR7–11 and Amylo-Glo on NFTs showed a virtual overlap instead of Aβ plaques, which confirmed the selective fluorescence labeling behavior of TNIR7s for Tau pathologies.

Figure 2.

In vitro staining of TNIR7–9 and TNIR7–11 (shown in magenta in the first column) on brain sections from an advanced AD patient (95-year-old, female). The locations of Aβ plaques and NFTs were confirmed using the commercial dye Amylo-Glo (shown in blue, in the second column). The normalized intensity profiles of TNIR7s (magenta line) and Amylo-Glo (blue line) are shown in the last column, which indicated colocalization foci in NFTs but no overlap in Aβ plaques. Scale bar: 200 μm.

We previously demonstrated that TNIR7s could selectively bind to NFTs in AD brain sections. To investigate their ability to detect Tau in cells, we tested human neuroblastoma SH-SY5Y cells with okadaic acid, a proven cell culture model for simulating Tau phosphorylation.^6,24,25 As shown in SI, Figure S13, TNIR7–12 produced obvious fluorescence spots in the cytoplasm of the treated cells, indicating its ability to penetrate the cell membrane and bind to the phosphorylated Tau with a high contrast. Moreover, we performed in vitro ThT staining on the cells following TNIR7–12 incubation to identify induced-Tau pathologies. As expected, the merged images in SI, Figure S13 confirmed the colocalization of TNIR7–12 with ThT staining. These results clearly indicate that TNIR7s are capable of detecting Tau aggregates in cells.

Next, to assess the potential for in vivo applications, we first quantified the ability of TNIR7–9, TNIR7–11, and TNIR7–12 to penetrate the blood–brain barrier (BBB). As shown in SI, Table S2, TNIR7–9 and TNIR7–11 displayed satisfactory brain uptakes at 2 min postinjection, with 3.68% ID/g and 4.57% ID/g, respectively, after tail vein injection. However, both probes were rapidly washed out from the healthy brains with clearance ratios (brain_{2 min/60 min}) of 4.2 and 10.9, respectively. Notably, TNIR7–9 and TNIR7–11 exhibited slightly higher washout ratios and similar initial brain uptakes compared to the original compounds TNIR7–1A and TNIR7–1B, suggesting that decreasing the clogP values to the range of 2.0–3.5 was conducive to optimizing brain kinetics, although this strategy has some limitations. In contrast, TNIR7–12, which has one more trans double bond, displayed poor brain uptakes, with a peak value of only 0.87% ID/g at 10 min, likely due to its unsatisfactory lipophilicity (clogP = 3.97), as reported in our previous studies.²¹

To further validate their BBB penetrability, in vivo NIR imaging of TNIR7–9 and TNIR7–11 was conducted in male nude mice (n = 2) to evaluate their brain kinetics. Consistent with previously reported methods, the two probes were dissolved in a solution containing 20% DMSO and 80% 1,2-propanediol, and the injected dose was maintained at 0.5 mg/kg. As shown in Figure 3, intense fluorescence signals were observed at 2 min postinjection, with TNIR7–11 demonstrating an increased washout ratio from the brain (brain_{2 min/60 min} = 10.9). Thus, the introduction of the hydroxyethyl group effectively facilitated fast brain clearance, potentially expanding the use of these probes for feasible biological evaluations.

Figure 3.

In vivo NIR fluorescent images of TNIR7–9 and TNIR7–11 in nude mice (BABL/c, 5-week-old, male) at various representative time points and the quantitative analysis of fluorescence signals in the brain.

Based on the data obtained above, we proceeded to substitute the hydroxyethyl in TNIR7–9 with fluoroethyl, enabling ¹⁸F-radiolabeling via efficient nucleophilic substitution of the terminal tosyl group, thereby exploring the clinical potential of PET tracers (Figure 4A). As shown in Figure 4B, the results of in vitro saturation binding assays using synthetic protein aggregates demonstrated that TNIR7–13 retained a high binding affinity (K_d = 11.8 nM) and selectivity (SI = 2.2).

Figure 4.

Binding characteristics of TNIR7–13 to Tau. (A) Chemical structure of the fluorinated cycloheptatriene-BODIPY compound TNIR7–13. (B) The saturation binding curves of TNIR7–13 binding to Tau-K18Δ280 aggregates. (C) Double staining of the brain sections from an advanced AD patient (102-year-old, female) using TNIR7–13 and Amylo-Glo. The top row shows NFTs, including mature tangles (arrowhead) and ghost tangles (asterisk). The bottom row shows dystrophic neurites in the neuritic plaques (NPs).

In a manner similar to TNIR7–9, we evaluated the specific reactivity of TNIR7–13 with pathological Tau lesions using in vitro fluorescence labeling of AD brain sections. Encouragingly, we observed remarkable fluorescent signals upon binding to different types of Tau tangles, including the mature tangles (indicated by arrowhead) and ghost tangles (indicated by asterisks), as well as dystrophic neurites in neurtic plaques (NPs), demonstrated the strong binding of TNIR7–13 to Tau pathologies (Figure 4C). Furthermore, these intense fluorescent patterns of TNIR7–13 overlapped with the commercial dye Amylo-Glo. Conversely, TNIR7–13 did not exhibit significant fluorescence spots with the Aβ plaques on the AD brain section (95 years old, female, frontal lobe), unlike the fluorescent mapping observed with DANIR 3b(23) (SI, Figure S12). Based on these findings, we are confident that the fluorinated cycloheptatriene-BODIPY compound TNIR7–13 has potential for PET imaging of the Tau burden.

In conclusion, our study successfully developed a series of cycloheptatriene-BODIPY derivatives for the detection of native Tau tangles in the AD brain. TNIR7–9 and TNIR7–11 demonstrated high binding affinities toward Tau aggregates and specific visualization of Tau depositions on AD brain sections without responding to Aβ plaques. Compared to previously reported compounds, TNIR7–9 and TNIR7–11 exhibited faster brain clearance rates. TNIR7–13, modified with a fluorine atom, was able to detect different conformations of Tau lesions with high contrast on AD brain sections, suggesting its potential for PET imaging of Tau burden. Further, PET research based on TNIR7–13 will be reported in the future. Our findings provide a promising direction for the development of diagnostic and therapeutic strategies targeting Tau pathology in AD and other tauopathies.

Glossary

Abbreviations

AD

Alzheimer’s disease

Aβ

β-amyloid

BBB

blood–brain barrier

BODIPY

borondipyrromethene

BSA

bovine serum albumin

CTE

chronic traumatic encephalopathy

DS

Down syndrome

HRMS

high-resolution mass spectra

ICT

intramolecular charge transfer

NFTs

neurofibrillary tangles

NIR

near-infrared

NIRF

near-infrared fluorescence

NPs

neuritic plaques

OA

okadaic acid

PART

primary age-related tauopathy

QYs

quantum yields

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.3c00248.

• Experimental details, absorption and fluorescence spectra, saturation binding curves, in vitro fluorescent images, ¹H NMR, ¹³C NMR, and HRMS spectra (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Y.L. and C.T. contributed equally.

The authors declare no competing financial interest.