Characterization of bone marrow derived mesenchymal stem cells in suspension

Introduction Bone marrow mesenchymal stem cells (BMMSCs) are a heterogeneous population of postnatal precursor cells with the capacity of adhering to culture dishes generating colony-forming unit-fibroblasts (CFU-F). Here we identify a new subset of BMMSCs that fail to adhere to plastic culture dishes and remain in culture suspension (S-BMMSCs). Methods To catch S-BMMSCs, we used BMMSCs-produced extracellular cell matrix (ECM)-coated dishes. Isolated S-BMMSCs were analyzed by in vitro stem cell analysis approaches, including flow cytometry, inductive multiple differentiation, western blot and in vivo implantation to assess the bone regeneration ability of S-BMMSCs. Furthermore, we performed systemic S-BMMSCs transplantation to treat systemic lupus erythematosus (SLE)-like MRL/lpr mice. Results S-BMMSCs are capable of adhering to ECM-coated dishes and showing mesenchymal stem cell characteristics with distinction from hematopoietic cells as evidenced by co-expression of CD73 or Oct-4 with CD34, forming a single colony cluster on ECM, and failure to differentiate into hematopoietic cell lineage. Moreover, we found that culture-expanded S-BMMSCs exhibited significantly increased immunomodulatory capacities in vitro and an efficacious treatment for SLE-like MRL/lpr mice by rebalancing regulatory T cells (Tregs) and T helper 17 cells (Th17) through high NO production. Conclusions These data suggest that it is feasible to improve immunotherapy by identifying a new subset BMMSCs.

BMMSCs are thought to be derived from the bone marrow stromal compartment, initially appearing as adherent, single colony clusters (colony-forming unitfibroblasts [CFU-F]), and subsequently proliferating on culture dishes [25]. To date, the CFU-F assay has been considered one of the gold standards for determining the incidence of clonogenic BMMSC [26,27]. Since BMMSC are a heterogeneous population of stem cells, it is critical to identify whether BMMSC contain unique cell subsets with distinctive functions, analogous to the

Isolation of mouse bone marrow mesenchymal stem cells (BMMSCs)
The single suspension of bone marrow derived all nucleated cells (ANCs) from femurs and tibias were seeded at a density of 15 × 10 6 into 100 mm culture dishes (Corning, NY, USA) at 37°C and 5% CO2. Nonadherent cells were removed after two days and attached cells were maintained for 16 days in alpha minimum essential medium (α-MEM, Invitrogen, Grand Island, NY, USA) supplemented with 20% fetal bovine serum (FBS, Equitech-bio, Kerrville, TX, USA), 2 mM L-glutamine, 55 μM 2-mercaptoethanol, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen). Colony-forming attached cells were passed once for further experimental use.

Preparation of Extracellular Matrix (ECM) coated dishes
ECM coated dishes were prepared as described previously [28]. Briefly, 100% confluence of BMMSCs was cultured in medium with 100 nM L-ascorbic acid phosphate (Wako Pure Chemical, Richmond, VA, USA). After two weeks, cultures were washed with PBS and incubated with 0.005% Triton X-100 (Sigma) for 15 minutes at room temperature to remove cells. The ECM was treated with DNase I (100 units/ml; Sigma) for 1 hour at 37°C. The ECM was washed with PBS three times and stored in 2 ml of PBS containing 100 U/ml penicillin, 100 μg/ml streptomycin and 0.25 μg/ml fungizone (Invitrogen) at 4°C.

Isolation of BMMSCs in culture suspension (S-BMMSCs)
Bone marrow-derived ANCs (15 × 10 6 ) were seeded into 100 mm culture dishes and cultured for two days. The culture supernatant with floating cells was collected and centrifuged to obtain putative non-attached BMMSCs. The cells were re-seeded at indicated numbers on ECMcoated dishes. After 2 days, the floating cells in the cultures were removed with PBS and the attached cells on ECM were maintained for an additional 14 days. Colony-forming attached cells were passed once and subcultured on regular plastic culture dishes for further experiments. For some stem cell characterization analyses, we collected SSEA4 positive S-BMMSCs using the MACS magnetic separation system (Milteny Biotech, Auburn, CA, USA) and expanded in the cultures.

