Study design
The animal protocol for this study was approved by the Stanford University Administrative Panel on Laboratory Animal Care (APLAC). All experiments were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals. Two series of experiments were carried out, one for histological analysis and one for FCM analysis.
Thirty mice with brain tumors were divided into six groups that received the following treatments prior to histology analysis:
-
Group 1: No MRgFUS - control group - animals were euthanized when the tumors were 4–5 mm in diameter (n = 5).
-
Group 2: MRgFUSx1 - one session of MRgFUS when the tumors were 4mm in diameter and the animals were euthanized 2 days after MRgFUS (n = 5).
-
Group 3: MRgFUSx2 - two sessions of MRgFUS every other day starting when the tumors were 3–4 mm in diameter and euthanized at 2 days after the second session of MRgFUS (n = 5).
-
Group 4: MRgFUSx3 - three sessions of MRgFUS, once every other day, starting when the tumors were a 2-3mm in diameter, euthanized at 2 days after the last session of MRgFUS (n = 5).
The other ten animals were divided into 2 groups that received the following treatments prior to FCM analysis.
-
Group 5: Tumor_No FUS, animals were euthanized when the tumors were 4–5 mm in diameter (n = 5).
-
Group 6: Tumor_FUS, three sessions of MRgFUS, once every other day, starting when the tumors were a 2-3mm in diameter, euthanized at 2 days after the last session of MRgFUS (n = 5).
Three naïve animals without brain tumors were euthanized to collect brain tissue for FCM analysis.
Cell Line and Culture
The GL26 mouse glioma cell line was provided by Dr. Ramasamy Paulmurugan and maintained in media consisting of DMEM, 10% FBS, 100ul/ml Penicillin-Streptomycin, and 4mM L-glutamine.
Mice
All mice were housed in specific pathogen-free conditions at a barrier facility at Canary Center at Stanford University School of Medicine (Stanford, California). All animal handling, surveillance, and experimentation was performed in accordance with and approval from the Stanford University Administrative Panel on Laboratory Animal Care.
Homozygous Ccr2RFP/RFP mice (JAX stock #017586) (27) and Cx3cr1-GFP mice (JAX stock #005582) (28) on a C57BL/6 background were purchased from the Jackson Laboratory and intercrossed to yield Ccr2RFP/wtCx3cr1GFP/wt animals. To confirm the establishment of heterozygous mice, ear snips were collected when the offspring were 15 through 18 days old, and genotyping was performed using a commercial assay service (Transnetyx, Inc). Previous studies (29) have shown that implantation of murine GL261 glioma cells to the Ccr2RFP/wtCx3cr1GFP/wt dual knock−in mice allowed for efficient evaluation of the myeloid cells, known to constitute the majority of CD45 + immune cells in gliomas. Immunohistochemistry allows for discrimination of CX3CR1-GFP + microglia and TAMs derived thereof and CCR2-RFP + blood-derived monocytes and monocyte-derived TAMs within the tumor as well as in surrounding adjacent brain .
Orthotopic syngeneic model of mouse brain tumors
Mouse glioma GL26 cells dissociated into single-cells suspensions were orthotopically injected into the brain of 8 to 10-week-old Ccr2RFP/wtCx3cr1GFP/wt mice using sterotactic injection. In brief, mice were anesthetized with 3% isoflurane (Minrad International, Buffalo, NY, USA) in an induction chamber. Anesthesia on the stereotactic frame (David Kopf Instruments, Tujunga, CA, USA) was maintained with 2% isoflurane/L oxygen delivered through a nose adaptor. A burr hole was placed 1.7 mm lateral and 2 mm posterior of bregma. A blunt-ended needle (75N, 26s/2”/2, 5 µL; Hamilton Co., Reno, NV, USA) was lowered into the burr hole to a depth of 3.5 mm below the dura surface and retracted 0.5 mm to form a small reservoir. Using a microinjection pump (UMP-3; World precision Instruments, Sarasota, FL, USA), 4x105 GL26 cells were injected in a volume of 3 µL at 30 nL/s. After leaving the needle in place for 1 minute, it was retracted at 3 mm/min. The cranial injection site was sealed using biodegradable glue. Tumor formation was followed by MRI using a 3-tesla scanner from MR Solutions. From 1-week post tumor implantation, T2- Fast spin echo (FSE, repetition time/echo time [TR/TE] = 4800/68 ms, flip angle 90°, 2 averages, field of view 28 mm, matrix size = 256x248, slice thickness 1.0 mm) images were acquired once every 2 days to monitor the growth of the tumor. T2*-weighted gradient echo images (repetition time/echo time [TR/TE] = 391/13 ms, flip angle 20°, 3 averages, field of view 28 mm, matrix size = 256x256, slice thickness 1.0 mm) were obtained in order to identify possible hemorrhage.
