Abstract
Metastatic prostate cancer (PCa) poses a significant therapeutic challenge with high mortality rates. Utilizing CRISPR-Cas9 in vivo, we target five potential tumor suppressor genes (Pten, Trp53, Rb1, Stk11, and RnaseL) in the mouse prostate, reaching humane endpoint after eight weeks without metastasis. By further depleting three epigenetic factors (Kmt2c, Kmt2d, and Zbtb16), lung metastases are present in all mice. While whole genome sequencing reveals few mutations in coding sequence, RNA sequencing shows significant dysregulation, especially in a conserved genomic region at chr5qE1 regulated by KMT2C. Depleting Odam and Cabs1 in this region prevents metastasis. Notably, the gene expression signatures, resulting from our study, predict progression-free and overall survival and distinguish primary and metastatic human prostate cancer. This study emphasizes positive genetic interactions between classical tumor suppressor genes and epigenetic modulators in metastatic PCa progression, offering insights into potential treatments.
Introduction
Prostate cancer (PCa) is the second most common cancer in man and the incidence is continuously increasing1. PCa is notoriously heterogeneous, and the progression from an indolent disease to advanced cancer usually takes several years2,3,4,5. The molecular alteration that drives disease progression is still not fully understood, even though genetic sequencing of patients’ samples and pre-clinical models have revealed essential mechanisms2,3,6,7,8. Genetically engineered mouse models of PCa have displayed gene functions in the development of prostatic intraepithelial neoplasia (PIN) and PCa9,10,11,12. Intercrossing of mouse strains with specific loss of different tumor suppressor genes in the mouse prostate has shown cross-talks between these factors and accelerated cancer progression. For instance, loss of Rb1 in combination with Trp53 and Pten results in the formation of neuroendocrine tumors at 4 months and loss of Smad4 with Trp53 and Pten forms adenocarcinoma at 4 months with metastasis formation in multiple organs13,14. Intercrossing of multiple mouse strains with conditional loss of specific genes in the prostate tissues is time-consuming. However, this can be overcome by combining CRISPR/Cas9 technology with adeno-associated virus (AAV) delivery system. AAV system is used for gene therapy as it has a low risk of insertional mutagenesis and elicits low immune responses, comparing to other vectors15. We have successfully developed a strategy to mutate multiple genes simultaneously in the epithelia of the mouse prostate16,17. This method allowed clonal expansion of tumor cells under Darwinian selection as seen in human cancer18,19.
Extensive genomic sequencing of PCa over the last decade has identified multiple essential loss-of-function mutations. This includes common mutations not only in tumor suppressor genes such as TP53 and PTEN but also in genes, of which functions are less understood in the context of PCa3,6,7. For instance, mutations in epigenetic factors such as lysine methyltransferases (KMT) genes are found in a significant amount of samples3,6. Mutations in epigenetic factors such as ZBTB16, KMT2C, and KMT2D are often found in advanced PCa, but the impact on PCa of such alterations remains unclear. Similarly, mutation in STK11 can occur in PCa, and data from Stk11 deficient mice have revealed a possible tumor suppressive function in PCa20,21,22,23. While germline mutations in RNASEL are less common, they have been associated with a predisposition to prostate cancer24,25,26. Moreover, RNASEL has been linked to other cancer types, suggesting a potential role in cancer incidence and initiation27,28.
In this study, we aimed to seek the genetic as well as epigenetic basis for PCa progression and onset of metastatic disease. For this, we applied CRISPR to simultaneously mutate five tumor suppressor genes Trp53, Pten, Rb1, Stk11, and RnaseL, in the mouse prostate by AAV delivery. This generated a rapid invasive and androgen-independent tumor where mice reached humane endpoint at eight weeks after initiation but without the formation of metastasis. Additionally, CRISPR guides targeting Zbtb16, Kmt2c, and Kmt2d were included together with the aforementioned five genes to mutate eight genes simultaneously. The tumor progression was unchanged, but all mice developed lung metastasis, and further investigation revealed that loss of Kmt2c was essential for metastasis formation. With whole genome sequencing (WGS) and RNAseq, we addressed that tumor progression was driven by alteration in gene expression in classical pathways but not by additional somatic mutations. Furthermore, the loss of Kmt2c resulted in the upregulation of a highly conserved region covering a unique gene cluster, which has not previously been associated with tumor development. Remarkably, the disruption of two genes in this cluster, Odam and Cabs1, dismissed secondary tumor formation, revealing their implication in metastatic formation. Overall, this study showed that loss of multiple tumor suppressor genes accelerated PCa progression and that loss of Kmt2c initiated the onset of metastatic disease through regulation of a unique gene cluster.
Results
Loss of multiple tumor suppressor genes accelerates PCa progression
Whole exome and genome sequencing (WES/WGS) has revealed genomic alteration in a large number of genes that are associated with prostate cancer, including TP53, PTEN, RB1, STK11, and RNaseL3,6,7. Analysis of WES data of 494 PCa patients from the cancer genome atlas (TCGA) and 1013 patients from Memorial Sloan Kettering Cancer Center (MSK) and Dana-Farber Cancer Institute (DFCI) confirmed that all 5 genes are mutated in prostate cancer with the highest prevalence of TP53, PTEN, and RB1 (Fig. S1). Furthermore, we observed mutational co-occurrence between these tumor suppressor genes with no mutual exclusivity detected, and the alteration was associated with a worse prognosis (Fig. S1). To evaluate the positive genetic interaction between loss of multiple tumor suppressor genes in prostate cancer progression, we applied CRISPR technology to mutate these five targets simultaneously in the same somatic cell of a mouse prostate. We constructed a viral vector with unique sgRNAs for Pten, Trp53, Rb1, Stk11, and RnaseL (hereafter 5 g), and generated AAV particles for in vivo applications. As controls, viral constructs with only a guide for Pten (sgPten) or a non-targeting guide RNA (sgNT) were produced (Fig. 1A). We delivered viral particles to the murine anterior prostate by surgical injection to Cas9-EGFP flox mice, which were bred to the prostate-specific Cre line, PB4Cre, for limiting expression of Cas9 and EGFP to the prostatic tissues. Hereby, cancer was induced specifically to the mouse prostate, and undesired tumor inductions to other organs were avoided.
