Small size, big impact: how studies of small DNA tumour viruses revolutionized biology

Abstract

Intense study of three families of small tumour viruses with double-stranded DNA genomes, carried out over 50 years, has had a profound impact on biology. The polyomaviruses and papillomaviruses have circular DNA genomes of approximately 5000 and approximately 8000 base-pairs, respectively, and thus encode only a handful of proteins. Adenoviruses have a 32 000-base-pair linear DNA genome, still far smaller than the three billion-base-pair human genome. Members of all three virus families can transform cultured cells to tumorigenicity and cause tumours in experimental animals. Several human papillomaviruses (HPV) and at least one polyomavirus are oncogenic in humans. Early analysis of these viruses, particularly the polyomavirus SV40, led to the development of many powerful experimental tools, including restriction mapping, site-directed mutagenesis, gene transfer, genome-wide sequencing and recombinant DNA. These tools have since been refined and used to study cellular genes, revolutionizing our understanding of biology. These tools were also applied to the viruses themselves. Analysis of the virus life cycle and the effect of these viruses on cells yielded important new insights into many aspects of gene expression, DNA replication, cell biology and carcinogenesis. These studies have also led to vaccination strategies to prevent infection and cancer in humans.

This article is part of the theme issue ‘Silent cancer agents: multi-disciplinary modelling of human DNA oncoviruses’.

1. Introduction

Over the past 50 years, our understanding of genes and cells has undergone a revolution. Studies of small DNA tumour viruses—specifically polyomaviruses, papillomaviruses and adenoviruses—played an outsized role in this revolution. This review presents some of the historical highlights of research on these viruses, with a focus on the development of powerful tools to study viral (and later cellular) genomes and the insights that flowed from applying these tools to the viruses themselves. The reader is also referred to two earlier, excellent short reviews that consider some aspects of these topics [1,2]. Studies of hepatitis B virus (HBV), another small DNA virus that can cause cancer in humans, did not play a major role in the revolution led by studies of the polyomaviruses, papillomaviruses and adenoviruses. Readers interested in the history of HBV research are directed to the comprehensive review by Block and colleagues [3].

2. Setting the stage

The late 1960s was a heady time in genetics. It was a century after Mendel's discoveries (and more than 60 years after their rediscovery) and a quarter century after Oswald Avery discovered that DNA carried the genetic information. Studies of sickle cell anaemia had shown that a mutant gene resulted in the synthesis of a protein with an altered amino acid sequence, and the double-helical structure of DNA provided a likely explanation: the linear nucleotide sequence of DNA was somehow converted into the amino acid sequence of proteins. Messenger RNA had recently been discovered and provided a conduit for this information flow, formalized by Francis Crick as the central dogma—DNA to RNA to protein, and the biochemists had cracked the genetic code and revealed the fundamental mechanism of protein synthesis.

There was a looming problem, though. The information content of DNA, carried by the sequence of nucleotides, did not lend itself to biochemical analysis. DNA is, frankly, a fairly boring molecule. Unlike proteins, which exist in seemingly almost unlimited variety, DNA is a bland, monotonous chemical that barely differed from one organism to another, an objection that spawned the initial resistance to Avery's momentous discovery. It was easy to think of DNA as a featureless railroad track: the rails are the sugar-phosphate backbone and the ties are the base-pairs held together by hydrogen bonds. Genetic crosses had dissected genes at ever finer resolution, but could genes be understood at the biochemical level?

3. Powerful new tools

The solution to this dilemma came from an unexpected source: viruses that cause cancer. In 1911, a Long Island farmer brought a chicken bearing a sarcoma to Peyton Rous at the Rockefeller Institute in New York City, and Rous discovered that the tumour was transmissible to healthy birds [4]. Astoundingly, the material responsible for tumour formation retained activity in cell-free extracts and could pass through a filter that retained bacteria—it was a virus! The Rous sarcoma virus, as it is now known, contains an RNA genome, but Rous and Beard discovered in the mid-1930s that a papillomavirus isolated by Richard Shope could cause cancer in rabbits [5], and it was later discovered that the murine polyomavirus caused a plethora of tumours in mice, hence its name: poly-oma, ‘many tumours’ [6,7]. Unlike Rous sarcoma virus, papillomaviruses and polyomaviruses contain DNA genomes.

