Publication highlights

This work reveals that amino acids, rather than fructose 1,6-bisphosphate, are the relevant cellular regulators of pyruvate kinase M2 (PKM2), a critical node in cancer metabolism. It further elucidates the molecular mechanism of PKM2 regulation by amino acids with a new algorithm that predicts allosteric pathways in proteins, a major and difficult problem in structural biology.

This study was the first to demonstrate that RAD51 paralogues bind to and structurally remodel the pre-synaptic RAD-51-ssDNA filament to a stabilised, “open”, and flexible conformation, which facilitates strand exchange with the template duplex. We showed that RAD51 paralogues act by binding the end of the presynaptic filament, which induces a conformational change that stabilises RAD-51 bound to ssDNA and primes the filament for strand exchange. These observations established for the first time the underlying mechanism of HR stimulation by Rad51 paralogues and revealed a new paradigm for the action of HR mediator proteins.

Evidence suggested that the telomere adopts a lasso-like t-loop configuration, which safeguards chromosome ends from being recognised as DNA double strand breaks. However, the regulation and physiological importance of t-loops in end-protection was uncertain. This study uncovered a phospho-switch in TRF2 that coordinates the timely assembly and disassembly of t-loops during the cell cycle, which protects telomeres from replication stress and an unscheduled DNA damage response. These results were the first to definitively establish the t-loop as a physiologically important structure required to suppress checkpoint activation at telomere ends.

Like many developing tissues, the vertebrate neural tube is patterned by antiparallel morphogen gradients. Using quantitative gene expression and signalling measurements we derived and validated a characteristic decoding map that relates morphogen input to the positional identity of neural progenitors. This revealed a strategy that minimises patterning errors in response to the joint input of noisy opposing gradients. The study illustrates how we integrate quantitative data, developmental and microfluidic experiments with phenological and mechanistic models.

Lentiviral IN proteins are notoriously poorly behaved in vitro, and the HIV 1 intasome has eluded structural biologists for over two decades. Prior research resulted in a collection of partial crystal and NMR structures that did not explain how lentiviral integrase synapses viral DNA ends. This paper described the first structure of the lentiviral intasome, solving the long-standing mystery and reconciling years of HIV-1 integrase structural biology and biochemistry.

Here, we described a cryo-EM structure of a retroviral intasome in a functional complex with a nucleosome. The structure revealed a multivalent interface of the viral integration machinery and chromatin, involving both gyres of nucleosomal DNA and histones. Whilst the histone octamer remains intact, the DNA is lifted from its surface to allow for strand transfer at highly preferred integration sites. These data provided a unique snapshot of an enzyme recognizing and acting upon nucleosomal DNA. The structure was the first to illustrate nucleosome flexibility facilitating a biological process and, as such, had far-reaching implications for chromosome biology.

Oesophageal adenocarcinoma shows high genetic heterogeneity making the identification of cancer drivers challenging. We developed a machine learning algorithm to identify cancer drivers in 261 oesophageal adenocarcinomas. Although most predicted drivers were rare or patient-specific, they all perturbed well-known cancer pathways. Using the recurrence of the same pathway perturbations rather than individual genes, we stratified patients into six groups different for their clinical features. We validated experimentally the contribution of these genes to disease progression and revealed acquired dependencies exploitable in therapy. This study described a new way to identify cancer drivers that we have recently further developed for application in precision oncology.

The MCM replicative helicase is loaded onto duplex DNA as a double hexamer. Here we use time-resolved cryo-EM to show that ORC binds to its high affinity binding site to load the first MCM hexamer. ORC then releases this site and it, or another ORC molecule then binds the B2 element, which contains a degenerate ORC binding site. This binding is stabilised by a novel interaction between the Orc6 subunit of ORC and the N-terminus of the MCM hexamer. ORC then recruits and loads the second hexamer by the same mechanism as the first hexamer. We employed newly developed in silico reconstitution approaches to describe the full context of the helicase loading reaction, studied on a near-native, chromatinised origin of replication. This study radically changes our approach to investigating chromosome replication with cryo-EM.