Colony forming unit-fibroblastic (CFU-F) assay
One million cells of ANCs from bone marrow were seeded on a T-25 cell culture flask (Nunc, Rochester, NY, USA). After 16 days, the cultures were washed with PBS and stained with 1% toluidine blue solution in 2% paraformaldehyde (PFA). A cell cluster that had more than 50 cells was counted as a colony under microscopy. The colony number was counted in five independent samples per each experimental group.

Cell proliferation assay
The proliferation of BMMSCs and S-BMMSCs was performed using the bromodeoxyuridine (BrdU) incorporation assay. Each cell population (1 × 10 4 cells/well) was seeded on two-well chamber slides (Nunc) and cultured for two to three days. The cultures were incubated with BrdU solution (1:100) (Invitrogen) for 20 hours, and stained with a BrdU staining kit (Invitrogen). BrdU-positive and total cell numbers were counted in ten images per subject. The BrdU assay was repeated in five independent samples for each experimental group.

Population doubling assay
A total of 0.5 × 10 6 cells of BMMSCs and S-BMMSCs was seeded on 60 mm culture dishes at the first passage. Upon reaching confluence, the cells were passaged at the same cell density. The population doubling was calculated at every passage according to the equation: log 2 (number of harvested cells/number of seeded cells). The finite population doublings were determined by cumulative addition of total numbers generated from each passage until the cells ceased dividing.

In vivo bone formation assay
A total of 4.0 × 10 6 cells was mixed with hydroxyapatite/tricalcium phosphate (HA/TCP) ceramic powders (40 mg, Zimmer Inc., Warsaw, IN, USA) and subcutaneously transplanted into eight-week-old immunocompromised mice. After eight weeks, the transplants were harvested, fixed in 4% PFA and then decalcified with 5% ethylenediaminetetraacetic acid (EDTA; pH 7.4), followed by paraffin embedding. The paraffin sections were stained with H & E and analyzed by an NIH Image-J. The newly-formed mineralized tissue area from five fields was calculated and shown as a percentage to total tissue area.

In vitro osteogenic differentiation assay
BMMSCs and S-BMMSCs were cultured under osteogenic culture conditions containing 2 mM β-glycerophosphate (Sigma), 100 μM L-ascorbic acid 2-phosphate and 10 nM dexamethasone (Sigma). After induction, the cultures were stained with alizarin red or alkaline phosphatase.

In vitro adipogenic differentiation assay
For adipogenic induction, 500 nM isobutylmethylxanthine, 60 μM indomethacin, 500 nM hydrocortisone, 10 μg/ml insulin (Sigma), 100 nM L-ascorbic acid phosphate were added to the culture medium. After 10 days, the cultured cells were stained with Oil Red-O and positive cells were quantified by using an NIH Image-J. Total RNA was also isolated from cultures after 10 days induction for further experiments.

Reverse transcriptase polymerase chain reaction (RT-PCR) analysis
Extraction of total RNA and RT-PCR were performed according to standard procedures. Primer information is described in Additional materials and methods [see Additional file 1].

Western blotting analysis
A total of 20 µg of protein was used and SDS-PAGE and western blotting were performed according to standard procedures. Detailed procedures are described in Additional materials and methods [see Additional file 1]. β-actin on the same membrane served as the loading control.

Hematopoietic differentiation of BMMSCs and S-BMMSCs
BMMSCs and S-BMMSCs were cultured onto 35 mm low attach culture dishes (2 × 10 4 /dish, STEMCELL Technologies, Vancouver, BC, V5Z 1B3, Canada) under hematopoietic differentiation medium (STEMCELL Technologies) with or without erythropoietin (EPO; 3 U/mL) for seven days. Whole bone marrow cells and linage negative bone marrow cells (Linage-cells) were used as positive controls. The results are representative of five independent experiments.

