In vivo detection of tau fibrils and amyloid β aggregates with luminescent conjugated oligothiophenes and multiphoton microscopy

Mouse experiments were performed with the approval of the Massachusetts General Hospital Animal Care and in compliance with the National Institutes of Health guidelines for the use of experimental animals. Mice used included: 1) APPswe:PSEN1∆E9 double transgenic (Tg) mice (APP:PS1) (The Jackson laboratory, B6.Cg-Tg (APPswe, PSEN1dE9)85Dbo/Mmjax), expressing both the human APP gene carrying the Swedish mutation K594n/M595 L and the exon 9 deletion mutation in the PS1 gene. In this model, amyloid pathology (Aβ plaques and CAA) start to deposit around 5-months of age [7]. Six- to 11-month-old APP:PS1 Tg mice of either sex, along with age-matched Wt littermate controls were used; 2) 8–11 month-old rTg4510 Tg mice of either sex (The Jackson laboratory, Tg (Camk2a-tTA)1Mmay Fgf14Tg (tetO-MAPT*P301L)4510Kh strain that expresses the tetracycline-controlled transactivator protein (tTA) [8]), along with age-matched Wt littermate controls were used, and 3) APP:PS1-rTg4510 (FVBB6F1rTg4510(App/PSEN1)85) (APP:PS1-rTg4510 mice), which exhibits mixed pathology of amyloid β plaques and NFTs [9]. Mice were socially housed at 3–4 animals per cage with ad libitum access to food and water on a 12/12 h light/dark cycle with controlled conditions of temperature and humidity.

Cranial window surgery was performed as previously described [10] with minor modifications. Mice were anesthetized with 1.5% (vol/vol) isoflurane and placed in a stereotactic apparatus. A piece of skull (measuring 3 mm in diameter) over the left somatosensory cortex was removed, replaced with a 5 mm diameter glass coverslip and fixed with a mixture of Krazy Glue and dental cement [11]. Body temperature was maintained at 37C throughout the full procedure by using a heated pad. Mice were given buprenorphine (0.1 mg/kg) for 3 days following surgery, and were allowed to recover for at least 3 weeks before they were imaged.

HS-84 and HS-169 were synthesized as described previously [3] and diluted with PBS to a stock concentration of 5 mg/mL. To label amyloid pathology and NFTs in vivo, 150 nmol of LCO in 150 μl PBS [3] was intravenously delivered via retro-orbital injection [12]. A subset of APP:PS1 Tg mice were injected IP with Methoxy-X04 (363 nmol in 280 μl PBS) 24 h before the imaging session to produce high contrast images of Aβ plaques (emission 460–500 nm) [13].

To label NFTs in vivo with Thiazine Red in rTg4510 mice [14], dura matter was removed and 0.5 mg/ml Thiazine Red in PBS was topically applied onto the brain for 1 h, and then thoroughly washed with PBS. An 8 mm cranial window was implanted and sealed with a mixture of Krazy Glue and dental cement. HS-84 was injected 1 week before the experiment.

In vivo imaging was performed on anesthetized mice (1.5% isoflurane). Images of amyloid pathology, NFTs and dextran angiograms were obtained using an Olympus FluoView FV1000MPE multiphoton laser-scanning system mounted on an Olympus Bx61WI microscope and an Olympus 25x dipping objective (NA = 1.05). A Deep-See Mai Tai Ti:Sapphire mode-locked laser (Mai Tai; Spectra-physics) generated two-photon excitation at 800 nm, and three photomultiplier tubes (PMTs) (Hamamatsu) collected emitted light in the range of 380–480, 500–540 and 560–650 nm [15]. Settings and laser power remained unchanged throughout the different imaging sessions. Either Texas Red dextran or fluorescein dextran (70,000 Da MW; 12.5 mg/mL in PBS; Molecular Probes) was retro-orbitally injected before every imaging session to provide a fluorescent angiogram. The brain was imaged at depths of up to 200 μm from the surface of the brain. Three to eight cortical volumes (Z-series, 127 μm × 127 μm, 200–300 μm depth) were acquired per mouse, at a resolution of 512 × 512 pixels. Imaged volumes were randomly chosen. Images were exported and processed using the Fiji package of ImageJ (National Institutes of Health). Fluorescence intensity of each Aβ plaque, NFT or individual blood vessel was quantified using the ImageJ measure tool. Images presented in the figures are either single slices or maximum intensity image projections of the 3D volumes.

