Lethal clues to cancer-cell vulnerability

Cancer cells often have mutations in anticancer genes that make their survival dependent on other genes. The gene-editing approach CRISPR–Cas9 offers a way to identify such vulnerabilities.
Felix Y. Feng is in the Department of Urology at the University of California, San Francisco, California 94158, USA; at the Helen Diller Family Comprehensive Cancer Center, San Francisco, California; and in the Department of Medicine and the Department of Radiation Oncology, University of California, San Francisco.

Search for this author in:

Luke A. Gilbert is in the Department of Urology at the University of California, San Francisco, California 94158, USA, and at the Helen Diller Family Comprehensive Cancer Center, San Francisco, California.

Search for this author in:

Suitable protein targets are needed to develop new anticancer drug-based treatments. Writing in Nature, Behan et al.1 and Chan et al.2, and, in eLife, Lieb et al.3, report that certain tumours that have deficiencies in a type of DNA-repair process require an enzyme called Werner syndrome ATP-dependent helicase (WRN) for their survival. If inhibitors of WRN are found, such molecules might be promising drug candidates for further testing.

Imagine a scenario in which scientists could perform an experiment that reveals how almost every gene in the human genome is dysregulated in cancer. Even better, what if such an investigation also offered a road map for how to select a target when trying to develop treatments that take aim at cancer cells, but are non-toxic to normal cells? A type of gene-editing technology called CRISPR–Cas9 enables just that in an approach termed functional genomics. Using this technique, the function of almost every gene in cell-based models of cancer (comprising human cells grown in vitro or in vivo animal models) can be perturbed, and the effect of each perturbation on cancer-cell survival can be measured.

CRISPR can be used to mutate, repress or activate any targeted human gene4,5. In functional genomics, gene function is assessed in a single experiment by growing a large number of cells and then perturbing one gene in each of the cells. The approach is aided by measuring the concentration in each sample of the DNA sequence that encodes an engineered RNA molecule (termed a single-guide RNA; sgRNA) that is needed for the CRISPR gene-editing process. When the DNA that encodes a particular sgRNA is present in a sample of cancer cells in such an experiment, this means that the gene that the sgRNA targets is not required for cell survival. However, if the sgRNA-encoding DNA sequence is not detected, this indicates that the gene targeted by the sgRNA is required for cancer-cell viability, and those cells containing it died, thereby eliminating the sgRNA-encoding sequence from the sample. This approach offers the possibility of systematically searching for genes that are crucial to tumour-cell survival in large collections of cancer cells that are representative of the diversity of tumour types in humans.

Behan et al. report their development of an online database that they call Project Score. It is a platform for cancer researchers that amalgamates Behan and colleagues’ large-scale data for genome-wide gene editing by CRISPR with previously published genomics information about the cancer models used. The resource consists of data for more than 5 million CRISPR-mediated perturbations undertaken to prevent the expression of individual genes (generating what are known as gene knockouts) in 324 cell-based models of cancer, representing 30 types of cancer in humans. This systematic effort has enabled the identification of genes on which cancer cells depend for survival, as well as those that drive cancer-cell proliferation.

In the database, the authors’ integrated analysis of this functional-genomics data is provided together with other data about the samples. For example, there is information about tumour-cell genomes, such as the sequence of the whole genome or that of the genome’s protein-coding regions. Also available are the gene-expression profiles for each cell-based cancer model, which reveal vulnerabilities in cancer cells that are associated with the specific tumour-driving alterations that exist naturally in each model. These include mutations in types of gene known as oncogenes and tumour suppressors that, respectively, are associated with promoting or blocking cancer formation. The platform developed by Behan et al. complements a similar effort called the Cancer Dependency Map6,7, and together these resources should accelerate cancer research. For example, these approaches could guide target prioritization for the development of treatments that might be effective in multiple types of cancer.

Many cancers in humans are driven by genetic mutations known as loss-of-function mutations, which inactivate genes so that they no longer encode functional proteins. In cancers, such mutations commonly occur in tumour-suppressor genes. Because these mutations result in the absence of a protein, they do not offer an opportunity for directly targeting an abnormal protein with drugs. However, loss-of-function mutations can render cancer cells dependent on specific genes through a principle known as synthetic lethality8,9. Synthetic lethality is a relationship between two genes in which the loss of both gene A and gene B is lethal, whereas the loss of either gene can be tolerated by cells. If we could understand how the inactivation of tumour-suppressor genes rewires cancer cells to drive tumour progression, this knowledge could be used to implement strategies that enable the specific killing of such mutated cancer cells.

