A complex interplay of intra- and extracellular factors regulates the outcome of fetal- and adult-derived MLL-rearranged leukemia

Infant and adult MLL1/KMT2A-rearranged (MLLr) leukemia represents a disease with a dismal prognosis. Here, we present a functional and proteomic characterization of in utero-initiated and adult-onset MLLr leukemia. We reveal that fetal MLL::ENL-expressing lymphomyeloid multipotent progenitors (LMPPs) are intrinsically programmed towards a lymphoid fate but give rise to myeloid leukemia in vivo, highlighting a complex interplay of intra- and extracellular factors in determining disease subtype. We characterize early proteomic events of MLL::ENL-mediated transformation in fetal and adult blood progenitors and reveal that whereas adult pre-leukemic cells are mainly characterized by retained myeloid features and downregulation of ribosomal and metabolic proteins, expression of MLL::ENL in fetal LMPPs leads to enrichment of translation-associated and histone deacetylases signaling proteins, and decreased expression of inflammation and myeloid differentiation proteins. Integrating the proteome of pre-leukemic cells with their secretome and the proteomic composition of the extracellular environment of normal progenitors highlights differential regulation of Igf2 bioavailability, as well as of VLA-4 dimer and its ligandome, upon initiation of fetal- and adult-origin leukemia, with implications for human MLLr leukemia cells’ ability to communicate with their environment through granule proteins. Our study has uncovered opportunities for targeting ontogeny-specific proteomic vulnerabilities in in utero-initiated and adult-onset MLLr leukemia.

Systems.In all flow cytometry and FACS experiments, cells were incubated with 7AAD (Merck) briefly before analysis to exclude dead cells.All flow cytometry and FACS experiments were performed on BD FACSAriaIIu (70 µm nozzle), BD FACSAriaIII (70 or 85 µm nozzle), BD LSRFortessa or BD LSRFortessa X-20 instruments at the FACS Core Facility at Lund Stem Cell Center.Data analysis was performed in FlowJo (BD).

Sample preparation for cellular proteome analysis
FACS-sorted cells were collected in ice-cold Hank's Balanced Salt Solution (HBSS), centrifuged, and stored as dry pellets at -80 °C until further use.Pellets corresponding to 40,000 cells were processed using in-StageTip (iST) NHS sample preparation kit (PreOmics) in accordance with manufacturer's protocol.Digested peptides were labelled using TMTpro reagents (Thermo Scientific).Immediately before use, TMTpro reagents were equilibrated to room temperature (RT).Vials containing 0.5 mg of TMT label were dissolved in 20 µl of anhydrous acetonitrile (ACN).Labelling was performed by addition of 5 µl dissolved TMTpro to each sample and incubation for 1 hour at RT.Following desalting, labelled peptides were combined and dried by vacuum centrifugation.High-pH-reverse phase (HpH-RP) prefractionation was carried out as previously described 2,3 .Fractionated samples were dried by vacuum centrifugation and stored at -20 °C until further use.Prior to LC-MS analysis, samples were dissolved in MS loading buffer (4% ACN/0.1% formic acid (FA)).

