Rational design and synthesis of a novel BODIPY-based probe for selective imaging of tau tangles in human iPSC-derived cortical neurons

Abstract

Numerous studies have shown a strong correlation between the number of neurofibrillary tangles of the tau protein and Alzheimer's disease progression, making the quantitative detection of tau very promising from a clinical point of view. However, the lack of highly reliable fluorescent probes for selective imaging of tau neurofibrillary tangles is a major challenge due to sharing similar β-sheet motifs with homologous Amyloid-β fibrils. In the current work, we describe the rational design and the in silico evaluation of a small-size focused library of fluorescent probes, consisting of a BODIPY core (electron acceptor) featuring highly conjugated systems (electron donor) with a length in the range 13-19 Å at C3. Among the most promising probes in terms of binding mode, theoretical affinity and polarity, BT1 has been synthesized and tested in vitro onto human induced pluripotent stem cells derived neuronal cell cultures. The probe showed excellent photophysical properties and high selectivity allowing in vitro imaging of hyperphosphorylated tau protein filaments with minimal background noise. Our findings offer new insight into the structure-activity relationship of this class of tau selective fluorophores, paving the way for boosting tau tangle detection in patients possibly through retinal spectral scans.

Figure 1

Chemical structures of BODIPY based fluorophores. (A) Chemical structures of TAU1 and BT1. (created by ChemDraw version 20.1.0.112 (112); PerkinElmer Inc) (B) Molecular docking of BT1 against the crystallographic structure of the PHF6 fragment: side and diagonal view of the molecule in the protein tunnel. (PyMOL Software. The PyMOL Molecular Graphics System, Version 2.2.0 Schrödinger, LLC. URL: https://pymol.org/2/).

Figure 2

Two-step synthetic strategy for the preparation of BT1. (ChemDraw software version 20.1.0.112 (112); PerkinElmer Inc).

Figure 3

Fluorescence analysis of BT1. (A) Sample plot showing the emission spectra of synthesized BT1 (100 μM) in PBS buffer pH 7.4 and 1% DMSO according to the experimental conditions used for the immunostaining in human iPSC derived cortical neurons. The emission spectra were recorded at different λ_ex as indicated in the legend on the right. (B) Representative plot showing the time-course of BT1 (20 μM) (λ_ex = 520 nm) in the presence of unfibrillated K18 tau protein (70 μM) (t0) and fibrillated K18 upon induction with heparin (70 μM) after 1 h, 1 day, 2 days and 3 days at 37 °C. (C) representative graph reporting the time-course of fluorescence of BT1 (20 μM) (λ_ex = 520 nm) in the presence of unfibrillated BSA (20 μM) (t0) and fibrillated BSA upon heating at 62 °C after 1 h, 2 h, 4 h and 8 h. Note that only the presence of fibrillated TAU enhances BT1 fluorescence. (GIMP software "The GIMP Development Team, 2019. GIMP, URL: https://www.gimp.org).

Figure 4

Molecular and cellular characterization of human iPSC derived NGN2 cortical neurons. (A) Schematic representation of the doxycycline-based strategy used for inducing the tuned expression of human NGN2 transcript. (B) Representative transmitted image of human iPSC-derived cortical neurons at day 30 in vitro. Scale bar 200 µm. (C) Real-time qRT-PCR analysis of specific molecular markers expression in human iPSC-derived cortical neurons at different time points of differentiation (n = 3 differentiation batches). (D) Representative images of immunostaining for NeUN (red), and MAP2 (green) expression in human iPSC-derived cortical neurons at day 30. Nuclei are stained in blue. Scale bar: 100 µm. (E) Representative images of immunostaining for total tau protein (HT7, red), and TUJ1 (green) expression in human iPSC-derived cortical neu-rons at day 30. Nuclei are stained in blue. Scale bar: 100 µm. (ImageJ software bundled with Java 1.8.0_172 software; URL: https://imagej.nih.gov/ij/).

