|Home | About | Journals | Submit | Contact Us | Français|
Glioblastoma multiforme (GBM) is one of the most dreaded cancer diagnoses due to its poor prognosis and the limited treatment options. Homozygous deletion of the p16INK4a/p14ARF/p15INK4b locus is among the most common genetic alterations in GBM. Two recent studies have shown that deletion and mutation of another INK4 family member, p18INK4c, also drives the pathogenesis of GBM. This minireview will discuss the known roles for p18INK4c in the initiation and progression of cancer and suggest opportunities for future studies.
Glioblastoma multiforme (GBM) is one of the most devastating human cancers, notorious for its fast growing nature, infiltrative growth, resistance to radiotherapy, and rapid progression from diagnosis to death (1). Homozygous deletion of the p16INK4a/p14ARF/p15INK4b locus on chromosome 9p21.3 is a signature genetic event that drives the pathogenesis of GBM (2, 3). Deletion of this locus is thought to be strongly selected for during the pathogenesis of GBM and other tumor types because it simultaneously inactivates three tumor suppressor genes: the p16INK4a and p15INK4b cyclin-dependent kinase (cdk) inhibitors and p14ARF, which modulates the activity of the p53 tumor suppressor (4, 5).
p16INK4a is the prototype member of the INK4 family of cdk inhibitors, which includes p16INK4a, p15INK4b, p18INK4c, and p19INK4d. These four proteins have ~30% amino acid sequence similarity and share potent cyclin-dependent kinase (cdk) inhibitory activity, but are thought to differ from each other in their patterns of expression and their relative affinities for different cdks. Of the four genes, p16INK4a plays the most prominent role as a tumor suppressor gene in human cancer, as it is deleted or mutated in the majority of common human tumor types and the cause of an inherited cancer predisposition syndrome. In contrast, p15INK4b and p18INK4c are thought to play more limited roles as human tumor suppressors, and p19INK4d is thought not to be involved in cancer pathogenesis.
Two recent studies have shown that deletions and mutations of p18INK4c, together with deletions and mutations of the p16INK4a/p14ARF/p15INK4b genes, drive the pathogenesis of human GBM (6, 7). When taken together with other prior studies identifying deletions and mutations of p18INK4c in several other tumor types and showing that p18INK4c-deficient mice are tumor prone, these recent studies have suggested that inactivation of p18INK4c may play a perhaps underappreciated role in human cancer pathogenesis. As such, this review will provide a brief history of the discovery of p18INK4c, summarize the studies that link it to cancers in mice and humans, and provide a forward-looking assessment of the role of p18INK4c inactivation in the pathogenesis of GBM (Fig. 1).
The human p18INK4c gene was initially discovered in 1994 by Guan and colleagues (8) in a yeast two-hybrid screen to identify proteins that interacted with human cdk6. The mouse homologue was reported the following year by Hirai and colleagues (9). Human p18INK4c is a 168-amino-acid cdk-interacting protein with multiple ankyrin repeats and substantial similarity to both human Notch and the yeast cdk inhibitor PHO8. p18INK4c bound with high affinity to cdk6, with intermediate affinity to cdk4, and did not bind to cdk2. This discovery of a new p16INK4a homologue immediately raised the exciting possibility that p18INK4c might be a new tumor suppressor gene commonly inactivated in a wide range of human cancers. However, initial genetic analysis of p18INK4c status in >100 human tumor samples derived from >20 different tissues failed to identify homozygous deletions and found only a single sample with a putative somatic missense mutation (10).
Detailed biochemical studies provided further information on the role of p18INK4c in inhibition of cdks. Initial studies suggested that p18INK4c preferentially inhibited cdk6-containing complexes whereas p16INK4a preferentially inhibited cdk4-containing complexes (8). This suggested that despite their status as homologues, these proteins have evolved for the purpose of providing subtly distinct negative regulation of cdks in the human cell cycle. However, this notion that INK4 proteins have distinct binding specificities for cdk4 and cdk6 remains somewhat controversial (11).
In contrast to the early data suggesting that p18INK4c might play a limited role in human cancer pathogenesis, data from knockout mouse studies have shown a clear role for p18INK4c in cellular proliferation, organ size, and cancer. p18INK4c-deficient mice are viable but show a variety of intriguing phenotypes. Franklin and colleagues (12) reported that p18INK4c-deficient mice displayed pronounced gigantism, organomegaly, lymphoma, and pituitary hyperplasia, phenotypes that were strikingly reminiscent of those seen in Rb−/− mice. Shortly thereafter, Latres and colleagues (13) reported that p18INK4c-deficient mice also displayed deregulated epithelial cell growth in the kidneys and breast. Interestingly, Bai and colleagues reported that p18INK4c heterozygous (+/−) knockout mice developed tumors at an accelerated rate when treated with a chemical carcinogen. These tumors retained the remaining allele of p18INK4c, showing that p18INK4c is a haploinsufficient tumor suppressor in mice (14).
