Participants, including six men and six women [age 48–85 years (average 70 years)], were recruited from the hematology clinic. All participants signed informed consent prior to participation in the study. Ethical approval was obtained from the Ethic Committee (IORG0002175, IRB00002705, and FWA00029052) of the General University Hospital in Prague (156/22 S-IV). Samples were collected in accordance with the Declaration of Helsinki.

BM samples were obtained by aspiration of 2–4 mL from the iliac crest after administration of the local anesthetic (lidocaine, 10 mg/mL) into a 20 mL syringe and mixed 1:1 with heparin (cat. number 0093746) (100 U/mL) or collected in EDTA-coated tubes (BD Vacutainer K2E EDTA; cat. number 367864). Low-density mononuclear cells were isolated through centrifugation using a Lymphoprep density gradient (density = 1.077 ± 0.001 g/mL) and then selected through the process of plastic adherence. Cells were then cultured at 1 × 10⁵ cells/cm² (1 × 10⁶ cells per chamber slide) in minimal essential media (MEM, Gibco) containing 10% fetal bovine serum (FBS, Gibco) and 1% penicilin/streptomycin (P/S, Gibco), incubated at 5% CO₂ at 37°C, and then, nourished by completely changing the medium once a week along with the passage at 70% confluence. Cultivated cells were sub-cultured and further studied under differentiation conditions to induce adipogenesis and osteogenesis, respectively. Due to the limited number of hBMSCs, it was not possible to test every donor sample by all testing methods. Specification of samples used in individual experiments is provided in the description of each experiment.

After hBMSC isolation, cells from eight selected donors (donors 1, 2, 4, 5, 6, 7, 8, and 9) were seeded for CFU-f 0.5 mil. cells/25 cm² flask and cultivated in growth media. After 14 days in culture, colonies displaying more than 50 cells were counted using crystal violet staining (Merck).

Isolated hBMSCs from six selected donors (donors 4, 5, 6, 7, 8, and 9) were plated in the 12-well plate in triplicate at a density of 5,000 cells/well in a standard growth medium. Cell number was evaluated after 1, 3, 6, 9, and 12 days. Cells were washed with PBS, detached by trypsinization, and then, manually counted using the Bürker–Türk counting chamber.

Primary hBMSCs from two selected donors (donors 8 and 9) from passage 2 were used for analyzing their differentiation capacity as previously described (Tencerova et al., 2019).

hBMSCs were seeded at a density of 20,000 cells/cm². When the cells reached 80% confluence, MEM (Gibco) supplemented with 10% FBS (Gibco) and 1% P/S (Gibco) was added to the control cells, and the rest of the cells were cultivated with osteoblast (OB) induction medium consisting of 10 mM β-glycerophosphate (Merck), 10 nM dexamethasone (Merck), 50 μg/ml vitamin C (Wako), and 10 nM vitamin D3 (Merck). The medium was changed every second day for 7 days (ALP activity) and 11 days (Alizarin Red S staining was provided).

Mineralization of cell matrix at day 11 was measured using Alizarin Red S staining. Cells were fixed with 70% ice-cold ethanol for minimum 1 h at −20°C after which Alizarin Red S solution (Merck) was added. The cells were stained for 10 min at room temperature (RT). Excess dye was washed with distilled water followed by PBS.

ALP activity and cell viability assay were quantified at day 7 of OB differentiation in order to normalize the ALP activity data to the number of viable cells. Cell viability assay was performed using CellTiter-Blue Assay Reagent (Promega) at fluorescence intensity (579_Ex/584_Em). ALP activity was determined by absorbance at 405 nm using p-nitrophenyl phosphate (Sigma-Aldrich) as a substrate.

Cells were plated at a density of 30,000 cells/cm². For adipocyte (AD) differentiation, DMEM (Gibco) was used, containing 10% FBS (Gibco), 1% P/S (Gibco), 100 nM dexamethasone (Merck), 450 μM 3-isobutyl-1-methylxanthine (IBMX) (Merck), 1 µM BRL (Merck), and 3 μg/mL insulin (Merck). The medium was changed every third day for 10 days (followed by Oil Red O and Nile Red staining).

At day 10 of differentiation, cells were rinsed with PBS and fixed in 4% paraformaldehyde (Merck) for 10 min at room temperature. After fixation, cells were briefly rinsed with 3% isopropanol solution (Merck), and lipid droplets were stained with Oil Red O solution (Merck) for 1 h at room temperature.

