Adult stem cells (ASCs), which possess the ability to self-renew and regenerate tissue, are of significant value for the development of cellular therapies, tissue engineering tools, and drug screening models. Conventional protocols for ASC enrichment generate a small number of cells that do not represent the total ASC population of tissues. We avoided these conventional methodologies and used a different approach to identify early lineage adult (ELA) stem cells, a subpopulation of ASCs 4-6 μm in diameter, in the synovial fluid of osteoarthritic patients. These cells lack cell surface markers expressed by other ASC (e.g. CD34, CD73, CD105, SSEA-1, CXCR4). However, RT-PCR studies demonstrated expression of pluripotency genes such as NANOG, OCT4, REX1, KLF4, STELLA, and SOX. When cultured in adipogenic, chondrogenic, or osteogenic differentiation media, ELA cells differentiated into fat, cartilage, and bone tissue, respectively. Quantitative PCR analysis revealed unique molecular signatures consisting of tissue-specific and non-tissue specific mRNA in the differentiated tissues, suggesting a continuum of mRNA expression. Furthermore, the ELA cell population shared unique gene sets with embryonic stem cells, mesenchymal stem cells, and induced pluripotent stem cells. Some of these genes are unique to neuronal, cardiac, pancreatic, and hepatic progenitor cells, while others, such as mucins, ICAM, and tetraspans, have tissue-specific cell functions. ELA cells also demonstrated strong in vitro immunomodulatory properties by inhibiting T cell proliferation, inducing CD4+/ CD25+ regulatory T cells, and inhibiting natural killer cell activity. Collectively, these observations suggest that ELA cells might be useful for cell-based regenerative therapies and the treatment of systemic diseases with immunological etiologies.
Keywords: Osteoarthritis; Synovial fluid; Bone marrow; Adult Stem Cells (ASCs); Self-Renewal; Heterogeneity; Multipotent,
Differentiation, Transcriptome, Early Iineage Adult (ELA) Stem Cells; Molecular Signatures
Stem cells have the remarkable capacity to self-renew, differentiate into multiple cell lineages, and reconstitute tissue in vivo . Embryonic stem cells (ESCs), a pluripotent cell type, are established from early embryonic cells and possess the ability to differentiate into all three germ layers [2-5]. In contrast, adult stem cells (ASCs) are found in the developing fetus during tissue renewal and postnatally [6,7]. Hematopoietic stem cells (HSCs), one of the most characterized types of ASCs, have been studied for over 50 years and are known progenitors of various blood cell types [8-10]. HSCs have been used clinically to reconstitute bone marrow (BM) cells destroyed by BM ablation therapy for cancer [11,12]. There is also a heterogeneous population of non-hematopoietic stem cells in the BM. In particular, mesenchymal stem/progenitor cells (MSCs) are also thought to originate from the BM and comprise 0.01-0.001% of nucleated BM cells . MSCs are found in the peripheral blood, umbilical cord blood, adipose tissue, skeletal muscle, liver, lungs, synovium, dental pulp, apical papilla, amniotic fluid, and fetal blood . Because MSCs are found in extremely low numbers in the BM, sustained ex vivo culture on tissue culture plastic is required to generate sufficient cell numbers for phenotypic characterization . MSCs most commonly express surface markers such as CD29, CD44, CD49a-f, CD51, CD73, CD105, CD106, CD166, and Stro1 and lack expression of hematopoietic lineage markers such as CD11b, CD14, and CD45 . MSCs are multipotent ASCs capable of differentiating into various mesodermal tissues, such as adipose, cartilage, and bone . Other groups have reported that MSCs are capable of differentiating into ectodermal and endodermal tissues, such as lung, skin, pancreas, and liver tissue . It has been hypothesized that MSCs may not directly repair tissue but instead secrete trophic factors that decrease cell death, recruit immune cells to the injury site, and promote healing [18,19]. Presently, MSC immunomodulatory properties are being assessed in clinical trials to determine their efficacy in treating a variety of immune-related diseases .
