Common and different alterations of bone marrow mesenchymal stromal cells in myelodysplastic syndrome and multiple myeloma.

Abstract Objective The objective of this study was to explore characteristics of bone marrow mesenchymal stromal cells (BM‐MSCs) derived from patients with myelodysplastic syndrome (MDS) and multiple myeloma (MM). Methods BM‐MSCs were recovered from 17 of MDS patients, 23 of MM patients and 9 healthy donors and were passaged until proliferation stopped. General characteristics and gene expression profiles of MSCs were analysed. In vitro, ex vivo coculture, immunohistochemistry and knockdown experiments were performed to verify gene expression changes. Results BM‐MSCs failed to culture in 35.0% of patients and 50.0% of recovered BM‐MSCs stopped to proliferate before passage 6. MDS‐ and MM‐MSCs shared characteristics including decreased osteogenesis, increased angiogenesis and senescence‐associated molecular pathways. In vitro and ex vivo experiments showed disease‐specific changes such as neurogenic tendency in MDS‐MSCs and cardiomyogenic tendency in MM‐MSCs. Although the age of normal control was younger than patients and telomere length was shorter in patient's BM‐MSCs, they were not different according to disease category nor degree of proliferation. Specifically, poorly proliferation BM‐MSCs showed CDKN2A overexpression and CXCL12 downregulation. Immunohistochemistry of BM biopsy demonstrated that CDKN2A was intensely accumulation in perivascular BM‐MSCs failed to culture. Interestingly, patient's BM‐MSCs revealed improved proliferation activity after CDKN2A knockdown. Conclusion These results collectively indicate that MDS‐MSCs and MM‐MSCs have common and different alterations at various degrees. Hence, it is necessary to evaluate their alteration status using representative markers such as CDKN2A expression.


| INTRODUC TI ON
Therefore, BM-MSCs are considered as active players in the pathophysiology of haematologic malignancies rather than passive by-standers in haematopoietic microenvironment.
They are even considered, as possible therapeutic targets. 8   To calculate population doubling time (PDT) of each passage, cells were counted using a disposal hemocytometer C-Chip (SystemBükerTürk; Incyto, Cheonan, Korea). Passaging was proceeded at a seeding concentration of 1 × 10 5 cells per dish. The passaging was repeated until BM-MSCs stopped proliferating. PDT was calculated using an algorithm available online (http://www. doubl ing-time.com/compu te.php). We measured cellular fraction in apoptosis and necrosis at P3 using an Apoptosis/Necrosis Detection Kit (ab176749, Abcam, Cambridge, MA, USA) by flow cytometry (FACSCalibur, BD Biosciences) and CellQuest™ Proversion 6.0 software (BD Biosciences) according to manufacturer's instruction.

| Telomere length analysis
We measured telomere length of BM-MSCs according to our previous protocol. 11 Telomere-specific primers and the 36b4 primers were used.
All PCRs were performed on the Rotor-Gene Q real-time instrument (Qiagen). The average telomere length in a cell was calculated as the telomere-to-single copy gene (T/S) ratio using Rotor-Gene Q software 2.0.2.

| In vitro oesteogenic, chondrogenic, adipogenic, neurogenic and cardiomyogenic differentiation
We seeded 1 × 10 5 BM-MSCs at P3 into each well of a 6-well plate (Nunc, Shanghai, China). Culture medium was changed into differentiation medium when cells reached 70% confluency. After culturing for three weeks, mesodermal differentiation was analysed after special staining with the same procedure as described in our previous study. 12  and analysed after staining with 2% Alizarin Red Solution (ScienCell).
In addition, we tried to differentiate BM-MSCs into neural cells and cardiomyocytes according to our previous protocol. 13 The medium was replaced with Neural Induction Medium and supplement (Gibco). After two weeks, cells were stained with anti-SOX2 (Abcam, Cambridge, MA, USA) and anti-Nestin antibody (Abcam) and observed using a conformal system. Cardiomyocyte differentiation medium A (Gibco) was replaced when cells reached 70% confluency. After two days, cardiomyocyte differentiation medium B (Gibco) was applied for two days. Cells were then cultured in cardiomyocyte maintenance medium (Gibco) for two weeks. Differentiation was confirmed using a C2 + confocal system (Nikon, NY, USA) after staining with anti-alpha actinin (Abcam) and anti-cardiac troponin T antibody (Abcam). MSCs from four healthy donors (1 × 10 4 ) were seeded into 100 mm plates and incubated at 37°C in CCM for 1 day. These MSCs were then cultured alone (unprimed) or cocultured (primed) with 2.5 × 10 4 CD34 + HSCs, SKM1, or IM-9 cell lines. Primed and unprimed MSCs were harvested after 10 days and subjected to RNA sequencing.