Viruses are ideal for genetic analysis because they replicate rapidly, have huge population sizes and often mutate with abandon. And they are small. Instead of the enormous number of genes in a cell, many viruses contain a mere handful of genes. Importantly, viruses have an extracellular phase, so the virus particle and its genome can be isolated in pure form from infected cells. Early biophysical studies of papillomavirus and polyomavirus genomes revealed that they consist of double-stranded DNA, like cell genomes, but are small and circular [8,9]. Instead of a transcontinental railroad, it was a child's model train set. But even with this greatly simplified system, it was not clear how biochemical analysis could be extended to individual genes.

The key advance emerged from the discovery that bacterial cells contain systems to protect themselves from invaders such as bacteriophage. Based on insightful genetic and biochemical studies, Werner Arber and Matthew Meselson deduced that many bacterial species inactivated (‘restricted’) foreign DNA, but once the DNA was passaged through the bacteria, it was modified so that it could now infect cells with impunity [10,11]. This restriction/modification system is populated by restriction endonucleases that degrade the incoming foreign DNA and by enzymes that chemically modify the surviving DNA so that cleavage is inhibited during the next round of infection. Bacteria have evolved multiple defence mechanisms to protect themselves from foreign RNA and DNA—restriction/modification, RNA interference and CRISPR. It is striking that all these mechanisms have been developed into powerful experimental tools to manipulate DNA, regulate gene expression and edit cellular genomes in situ.

In 1970, Hamilton Smith and his then-postdoctoral fellow Thomas Kelly purified and characterized a restriction endonuclease from Hemophilus influenzae and showed for the first time that a restriction endonuclease cleaved DNA at specific, short sequences of nucleotides [12]. (Their enzyme preparation, unbeknownst to them at the time, was actually a mixture of two endonucleases, but bacteriophage T7 DNA used in these experiments fortuitously contained cleavage sites for only one of them, greatly simplifying the analysis.) Smith told his virologist colleague, Daniel Nathans, about this finding, and Nathans realized that these enzymes could be used to generate specific fragments of DNA for analysis. By analogy to the specific proteases used in protein analysis, they could be ‘the trypsins and chymotrypsins for DNA’ [13]. To test this idea, Nathans digested DNA from the simian polyomavirus, SV40, a small DNA virus that causes tumours in hamsters and was originally discovered as a contaminant of poliovirus vaccines [14]! In a landmark paper published in Proceedings of the National Academy of Sciences in 1971, Danna and Nathans showed that restriction endonuclease cleavage generated a small number of discrete DNA fragments that could be separated by polyacrylamide gel electrophoresis (figure 1) [15]. Each fragment was present in a single copy of the circular viral genome and, in aggregate, the fragments represented the entire approximately 5000-base-pair genome. In the visionary discussion of this paper, Danna and Nathans outlined how a physical map of the viral genome could be constructed from these DNA fragments, onto which genetic elements could be located.

Figure 1.

Restriction endonucleases generate discrete DNA fragments. Radiolabelled SV40 DNA was digested with restriction endonucleases from Hemophilus influenzae and subjected to polyacrylamide gel electrophoresis followed by autoradiography. Origin of the gel is at the left. Reproduced from Danna & Nathans [15] with permission.

