Single cell approaches resolve the molecular network driving malignant hematopoie6c stem cell self-renewal David Kent WT/MRC Cambridge Stem Cell Ins6tute University of Cambridge, UK MPN EuroNet 17 May, 217
Stem Cell Fate Choice Balance is key how is it established and maintained? Population asymmetry is required for homeostasis What are the implications for disease?
Stem Cell Fate Choice
Stem Cell Fate Choice CANCER Mixed population of normal and diseased DEGENERATION Fewer (functional) stem cells What are the molecular regulators of disease?
Single cell approaches can add resolu6on Single cell transplanta6ons can defini6vely demonstrate func6onal proper6es (Nakauchi, Eaves, Goodell, Morrison, Skoda, etc) Single cell gene expression has improved drama6cally and has been applied to HSPCs (Amit, Regev, Ebert, GoSgens, etc) 1. Clonal assays in pa-ents can track tumour evolu-on (Ortmann et al., NEJM, 215) 2. Index sor6ng can link mul6ple single cell assays (Schulte et al., Exp. Hemat 215.; Wilson et al., Cell Stem Cell 215; Nestorowa et al., Blood 216) 3. Single cell gene expression and func6onal assays in gene6c models can iden6fy drivers of malignant self-renewal
Chronic Myeloproliferative Neoplasms (MPNs) Why do we study MPNs? Clonal Early stage of tumours Stem Cell Chronic condition Transformation to more serious disease (AML) Polycythaemia vera (>95%) Progenitors Differentiated Cells Essential thrombocythaemia (~5%) Myelofibrosis (>5%) JAK2 V617F
Cancer and its mutations: Does the order matter? 1 1 2 1 2 3 CANCER 3 3 2 3 2 1 CANCER Malignant properties are the sum of the mutations acquired (ie., 1+2+3) However, virtually nothing is known about whether the order of acquisition matters. =??? 1 1 2 2 2 1
So does the order of mutation acquisition make any difference?
TET2 loss of function is the most common collaborating mutation in MPNs 1% of patients harbour a collaborating TET2 mutation Multiple clones are detectable within the same patient Order of mutation acquisition can be determined by colony assay TET2 TET2 =??? JAK2 JAK2 TET2 JAK2 PV (4) PV (4) PV (7) PV (7) ET ET (5) (5) ET (2) MF MF ET (2) MF MF (3) (3) (3) TET2 JAK2 TET2-first (n=12) JAK2-first (n=12) Ortmann / Kent et al., New Eng J. Med 215
Order of acquisition impacts disease phenotype JAK2-first patients present at an earlier age Age at diagnosis (years) 9 8 7 6 5 4 TET2-first ** JAK2-first % V617F homozygous / total mutant colonies have more homozygous V617F colonies 1 8 6 4 2 PV1 PV2 PV6 PV7 ET1 ET2 ET4 ET6 ET7 TET2-first *** PV3 PV4 PV5 PV8 PV9 PV1 PV11 JAK2-first ET3 ET5 Relative number of progenitors (in CD34+/CD38+ cells).8.6.4.2 Normal TET2-first ` JAK2-first Myeloid Prog Mega/Eryth Prog Gran/Mac Prog and have skewed progenitor production
V617F affects stem and progenitor cells differently HSCs 2 fold in frequency 8 fold in long-term self renewal Progenitors Increased #s (especially MEPs) Increased CFCs produced in vitro / in vivo 8 Number of HSCs (E-SLAM) per 1, BM WBCs 6 4 2 P a i r e d t t e s t P v a l u e. 28 Donor chimerism (%) from 1 E-SLAM HSCs Colonies per 1 input HSCs * * E GM GEMM Colony Type Kent et al., PLoS Biol 213 ; *Li/*Kent et al., Blood 215
