1. Introduction

Strokes in a deep brain area, such as basal ganglia, internal capsule, thalamus, and brain stem, are called small vessel occlusive disease, or lacunar stroke [1–5]. Lacunar stroke is an important subtype of ischemic stroke, accounting for more than 25% of cerebral stroke [6]. Besides, due to their strategic location, lacunar infarcts often cause serious neurological deficits [2,7]. Animal stroke models provide an essential tool for understanding the complex pathophysiology of stroke and testing new neuroprotective and recanalization strategies for therapies in preclinical settings [8]. While various animal models have been suggested to investigate the mechanism of focal cerebral ischemia and to develop its therapies [8]. It has been challenging to develop lacunar infarction models due to the technical difficulty in inducing targeted occlusion of small subcortical vessels [9]. Developing animal model for lacunar strokes, therefore, requires a highly selective and accurate local infarction induction technique that is applicable even in the deep brain.

Endovascular filament MCAO is most commonly used method of ischemic strokes [10]. However, the MCAO is not suitable for inducing subcortical ischemic stroke because it is performed by directly occluding the superficial MCA branches via craniotomy. Embolic occlusion also can be used to induce the smaller infarcts by directly injecting artificial microspheres less than 50 µm or autologous blood clots into internal carotid artery [11]. Unfortunately, these embolism occlusion models have failed to translate animal data into humans due to their variability and limitations in neuroprotective studies. In fact, the mechanical occlusion methods, including endovascular filament MCAO and embolism occlusion, failed to adequately reflect the hemodynamic aspects of thrombolytic reperfusion, which could potentially alter brain tissue responsiveness to neuroprotective treatments [8].

Endothelin-1 occlusion has been widely used to induce subcortical focal ischemic stroke [12–14]. It is possible to target a specific part of the deep brain through stereotaxic injection and simply controlling a size of the infarction in a dose-dependent manner as a potent vasoconstrictor [12]. However, the method may unwantedly affect several microvessels. Moreover, direct histological changes to vasoconstriction during its procedures have not yet been identified [9,13,14].

A promising method to induce focal ischemic stroke is photothrombosis. Photothrombosis is performed by irradiation a laser beams in a specific wavelength range to a certain brain region after systemic injection of a photosensitive dye such as Rose Bengal [15]. The main mechanism of photothrombosis is that dye-sensitized photooxygenation causes endothelial damage and then platelet adhesion and aggregation to form blood clots to block brain blood vessels. Photothrombosis can also control thrombus formation with an optical resolution, and with high reproducibility in a living animal [16]. In addition to photochemical embolization of cortical microvessels causing focal cortical infarction, laser beams can directly irradiate specific vessels to induce cerebral ischemia in the supply region [17]. However, most of the photothrombotic stroke studies published to date have used two-photon microscopy [18] or confocal microscopy [19,15], which are confined only to the cerebral cortex due to their limited light transmission to brain tissue. Thus, a development of photothrombosis based methodologies for accurate subcortical occlusion has been insignificant yet.

The purpose of this study is to apply minimally invasive fiber bundle endomicroscopy (FBEµ) to targeted photothrombosis. The FBEµ can be applied to create targeted subcortical occlusion through stereotactic insertion using a fiber bundle probe into specific deep brain regions in live mice. Our FBEµ also includes a patterned laser stimulation system and a real-time dual-color fluorescence imaging system. A patterned laser stimulation system can be used to selectively induce targeted photothrombosis. Another part of our FBEµ, a dual-color fluorescence imaging system, can also be used to simultaneously observe clot formation and blockage of blood flow during photochemical reactions. First, photothrombosis is performed by selecting blood vessels distributed in the barrel cortex located on the brain surface of living mice. Based on the results, we develop a subcortical small blood vessel occlusion by targeting blood vessels in the anterior pretectal nucleus of the thalamus. TTC staining and post hoc histology are performed on day one after occlusion to identify small vessel occlusion lesions in all groups.

2. Material and methods

2.1 System setup of a FBEµ

The principle and detailed configuration of the optical system can be found in our previous work [20–23]. In brief, real-time FBEµ can be divided into three parts, as shown in Fig. 1(A): fiber bundle probe, dual-color fluorescence imaging system, and 532-nm laser patterned stimulation system. The optical system used an optical fiber bundle with an image diameter of 320-µm (FIGH-10-350S, 1 m, Fujikura Europe Ltd.). The light transmission property between the two sides of the fiber bundle allows easy and accurate targeting of specific blood vessels in the deep brain area [20,23]. The FBEµ equipped with a digital delay pulse generator (505, Berkeley Nucleonics Corp.) enables simultaneous fluorescence imaging and optical stimulation without interference by controlling the number of pulses (single, double, or multiple), pulse width, pulse delay, and pulse interval of the CCD camera and 532-nm laser. As shown in Fig. 1(B), the CCD camera and 532-nm laser were synchronized and controlled by the digital delay pulse generator. The control signal timing was referenced to the delay zero point (T₀) of the internal trigger output from the digital delay pulse generator every 5 seconds. The CCD camera ran on every output pulse (T₁) of the digital delay pulse generator for 300 milliseconds and takes 121 frames per experiment. Additionally, the shutter in front of the broadband light source was controlled by the CCD’s exposure TTL signal output, which opened the shutter when the camera was exposed. The digital delay pulse generator output pulses for triggering the 532-nm laser (T₂) were maintained every 5 seconds with a delay of 500 milliseconds per pulse with 4500 milliseconds pulse width. After the first internal trigger (T₀), the laser pulse stopped after outputting a 120-pulse shot (T₂).

