Mice

Mice containing the Sin3B-flox (Sin3BF) allele have been previously described. To generate hematopoietic specific deletion of Sin3B, Sin3BF/F mice were intercrossed to Vav1-iCre mice, which is active at embryonic day 11.5 (E11.5). Ptprca; Pepcb (CD45.1) congenic mice were purchased from The Jackson Laboratory and bred in-house to use as recipients for competitive transplantation experiments. All mice were kept on an inbred C57BL/6 background. Mice were housed in pathogen-free barrier facilities with a 12-h light/dark cycle and given food and water ad libitum. Mice were administered 5-Fluorouracil (Invivogen) via intraperitoneal injection at 100 mg/kg body weight. For competitive transplantation assays, recipient mice at 8–10 weeks of age were lethally irradiated (total body irradiation) with 12 Gy of γ-irradiation with a MultiRad 350 X-Ray Irradiator (Faxitron®). Mice were given 2 doses of 6 Gy of irradiation at least 3 h apart. Mice were maintained on sterile, acidified water supplemented with Sulfamethoxazole and Trimethoprim for 2 weeks following irradiation, replenishing the antibiotics after a week. Equal numbers of male and female mice were used in all experiments unless specified otherwise. All animal experiments and protocols were approved by the New York University Grossman School of Medicine Institutional Animal Care and Use Committee.

Flow cytometry and cell sorting

To isolate indicated cell populations, mice were humanely sacrificed via CO2 inhalation and cervical dislocation was used as a secondary means of euthanasia. Femurs, tibiae, and pelvis were isolated from mice, and if increased numbers of cell were required, sternum, humeri, and vertebrae were dissected as well. Whole bone marrow was isolated from bones (femurs, tibiae, pelvis) through spinning in a microcentrifuge for 8 s into FACS-E buffer (1× phosphate buffered saline [PBS] supplemented with 2% fetal bovine serum [FBS] and 25 mM ethylenediaminetetraacetic acid [EDTA]) or through crushing in a mortar and pestle (sternum, humeri, vertebrae).

For whole bone marrow analysis, whole bone marrow was incubated in Ammonium-Chloride-Potassium (ACK) lysis buffer for 5 min on ice to remove erythrocytes. Cells were then resuspended in FACS buffer (1× PBS supplemented with 2% FBS) and incubated with a cocktail of biotinylated antibodies against lineage markers and Rat IgG (20 μg/mL) for 30 min on ice. Cells were washed and then incubated with HSPC antibodies conjugated to fluorophores and Rat IgG for 90 min. Cells are washed and resuspended in FACS buffer carrying 4′,6-diamidino-2-phenylindole (DAPI, 500 ng/mL) to mark dead cells and analyzed on either a Bectin, Dickinson, and Company (BD™) LSR II UV (equipped with 355 nm, 407 nm, 488 nm, 561 nm, 633 nm lasers) or a BD™ LSR II HTS (equipped with 407 nm, 488 nm, 561 nm, 633 nm lasers). Data collection was done using BD FACSDiva™ software and.fcs files were formally analyzed with FlowJo (FlowJo, BD). All flow cytometry experiments contained single color controls for compensation and gating.

For fluorescence activated cell sorting, whole bone marrow was first blocked with TruStain FcX™ PLUS (anti-mouse CD16/32) (50 μg/mL) for 5 min on ice in MACS Buffer (1× PBS supplemented with 1% FBS, 1% bovine serum albumin [BSA], and 2 mM EDTA that was sterile filtered through 0.22 μm filter and de-gassed). Then, cells were incubated with anti-mouse CD117 microbeads (20 μL for femurs, tibiae, and pelvis, 40 μL if also isolating cells from sternum, humeri, and vertebrae) for 15 min on ice. Cells were washed in MACS buffer and then filtered through a 40 μm mesh before being loaded onto a Miltenyi Biotec MS column placed in a miniMACS separator. Flowthrough containing CD117− cells was discarded. MS column was washed with MACS buffer and then flowthrough discarded. Column was then removed from magnet and placed in a microcentrifuge tube. MACS buffer was loaded and plunger used to gently expel cells from the column. This CD117+ enriched fraction was then stained as previously described for lineage markers and HSPC markers before being resuspended in FACS buffer containing DAPI and passed through a 40 μm filter again and sorted on a BD FACSAria™ II (equipped with 355 nm, 407 nm, 488 nm, 561 nm, 633 nm lasers) or a BD FACSAria™ IIu SORP (equipped with 355 nm, 407 nm, 488 nm, 561 nm, 633 nm lasers) utilizing a 100 μm nozzle. Single color controls were used for compensation and florescence minus one controls were utilized to set gates before sorting using FACSDiva software. Cells were sorted into 1× PBS supplemented with 2% FBS. Analysis of sorting data was accomplished with FloJo software. Visualization and statistical analysis was computed after exporting data to GraphPad Prism9 software.

