ABSTRACT:The BCR::ABL1 tyrosine kinase inhibitors (TKI) in chronic myeloid leukemia (CML) represent a paradigm for molecularly targeted therapy. However, clinical outcomes (rate/depth of response, treatment-free remission [TFR], progression to blast crisis [BC]) and adverse events vary among patients. While additional somatic mutations have been invoked to explain varying clinical outcomes, we here propose a complementary perspective based on single-cell omics (sc-omics) approaches that have enabled unprecedented resolution of the cellular ecosystems, including their composition, interactions, and activity. In patients who were treatment-naïve and in chronic phase (CP), this has revealed differences in the growth-rate of BCR::ABL1+ clones, ratio of TKI-insensitive leukemic stem cells (LSC) to residual hematopoietic stem cells (HSC), and immune cell composition, factors that collectively contribute to variability in therapy efficacy. Together these findings suggest that cellular heterogeneity serves as a foundation of clinical outcome in CML. Patients who remain in CP exhibit an erythroid signature in LSC, while those progressing to BC manifest an inflammatory profile, additional mutations, and expansion of early progenitors. Deep responders with active natural killer, and regulatory T cells are more likely to sustain TFR. Similarly, the outcomes of donor lymphocyte infusion after allogeneic stem cell transplant are heterogeneous, and reflect differences in preexisting T-cell clonotypes, their expansion, and interaction with leukemic cells in responders vs nonresponders. Here, we summarize key insights from sc-omics in CML, and propose an actionable road map to further leverage these technologies. This includes mechanistically explaining heterogeneity, predicting therapy response and BC, tracking leukemogenic clones longitudinally, targeting TKI-insensitive LSC, and restoring hematopoiesis from residual HSCs.

The advent of tyrosine kinase inhibitors (TKIs) as treatment of chronic myeloid leukemia (CML) is a paradigm in molecularly targeted cancer therapy. Nonetheless, TKI-insensitive leukemia stem cells (LSCs) persist in most patients even after years of treatment and are imperative for disease progression as well as recurrence during treatment-free remission (TFR). Here, we have generated high-resolution single-cell multiomics maps from CML patients at diagnosis, retrospectively stratified by BCR::ABL1 IS (%) following 12 months of TKI therapy. Simultaneous measurement of global gene expression profiles together with >40 surface markers from the same cells revealed that each patient harbored a unique composition of stem and progenitor cells at diagnosis. The patients with treatment failure after 12 months of therapy had a markedly higher abundance of molecularly defined primitive cells at diagnosis compared to the optimal responders. The multiomic feature landscape enabled visualization of the primitive fraction as a mixture of molecularly distinct BCR::ABL1 + LSCs and BCR::ABL1 - hematopoietic stem cells (HSCs) in variable ratio across patients, and guided their prospective isolation by a combination of CD26 and CD35 cell surface markers. We for the first time show that BCR::ABL1 + LSCs and BCR::ABL1 - HSCs can be distinctly separated as CD26 + CD35 - and CD26 - CD35 + , respectively. In addition, we found the ratio of LSC/HSC to be higher in patients with prospective treatment failure compared to optimal responders, at diagnosis as well as following 3 months of TKI therapy. Collectively, this data builds a framework for understanding therapy response and adapting treatment by devising strategies to extinguish or suppress TKI-insensitive LSCs.

The emerging notion of hematopoietic stem and progenitor cells (HSPCs) as a low-primed cloud without sharply demarcated gene expression programs raises the question on how cellular-fate options emerge and at which stem-like stage lineage priming is initiated. Here, we investigate single-cell chromatin accessibility of Lineage(-), cKit(+), and Sca1(+), (LSK) HSPCs spanning the early differentiation landscape. Application of a signal-processing algorithm to detect transition points corresponding to massive alterations in accessibility of 571 transcription factor motifs reveals a population of LSK FMS-like tyrosine kinase 3 (Flt3)(int)CD9(high) cells that concurrently display stem-like and lineage-affiliated chromatin signatures, pointing to a simultaneous gain of both lympho-myeloid and megakaryocyte-erythroid programs. Molecularly and functionally, these cells position between stem cells and committed progenitors and display multi-lineage capacity in vitro and in vivo but lack self-renewal activity. This integrative molecular analysis resolves the heterogeneity of cells along hematopoietic differentiation and permits investigation of chromatin-mediated transition between multipotency and lineage restriction.