 

Fig. 1. Schematic of entire FBEµ including a control system for real-time fluorescence imaging and laser patterned photothrombosis. (A) System schematic diagram of FBEµ. It combined a fiber bundle probe, a dual-color fluorescence imaging system, and a 532-nm laser patterned system. (B) The CCD camera and 532-nm laser were synchronized and controlled by the digital delay pulse generator. Control signal timing was relative to the delay zero point (T₀) of the digital delay pulse generator internal trigger. The CCD camera ran on every output pulse (T₁) of the digital delay pulse generator. Also, the shutter in front of the broadband light source was controlled by the CCD’s exposure TTL signal output, which opened the shutter when the camera was exposed. The digital delay pulse generator output pulses (T₂) triggered a 532-nm laser. (C) Dual-color raw fluorescence image of the central indicator pattern obtained from FBEµ using green fluorescence reference slide. The dual-color image was the same sample fluorescence image just optically split using a dichroic mirror 2 into two emission ranges: 520 nm to 550 nm (top left, green fluorescence) and 573 nm to 613 nm (bottom right, red fluorescence). (D) Cropped, magnified, and merged images inside the white boxes in Figure C after interpolation to remove pixelation. The upper-left and lower-right box images are shown in green and red pseudo-colors, respectively. (E) Cropped and magnified green fluorescence image of an interpolated 532-nm laser circular pattern image. The top center represents the laser pattern diameter. Lens 1: f = 180 mm; Lens 2: f = 150 mm; Lens 3: f = 75 mm; Lens 4: f = 180 mm; Dichroic mirror 1: XF2041; Dichroic mirror 2: 660dclp; Ex filter: 479-25/585-25; Em. Filter 1: 525/50Em.; and Em. Filter 2: 593/40Em. Abbreviations: SLM: spatial light modulator and PBS: polarizing beam splitter. Scale bars, 50 µm.

Download Full Size | PDF

In this study, all photochemical stimulation was limited to 10 minutes because plasma leakage due to blood-brain barrier (BBB) breakage beyond thrombus occlusion was observed when excessive stimulation was applied for more than 10 minutes. Also, the dual-color fluorescence imaging was performed in two wavelength ranges of 520-550 nm and 573-631 nm, as shown in Fig. 1(C). This allows for monitoring thrombus formation, blood flow, and blood vessel morphology that change concurrently with photothrombosis. All images and visualizations after Fig. 1 were displayed as one composited image frame by cropping and merging two spatially separated fluorescence image frames, as shown in Fig. 1(D).

In addition, laser pattern stimulation from 2 µm to 320 µm in diameter with a light intensity of 125 mW/mm² can be performed using a reflective spatial light modulator (SLM, LC-R720, Holoeye) and a 532-nm diode-pumped solid-state (DPSS) laser (MGL-III-532-300 mW, CNI Optoelectronics Tech. Co.). In this study, full-field photothrombosis and sham experiments were performed using a 320-µm diameter circular pattern, and targeted photothrombosis was performed using a 100-µm and a 50-µm diameter circular patterns, as shown in Fig. 1(E). The 320-µm pattern covers the entire field of view of the image, and the circular laser patterns with 100-µm and 50-µm diameters sufficiently cover the diameter of the target vessel.

2.2 Brain surgery and Rose Bengal photothrombosis

All animal experiments were approved by the Institutional Animal Care and Use Committee of the Korea Institute of Science and Technology (KIST-5088-2022-01-007). 10-12 weeks old C57BL/6NTac male mice (approximately 25-27 grams) prepared from DBL (Republic of Korea). The mice were anesthetized by the mixture consisting of 2,2,2-tribromoethanol (250 mg/kg, #T48402, Sigma–Aldrich) and xylazine (16 mg/kg, Unixylazine, UNIBIOTECH) through an intraperitoneal (IP) injection. To maintain a stable anesthesia state, the mice were regularly treated with 100 µL of the mixture per hour. The mixture consisting of 100 µL of Rose Bengal (160 mg/kg, #330000, Sigma–Aldrich), 100 µL of Rhodamine-B dextran (180 mg/kg, #R9379, Sigma–Aldrich), and 50 µL of DyLight488-conjugated anti-GPIbβ antibody (0.2 mg/kg, #X488, Emfret Analytics) was administrated to mice via intravascular (IV) tail injection after 30 minutes’ post-anesthesia. Each substance was responsible for thrombosis induction, plasma staining, and activated platelet staining, respectively. Cranitomical surgery was conducted after head fixation based on a stereotaxic atlas (#51730, Stoelting Co.). The coordinates of the craniotomy were as follows: -1.5 mm for anterior-posterior (AP) and 3.5 mm for medial-lateral (ML) in the barrel cortical occlusion model, and -2.0 mm for AP and 1.0 mm for ML and -2.7 mm for dorsal-ventral (DV) in the deep brain occlusion model. These stereotaxic coordinates were established by Franklin and Paxinos [24]. A 1.5 mm × 1.5 mm cranial window was exposed around the above coordinates using drilling. For the cortical occlusion, the fiber bundle was in contact with target blood vessels on the brain surface distributed within the window. On the other hand, in the subcortical occlusion, the fiber bundle was carefully inserted into the deep brain at a speed of about 4.6 µm/s according to the real-time image and placed in the target blood vessels. After inducing a photothrombotic occlusion for 10 minutes, the fiber bundle is slowly removed. After suturing the scalp and applying povidone-iodine, enrofloxacin (10 mg/kg, Baytril), and carprofen (2.5 mg/kg, Rimadyl) were administered to mice through IP injection.