MRgFUS set up and treatment protocol
MRgFUS was delivered to open the BBB. The MRgFUS system (Image Guided Therapy, Pessac, France) was configured as described in previous studies (30, 31). The system included an MR-compatible, pre-focused, eight-element annular array, 1.5-MHz transducer (spherical radius = 20+/-2 mm, active diameter = 25 mm [focal ratio = 0.8]; Imasonic, Voray sur l’Ognon, France), which was connected to a phased array generator and radiofrequency power amplifier. The transducer and animals were prepared as described in a previous study (32). The membrane in front of the transducer was filled with degassed water and acoustic gel was applied between the transducer and skin. For sonication, the animals were placed in a prone position and maintained in that position using a bite bar and ear bars. The scalp hair was shaved and removed with depilatory cream. The experimental apparatus in this study is shown in Fig. 1.
Definity® Microbubbles (mean diameter range: 1.1–3.3 µm, mean concentration of 1.2 × 1010 bubbles per mL, diluted by 1:20 using 1×PBS, 300 µL/kg, Lantheus Medical Imaging, MA, USA) were injected through a catheter placed in tail vein just before sonication (1.5 MHz, pulse duration 20-ms, duty cycle of 2%, 1-Hz pulse repetition frequency, 90-s duration per sonication). Multiple sonications were administered in the vicinity of the targeted area of the brain by moving the sonication zones slightly rostro-caudally and medio-laterally targeting the brain tumors. An MR-compatible motorized positioning stage was used to move the transducer in the rostral-caudal and medial-lateral directions. After determining the coordinates of the focal point within the MRI space, treatment planning MRI was acquired, and the focal region was positioned within the tumor. Ultrasound bursts were then applied at peak negative pressure of 0.5MPa.
MRI Data Collection
On the day prior to, and immediately post MRgFUS, a set of MRI data including T2-FSE, T2*-weighted gradient echo, dynamic contrast-enhanced (DCE), and post-contrast T1-weighted images was obtained.
The pre and post-MRgFUS T2-FSE images were acquired in order to assess the size and location of the resulting lesions, T2*-weighted gradient echo images were obtained to identify possible hemorrhagic complications from the MRgFUS procedure. In order to obtain quantitative measurements of BBB permeability, a bolus of gadodiamide contrast (gadobenate dimeglumine; Multihance, Bracco Diagnostics Inc., Monroe Township, NJ 08831, USA) was injected intravenously for DCE imaging (TR/TE = 34/3 ms, average = 1, FOV 28 mm2, flip angle 20°). Post-contrast T1-weighted imaging (TR/ TE = 620/12 milliseconds, 2 averages, field of view = 28 mm, matrix size = 256x244, slice thickness = 1 mm) was utilized to confirm the opening of the BBB after DCE imaging. Images were reviewed and analyzed using the Horos DICOM viewer. Using FDA-approved commercial software NordicICE (Nordic Neuro Lab, Bergen, Norway), Ktrans maps were computed by using a pipeline inspired by that of Anzalone et al (33). Notably, local AIF adapted for mice studies were extracted from the signal curves using a blind deconvolution method (34).
Tissue Preparation and Analysis
Mice were euthanized with inhalation of 3% isoflurane 2 days after the last MRgFUS session and perfused through the left ventricle at 15 mL/min for 1 min with 0.9% NaCl and then for 30 min with 4% paraformaldehyde in 0.1 M phosphate buffer solution (PBS, pH 7.4). Brains tissues were post-fixed overnight at 4°C and then transferred into 30% (w/v) sucrose in PBS. After equilibrating in the 30% sucrose solution, the brains were sectioned coronally (30 µm) with a sliding microtome. Sections were collected in 30% ethylene glycol and 25% glycerol in 50 mM PBS and stored at 20˚C until used. A 1-in-6 series of sections were collected for nuclear counterstaining with Invitrogen Hoechst 33342 dye. Coronal brain sections containing tumors were acquired with a NanoZoomer Digital Pathology slide scanning system (Hamamatsu Photonics, K.K., Japan).