At the end of the last imaging session, mice were euthanized under CO₂, perfused with PBS, and brains were removed, flash frozen and sliced into 20 μm coronal sections on a cryostat (Leica). To validate labelling by HS-84 and HS-169, slices were exposed to either 0.005% Thioflavin S (ThioS) or 0.005% Thiazine Red in EtOH for 5 min and then washed with 80% EtOH followed by TBS. Slices were mounted with DAPI Vectashield and subjected to confocal imaging to address the colocalization of either HS-84/Thiazine Red or HS-169/ThioS. Additionally, sections were subjected to heat induced epitope retrieval in 10 mM citric acid, 0.05% Tween 20, pH 6.0, permeabilized with 0.5% triton X-100 and incubated with anti-tau antibodies against Alz50 (kindly provided by Dr. Peter Davies [16], 1:100) or 6E10 (anti-β-amyloid, 1–16 antibody, Biolegend, 1:100) overnight at 4 C. Appropriate secondary antibodies (goat anti-mouse IgM heavy chain secondary antibody, Alexa Fluor 568 conjugate or Alexa Fluor 647 conjugate respectively, Invitrogen, 1:400) were applied and incubated for 1 h at room temperature. Slices were mounted with DAPI Vectashield (Vector Laboratories) and subjected to fluorescence imaging. Images were recorded using the VS120 Virtual Slide Microscope system (Olympus). Ex vivo staining/co-staining with both LCOs was carried out in APP:PS1 and APP:PS1-rTg4510 free floating sections. Sections were stained for 5 min in 0.005% (w/v) HS-84 and/or HS-169 prepared in 50% EtOH in TBS, and then washed with 80% EtOH followed by TBS. Slices were mounted with Vectashield and subjected to confocal imaging.

First, we addressed whether the delivery of HS-84 (Fig. 1a) to the mouse brain would specifically label NFTs and whether its fluorescence could be detected with multiphoton microscopy. rTg4510 Tg mice were used as a model of tauopathy. This mouse model expresses mutant human tau (P301L mutation) directed to the forebrain, and develops NFT pathology in the cortex, resembling the human AD type of tauopathy [8]. Wt littermates were used as control. HS-84 was injected intravenously (retro-orbitally) 1 week prior to the imaging session in mice with a cranial window already implanted (Fig. 1b). Dextran Texas Red (70,000 Da MW) was systematically injected to create a fluorescent angiogram [17]. HS-84 was excited at 800 nm. Figure 1c shows representative images of NFTs selectively labeled with HS-84. No significant fluorescence could be detected in the Wt mice. To validate that HS-84 was labelling NFTs in the mouse brain in vivo, we co-labelled with Thiazine Red, a dye previously described to selectively stain NFTs [18], in the living brain. rTg4510 Tg mice previously injected with HS-84 were subjected to craniotomy and dura removal. Thiazine Red (0.5 mg/ml) was directly applied onto the brain and then the mouse was imaged with multiphoton microscopy. Thiazine Red was topically applied to the brain since it was found not to cross the BBB. Colocalization of both dyes supports the fact that HS-84 stains NFTs and that it can be imaged with intravital multiphoton microscopy (Fig. 1d). To further validate HS-84 labelling NFTs in the mouse brain, post-mortem staining with Thiazine Red and an antibody directed against a pathological form of tau (Alz50) was performed to probe colocalization (Fig. 1e and Additional file 1: Figure S1). To determine the optimal time window of imaging after HS-84 injection, we next performed longitudinal multiphoton imaging of the same volumes in the same mouse at 7 different time points within a 6 week period. rTg4510 mice with a cranial window previously implanted were injected with 150 nmol HS-84 and imaged at t = 0 (right after injection), and later times (Fig. 2a). The optimal imaging time (time with strongest fluorescent labelling) was determined to be 1 week after HS-84 injection in the rTg4510 Tg mouse (Fig. 2b). The calculated fluorescence intensity did not include the blood vessels. It was observed that a single injection was enough to follow the formation of new NFTs, since HS-84 remained in the circulation (Fig. 2c), and stained newly formed NFTs for at least 3 consecutive days (Fig. 2d), which were labeled within 24 h in this mouse model. We also attempted to image NFTs labelled with the LCO HS-169 following the same procedure. No cellular fluorescence could be detected with multiphoton microscopy after intravenous injection (Additional file 2: Figure S2), suggesting the lack of (or weak) in vivo binding of HS-169 to NFTs, although it has been reported that HS-169 labels NFTs ex-vivo [3]. To test the hypothesis of a weaker labeling of HS-169 to the NFTs, we stained tissue sections from APP:PS1-rTg4510 Tg mice [9], which exhibit both plaques and tangles. Staining was performed with either HS-84 or HS-169 at same concentration, washing times and settings in the microscope. We could observe that both LCOs bound easier to amyloid β plaques and therefore the fluorescence obtained from plaques was more intense (Additional file 2: Figure S2C). On the other hand, we needed to increase laser % and gain in order to observe NFTs, and even more when labeled with HS-169 (notice and increased background in Additional file 2: Figure S2C, saturated, right). This suggests that likely the concentration of HS-169 injected in vivo was not enough to observe the NFTs under two photon microscopy, whereas it was sufficient for HS-84. In order to avoid toxicity, we decided to not inject the LCO more concentrated.

