In silico investigation of the haematological niche during normal and malignant haematopoiesis

Abstract

Eradicating leukaemia stem cells (LSCs) and curing stem cell-derived cancers is hugely challenging. This is exemplified in chronic myeloid leukaemia (CML), where the t(9;22)(q34;q11) chromosomal translocation results in an abnormal chimeric BCR-ABL1 tyrosine kinase in a haematopoietic stem cell (HSC). Treatment of CML with tyrosine kinase inhibitors (TKIs) targeting the BCRABL1 protein is considered an exemplar of precision medicine, but unfortunately, TKIs do not cure most patients. Hence, unmet clinical needs arise in CML through the persistence and propagation of LSCs that are not eradicated by TKI treatment as they yield a source of disease recurrence.

The persistence of LSCs after TKI treatment is related to tyrosine kinaseindependent mechanisms which include intrinsic properties of LSCs determined by epigenetic alterations, dysregulated transcriptional networks, and mitochondrial/metabolic changes. Importantly, both leukemic and normal HSCs share these properties along with quiescence and self-renewal ability supported in specialised haematopoietic bone marrow (BM) niches. The BM mesenchymal cell (BMSC) plays an essential niche role in HSC and LSC maintenance. Therefore, a deeper understanding of BMSC-dependent regulation of CML LSCs is required to determine how CML manipulates the haematopoietic niche and protect LSCs from TKI treatment.

In this thesis, I set out to identify genes downstream of established autocrine signalling loops which drive the persistence of CML LSCs in a BCRABL1 independent manner. I examined seventeen publicly available CML RNA-seq datasets obtained from the Gene Expression Omnibus for novel gene signatures (Chapter 3). To understand the possible bone marrow niche cell interactions, dataset samples from human primary CD34⁺ enriched normal donor and CML samples (Greater Glasgow and Clyde National Health Service) cultured alongside human stromal, myeloid, and Lymphoid CML cell lines were also examined. A Fluidigm platform was used for LSC-niche co-culture samples to verify the autocrine signalling loops identified from published RNA sequencing data (Chapter 4).

Collectively, the analyses inferred the expected widespread disturbances of haemopoietic networks. Gene sets enriched in CML LSCs downstream of previously reported axes included IL-6, TGF-β-BMP, TNF-α as well as PI3KBMP and NF-κB- expression pathways. During this study, significantly expressed genes from both human CML LSCs and a murine model of CML were mined for overlapping genes of novel significance. These datasets allowed us to compare normal HSCs, highly proliferative LSCs, quiescent BCR-ABL⁺ LSCs, as well as CML cell lines, and murine CML cells.

Gene enrichment for Pituitary tumour-transforming gene 1 (PTTG1) and its interacting protein (PTTG1IP) were found and CML co-cultures showed increased expression of PTTG1 transcriptional targets, c- Myc, p53, FGF2, p21, prolactin and MMP2. To confirm whether the cell source of expression was the LSC, or niche cells co-culture data were re-examined. We hypothesised that the BMSC niche harbours and selectively maintains LSC persistence through cellular secretion. Our findings reveal a role for the secretory BMP/TGF-β superfamily, particularly Activin signalling, during the progression of CML. These correspond with earlier observations in CML that propose a switch from BMP to ActivinA (INHBA) signalling during treatment resistance and progressive disease. Throughout the in-silico study, we explored the role of the TGF-β superfamily and ActivinA associated genes within the CML BM niche and how they could be involved in the augmentation of a PTTG1-LSC gene network.

In Chapter 4-5 we report an unanticipated finding. When CML cells were cocultured with BMSCs, macrophage chemotaxis and activation pathways were upregulated. This highlights CMLs ability to recruit immune cells to their specialised niche. We explored CML immune cell datasets focusing on macrophages from CML and normal healthy BM samples. CML exposed macrophages provided strong evidence that the niche can provide AcitivinA secretion that drives Activin signalling during CML. Thus, macrophages could pose an interesting therapeutic target along with other niche-related adjuvant approaches for TKI–resistant patients or following treatment discontinuation.

To unravel the insights, we observed during our in-silico analyses, we sought to determine the in vivo effects of the niche on CML-LSCs and the potential synergistic effect of the BM towards disease progression and relapse during therapy (Chapter 2). To achieve this, we employed a CML mouse model (SCLtTA×BCR–ABL1) in which BCR–ABL1 expression is mainly targeted to the haematopoietic population. Through this approach, our data established that BCR–ABL1 expression produced a myeloproliferative disorder (with a similar temporal onset) after BM transplantation into a large cohort of mice. From this observation, we could produce varying clinical scenarios that represent significant time points during the development and transformation of the CML BM microenvironment. With these cohorts, we developed protocols for bone digestion and whole-bone mount imaging allowing for endosteal and central marrow cell visualisation, providing the basis for single-cell studies localizing niche constituents within the BM. In summary, the re-analysis of existing CML transcriptomic datasets provided insights into BM niche pathways that are potentially involved in promoting the survival of a subset of quiescent BCR-ABL1⁺ LSCs via a BCR-ABLindependent mechanism.