This paper shows that loading of the MCM double hexamer is a quasi-symmetrical reaction: two ORC molecules bound at two opposing sites of different affinity each recruit and load a single hexamer. The distance between the ORC binding sites is not critical. Subsequent work has provided further evidence for this from cryo-EM.

In this and the accompanying paper (Yeeles et al. 2017 Mol Cel 65, 105-116) we describe the reconstitution of full chromatin replication. We first identified all of the factors required for complete and rapid replication of naked DNA. Then we identified and characterised factors required to replicate chromatinised templates. We showed FACT is essential for chromatin replication, whilst nucleosome remodellers and histone acetylases help chromatin replication. In addition, chromatin enforces origin specificity and Okazaki fragment processing. Finally, we found that histones are efficiently transferred to nascent DNA.

This work establishes for the first time a link between oncogenic RAS signalling and increased immuno-suppressive expression of the immune checkpoint protein PD-L1. RAS signalling results in phosphorylation and inactivation of TTP, a factor involved in degrading PD-L1 mRNA transcripts. As TTP inactivation causes accumulation of PD-L1 mRNA, interfering with the RAS pathway increases TTP binding to AU-rich elements of the transcripts, decreases PD-L1 protein production, and leads to enhanced antitumor immunity.

Metastatic melanoma is a lethal disease, in part because of rapid acquisition of resistance to therapy. Using genetically engineered mouse models, we demonstrate that the activating RAC1 P29S mutation, present in up to 5% of melanoma patients, cooperates with BRAF as a driver of melanoma initiation and promotes BRAF inhibitor resistance. The critical RAC1 effector pathway in melanoma is shown to be the transcription factor complex SRF/MRTF, which initiates a switch to a mesenchymal-like state characterized by therapy resistance. Therapeutic targeting of SRF/MRTF may have potential to reverse BRAF inhibitor resistance in melanoma patients bearing the oncogenic RAC1 P29S mutation

KRAS is the most commonly mutated oncogene in human lung cancer, but direct targeting of RAS proteins has proved difficult. A recently developed inhibitor of G12C mutant KRAS protein inhibits lung cancer progression in mouse models but does not provide durable regressions. By studying signalling pathways required for survival of KRAS mutant cells, we demonstrate a strong and selective potentiation of the effects of G12C KRAS inhibitors when mTOR and/or IGF1R are also inhibited. Using mutant specific G12C KRAS inhibitors rather than MEK inhibitors in these combinations is associated with greater specificity and lower toxicity. We propose that adding IGF1R and mTOR inhibitors will increase the impact of G12C KRAS inhibitors in clinical trials.

Through the use of aPKC inhibitors and genetic mutations, we demonstrate that aPKC cycles between distinct PAR-3 and CDC-42 dependent states, which define, respectively, the ability of the aPAR network to respond to spatial cues and to displace pPAR proteins from the membrane. We further show that cue sensing depends crucially on the oligomeric nature of the PAR-3 state, that the integrity of this cycle is required for coupling of cue-sensing and effector functions of the aPAR network, and that this cycle is enforced by activity of aPKC.

A key requirement for patterning networks is that the scale of pattern be appropriately matched to the size of the system to be patterned. Through a combination of theory and experiment, we show that failure of the PAR network to scale with cell size restricts stable cell polarity to a specific size range and imposes a minimum cell size threshold for polarity. Experimental alteration of cell size indicates that embryos are sensitive to this size threshold. We thus propose a general strategy by which cells can use intrinsic length scales of patterning networks to enable size-dependent decision making.

This paper is a continuation of our major research theme on how dividing stem cells in the CNS are able to resist environmental stresses that shut down proliferation in most other developing tissues. It reports the first identification, in any species, of lipid droplets as protectors of stem cells. We discovered that hypoxia induces lipid droplets in the neural stem cell niche and that these protect the neural stem cells themselves from damaging polyunsaturated fatty acid (PUFA) peroxidation reactions. This study laid the foundation for our current mechanistic studies into the antioxidant functions of lipid droplets during development and tumorigenesis.