Measurement of nitric oxide production
BMMSCs (0.2 × 10 6 /well) were cultured on 24-well plates with or without cytokines (IFNγ, 25 ng/ml; IL-1β, 5 ng/ml, R&D Systems, Minneapolis, MN, USA) and chemicals (L-NMMA, 1 mM; 1400 W, 0.2 mM) at the indicated concentration and days. The supernatant from each culture was collected and nitric oxide concentration measured using a Total Nitric Oxide and Nitrate/Nitrite Parameter Assay kit (R&D Systems) according to the manufacturer's instruction.

Allogenic mouse S-BMMSC transplantation into MRL/lpr mice
Under general anesthesia, C3H/HeJ-derived BMMSCs, S-BMMSCs, L-NMMA pre-treated BMMSCs (1 mM for five days), or CD34 + /CD73 + double sorted cells (0.1 × 10 6 cells/10 g body weight) were infused into MRL/lpr mice via the tail vein at 10 weeks of age (n = 6 each group). In the control group, MRL/lpr mice received PBS (n = 5). All mice were sacrificed at two weeks post transplantation for further analysis. The protein concentration in urine was measured using a Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA).
Measurement of autoantibodies, albumin, soluble runtrelated NF-B ligand (sRANKL) and C-terminal telopeptides of type I collagen (CTX) Peripheral blood serum samples were collected from mice. Autoantibodies, sRANKL and CTX were analyzed by ELISA using commercially available kits (anti-dsDNA antibodies and ANA; alpha diagnostics, albumin and sRANKL; R&D Systems, CTX; Nordic Bioscience Diagnostics, Herlev, Rigion Hovedstaden, Denmark) according to their manufactures' instructions. The results were averaged in each group. The intra-group differences were calculated between the mean values.

Statistics
Student's t-test was used to analyze statistical difference. P values less than 0.05 were considered significant.

Results
A subset of BMMSCs lacks the ability to adhere to plastic culture dishes (S-BMMSCs) but attaches to extracellular cell matrix (ECM)-coated culture dishes To determine whether a subset of BMMSCs remain in culture suspension, ANCs (15 × 10 6 cells) from bone marrow were plated onto regular plastic culture dishes for two days and all non-attached cells were subsequently transplanted into immunocompromised mice subcutaneously using HA/TCP as a carrier. At eight weeks post-transplantation, newly formed bone was identified in the transplants by H & E staining ( Figure 1A), suggesting that the BMMSC culture suspension may contain cells with a capacity to differentiate into bone forming cells. In vitro studies indicated that ECM produced by culture-expanded BMMSCs (BMMSC-ECM) could capture higher numbers of CFU-Fs when compared to plastic cultures [see Additional file 1, Figure S1] [28]. Thus, we collected culture supernatant with floating cells at two days post CFU-F culture and re-loaded it onto BMMSC-ECM-coated dishes ( Figure 1B). A subset of BMMSCs in the suspension (S-BMMSCs) was able to adhere to the BMMSC-ECM and form CFU-F ( Figure 1B), at a lower incidence compared to the number of CFU-F generated from regular BMMSCs ( Figure 1C). In order to characterize the stem cell properties of S-BMMSCs, we collected SSEA4-positive S-BMMSCs and assessed their proliferation rate by BrdU incorporation. We found that S-BMMSCs had a significantly elevated BrdU uptake rate compared to regular BMMSCs ( Figure 1D). In addition, we used a continuous cell culture assay to indicate that SSEA4-positive S-BMMSCs acquired a significantly increased number of population doublings ( Figure 1E). These data imply that S-BMMSCs are distinct from regular BMMSCs in terms of attachment, proliferation, and self-renewal.