2.3 Assessment of histologic damage

Infarcts in the brain were evaluated using 2,3,5-triphenyltetrazolium chloride (TTC) staining and post-hoc histology after 1-day post-occlusion. Especially, post-hoc histology was utilized to identify vessels and thrombus formation after fiber bundle insertion injury and photothrombotic occlusion. Briefly, the mice were anesthetized by the mixture consisting of 2,2,2-tribromoethanol (250 mg/kg, #T48402, Sigma–Aldrich) and xylazine (16 mg/kg, Unixylazine) through the IP injection to acquire the whole brain. Anesthetized mice were humanely sacrificed, and whole brains were obtained. The brain slices sectioned into 1-mm-thick were stained with the 1% TTC solution (500 mg/mL, #T8877, Sigma-Aldrich) for 30 minutes at 37 °C. The images of samples were acquired using the microscope (Nikon, SMZ18), and quantification of infarcts in the ipsilateral hemisphere was performed using NIH ImageJ software. In Fig. 3(A), infarct lesions were selected based on the same intensity threshold to select the infarct area under the same conditions in all subjects. Threshold criteria: full range of image histogram (excluding background), 0 to 255; whole brain, 90∼255; and infarct lesions, 180∼255.

For post-hoc histology, the TTC-stained brain sections were fixed in paraformaldehyde (PFA) 4% overnight. Acquisition of standardized images was performed with a microscope (Eclipse Ti-U, Nikon Corp.) equipped with 4 × (CFI Plan Fluor 4x, Nikon Corp.) and 10 × (CFI Plan Fluor 10x, Nikon Corp.) objective lenses and an XY scanning stage (MLS203, Thorlabs, Inc.). Images were captured with a Hamamatsu electron multiplying charge-coupled device (EM-CCD) digital camera (C9100-23B, Hamamatsu Photonics K.K.) using HCImage Live image-processing software (Hamamatsu Photonics K.K.). Mosaic images were captured and created using an EVOS M7000 microscope imaging system (Thermo Fisher Scientific).

2.4 Laser speckle contrast imaging (LSCI)

Cerebral blood flow (CBF) was monitored using a laser speckle blood flow imager (Omegazone, OZ-2, Japan). CBF of 200 frames was measured for 1 minute in the 1.6 mm × 1.6 mm region of interest (ROI) in both hemispheres before and 1-day post-occlusion, as shown in Fig. S1, CBF was quantified as the percentage for the contralateral hemisphere.

2.5 Image post-processing

Post-image processing was performed to remove the fiber bundle pattern and improve the image resolution for diagnosis or analysis. In this paper, we used the interpolation method using the local thresholding binarization method of our previous study [25]. Briefly, image post-processing was divided into three stages: binarization, peak detection, and interpolation. First, binarization was used to separate the core and cladding from the fiber bundle endomicroscopic image. As shown in Fig. 1(C), in raw images, the core tended to be brighter than the surrounding cladding because it was the path that carried the intensity information of the ROI. The core and cladding were separated using binarization, and only the pixels selected as the core went through the peak detection process. In this paper, we used local thresholding proposed by Bernsen [26] for binarization. Then, the peak points were obtained using an algorithm based on the work of Elter et al. [27]. Furthermore, the MATLAB “findPeaks” function was used to take only peaks above a specific value. Finally, the fiber bundle endomicrosopic image was reconstructed using linear interpolation (MATLAB “scatteredInterpolant” function) based on the peak point map obtained in the peak detection step [28]. As shown in Fig. 1(D), the image visibility was improved by removing the fiber bundle pattern through interpolation. The image process was carried out using MATLAB R2019b (MathWorks, USA) on a PC with the following configuration: Intel CoreTM i7-1065G7CPU @ 1.30 GHz, 16GB RAM, 64-bit OS (Window 10).

2.6 Data analysis

Image analysis was performed using Image-Pro Premier 9.1 (Media Cybernetics, Inc.) and ImageJ 1.53q (Wayne Rasband, US National Institutes of Health). In addition, Image-Pro Premier 9.3 was used to coat a pseudo-color, merge images, animate the frame rate, and combine image files into a mosaic. Adobe Photoshop CS6 was used to insert scale bars and edit the videos. Statistical tests were performed with Microsoft Excel (Microsoft) and Origin 2020 (OriginLab Corp.).