In order to analyze the histological images, a custom analysis pipeline was setup in order to provide a tool able to quantify in a semi-automated manner the number of GFP and RFP labeled cells per mm2 on the brain sections. On the images of the stained sections of the brain through Nanozoomer, a Region of Interest (ROI) was drawn around the brain tumor with the Freehand drawing tool of NDP.view2 software (Hamamatsu Photonics, K.K., Japan) by the the investigators blinded to the identity of the animals and sections they were analyzing. In order to minimize the bias in quantifying the GFP-labeled cells and the RFP-labeled cells, the red and green channel were set both at full dynamic ranges of 200% while the dynamic range of blue channel was switched to 0% and turned off. A simple filter was used to improve the resolution of the image and a magnification factor of 1.8 to 2.5% was applied to the ROIs. Then, the images were exported in .jpg files and analyzed in FIJI (ImageJ). FIJI is an open source software commonly used for biomedical image analysis (35). Upon loading the images in FIJI, the background was first removed using the Rolling Ball Radius algorithm (36, 37). The red and the green channels were then split, and considering the different sizes of the RFP-labeled cells and the GFP-labeled cells, two different radii were adopted to subtract the red and the green components from the background. Otsu thresholding was used on the red and green channel images separately to distinguish positive from negative signals (38, 39). The group above the threshold (automatically computed based on signal intensity in the gray scale) was recognized as the effective signal from the labeled cells and the group below the threshold was recognized as background noise and discarded. The application of a binary mask was followed by the watershed separation as a robust segmentation method, based on the average size of recognized single cells (40). Finally, a simple particle counting method was applied to the resulting images (for the green channel and the red channel) taking into account the average size of the single cells in order to quantify the number of GFP and RFP cells per mm2. T-test was used to compare the number of GFP and RFP cells per mm2 from two different groups.
FCM analysis
Mice were deeply anesthetized with 1–4% isofluorane through a nosecone, and transcardially perfused with ice-cold PBS. Brain specimens (left and right brain parenchyma from naïve animals; left and right brain parenchyma, and tumor tissue from the animals with gliomas) were dissected and dissociated into single-cell suspensions using the Brain Tumor Dissociation Kit (Miltenyi Biotech, Catalog # 130-095-942). Cells were then resuspended in ice-cold FCM buffer containing HBSS without Calcium and Magnesium, 2% FBS, and 10mM HEPES. Zombie Nir (Biolegend) staining was applied to exclude dead cells (20 min at 4°C degree), followed by a rinse using an ice-cold FCM buffer. Commercially available rat-anti-mouse CD16/32 antibody (Biolegend) staining for Fc-blocking (20 min at 4°C degree) followed by washing in ice-cold FCM buffer were employed to eliminate nonspecific binding. The cells were then stained with fluorochroma-conjugated antibodies for 30 min at 4°C degree in dark (Table 1), followed by washing in ice-cold FCM buffer. All data were collected on a BD LSR flow cytometer and analyzed using FlowJo 10 software v10.6.1 (Tree Star Inc.).
Two gating strategies were utilized to classify and identify the immune cells in brain and tumor immune microenvironement:
Strategy #1: The cell membrane markers were utilized to classify the immune cells in brain tissues and in brain tumors, TAMs including monocytes, blood-derived macrophages, microglia, and microglia-derived macrophages were classified from myeloid cells excluding Ly6G + neutrophils. Subsequently, the expressions of CX3CR1-GFP and CCR2-RFP were analyzed from the above four groups (Supplementary figure.1). Gating based on CD80 and CD206 was applied to show the pro-inflammatory (CD80 + CD206-) and anti-inflammatory (CD80-CD206+) polarization of TAMs.
Strategy #2, reverse gating: Genetic gating with CX3CR1-GFP and CCR2-RFP was applied to the myeloid cells, followed by the analysis with Ly6C and F4/80 to identify the subgroups of CX3CR1-GFP and CCR2-RFP expressing cells (Supplementary figure.2).