Sample preparation for EF proteome analysis
The EF collection protocol was modified from previous procedures 4,5 .One femur and one tibia per mouse (WT) were punctured at both ends, placed on in-house made '3-trap' tubes of a PCR tube, a 0.5 ml tube, and a 1.5 ml tube, where the bottoms of the PCR tube and the 0.5 ml tube were punctured with a 21G needle.50 µl sterile PBS was added to the PCR tube, and the samples were centrifuged at low speed (300 g) for 5 minutes at 4°C.Bones were taken out from the PCR tube and the '3-trap' tubes were centrifuged for another 2 minutes at 500 g, 4°C to collect the BMEF.FLs from 9 WT embryos (E14.5) were washed with sterile PBS and placed into a 40 µm cell strainer.50 µl sterile PBS was injected into the FL before centrifugation at 400 g for 10 minutes at 4°C to obtain the FLEF.BMEF and FLEF samples were next cleared at 2000 g for 20 minutes at 4°C, snap frozen and kept at -80°C for further processing.Protein concentration was estimated with BCA assay (Thermo Scientific).The nine FLEF samples were pooled into four samples.BMEF and FLEF samples were transferred into Amicon 3K filters and concentrated at 14000 g, room temperature until 20 µl of concentrated sample remained in the column.Filters were then inverted into new collection tubes and centrifuged at 1000 g for 2 minutes at room temperature.For proteomic analysis, approximately 50 µg of concentrated EF samples were diluted in 50 mM ammonium bicarbonate (Sigma) with 0.1% RapiGest (Waters) to denature proteins and shaken on a thermomixer (Eppendorf) at 400 rpm for 15 minutes decreasing the temperature from 80°C to 56°C, at 400 rpm, followed by reduction of disulfide bonds with 0.1M dithiothreitol at 56°C, cysteine alkylation with 0.2 M iodoacetamide at room temperature, and digestion overnight at 37°C with sequencing grade modified trypsin (enzyme:protein ratio 1:50, Promega).Digested peptides were acidified with 10% trifluoroacetic acid (TFA) and RapiGest was precipitated by incubation at 37°C.Peptides were desalted, dried by vacuum centrifugation, and resuspended in MS loading buffer prior to LC-MS analysis.Peptide concentrations were measured at 280 nm (Nanodrop 2000, Thermo Scientific) and 1 µg peptide was injected for LC-MS/MS analysis.

Polyvinyl alcohol (PVA) culture of LMPPs
FACS-sorted 200 fetal or 1000 adult LMPPs per well were cultured in a round-bottom 96-well plate in serum-free media composed of Ham's F12 media (Gibco), 1% Insulin-Transferrin -Selenium-Ethanolamine (ITSX, Gibco), 10 mM HEPES, 1% Penicillin-Streptomycin-Glutamine (Gibco), 100 ng/mL murine TPO (Peprotech), 10 ng/mL murine SCF (Peprotech), 1 mg/mL PVA (Sigma Aldrich) and 25 ng/mL Flt3l (Stem Cell Technologies) for adult LMPPs or 10 ng/mL Flt3l for fetal LMPPs.MLL::ENL expression was induced by addition of 1 µg/mL DOX.Cells without DOX treatment were seeded as control (also called WT).For Fbln1 and Fn1 assays, 2 µg/mL recombinant human Fbln1 and/or 6 µg/mL recombinant Fn1 (R&D Systems) were added to cultures with DOX.An additional 0.5 µg/mL DOX was added at day 4 of cultures.For fetal LMPPs, media was collected on day 6 of culture, and on day 8 for the adult counterpart.Cells were not passaged during the cultures, and conditioned media were collected without disturbing the cells.PVA media without the cells were collected as negative control of the secretome.

Secretome and phospho-proteome analysis of human cell lines
For the secretome and phospho-proteome analysis of human leukemia cells, THP-1 and KOPN-8 cells were washed three times with PBS and then cultured at 2x10 6 cells per well in serum-free RPMI 1640 (Gibco) media.Recombinant human IGF2 (R&D Systems) was added at 10 µg/mL to the cultures.Conditioned media were collected 17 hours after treatment and the cells were washed with PBS, and collected into new tubes for the phospho-proteome analysis.
Media without the cells were collected as negative control of the secretome.