3. Results

3.1 In vivo real-time imaging of activated platelets and plasma during thrombogenesis by full-field and targeted photothrombosis in barrel cortex

We first performed photothrombosis in the barrel cortex located in the superficial layer of the brain of living mice. As shown in Fig. 2(A and B), the sham and full-field photothrombosis were performed using a 320-µm circular laser pattern in Fig. 1(E). As shown in Fig. 2(A), the same experimental protocol as for full-field photothrombosis was applied in the sham. However, no activated platelet accumulation due to thrombus formation was observed even after 10 minutes of 532-nm laser irradiation, and plasma flow was maintained (see Visualization 1).

 

Fig. 2. Time-lapse images of in vivo real-time imaging during cortical photothrombotic occlusion. This figure selectively shows image frames at 5 s, 25 s, 50 s, 100 s, 200 s, 400 s, and 600 s out of a total of 121 frames (n = 5). See Visualization 1 and Visualization 5 for full images of each experiment. The upper left inset shows photothrombosis in the barrel cortex of a living mouse brain. (A-E) Dual-color imaging was performed after tail-vein injection of a mixture of Rose Bengal with Rhodamine-B dextran and DyLight 488-conjugated activated platelet antibody in all mice except for sham mice, which were replaced with normal saline instead of Rose Bengal. The fluorescence images were collected in both the 520∼550 nm (green pseudo-color) and 573∼631 nm (red pseudo-color) channels. The green pseudo-color represents activated platelets that form thrombi generated by photochemical stimulation inside blood vessels, and the red pseudo-color represents plasma flowing through the blood vessels. The direction of blood flow is indicated by a yellow arrow and is not indicated from the point when a blood clot blocks the blood vessel. The blue dotted circles indicate the 532-nm laser pattern used for each experimental group. The diameter of each vessel visible in the overall image is indicated by a white number and was measured from the corresponding white dotted line. (A and B) Full-field photothrombosis and sham experiments were performed using a 320-µm diameter laser circular pattern. (C-E) Targeted photothrombosis was performed using 100-µm and 50-µm diameter laser circular patterns. The 320-µm pattern covers the entire field of view in the image. Depending on the target vessel location, the 100-µm and 50-µm diameter laser circular pattern positions can be freely moved and positioned within the image area. Each experiment was stopped when it was determined that the blood clot was sufficiently hardened in the blood vessel to block blood flow. All experiments in this study were limited to 10 minutes and experimentally obtained time was judged sufficient to complete occlusion after thrombus formation under our experimental conditions. Scale bar, 50 µm.

Download Full Size | PDF

On the other hand, in full-field photothrombosis, precipitation of activated platelets was observed as soon as photochemical stimulation began, and the green fluorescence images after 25 seconds show that activated platelet accumulation was initiated at the laser irradiation site corresponding to the entire image area, as shown in Fig. 2(B) and Visualization 2. After about 100 seconds, the green fluorescence images show that activated platelets accumulate throughout the irradiated brain tissue and spread in the direction of blood flow in the venous vessel with a diameter of 78 µm, completely blocking blood flow.

Next, we performed targeted photothrombosis to select a single blood vessel distributed in the barrel cortex of living mice using 100-µm and 50-µm circular laser patterns, as shown in Fig. 1(E). Before performing targeted photothrombosis, the target blood vessel and photostimulation site were determined through brain fluorescence images of each experimental mouse, and a pattern that could cover the corresponding diameter was selected and placed, as shown in Fig. 2(C−E). In the targeted #1 mouse, an arterial blood vessel with a diameter of 68 µm was targeted, and this blood vessel branched into two blood vessels of 23 µm and 68 µm, as shown in Fig. 2(C).

The real-time dual-color fluorescence imaging confirmed that a thrombus was formed at the point where the photothrombosis started and that the thrombus spread according to the direction of blood flow. After 150 seconds, a red fluorescently labeled plasma defect at the site of thrombus accumulation indicates blockage of blood flow (see Visualization 3). As shown in Fig. 2(D), the targeted #2 mouse selectively photostimulated the lower right blood vessel among two adjacent blood vessels using a 50-µm pattern. The fluorescence image shows that a thrombus was formed only in the photostimulated blood vessel without affecting adjacent blood vessels, and blood flow was completely blocked for almost 450 seconds (see Visualization 4). Finally, the targeted #3 mouse underwent targeted photothrombosis targeting relatively small blood vessel, as shown in Fig. 2(E). The target vessel was one of the capillaries integrated into a 30 µm venous vessel with a diameter of 21 µm. The photothrombosis proceeded with a laser pattern with a diameter of 100-µm, and it was confirmed that a thrombus was formed inside the target blood vessel in about 350 seconds, and the flow of the blood vessel was stopped (see Visualization 5).

3.2 Measurement of ischemic occlusion lesions day one after cortical photothrombotic occlusion

To determine whether photothrombosis ultimately leads to ischemic occlusion in full-field photothrombosis, LSCI was performed in mice before and day one after occlusion. Fig. S1(A-C) show that ipsilateral infarction due to ischemic stroke was identified only in full-field photothrombosis. However, LSCI was not confirmed in the subsequent experimental group because it was limited to the brain surface and lacked the resolution to identify even tiny blood vessels.