Figure 5

BT1 and TAU1 staining of hyperphosphorylated tau in iPSCs derived cortical neurons. (A) Top: Representative fluorescence images of control human iPSC-derived cortical neurons at 30 days in vitro incubated with BT1 (100 μM) for 30 min at 37 °C and then stained for AT8 (green) and T22 (magenta). Bottom: Representative fluorescence images of human iPSC-derived cortical neurons treated with okadaic acid (50 nM) for 2 h before incubation with BT1 (100 μM) for 30 min at 37 °C and relative staining for AT8 (green) and T22 (magenta). Images were acquired on an Olympus iX73 microscope equipped with an X-Light V3 spinning disc head using a 40 × magnification. Scale bar: 100 µm. (B) Top: Representative fluorescence images of control human iPSC-derived cortical neurons at 30 days in vitro incubated with TAU1 (100 μM) for 30 min at 37 °C and then stained for AT8 (green) and T22 (magenta). Bottom: Representative fluorescence images of human iPSC-derived cortical neurons treated with okadaic acid (50 nM) for 2 h before incubation with TAU1 (100 μM) for 30 min at 37 °C and relative staining for AT8 (green) and T22 (magenta). Images were acquired on an Olympus iX73 microscope equipped with an X-Light V3 spinning disc head using a 40 × magnification. Scale bar: 100 µm. (ImageJ software bundled with Java 1.8.0_172 software; URL: https://imagej.nih.gov/ij/).

Figure 6

BT1 binds to hyperphosphorylated and oligomeric tau in OA treated neurons. (A) Bar charts showing the fluorescence intensity quantification of (left) T22 signal (**p = 0.002, MW test; n = 53/3, fields of view/batches) and (right) AT8 signals (****p < 0.0001, MW test; n = 53/6/3, fields of view/batches) in control condition and after the treatment with okadaic acid (50 nM) for 2 h. (B) Left, Manders’s colocalization in-dex of T22 staining with TAU1 (green) and BT1 (orange) fluo-rescence signal in control condition (p = 0.48, t-test; n = 25/3, fields of view/batches) and after the treatment with okadaic acid (50 nM) for 2 h (*p < 0.017, MW test; n = 25/3, fileds of view/batches). Right, Manders’s colocalization index of AT8 staining with TAU1 (green) and BT1 (orange) fluorescence signal in control condition (p = 0.408, MW test; n = 25/3, fields of view/batches) and after the treatment with okadaic acid (50 nM) for 2 h (****p < 0.0001, MW test; n = 25/3, fileds of view/batches), as determined using the custom-made MATLAB code. (C) On the left, Pearson’s correlation index of T22 staining with TAU1 (green) and BT1 (orange) fluorescence signal in control condition (***p = 0.0008, t-test; n = 25/3, fields of view/batches) and after the treatment with okadaic acid (50 nM) for 2 h (p = 0.423, t-test; n = 25/3, fileds of view/batches). On the right, Pearson’s correlation index of AT8 staining with TAU1 (green) and BT1 (orange) fluorescence signal in control condition (p = 0.056, MW test; n = 25/3, fields of view/batches) and after the treatment with okadaic acid (50 nM) for 2 h (****p < 0.0001, MW test; n = 25/3, fileds of view/batches), as determined using the custom-made MATLAB code. (Matlab software, version 2021a; URL: https://it.mathworks.com/products/matlab.html?s_tid=hp_products_matlab).

Figure 7

Schematic representation of MATLAB based algorithm for image analysis. (A) The maximum z-projection is calculated for each z-stack, then the background level is automatically retrieved and subtracted. The neurite structure mask is de-fined and applied as a filter to exclude unwanted signals located not in neurite structures. (B) An iterative thresholding procedure is used to binarized the image. Starting from the threshold value Thr = 0.2 (20% of the maximum signal) at each iteration, the threshold limit is increased by 0.15 units. The entire threshold image collection is combined to get a well-resolved binarized image. (C) The colocalization of the probe (BT1) with the antibodies (T22 and AT8) is calculated. Two well-known colocalization methods are exploited: the scatter plots (orange and green), which allow visualizing the correlation measured by PC coefficient, and the merged binary images (red/yellow and green/yellow), which allow to visualize the co-occurrence measured by M1 coefficient. (Matlab software, version 2021a; URL: https://it.mathworks.com/products/matlab.html?s_tid=hp_products_matlab).