The advent of single cell (Sc) genomics has challenged the dogma of haematopoiesis as a tree-like structure of stepwise lineage commitment through distinct and increasingly restricted progenitor populations. Instead, analysis of ScRNA-seq has proposed that the earliest events in human hematopoietic stem cell (HSC) differentiation are characterized by only subtle molecular changes, with hematopoietic stem and progenitor cells (HSPCs) existing as a continuum of low-primed cell-states that gradually transition into a specific lineage (CLOUD-HSPCs). Here, we combine ScRNA-seq, ScATAC-seq and cell surface proteomics to dissect the heterogeneity of CLOUD-HSPCs at different stages of human life. Within CLOUD-HSPCs, pseudotime ordering of both mRNA and chromatin data revealed a bifurcation of megakaryocyte/erythroid and lympho/myeloid trajectories immediately downstream a subpopulation with an HSC-specific enhancer signature. Importantly, both HSCs and lineage-restricted progenitor populations could be prospectively isolated based on correlation of their molecular signatures with CD35 and CD11A expression, respectively. Moreover, we describe the changes that occur in this heterogeneity as hematopoiesis develops from neonatal to aged bone marrow, including an increase of HSCs and depletion of lympho-myeloid biased MPPs. Thus, this study dissects the heterogeneity of human CLOUD-HSPCs revealing distinct HSPC-states of relevance in homeostatic settings such as ageing.

Hematopoietic Stem Cells (HSCs) that give rise to all lineages of the blood are defined as a lineage-negative (Lin-), c-Kit+, Sca-1+(LSK) population in mouse. The expression of Flt3 on LSK cells is accompanied by loss of self-renewal capacity, and transition to lymphoid-primed multipotent progenitors (LMPPs). The magnitude of Flt3 surface abundance has been shown to be correlated with lineage potential, where megakaryocyte/erythroid (MegE) potential is incompatible with high Flt3 levels (Buza-Vidas et al. 2011). However, it is unclear whether MegE capacity is marked by Flt3 itself or restricted to a specific population within the Flt3 spectrum. Interestingly, it has been shown that commitment to MegE lineages is an early event in HSC differentiation and that MegE progenitors share several cell-surface markers with HSCs. One such marker is CD9, which is highly expressed on megakaryocyte progenitors (Nakorn, Miyamoto, and Weissman 2003)as well as on the most primitive stem cells in mouse (Karlsson et al. 2013). In fact, we have recently shown that CD9 captures all long-term reconstitution activity of the murine LSK population, even in cell fractions that lacks some of the most established HSC markers .To compare CD9- and Flt3 expression we performed FACS analysis of the LSK compartment and plotted CD9 high- and low cells on top of the CD34/Flt3 expression profile. Interestingly we found that even if CD9 expression were predominantly high in LT-(CD34-Flt3-) and ST-HSCs (CD34+Flt3-), CD9highcells reached all the way into the Flt3+population but were lost in LMPPs, as defined by the 25% highest Flt3 expressing cells. To addresswhether CD9highcells represent a distinct subpopulation within Flt3intcells, LSKCD34+Flt3int(Flt3int) and LSKCD34+FLT3intCD9high(CD9high) populations, were subjected to single-cell RNA-Sequencing analysis. Interestingly,our results demonstratethatCD9highcells are enriched in the most primitive cluster of the Flt3intpopulation. This was further validated in transplantation assays where only CD9highFlt3intcells reconstituted irradiated recipients. However, long-term reconstitution was limited to lymphoid lineages, while myeloid engraftment was only observed at early time points and self-renewal activity was absent. To address the clonal capacity of CD9highcells, we designed a protocol to assess the potential of single-cells to generate B-, Myeloid (My)- and Erythroid (Er) cells in vitro.In these conditions 68% of clones generated from CD9highcells were bipotent (55% My/B and 13% My/Er). In contrast, the majority of CD9lowcells were unipotent (32% My- and 52% B cells).Moreover,the colony forming capacity for the CD9highpopulation was significantly higher compared to CD9lowcells and when assayed for colony-forming unit-megakaryocyte (CFU-MK) capacity, Meg potential was exclusively observed in LT-HSCs, ST-HSCs as well as the CD9highFlt3intpopulation, while neither CD9lownor LMPPs generated any megakaryocyte colonies. In an attempt to improve the efficiency of MegE differentiation, single-cell clonogenic assays were performed in suspension cultures. Here,multipotency including MegE potential was gradually down regulated with increasing Flt3 cell surface expression, and decreasing CD9 expression. Together our results shows thatCD9 captures the most primitive cells within the LSKFlt3+population, including multipotent cells with MegE differentiation potential, thus resolving the heterogeneity of Flt3+hematopoietic progenitors and identifying a multipotent LSKCD9+Flt3+population distinct from LMPPs. Disclosures No relevant conflicts of interest to declare.