Next, we evaluated whether LCOs labelled β-amyloid pathology in vivo and whether it could be detected with multiphoton microscopy. Whereas there is a wide amount of fluorescent dyes for staining Aβ pathology ex-vivo [19], few of them can be imaged in vivo with multiphoton microscopy, with most of them emitting in the blue or green regions of the spectrum. The most standardised dye to image Aβ core plaques and CAA with multiphoton microscopy is Methoxy-X04 [13], which can be excited at 800 nm. On the other hand, most of the reporters used for in vivo imaging, such as the calcium indicator GCaMP [20], are based on GFP, making it difficult to multiplex with different probes. Interestingly, HS-169 (Fig. 3a) has a redshifted emission [3], making it suitable for labelling Aβ plaques and CAA and multiplexing with popular in vivo GFP-based reporters. As a model of cerebral amyloidosis, we used APPswe:PS1dE9 (APP:PS1), which generates Aβ peptides and progressively accumulates Aβ deposits as soon as 4–6 months of age [7]. HS-169 was injected intravenously 1 day prior to the imaging session in APP:PS1 mice with a cranial window already implanted (Fig. 3b). Fluorescein dextran (70,000 Da MW) was systematically injected to create a fluorescent angiogram. Figure 3c shows representative multiphoton microscopy images of dense core plaques and CAA pathology labelled with HS-169. Both were very clearly stained with HS-169, however, diffuse plaques were not clearly stained with HS-169. Wt littermates showed the LCO in the vessels, but no amyloid-like pathology was present, as expected. Moreover, colocalization of HS-169 and Methoxy-X04 in the plaques was observed in vivo with multiphoton microscopy in the APP:PS1 mice (Fig. 3d). To confirm that HS-169 labelled amyloid pathology in vivo, post-mortem ThioS staining was performed to probe colocalization of both dyes (Fig. 3e). In addition, LCO HS-169 colocalized with Methoxy-X04 after ex-vivo staining, confirming the results obtaining in vivo (Fig. 3f). To determine the optimal time of imaging amyloid pathology after HS-164 injection, we performed longitudinal multiphoton imaging of the same volumes for 6 consecutive weeks. APP:PS1 Tg mice with a previously implanted cranial window were injected with 150 nmol HS-169 and imaged at t = 0 (right after injection), and subsequent time points (Fig. 4a). The optimal imaging time was determined to be 24-72 h after HS-169 injection in the APP:PS1 Tg mouse (Fig. 4b). HS-169 also remained in the circulation for this period (Fig. 4c). LCO HS-84 was also injected in the APP:PS1 to label Aβ plaques and CAA and imaged with multiphoton microscopy (Additional file 3: Figure S3). We determined that HS-84 also labelled both Aβ plaques and CAA (Additional file 3: Figure S3A). Ex-vivo colocalization of HS-84 with Thiazine Red confirmed the HS-84 staining in vivo. Interestingly, when both LCOs (HS-84 and HS-169) were simultaneously injected into an APP:PS1 mouse, they did not completely colocalize. HS-169 seemed to label dense core plaques more vividly, whereas HS-84 labelled diffuse plaques and CAA more intensively (Fig. 5). Thus, this labelling with the two different LCOs showed heterogeneity among plaques (insets in Fig. 5). This heterogeneity was also observed ex-vivo, as shown in APP:PS1 tissue sections after immunohistochemistry with 6E10 antibody, followed by co-staining with both LCOs used in the same concentration and equivalent washing times. Two different examples are shown in Fig. 5c. Surprisingly, 6E10 staining also differed between plaques. A different example is shown in Fig. 5d, where at least two different types of Aβ plaques can be observed in the same area from the same mouse. Note that HS-169 staining (red) is more prominent in the core of the Aβ plaque, confirming our results in vivo.