Nathans et al. constructed such a ‘cleavage’ or ‘restriction’ map by placing the DNA cleavage products in relation to each other, revealing that each fragment had a fixed position in the viral genome [16]. Genes and signals could then be located on the map. First to fall was the origin of SV40 DNA replication, the first signal assigned to a discrete DNA segment in eukaryotic cells [17]. With George Khoury and others, he also mapped the early and late viral transcription units that encoded viral replication and transformation functions [18,19]. A number of other laboratories soon joined the effort to construct and analyse these maps, initially including those led by Paul Berg and Joseph Sambrook for SV40, and Clyde Hutchinson for bacteriophage ϕX174 [20–22]. The ultimate physical map is the nucleotide sequence of the genome, which was determined within five years for SV40 and ϕX174 [23–25]. The genomic sequencing era was born.

As well as being used to construct restriction maps and locate genes and other genetic elements on these maps, restriction endonucleases were used to determine the positions of mutations in the SV40 genome and, later, to construct mutations at predetermined sites in the genome [26–28]. Hit-or-miss random mutagenesis and selection were supplemented with, and to a large extent replaced by, construction of mutations in specific genes and signals to test hypotheses regarding function. This was the ‘New Genetics’ described by Nathans in 1978, when he shared the Nobel Prize with Smith and Arber for the discovery and application of restriction endonucleases [13].

Initially, the use of restriction endonucleases was largely confined to viral genomes because digestion of cellular DNA generated millions of DNA fragments, a mixture far too complex for meaningful analysis. An approach was needed to convert individual cellular genes into short DNA segments that could be isolated in pure form, like viral DNA purified from virus particles. Molecular cloning solved the problem. SV40-sized bits of DNA were inserted into bacterial plasmid and phage vectors and propagated in bacteria. By isolating DNA from individual bacterial colonies or plaques carrying these vectors, a clonal population of identical DNA molecules was obtained. In principle, a limitless supply of any gene was in hand. And not surprisingly, the first DNA segment from eukaryotic cells to be inserted into a bacterial vector was the SV40 genome itself [29]. Another Nobel Prize, this one for Paul Berg. Other key tools developed to study viral genes about the same time were transfection techniques to deliver viral DNA into cells so their function could be assayed [30,31]. Transfer of adenovirus oncogenes generated HEK293 cells, a laboratory workhorse with many applications, including the production of retroviruses and papillomaviruses [32]. With the development of increasingly powerful and sophisticated techniques to clone, sequence, manipulate and analyse DNA, the approaches invented to study small viral genomes could be applied to cellular genes. These studies revolutionized our understanding of biology and allowed the production of useful proteins on an industrial scale.

4. Fantastic discoveries in basic biology

The small DNA tumour viruses were not just handmaidens for the development of useful tools to study cellular DNA. Analysis of the viruses themselves repeatedly revealed fundamental aspects of biology that apply to cellular genes as well. One of the most striking and unexpected discoveries was made by Philip Sharp and Richard Roberts and their colleagues while analysing mRNAs encoding the late proteins of adenovirus [33,34]. Instead of the expected co-linear relationship between the gene and its message, certain segments of the primary transcript were removed during biogenesis of the mature mRNA. This process, known as mRNA splicing, is widespread in higher eukaryotes and its discovery fundamentally changed our understanding of cellular gene expression. Another Nobel Prize for the DNA tumour virologists. Alternative splicing, in which the same primary transcript can use different splice donor and acceptor sites to generate mRNAs that encode different protein products, was discovered by studies of the mRNAs produced from the SV40 major early region [35].