In chronic myeloid leukemia (CML), a rare subset of leukemic stem cells (LSC) persists in patients responding to conventional tyrosine kinase inhibitor (TKI) therapy. The failure to eradicate these LSCs results in indefinite therapy dependence and a risk of leukemic relapse. However, the conventional LSC compartment (Lin-CD34+CD38-) is highly heterogeneous where only a subpopulation is believed to be functional, TKI-insensitive LSCs. Previously, using single-cell gene expression analysis we characterized the heterogeneity within the LSC population (Lin-CD34+CD38-) in CML patients using a selected panel of 96 primers. Interestingly, by comparing LSC heterogeneity at diagnosis with the heterogeneity following 3 months of TKI therapy we uncovered a therapy-insensitive, quiescent subpopulation, which could be isolated at high-purity using a combination of the surface markers: Lin-CD34+CD38-CD45RA-cKIT-CD26+ (Warfvinge, Geironson, Sommarin et al., 2017). Here, we expand the single-cell analysis of CML LSC populations to include combined immunophenotype-/RNA sequencing analysis (CITE-seq). CITE-seq allows for unbiased, further in-depth transcriptome analysis as wells as immunophenotypic characterization by pre-staining cells with a panel of DNA-barcoded antibodies prior to sequencing. DNA-barcoded antibodies convert the protein expression into readable sequences through unique oligo-conjugates as identifiers. Using CITE-seq with a panel of 44 distinct surface markers designed to immunophenotypically differentiate between stem/progenitors cells and leukemic clones we simultaneously characterize the molecular and immunophenotypic heterogeneity within Lin-CD34+/Lin-CD34+CD38- CML stem/progenitor compartment at diagnosis. Additionally by comparing the LSCs transcriptome from patients with different therapeutic outcome after 12 months of therapy we describe how differences in heterogeneity and the presence of immunophenotypic therapy-insensitive LSCs at diagnosis (Lin-CD34+CD38-CD45RA-cKIT-CD26+) contribute to therapy response. Disclosures Richter: Novartis: Consultancy; Pfizer: Consultancy, Research Funding.

Building molecular correlates of drug resistance in cancer and exploiting them for therapeutic intervention remains a pressing clinical need. To identify factors that impact drug resistance herein we built a model that couples inherent cell-based response toward drugs with transcriptomes of resistant/sensitive cells. To test this model, we focused on a group of genes called metastasis suppressor genes (MSGs) that influence aggressiveness and metastatic potential of cancers. Interestingly, modeling of 84 000 drug response transcriptome combinations predicted multiple MSGs to be associated with resistance of different cell types and drugs. As a case study, on inducing MSG levels in a drug resistant breast cancer line resistance to anticancer drugs caerulomycin, camptothecin and topotecan decreased by more than 50-60%, in both culture conditions and also in tumors generated in mice, in contrast to control un-induced cells. To our knowledge, this is the first demonstration of engineered reversal of drug resistance in cancer cells based on a model that exploits inherent cellular response profiles.