Hematopoietic stem and progenitor immunophenotypes

The following markers were used for the indicated cell types: LT-HSC: L-S+K+Flk2-CD48-CD150+; ST-HSC: L-S+K+Flk2-CD48-CD150-; MPP2: L-S+K+Flk2-CD48+CD150+; MPP3: L-S+K+Flk2-CD48+CD150-; MPP4: L-S+K+Flk2+CD48+CD150-; L (Lineage) markers: CD3, CD4, CD8a, CD11b, B220, Gr1, IL7Rα, Ter-119; S: Sca-1 (Ly6a); K: c-Kit (CD117); Flk2: Flt-3 (CD135).  

Competitive transplantation assay

Recipient CD45.1 mice at 8–10 weeks of age were irradiated the day before transplantation experiments, with split doses of irradiation at least 3 h apart. Donor Sin3BF/F or Sin3BH−/− were used at 6–8 weeks of age. Whole bone marrow was isolated erythrocytes lysed as described in Flow Cytometry and analysis section. Whole bone marrow cells were counted manually using a hemacytometer. 1 × 106 donor wild-type or Sin3B−/− cells were mixed at a 1:1 ratio with wild-type competitor CD45.1. Cells were washed with 1× PBS to remove traces of serum and 2 × 106 cells were resuspended in 100μL sterile 0.22 μm filtered 1× PBS and transplanted into mice via retroorbital injection using a 31G, 6 mm insulin syringe (BD). After 8 weeks, mice were sacrificed and whole bone marrow was isolated and stained as described above for HSPC markers with the addition of antibodies to distinguish between CD45.1 and CD45.2 alleles. Data was analyzed using FloJo and statistically analyzed in Prism9.

EdU incorporation assay

LT-HSCs were sorted as described above, and cultured in 96 well round bottom plates containing 100 μL of HSPC media (5% FBS, Stem Cell Factor [SCF, 25 ng/mL], Interleukin-11 [IL-11, 25 ng/mL], FMS-like tyrosine kinase 3 ligand [Flt-3L, 25 ng/mL], Thrombopoietin [TPO, 25 ng/mL], Interleukin-3 [IL-3, 10 ng/mL], Granulocyte–macrophage colony-stimulating factor [GM-CSF, 10 ng/mL], Erythropoietin [EPO, 4 Units/mL], 1% penicillin G/streptomycin, 2% GlutaMAXTM, 55 μM 2-mercaptoethanol in Iscove’s Modified Dulbecco’s Medium [IMDM]) for indicated time periods in a 37 °C humidified water- jacketed cell culture incubator with 5% CO2. At timepoints, cells were given 100 μL of fresh HSPC media containing 20 μM 5-Ethynyl-2′-deoxyuridine (EdU) for a final concentration of 10 μM. Cells were incubated for one hour, and then plated onto poly-L-lysine coated #1.5 coverslips placed in individual wells of 12 well plates. Cells were allowed to attach for 15 min at room temperature (RT). Then 800μL of BD Cytofix™ buffer was added to fix cells for 10 min at RT with gentle agitation. Then 200 μL of 1 M Glycine in ddH2O was added to quench fixation. Cells were washed three times with 1× PBS before proceeding to EdU staining.

Cells were processed with the Click-iT® Plus EdU Imaging Kit. After washing, cells were permeabilized with Triton X-100 Buffer (0.5%[v/v] Triton X-100; 20 mM HEPES-KOH, pH 7.9; 50 mM NaCl; 3 mM MgCl2; 300 mM sucrose; 0.05% [w/v] NaN3 in ddH2O) for 10 min at RT with gentle agitation. Cells were washed twice with IF Washing Buffer (1% FBS; 1% BSA; 0.1% Triton X-100; 0.1%[v/v] Tween-20; 0.05% NaN3 in 1× PBS) before being incubated with Click-iT® reaction cocktail containing Alexa Fluor® 488 picolyl azide. Cells were incubated with reaction cocktail for 30 min at RT protected from light. Samples were then washed with IF Washing buffer containing 500 ng/mL DAPI, and then washed 3 more times with IF Washing buffer, and then once with 1× PBS, before being mounted on slides with Vectashield® (Vector Labs). Slides were sealed with commercially available clear nail polish and allowed to dry before being imaging on an inverted Zeiss LSM 700 Laser Scanning Confocal Microscope (equipped with 405 nm, 488 nm, 555 nm, 639 lasers) using a 63× plan apochromat 1.4 oil objective. Zen software was used to acquire images, using 3.0 zoom and preventing saturation of images. Images were exported to FIJI (Fiji is just ImageJ) to quantify proportion of cells staining positively for EdU. Quantification was exported to Prism9 for additional analysis and visualization.