The weight of the prostate was evaluated four weeks after tumor initiation, and no difference was observed between control groups and 5 g samples. However, this was significantly altered at 6 weeks after injection, where prostates with loss of five genes were enlarged (Fig. 1B, C). At around eight weeks, prostates with five mutations had become very enlarged, and mice were sacrificed as a humane endpoint. In comparison, loss of Pten alone or a non-targeting gRNA had not compromised the mice at this stage (Fig. 1D).
To assess if the rapid tumor progression was driven by the CRISPR-induced mutations in the five target genes, biopsies underwent Sanger sequencing at the target site. Mutations in Pten were detected for all samples, revealing that loss of Pten is required for initiation of PCa, as shown before16,18. Mutations in Trp53 and Rb1 were also detected for all the samples, but Stk11 and RnaseL were only mutated in a subset of samples, revealing that clones with different mutation profiles were present as a consequence of imperfect CRISPR/Cas9 generated mutations (Fig. 1E, S2). Histology assessment of the prostates revealed low-grade PIN in sgPten control samples at eight weeks, whereas 5 g samples contained areas of high-grade PIN at four weeks10. At 6 weeks, 5 g tumors progressed to invasive cancer, but no metastasis formation was identified in lung, lymph node, or liver before humane endpoint at 8 weeks (Fig. 1F). Immunohistochemistry (IHC) for phospho-Akt (pAkt) revealed increased levels due to the loss of Pten and staining for basal cell marker p63 showed intact basal structure in control samples. However, this feature was lost in 5 g samples at 8 weeks, as seen in human PCa29. Tumors were positive for E-cadherin and negative for Synaptophysin (SYP), showing epithelial origin of the tumors (Fig. 1G). Cells positive for proliferation marker Ki67 were significantly increased in 5 g samples compared to sgPten control in agreement with larger tumor burden (Fig. 1G, H). To assess if the rapidly growing tumors will respond to androgen deprivation, mice were castrated 5 weeks after tumor initiation, and samples were assessed at 8 weeks. No difference in tumor size was observed between non-castrated and castrated mice, even though two mice did present smaller tumors (Fig. 1I, J). Histological assessment of the tumors from the castrated mice showed an invasive tumor with a high proliferation index, similar to the samples from intact mice (Fig. 1K). Overall, loss of Pten, Trp53, Rb1, Stk11, and RnaseL in the murine prostate drives tumor formation to invasive cancer in less than 8 weeks, revealing the existence of positive genetic interaction between these gatekeeper genes in prostate cancer progression.
Loss of epigenetic factors drives metastatic formation
Epigenetic factors are frequently mutated in human PCa but their implication on PCa is still not fully understood. Especially, mutations in KMT2C, KMT2D, and ZBTB16 are found in primary PCa and associated with poorer survival, as well as mutation burden for KMT2C is increased in metastatic lesions compared to primary PCa (Fig. S3). We sought to understand the implication of epigenetic alterations in PCa progression for these three genes. To this end, we cloned CRISPR guides targeting Kmt2c, Kmt2d, and Zbtb16 together with a sgRNA for Pten in an AAV backbone. Furthermore, we assembled the fragments targeting three epigenetic factors into 5 g vector, to mutate eight genes simultaneously in the murine prostate (hereafter called 8 g) (Fig. 2A). AAV particles containing 8 sgRNAs were injected to the anterior mouse prostate, tumor progression was rapid and similar to mice that received AAV particles containing 5 g (Fig. S4). At 4 weeks after tumor initiation the weight of the prostate was not altered, but at 6 weeks a large tumor developed, and ~8 weeks after initiation, mice reached humane endpoint (Fig. 2B–D, S4). Initiation of PCa by the AAV particles containing the epigenetic factors in combination with sgPten did not accelerate tumor progression when compared to only loss of Pten (Fig. S4). Furthermore, mice with loss of Pten, Trp53 and Rb1 had slower tumor progression, revealing that loss of Stk11 and RnaseL accelerated tumor development (Fig. S4G–J). A group of mice was castrated 5 weeks after tumor initiation, and deprivation of testosterone did not impair tumor size, even though two mice had smaller tumors, and expression of the AR-regulated gene Klk4 was downregulated (Fig. 2C, D, S5). Hereafter, sanger sequencing was performed on the tumor samples at the target regions of the guides, to analyze indel frequency. This confirmed that all guides induced indel formations, and most samples had mutations in all eight genes (Fig. 2E). Histological assessment of the tumors showed areas of high-grade PIN 4 weeks after initiation and invasive cancer at 6 to 8 weeks. This was similar to the tumors induced by the injection with 5 g virus but more advanced compared to sgPten control samples (Fig. 2F). Evaluation of the proliferation by Ki67 staining revealed increased proliferation in 8 g group when compared to sgPten controls, but a slight decrease compared to tumors with loss of five tumor suppressor genes (Fig. 2G, H)….