Halazonetis Lab

Research Projects

Oncogene-induced DNA replication stress

The long-term goal of our research is to understand cancer at the molecular level, aiming to contribute towards the development of novel, non-toxic therapies. While this goal is very ambitious, I believe that through the concerted efforts of many dedicated laboratories, this goal will be achieved.

A first step for effective therapies is to identify differences between normal cells and cancer cells. A few years ago our laboratory, in collaboration with Vassilis Gorgoulis (University of Athens Medical School) and, independently, the laboratory of J. Bartek (Danish Cancer Society) observed that human precancerous lesions and established cancers are characterized by the presence of DNA double-strand breaks (DSBs) and DNA replication stress. In contrast, in normal cells, even from highly replicating tissues, we did not observe signs of DNA damage. Thus, the presence of DNA replication stress can distinguish normal cells from cancer cells.

Our current model to explain the presence of DNA damage in human cancers posits that oncogenes interfere with the process of DNA replication, leading to fork stalling and fork collapse. Fork collapse, i.e. the dissociation of the replisome from the replication fork, is often associated with the formation of DNA DSBs. In turn, these breaks activate p53, which, in tun, induces apoptosis or senescence. Thus, the DNA damage response pathway via p53 restrains cancer development. However, once p53 is inactivated, the precancerous lesions escape apoptosis and senescence and progress to become cancers. Another consequence of replication fork collapse and DNA DSBs is genomic instability arising from error-prone repair. Therefore, our model can explain two key features of cancer: why p53 is the most frequently mutated gene in human cancers and why cancers have genomic instability (Figure 1).

Figure 1. Model for oncogene-induced DNA replication stress in human cancers selecting for p53 mutations and driving genomic instability.
Figure 1. Model for oncogene-induced DNA replication stress in human cancers selecting for p53 mutations and driving genomic instability.

Oncogene-induced DNA replication stress leads to a mutator phenotype in human cancers

Most human cancers are characterized by the presence of thousands of point mutations in their genome. The mechanisms accounting for the presence of these mutations in sporadic cancers are still being deciphered. According to one hypothesis, point mutations accumulate during normal aging and the mutations in cancers simply reflect this process. Others have argued that the rate of acquisition of mutations is higher in precancerous cells and cancers than in normal cells and tissues. To distinguish between these two models, we sequenced the exomes of human colon adenomas, which are precancerous lesions. The number of mutations in these lesions was not proportional to the age of the patient, but rather was proportional to the size of the adenoma, arguing that mutations were acquired in the precancerous stage. Based on the distribution of the mutations in the genome we propose that stalled or collapsed replication forks in precancerous cells result in exposure of single-stranded DNA, which is highly susceptible to mutagenesis (Figure 2).

Figure 2. Mechanism for increased rate of acquisition of point mutations due to DNA replication stress in human precancerous lesions.
Figure 2. Mechanism for increased rate of acquisition of point mutations due to DNA replication stress in human precancerous lesions.

Break-induced replication is a major pathway for repair of collapsed DNA replication forks in human cancers

Collapsed replication forks in human cancers need to be repaired to allow DNA replication to resume. We have identified break-induced replication (BIR) as a major pathway for repair of collapsed forks in cancer cells. BIR is a variant homologous recombination pathway that utilizes POLD3 and POLD4, two subunits of DNA polymerase delta, to initiate replication at collapsed forks. Depletion of POLD3 or POLD4 inhibits proliferation of cancer cells, without affecting normal cells. Repair by BIR should normally not lead to gross genomic rearrangements, but occasionally segmental genomic duplications with microhomology junctions can arise. This specific type of duplication is very common in human cancers, suggesting that BIR is relevant in vivo. Interestingly, replication forks initiated by BIR differ from normal forks initiated at origins of replication. BIR-initiated forks apparently retain the D-loop formed during strand invasion and replicate the DNA in a conservative manner (Figure 3). It is likely that the differences between BIR-initiated and origin-initiated forks could be eventually exploited for cancer therapies.

Figure 3. Comparison of BIR-initiated and origin-initiated forks. BIR-initiated forks retain the D-loop formed during strand invasion and replicate DNA in a conservative manner.
Figure 3. Comparison of BIR-initiated and origin-initiated forks. BIR-initiated forks retain the D-loop formed during strand invasion and replicate DNA in a conservative manner.

Adult Tissue-Specific Stem Cells

In recent years the laboratory of Hans Clevers (Hubrecht Institute) has succeeded in growing in vitro intestinal organoids, as well as organoids from other mouse and human tissues. We are also culturing these organoids, as a system to study cancer development (Figure 4). We aim to recapitulate the early steps of cancer development by introducing oncogenes into intestinal organoids. We should then be able to monitor the induction of DNA replication stress and genomic instability.

Figure 4. Intestinal organoid grown in tissue culture.
Figure 4. Intestinal organoid grown in tissue culture.

Protein Three-Dimensional Structures

The function of a protein depends on its structure. Thus, elucidating the structure of a protein can help the understanding of its function. As an example of this approach, we recently solved the crystal structure of a p53 tetramer bound to DNA. The structure reveals a conformational switch upon sequence-specific DNA binding (Figure 5). Using biophysical methods we have shown that the conformational switch allows p53 to have a thousand-fold longer residence time on specific DNA, as compared to non-specific DNA. Based on our measurements, the increased residence time on DNA is the most critical factor for discriminating between specific and non-specific DNA sites.

Figure 5. Structure of a human p53 tetramer bound to specific DNA.
Figure 5. Structure of a human p53 tetramer bound to specific DNA.