Most ASC studies focus on BM-derived stem cells and use discontinuous density gradients, such as Ficoll-Paque and Lymphoprep, and plastic adherence to enrich for ASCs . Although density gradients effectively separate debris, platelets, and red blood cells (RBCs) from the mononuclear cells in the buffy layer, they also inadvertently discard a subset of ASCs . To avoid potential discrepancies in the isolation of ASCs, we decided to forgo the use of discontinuous gradients and prolonged culture on tissue culture plastic to harvest ASCs. We chose synovial fluid (SF) as a source due to low RBC contamination. To isolate ASCs, we used time sedimentation of diluted SF. Cells in the enriched ASC population measured 4-6μm in diameter (mean = 5μm). Flow cytometry and gene expression analysis suggested that this ASC population expresses genes and proteins generally thought to be restricted to ESCs. In addition, these cells did not express MHC class II, CD44, CD45 or CD49, but unlike our prior studies they were smaller and had minimal MHC class I expression. Semi-quantitative PCR studies showed expression of embryonic transcription factors such as OCT4, REX1, NANOG and SOX2, suggesting pluripotency. We therefore named these cells early lineage adult (ELA) stem cells, since they shared the gene expression profiles of the cells described in earlier studies.
In this study, we investigated the ability of ELA cells to self-renew and differentiate into multiple lineages. Moreover, we optimized the isolation, culture, and expansion conditions for these cells in vitro. We show that ELA cells can differentiate into adipose, cartilage, and bone lineages, and that they also express genes from other cell types. In addition, we determined that ELA cells are potent modulators of the immune response, potentially by inhibiting T cell proliferation, inducing regulatory CD4+/CD25+ T cells, and inhibiting natural killer (NK) cell activity. Thus, like MSCs, ELA cells might participate in various regenerative processes, or they might work concomitantly with MSCs, to allow new therapies for a wide range of common and orphan diseases .
Materials and Methods
SF isolation and cell culture
SF was extracted from the knees of patients diagnosed with osteoarthritis (OA) following appropriate institutional approved protocols. Within 24 h of harvest, the SF was diluted 10:1 in dilution buffer (phosphate-buffered saline (PBS) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT)) and 10 mM ethylenediaminetetraacetic acid (EDTA; Gibco, Grand Island, NY). To extract the cellular component, the diluted SF was spun at 500 x g for 30 min, and the pellet was resuspended in dilution buffer. This process was repeated twice at 300 x g for 30 min, and the final pellet was resuspended in Hank’s balanced salt solution (HBSS; Gibco). The pelleted cells were either directly analyzed or subjected to culture expansion and differentiation. Samples that underwent culture expansion and differentiation were suspended in growth medium (MSCGM™ Human Mesenchymal Stem Cell Growth BulletKit™ Medium or MSCGM-CD™ Mesenchymal Stem Cell Chemically Defined Medium with or without 1% FBS; Lonza, Basel, Switzerland) and plated at a concentration of 3000 cells per cm2 (225,000 cells total) in a T-75 vented tissue culture flask (BD Biosciences, San Jose, CA). All culture media were supplemented with 100 U/ ml penicillin and 1000 U/ml streptomycin (PCN-Strep; Gibco) and exchanged every 48 h. Once cells reached >80% confluence, they were harvested with Trypsin-EDTA and replated into new T-75 flasks. Cells were cultured at 37oC with 5% CO2 for all experiments. Samples were either immediately seeded into cell culture or mixed 1:1 with freezing medium composed of Dulbecco’s Modified Eagle’s Medium (Gibco) supplemented with 20% FBS and 20% dimethylsulfoxide (Sigma-Aldrich) and stored in liquid nitrogen at less than -150°C.