TTC staining and post hoc histology were performed on day one after photothrombotic occlusion to measure ischemic occlusion lesions in all experimental groups. Experimental data in Fig. 3 is obtained from the same object as Fig. 2. As shown in Fig. 3(A), the red dotted line indicated that photothrombosis was performed at −1.5 mm anterior to bregma and 3.5 mm lateral to the right hemisphere. Also, the infarct lesion and the whole brain were divided into white and red pseudo-colors in the TTC-stained image for more precise identification. After full-field photothrombosis, the most prominent infarct lesions appeared in the brain slices corresponding to -1.0 mm and -2.0 mm from the bregma, while other infarct lesions were distributed throughout the brain. Because the barrel cortex has many pial arteries and penetrating arteries that affect the whole brain [29], if a thrombus blocks these arterial vessels, the penetrating arteries that branch from these arteries are also blocked. For this reason, it is presumed that the infarcted lesions appeared wider than the laser irradiation site.

When a single blood vessel is targeted and occluded, as in the targeted photothrombosis of targeted #3 mouse, it was confirmed that a more localized infarct site occurred due to blockage of a downstream branching vessel localized to the target vessel. As shown in Fig. 3(A), TTC-stained brain slices could identify infarct lesions limited to brain slice points of -1.0 mm and -2.0 mm corresponding to target locations, in contrast to full-field photothrombosis. This means that a focal ischemic infarction was successfully created. On the other hand, in the sham, it was challenging to identify the region judged to be an infarct lesion.

 

Fig. 3. 2,3,5-triphenyltetrazolium chloride (TTC) staining and post hoc histology for evaluation of histological damage day one after cortical photothrombotic occlusion (n = 3). (A) Representative lesion reconstructions from each infarct group superimposed on the normal area of the coronal section. The upper left inset shows the somatosensory barrel cortex (S1BF), a photothrombotic target area in the anatomy of the mouse brain, schematically depicted in the coronal plane. Subjects were sham, full-field photothrombosis, and targeted photothrombosis. Sham and full-field photothrombosis were performed using a 320-µm diameter laser circular pattern, and targeted photothrombosis was performed using a 100-µm diameter laser circular pattern. TTC-stained brain slices of targeted photothrombosis were those of targeted #3 mouse in Fig. 2. The lesion reconstructions of the brain coronal section image were processed according to the threshold method to identify the infarct area manually. Normal and infarct areas are color-coded and marked in red and white, respectively. The numbers on the lower left are the anterior to posterior coordinates relative to bregma. The red dotted line represents the anterior coordinates for the bregma of the photothrombotic target vessel location. The anterior coordinates are -1.5 mm. Scale bar, 3 mm. (B, D, and F) Three representative TTC-stained brain sections show lesions (white tissue) from mice that received sham, full-field photothrombosis, and targeted photothrombosis. The section was about -1.0 mm from bregma. Scale bars, 1 mm. (F) TTC-stained brain slices of targeted photothrombosis were those of targeted #1 mouse in Fig. 2. (C, E, and G) Post hoc histological fluorescence images are magnified images of the red dotted boxes in the corresponding Figure B, D, and F. Activated platelets were detected with DyLight488-conjugated anti-GPIbβ antibody, and plasma was detected using rhodamine-B dextran. Platelets are displayed in a green pseudo-color, and plasma is displayed in a red pseudo-color. Scale bars, 500 µm. (H) Three enlarged images of the red dotted box in the ipsilateral image of the corresponding Figure G. Scale bar, 50 µm.

Download Full Size | PDF

Then post hoc histological fluorescence images were measured after TTC staining, as shown in Fig. 3. In all experimental groups, fluorescence images were measured in brain slices corresponding to the -1.0 mm position of bregma in Fig. 3(A). As shown in Fig. 3(B, C), the fluorescence image of the sham mouse did not observe a significant difference with the ipsilateral side compared to the contralateral side. On the other hand, brain slices that underwent full-field photothrombosis show that high fluorescence signals were observed in both red and green in the infarct region, suggesting thrombus and plasma leakage due to BBB breakage following vascular occlusion, as shown in Fig. 3(D, E). In the brain slice that underwent targeted photothrombosis of the targeted #1 mouse, plasma leakage following BBB disruption and thrombus accumulation in the infarct lesion were more clearly observed with limited red and green fluorescence within the occluded vessels as shown in Fig. 3(G and H).

3.3 Deep brain insertion imaging and targeted photothrombosis using Real-time FBEµ in living mouse for subcortical small vessel occlusion

Based on the results of cortical small vessel occlusion, we developed a subcortical small vessel occlusion model by targeting blood vessels in the anterior pretectal nucleus of the thalamus. To accurately target blood vessels, real-time fluorescence imaging was performed while the fiber bundle probe was inserted at a speed of approximately $4.6{\;\ \mathrm{\mu} \mathrm{m}}/\textrm{sec}.$, as shown in Fig. 4(A). If the vessel is ruptured due to excessive insertion, the red fluorescence intensity will increase rapidly due to the outflow of Rhodamine-B stained plasma [20]. This is undesirable because there is a risk of future hemorrhagic stroke as well as leakage of plasma over the entire imaging area, making imaging impossible. Therefore, we inserted very slowly and stopped fiber insertion as soon as we found the target vessel to prevent further brain damage and induction of hemorrhagic stroke.