| Analysis of gene expression changes by RNAsequencing
One gram of total RNA was processed to prepare mRNA sequencing library using a TruSeq stranded mRNA sample preparation kit (Illumina, San Diego, CA, USA) according to the manufacturer's instruction. Products were then purified and enriched with PCR to create the final cDNA library. Sequencing of the prepared library was conducted on a Nextseq system (Illumina) with 75 bp paired-end reads. Mapping of high quality reads on Human reference genome (hg38), gene counting and differential analysis were performed using Strand NGS v.2.9 (Strand Genomics, CA, USA). For normalization of gene counting, DESeq algorithm was applied. 15 To determine DEGs, cut-off values for fold changes were log2 ratio ≥ 1 or ≤ −1, with raw read count ≥ 10. As biological replicates were not used in this study, cut-off for statistical significance was not applied.

| Mutation analysis by next-generation sequencing (NGS)
We analysed and compared genetic mutations in patient's malig- Bound peroxidase was then visualized after reacting with 3,3'diaminobenzidine (DAB substrate). Counterstaining was performed with haematoxylin.

| CDKN2A knockdown (KD) experiment and real-time cell monitoring
Since the CDKN2A might be important in proliferation activity in patient's BM-MSCs, we performed CDKN2A KD experiment and analysed proliferation activity using a real-time cell monitoring system.

| Data analysis
Data are presented as mean ± standard deviation (SD) of at least three independent determinations. Statistical differences between groups were determined using Student's t test or one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc test for multiple comparisons. Pearson's chi-squared test and Pearson's correlation was utilized to analyse the relationship between analytic data.

| Impaired proliferation activity and differentiation potential of patient's BM-MSCs
MSCs were successfully isolated and cultured from all BM samples from healthy donors. However, we did not observe BM-MSC prolif- The donor age was inversely correlated with T/S ratio (R = −0.587, P < .001) and CXCL12 expression (R = −0.493, P = .004) ( Figure S1A).
Although the age of normal control was younger than patients (P < .001), it was not different between disease category nor among MSCs continued to proliferate after P6, stopped to proliferate before P6, and culture failure in patients ( Figure S1B). In addition, T/S ratio was lower in patients than normal control (P = .001), but it was not different according to disease category nor degree of proliferation ( Figure S1C). CXCL12, a gene required for haematopoietic stem-cell maintenance 16 expression was also significantly lower in patients than normal control (P = .004), but it was not different between disease category in patients ( Figure S1D). Then, we categorized disease state according to revised international prognostic system (IPSS-R) and international staging system (ISS) for MDS and MM, respectively.
Patients were grouped as high-risk (IPSS-R high/very high and ISS stage II/III) and low-risk (IPSS-R very low/low/intermediate and ISS stage I). Interestingly, there was significant difference of proliferation activity between two risk groups (P = .015) (Table S1). vs. 2.7% ± 0.9%, P = .036) ( Figure S2). In the aspect of differentiation, osteogenic potential was decreased in patient's BM-MSCs and adipogenic potential was decreased in MM-MSCs. Chondrogenic potential of patients' BM-MSCs was similar to that of normal BM-MSCs ( Figure S3).

| Different gene expression and functional alterations in patient's BM-MSCs
By disease category, genes associated with "cell morphogenesis" and "neuron development" were highly expressed in MDS-MSCs while genes associated with "heart development" and "receptor-

| Gene expression alteration of normal BM-MSC after coculture with MDS and MM cell lines
Ex vivo coculture experiment showed that the gene expression profile of normal BM-MSCs was changed after coculture with MDS or MM cell lines (Table 3). In SKM1-primed BM-MSCs, genes associated with "acute-phase response," "positive regulation of angiogenesis" and "neuromuscular process controlling posture" were upregulated. In IM-9-primed BM-MSC, genes associated with "cell adhesion," "positive regulation of ERK1 and ERK2 cascade," "apoptotic process," "programmed necrotic cell death," "|MAPK cascade," "angiogenesis" and "heart development" were upregulated while genes associated with "ossification," "osteoblast differentiation" and "positive regulation of Wnt signalling pathway" were downregulated.

| Mutations mutually exclusive between patient's BM-MSCs and malignant cells
Patient's malignant cells showed one or two disease-associated mutations (Table 4). However, their BM-MSCs did not harbour the same genetic mutations. We detected an NF1 mutation c.8087C>T