Sample preparation for secretome analysis
The collected media from the Fbln1 and/or Fn1-treated LMPPs, DOX-treated and untreated LMPPs, media without cells, and IGF2-treated and untreated THP-1 and KOPN-8 cells were centrifuged at 350 g for 5 min at 4°C.The supernatant was transferred to LoBind Tubes and stored at -80°C until further analysis.Four volumes of ice-cold acetone were added to the samples, vortexed, and incubated for 1 hour at -20°C.The samples were centrifuged at 15,000 g for 10 min at RT and the supernatant was discarded.The precipitated pellets were dissolved in 50 mM ammonium bicarbonate (Sigma) to denature proteins followed by reduction of disulfide bonds with 0.1M dithiothreitol at 56°C, cysteine alkylation with 0.2 M iodoacetamide at room temperature, and digestion overnight at 37°C with sequencing grade modified trypsin (enzyme:protein ratio 1:50, Promega).Digested peptides were acidified with 10% trifluoroacetic acid (TFA), desalted, and dried by vacuum centrifugation.The dried peptides were resuspended in MS loading buffer prior to LC-MS analysis.Peptide concentrations were measured at 280 nm (Nanodrop 2000, Thermo Scientific) and the sample was injected twice.

Sample preparation for phospho-proteome analysis
Phosphopeptide enrichment for the IGF2-treated and untreated THP-1 and KOPN-8 cells was performed as described previously 6 using the Pierce High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo) according to manufacturer instructions.The unbound fraction and washes (flow-through) from the enrichment were combined for the corresponding proteome analysis.

Suspension culture assays
For suspension culture experiments, fetal and/or adult iMLL::ENL LMPPs were FACS-sorted into 96-well plates.Cells were cultured in complete medium supplemented with 10 ng/mL SCF, 10 ng/mL Flt3l, 10 ng/mL IL-7 and 5 ng/mL IL-3 for 3 or 4 days.MLL::ENL expression was induced by addition of 1 µg/ml DOX.For IGF2 assays, recombinant mouse or human IGF2 purchased from R&D Systems was added at 50 or 2000 ng/mL (mouse) and 2000 ng/mL (human) to the cultures, and cells analyzed by flow cytometry at day 4 for mouse cells and day 2 for human cells.To partially block the Igf2r, cells were treated with 2000 ng/ml IGF2R (R&D Systems) for 30 minutes prior to the IGF2-treatment.For the Fn1 and Fbln1 assays, the cells were treated with recombinant human Fbln1 (2 µg/ml; R&D Systems) and/or recombinant human Fn1(6 µg/ml; R&D Systems) and analyzed by flow cytometry at day 4 for mouse cells and day 2 for human cells.

Ex vivo lineage-potential analysis
Starting from day 4 of culture, cells were collected from wells every 3 days for flow cytometric analysis.Prior to flow cytometric analysis, cells were incubated with FC-block and surfacestained with fluorophore-conjugated antibodies against cKit, Flt3, CD11b, CD19, Ly6G, B220, NK1.1 and CD11c.

In vivo leukemia analysis
Fetal and adult MLL::ENL LMPPs were FACS-sorted and frozen at -80°C together with unfractionated WT BM in fetal bovine serum (FBS) containing 10% dimethyl sulfoxide (DMSO).For neonatal recipients, pregnant WT females were put on a DOX-containing diet at E18.5.WT adult recipients were put on a DOX-containing diet 4 days prior to transplantation.Neonatal (24-48 h old) and adult recipients were pre-conditioned by irradiation with 350 and 900 cGy, respectively.Frozen cells were thawed and resuspended in sterile PBS and a volume corresponding to 2000 MLL::ENL LMPPs and 300,000 WT support cells was injected intravenously via the facial vein (neonates) or tail vein (adults).Disease progression was assessed in peripheral blood (PB) every 3 weeks starting from 4 weeks post transplantation by Sysmex and flow cytometry.For flow cytometry analysis, red blood cells (RBCs) were lysed by incubating PB cells with ammonium chloride solution for 10 min at RT.After removal of RBCs, PB cells were incubated with FC-block and surface-stained with fluorophoreconjugated antibodies against CD45.1 (A20; BioLegend), CD45.2 (104; BioLegend), CD11b, Ly6G, CD115 (AFS98; BioLegend), B220 and CD3 (145-2C11; BioLegend) prior to flow cytometric analysis.Mice were euthanized when they became moribund.In addition to PB, spleen and BM were collected and analyzed by flow cytometry to determine disease type.
Following RBC lysis, spleen and BM cells were incubated with FC-block and surface-stained with fluorophore-conjugated antibodies against CD45.1,CD45.2, CD11b, Ly6G, B220, CD19, cKit, Flt3, CD3 and Ter119 (TER119; BioLegend) prior to flow cytometric analysis.In cases where mice died prior to euthanasia could be performed, spleen size and weight were examined to determine if leukemia was the cause of death.These animals were included in survival analyses but not subjected to flow cytometric analysis due to extensive necrosis in the tissues.

