Oncogenic State and Cell Identity Combinatorially Dictate the Susceptibility of Cells within Glioma Development Hierarchy to IGF1R Targeting

Abstract Glioblastoma is the most malignant cancer in the brain and currently incurable. It is urgent to identify effective targets for this lethal disease. Inhibition of such targets should suppress the growth of cancer cells and, ideally also precancerous cells for early prevention, but minimally affect their normal counterparts. Using genetic mouse models with neural stem cells (NSCs) or oligodendrocyte precursor cells (OPCs) as the cells‐of‐origin/mutation, it is shown that the susceptibility of cells within the development hierarchy of glioma to the knockout of insulin‐like growth factor I receptor (IGF1R) is determined not only by their oncogenic states, but also by their cell identities/states. Knockout of IGF1R selectively disrupts the growth of mutant and transformed, but not normal OPCs, or NSCs. The desirable outcome of IGF1R knockout on cell growth requires the mutant cells to commit to the OPC identity regardless of its development hierarchical status. At the molecular level, oncogenic mutations reprogram the cellular network of OPCs and force them to depend more on IGF1R for their growth. A new‐generation brain‐penetrable, orally available IGF1R inhibitor harnessing tumor OPCs in the brain is also developed. The findings reveal the cellular window of IGF1R targeting and establish IGF1R as an effective target for the prevention and treatment of glioblastoma.

(J) Schematic of the Mir155 construct used to knock down endogenous mouse IGF1R. miRNA sequences against luciferase were used as the non-specific control. Of note, these miRNA sequences were inserted into the artificial beta-globin intron of an EGFP coding sequence to enable the visualization of transfected cells. directly isolated from the CKO_NG2-Cre ER mouse model. FL, full length of IGF1R. B, betachain of IGF1R. OE, over-exposed.

(D)
The percentage of labeled mutant OPCs among all labeled mutant cells in multiple brain regions at three different time points after tamoxifen treatment. The WT_Nestin-Cre ER model at 43 dpi was used as the reference. Data averaged across all these regions are provided on the right. N=3 mice for each group.
(E) Representative images of the brain sections from the mouse models indicated. The images were obtained from the CC region (region #6 in C). The arrows point to the cells coexpressing all of the markers indicated. Importantly, in the WT brain, adult NSCs rarely gave rise to OPCs even at 43 dpi. Scale Bar: 100 μm.
(F) Schematic of the tamoxifen administration schedule for the mouse models used in this figure.

Genotyping standards for all mouse models used in this paper
To ensure the correct genotyping information, all mice used in the study were genotyped twice: the first time were performed before weaning; and the second time was done upon analysis.

Mouse tissue preparation and histology
After anesthesia, mice were briefly perfused with cold PBS and then thoroughly perfused with 4% paraformaldehyde. Brains were isolated and post-fixed in 4% PFA before dehydrated in 30% sucrose. Fixed brain tissues were embedded into optimal cutting temperature (O.C.T.) and snap-frozen on dry ice before preserved in a -80℃ refrigerator and used for cryosection.
To collect fresh tissues for Western Blot, qPCR, RNA-Seq, sc-RNA sequencing and cell culture, tissues were acutely collected from deeply anesthetized mice without perfusion. A fluorescence stereoscope was used to facilitate visualization and collection of tumor tissues whenever necessary. Fresh tissues were either snap-frozen in liquid nitrogen or directly dissociated into single cells for sc-RNA seq and/or the primary cell culture. A small piece of adjacent tumor tissue was usually collected and fixed into PFA for routine histological analyses.

BrdU staining
For BrdU staining, tissue sections, after three washes in 1x PBS (10 minutes each), were immersed in 1.5M HCl (in 1x PBS) for 30 minutes at 37°C. Following HCl treatment, tissue sections were washed three times with 1x PBS (10 minutes each) before incubated with the blocking solution (0.3% Triton X-100 in 1x PBS, 5% Normal Donkey Serum, 0.1% NaN 3 ) and gone through routine staining procedure as described. [7,25]

CC1-antibody staining
To remove the nonspecific staining background signals generated by the mouse anti-CC1 monoclonal antibody (Millipore, #OP80), we used donkey-anti-mouse monovalent Fab fragment to block the tissue sections before primary antibody staining as described previously.

MEF cell line preparation and culture
To prepare MEFs, timed matings were set up (with the plug day as E0.5) and the embryos were collected at E13.5. As the embryos from the same litter harbored distinct genotypes, the MEFs from each embryo were prepared and used separately. For each embryo, the head, arms, legs and viscera were tossed and the remained torso was minced and digested by TrypLE (Gibco, #12605028) and dissociated into single cells before cultured and maintained in DMEM (HyClone, #SH30022.01B), supplemented with Penicillin/Streptomycin and 10% FBS. The remained tissues were used for genotyping. As the NG2-Cre ER was not expressed in the MEFs, to knock out the Trp53 flox and the NF1flox allele in the mutant MEFs, we transiently transfected the cultured MEFs with a plasmid (PIGGY-Cre) that expressed Cre recombinase. The success of recombination was validated by genotyping and the strong expression of tdTomato in these cells.

