Human Bone Marrow Organoids for Disease Modeling, Discovery, and Validation of Therapeutic Targets in Hematologic Malignancies

Development of human bone marrow organoids that capture key cellular, molecular, and architectural features of hematopoietic tissue enables disease modeling and target screening using primary cells from patients with myeloid and lymphoid blood cancers.


Supplementary Materials and Methods iPSC Culture and Differentiation
A Gibco Human Episomal iPSC (Thermo Fisher Scientific Cat#A18945) line was maintained in StemFlex medium (Thermo Fisher Scientific Cat # A3349401) and on Geltrex (Thermo Fisher Scientific Cat#A1569601)-coated 6-well plates. The iPSC line was karyotyped prior to use (1) and potency markers assessed upon expansion and freezing. Cells were passaged as clumps using EDTA at 0.02% in PBS (0.5mM, Sigma Cat#E8008), and were freshly thawed or passaged for differentiation and maintained in StemFlex supplemented with RevitaCell (Thermo Fisher Scientific Cat#2644501). Cultures were maintained at 37°C and 5% CO2. Differentiations were initiated with iPSCs between passages 5 and 30. Cell aggregates were then collected by gravitation and re-suspended in Phase II medium for a further 48 hours (d3-5). Phase II medium (APEL2 supplemented with BMP-4, FGF2, and VEGFA at 50ng/mL; human Stem Cell Factor (hSCF, StemCell Technologies Cat#78062) and Fms-like tyrosine kinase-3 Ligand (Flt3, StemCell Technologies Cat#78009) at 25ng/mL. On d5 cells were collected by gravitation for hydrogel embedding. Hydrogels were composed of 60% collagen (either type I, type IV, or an equal parts type I+IV mix) and 40% Matrigel. Hydrogels were prepared on ice as per manufacturer's instructions and were comprised of Reduced Growth Factor Matrigel (Corning, Cat#354230) supplemented with 1mg/mL human collagen type I (Advanced Biomatrix, Cat#5007) and human collagen type IV (Advanced Biomatrix, Cat#5022) as per designated gel composition. Hydrogel mixes were neutralised with 1N NaOH. An 0.5mL cell-free base layer was added and allowed to polymerise for 2 hours, before a further 0.5mL layer of gel supplemented with gravitated cell aggregates was added and also left to polymerise for 2 hours at 37°C and 5%CO2. Fully polymerised gels with cell aggregates were then supplemented with Phase III media comprised of VEGFA at either 50ng or 25ng/mL, VEGFC (where relevant) at 50 or 25ng/mL, FGF2, BMP4, hSCF, Flt3, Erythropoeitin (EPO, StemCell Technologies, Cat#78007), Thrombopoeitin (TPO, StemCell Technologies, Cat#78210), Granuolocytic Colony-Stimulating Factor (G-CSF, StemCell Technologies, Cat#78012), at 25ng/mL, and Interleukin-3 (IL3, StemCell Technologies, Cat#78194) and Interleukin-6 (IL6, StemCell Technologies, Cat#78050) at 10ng/mL. Media was replenished every 72 hours.
Whole organoid blocking solution was further supplemented with Triton X100, Tween, and Sodium deoxycholate as described by Wimmer et al (3).

Microscopy and Image Analysis
Confocal microscopy was performed using a Zeiss LSM880 confocal AiryScan microscope with either a 25X LD LCI plan apo 0.8 NA dual immersion (420852-9871-000) or 40x C-APO NA 1.2 water immersion objective (421767-9971-711) as described previously (1). Confocal images were acquired as representative Z-stacks (with Z-resolution set to Nyquist requirements), and presented as maximum intensity projections (Fiji)(4) where stated. Histological preparations (reticulin and H&E, details provided in supplementary materials and methods) were imaged using a Zeiss AxioScan.Z1 slide scanner. Image analysis was performed in Fiji. For measurements of sprout radii, brightfield images acquired on an Evos (Thermo Fisher Scientific) desktop microscope. Sprout radii were measured manually by drawing and measuring a line from the centre to the tip of the sprout across 3 independent biological replicates, with between 30-50 sprouts measured per replicate. To measure the proximity of megakaryocytes to organoid blood vessels, 250μm x 50μm volumes of individual organoids were acquired using cleared whole mount organoids imaged by confocal microscopy, as previously described. CD41 labelled megakaryocytes within 5μm of UEA1 labelled vessels were counted as 'vessel-associated MKs' within a maximum intensity projection of each imaged volume (5). Cryopreserved cells pooled from 15 organoids from 3 differentiations from both VEGFA and   VEGFA+C protocols were thawed, stained with DAPI to exclude non-viable cells, and DAPI-live   cells sorted on a Becton Dickinson Aria Fusion with 100nm nozzle as per recommendations in the 10x Genomics Single Cell Protocols -Cell Preparation Guide. 10,000 live cells per sample were sorted into 2μL PBS/0.05% BSA (non-acetylated) and the cell number/volume adjusted to the target for loading onto the 10x Chromium Controller. Samples were processed according to the 10x protocol using the Chromium Single Cell 3′ library and Gel Bead Kits v3.1 (10x Genomics). Cells and reagents were prepared and loaded onto the chip and into the Chromium Controller for droplet generation. Reverse transcription was conducted in the droplets and cDNA recovered through demulsification and bead purification. Pre-amplified cDNA was used for library preparation, multiplexed and sequenced on a Novaseq 6000. Details on data processing ana scRNAseq data processing and analysis