Order of acquisition impacts HSC function CD34 CD38 CD9 CD45RA Single HSCs * 1d Secondary colony assay + genotyping Number of Secondary Cols. (per starting HSC) 6 4 2 ** Genetically similar HSCs (Tet2 and Jak2 mutations) Different functional capacity TET2-first (n=2) JAK2-first (n=4) WT single mut. T + J mut. Ortmann / Kent et al., New Eng J. Med 215
Loss of TET2 prevents JAK2V617F from making the same set of gene expression changes WT JAK2 ~12 genes upregulated ~8 genes downregulated TET2 TET2 JAK2 Many shared changes Distinct subset that are altered differently Ortmann / Kent et al., New Eng J. Med 215
Expanding the patient cohort using an order prediction algorithm >1 patient samples (UK, Italy) Screen using RNA bait set (111 genes implicated in myeloid malignancies) 9 patients with JAK2V617F and a TET2 mutation Identification of regions with LOH (using normalised coverage and B-allele frequencies (1 genome dataset)) Determination of allele frequency and order calling, excluding: No significant difference between TET2/JAK2 frequency Possible biclonal disease (e.g., sum of frequencies is less than 1) 24 patients where order can be confidently assigned Ortmann / Kent et al., New Eng J. Med 215
Follow-up cohort reveals disease association bias and increased risk of thrombotic events No. patients with diagnosis 2 155 1 5 * 5 TET2 FIRST * JAK2 FIRST PV ET Proportion without thrombosis TET2-first (n=18) JAK2-first (n=3) P=.2 TET2-first JAK2-first Days from diagnosis JAK2-first patients are more likely to have PV and have an increased frequency of thrombotic events Ortmann / Kent et al., New Eng J. Med 215
Single and double mutant clones are more sensitive to ruxolitinb in culture 14 days colony counts genotyping TET2 JAK2 1 ** * Mutant colonies (% of total colonies) 8 6 4 2 Rux - + - + - + - + TET2-first JAK2-first Double mutant clones from patients with different order are differently sensitive to treatment Single Double Ortmann / Kent et al., New Eng J. Med 215
Acquisition order impacts clonal evolution Single stem cell derived clone TET2-first HSC Prog Diff Excess Delayed disease phenotype Smaller homozygous clones 1 2 3 JAK2-first wildtype TET2 mutant Earlier presentation Larger homozygous clones JAK2 het 1 2 3 JAK2 hom time
Summary Order of mutation acquisition impacts clonal evolution and stem cell function: Timing of disease presentation Relative composition of progenitor compartment Expansion ability of most mature clone
Single cell approaches can add resolu6on Single cell transplanta6ons can defini6vely demonstrate func6onal proper6es (Nakauchi, Eaves, Goodell, Morrison, Skoda, etc) Single cell gene expression has improved drama6cally and has been applied to HSPCs (Amit, Regev, Ebert, GoSgens, etc) 1. Clonal assays in pa6ents can track tumour evolu6on (Ortmann et al., NEJM, 215) 2. Index sor-ng can link mul-ple single cell assays (Schulte et al., Exp. Hemat 215.; Wilson et al., Cell Stem Cell 215; Nestorowa et al., Blood 216) 3. Single cell gene expression and func6onal assays in gene6c models can iden6fy drivers of malignant self-renewal
How do we study the proper6es of a seed? Different contaminating cells HSC3 Hypothesis True HSCs have a common molecular program Same cells (same genes?)