 

Fig. 4. Time-lapse images of in vivo real-time imaging during deep brain photothrombotic occlusion. The upper left inset shows the coordinates of the deep brain photothrombotic target region in vivo in the anatomy of the mouse brain schematically depicted in the coronal plane. The bottom right is the bregma coordinates (n = 1). (A) Image frames in the first and second rows were achieved while targeting deep cerebral vessels by inserting a fiber bundle probe into a living mouse brain. The fiber bundle probe insertion rate was 4.6 µm/s. 5 s, 25 s, 50 s, 100 s, 150 s, 200 s, 250 s, 300 s, 350 s, 400 s, 450 s, 500 s, 550 s, and 590 s video frames were selected out of a total of 119 frames (see Visualization 6). (B) The third row is an image frame obtained while performing targeted photothrombosis in vivo using a laser circular pattern with a diameter of 100-µm to select specific blood vessels in the deep brain. Selectively display 5 s, 25 s, 50 s, 100 s, 200 s, 400 s, and 600 s image frames out of a total of 121 frames (see Visualization 7). The green pseudo-color represents activated platelets that form clots generated by photochemical stimulation inside the blood vessel, and the red pseudo-color represents plasma flowing through the blood vessel. The direction of blood flow is indicated by a yellow arrow and is not indicated from the point when a blood clot blocks the blood vessel. The blue dotted circles indicate the 532-nm laser pattern and diameter. The diameter of each vessel visible in the overall image is indicated by a white number and was measured from the corresponding white dotted line. Scale bar, 50 µm.

Download Full Size | PDF

Real-time dual-color fluorescence imaging shows that the target vessel begun to appear for about 350 seconds, from which point we insert the fiber bundle more slowly and wait for the vessel to make proper contact with the fiber bundle imaging plane, as shown in Fig. 4(A).

Targeting the vessels was completed at 590 seconds (see Visualization 6). At this time, the depth is 2.7 mm, and the targeted blood vessels are located in the anterior pretectal nucleus of the dorsal part of the thalamus. As shown in Fig. 4(B), within a few minutes, vascular bifurcation sites were targeted with a circular pattern of 100-µm, and targeted photothrombosis was performed for 10 minutes. Based on the dual-color fluorescence image, it was confirmed that blood flow was blocked in one horizontally flowing blood vessel in about 100 seconds, the flow of the left branch blood vessel directed vertically upward stopped at 225 seconds, and the remaining blood vessels were blocked entirely in 300 seconds (see Visualization 7).

3.4 Measurement of photothrombotic occlusion lesions day one after subcortical small vessel occlusion

When TTC staining was performed on day one after occlusion, no infarct lesions were identified after photothrombotic occlusion, as shown in Fig. 5(A). This indicates that occlusion of a small vessel in the deep brain may take at least a day or more to cause an acute infarction identifiable by TTC staining. However, as a result of confirming the post hoc fluorescence image of Fig. 5(B), it could be confirmed more clearly that a thrombus was formed in the lower extremity capillaries of the target blood vessels at the exact target location. The occlusion lesions were identified by the fluorescence signals in brain slices corresponding to -2.0 mm and -3.0 mm from bregma, and the location of the lesions appeared only in a minimal range, as shown in Fig. S2. In addition, no lesions associated with hemorrhagic stroke or other brain damage resulting from fiber bundle insertion were identified. This occlusion lesion is localized in the brain area corresponding to the anterior pretectal nucleus of the dorsal part (APTD), a sub-region included in the thalamus [24].

 

Fig. 5. TTC staining and post hoc histology for evaluation of histological damage day one after deep brain photothrombotic occlusion (n = 1). (A) TTC-stained brain slices of targeted photothrombosis were those of mice in Fig. 4. The numbers on the right are the anterior to posterior coordinates relative to bregma. Brain coronal section images were taken from both sides of each slice. The upper right inset shows the coordinates of the deep brain photothrombotic target region in vivo in the anatomy of the mouse brain schematically depicted in the sagittal plane. The red line points to the portion of the TTC-stained coronal brain slices corresponding to the photothrombotic target area approximately -2.0 mm from bregma. Scale bar, 2 mm. (B) Post hoc histological fluorescence images were measured from the back of a coronal brain slice corresponding to -2.0 mm from bregma in Figure A. The yellow dashed box indicates the fiber probe insertion track. The white dotted line indicates the region of the anterior pretectal nucleus in the dorsal part of the thalamus. A magnified image of the ipsilateral hemisphere in the red box area shows a small infarct lesion differently from the relative contralateral hemisphere. Scale bars, 1 mm (center) and 50 µm (right).

Download Full Size | PDF

We also confirmed whether it was possible to repeatedly induce thrombus by selectively targeting blood vessels in APTD of other mice as well.