C D E
Expansion in regular dish To examine the multipotent differentiation potential, we showed that S-BMMSCs are analogous to BMMSCs in their expression of alkaline phosphatase (ALP), mineralized nodule accumulation under the osteogenic inductive cultures, and bone regeneration when transplanted into immunocompromised mice using HA/TCP as a carrier (Figures 2A and 2B). Furthermore, we showed that S-BMMSCs were similar to regular BMMSCs in forming Oil red-O positive fat cells under adipogenic inductive conditions, which was associated with expression of the adipogenic genes, peroxisome proliferator-activated receptor gamma 2 (pparγ2) and lipoprotein lipase (lpl) (Figures 2C and 2D). Parallel studies showed a similar capacity between S-BMMSCs and regular BMMSCs to differentiate into chondrocytes under chondrogenic inductive conditions, associated with the expression of proteoglycan, trichrome positive collagen, and type II collagen ( Figure 2E). Collectively, these data confirm that S-BMMSCs are a subset of BMMSCs.
S-BMMSCs express CD34, but are distinct from hematopoietic stem cells By flow cytometric analysis, S-BMMSCs expressed mesenchymal stem cell markers at the same level as regular BMMSCs ( Figure 3A). Interestingly, 23.4% of S-BMMSCs expressed CD34, a hematopoietic stem cell (HSC) and endothelial cell marker, whereas 0.2% of BMMSCs expressed CD34 ( Figure 3A). BMMSCs (21.4%) and S-BMMSCs (31.2%) expressed CD45, another hematopoietic marker, at passage 2 ( Figure 3A). Both BMMSCs and S-BMMSCs were negative to CD11b antibody staining (data not shown), excluding the possibility that S-BMMSCs are derived from monocyte/macrophage lineage cells. Importantly, CD34 + S-BMMSCs co-expressed BMMSC-associated markers CD73 or Octamer-4 (Oct4), as evidenced by flow cytometric analysis ( Figure 3B). Western blot analysis confirmed that S-BMMSCs expressed CD34, CD73, and CD105 ( Figure 3C), and regular BMMSCs expressed CD73 and CD105 but lacked CD34 expression ( Figure 3C). Whole bone marrow cells (BMC) were used as positive control. S-BMMSCs also showed a continued expression of CD34 from passage one to five; however, the expression levels appear reduced after passage three ( Figure 3D). In order to further verify CD34 expression in S-BMMSCs, immunocytostaining analyses were performed to show co-expression of CD34 with mesenchymal markers CD73 ( Figure 3E) in contrast to regular BMMSCs that were negative for anti-CD34 antibody staining ( Figure 3E).
It is generally believed that CD34 expression is associated with HSCs and endothelial populations. HSCs can differentiate into all the blood cell lineages and rescue lethally irradiated subjects. Thus, we cultured S-BMMSCs and regular BMMSCs in hematopoietic differentiation medium and determined that these mesenchymal cells failed to differentiate into a hematopoietic cell lineage compare to bone marrow cells that formed myeloid and erythroid colony forming clusters ( Figure 3F). In addition, CD45 -CD34 -BMMSCs showed an ability similar to that of S-BMMSCs in colony forming and expressing surface marker as MSC [see Additional file 1, Figure S2]. Furthermore, we infused S-BMMSCs systemically to rescue lethally irradiated  mice and found that S-BMMSCs, but not regular BMMSCs, could extend the lifespan of lethally irradiated mice [see Additional file 1, Figure S3]. However, S-BMMSCs failed to rescue lethally irradiated mice, as shown in the whole bone marrow cell group [see Additional file 1, Figure S3]. These data provid further evidence that CD34 expression in S-BMMSCs is not due to HSC contamination.

