p53 transcriptional activity as a tool to uncover novel and diverse druggable targets in cancer

Abstract

The transcription factor p53 is one of the most studied tumour suppressors with over 90 000 publications in PubMed referring to the protein. It is also the most frequently mutated gene across all cancer types with around 50% of cancers presenting as mutant p53, and when it is not mutated, it is frequently inactivated to circumvent its tumour suppressor function. Therapeutic targeting of both mutant and wild-type p53 has been a key focus ever since its first discovery as “the guardian of the genome”. For our drug development programme, we have focused on visualising the induction of p53 transcriptional activity as a readout for a desirable phenotype. This screen used two stably transfected reporter cell lines, the T22 murine fibroblasts, and the ARN8 human melanoma cell line. Using this forward chemical genetic approach, we have entered into our drug development programme in a target-blind manner.

For Paper I we screened 30 000 compounds in both T22 and ARN8 cells and selected those that were capable of increasing p53 transcriptional activity in the ARN8 tumour cells, but not in the T22 murine fibroblasts. We selected a compound from the hits that had a drug-like structure as well as possessing a chiral centre and christened it HZ00. HZ00 was found to induce p53 protein in a dose-dependent manner, selectively kill tumour cells whilst inducing a reversible G1 arrest in normal human dermal fibroblasts (HNDFs), and increase p53 synthesis at early timepoints without stabilising the protein or increasing levels of p53 mRNA. HZ00 also synergised with the inhibitor of p53 degradation, nutlin 3, both in vitro and in vivo in a tumour xenograft model. Following target deconvolution using a knowledgebased approach we identified DHODH, a key enzyme in the de novo pyrimidine nucleotide synthesis pathway, as the target of HZ00. At this point we re-screened 30 000 compounds in ARN8 cells that were previously screened in the T22 cell line for another study. We found that those that were able to activate p53 in ARN8 cells also largely inhibited DHODH. This yielded 12 other chemotypes capable of inhibiting DHODH. At this point we tested HZ00 analogues and identified a much more potent compound we named HZ05. HZ05 phenocopied HZ00 and demonstrated enantiomer-selective inhibition of DHODH with (R)-HZ05 inhibiting DHODH with an IC50 of 11 nM. We obtained a crystal structure of (R)-HZ05 in complex with DHODH and found that it occupied the same quinone tunnel as the known inhibitors brequinar and teriflunomide (A77 1726). HZ05 caused a number of tumour cells to accumulate in S-phase. We found that a slower cycling cell line, U2OS, required pretreatment with HZ05 to accumulate cells in S-phase prior to treatment with nutlin 3a to achieve tumour cell kill, as co-treatment resulted in G1 arrest. We therefore theorised that accumulating cells in S-phase with high levels of p53 predisposed them to cell kill upon application of a blocker of p53 degradation.

The first sets of compounds found back in 2008 by the Laín laboratory were the tenovins. Tenovin 1 was the first compound identified from the screen, which used the T22 murine fibroblasts to establish its ability to activate p53 transcriptional activity in the reporter assay. Tenovin 1 was, however, not particularly soluble and therefore a more soluble analogue called tenovin 6 was synthesised. Tenovin 6 elicited many of the same cellular phenotypes as tenovin 1, and therefore target identification was conducted using tenovin 6. Tenovin 6 was subsequently identified as an inhibitor of SirT1 and SirT2 in a yeast genetic screen, biochemical assays and further target validation in mammalian cells. Tenovin 1 and 6 displayed a very similar profile – they both induced p53 transcriptional activity and both increased acetylation of both p53 and tubulin. This is where the similarity ends, however, as it was discovered, through extensive structure-activity relationship studies, that the targeting profiles of both molecules was markedly different. In Paper II we built upon previous studies that identified tenovin 6 as a compound capable of inhibiting autophagy. In this paper we conducted structure-activity relationships using tenovin analogues to understand the mechanism by which tenovins affect autophagy. We confirmed that tenovins capable of perturbing autophagy do so by inhibition of autophagic flux, in a similar manner to chloroquine, by raising the pH of lysosomes. We also isolated the portion of the molecule, a tertiary amine at the end of an aliphatic chain, as the reason for blockage of autophagic flux. Finally, we found that blockage of autophagic flux by tenovins is required to eliminate tumour cells in culture and that this blockage of autophagy is capable of killing mutant B-Raf tumour cells arrested in G1 by vemurafenib treatment.

In Paper III we further explored the targeting profile of the tenovins and tested whether tenovins were capable of inhibiting DHODH. We found that tenovins 1 and 6 were capable of inhibiting DHODH at 113 nM and 500 nM respectively. We also conducted a thermal shift assay and identified tenovins 1, 6 and 39OH as being capable of interacting with DHODH in vitro. We then obtained a crystal structure of tenovin 6 occupying the same quinone tunnel as HZ05, brequinar and teriflunomide. Phenotypically, tenovin 1 and 33 had their ability to induce p53 transcriptional activity ablated upon addition of either uridine or orotate, but not dihydroorotate, whilst tenovin 6 had its ability to induce p53 transcriptional activity partially prevented by addition of uridine or orotate. Tenovin 39 and 39OH displayed no difference upon supplementation. Tenovin 1 and 33 also had their growth inhibitory effect markedly reduced upon orotate or uridine supplementation, but no other tenovin, including 6, showed any effect of supplementation. We also discovered another target of the tenovins – the ability to inhibit nucleoside uptake. We discovered that uridine uptake was blocked by tenovin 6, 33, 39, 39OH and 50. This paper, therefore, highlights the shifting targeting profile of the tenovins due to small molecular changes and that a phenotypic readout may remain static even as the targeting profile changes, as well as highlighting both the benefits and cautions of targeting multiple disparate targets in cells.

Unlike our other projects, Paper IV focused on understanding the structure and function of DHODH. We studied a purified DHODH lacking the transmembrane domain using native protein nano-electrospray mass spectrometry (nESI-MS). Firstly, we identified MS conditions that allowed for the DHODH to spray and isolated a high m/z range that corresponded to the molecular weight of the enzyme plus the bound FMN cofactor. Ion mass spectrometry was conducted to differentiate between the holo- and apo-DHODH, with the holo-DHODH corresponding to a compact formation suggesting that folded DHODH with FMN present can be preserved in the gas phase. We next incubated lipids that constitute the human mitochondrial membrane with DHODH and analysed the interaction in the gas phase. Complexes with both PE and CDL were evident, but complexes with PC were not easily detected. The next finding was that an intact protein-cofactor complex was required for the DHODH inhibitor, brequinar, to bind thus confirming that brequinar binding to DHODH is not random, but requires properly structured DHODH. Finally, MD simulations were conducted using both full length and truncated protein associated with a model PE bilayer. These models established that DHODH sits on the surface of the lipid bilayer loosely and is anchored in place by the transmembrane helix and this anchorage holds DHODH in the correct orientation to allow insertion of coenzyme-Q10 into the quinone tunnel of DHODH.