Nile Red Staining was performed as previously described (Benova et al., 2022). Cells were cultured in polystyrene flat-bottom 96-well tissue culture-treated black microplates (BRANDplates^®, cellGrade™, Brand, DE). Nile Red dye (Merck) working solution was prepared from a stock solution of 1 mg/ml. Cells were washed with PBS (Gibco). Dye was added directly to the cells (5 μg/ml in PBS), incubated for 10 min at room temperature in the dark, and then, washed twice with PBS. The fluorescent signal was measured using a Cytation 3 cell imaging multimode plate reader (BioTek) using an excitation of 485 nm and an emission of 572 nm. The fluorescent signal of Nile Red stain was normalized to cell viability signal measured by Cell Titer-Blue Assay Reagent (Promega) mentioned previously.

The cells obtained from two selected donors (donors 8 and 9) were stained by vital fluorescence stains [NucBlue™ and MitoTracker™ Red CM-H2Xros (Thermo Fisher Scientific, Waltham, MA, United States)]. The stains were added according to the manufacturer’s recommendations directly into the culturing flask and incubated for 30 min. Furthermore, the medium was changed to fresh MEM supplemented with 10% FBS (Gibco) + 1% P/S + 1% GlutaMax + 1% MEM non-essential amino acid solution (NEAA) + 1% sodium pyruvate. Cells were examined by fluorescence microscopy in two steps: 1) screening at ×20 magnification to locate the cells and 2) observation at ×40/×60 magnification for detailed cytomorphological analysis.

Total RNA obtained from two selected donors (donors 8 and 9) was isolated using TRI Reagent (Merck), and RNA concentration was measured using NanoDrop spectrophotometer. cDNA synthesis was performed from 1 µg of total RNA using the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher) according to the manufacturer’s protocol. Quantitative real-time PCR was performed using the LightCycler^® 480 SYBR Green I Master (Roche) with specific primers listed in Supplementary Table S1. RT-qPCR data were normalized to β-actin housekeeping gene expression.

Total RNA obtained from two selected donors (donors 10 and 11) was isolated, and RNA-seq was performed on hBMSCs obtained from heparin- and EDTA-coated tubes. One set of samples was analyzed right after the isolation (non-cultivated hBMSCs), and a second set was analyzed 13 days after in vitro cultivation until cells were confluent (cultivated hBMSCs).

The quality of RNA was measured using the Fragment Analyzer RNA Kit (Agilent Technologies, DNF-471), and samples with RQN value > 7 were used for library preparation. RNA sequencing libraries were prepared from 50 ng of total RNA using QuantSeq 3’mRNA-Seq Library Prep FWD for Illumina (Lexogen, 015.96) with the UMI Second Stranded Synthesis Module for QuantSeq FWD (Lexogen; 081.96) according to the manufacturer’s protocol. Final libraries were pooled together and sequenced on Illumina NextSeq^® 500/550 (High Output Kit v2.5; 20024906) with 86 bp single-end reads. Approximately 20 M reads per sample were obtained. First, the quality of raw reads was checked using FastQC v0.11.9, and the potential contamination was screened using FastQ_Screen v0.11.1 (Wingett and Andrews, 2018). Then, the 6 bp-long UMIs followed by “TATA” spacer were added to reads name using umi_tools extract function (https://genome.cshlp.org/content/early/2017/01/18/gr.209601.116.abstract). The low-quality reads and adaptor sequences were removed using TrimmomaticSE v0.36 with parameters “HEADCROP:4 ILLUMINACLIP: Lexogen_quantseq.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36.” Ribosomal and mitochondrial reads were removed using SortMeRNA v2.1b (Kopylova et al., 2012). BAM files with alignment were created with STAR v2.7.0f (https://pubmed.ncbi.nlm.nih.gov/23104886/) (reference Homo sapiens genome version GRCh38). Reads that were mapped to more than one place in the genome were not processed. The count tables were generated using the script HTseq-count v0.11.4 (https://academic.oup.com/bioinformatics/article/31/2/166/2366196) with the annotation version GRCh38.87 and parameter “-m union.” ENSEMBL-IDs were used as identifiers of transcripts.