sc Index sor6ng links func6onal data with transcrip6onal data Single cell gene expression Index Sort 11 common cell surface markers Laser sc Single cell functional data Schulte et al., Exp Hem 215
sc sc 9 populations (FACS) 21 cells for each population 48 genes in 189 sc single cells (Fluidigm) Wilson / Kent / Buettner et al., Cell Stem Cell 215
Iden6fica6on of Molecular Overlap HSCs Popula-on % Durable Stem Cells Weigh-ng HSC1 5.352 HSC2 4.281 HSC3 25.176 HSC4 2.141 HSC5 7.49 HSCs with molecular overlap (MolO) Wilson / Kent / Buettner et al., Cell Stem Cell 215
High Sca1 associates with MolO HSCs vs. No molecular overlap Notch Gfi1 Gata1 Ikzf1 Itga2b Sca1 Sca1 Wilson / Kent / Buettner et al., Cell Stem Cell 215
Sca1 expression levels separate LT-HSCs 1 cell transplants Single cell transplants 15 / 29 52%
sc sc Index sor6ng links func6onal data with transcrip6onal data Single cell RNA-seq data Index Sort 11 common cell surface markers Laser Single cell transplantation data Wilson / Kent / Buettner et al., Cell Stem Cell 215
SuMO score Low Functional HSCs cluster with high MolO scores High t-sne plot 11 Surface Marker Parameters
Gene expression analysis of SuMO cells iden6fies Procr (EPCR) Good overlap with MolO scoring scheme (shared genes = green)
New SLAM Sca hi EPCR hi strategy achieves 67% HSC purity at single cell level SCA CD48 CD15 EPCR * 26/39 (67%) LT-HSCs FSC FSC
Index sor6ng is a powerful tool for dissec6ng cell popula6on heterogeneity Single cell func6onal assays required Broadly applicable across most cell systems Mouse progenitor cell gene expression networks Human progenitor cell lineage fate choice
Single cell approaches can add resolu6on Single cell transplanta6ons can defini6vely demonstrate func6onal proper6es (Nakauchi, Eaves, Goodell, Morrison, Skoda, etc) Single cell gene expression has improved drama6cally and has been applied to HSPCs (Amit, Regev, Ebert, GoSgens, etc) 1. Clonal assays in pa6ents can track tumour evolu6on (Ortmann et al., NEJM, 215) 2. Index sor6ng can link mul6ple single cell assays (Schulte et al., Exp. Hemat 215.; Wilson et al., Cell Stem Cell 215; Nestorowa et al., Blood 216) 3. Single cell gene expression and func-onal assays in gene-c models can iden-fy drivers of malignant self-renewal
JAK2 V617F models: dosage, disease and defective HSCs Haematocrit (%) 1 9 8 7 6 5 4 3 2 1 Platelets (x1 3 )/µl 25 2 15 1 5 WT Het Hom WT Het Hom WT Het Hom ET-LIKE: increased platelets, mild increase in red cells Kent et al., PLoS Biology 213; Li et al., Blood 21; *Li/*Kent et al., Blood 214
JAK2 V617F models: dosage, disease and defective HSCs Haematocrit (%) 1 9 8 7 6 5 4 3 2 1 Platelets (x1 3 )/µl 25 2 15 1 5 WT Het Hom WT Het Hom WT Het Hom PV-LIKE: increased red blood cells + splenomegaly Kent et al., PLoS Biology 213; Li et al., Blood 21; *Li/*Kent et al., Blood 214
1 9 8 7 6 5 4 3 2 1 25 Platelets (x13)/µl Haematocrit (%) JAK2V617F models: dosage, disease and defective HSCs 2 15 1 5 WT Het Hom WT Het Hom WT Het Hom MF Transforma6on Kent et al., PLoS Biology 213; Li et al., Blood 21; *Li/*Kent et al., Blood 214
JAK2V617F reduces stem cell number and function 8 Number of HSCs (E-SLAM) per 1, BM WBCs 6 4 2 P a i r e d t t e s t P v a l u e. 28 Donor chimerism (%) from 1 E-SLAM HSCs 2 fold in frequency 8 fold in long-term self renewal Kent et al., PLoS Biology 213
Model HSC JAK2 V617F Old JAK2 V617F cells accumulate DNA damage Relative intensity of γ-h2ax foci to DAPI.5.4.3.2.1 * Whole BM Control * Stem/Prog JAK2 V617F Transformed mice recover HSC activity differentiation clonal exhaustion Impaired HSC function proliferation Additional Mutations Relative HSC activity Clonal restraint Clonal expansion WT Jak2 WT Jak2 WT Trans Jak2 Young Old
What are the additional drivers in patients and how do they alter HSC properties?