In this study, in order to perform targeted photothrombosis using a 100 µm laser circular pattern, small blood vessels with a diameter less than 50 µm were targeted in the APTD region located at the same stereotactic coordinates of the brain as the experiments performed in Figs. 4 and 5. As shown in Fig. 6(A, B, E, and F), a small blood vessel located in the APTD was targeted, and patterned photothrombosis was induced within 10 minutes using real-time dual-color fluorescence imaging and patterned laser stimulation simultaneously. Also, acute infarction was not seen in the TTC-staining, but the positive fluorescence signal due to the accumulation of thrombus and leakage of the BBB in the APTD region was accurately confirmed on the post hoc fluorescence images, as shown in Fig. 6(C, D, G, and H). Moreover, the occluded lesions were confirmed by the fluorescence signal of the brain slices corresponding to -2.0 mm and -3.0 mm from the bregma, and the location of the lesion appeared only in the minimum range. No lesions associated with hemorrhagic stroke or other brain damage resulting from fiber bundle insertion were identified. As shown in Fig. 6, it could confirm that it was possible to precisely target the lower extremity capillaries of the target blood vessels in the APTD and repeatedly create a thrombus at the corresponding location.

 

Fig. 6. Time-lapse images, TTC staining, and post hoc histology of in vivo anterior pretectal nucleus of the dorsal part (APTD) photothrombotic occlusion (n = 2). The upper left inset shows the coordinates of the APTD in vivo in the anatomy of the mouse brain schematically depicted in the coronal plane. The bottom right is the bregma coordinates. (A-D) first object (E-H) second object. (A, B, E, and F) The image fames were obtained while performing targeted photothrombosis in vivo using a laser circular pattern with a diameter of 100-µm to select specific blood vessels in the APTD in the deep brain. The green pseudo-color represents activated platelets that form clots generated by photochemical stimulation inside the blood vessel, and the red pseudo-color represents plasma flowing through the blood vessel. The blue dotted circles indicate the 532-nm laser pattern and diameter. The diameters of each vessel visible in the overall image are indicated by a white number and were measured from the corresponding white dotted line. (A and E) Selectively display 5 s and (B and F) 600 s image frames out of a total of 121 frames. (C, D, G, and H) TTC staining and post hoc histology on day one after APTD photothrombotic occlusion. In the TTC staining brain slices, the green dotted box indicates the track of fiber probe insertion, and the green dotted circle indicates the region for targeting a small vessel. In mosaic fluorescence images, the white dotted line indicates the region of the APTD in the thalamus. Magnified images of the ipsilateral hemisphere in the red box and yellow box areas show a small infarct lesion. Scale bars, (B and F) 50 µm, (C-H) 1 mm (TTC stained brain slices and mosaic fluorescence images), 200 µm (red box), and 50 µm (yellow box).

Download Full Size | PDF

Therefore, we can clearly assert that our FBEµ can repeatedly and precisely target blood vessels located in the specific region of the deep brain, such as the APTD, to induce targeted photothrombosis, generating a model of subcortical small vessel occlusion.

4. Discussions

In this study, we showed that a subcortical small vessel occlusion model can be created by combining a real-time FBEµ and Rose Bengal photothrombosis. The results demonstrate that our techniques can induce small vessel occlusion in the deep brain as well as the cortical surface in living mice. The FBEµ allowed a minimally invasive access to certain critical deep brain regions in mice through stereotaxic insertion. In addition, it was possible to select a specific blood vessel and to simultaneously observe the formation of a thrombus and blockage of blood flow by using a real-time dual-color fluorescence image while photochemically stimulating the target blood vessel with patterned laser light.

Before photothrombosis, a mixture of fluorescently labeled activated platelet antibody and Rhodamine-B dextran with Rose Bengal was injected into the tail vein of mice. The sham group was injected with saline instead of Rose Bengal. Rose Bengal, a photosensitive dye, is activated by light illumination of a specific wavelength including 532-mm to form singlet oxygen and peroxide, which causes endothelial damage, platelet activation, and thrombus aggregation [16]. It also triggers rapidly evolving ischemic cell death in brain regions covered with illuminated vascular regions that eventually lead to ischemic stroke. Our results show that when thrombus formation was initiated with Rose Bengal photothrombosis, the accumulation of activated platelet antibodies was indicated by an increase in the green fluorescence signal. At the same time, the flow of plasma inside the blood vessel was disturbed by the thrombus, and the red fluorescence signal of the plasma stained with Rhodamine-B dextran gradually decreased. This helps to accurately and quickly determine whether a vessel occlusion is complete. It can also identify changes in blood flow, shape, and secondary thrombus formation in the surrounding blood vessels.

In addition, in the early stage of photothrombosis, clots begin to form along the vessel wall, which is unstable and is easily removed by the bloodstream [19]. Therefore, sufficient time and photochemical stimulation are required for the blood clots to accumulate enough to block the target blood vessel. The larger diameter vessels with relatively high blood flow and greater fluidity may take longer than smaller vessels and may be difficult to close completely. Most optical systems for Rose Bengal photothrombosis published to date have used a modified confocal microscopy or two-photon microscopy system that uses a small laser spot focused on the vessel wall [17,19,30].