S-BMMSCs transplantation ameliorates multiple organ dysfunctions in MRL/lpr mice
Since the immunomodulation property of MSCs is one of the essential factors for MSC characterization, allogenic S-BMMSC transplantation into MRL/lpr mice was performed ( Figure 4A). Two weeks after transplantation, both S-BMMSCs and BMMSCs were capable of ameliorating SLE-induced glomerular basal membrane disorder (yellow arrow, Figure 4B) and reducing the urine protein level ( Figure 4C). It appeared that S-BMMSCs were superior compared to BMMSCs in terms of reducing the overall urine protein levels ( Figure 4C). As expected, MRL/lpr mice showed remarkably increased levels of autoantibodies, including anti-double strand DNA (dsDNA) IgG and IgM antibodies (Figures 4D and 4E) and anti-nuclear antibody (ANA; Figure 4F) in the peripheral blood serum. Although S-BMMSC and BMMSC infusion showed significantly decreased serum levels of anti-dsDNA IgG, IgM antibodies and ANA in peripheral blood ( Figures 4D-F), S-BMMSCs showed a superior therapeutic effect in reducing anti-dsDNA IgG antibody and ANA levels when compared to BMMSCs ( Figures  4D and 4F). Additionally, decreased serum albumin levels in MRL/lpr mice were recovered by S-BMMSC and BMMSC infusion ( Figure 4G) but S-BMMSC treatment resulted in a more significant recovery than BMMSC treatment ( Figure 4G). Next, flow cytometric analysis revealed that S-BMMSC showed more effectiveness in recovering the decreased level of CD4 + CD25 + Foxp3 + Tregs and increased the number of CD4 + IL17 + IFNγ -Tlymphocytes (Th17 cells) in peripheral blood when compared to BMMSCs (Figures 4H, 4I). In addition, highly passaged mouse S-BMMSCs failed to inhibit Th17 differentiation in vitro (data not shown) suggesting that mouse S-BMMSCs probably lose their immunomodulation property under long culture expansion. Furthermore, we showed that S-BMMSCs were superior to BMMSCs in terms of reducing increased numbers of tartrate-resistant acid phosphatase (TRAP) positive osteoclasts in the distal femur epiphysis of MRL/lpr mice [see Additional file 1, Figure S4A], elevated serum levels of sRANKL, a critical factor for osteoclastogenesis [see Additional file 1, Figure S4B] and bone resorption marker CTX [see Additional file 1, Figure S4C]. These data suggest that S-BMMSCs exhibit a superior therapeutic effect for SLE disorders compared to regular BMMSCs.