Liquid chromatography and mass spectrometry
LC-MS analyses were carried out on an Orbitrap Exploris 480 MS instrument with a reverse phase UltiMate 3000 UHPLC system via an EASY-Spray ion source equipped with FAIMS Pro (all Thermo Fisher Scientific).For TMTpro labeled peptides, each fraction was injected twice and separated using a 120 min linear gradient separation followed by tandem MS.
Injected peptides were loaded onto a trap cartridge (Acclaim PepMap C18, 5 μm, 300 µm x 5 mm, Thermo Fisher Scientific), followed by gradient elution of peptides on an EASY-Spray analytical column (2 μm particle size, 75 μm inner diameter x 500 mm length, Thermo Fisher Scientific) using 0.1% (v/v) FA in LC-MS-grade water (solvent A) and 0.1% FA in 80% ACN (solvent B) as the mobile phases.Peptides were loaded with a constant flow of solvent A at 5 μl/min onto the trapping column and eluted via the analytical column at a constant flow of 300 nl/min.Equilibration of the column was performed for 14 minutes at a 2% concentration of solvent B. During the elution step, the percentage of solvent B was increased in a linear fashion from 2% to 4% in 3 minutes, to 16% in 83 minutes, to 25% in 45 minutes, and finally to 85% in an additional 5 minutes.FAIMS compensation voltage (CV) was set to -40, -60, and -80.
The spray voltage was set at 2.1 kV and the ion transfer tube temperature was set at 275°C.Data acquisition was carried out using a data-dependent 'Top speed' MS2-method (1 s/CV).
The RF lens was set to 40%.The full MS scan was performed in the Orbitrap in the range of 400 to 1400 m/z at a resolution of 120,000 at full-width-half-max (FWHM) using an automatic gain control (AGC) of 300% and a maximum ion accumulation time of 50 ms.The intensity threshold was set to 5.0 e4 and mass tolerance to 10 ppm.The most intense ions selected in the first MS scan were isolated for higher-energy collision-induced dissociation (HCD) at a precursor isolation window width of 0.7 m/z, an AGC of 200%, a maximum ion accumulation time of 120 ms and a resolution of 45,000 FWHM.The first mass and the normalized collision energy were set to 110 m/z and 32%, respectively.EFs, secretome, phospho-enriched, and phospho-unbound (flow-through/proteome) samples were analyzed by data-independent acquisition (DIA).Digested peptides were loaded onto a trap cartridge (Acclaim PepMap C18, 5 mm particle size, 0.3 mm inner diameter x 5 mm length, Thermo Fisher Scientific) and separated by EASY-Spray analytical column (2 mm particle size, 75 mm inner diameter x 500 mm length, Thermo Fisher Scientific).Each sample was injected twice and eluted with a linear gradient ranging from 2-25% Solvent B (0.1% FA in 80% ACN) over 100 min, 25-40% B over 20 min, 40-90% B over 2 min and held at 90% B for 5 min at a constant flow rate of 300 nl/min at 45°C.FAIMS compensation voltages (CV) were set to -45 and -60.The spray voltage was set at 2.1 kV and the ion transfer tube temperature was set at 275°C.For DIA analysis, peptides were analyzed with one full scan (340-1,300 m/z, R = 120,000) at a normalized AGC target of 300%, followed by 29 (EF and phospho-enriched samples) or 20 (secretome and phospho-unbound (proteome) samples) DIA MS/MS scans (350-1,050 m/z) in HCD mode (isolation window 23.4 (EF and phosphoenriched samples) or 35 m/z (secretome and phospho-unbound (proteome) samples), 1 m/z window overlap, normalized collision energy 30%), with fragments detected in the Orbitrap (R = 15,000).All data were acquired in positive polarity and MS/MS were acquired in centroid mode (except for the phospho-enriched samples which were acquired in profile mode).