This is also why most photothrombosis studies are limited to small blood vessels such as capillaries. On the other hand, the laser pattern stimulation of our FBEµ system can stimulate a wider area compared to the focused laser spot stimulation of other optical systems. Therefore, our FBEµ system can allow for rapid aggregation and occlusion even in relatively large cerebral vessels.

In this study, only circular stimulation patterns were used, but the laser pattern stimulation system of FBEµ can be used for optical stimulation for blood vessels of different shapes with various diameters from 2 µm to 320 µm [20]. This means that selective photothrombosis is possible for vessels of various shapes and diameters, which were difficult to achieve with conventional photothrombosis technique.

Although there have been imaging studies using various fluorescent antibodies in the existing thrombosis studies, few studies have confirmed plasma and blood clots in the brain at the same time as in this study [31]. The dual-color fluorescence imaging system can significantly help to understand direct neuropathological causality by enabling complementary analysis using two or more bio-images and acquiring various biological information from the same object. The application of indicators capable of representing various ligands involved in the mechanism of thrombus adhesion, such as the platelet glycoprotein Ib-V-IX system [32], will enable the study of novel thrombotic stroke mechanisms. Our fluorescence imaging system enables cellular imaging with a resolution of 2 µm [20,23], which can be used to the observation of pathophysiological changes in small vessel diseases due to the interaction of the neurovascular complex composed of peripheral neurons, non-neuronal cells, and vascular cells. It would also be applicable in long-term monitoring studies needed to investigate the effects of small vessel diseases progression on cognitive and functional changes after stroke [33], the relationship between small vessel progression and BBB dysfunction [5] and the vascular penumbra [34].

Small subcortical strokes are caused by blockage of one of the small penetrating terminal arterials at the base of the brain and have traditionally been thought to result from fibrin degeneration [1]. However, there is debate in the literature that embolism is associated with the etiology of lacunar infarction [35]. Some argue that it is important to investigate subcortical infarction for underlying embolism [36]. In order to elucidate the exact etiology of subcortical infarction, the development of appropriate animal models and methodologies is highly required. Our system and methods can further develop an appropriate thromboembolic stroke model that prevents excessive endothelial damage and rapid destruction of the BBB by photochemical reactions and induces only cytotoxic edema, similar to the pathogenesis of clinical ischemic stroke [16]. Above all, our FBEµ system can also be proposed as an optical technique that can maximize many methodological advantages of photothrombosis, including uniform infarct location and size, low mortality, and relatively simple surgery [19].

In the subcortical small vessel occlusion model, we targeted small vessels in APTD, a sub-region included in the thalamus. The anterior pretectal nucleus is a diencephalic nucleus that projects on both the posterior thalamic nucleus and the zona incerta, suggesting that it has been regarded as a mediator participating in the pathophysiology of central pain syndrome [37,38]. Therefore, the small vessel occlusion model obtained in this study can be applied to the pathophysiological study of pain syndrome caused by abnormal anterior nuclear activity due to small vessel occluded disease.

In addition, we did not observe the prognosis of secondary hemolytic stroke, which can be caused by the direct insertion of the fiber bundle probe into the brain. These results demonstrate that our FBEµ is suitable for generating small vessel occlusion in the deep brain, such as in the basal ganglia (thalamus, globus pallidus, and caudate nucleus), the pons, and the subcortical white matter structures (corona radiate and internal capsule), with a minimally invasive approach. These anatomical sites correspond to lacunar stroke or lacunar infarction lesions at the thalamoperforant arteries, the basilar artery, the lenticulostriate arteries, the anterior choroidal artery, pericallosal artery, and the recurrent artery of Heubner from the anterior cerebral artery [2,7,9,39,40].

However, our technique has some limitations to consider. Our system is a stereotaxic-based technique that target blood vessels distributed in specific brain region, such as the APTD. However, since such a technique target specific brain regions by coordinates of the normalized brain, we could not account for differences in vascular distribution across subjects. Moreover, since our system could observe relatively narrow F.O.V. of 320 µm, it has a disadvantage in confirming whether the thrombus was formed downstream or in the surrounding blood vessels outside the F.O.V. Therefore, to complement our technique, future studies may integrate our system with a real-time in vivo whole-brain microvascular imaging system, such as the recently developed ultrasound localization microscopy [41,42] or ultrafast photoacoustic microscopy [43]. The combined system can check the blood flow along with the anatomical structure of blood vessels distributed in a targeted subcortical layer in real-time. Moreover, the advanced technique may allow deep cerebral blood vessel selective thrombus based on the customized photothrombosis considering subject-wise brain vessel morphology rather than the anatomical coordinates of the brain area. It is expected to enable research on specific cerebral blood flow-based brain cognitive disorders such as vascular dementia in the future [44,45].

5. Conclusions

In this study, we successfully created a subcortical small vessel occlusion model using a FBEµ for targeted photothrombosis with minimal invasiveness. Our FBEµ system made it possible to precisely target specific blood vessels in deep brain and simultaneously observe the clot formation and blood flow blockage inside the target blood vessel during photochemical reactions. These results suggest that our system and methods can further develop various clinical animal models for etiology and therapeutic studies of lacunar stroke.