The possibility of synthetic lethality pointing the way to fresh therapeutic approaches is exemplified by studies that demonstrate that combining loss-of-function mutations in the tumour-suppressor genes BRCA1 or BRCA2 with the inhibition of a type of enzyme called PARP imparts synthetic lethality9. BRCA1 and BRCA2 are often mutated in breast, ovarian and prostate cancers, and clinical trials that enrolled people with BRCA mutations have resulted in the approval of three PARP inhibitors for cancer treatment9. These and other successes have inspired many efforts to search for synthetically lethal genetic relationships involving commonly mutated tumour-suppressor genes such as TP53, PTEN and RB1. Despite the enormous potential for such efforts to yield drug targets, a challenge remains: how can researchers systematically search for synthetically lethal gene relationships? Now, functional-genomics approaches offer a way to proceed.

The CRISPR knockout data sets from Project Score and the Cancer Dependency Map were analysed by Behan et al. and Chan et al., respectively. Both groups found that WRN, a type of helicase that can unwind DNA and is a member of the RecQ family of proteins, is essential in cancers that have a type of genomic alteration called microsatellite instability (MSI). Lieb et al. found a similar synthetically lethal relationship between MSI and WRN using a functional-genomics approach involving a technique called RNA interference.

MSI is a common driver of cancer progression in a range of tumour types, including colon, gastric, endometrial and ovarian cancers. It arises when errors occur in a DNA-damage repair system called DNA mismatch repair. Inactivation of a number of different genes, for example, MLH1 and MSH2, can cause a deficiency in mismatch repair. Behan and Chan, and their respective colleagues, found that mutations in genes required for mismatch repair caused synthetic lethality if the gene that encoded WRN was also inhibited. They characterized this synthetic lethality using experiments that measured cellular DNA-repair defects and cell-death mechanisms in cells studied in vitro and in vivo. This discovery of a strong and specific synthetically lethal dependency represents a major step forward for efforts to develop approaches to treat cancers that have MSI.

The exact molecular mechanisms that underlie the specificity of this synthetically lethal interaction remain to be determined. For example, why does this WRN dependency occur only with tumours that have MSI, and not with tumours that have other forms of genomic instability? Interestingly, this genetic interaction is highly specific; experiments by Behan et al., Chan et al. and Lieb et al. demonstrated that repression of WRN, but not repression of the four other RecQ helicases that function in the same pathways as WRN, is synthetically lethal in cancers that have deficiencies in DNA mismatch repair. Next-generation functional-genomics approaches promise higher-resolution characterization of individual genetic interactions, which could reveal not only the genes that are involved in a process, but also how those genes function to affect the cell. Such approaches should also enable scientists to elucidate the mechanisms that underlie a particular example of synthetic lethality10,11.

WRN has both helicase activity and exonuclease activity (the ability to remove nucleotides from a strand of DNA). Behan et al., Chan et al. and Lieb et al. demonstrate that the disruption of WRN’s helicase activity, but not its exonuclease activity, is required for the synthetically lethal effect that they observed. There is a possibility that WRN could be targeted by small-molecule inhibitors. Further studies might enable the development of potent and specific WRN helicase inhibitors that could be tested in cancers that have MSI. These discoveries exemplify how a combined genomics and functional-genomics approach — characterizing the genetic alterations that are already present in cancer models and then assessing the effects of experimentally induced perturbations in further genes — can reveal important cancer-cell dependencies and provide a pathway towards therapeutic innovation.

Nature 568, 463-464 (2019)

doi: 10.1038/d41586-019-01086-w


  1. 1.

    Behan F. M. et al. Nature 568, 511–516 (2019).

  2. 2.

    Chan E. M. et al. Nature 568, 551–556 (2019).

  3. 3.

    Lieb, S. et al. eLife (2019).

  4. 4.

    Wang, H., La Russa, M. & Qi, L. S. Annu. Rev. Biochem. 85, 227–264 (2016).

  5. 5.

    Komor, A. C., Badran, A. H. & Liu, D. R. Cell 168, 20–36 (2017).

  6. 6.

    Tsherniak, A. et al. Cell 170, 564–576 (2017).

  7. 7.

    Meyers, R. M. et al. Nature Genet. 49, 1779–1784 (2017).

  8. 8.

    Hartman, J. L. IV, Garvik, B. & Hartwell, L. Science 291, 1001–1004 (2001).

  9. 9.

    Ashworth, A. & Lord, C. J. Nature Rev. Clin. Oncol. 15, 564–576 (2018).

  10. 10.

    Horlbeck, M. A. et al. Cell 174, 953–967 (2018).

  11. 11.

    Adamson, B. et al. Cell 167, 1867–1882 (2016).

Download references

Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up