The analysis of SV40 also revealed many other new aspects of gene expression. Transcription of eukaryotic genes was thought to be controlled by the activity of promotors, but studies of SV40 showed that some viral sequences could stimulate transcription of viral and heterologous genes at a distance, and that the orientation of these elements relative to the direction of transcription was not fixed [36,37]. Analysis of these sequences identified a new regulatory element, transcriptional enhancers, which stimulated transcription in a position- and orientation-independent fashion. Further in vitro studies by Robert Tjian and colleagues identified site-specific DNA binding proteins that bind the SV40 enhancer and mediate transcriptional activation [38,39]. In order to enter the nucleus to bind DNA, most transcription factors and other nuclear proteins contain a short protein segment comprised of basic amino acids that directs them to the nucleus. The first such nuclear localization sequence (NLS) was discovered in SV40 Large T antigen, normally a nuclear protein, because some T antigen mutants displayed cytoplasmic distribution [40,41]. mRNA synthesis is not only controlled by transcription initiation at the 5′ end of genes and by splicing in internal mRNA segments. Fitzgerald and Shenk identified a novel AU-rich regulatory element in the 3′ end of the SV40 early region that specified the site of mRNA polyadenylation [42]. Most eukaryotic genes contain this element, typically AAUAAA, which controls mRNA stability and other aspects of mRNA synthesis and function. Thus, many key features of eukaryotic gene expression—transcription factors, nuclear localization signals, transcriptional enhancers, mRNA splicing, alternative splicing and control of polyadenylation—were discovered by studies of small DNA tumour viruses.

Crucial features of eukaryotic DNA replication were also elucidated by studies of these viruses. These studies were made possible by the development of in vitro viral DNA replication systems by Thomas Kelly, who had previously discovered site-specific cleavage by restriction endonucleases with Hamilton Smith. First, he developed a system for replication of adenovirus DNA, in which viral DNA was added to extracts of uninfected cells and DNA synthesis was monitored by incorporation of radiolabelled nucleotides [43]. He showed that adenovirus DNA replication proceeded by an unusual protein-priming mechanism. A few years later, he developed a similar in vitro system for SV40 DNA replication [44]. SV40 DNA replication in this system displayed several features that defined DNA replication in infected cells: dependence on the viral replication origin identified by Nathans, requirement for viral large T antigen, and proteins from permissive monkey cells. Because replication of SV40 is thought to closely mimic cellular DNA replication, this finding galvanized the field, and Kelly, Gerald Hurwitz, and Bruce Stillman conducted tour-de-force biochemical experiments that identified and purified the proteins required for SV40 DNA replication, so that the cell extract could be replaced with purified cellular proteins that supported replication [45]. This simplification allowed detailed mechanistic analysis of this complex process (e.g. [46]). These same proteins were later shown to be required for cellular as well as viral DNA replication. Similarly, SV40 large T antigen, a DNA helicase that binds to the viral replication origin and replication forks, could be replaced with cellular counterparts. Thus, analysis of this simple viral system led to fundamental insights into DNA replication, the central event in cellular propagation.

5. Tumour viruses as keys to understanding carcinogenesis

A prime rationale for studying tumour viruses was the hope that such studies would provide insight into mechanisms of cancer formation. Indeed, studies of small DNA tumour viruses, primarily in cultured cells, have had a profound impact on our understanding of carcinogenesis and other aspects of cell growth control. Normal cells have strict requirements for survival and proliferation, but cells expressing oncogenes display altered growth properties that resemble the properties of cells isolated from tumours. For example, while normal cultured fibroblasts display contact inhibition, require anchorage for growth, depend on serum in the growth medium and display a finite replicative lifespan, transformed cells expressing oncogenes escape contact inhibition and form foci, form colonies in semi-solid medium, display reduced serum requirements and often grow indefinitely.

Although individual retroviral oncogenes can transform cells, DNA tumour viruses often contain multiple oncogenes. Experiments by Francois Cuzin and colleagues identified large T antigen as the major oncogene of murine polyomavirus, but Robert Kamen et al. assigned this activity to middle T antigen, which shares some sequences with large T but is mostly a totally different protein owing to alternative splicing [47,48]. This conflict was resolved with the realization that the two groups used different assays for transformation. Cuzin et al. assayed induction of serum independence, whereas Kamen et al. measured focus formation. Follow-up studies showed that transformation was not an all-or-nothing phenomenon and that different oncogenes had different activities. Importantly, co-transfer of both genes resulted in the fully transformed phenotype [47]. Alex Van der Eb and colleagues showed that transformation by adenovirus also required two viral oncogenes, E1A and E1B [49], and similar cooperation was soon reported for myc and ras oncogenes and often occurs in human cancer [50].