TET2 loss of function is the most common collaborating mutation in MPNs 1% of patients harbour a collaborating loss of function TET2 mutation Known epigenetic regulator, mutant mice have more HSCs Multiple clones are detectable within the same patient TET2 KO gives an HSC self-renewal advantage to JAK2V617F cells Chen et al., Blood 215
TET2 KO stabilises transplantable MPN phenotype Proliferation (MPN phenotype) Self-renewal (2 o transplantation) Haematocrit (%) Proportion HSCs completed 1 st division 1 8 1 5 6 4 2 *** *** *** *** *** ** *** 8 16 24 Time (weeks) *** *** 24 48 72 96 12 Time (hours) WT JAK HOM TET HOM DOUBLE WT JAK HOM TET HOM DOUBLE n=572 n=437 n=563 n=432 Donor (%) 8 6 4 2 T m m t Jak WT Hom WT Hom Tet WT WT Hom Hom Mairi Shepherd m
TET2 KO stabilises transplantable MPN phenotype Proliferation (MPN phenotype) Self-renewal (2 o transplantation) Haematocrit (%) 1 8 6 4 2 ** *** *** 8 16 24 Time (weeks) WT JAK HOM TET HOM DOUBLE Donor (%) 8 6 4 2 t T m m m Jak WT Hom WT Hom Tet WT WT Hom Hom Double phenotype + HSC self-renewal Jak Hom fading phenotype; HSC defect Tet Hom HSC SR; no phenotype WT HSC SR; no phenotype Mairi Shepherd
Identifying the molecules driving the difference Bulk gene expression of enriched (KSL) and purified (E-SLAM) HSCs were uninformative Single E-SLAM HSCs 43 HSC genes x 1 single HSCs Single cell gene expression (Moignard et al., 213)
Principal Components Analysis identifies Meis1, Pbx1, Smarcc2, Bmi1, Sfpi1, and Runx1 as key molecules 1 1 1 1 2 3 PC1 PC2 Wild Type JAK2 V617F
Principal Components Analysis identifies Meis1, Pbx1, Smarcc2, Bmi1, Sfpi1, and Runx1 as key molecules.6 Prdm16 Vwf Sfpi1.4 Smarcc2 Meis1 PC2.2 Gata3 Bmi1. Gfi1 Tet2 Procr Itga2b Gata2 Dnmt3a Erg Fli.1 Lyl1 Mpl Csf1r Cbfa2t3h Bptf kit Hoxb4 Tcf7 Nfe2 Foxo3a Sh2b3 Tal1 Myb Lmo2 Gata1 Smarcc1 Ezh2 Hhex Gfi1b Hoxa5 Mecom Ikzf1 Hoxa9 Trib3 Runx1 Bmi1 Sfpi1 Smarcc2 Pbx1 Meis1 Pbx1 Runx1..2.4 PC1 But. How do we know which genes are actually driving the difference?
Identifying the molecules driving the difference Single E-SLAM HSCs WT JAK HOM 43 HSC genes x 2 single HSCs TET KO JAK HOM / TET KO Single cell gene expression (Moignard et al., 213)
TET2 KO rescues HSC defect at molecular level and refines the candidate list of key regulators JAK HOM Double Mutant 2 1 1 2 2 1 1 2 PC1 PC2 Cell_type RR RRTKO TetKO WT JAK2 HOM 2 1 1 2 2 1 1 2 PC1 PC2 Cell_type RR RRTKO TetKO WT TET2 KO 2 1 1 2 2 1 1 2 PC1 PC2 Cell_type RR RRTKO TetKO WT WT
TET2 KO rescues HSC defect at molecular level and refines the candidate list of key regulators Vwf JAK HOM (proliferation) Gata3 TET KO (self-renewal) X Sfpi1 Runx1 Bmi1 X Runx1 Smarcc2 Bmi1 Gfi1 Procr Pbx1 Trib3 Hoxb4 Smarcc2 Mecom Lmo2 Sh2b3 Hoxa5 Myb Hoxa9 Cbfa2t3h Fli.1 Mpl Erg Smarcc1 Foxo3a Nfe2 kit Lyl1 Bptf Gata2 Tcf7 Csf1r Gata1 Ikzf1 Hhex Ezh2 Sfpi1 Itga2b Tal1 Dnmt3a Prdm16 Pbx1 (Meis1) Functional tests underway Gfi1b Double mutant (both)
Disease-associated mutations alter the balance of stem cell molecular subtypes JAK HOM (proliferation) Runx1, Pbx1, Bmi1, Meis1 TET KO (self-renewal) Vwf, Gata3, Prdm16 Double mutant (SR + DIFF) Gfi1b, Dnmt3a, Itga2b
Summary 1. Single cell biology reaches beyond bulk studies, allowing resolu6on of key molecular targets 2. JAK2 V617F mutant HSCs lack key HSC self-renewal regulators 3. TET2 rescue experiments iden6fy drivers of HSC defect 4. HSC subtypes are differen6ally expanded (or absent) in mouse models