S-BMMSCs possess superior immunomodulatory functions via high nitric oxide (NO) production
Recently, immunomodulatory properties were identified as an important stem cell characteristic of BMMSCs, leading to the utilization of systemic infused BMMSCs to treat a variety of immune diseases [19][20][21]. Here, we found that S-BMMSCs exhibited a significantly increased capacity for NO production compared to regular BMMSCs when treated with IFNγ and IL-1β ( Figure 5A). It is known that NO plays a critical role in BMMSCmediated immunosuppression [see Additional file 1, Figures S5A-F] [29]. Therefore, we assessed the functional role of high NO production in S-BMMSC-associated immunomodulatory properties. Spleen (SP) cells were activated by anti-CD3 and anti-CD28 antibodies for three days and then co-cultured with S-BMMSCs or regular BMMSCs in the presence of the general NOS inhibitor, L-NMMA or the iNOS inhibitor, 1400 W, using a Transwell culture system. The efficacy of L-NMMA and 1400 W to inhibit NO production in BMMSCs was verified [see Additional file 1, Figures S6A and 6B]. Although both S-BMMSCs and regular BMMSCs were capable of inhibiting cell viability of activated SP cells, S-BMMSCs showed a marked inhibition of SP cell viability over that of regular BMMSCs ( Figure 5B). Moreover, both BMMSCs and S-BMMSCs induced SP cell apoptosis (Figure 5C). However, S-BMMSCs showed an elevated capacity in inducing activated SP cell apoptosis compared to regular BMMSCs ( Figure 5C). Interestingly, when L-NMMA and 1400 W were added to the cultures, the number of apoptotic SP cells was significantly reduced in both S-BMMSC and regular BMMSC groups ( Figure 5D and 5E). These in vitro experimental data suggested that NO production is an essential factor for BMMSCmediated immunomodulation.
Since up-regulation of CD4 + CD25 + Foxp3 + Tregs is required for immunotolerance [30], we tested Tregs upregulation property of S-BMMSCs and BMMSCs in an in vitro co-culture system. When naïve -T-cells were cocultured with S-BMMSCs or regular BMMSCs in the presence of IL-2 and TGF-β1, S-BMMSCs showed a significant up-regulation of Treg levels compared to regular BMMSCs ( Figure 5F). Both L-NMMA and 1400 W were able to inhibit BMMSC-and S-BMMSC-induced up-regulation of Tregs, as shown by flow cytometric analysis (Figures 5G and 5H). Interestingly, the regulation effect on Tregs was more significant in the S-BMMSC group compared to the BMMSC group ( Figure 5G and 5H). Moreover, both BMMSCs and S-BMMSCs could inhibit differentiation of Th17 in vitro, with a more prominent effect observed with S-BMMSC ( Figure 5I). These inhibitions of Th17 differentiation were abolished by L-NMMA ( Figure 5J) and 1400 W ( Figure 5K). These data further verified the functional role of NO in S-BMMSC-induced immunomodulatory effect.
In order to identify whether there are functional endogenous S-BMMSCs, we used fluorescence activated cell sorting (FACS) to isolate CD34 and CD73 double-positive cells from bone marrow ANCs which resulted in the recovery of 3.77% double-positive cells [see Additional file 1, Figure S7A]. These CD34 and CD73 double-positive cells exhibited mesenchymal stem cell characteristics, including the capacity to form single colony clusters of fibroblast-like cells [see Additional file 1, Figure S7B], which could differentiate into osteogenic cells in vitro [see Additional file 1, Figure S7C]. These data indicated the feasibility of this approach to isolate S-BMMSC-like cells directly from bone marrow. We found that CD34 + /CD73 + BMMSCs were analogous to S-BMMSCs in terms of having higher levels of NO production when compared to regular BMMSCs [see Additional file 1, Figure S7D] and reducing levels of urine *** *** *** *** *** protein, serum anti-dsDNA IgG and IgM antibodies in MRL/lpr mice (data not shown). These data indicate that endogenous S-BMMSCs could be isolated from bone marrow using CD34 and CD73 antibodies double sorting.
Additionally, we used the same BMMSC-ECM isolation approach to reveal the existence of human S-BMMSCs (hS-BMMSC) that possess stem cell properties including multipotent differentiation and self-renewal but lack expression of CD34 (data not shown). hS-BMMSCs showed elevated NO and kynurenine production which indicate high indoleamine 2,3-dioxygenase (IDO) activity when compared to regular BMMSCs [see Additional file 1, Figures S8A-C]. Thus, when activated T cells were co-cultured with hS-BMMSCs, AnnexinV-7 aminoactinomycinD (7AAD) double positive apoptotic SP cells were significantly elevated compared to BMMSCs [see Additional file 1, Figure S8D]. Discussion Adherent BMMSCs are able to proliferate and undergo osteogenic differentiation, providing the first evidence of CFU-F as precursors for osteoblastic lineage [25]. For over a few decades, the adherent CFU-F assay has been used as an effective approach to identify and select BMMSCs. In the current study, we showed that the adherent CFU-F assay collects the majority of clonogenic BMMSCs, but a subpopulation of BMMSCs is sustained in the culture suspension. This newly identified subpopulation of BMMSCs may be lost in the standard CFU-F assay for BMMSC isolation.