The counted data were analyzed using the R package DESeq2 v1.34.0 (Love et al., 2014). Rlog-transformed data were processed by the principal component analysis. Differentially expressed genes (DEGs) were identified by the command “DESeq” with default parameters. ENSEMBL-IDs were converted into gene symbol using the org.Hs.eg.db v3.14.0 database (Carlson, 2019). Genes with baseMean > 20 and adjusted p-value < 0.1 were considered DEGs; log₂FoldChange > 0.6 for upregulated DEGs and log₂FoldChange < 0.6 for downregulated DEGs. Functional Gene Ontology overrepresentation analysis (ORA) was conducted using the web tool WebGestalt 2019 (https://academic.oup.com/nar/article/47/W1/W199/5494758?login=false) using these parameters: organism—H. Sapiens, method of interest—ORA, functional database—Gene Ontology, biological process noRedundant, selected gene ID type—Ensembl gene ID, select reference set—genome, minimum number of genes for a category—five, maximum number of genes for a category—2000, multiple test adjustment—BH, significance level—TOP 10, number of categories expected from set cover—10, number of categories visualized in the report—40, and color in DAG—continuous.

hBMSCs obtained from two randomly selected donors (donors 8 and 11) were immunophenotyped using a panel of cell surface markers (Supplementary Table S2). Adherent cells were released from flasks using 0.05% trypsin/EDTA and incubated with the Fc-human blocking reagent followed by incubation with pre-conjugated antibodies for specific markers according to the manufacturer’s recommendations. The flow cytometry was performed by LSRII (BD Biosciences) and analyzed by FlowJo 10.8.0 analysis software.

Parallel measurement of the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) was performed using the Seahorse XFe24 Analyzer (Seahorse, Agilent). Primary hBMSCs obtained from two selected donors (donors 8 and 9) were seeded in the 24-well Agilent Seahorse XF cell culture microplate in 5-plicates at a density of 20,000 cells/well in the growth medium the day prior to the analysis. The next day, all wells were washed with 1 mL of Dulbecco’s modified Eagle’s medium (D5030, Merck) supplemented with 10 mM glucose, 4 mM glutamine, and 2 mM pyruvate (pH 7.4; 37°C); 500 µL of the same medium was pipetted, and the microplate was incubated at 37°C for 30 min. Meanwhile, the XFe24 sensor cartridge was prepared by injection of substrates according to the protocol to measure metabolic rates with endogenous substrates (basal) and after subsequent additions with final concentrations of 10 mM glucose (Merck), 1 µM oligomycin (Oligo) (Merck), 2 µM carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP) (Merck), and the mixture of inhibitors of 1 µM rotenone (Rot) (Merck), 1 μg/mL of antimycin A (AA) (Merck), and 100 mM 2-deoxyglucose (2DG) (Merck) (2DG+AA+Rot). The Seahorse data were analyzed using Wave Software 2.6.1 (Agilent). The data were normalized by the cell number determined by Hoechst 33342 staining of cell nuclei (final concentration 5 μg/mL) (Thermo Fisher), which was performed immediately after the measurement using the Cytation 3 cell imaging multimode plate reader (BioTek) and processed using Gen5 software (BioTek).

Global lipidomic and metabolomic profiling of human BM plasma samples was conducted using a combined untargeted and targeted workflow for lipidome, metabolome, and exposome analysis (LIMeX) (Janovska et al., 2020; Tsugawa et al., 2020; Sistilli et al., 2021) with some modifications. Extraction was performed using a biphasic solvent system of cold methanol, methyl tert-butyl ether (MTBE), and 10% methanol. Four different liquid chromatography–mass spectrometry (LC–MS) platforms were used for profiling: 1) lipidomics of complex lipids using reversed-phase liquid chromatography with mass spectrometry (RPLC–MS) in the positive ion mode, 2) lipidomics of complex lipids in RPLC–MS in the negative ion mode, 3) metabolomics of polar metabolites using hydrophilic interaction chromatography with mass spectrometry (HILIC–MS) in the positive ion mode, and 4) metabolomics of polar metabolites using RPLC–MS in the negative ion mode. Sample preparation and data processing were performed as described in the previous publication (Benova et al., 2022). The raw metabolomic data with peak height intensities of annotated metabolites are presented in Supplementary Table S3.

The statistical significance of the differences between the experimental groups (heparin vs. EDTA) was determined by paired Student’s t-test using GraphPad Prism 9.3.1 software. Data are presented as means ± SEM. p-value < 0.05 was considered statistically significant. Negative binomial generalized linear model fitting and Wald statistics with the Benjamini–Hochberg test were used for bulk RNA sequencing. Metabolomics and lipidomics data were processed using MetaboAnalyst (Pang et al., 2022). All the statistical details of experiments can be found in the figure legends.

To investigate whether the collection of human BM aspirates in tubes coated with heparin or EDTA anticoagulant affects the cellular properties of isolated hBMSCs, we enrolled 12 random donors (6M/6F) from the hematology clinic. These BM samples were divided into heparin-coated and EDTA-coated blood tubes to obtain paired samples of BM aspirate for further cellular and molecular analyses of isolated hBMSCs. The basic characteristics of enrolled donors are summarized in Table 1.