The mechanism of action of NME2, a widely accepted metastasis-suppressor gene, is poorly understood. Recently we found that NME2 directly regulates transcription of the c-MYC proto-oncogene. This prompted a genome-wide study to ascertain whether NME2 exerts its anti-metastatic action through transcriptional regulation. Chromatin-immunoprecipitation followed by massively parallel sequencing (ChIPseq) along with transcriptome profiling uncovered a network of genes involved in intercellular contact, focal adhesion and actin assembly under direct transcriptional control of NME2. In line with this, NME2-depleted cells displayed increased focal adhesion points and altered actin stress fiber organization. Our findings demonstrate that NME2 regulates transcription of a key focal adhesion factor vinculin and its localization within adhesion foci. NME2-depleted A549 lung cancer cells showed higher invasiveness in vitro and seeded more metastases in vivo. Consistent with these findings, expression of several NME2-transcriptional target genes related closely to advanced tumor stages with metastatic proclivity, and NME2 levels predicted patient survival.

Previous studies have analyzed patterns of transcription, transcription factor (TF) binding or mapped nucleosome occupancy across the genome. These suggest that the three aspects are genetically connected but the cause and effect relationships are still unknown. For example, physiologic TF binding studies involve many TFs, consequently, it is difficult to assign nucleosome reorganization to the binding site occupancy of any particular TF. Therefore, several aspects remain unclear: does TF binding influence nucleosome (re)organizations locally or impact the chromatin landscape at a more global level; are all or only a fraction of TF binding a result of reorganization in nucleosome occupancy and do all TF binding and associated changes in nucleosome occupancy result in altered gene expression? With these in mind, following characterization of two states (before and after induction of a single TF of choice) we determined: (i) genomic binding sites of the TF, (ii) promoter nucleosome occupancy and (iii) transcriptome profiles. Results demonstrated that promoter-proximal TF binding influenced expression of the target gene when it was coupled to nucleosome repositioning at or close to its binding site in most cases. In contrast, only in few cases change in target gene expression was found when TF binding occurred without local nucleosome reorganization.

Tumor metastasis refers to spread of a tumor from site of its origin to distant organs and causes majority of cancer deaths. Although >30 metastasis suppressor genes (MSGs) that negatively regulate metastasis have been identified so far, two issues are poorly understood: first, which MSGs oppose metastasis in a tumor type, and second, which molecular function of MSG controls metastasis. Herein, integrative analyses of tumor-transcriptomes (n = 382), survival data (n = 530) and lymph node metastases (n = 100) in lung cancer patients identified nonmetastatic 2 (NME2) as a key MSG from a pool of >30 metastasis suppressors. Subsequently, we generated a promoter-wide binding map for NME2 using chromatin immunoprecipitation with promoter microarrays (ChIP-chip), and transcriptome profiling. We discovered novel targets of NME2 which are involved in focal adhesion signaling. Importantly, we detected binding of NME2 in promoter of focal adhesion factor, vinculin. Reduced expression of NME2 led to enhanced transcription of vinculin. In comparison, NME1, a close homolog of NME2, did not bind to vinculin promoter nor regulate its expression. In line, enhanced metastasis of NME2-depleted lung cancer cells was found in zebrafish and nude mice tumor models. The metastatic potential of NME2-depleted cells was remarkably diminished upon selective RNA-i-mediated silencing of vinculin. Together, we demonstrate that reduced NME2 levels lead to transcriptional de-repression of vinculin and regulate lung cancer metastasis.

Lung cancer is one of the deadliest cancers worldwide, with the highest incidence and mortality amongst all cancers. While the prognosis of lung cancer is generally grim, with 5-year survival rates of only 15%, there is hope, and evidence, that early detection of lung cancer can reduce mortality. Today, only computed tomography screening has shown to lead to early detection and reduction in mortality, but is limited by being anatomic in nature, unable to differentiate between inflammatory and neoplastic pathways, and therefore, susceptible to false positives. There is increasing interest in biomarkers for lung cancer, especially those that predict metastatic risk. Some biomarkers like DNA mutations and epigenetic changes potentially require tissue from the at-risk site; some like serum proteins and miRNAs are minimally invasive, but may not be specific to the lung. In comparison, emerging biomarkers from exhaled breath, like volatile organic compounds (VOC), and exhaled breath condensate, e.g., small molecules and nucleic acids, have the potential to combine the best of both. This mini review is intended to provide an overview of the field, briefly discussing the potential of what is known and highlighting the exciting recent developments, particularly with miRNAs and VOCs.

