To be an oncogene or a tumor suppressor? The answer lies in the dose of p27

From the Kemp lab, Human Biology Division

The cyclin-dependent kinase (CDK) inhibitor p27kip1 (CDKN1B) is known to be a tumor suppressor, however, it may also function as an oncogene within the cytoplasm by affecting cell motility and metastasis.  This activity was detected when the CDK inhibitory and cyclin-binding domains of p27 were mutated, resulting in the inability of p27 to bind to cyclins and CDKs to affect the cell cycle.  Nevertheless, the mechanism of oncogenic action of p27, and the strong association between its expression and patient mortality remains unknown. Most critically, it is unclear if p27 supports or blocks metastasis.

As a tumor suppressor, p27 has been shown to be haploinsufficient, with loss of only one allele being sufficient to cause tumorigenesis.  However, literature on the oncogenic role of p27 has been inconsistent.

Historically, a basic tenet of disease pathophysiology is that higher doses of a bad stimulant are usually thought to cause greater harm.  However, that is not necessarily always true.  Nonmonotonic dose-response curves, meaning a non-linear dose-response interaction, are widely observed in biology, especially in studies that involve responses to hormones and endocrine-disrupting chemicals. In these instances, the shape of the curve reverses as the dose increases. Could this also apply to p27 in the context of cancer?  Dr. Kemp (Human Biology Division) and his collaborators at New Mexico State University believe so.  They hypothesize that a nonmonotonic function of p27 exists for its role in cell migration, and suggest that both cell migration and invasion could be active at an intermediate dose of p27. The results of their study were recently published in the journal Cancer Informatics.

To do this, they turned to an in vivo model, and analyzed gene expression at the mRNA level in chemically induced mouse squamous cell papillomas derived from p27 null (corresponds to a low dose), heterozygous (corresponds to an intermediate dose), and wild-type (corresponds to a high dose) mice to capture potential nonmonotonic effects.  These transcriptomic responses were further characterized using a novel informatics method for functional pathway analysis.  

Existing pathway analysis approaches are limited in their abilities to determine dependency between pathway responses and dosage. To overcome these limitations, the authors came up with a novel method called the functional pathway dose-response analysis, which has the ability to test whether the response of a pathway is a function of the dosage of a gene.  This mathematical approach is rooted in the concept that for a gene to be functionally dependent on the dose of the stimulant, the expression value of the gene has to be uniquely determined by the dose. They applied this approach to characterize how genes and functional pathways respond to different doses of the Cdkn1b gene in mouse papillomas.

Interestingly, 16 out of 17 pathways known to play important roles in tumorigenesis from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database were more highly expressed at the intermediate gene dosage of p27 at a false discovery rate of 10%, strongly suggesting that the intermediate dose of p27 can promote cancer. On further examination of the member genes in these cancer pathways, Cyclin D1 (Ccnd1)—a cell cycle regulator—emerged as the most responsive gene in several p27-responsive cancer pathways. Its response pattern suggests that Ccnd1 is active at the intermediate dose of p27.  In humans, Ccnd1 has also been reported as an oncogene that is overexpressed in many tumors. This result implies that intermediate levels of p27 may have pro-oncogenic properties, and is further enhanced with additional evidence from The Cancer Genome Atlas (TCGA) showing the dominance of CDKN1B heterozygous somatic mutations in thousands of samples from diverse tumor types.

This study has revealed, using an in vivo tumor model that a large proportion of genes in known pathways responded nonmonotonically to p27 dosage, and corroborates literature describing both tumor suppressing and oncogenic activities of p27.  The authors also show evidence for a novel approach to determine directional influence from dose to response and uncover previously overlooked potential causal relationships. In the context of p27, the oncogenic mechanism resulting from nonmonotonic effect of gene expression could regulate cell migration leading to tumor metastasis. The authors speculate that the transition from an oncogenic to tumor suppressing cellular context may be induced when the dose of p27 diverges from the intermediate range. Dr. Kemp also alluded that this nonmonotonic effect of gene expression is almost certainly a general phenomenon in cancer, also evident in a p27 mouse model of prostate cancer published in 2004. Going forward, the Kemp lab is planning on using functional genomics approaches to identify new drug targets to selectively kill cells with aberrant p27 expression.  

Decorative Image
(A) CyclinD1 gene expression follows the pattern of Low-High-Low, in response to increasing dose of p27, suggesting that Ccnd1 is highly expressed at the intermediate level of p27 dosage. (B) Model for the relationship of p27kip1 gene dosage and CyclinD1 expression in prostate cancer. Enhanced prostate cancer tumorigenesis was observed at an intermediate dose of p27. Adapted from Gao et al, 2004. Figures provided by Dr.Kemp and adapted by Y Lim

Nguyen HH, Tilton SC, Kemp CJ, and Song M. 2017. Nonmonotonic pathway gene expression analysis reveals oncogenic role of p27/Kip1 at intermediate dose. Cancer Informatics. (16):1-13. doi:10.1177/1176935117740132

Funding was provided by the National Institutes of Health, in particular the NCI Partnership for the Advancement of Cancer Research between the Fred Hutch and New Mexico State Uuniversity.

Research reported in the publication is a collaboration between the Fred Hutch and New Mexico State University.