Julian Downward: Projects

The laboratory is interested in the mechanisms by which cancer cells become addicted to growth and survival signals generated by activated oncogenes and loss of tumour suppressor genes. We particularly focus on identifying unique dependencies of oncogene addicted cancer cells that might be targetable in the therapy of human cancer.

Investigation of mechanisms of oncogene driven transformation and drug resistance

Much of the work in the group has focused on the RAS family of oncogenes and the signalling pathways that they control. RAS genes are activated by point mutation in some 20 per cent of all human tumours and are known to play a key role in the establishment of the transformed phenotype. While the signalling pathways activated by RAS are well characterised, it remains a major challenge to identify what proteins are selectively important in the establishment and maintenance of the RAS transformed phenotype and may therefore act as potential therapeutic targets for cancer treatment.

We have employed functional genomics approaches using gene silencing by genome-scale libraries of RNA interference agents to investigate this, along with the use of genetically engineered mouse models of human cancer.

When cells become progressively transformed during the evolution of cancer, they suffer stresses that are not seen by normal cells and become increasingly dependent on stress management pathways. This means that the tumour cells show a unique set of dependencies, both on the oncogenic drivers and also on stress handling pathways, sometimes termed oncogene addiction and non-oncogene addiction, respectively.

We have investigated these dependencies by RNAi screening, comparing a cancer cell line containing an activated KRAS allele with a normal ('isogenic') derivative in which this has been removed (Wang et al., 2010; Oncogene. 29(33): 4658-70), and also using a panel of 30 or so cancer cell lines, half of which were mutant and half wild type for KRAS (Steckel et al., 2012; Cell Research. 22: 1227).

This approach uncovered proteins whose therapeutic targeting might be expected to provide differential toxicity towards KRAS mutant tumour cells. One example of a determinant of non-oncogene addiction found in this way, the transcription factor GATA2, has been investigated in detail. Using genetic mouse models, we could show that the development and continued maintenance of KRAS induced lung cancer is uniquely dependent on the expression of GATA2: when this is blocked, the lung adenocarcinomas regress completely. As a transcription factor, GATA2 itself is not likely to be a good drug target. However, using a range of methods to investigate the transcriptional programme controlled by GATA2, we identified a number of downstream pathways that can be inhibited by existing drugs.

Using a combination of two such drugs we have been able to show an impressive therapeutic response in KRAS induced mouse lung cancer models (Kumar et al., 2012; Cell. 149(3): 642-55). We are now investigating the possibility of studying this drug combination in a clinical setting.

The role of phosphatidylinositol 3-kinase in RAS-driven oncogenesis

RAS proteins signal through direct interaction with a number of effector enzymes, including type I phosphatidylinositol 3-kinases (PI3Ks).

Mice with mutations in the RAS binding domain (RBD) of the Pik3ca gene encoding the PI3K catalytic p110α isoform are highly resistant to endogenous KRAS oncogene induced lung tumourigenesis and HRAS oncogene induced skin carcinogenesis (Gupta et al., 2007; Cell. 129(5): 957-68).

The interaction of RAS with p110α is thus required in vivo for RAS-driven tumour formation. We have also generated mice with inducible expression of the inactivating mutation in the RAS binding domain of p110α so that the requirement of this interaction for maintenance of established tumours can be assessed.

Blocking the RAS/p110α interaction causes partial regression and stasis of tumours, although more complete regression requires coordinate inhibition of the MEK pathway (Castellano et al., Cancer Cell. In Press).

While combined inhibition of the RAS effector pathways MEK and PI3K can cause impressive tumour regression, this combination has high toxicity that may be problematic in the clinic. We have sought to find less toxic ways of inhibiting PI3K in KRAS mutant lung cancer cells using a drug library screen and have found that inhibition of IGF1 receptor allows this. It appears that the activation of PI3K by mutant KRAS in lung cancer cells is dependent on basal signalling by IGF1 receptor. The combination of MEK and IGF1 receptor inhibition shows potential in preclinical models of KRAS mutant lung cancer (Molina-Arcas et al., 2013; Cancer Discovery. 3(5): 548-63).

We have also created a mouse with inactivating mutations in the RAS binding domain of p110β the other ubiquitously expressed PI 3-kinase catalytic subunit isoform, and are testing the effects of this mutation on tumour initiation and maintenance, especially in the context of PTEN deletion, where p110β is thought to be particularly important.

Our investigations with p110β have led us to the surprising observation that this isoform is not controlled by direct interaction with RAS, unlike p110α, γ and δ, but rather that the RBD of p110β interacts directly with a number of other small GTPases with distinct biological function - the RAC and CDC42 proteins. This has led us to a significantly revised model of how extracellular stimuli, especially those signalling through G protein coupled receptors, activate the PI 3-kinase activity of p110β, and the importance of this mechanism in cancer metastasis and also fibrosis (Fritsch et al., 2013; Cell. 153(5): 1050-63).

Figure 1

Figure 1. Differential interaction of RAS and RHO subfamily GTPases with different type I PI3K isoforms. RAS subfamily proteins, including the oncoproteins KRAS, HRAS and NRAS, interact with PI3K p110α, γ and δ isoforms, whereas the RHO subfamily proteins RAC and CDC42 interact with PI3K p110β isoform. In all cases, the interaction is dependent on the GTPase being in the activated GTP-bound conformation, acts through the conserved RBD region of p110 and results in activation of the PI3K enzymatic activity (Fritsch et al., 2013).


Julian Downward

+44 (0)20 379 61838

  • Qualifications and history
  • 1986 PhD in Natural Sciences, Imperial College, London University, UK
  • 1986 Postdoctoral Research Fellow, Massachusetts Institute of Technology, USA
  • 1989 Established lab at the Imperial Cancer Research Fund (in 2002 the Imperial Cancer Research Fund became Cancer Research UK)
  • 2015 Associate Research Director, the Francis Crick Institute, London, UK