The paired samples of isolated hBMSCs from heparin- and EDTA-coated tubes obtained from the same donors were analyzed for their basic cellular characteristics (Figure 1A; Table 2). While the yield of hBMSCs after isolation was not different between heparin- vs. EDTA-coated tubes, the yield of hBMSCs after passage 0 (when they reach the confluence in a flask) was higher in the samples obtained from heparin-coated tubes compared to EDTA-coated tubes, with a trend toward significance (p = 0.062) (Figures 1B, C). Furthermore, the number of CFU-f, an in vitro surrogate marker of the stem cell potency of hBMSCs, was higher in all heparin-isolated hBMSCs in comparison to EDTA-isolated hBMSCs (p = 0.017) (Figures 1D–F), which might suggest a presence of more potent stem cell populations in heparin-isolated samples. On the other hand, the short-term proliferation rate of hBMSCs was not affected by the type of anticoagulants used to prepare BM aspirates (Figure 1G). In addition, microscopic analysis of cell morphology in the cultivated hBMSCs did not reveal any significant changes between heparin- and EDTA-isolated hBMSCs (Figure 1H; Supplementary Figure S1A). Moreover, using fluorescent markers for the nucleus (NucBlue ™) and active mitochondria (MitoTracker ™ Red CM-H2Xros) did not show any profound differences between heparin- and EDTA-isolated hBMSCs (Figure 1I).

To further understand the impact of used anticoagulants on the molecular characteristics of isolated hBMSCs, we performed RNA sequencing (RNA-seq) of paired samples of heparin- and EDTA-isolated hBMSCs. In addition, we investigated whether hBMSC expansion in the culture changes their transcriptomic profiles depending on the type of anticoagulants used in the isolation process (Figure 2A).

Principal component analysis (PCA) showed minimal separation between paired hBMSCs from heparin and EDTA samples, which were analyzed directly after isolation (non-cultivated) (Figure 2B). From around 10,000 genes identified in non-cultivated hBMSCs, comparative analysis revealed only 141 significant DEGs (Figure 2C; Supplementary Table S4), from which 94 were downregulated and 47 were upregulated in heparin-isolated BMSCs compared to EDTA-isolated BMSCs. This number of differently expressed genes between non-cultivated heparin-isolated vs. EDTA-isolated hBMSCs suggests no significant impact of the used anticoagulants on the transcriptomic profile of hBMSCs, if measured directly after the isolation process.

Similarly, minimal changes were observed in paired samples of hBMSCs cultivated in growth media after isolation from heparin-coated and EDTA-coated tubes. From the total of 11,407 analyzed genes, we found only 12 upregulated and 21 downregulated genes in cultivated hBMSCs between heparin and EDTA samples (Figure 2D), suggesting that used anticoagulants did not show any significant differences between the transcriptome of cultivated hBMSCs obtained from heparin- and EDTA-coated tubes.

On the other hand, we found that cultivation had a much bigger effect on molecular characteristics of hBMSCs compared to used anticoagulants in the isolation procedure. The PCA plot of all cultivated and non-cultivated hBMSCs (Figure 2B; Supplementary Table S4) showed a clear separation without differences between the types of anticoagulants. The volcano plot revealed 11,560 significant detect genes, where 3,969 were upregulated and 3,459 downregulated in cultivated compared to non-cultivated hBMSCs (adjusted p-value < 0.100 and the absolute value of log2 fold change > 0.6) (Figure 2E; Supplementary Table S4). In order to elucidate the functionality of differentially expressed genes, we performed functional Gene Ontology overrepresentation analysis. Genes that exhibited higher expression with cultivation compared to non-cultivated cells (Figures 2F, G) were annotated in the GEO category for mitochondrial respiratory chain complex assembly and mitochondrial gene expression (TIMMDC1—translocase of inner mitochondrial membrane domain containing 1 and UQCC1—ubiquinol-cytochrome c reductase complex assembly factor 1), generation of precursor metabolites, and energy (GLRX2—glutaredoxin 2 and UGDH—UDP-glucose 6-dehydrogenase). Transcripts involved in the ribonucleotide metabolic process (COX14—cytochrome c oxidase assembly factor COX14 and NDUFB5—NADH:ubiquinone oxidoreductase 5) were also enriched in this subset. The remodeling of the mitochondrial and metabolic infrastructure matches the evolving bioenergetic requirements, as cells have a high energy demand. However, the genes involved in cell proliferation (e.g., CCNA2, CCNB1, and MKI67) (Supplementary Figure S2A) were not differently expressed between non-cultured and cultured cells. The genes that were downregulated with the in vitro cultivation (Figures 2F, G) are related to decreased inhibitory factors on cell cycle (CDKN1B—cyclin-dependent kinase inhibitor 1B and TFDP2—transcription factor Dp-2), chromatin remodeling (KDM7A—lysine demethylase 7A and HDAC4—histone deacetylase 4), and decreased inflammatory properties of the cells (CD6—cluster of differentiation 6; VSIR—V-set immunoregulatory receptor; and CD38—cluster of differentiation 38) suggesting the maintenance of stable expression states of transcription factor genes, the control of stem cell activity, higher proliferation, and active metabolic status of the stem cells.

