Commentary on Alfred G. Knudson, Jr.: “Hereditary Cancer, Oncogenes, and Antioncogenes” Free

Visit the Cancer Research 75^th Anniversary timeline.

See related article by Knudson, Cancer Res 1985;45:1437–43.

The original studies, reviews, and essays of Alfred G. Knudson, Jr. have shaped a conversation around the process of oncogenesis, and more specifically tumor suppression, for more than five decades. As a body of work, Knudson's studies and writings embody the full spirit of Einstein's quote above, exemplifying the time-honored scientific concept that careful work, solid data, attention to the work of the field as well as one's own, and a large dose of insight and imagination synergize to produce major conceptual advances (wholeheartedly acknowledging that imagination alone is not key to progress, but imagination rooted in the fertile soil of experimentation and collaboration can be transformational). Knudson, of course, is best known for the “two-hit hypothesis” that predicted loss of two alleles of a tumor-suppressive gene, or antioncogene, as they were first known, would be sufficient to instigate hereditary pediatric tumors, with retinoblastoma the signature example (1). With time and tireless study by myriad researchers, the fundamental truth of this concept has come to light, particularly for retinoblastoma. Nevertheless, the route to our current understanding of cancer genetics has been arduous and has revealed both important nuances and new, overarching concepts that have provided both conundrums and opportunities for therapy well beyond the wildest dreams of decades past.

Knudson's essay in Cancer Research in 1985 entitled “Hereditary Cancer, Oncogenes, and Antioncogenes” (2) reflected an evolution of thought embracing the burgeoning appreciation of a genetic etiology of cancer, together with exciting support for the two-hit hypothesis at the molecular level. By this point, viral oncogenes and their cellular counterparts had been identified and well established as drivers of transformation in cell culture, and the concept of oncogene “classes” was extant (3), underscoring the idea articulated in this review that oncogenes may be individually poor at driving the full process of carcinogenesis on their own, hence their potential rarity as hereditary cancer genes. Most importantly, in 1985, Knudson was able to point to molecular evidence in support of the two-hit hypothesis, or at the very least the existence of discrete genetic elements with the characteristics of “antioncogenes,” in numerous pediatric and hereditary cancers, including retinoblastoma, Wilm's tumor, neurofibromatosis, hereditary polyposis of the colon with associated colon carcinoma, and a sarcoma and breast cancer syndrome that became known as Li-Fraumeni (4).

Although each of these diseases showed genetic patterns that strongly supported the concept of recessive cancer genes, consistent with functions expected of antioncogenes, Knudson notes that most such syndromes produce rare clonal tumors in tissues otherwise free of major indicators of pathology arising from the heterozygous state, with rare exceptions such as polyposis. This observation supported the concept that the conditions caused by hereditary antioncogenes in many cases are not truly cancer syndromes, but rather cancer-susceptibility syndromes, in which a primed cell, or pool of cells, having lost a functional copy of the wild-type antioncogene allele, could serve as a fertile progenitor for further mutations that lead to frank cancer, thus unifying the concept of hereditary antioncogenes with the concept that multiple “hits” may be required for many cancers to display overt pathology. Indeed, this concept also illustrates a key concept in genetically recessive cancer genes, which often exert dominant phenotypes, as the cells that lose the second allele are advantaged in cancer formation, rather than disadvantaged, as might be the case with a recessive gene causing a profound developmental abnormality when lost in both copies in the germline of an affected individual, but little effect when lost in both copies in a small subset of somatic cells later in development or in adulthood.

However, as noted in Knudson's analysis in 1971, retinoblastoma remained particularly distinct from other syndromes, presenting as a highly penetrant, multifocal pediatric cancer that strongly supported the need for only two “hits,” loss of both alleles of a putative retinoblastoma gene, to produce an overt tumor. By 1985, such a gene was tantalizingly close; its identity supported by inherited, visible chromosomal alterations at 13q14 in many affected individuals, yet others remained somewhat of a conundrum, as major alterations of the second 13q14 locus or losses of the entirety of the chromosome could not be found in many tumors, interpreted by some to indicate a second, perhaps collaborating retinoblastoma gene (5). As Knudson clearly articulates in this review, keys to this mystery lay in the unassuming esterase D enzyme, the gene that is tightly linked to the retinoblastoma syndrome and that has electrophoretic isoforms derived from polymorphic alleles that allow individual chromosomes 13 to be tracked in unaffected somatic versus tumor cells. Coupled with the new technology of RFLP analysis, retinoblastoma disease–associated alleles could with confidence be found to be retained in tumors at the expense of the wild-type chromosome 13. Such evidence strongly supported the existence of the “13q^rb” chromosome, or that bearing a cytogenetically undetectable deletion or even point mutation that conferred the tumor phenotype, and likely defined the retinoblastoma gene itself, long a “holy grail” that would become the first isolated human anti-oncogene, or tumor suppressor gene.