Immunofluorescence of LT-HSCs

LT-HSCs from Sin3BF/F and Sin3BH−/− mice were isolated via FACS and plated on poly-l-lysine coverslips as described above for EdU labeling. After permeabilization, cells were washed twice with 1× PBS, and then blocked with 1× PBS supplemented with 5% (v/v) goat serum and 0.1% (v/v) Tween-20 for one hour at RT with gentle agitation. Coverslips were then flipped onto 100 μL droplets of blocking buffer (10% FBS, 2.5% BSA, 0.1% Tween-20, 0.1% Triton X-100, 0.05% NaN3 in 1× PBS) containing primary antibody at a 1:100 dilution. Cells were incubated with primary antibody for 1 h at RT. Coverslips were then returned to individual wells of 12 well plate and washed 3 times with IF washing buffer. Coverslips were flipped again onto droplets of blocking buffer containing secondary antibody. Goat anti-rabbit Alexa Fluor 488 or Goat anti-mouse Alexa Fluor 594 antibodies (Invitrogen) at a 1:400 dilution were used and cells were incubated for 1 h at RT protected from light. Cells were then washed 3 times in IF buffer, with the first wash carrying DAPI (500 ng/mL). Coverslips were washed once with 1× PBS before being mounted on slides in Vectashield. Samples were sealed with nail polish and imaged using a Zeiss 700 as detailed above. Z-stacks were taken and the widest slice of each cell was utilized for quantification using FIJI. A mask was manually drawn using DAPI to quantify the signal for the indicated antibody within the nucleus. A small area not containing cells was quantified to determine background. Corrected fluorescence was determined by the following formula: CorrFluor = Sample – (Area × Background). All data was exported to Prism9 for statistical analysis.

Single cell RNA-sequencing: cDNA library preparation and sequencing

For homeostasis dataset, whole bone marrow was isolated as described above, and antibodies against Lineage, Sca1, and cKit were for 2 Sin3BF/F and 2 Sin3BH−/− male mice. LSKs were pooled and cells were blocked again with TruStain FcX PLUS for 5 min before wild-type cells were incubated with TotalSeq™-B0096 anti-mouse CD45 antibody and knockout cells with TotalSeq™-B0157 anti-mouse CD45.2 antibody for 15 min. Cells were washed with FACS buffer before being manually counted with a hemacytometer. The 10× Genomics Chromium Single Cell 3′ v3 kit was used to generate single cell suspensions. After counting, 2.5 × 104 cells from each genotype were combined with MasterMix and loaded onto a Chromium Chip along with the gel beads and partitioning oil. Gasket was carefully placed over chip and was loaded into a Single Cell Controller. GEMs were carefully pipetted out and visually inspected before being placed in a thermal cycler for cDNA synthesis with the following parameters (Step 1—53 °C for 45 min; Step 2—85 °C for 5 min; Step 3—4 °C: Hold).

Reaction was stored at − 20 °C until cDNA library preparation. Quality of cDNA was checked with an Agilent 2100 Bioanalyzer. Libraries were generated via using the 10× Genomics 3′ GEM protocol with HTO primers and sequenced on an Illumina NovaSeq.

For dataset of LSKs at stress, the same workflow was utilized, except mice were first given a single dose of 5-fluorauracil (100 mg/kg) injected intraperitoneally. After 9 days, mice were sacrificed and bone marrow was processed and stained as previously described and LSKs were isolated via FACS utilizing the same strategy.

Single cell RNA-Seq: analysis

After sequencing, reads from cDNA and Hashtag oligos (HTOs) were demultiplexed and aligned using the 10× Genomics CellRanger software. This generated a matrix file, features file, and barcodes file that was imported into R. The Seurat package was used for downstream analysis. Briefly, the output files from CellRanger were used to generate a Seurat object. HTO counts were extracted and added to the metadata of captured cells. Data was normalized and the HTODemux function was used to classify cells. Only singlets were kept for downstream analysis. Quality control was used to filter to calculate the distribution of genes per cell identified, as well as overall counts and the proportion of reads coming from the mitochondria. Cells in the top and bottom 2% of these metrics were filtered out. Cells found to be expressing any lineage markers used in the FACS isolation were also removed. In addition, cells expressing genes related to biased and lineage-primed HSCs were also removed.