Because of differences in the BMSC yield and number of CFUs-f of cultivated heparin- and EDTA-isolated hBMSCs, we further analyzed the expression profile of cell surface markers in non-cultivated and cultivated hBMSCs obtained from heparin- and EDTA-coated tubes (Figure 3A). Using the panel of used stem cell markers (ALPL, LepR, CD44, and CD73) (Supplementary Figure S2A), no changes were found between non-cultivated hBMSCs obtained from heparin- and EDTA-coated tubes (Figure 3B; Table 3), and the same was observed in cultivated hBMSCs (Figure 3C; Table 4). The presence of cell surface markers (CD44 and CD73) differed between non-cultivated and cultivated hBMSCs, but they were not affected by the type of anticoagulants used in the isolation process. These data were also confirmed with RNA-seq analysis (Supplementary Figures S3A, B).

Thus, these data suggest that heparin and EDTA anticoagulants did not affect the composition of isolated hBMSCs based on the basic panel of stem cell surface markers, and differences were detected only between non-cultivated and cultivated hBMSCs.

In order to understand whether a type of anticoagulant may affect the differentiation capacity of cultivated hBMSCs, we tested the differentiation potential of heparin and EDTA hBMSCs in the osteogenic (OB) and adipogenic (AD) conditions in vitro (Figure 4A).

As previous findings documented that heparin stimulates osteogenic potential in mouse BMSCs (Ling et al., 2010), we hypothesized that heparin-isolated hBMSCs might have increased OB differentiation capacity in vitro compared to EDTA-isolated cells. However, Alizarin S (AZR) staining (Figure 4B), ALP activity (Figure 4C; Table 5), and the gene expression of OB markers (RUNX2, BGLAP, and COL1A1) in OB-differentiated hBMSCs (Figure 4D) did not show significant changes between heparin- and EDTA-isolated hBMSCs in different donors. In addition to AD differentiation of hBMSCs measured by Nile Red staining (Figures 4E, F), the gene expression of AD genes (PPARγ, CEBPA, ADIPOQ, and FABP4) (Figure 4G) also did not show any changes between heparin- and EDTA-isolated samples. Thus, these in vitro validations of hBMSC differentiation potential suggest no significant difference between samples obtained from heparin- and EDTA-coated tubes.

To further investigate the effects of used anticoagulants on functional characteristics (cellular metabolism) of isolated hBMCSs, we performed bioenergetic profiling of heparin- and EDTA-isolated hBMSCs (from two selected donor samples) using Seahorse technology (Figure 4H). We obtained simultaneous measurements of mitochondrial function via OCR (Supplementary Figure S4A) and glycolysis via ECAR (Supplementary Figure S4B).

Glycolysis measurements did not reveal any significant changes in glycolytic capacity and glycolytic reserve between heparin and EDTA samples (Figure 4I). In addition, the measurements of cellular respiration also did not show any changes in maximal respiration induced by FCCP or spare capacity.

Overall, these findings assume that different types of anticoagulants in the BM collection procedure of hBMSCs do not influence the functional performance and quality of isolated hBMSCs. Thus, these types of samples might be comparable for future multicentric analysis.

In order to investigate the effects of heparin and EDTA on the lipidome and metabolome of BM plasma from the paired samples from the same patient, global lipidomic and metabolomic analyses were performed using a multiplatform liquid chromatography–mass spectrometry-based (LC–MS) approach. Overall, 470 polar metabolites and simple and complex lipids were annotated in these samples. For most annotated polar metabolites and lipids (95%), no differences between heparin and EDTA groups were observed (Figure 5A) except for a group of organic acids (isocitric acid, cis-aconitic acid, malic acid, aspartic acid, and α-ketoglutaric acid) that were increased in EDTA samples (p < 0.001). On the other hand, some organic acids (citric acid, lactic acid, glyceric acid, and α-hydroxyglutaric acid), amino acid derivatives (e.g., N-acetyl forms of alanine, leucine, valine, and phenylalanine), and dipeptides were increased in heparin samples (p < 0.001) (Figure 5B).

Thus, these data showed that the metabolomic profile in BM plasma is minimally affected between heparin- and EDTA-coated paired samples.