The mechanism of action of middle T antigen, a membrane-anchored protein, was soon determined. Co-immunoprecipitation experiments performed independently by Thomas Benjamin, Alan Smith and Tony Hunter revealed that middle T antigen was not an enzyme but rather was constitutively associated with a novel kinase activity in infected cells [51–53]. Remarkably, the middle T-associated kinase transferred phosphate to tyrosines, not serines or threonines, on its protein substrates (figure 2) [53]. This was the first example of tyrosine kinase activity, which has since been shown to play a major role in cell signalling and cancer. Further experiments identified this middle T-associated tyrosine kinase as the product of pp60^c-src proto-oncogene, which had recently been identified as the progenitor of the oncogene present in Rous sarcoma virus [54]. Joan Brugge et al. and Sara Courtneidge then showed that middle T binding activated the kinase activity of pp60^c-src, correlating with middle T antigen transforming activity [55,56]. These experiments established that middle T antigen transforms cells by activating a cellular oncoprotein. It was later shown that once recruited to middle T by protein phosphatase 2A (PP2A), activated pp60^c-src phosphorylated multiple tyrosine residues in the cytoplasmic domain of middle T antigen, generating docking sites for cellular signalling proteins containing SH2 domains (reviewed in [57]). Thus, middle T antigen mimicked an activated growth factor receptor tyrosine kinase, which typically autophosphorylates cytoplasmic tyrosines in response to ligand binding to generate a signalling complex. Analysis of the SH2 proteins associated with middle T antigen identified phosphoinositol 3′ kinase and the novel PI(3,4)P2 metabolite [58,59], which is important in signalling and cancer.

Figure 2.

Discovery of tyrosine phosphorylation. Middle T antigen was immunoprecipitated from polyomavirus-infected cells, and immunoprecipitates were labelled by incubation with ³²P. Middle T antigen was purified, and the products of partial acid hydrolysis were separated by electrophoresis (a) or on two dimensions by electrophoresis and chromatography (b). Phosphorylated amino acids were detected by autoradiography and identified by reference to standards indicated by dotted circles. P.THR, phospho-threonine; P.TYR, phospho-tyrosine; P.SER, phospho-serine. Reproduced from Eckhart et al. [53] with permission.

The bizarre bovine papillomavirus (BPV) E5 protein uses a related mechanism to transform cells. Only 44-residues long and extremely hydrophobic, E5 is essentially a free-standing transmembrane domain [60]. The E5 protein binds specifically to the transmembrane domain of the platelet-derived growth factor (PBGF) β receptor, causing sustained ligand-independent activation of the receptor and cell transformation [61–63]. Thus, while middle T antigen transforms cells by mimicking an activated growth factor receptor, E5 activates a cellular growth factor receptor. These experiments established the general principle that small transmembrane proteins can act in trans by binding to the transmembrane domains of larger cellular proteins and modulating their activity. Based on this insight, artificial transmembrane proteins modelled on BPV E5 have been constructed that specifically modulate various cellular targets. Called traptamers (for transmembrane protein aptamers), these artificial proteins can cause cellular transformation by activating the PDGF receptor, induce erythroid differentiation by activating the erythropoietin receptor or inhibit HIV infection by downregulating the HIV co-receptor, CCR5 (e.g. [64]). Some of these traptamers are chemically the simplest biologically active proteins described [65].