Analysis of chromatin-immunoprecipitation followed by sequencing (ChIP-seq) usually disregards sequence reads that do not map within binding positions (peaks). Using an unbiased approach, we analysed all reads, both that mapped and ones that were not included as part of peaks. ChIP-seq experiments were performed in human lung adenocarcinoma and fibrosarcoma cells for the metastasis suppressor non-metastatic 2 (NME2). Surprisingly, we identified sequence reads that uniquely represented human telomere ends in both cases. In vivo presence of NME2 at telomere ends was validated using independent methods and as further evidence we found intranuclear association of NME2 and the telomere repeat binding factor 2. Most remarkably, results demonstrate that NME2 associates with telomerase and reduces telomerase activity in vitro and in vivo, and sustained NME2 expression resulted in reduced telomere length in aggressive human cancer cells. Anti-metastatic function of NME2 has been demonstrated in human cancers, however, mechanisms are poorly understood. Together, findings reported here suggest a novel role for NME2 as a telomere binding protein that can alter telomerase function and telomere length. This presents an opportunity to investigate telomere-related interactions in metastasis suppression.

Non-metastatic 23 [NM23/nucleoside diphosphate kinases ( NDPK )] genes are the first discovered metastasis suppressor genes. More than two decades of research has demonstrated their roles in a variety of biological processes with NME1 and NME2 being most studied in the context of metastasis suppression. Although NME1 and NME2 share >85% homology at amino acid level, they show redundant as well as unique molecular functions. Phenotypic analyses of knockout (KO) mice for NM23 members ( NDPK-A, B ) and compound KO (A as well as B) showed requirement of both proteins in hematopoiesis suggesting shared functions in development disease. Several reviews have discussed NME1, however the role of NME2 appears to be relatively less understood in the context of metastasis suppression. Here, we focus on NME2 and by meta-analysis of gene expression from multiple tumor types, and survey of in vivo and vitro studies, suggest the possibility that NME2 may be one of the key factors in metastasis. This along with the relevance of normal physiological functions of NME2 in the context of metastasis is discussed. We further examined the genetic and epigenetic features of NME2 and NME1 gene promoters and found aspects of transcription control that could be unique to NME2/NME1. Findings on signaling pathways and small molecules which regulate the expression of NME2 that could be therapeutically important are also discussed.

Regulatory influence of the G-quadruplex or G4 motif present within the nuclease hypersensitive element (NHE) in the promoter of c-MYC has been noted. On the other hand, association of NM23-H2 to the NHE leads to c-MYC activation. Therefore, NM23-H2 interaction with the G4 motif within the c-MYC NHE presents an interesting mechanistic possibility. Herein, using luciferase reporter assay and chromatin immunoprecipitation we show NM23-H2 mediated c-MYC activation involves NM23-H2-G4 motif binding within the c-MYC NHE. G4 motif complex formation with recombinant NM23-H2 was independently confirmed using fluorescence energy transfer, which also indicated that the G4 motif was resolved to an unfolded state within the protein-bound complex. Taken together, this supports transcriptional role of NM23-H2 via a G4 motif.

Using a combination of in silico and experimental approaches, we present evidence that the G-quadruplex (G4) motif (an alternative higher-order DNA conformation) has regulatory potential. Genome-wide analyses of 99980 human, chimpanzee, mouse, and rat promoters showed enrichment of sequence with potential to adopt G4 (potential G4 or PG4) motifs near transcription start sites (TSS; P < 0.0001), supporting earlier findings. Interestingly, we found >700 orthologously related promoters in human, mouse, and rat conserve PG4 motif(s). The corresponding genes have enriched (z score > 4.0) tissue-specific expression in 75 of 79 human tissues and are significantly overrepresented in signaling and regulation of cell-cycle (P < 10(-05)). This is supported by results from whole genome expression experiments in human HeLa S3 cells following treatment with TMPyP4 [5,10,15,20-tetra(N-methyl-4-pyridyl) porphine chloride], which is known to bind the G4 motif inside cells. Our results implicate G4-motif mediated regulation as a more general mode of transcription control than currently appreciated.