MS raw data processing and protein identification
The MS raw files from the TMTpro experiment were searched in Proteome Discoverer (version 2.5, Thermo Scientific) against the Swissprot mouse database together with isoforms of the human MLL protein, peptide sequences for the MLL::ENL fusion protein, as well as commonly observed contaminants and reversed sequences for all entries using the Sequest HT node.The enzyme was set to trypsin with up to two missed cleavages.Cysteine acetylhypusinylation, N-terminal and lysine TMTpro were set as static modifications whereas methionine oxidation, N-terminal acetylation, were set as dynamic modifications.TMT batchspecific isotopic correction factor was applied in the reporter ion quantification.MS1 mass tolerance was set to 10 ppm and MS2 to 0.02 Da.The false discovery rate for peptide-spectrum matches (PSMs) was set to 0.01 using the Percolator node.The co-isolation threshold was set to 50.Proteins were quantified based on the average corrected TMT reporter ion intensities from two technical replicates per sample.
The MS data of the single-shot BMEF and FLEF samples, secretome samples, and phosphoproteome samples were searched with 'directDIA' in Spectronaut (version 17 and 18, Biognosys AG) against the mouse or human SwissProt reference proteome along with commonly used contaminants.Searches used carbamidomethylation as fixed modification and acetylation of the protein N-terminus and oxidation of methionines or 'STY' for the phosphoenriched samples, as variable modifications.The Trypsin/P proteolytic cleavage rule was used, permitting a maximum of 2 missed cleavages and a minimum peptide length of 7 amino acids.
Data filtering was set to Q-value.'Cross run normalization' was enabled with Normalization Strategy set to 'local normalization' based on rows with 'Identified in All Runs (Complete)'.
Normalization filter type was set to 'Phospho (STY)' additionally for the phospho-proteome analysis.The Q-value thresholds were set to 0.01 at PSM, peptide, and protein levels.

MS statistical analyses and bioinformatics analysis
Statistical analysis of the TMTpro quantification was performed using MSStatsTMT (version 2.4.1) 7 in R. PSM results from Proteome Discoverer were exported and converted into MSstatsTMT-compatible format using 'PDtoMSstatsTMTFormat' function.PSMs were filtered with peptide percolator q-value < 0.01.Only unique peptides were used for protein quantifications.Protein summarization was performed using the 'msstats' method and global median normalization was performed.Differential expression analysis was performed using moderated t-tests with Benjamini-Hochberg (BH) multiple hypothesis correction.Proteins with adjusted p-value less than 0.05 between MLL::ENL and WT in fetal or adult were considered as differentially expressed.Principal component analysis (PCA) was performed using the top 300 most variably expressed proteins using the DESeq, statmod and PCAtools packages in R. Mapping of PC loadings to transcriptome data from BloodSpot was carried out using the ''normal mouse hematopoiesis'' dataset 34 .A radar plot was generated using min-max scaled median values of marker genes in each cell type.
For the DIA datasets, 'features' were exported from Spectronaut for statistical analysis with MSstats 7 (version 4.4.1 and 4.8.7).For the phospho-proteome analysis, phospho-enriched and unenriched datasets were analysed together with MSstatsPTM (2.4.1) 8 .Contaminants were filtered and features were converted to MSStats format for downstream processing.For the secretome datasets, proteins identified in the respective control media were also manually excluded from downstream processing.Uninformative features and outliers were removed (one of six FL WT and one of six FL MLLr) and missing values were imputed for both the proteome and phospho-proteome analysis.Phospho-site level quantification was performed for the phospho-proteome datasets.For the EF dataset, MSStats group comparison was done for FLEF versus BMEF, while for the secretome datasets of fetal and adult LMPPs, the comparisons were MLL::ENL versus WT, as well as MLL::ENL versus Fn1, MLL::ENL versus Fbln1, and MLL::ENL versus Fn1+Fbln1.The secretomes of IGF2 treated THP-1 and KOPN-8 cells were compared with the experimental control.The BH method was used to account for multiple testing.Differentially expressed proteins were selected with adjusted p-value less than 0.001 and a fold change of more than 2 between FLEF and BMEF, adjusted p-value < 0.05 for the secretome, adjusted p-value < 0.05 and a fold change of more than 1.5 for the phosphoproteome analysis.Protein or phospho-site level abundance per sample or condition was used for further analysis and plotting.Protein subcellular localization was predicted with DeepLoc 2.0 9 and proteins with annotation 'extracellular' were extracted.Gene set enrichment and overrepresentation analysis were performed with ClusterProfiler 10 .Ligand-receptor interactions were curated by mapping the identified intracellular (TMTpro MLL::ENL vs WT) and extracellular (BMEF vs FLEF) proteins to the mouse ligand-receptor pairs downloaded from CelltalkDB 11 (http://tcm.zju.edu.cn/celltalkdb/) and Cellinker 12 (http://www.rna-society.org/cellinker/)database.Receptors were also mapped in our previous study 13 and the corresponding ligands were retrieved.The resulting networks and subnetworks were visualized in Cytoscape 14 .For the secretome data, protein-protein interactions were retrieved from STRING 15 .

