Recent Progress 2015
A key focus of the group’s work is the identification of genes selectively required for the function of leukaemia cells but not normal bone marrow cells: this is an important strategy to pinpoint new therapeutic targets for drug development. Over the past 12 months there has been significant progress in two projects which will lead to publications in 2015. In the first, we have identified an unexpected mechanism by which inhibitors of LSD1 promote myeloid differentiation. In the second, we have identified FOXC1 as a transcriptional regulator which is aberrantly derepressed in human acute myeloid leukaemia to functional effect.
Lysine Specific Demethylase 1 (LSD1) is one of a number of epigenetic regulators which have recently emerged as candidate therapeutic targets in cancer. It was initially identified as a core component of a RCOR1 (CoREST) histone deacetylase (HDAC) corepressor complex and later found to have lysine-specific demethylase activity. LSD1 is a flavin adenine dinucleotide (FAD) dependent homologue of the amine oxidase family which is known to demethylate monomethyl or dimethyl lysine 4 (K4) of histone H3, releasing hydrogen peroxide and formaldehyde.
The interest in LSD1 as a therapeutic target in cancer arose from the observation of its high level expression in poor prognosis sub-groups of prostate, lung, brain and breast cancer, as well as in certain haematological malignancies. The first drug found to inhibit LSD1 was tranylcypromine (TCP), a monoamine oxidase inhibitor used in the treatment of depression. TCP is a mechanism-based suicide inactivator of LSD1 which covalently attaches to the N(5) and C(4a) residues of the isoalloxazine ring of FAD, which is itself located deep within the active site of LSD1. To improve the potency and selectivity of TCP towards LSD1, derivatives active in the nanomolar range have been developed and these show promise as differentiation-inducing agents in pre-clinical studies in acute myeloid leukaemia (AML), as we previously reported in Cancer Cell in 2012. With LSD1 inhibitors already in early phase clinical trials, and others in development, an appreciation of their mechanism of action is essential. The assumption has been that LSD1 contributes to gene repression by removing monomethyl and dimethyl histone marks from lysine 4 of histone H3 and that this is the key activity targeted for potential therapeutic effect. However, in work led by James Lynch and Gary Spencer, we observed that drug-induced changes in transcription preceded changes in histone modifications targeted by LSD1 and that AML cell proliferation did not require the catalytic activity of LSD1. Instead, concomitant with up regulation of a myeloid differentiation program, a tranylcypromine-derivative inhibitor induced physical separation of LSD1 from the transcription factors GFI1 and MYB, and more generally from chromatin (Figure 1). Physical separation of GFI1 from LSD1 was required for drug-induced differentiation, and LSD1 sustained the association of MYB with chromatin. Complete resistance to the effects of LSD1 inhibition could be achieved by co-induction of MYB and a GFI1 DNA binding domain-LSD1 fusion protein. Thus, pharmacological inhibition of LSD1 promotes differentiation by disabling the activity of key myeloid transcription factors, through abrogation of protein:protein interactions rather than blockade of histone demethylase activity. A report detailing these discoveries is currently out for review.
In the clinical arena, there is ongoing progress (in collaboration with Oryzon Genomics) with our first-into-man Phase 1 trial of a first-in-class LSD1 inhibitor, ORY1001. The trial began its recruitment in June 2014 and is proceeding satisfactorily at The Christie NHS Foundation Trust, as well as at sites throughout Spain.
Figure 1. Model of the mechanism of action of LSD1 inhibitors in myeloid leukaemia. OG86, an exemplar tranylcypromine-derivative inhibitor of LSD1, irreversibly binds LSD1 leading to its physical separation from transcription factors and chromatin, and resulting loss of MYB from chromatin.
To identify transcriptional regulators expressed in human acute myeloid leukaemia haematopoietic stem and progenitor cells (AML HSPC) but not normal HSPC, we analysed the expression levels of known or candidate transcription factor genes in recently published datasets. Of those exhibiting significantly higher expression in AML HSPC versus HSPC, FOXC1 was among the most highly up regulated genes in each study when ranked by fold-change increase in expression (Figure 2). FOXC1 is a member of the forkhead box family of transcription factors which regulate processes such as development and differentiation. In keeping with a requirement for FOXC1 in mesenchymal differentiation, Foxc1 null mice die perinatally with skeletal, cardiac and renal abnormalities, hydrocephalus, iris hypoplasia and open eyelids. Humans with germ line mutations in FOXC1 develop the Axenfeld-Rieger syndrome which includes developmental anterior segment abnormalities of the eye.
Critically, FOXC1 is not expressed in the haematopoietic system and so a question was raised as to whether its derepression in ~15-20% of cases of human AML contributes to transformation. In work led by Tim Somerville, we have shown that shRNA-mediated knockdown of FOXC1 in human AML cells impairs clonogenic potential and induces myeloid differentiation, whereas normal HSPC are spared. In forced expression experiments, FOXC1 induced a transient enhancement of both the clonogenic potential and myeloid differentiation block of normal murine HSPCs cultured in serial replating assays, and myeloid skewing in in vivo transplantation assays. In silico and qPCR analyses showed that high level FOXC1 expression strongly associates with high level HOXA9 expression in human AML. We demonstrated this co-expression to be of functional significance because retroviral co-overexpression of FOXC1 and HOXA9 in murine HSPC strongly enhanced their clonogenic potential and myeloid differentiation block in serial replating assays, and accelerated leukaemia initiation in in vivo transplantation assays. These data demonstrate that FOXC1 functions to accelerate and enhance the development of AML in collaboration with HOXA9. Thus FOXC1 is a hitherto unappreciated transcriptional regulator in human AML which is inappropriately derepressed to functional effect. A report detailing these discoveries is also currently out for review.
Figure 2. Expression of FOXC1 in human AML. Transcription factor genes differentially expressed in AML HSPCs versus normal HSPCs were identified using an unpaired t-test (with p<0.005) and were ranked according to the mean fold change increase in expression. Heat maps show the most highly up regulated genes in the indicated datasets, in order of rank (indicated by number to the left of each heat map row).
We welcome to the lab Alba Maiques-Diaz and Gauri Deb who have recently completed their doctoral studies at CNIO Madrid, Spain and IIT Guwahati, India respectively. We also look forward to Isabel Romero joining the lab in March 2015 following her productive period of doctoral research at the University of Salamanca, Spain.