With the discovery that middle T antigen activated an oncogene product, the search was on for other cellular oncoproteins that mediate viral transformation. A likely candidate for this approach was the adenovirus E1A oncoprotein. E1A shares a short, conserved sequence with SV40 large T antigen, which also transforms cells. This conserved region 2 (CR2) was required for E1A-mediated transformation, and Elizabeth Moran showed that it could be replaced with the corresponding sequence from T antigen, implying that it acted in a modular fashion, perhaps by binding to and activating a cellular protein that caused transformation [66]. When Edward Harlow immunoprecipitated the E1A oncoprotein from radiolabelled adenovirus-infected cells, a number of associated cellular proteins were immunoprecipitated in addition to the viral protein. Teaming up with Robert Weinberg, they soon identified some of the co-precipitating proteins as the products of the retinoblastoma (RB) gene and other members of the RB gene family [67]. The RB protein had been previously implicated in a hereditary human cancer, but the genetics of RB formation suggested that loss of RB function, not gain of function, was responsible for tumour formation. So, the RB gene was not an oncogene, but an ‘anti-oncogene’ that encoded a tumour suppressor protein. This result caused a paradigm shift: viral oncoproteins can act not only by activating the products of oncogenes, but also by inactivating the products of tumour suppressor genes.

E1A is not alone in binding the RB protein. David Livingston and others soon showed that CR2 of SV40 large T antigen also binds the RB protein and that mutations in CR2 that disrupt this interaction inhibit transforming activity, providing genetic evidence that T antigen/RB binding is important for transformation [68]. The papillomaviruses then joined the party, when Edward Harlow and Peter Howley showed that RB binds the E7 oncoprotein from the high-risk human papillomaviruses (HPV), the virus types that cause cervical cancer, whereas E7 from low-risk HPV types shows greatly reduced RB binding [69]. More recently, Chang and Moore and colleagues, who previously discovered Kaposi sarcoma-associated herpes virus, sequenced tumour cell RNA to discover a new human tumour virus in the aggressive skin cancer, Merkel cell carcinoma [70]. Like SV40, Merkel Cell polyomavirus (MCPyV) encodes large T antigen that binds RB, an interaction that is important for the ability of this virus to stimulate cell growth [71,72].

Several DNA tumour virus oncoproteins also bind and neutralize a second tumour suppressor. The first key discovery, made independently by Lane & Crawford and Linzer & Levine, was that SV40 large T antigen formed a stable complex with a 54 kDa cellular protein, initially named non-viral T antigen [73,74]. By analogy to the situation of murine polyomavirus middle T antigen and pp60^c-src, it seemed likely that this cellular protein was an oncoprotein activated by large T antigen to induce transformation. Consistent with this view, the gene encoding this protein, now known as p53, could act as an oncogene (e.g. [75]). However, after some twists and turns, it became evident that p53 instead usually acts as a tumour suppressor [76–78]. In fact, more than half of all human cancers harbour inactivating or dominant-negative mutations in p53. As was the case for RB, oncoproteins from multiple DNA tumour virus families bind and inactivate p53. In addition to large T antigen, adenovirus E1B and E4orf6 and high-risk HPV E6 also neutralize p53 function to support transformation and tumorigenesis [79–81]. Studies of the mechanism of E6 action resulted in the identification of E6-associated protein (E6AP), the first mammalian ubiquitin E3 ligase. The E6 protein binds to E6AP, hijacking it to ubiquitinate p53 and tag it for proteasome-mediated degradation [82,83].

The central role of DNA tumour viruses in our understanding of cancer is illustrated nowhere more clearly than by the fact that studies of these viruses led to the discovery and characterization of RB and p53, the two major tumour suppressor pathways disrupted in human cancer. And the two pathways are connected. p53 is a transcription factor that activates expression of genes that regulate the activity of the RB protein. RB itself regulates the action of the E2F1 transcription factor and other members of the E2F family that control expression of genes required for cell cycle progression. E2F1 was discovered by Joseph Nevins and colleagues because of its ability to regulate transcription of the adenovirus E2 gene and has since been shown to play a major role in cell cycle control [84]. The discovery that three entirely distinct DNA tumour virus families all bind to RB and p53 highlights the importance of these interactions in viral replication, presumably to stimulate the cellular DNA replication apparatus to support viral DNA replication or to counter host defences against infection [1]. It should also be pointed out that in addition to the proteins mentioned here, DNA tumour virus oncogenes bind many other cellular proteins that are also involved in transformation and tumorigenesis.