Statistical analysis
For all other experiments, differences between groups were assessed by two-tailed Students' ttest (two groups) or one-way ANOVA with Tukey's post hoc test (three or more groups) using

Table S1. Differential protein expression of proteins in fetal and adult WT and leukemic
cells.Related to Figure 3. Table S2.GSEA enrichment of fetal and adult leukemic vs WT cells.Related to Figure 4. Table S3.Differential protein expression in extracellular fluids of fetal liver and adult bone marrow.Related to Figure 5. Table S4.GSEA enrichment of FLEF vs BMEF.Related to Figure 5. 6 and S6.

Table S6. Differential protein expression in the secretome of fetal and adult MLL::ENL and WT LMPPs and differential protein expression in the secretome of fetal and adult
LMPPs treated with Fbln1 and/or Fn1.Related to Figures 7 and S9.Table S7.Differential protein expression in the secretome of IGF2 treated and untreated THP-1 and KOPN-8 leukemic cell lines.Related to Figures 8 and S11.

Figure S6. Integrative analysis of the extra-and intracellular proteome of fetal and adult
HSPCs.(A) Protein network of extracellular ligands (triangles) of the proteomic comparison of FLEF and BMEF, and cell surface receptors (circles) of the cellular proteomic comparisons of fetal and adult HSPCs 10 .Edges represent known and predicted receptor-ligandinteractions 11,12 .sign.= significant, n.s.= not significant.(B) Protein network of extracellular ligands (triangles) of the proteomic comparison of FLEF and BMEF, and the receptors Itga4 and Itgb1 (circles) of the cellular proteomic comparisons of fetal and adult HSPCs 13 .Edges represent known and predicted receptor-ligand interactions 11,12 .sign.= significant, n.s.= not significant.Related to Figure 6 and Table S5.S6.S7 and S8.

Figure S2 .
Figure S2.Transplantation of fetal and adult iMLL::ENL LMPPs into neonatal and adult

Figure S4 .
Figure S4.Correlation in protein expression differences between WT and leukemic cells

Figure S5 .
Figure S5.Details of FLEF and BMEF proteomic analysis.(A) Workflow for the proteomic

Figure S7 .
Figure S7.Fn1 and Fbln1 treatment of fetal and adult pre-leukemic cells.(A) Workflow

Figure S9 .
Figure S9.Proteomic analysis of the secretome of fetal and adult pre-leukemic cells upon

Figure S10 .
Figure S10.IGF2 treatment differentially affects fetal and adult cells.(A, B) Frequency of