Subsequent studies in which p18INK4c-deficient mice were crossed with mice harboring targeted deletions of other tumor suppressor genes uncovered additional roles for p18INK4c in cancer pathogenesis. A subset of these studies are described here. Mice lacking both p18INK4c and p53 were predisposed to medulloblastomas, hemangiosarcomas, and other tumors not present in either of the parental strains (15). Deletion of p18INK4c in Ptc1+/− mice led to the rapid development of medulloblastoma with high penetrance, and further examination of these and related mice indicated that p18INK4c plays an important role in normal cerebellar development (16–18). Deletion of p18INK4c in PTEN+/− mice led to an enhanced predisposition to a variety of tumors when compared with mice with deletion of either gene alone, suggesting that these genes could synergize during the development of human malignancies harboring inactivation of both genes (such as GBM; ref. 19). Finally, mice lacking both p18INK4c and p16INK4a developed unusually aggressive pituitary tumors, suggesting that these GBM-related genes also synergize in tumor formation and that these INK4 family members could at least partially compensate for each other (20).
Despite the early negative results in human cancer samples, the potential promise of p18INK4c as a human tumor suppressor gene and the mouse studies implicating p18INK4c as a bona fide tumor suppressor gene prompted several groups to continue to search for homozygous deletions and inactivating point mutations in a variety of common human cancer types. These efforts continued to prove largely frustrating, in that deletions/point mutations of p18INK4c were identified rarely, albeit in a range of human tumor types including adenocarcinoma of the breast, multiple myeloma, acute lymphoblastic leukemia, oligodendroglioma, and meningioma (10, 21–30). Epigenetic inactivation of p18INK4c in human medulloblastoma has also been reported (16).
Of these tumor sample types, cell lines derived from multiple myeloma harbored by far the highest frequency of p18INK4c deletions, ~30%. Two studies in particular highlighted the frequency and effect of p18INK4c deletion in multiple myeloma. In 2002, Kulkarni and colleagues (23) identified homozygous deletions of p18INK4c in 6 of 16 multiple myeloma cell lines and showed that these deletions often occurred in tumor cells also harboring deletions of the p16INK4a/p14ARF/p15INK4b locus. In 2006, Dib and colleagues (24) confirmed and extended this work, identifying homozygous deletions of p18INK4c in 13 of 40 multiple myeloma cell lines, but in only 2% to 10% of primary tumor samples. There were two opposing possible explanations for these data: the worrisome possibility that the majority of these p18INK4c deletions in multiple myeloma cell lines could be artifacts of ex vivo culture, and the exciting possibility that deletions of p18INK4c could be present in a similarly high fraction of primary tumor samples as well but that their presence was masked by intratumoral heterogeneity. These same issues would reemerge in the more recent studies implicating p18INK4c in the pathogenesis of GBM.
In April 2008, Solomon and colleagues (6) and Wiedemeyer and colleagues (7) reported homozygous deletions of p18INK4c in GBM. These deletions were identified using copy number analysis in GBM cell lines, xenografts, and primary tumors. Notably, Wiedemeyer and colleagues also identified three missense mutations of p18INK4c that interfered with binding to cdk6. There were two similarities between the data presented in these studies and those previously reported for multiple myeloma. First, whereas p18INK4c deletions were identifiable in GBM primary tumor samples, they were most commonly found in GBM cell lines and xenografts. Second, as in multiple myeloma, GBM samples with deletions of p18INK4c often also harbored deletions of the p16INK4a/p14ARF/p15INK4b locus. Follow-up functional studies clearly showed that p18INK4c could play a tumor suppressor role in human and murine GBM, and Wiedemeyer and colleagues suggested that p18INK4c could functionally compensate for p16INK4a homozygous deletions during the pathogenesis of GBM.
One clear result of the GBM and multiple myeloma studies is that deletions of p18INK4c are much easier to identify in cultured samples than in uncultured samples. There are at least two possible explanations for this disparity.