Trp53/NF1 mutant and wild-type OPC culture
OPCs were enriched through immuo-panning using the anti-O4 antibody. Cells were

Real-Time qPCR
Total RNA was isolated by the Trizol Reagent and reversely transcribed using the oligo (dT) primer. qRT-PCR was performed on the CFX96 Touch™ Real-Time PCR Detection System using HieffTM qPCR SYBR® Green Master Mix. β-actin (ActB) and/or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were used as the internal control to normalize the expression of the interested genes. See Table S4 (Supporting Information) for primer information.

Quantification of tumor Spheres
Tumor cells were incubated with growth factors as indicated for 4-7 days. Human GBM spheres were cultured for 14 days, and 50% of the culture media was replaced for every 7 days. Images for each well were taken with Fluorescence Inverted Microscope System (OLYMPUS CKX53) in 10x magnifier. 4 images (one for each well) were analyzed for each condition. The size and/or the number of spheres were quantified and measured by Image J.
We defined tumor spheres as the cluster with more than 3 cells. The diagraph and statistical analysis were done with Graphpad Prism 5.

Quantification for the NG2-Cre ER mouse models in Figure 3
The cell proliferation rates were quantified by the percentage of the cells that incorporated BrdU. The mouse brains were sectioned coronally in 20 μm thickness and stained with BrdU (1:500), DsRed (1:100) and Olig2 (1:500). We used DAPI to determine the anatomical regions and collected images from the olfactory bulb (OB) and distinct cortical regions. We randomly collected images from 10 brain sections in the OB and 6 sections in the cerebral cortex. For each brain section, 4 regions were collected in the OB and 6 regions were collected in the cerebral cortex. Each image was taken by a 20x objective with 1x digital zoom by 2 μm optical sectioning (scanning depth 6 μm). We quantified the BrdU + , DsRed + , Olig2 + co-stained cells and the DsRed + , Olig2 + co-stained cells. These counts were used to calculate the proliferation rate of labeled OPCs (the percentage of BrdU labeling among all mutant OPCs) in the brains of distinct genotypes (see Figure 3C-3D). These values were averaged from N=3 brains and presented as mean ± SEM.

Quantification for the G/R ratios and the density of OPC-lineage cells in the MADM mouse models presented in Figure 4
To quantify MADM labeled OPC-lineage cells, the MADM brain tissues were sectioned sagittally in 20 μm thick slices. We stained the MADM brain sections with GFP (1:500), c-Myc (1:200) and Olig2 (1:500). N=4 brains were processed for each time point. We used DAPI to determine the brain regions. For each brain, 5 brain sections were examined and 5 cortical regions within each brain section were imaged. Therefore, 25 images were collected from each mouse brain. Each image was taken by a 20x objective with 1x digital zoom by 2 μm optical sectioning (scanning depth 6μm). We processed each image in two ways: First, we quantified all Olig2 + cells, including red, green, yellow and non-labeled ones. These data were used to calculate the G/R ratio between green and red Olig2 + cells (see Figure 4G) and the percentage of all four labeled cell populations with distinct genotypes (see Figure 4D-F).
Second, we calculated the area (mm 2 ) of each image to determine the density of all Olig2 + cells (see Figure 4H). These values were averaged from N=4 brains from each genotype and presented as mean ± SEM.

Quantification for the Nestin-Cre ER mouse models in Figure S7
To determine the distribution of the mutant OPCs (PDGFR + , tdT + ) in distinct brain structures, the mouse brains were sectioned coronally in 20 μm thick sections and stained with PDGFRantibody (1:250). The live fluorescence of tdTomato was strong enough.
Therefore, no DsRed antibody was used to enhance the signal. We used DAPI to determine brain anatomical structures and defined 10 brain regions covering major representative brain structures in both the gray and the white matter (see Figure S7C), which include the olfactory bulb (two random fields), the cerebral cortex (three random fields), the corpus callosum (two random fields, one is next to the SVZ and the other was in the midline), the lateral-SVZ, the striatum and the ventral pallidum. Each image was taken by a 20x objective with 1x digital zoom by 2μm optical sectioning (scanning depth 6μm). We counted all the PDGFR + , tdT + co-stained cells and tdT + only cells to calculate the percentage of mutant OPCs among all NSC-derived cells (see Figure S7D). We calculated the area using the Olympus Fluoview 1000 software. The number was averaged from N=3 brains and presented as mean ± SEM.

Quantification for the Nestin-Cre ER mouse models presented in Figure 5
To measure the proliferation rates of mutant adult NSCs and OPCs, we focused on 2 distinct brain regions: the lateral-SVZ and the corpus callosum next to the SVZ in the coronal brain sections. Brain tissues were sectioned coronally in 20 μm thick sections and stained with BrdU (1:500), DsRed (1:100) and Olig2 (1:500) or Sox9 (1:250). In order to quantify the number of mutant granule cells derived from NSCs, we stained the OB sections with NeuN (1:250). We used DAPI to determine brain regions such as the lateral-SVZ (the region where mutant NSCs reside), the corpus callosum next to SVZ (the region mutant OPCs reside) and the OB (the region NSC-derived granule cells reside). Each image in the SVZ region was