Toward a CRISPR-based mouse model of Vhl-deficient clear cell kidney cancer: Initial experience and lessons learned

Abstract

CRISPR is revolutionizing the ability to do somatic gene editing in mice for the purpose of creating new cancer models. Inactivation of the VHL tumor suppressor gene is the signature initiating event in the most common form of kidney cancer, clear cell renal cell carcinoma (ccRCC). Such tumors are usually driven by the excessive HIF2 activity that arises when the VHL gene product, pVHL, is defective. Given the pressing need for a robust immunocompetent mouse model of human ccRCC, we directly injected adenovirus-associated viruses (AAVs) encoding sgRNAs against VHL and other known/suspected ccRCC tumor suppressor genes into the kidneys of C57BL/6 mice under conditions where Cas9 was under the control of one of two different kidney-specific promoters (Cdh16 or Pax8) to induce kidney tumors. An AAV targeting Vhl, Pbrm1, Keap1, and Tsc1 reproducibly caused macroscopic ccRCCs that partially resembled human ccRCC tumors with respect to transcriptome and cell of origin and responded to a ccRCC standard-of-care agent, axitinib. Unfortunately, these tumors, like those produced by earlier genetically engineered mouse ccRCCs, are HIF2 independent.

Keywords: HIF2; PT2399; VHL; ccRCC; mouse models of ccRCC.

Fig. 1.

CRISPR/Cas9 gene editing strategy for generating a syngeneic Vhl-deficient ccRCC mouse model. (A) Schematic representation of Ksp-Cre AAV and Cre-less AAV vectors. (B) % gene editing of Vhl, Pbrm1, Keap1, and Tsc1 in 3T3-J2 Cas9 parental cells, 3T3-J2 Cas9 crude prep, 3T3-J2 Cas9 cells with GFP-AAV (control), 3T3-J2 Cas9 cells with AAV carrying Vhl, Pbrm1, Keap1, and Tsc1 sgRNAs. (C) Immunoblot of 3T3-J2 Cas9 cell preps shown in (B) indicating effective gene editing. (D) Lox-STOP-Lox-Cas9-GFP cassette in genetically modified Cas9/+ mice denoting Cre recombinase excision of loxP flanked STOP sequence.

Fig. 2.

Combined loss of Vhl, Pbrm1, Keap1, and Tsc1 is sufficient to drive ccRCC development. (A) Tumor-free survival of Ksp-Cre model vs. Cre-less AAV model. (B) Comparative overall survival of Ksp-Cre AAV model vs. Cre-less AAV model. N = 24 mice for KSP-Cre AAV model; n = 23 mice for Cre-less AAV model. Mice that died for reasons unrelated to tumor burden were censored. P value determined by the log-rank test, P < 0.0001. Tumor visualization by MRI in (C) Ksp-Cre AAV model and (D) Cre-less AAV model. Representative images of renal tumors in (E) Ksp-Cre AAV-injected LSL-Cas9/+ mice and Cre-less AAV-injected LSL-Cas9/+, Pax8 Cre mice. Immunohistochemistry of markers indicating targeted gene editing. (F) H&E staining of Cre-less AAV model tumor with adjacent normal kidney (upper region) showing a mixed clear cell and eosinophilic histology. Immunohistochemical staining of Cre-less AAV model for (G) GFP, (H) Baf180 (Pbrm1), (I) Ca9, (J) phospho-S6 (mTOR readout), and (K) Pax8. (Scale bar, 100 µm.)

Fig. 3.

Comparative gene expression analysis of Cre-less AAV model kidney tumors and TCGA human ccRCC. (A) Venn diagrams showing up-regulated and down-regulated genes in Cre-less AAV model tumors vs. TCGA ccRCC as shown in volcano plots. (B) Volcano plots displaying up-regulated and down-regulated genes in Cre-less AAV kidney tumors vs. normal mouse kidney (Left plot) and in TCGA ccRCC vs. normal human kidney (Right plot). (C) Hypoxia-related gene expression in Cre-less AAV mouse tumor vs. normal mouse kidney (Left panels) and in TCGA ccRCC vs. normal human kidney (Right panels). (D) GSEA showing significantly up-or down-regulated pathways in Cre-less AAV model kidney tumor vs normal mouse kidney and TCGA ccRCC vs normal human kidney.

Fig. 4.

Single-cell RNA sequencing of Cre-less AAV model kidney tumors and normal mouse kidney tissue. (A) UMAP of nonimmune cells from 4 Cre-less AAV kidney tumors, colored and labeled by cluster/cell type. (B) Heatmap showing similarity of tumor cells (split by sample) and stromal cells to normal kidney cell types. Colors represent predicted similarity of a tumor or stromal subset in a given sample (columns) to normal cell types (rows). (C) Left, UMAP of CD8+ T cells from all four normal kidney and four tumor samples, colored and labeled by cluster/cell type. Middle and Right, UMAPs of CD8+ T cells split by normal kidney and tumor samples and colored by relative point density. (D) Quantification of CD8+ T cell subset proportions relative to total CD8+ T cells in normal kidney and tumor samples. P-values determined by the two-sided Wilcoxon rank-sum test. (E) Violin plots showing normalized expression of indicated genes in CD8 Tox-Hi cells from normal kidney and tumor samples. Bonferroni-adjusted P-values are from a logistic regression differential expression analysis using Helicobacter infection status as a covariate (Materials and Methods).

Fig. 5.

Potential of Cre-less AAV model for therapeutic studies. Axitinib study in Cre-less AAV model demonstrates response to VEGF pathway therapeutic intervention (100 mg/kg, b.i.d., p.o., 5 × week, 7-wk duration; n = 10 mice/treatment group). (A) Mean tumor volume, axitinib treated vs. vehicle. (B) Tumor growth rate, axitinib treated vs. vehicle. (C) Immunohistochemical staining of CD31, Ki67, and Caspase 3 in kidney tumors from Cre-less AAV model treated with axitinib (100 mg/kg) or vehicle. P value, unpaired t test, two-tailed.