In this study, we have identified a novel subset of S-BMMSCs that failed to form adherent CFU-F in regular culture dishes, but were capable of adhering on mesenchymal stem cell-produced ECM and differentiating into osteoblasts, adipocytes and chondrocytes from both C3H/HeJ and C57BL/6J mice. S-BMMSCs coexpressed the HSC marker CD34 with the MSC markers CD73 and Oct4, excluding the potential of HSC contamination. Furthermore, S-BMMSCs were found to be distinct from HSC because they lacked the ability to differentiate into hematopoietic cell lineages in vitro and failed to rescue lethally-irradiated mice. The mechanism that may contribute to the up-regulated immunomodulatory function was associated with high NO production in S-BMMSCs and a NO-driven high Tregs level [36]. NO is a gaseous biological mediator with important roles in affecting T cell function [37]. This is the reason that S-BMMSCs showed a superior therapeutic effect in treating SLE mice.
One successful approach is to isolate cells that express specific molecules on their cell surfaces using monoclonal antibodies and cell sorting technologies. Enriched populations of BMMSCs have been isolated from human bone marrow aspirates using a STRO-1 monoclonal antibody in conjunction with antibodies against VCAM-1/CD106 [32], CD146 [11], low affinity nerve growth factor receptor/CD271, PDGR-R, EGF-R and IGF-1-R [38], fibroblast cell marker/D7-Fib [39] and integrin alpha 1/CD49a [40]. A more recent study has also identified molecules co-expressed by a CD271 + mesenchymal stem cell population including platelet derived growth factor receptor-β (CD140b), human epidermal growth factor 2/ErbB2 (CD340) and frizzled-9 (CD349) [41]. Further cell separation based upon multiparameter FACS identified a population of proposed mouse mesenchymal precursors with the composite phenotype Lin -CD45 -CD31 -Sca-1 + [42]. Another recent study also identified and characterized an alternate population of primitive mesenchymal cells derived from adult mouse bone marrow, based upon their expression of the SSEA-1 [43]. All approaches used for BMMSC purification and isolation will undergo ex vivo expansion to enrich cell numbers for tissue regeneration or systemic therapies by plastic adherent assay. In addition to identifying a novel sub-population of BMMSCs that possess enhanced immunomodulatory properties when compared to regular BMMSCs, we showed that CD34 + /CD73 + BMMSCs could be isolated directly from whole bone marrow and that CD34 + /CD73 + BMMSCs are endogenous S-BMMSCs with higher NO production, and are superior in treating SLE-like mice when compared to regular BMMSCs.
Recently, non-adherent bone marrow cells (NA-BMCs) were identified [44,45]. The NA-BMSCs could be expanded in suspension and gave rise to multiple mesenchymal phenotypes, including osteoblasts, chondrocytes, and adipocytes in vitro, suggesting the presence of non-adherent BMMSCs in primary CFU-F cultures [45]. Although it has been reported that the NA-BMCs can rescue lethally-irradiated mouse recipients, our data indicated that S-BMMSCs only showed improved survival lifespan without a complete rescue of lethally-irradiated mice, compared to whole bone marrow transplantation. While the mechanism of S-BMMSC-mediated lifespan extension in lethally-irradiated mice is unknown, it is possible that S-BMMSCs have a more active interplay with hematopoietic cells than regular BMMSCs. It has been reported that granulocyte colony stimulating factor might promote BMMSCs into the circulation in humans [46], suggesting that non-attached BMMSCs may exist in vivo for specific functional needs. Added evidence indicated that osteocalcin-positive cells in circulation were able to differentiate into osteoblastic cells when cultured in the presence of TGFβ [47]. However, it is unknown whether S-BMMSCs are associated with circulating mesenchymal stem cells initially identified in mice, and this is very rare in humans.

Conclusions
A new subset of BMMSCs (S-BMMSCs) which failed to adhere to culture dishes possesses similar stem cell properties as those seen in BMMSCs, including CFU-F, stem cell markers, osto-, adipo-, and chondro-genic differentiation. However, S-BMMSC showed distinct features including expression of CD34 and a superior immunomodulation property through high NO production. These findings suggest that it is feasible to improve immunotherapy by identifying new subset BMMSCs.

Additional material
Additional file 1: Figures S1 to S8 and Additional materials and methods. Figure S1. ECM coated dish could capture a greater number of CFU-F. CFU-f number in ECM coated dish compared to regular dish. Figure