One possibility is that deletions of p18INK4c are more easily identifiable in cultured samples because ex vivo growth enriches for p18INK4c-deficient cells already present in the tumor. In fact, Solomon and colleagues showed just this: Whereas one primary GBM tumor they studied clearly harbored a deletion of p18INK4c, identification of the deletion was obscured by intratumoral heterogeneity. However, that same deletion was easily identifiable in a first-passage xenograft derived from that tumor (6). That same primary tumor was homogeneous for deletion of the p16INK4a/p14ARF/p15INK4b locus, providing support for the notion that genetic heterogeneity, and not the presence of admixed nonneoplastic cells, obscured detection of the p18INK4c deletion in the primary tumor. Of note, it seems unlikely that the genetic heterogeneity displayed by these and other heterogeneous deletions was due to true haploinsufficiency because all p18INK4c deletions identified by Solomon and colleagues and Wiedemeyer and colleagues in genetically homogeneous cell lines seemed to be biallelic/homozygous. Based on these types of data and immunohistochemistry studies identifying loss of p18INK4c expression in ~40% of GBMs, Solomon and colleagues suggested that deletions of p18INK4c could be common in GBM primary tumors but that intratumoral heterogeneity might obscure their easy identification.
However, it is also a formal possibility that the deletions of p18INK4c identified by Solomon and colleagues and Wiedemeyer and colleagues in cultured GBM samples could have occurred de novo during ex vivo culture, and as such represent true culture artifacts. Importantly, however, both groups clearly identified deletions of p18INK4c in both cultured and uncultured samples, providing strong evidence that deletions of p18INK4c are not artifacts of culture. Furthermore, it is arguable whether any putative cancer-related gene has been shown to be deleted/mutated only in samples grown in an ex vivo setting, and some believe that the enthusiasm for the idea that ex vivo culture leads to artifactual genetic lesions exceeds the data to support it.
A related point brought up by Solomon and colleagues is that unlike the heterogeneous deletions of p18INK4c, GBMs are remarkably homogeneous with regard to deletions of the p16INK4a/p14ARF/p15INK4b locus. Emerging copy number analyses of GBM samples make it obvious that deletion of this locus is by far the most common homozygous deletion present in GBM (in >75% of samples), and that genetically heterogeneous uncultured primary GBMs tend to be homogeneous for deletion of this locus. This degree of homogeneity suggests that deletion of the p16INK4a/p14ARF/p15INK4b locus is unusually fundamental to the pathogenesis of GBM, and, when taken together with the heterogeneity of deletions in p18INK4c, suggests that deletion of the p16INK4a/p14ARF/p15INK4b locus is an early event in the pathogenesis of GBM, whereas deletion of p18INK4c is likely to be a later event associated with progression of GBM (Fig. 1). These data further suggest that deletion of the p16INK4a/p14ARF/p15INK4b locus could even be the initiating event in GBM, as this would simultaneously explain the remarkable homogeneity of these deletions and the shockingly explosive growth of primary GBM.
One intriguing observation is that deletions of p18INK4c generally coexist in the same GBM cells also harboring deletions of the p16INK4a/p14ARF/p15INK4b locus. There are several possible explanations for this surprising finding. Wiedemeyer and colleagues showed that depletion of p16INK4a led to transcriptional up-regulation of p18INK4c by the E2F1 transcription factor, suggesting that p18INK4c is induced to compensate for deletion of p16INK4a, and that this provides selective pressure for the loss of p18INK4c (7, 31). Such a model might be considered a “serial model,” in that there is a linear mechanistic relationship between the tumor suppression activities of these two genes. However, it is also worth noting that (a) deletions of p16INK4a are at least twice as common in GBM than deletions of p18INK4c; (b) GBMs are quite heterogeneous for deletions of p18INK4c yet remarkably homogeneous for deletions of the p16INK4a/p14ARF/p15INK4b locus; and (c) the cdk inhibitors encoded by these loci probably have distinct specificities for binding to different cdks. These observations suggest that INK4 genes could play biologically distinct roles in the initiation (p16INK4a) and progression (p18INK4c) of GBM. Such a model might be termed a “parallel model,” in that inactivation of these genes is mechanistically unrelated and loss of each gene has distinct biochemical and phenotypic sequelae during the pathogenesis of GBM.
Numerous studies have now implicated p18INK4c as a tumor suppressor whose genetic inactivation plays a role in the pathogenesis of GBM, multiple myeloma, and a variety of other common cancers. Deletions of p18INK4c are clearly less common than deletions of the p16INK4a/p14ARF/p15INK4b locus and seem to show substantial intratumoral heterogeneity. They often occur in tumors also harboring deletions of the p16INK4a/p14ARF/p15INK4b locus but probably occur later in tumorigenesis, raising fundamental questions about the relative contributions of these two loci to the initiation and progression of human neoplasia. Further studies are clearly warranted to better define the relative biochemical and phenotypic roles of these INK4 family members in the pathogenesis of GBM and other human tumor types.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.