| Correlation of CDKN2A expression and impaired proliferation activity in patient's BM-MSCs
Hierarchical cluster analysis of microarray data indicated that pa-  "cell migration," "regulation of vascular endothelial growth factor production," "regulation of angiogenesis" and "positive regulation of cell adhesion" were downregulated ( Figure 4A). It was noteworthy that these differently gene expressing groups were in line with proliferation activity. CDKN2A and TLR4 were highly expressed in the first group compared with those in the second group (P = .029 and P = .037, respectively). CXCL12 expression was significantly low in the second group (P = .029) ( Figure 4B). Interestingly, the value of CDKN2A expression showed a positive correlation with PDT of BM-MSCs at P4 (P < .001) ( Figure 4C).
Having observed higher CDKN2A expression in BM-MSCs that stopped to proliferate before P6, we postulated that the value of CDKN2A activity. To test this hypothesis, we implemented two of experiment.
First, we determined CDKN2A expression levels in patients whose BM-MSCs were not obtained because of culture failure. CDKN2A IHC was performed for patient's BM biopsy sample, because it was possible to observe the location and shape of immunoreactive cells ( Figure 4D).
CDKN2A was not or faintly stained in patient's BM whose MSCs continued to proliferate after P6. However, it showed slight to moderate cytoplasmic and nuclear staining on perivascular cells in BM whose MSCs stopped to proliferate before P6. Of note, we could observe intense CDKN2A accumulation in perivascular cells of BM that showed culture failure ( Figure S6). The CDKN2A immunoreactive cells were thought to be BM-MSC because they located on same locus as CD271 and/or CD146 immunoreactive cells ( Figure S7). These results together indicated that the degree of impaired proliferation activity of patient's BM. Additional

| D ISCUSS I ON
In this work, we evaluated characteristics of haematopoietic microenvironment in two types of haematologic malignancies and compared their characteristics. We selected MDS and MM to explore the similarity and difference of their characteristics because they represent myeloid and lymphoid neoplasms, respectively. They share similar nature such as late-onset age and disease progression from low-risk to high-risk. Patient's BM-MSCs exhibited a much lower proliferation activity, shorter telomere length and lower CXCL12 expression compared to normal BM-MSCs. When analyses were limited in patients, age and telomere length were not different according to degree of proliferation nor between disease category. We postulated that age was associated with physiological senescence of BM-MSCs including telomere shortening, decreased proliferation activity and haematopoiesis support function rather than premature impairment of proliferation which was seen in patients. [19][20][21][22] Interestingly, CXCL12 expression was low in poor proliferating MSCs, which functions HSC maintenance and regulation including quiescence and the ability to induce multilineage reconstitution. 23 Downregulation of the haematopoiesis associated genes was also observed in BM-MSCs primed by MDS and MM cell lines, which could result in multilineage peripheral blood cytopenia in patients. Further studies will be needed to define the correlation among age and CXCL12 expression with haematopoietic function in patients. [24][25][26] Global gene expression profile by microarray provided significant molecular pathways to explain the impaired proliferation activity in patient's BM-MSCs including upregulation of important genes in ERK1/ERK2 cascade and MAPK cascade. ERK activity is known to be correlated with increased β-galactosidase activity and induction of classical senescence-associated genes. 27 More recently, MAPK signalling pathway has been shown to be involved in the mechanism of BM-MSC apoptosis. 28   Interestingly, changes in the proliferation activity of BM-MSCs were closely associated with differentiation potential. We ob-   53 In plasma cell dyscrasias, disease evolution and progression is independently affected by the biology of the surrounding BM niche. 54 Numerous observations indicate that an altered BM microenvironment provides a nurturing niche that sustains haematologic malignancy and might even contribute to F I G U R E 5 Real-time monitoring of proliferation after CDKN2A knockdown (red line) using the xCELLigence assay compared with untreated (NOT, green), lipofectamine treated (Lipo, light blue) and negative control (NC)-siRNA transfected (blue) bone marrow mesenchymal stromal cells (BM-MSC) from myelodysplastic syndrome (MDS) and multiple myeloma (MM) for 72 hrs. Representative data by three technical replicates (A, B) and the mean and SE from the six independent donors (C, D). *P < .05, ** P < .01 the emergence and evolution of malignant clones. 55 Specifically, we found that CDKN2A expression was significantly increased in poorly proliferating MSCs and correlated with PDT. These findings led us to perform CDKN2A IHC on BM samples whose MSCs were not cultured. IHC showed that CDKN2A immunoreactivity was markedly increased in those samples. CDKN2A was induced by cellular stress through CDKN2A-RB pathway. If stress persisted, sustained activation of CDKN2A became engaged upon irreversible senescence. 56,57 It is noteworthy that patient's BM-MSCs revealed improved proliferation activity after CDKN2A KD. Downregulation of p16 INK4a using siRNA targeting CDKN2A increased cell division during clonal expansion 58 and neural stem cell self-renewal depended on repression of CDKN2A. 59 Taken together, we postulated that CDKN2A is a good candidate of therapeutic target to control BM microenvironment because early senescence is reversible through de-induction of CDKN2A. 60

| CON CLUS IONS
Our results highlighted that BM-MSCs from MDS and MM pa- Further studies are required to specify the functional and/or molecular alteration that might be therapeutic targets as well as recovery indicators.

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no competing interests.

AUTH O R CO NTR I B UTI O N S
YK and MK involved in conceptualization and design; S-SP, Y-JK and C-KM provided patient data and samples; HC, JK, AK and DK involved in experiments, collection and assembly of data; HC, JMK, MK and YK involved in data analysis and interpretation; HC, YK and MK edited and wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.