Next, LSKs were assigned a subset identify utilizing previously published transcriptional signatures of LT-HSCs, ST-HSCs, MPP2s, MPP3s, and MPP4s. To generate Uniform Manifold Approximation Projections (UMAPs) data was scaled and Principal Component Analysis (PCA) was computed. JackStraw was then used to calculate statistically significant PC’s to use for UMAP analysis. Differentially expressed genes were determined using the FindMarkers function. The dataset containing LSKs at stress was analyzed using the same workflow as just described.

To integrate the homeostasis and stress datasets, we utilized Seurat’s IntegrateData function. First, variable features were calculated for both datasets, and integration anchors were calculated using FindIntegrationAnchors. These anchors were then used for the IntegrateData function to generate the object containing cells from homeostasis and stress.

For pseudotime analysis, the Monocle3 package in R was used. First, the relevant Seurat object was converted into a CellDataSet format using the as, cell_data_set function from the SeuratWrappers package. The UMAP calculation and LSK subsets defined in Seurat were used in the learn_graph function of Monocle3 when determining cell trajectories. The order_cells function was used to determine pseudotime, with the node containing the most LT-HSCs manually selected as the beginning of the pseudotime trajectory. The graph autocorrelation analysis was completed using the graph_test function using the “principal_graph” that constituted the previously calculated trajectory in learn_graph and order_cells. Genes with a q_value < 0.05 were selected and modules of co-regulated genes as a function of pseudotime were determined via find_gene_modules. Aggregate expression of genes within individual modules was accomplished via the aggregate_gene_expression function and graphed. Modules were exported and gene lists uploaded to Enrichr to determine gene ontologies.

For cell cycle analysis, cell cycle scores utilizing previously published datasets were calculated using log10 transformed expression. LT-HSCs were ordered from M/G1, G0, G0/G1, G1/S, S, G2/M, and M. Cells were ordered from lowest to highest expression of their G0 score, and data were transformed into percentile ranks to normalize for cell number.

Assay for transposase-accessible chromatin using sequencing (ATAC-Seq)

To conduct ATAC-Seq on LT-HSCs, we utilized the ATAC-Seq kit from ActiveMotif with the following modifications. LT-HSCs from 2 mice per genotype in duplicate were pooled after sorting via FACS as previously described. The amount of Assembled Transposomes was scaled based on the number of cells we were able to isolate. After tagmentation for 30 min, the rest of the kit was followed per manufacturer’s instructions. Briefly, DNA Purification Binding Buffer was added to samples and transferred to a DNA binding column. Columns were washed and DNA was eluted for subsequent PCR amplification. Illumina’s indexed i7 and i5 Nextera primers were used to distinguish between samples. DNA was amplified using Q5 polymerase and specific primer combinations for 10 cycles with the following conditions on a thermal cycler (Step 1—72 °C, 5 min; Step 2—98 °C, 30 s; Step 3—98 °C, 10 s; Step 4—63 °C, 30 s; Step 5—72 °C, 1 min; Repeat Steps 3–5 nine more times for a total of 10 cycles; Step 6–10 °C, hold). SPRI (Solid Phase Reversible Immobilization) beads were used for clean-up. Beads were washed twice with ethanol and DNA was eluted from beads. Size distribution of libraries was determined using a TapeStation, and concentration with Bioanalyzer.

Our samples required additional PCR cycles and the same process was repeated on libraries for an additional 2 cycles before bead clean-up was repeated. Libraries were sequenced on an Illumina NovaSeq. Fastq files were run through FastQC and trimmed. Reads were then aligned using bowtie2 and duplicates were removed using sambamba. Bigwig files were generated using deeptools and peaks were called using MACS2.

Bed files containing peaks were imported into R and the DiffBind package was used for downstream analysis. Peaksets were read in and normalized before differential analysis was calculated using DESeq2. The HOMER package was used to annotate peaks and to calculate enrichment of DNA-binding factor motifs. The bedtools suite was utilized to compare peaksets to each other, with the intersect function used to directly compare lists of accessible peaks. Peaks were then fed into the GREAT tool using default parameters to determine putative genes regulated by the accessible chromatin peaks we identified. Finally, those genes were then used as input for Enrichr to determine Gene Ontology enrichment. Individual ATAC-Seq tracks were loaded by opening bigwig files in IGV.

For motif analysis, the HOMER function findMotifsGenome.pl was used for individual peak lists. Primed peaks list were taken from Martin, et al. [52] and first changed to an mm10 annotation format using the UCSC genome browser tool. Lists were directly compared to ATAC-seq peaks using bedtools.

Statistical considerations

Samples were compared using the statistical test indicated in figure legend.

Sample sizes were not determined with any formal power calculation.