A concept that emerged from the study of mouse models of cancer is oncogene addiction, the idea that cancer cells require continued expression of oncogenes, so that extinction of oncogenes causes the cells to cease proliferation or die [85]. This concept has been validated in human cancer patients with the introduction of effective therapy that targets activated oncogene products, such as βcR-abl in chronic myelogenous leukaemia. The recognition of oncogene addiction in mouse models was preceded by numerous studies showing that cervical carcinoma cells required continuous expression of the HPV oncogenes for growth in cell culture. The vast majority of cervical cancers express the HPV E6 and E7 oncogenes. In cervical cancer cell lines including HeLa cells, which contain integrated HPV18 DNA, these genes have been expressed for decades. Early RNA interference studies showed that repression of HPV E6/7 expression slowed the proliferation of cervical cancer cell lines [86]. More dramatic E6/7 repression was achieved by expression of the BPV E2 regulatory gene, resulting in restoration of RB and p53 activity and profound growth arrest (figure 3) [87,88]. Similar results were obtained in freshly derived cervical cancer cell strains, demonstrating that dependence on the HPV oncogenes is an intrinsic property of cervical cancer cells, not just a secondary response to long-term passage in culture. Such oncogene dependence suggests that strategies that interfere with E6/7 expression or action may have an anti-tumour effect.

Figure 3.

Cervical cancer cells are addicted to HPV oncogenes. HeLa cervical cancer cells were infected with a recombinant virus that expresses the bovine papillomavirus E2 transcription factor. At various times after infection, cell lysates were probed for expression of HPV18 E6/E7 oncogene mRNA, p53, and p105^Rb, as indicated. The arrowhead indicates the hypophosphorylated, transcriptionally active form of p105^Rb.

6. Vaccines to prevent cancer

Epidemiological studies have revealed that some human cancers are caused by DNA tumour virus infection. Numerically, the most important are the HPVs, which are responsible for essentially all uterine cervical cancer, variable amounts of other anogenital cancers, the majority of oropharyngeal cancer and some non-melanoma skin cancers. In aggregate, 5% of human cancer is thought to be triggered by HPV infection [89]. As is the case for other DNA tumour viruses in humans, most people infected with HPV do not develop HPV-associated cancer, and cancers arise years or decades after the initial infection. The identification and molecular cloning of HPV types associated with cervical cancer was pioneered by Harald zur Hausen, after years of unsuccessful attempts by many laboratories to implicate herpesviruses in these tumours [90,91]. These experiments led to the discovery of the role of HPV in this cancer and to a Nobel Prize for zur Hausen. As outlined in previous sections, molecular characterization of the HPV and other DNA tumour viruses has yielded great insight into the mechanism of action of these human carcinogens. These studies also showed that the HPV L1 major capsid protein could self-assemble to form non-infectious virus-like particles (VLPs) devoid of viral DNA, similar to earlier observations with polyomavirus VP1 [92–94]. Because these VLPs closely resemble the authentic virus capsid, they are effective vaccines, eliciting both humoral and cell-mediated immunity against L1 [95–98]. Extensive clinical trials and follow-up monitoring showed that these vaccines protect against infection by the HPV types in the vaccine and prevent the formation of genital warts and cervical precancerous lesions [99,100]. Because of the long time-lag between HPV infection and cancer formation, we will not know for years whether these vaccines actually prevent cancer. Nevertheless, it seems likely that our deep understanding of DNA tumour viruses, which has already elucidated fundamental mechanisms of carcinogenesis as well as many other aspects of biology, will also translate into effective approaches to prevent cancer in humans.

Data accessibility

This article has no additional data.

Competing interests

I have no competing interests.

Funding

Research in the DiMaio laboratory is supported by grants from the National Institutes of Health to D.D. (CA037157, AI102876, CA016038).