Single-cell RNA-sequencing
Demultiplexed FASTQ files were aligned to the human reference genome (GRCh38/hg38) using standard CellRanger (version 6.0.1) 'cellranger count' pipeline (10x Genomics). SingCellaR (6) (https://supatt-lab.github.io/SingCellaR.Doc/) was used for the downstream analysis. Data was first subject to quality control with the maximum percentage of mitochondrial genes, maximum detected genes and max number of UMIs set to 12%, 6,000, and 50,000, respectively. Minimum detected genes and UMIs were set to 300 and 500, respectively and genes with minimum expressing cells was set as 10. Raw expression matrix was then normalised and scaled and number of UMIs and percentage of mitochondrial reads were regressed out before a general linear model (GLM) was used to identify highly variable genes (7). Data were then subject to downstream analyses including principal component analysis (PCA), UMAP analysis (top 40 PCs were used, and n.neighbour = 120), and clustering using the Louvain method. Differentially expressed genes were calculated using 'identifyDifferentialGenes' function (min.log2FC = 0.3 and min.expFraction = 0.25). To compare cells from the two experimental conditions (VEGFA only and VEGFA+C), cells were downsampled so that each cell group had the same number of cells. Wilcoxon test of normalized UMIs was used to compare the gene expressions and Fisher's exact test was used to compare the cell frequency. The resulting P values from both tests were combined using Fisher's method and subsequently adjusted by Benjamini-Hochberg correction. 'runFA2_ForceDirectedGraph' function was used to identify the trajectories.
CellPhoneDB v 2.1.1 (https://github.com/Teichlab/cellphonedb) was performed for ligand-receptor interactions using normalized expression matrix of VEGFA +C as detailed by Garcia-Alonso et al. (8,9). Cell-cell interaction network between the different cell clusters from VEGFAC and Sankey plot demonstrating the interaction between TGFβ1, CXCL12, and CD44 ligands with their responding receptors from VEFGA and VEFGAC were plotted using a modified version of the CrossTalkeR R package (version 1.2.1) (10).
We applied Symphony (11) to map cells from VEGFA+C organoids to published scRNAseq datasets from human bone marrow (12) and fetal liver and bone marrow cells (6,13) respectively.
For the human bone marrow dataset, we first built the reference data using the normalized expression matrix using 'symphony::buildReference'. For the fetal liver dataset we used the pre-built reference provided by the Symphony developer. The 'mapQuery' and 'knnPredict' function were used to map the VEGFA+C cells onto the three reference datasets.

CellTrace labelling for viability and proliferation assays
Primary cells were labelled with CellTrace Far Red as indicated by the manufacturer. Briefly, cells were washed 1X with PBS and resuspended at 1x10 6

cells/mL in staining solution (CellTrace Far
Red 2µM in PBS). Cells were incubated in staining solution for 30min at 37°C. After incubation CellTrace was quenched with 5 volumes of PBS with FBS (10%), spun down and resuspended in the appropriate media.

Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)
Whole organoids were processed using either the Micro RNEasy Kit (Qiagen, Cat#74004) or Qiagen Mini RNA isolation kit (Qiagen, Cat#74104) according to the manufacturer's instructions.
Isolated RNA was quantified on the NanoDrop ND-100 (Thermo Scientific) and cDNA was prepared using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Cat# 4368814) or EvoScript Universal cDNA Master (Roche, Cat#07912374001) according to the manufacturer's instructions using standard cycling conditions. cDNA was diluted to 5ng before being combined with PowerUp SYBR Green Master Mix reagent (Applied Biosystems, Cat# A25742) and the relevant PrimeTime qRT-PCR primers (IDT), or performed using TaqMan™ Universal PCR Master Mix (Applied Biosystems) on StepOne plus machine (Applied Biosystem) (see Suppl. Table 6 for a full list of primers). The absolute expression of the respective genes was calculated using the ∆Ct method using GAPDH as an internal housekeeping control.

Luminex Assays
To assess the production of growth factors, organoids were washed and cultured in StemPro-34 (L-Glutamine only) without any added supplements or growth factors for 12 days. 50:50 media changes were performed at 72 hour intervals, and media was collected for Luminex assays at day 12. Supernatant from 12 organoids was collected and pooled for each repeat.
Luminex kits (LXSHAM-03, LXSAHM-28) were used for multiplexed proteomic assays as per the manufacturer instructions. No detectable signal was observed in cell-free medium.

Mutation detection by next generation sequencing (NGS)
A custom-made, targeted, hybridization, ultra-deep, next-generation sequence panel for the detection of a panel of commonly mutated genes in clonal hematopoiesis and myeloproliferative neoplasms was used to screen for mutations in samples from myelofibrosis patients post-organoid engraftment (14,15). The combined probe footprint size of the panel was 25,083 bp, and the exact target regions are listed in Suppl. Table 7. Library preparation was performed as per the Twist NGS Workflow manufacturer instructions, including enzymatic DNA fragmentation, end repair and dAtailing; universal adapters ligation; PCR amplification using UDI primers; bead-based, capture probes -pool hybridization; and post-hybridisation target enrichment. Sequencing was carried out on the Illumina NextSeq™ 500 System, using the NextSeq™ 500/550 Mid Output Kit v2.5 (150 Cycles).