The imagination of Knudson, and the hard work from his laboratory, and those of geneticists R.S. Sparkes, M.C. Sparkes, Wilson, Towner, Benedict, Murphree, and Yunis (6), Dryja (7), and others, who defined the genetic associations of retinoblastoma disease with chromosome 13q14 deletions, culminated just a year later in the first report identifying the retinoblastoma gene (8). Indeed, sentence two of Friend's seminal work identifying the retinoblastoma gene, “There is evidence for another class of oncogenes, in which tumor-predisposing mutations are recessive to wild-type alleles,” both captures the overarching, paradigm-shifting nature of their study and credits this Cancer Research review of Knudson a year earlier in setting the stage for this breakthrough. The identification of what is now known as RB-1 can thus be attributed to the knowledge and imagination of Knudson, as well as to the many others who labored on behalf of so many affected children to understand their tumor biology and genetics, as encapsulated in the 1985 review under discussion here.

Many years hence, the RB-1 gene and its encoded protein, although at first an enigma without indicators of function, have unleashed a torrent of studies leading to deep insight into the role of tumor suppressors, not only in retinoblastoma but in numerous other cancers where somatic mutation of RB-1 or its regulators is crucial to initiation or progression of disease, and indeed, the concepts that led to the cloning of RB-1 have provided cancer researchers with insights into the function of a plethora of other tumor suppressors, née antioncogenes, not the least of which is p53 (ironically cloned years before due to its association with a viral oncogene, SV40 T-antigen; ref. 9), but only definitively identified as an antioncogene in 1989 (10), despite intriguing evidence for its recessive nature in murine tumorigenesis (e.g., ref. 11). Indeed, identification of p53 mutation as a causative event in hereditary cancers of the Li-Fraumeni syndrome was secondary to appreciation of p53's widespread somatic loss in the same types of cancers in sporadic cases (12), as was notably first described in colon cancer (13, 14), but the principles articulated by Knudson apply, albeit both genes leave us with the conundrum of why certain tissues are susceptible to germline loss of certain tumor suppressors, whose somatic mutation leads to many more tumor types.

Illustrating the continued currency of concepts brought forth in the 1985 Cancer Research review, the conundrum described above is the subject of recent speculation by Knudson, whose imagination continues to wrestle with the broader implications of cancer as a disease of development (15). In the 1985 review, Knudson notes that tissue formation (from stem cells) involves commitment to restricted differentiation, proliferation of committed cells to generate tissue-specific progenitors, followed by acquisition of a terminal postmitotic state. He then posits: “Is it possible that oncogenes play a major role in the proliferation phase and antioncogenes play one in the differentiation phase of histogenesis?” As with identification of antioncogenes, Knudson's speculations appear very pertinent to ongoing studies of the normal role of both tumor suppressors and oncogenes, and again, retinoblastoma may serve as a leading example of cancer genes' roles in differentiation and histogenesis. As predicted years before, loss of both alleles in the germline of RB-1 in mice is an embryonic lethal event due to loss of function of certain tissues, notably including hematopoiesis (e.g., ref. 16), and indeed widespread somatic loss is also lethal due to disruptions of normal tissue functions (17). Loss of RB-1 in specific progenitor cells impairs their lineage commitment in vivo and in vitro (18, 19), consistent with a significant role for the retinoblastoma protein in potentiating the activity of tissue-specifying transcription factors (e.g., refs. 20, 21). Finally, on a broader scale, it is notable in this regard that all of the core developmental pathways, exemplified by the biochemical pathways conveying WNT, Notch, Hedgehog, and Hippo signaling, are key contributors to tumorigenesis, with both tumors suppressors and oncogenes among their ranks. As Maris and Knudson pointed out recently (15), an appreciation of the embryologic effects of heritable cancer mutations is likely to have relevance beyond the pediatric cancers to include sporadic cancers in older individuals, where somatic mutation of these genes is common. Although cutting-edge techniques describing the landscape of mutations and epigenetics changes leading to transcriptomic and proteomic changes in cancers may be necessary to reveal them, many such cancers will reveal as their cell of origin a stem-like cell with networks of embryologic origin. Targeting these cells with precision, combination therapy has great promise and owes its origins in no small part to the knowledge and imagination exemplified in Knudson's review in Cancer Research in 1985 that set the field on the path to discover the anti-oncogene.