Flow cytometric characterization of ELA cells
Antibody use was based on the minimal surface marker panel proposed by the International Society of Cellular Therapy . Fluorochrome-labeled antibodies against the following markers and matching isotype controls were obtained from BD Biosciences: CD44-PE, CD45-PE, CD49-PE, CD105-PE, CD133-PE, CD34-PE (clone 581), CD73-PE, CD90-PE, CD99-PE,CD235a-PE, MUC1-PE, HLA Class1-PE, HLA-DR-PE, SSEA1-PE, HLA-DR-PE, IgG1-PE, IgG1-FITC, IgG2a-PE, IgG2bkappa-PE, IgM-PE, CD4-FITC, CD8-FITC, CD69-PE, CD3-FITC, and CD25- PE. Anti-CXCR4-PE (CD184) antibody was obtained from R&D Systems (Minneapolis, MN). Anti-PD1-PE (CD279) antibody was obtained from eBiosciences (San Diego, CA). Upon confluence, cells from one 75 cm2 flask were harvested, washed, and counted. Cells were kept on ice and suspended in incubation buffer (Dibco’s PBS + 2% FBS + 1 mM EDTA). After centrifugation and aspiration of supernatant, 5 or 10 μl of antibody (depending on cell number) was applied directly onto the pellet. Cells were incubated at 4°C for 30-45 min, washed, resuspended, and analyzed in a FACSCalibur machine using CellQuest™ software (BD Biosciences). The pluripotent properties and ESC marker status of ELA cells were determined by intracellular staining using monoclonal antibodies against OCTA4- PE, RUNX2-PE, SOX9-PE, REX1-PE, NANOG-PE, and KLF4- PE with matching isotype controls and by RT-PCR of freshly isolated and culture-expanded cells. For intracellular staining, cells were permeabilized with Cytofix/Cytoperm solution (BD Biosciences) and thereafter subjected to intracellular staining with the appropriate antibodies.
Self-renewal properties of ELA cells
ELA cell cultures grew as monolayers. Following cell sorting, both marker-positive and marker negative cultures were set up in parallel. At days 3 and 5, cultures were rinsed with PBS, detached with trypsin-EDTA, centrifuged, and resuspended in media. Duplicate aliquots were placed into 96-well plates, and 10 μl of Cell Counting Kit-8 solution (Dojindo Molecular Technologies Inc., Gaithersburg, MD) was added to each well. Following 3 h incubation at 37°C, A450 was measured using a Victor5 Light Luminescence Counter (PerkinElmer Life Sciences, Boston, MA) and compared with standards of known cell numbers. To detect apoptotic cells, cultures were fixed and stained with the fluorescence-based ApoAlert DNA Fragmentation Assay Kit (BD Biosciences) following the manufacturer’s protocol.
Adipogenic, chondrogenic, and osteogenic differentiation of cells isolated from synovial fluid
ELA cells were suspended in chemically defined media with 1% FBS, passaged upon reaching 80% confluence, and plated in 75cm2 vented cell culture flasks at a concentration of 150,000/cm2, with a total volume of 25 ml per flask. After 20 passages, cells were plated into 12-well plates at a concentration of 200,000 cells/well and cultured in the appropriate differentiation media. For adipogenic differentiation, cells were cultured in StemPro Adipocyte Differentiation Media (Invitrogen) supplemented with PCN-Strep at a total volume of 1.5 ml/well. The media was changed every 48 h. After 21 d, the cells were harvested for histochemical staining and real-time quantitative PCR (qPCR). Differentiated cells were stained with fresh Oil Red O solution (Sigma-Aldrich) to verify adipocyte characteristics. For chondrogenic differentiation, cells were cultured in Osteocyte/Chondrocyte Differentiation Basal Medium (Invitrogen) with Chondrogenesis Supplement (Invitrogen). These cell cultures were then stained with Alcian Blue (Sigma-Aldrich) for chondrocyte detection. For osteogenic differentiation, cells were cultured in Osteocyte/Chondrocyte Differentiation Basal Medium (Invitrogen) with Osteogenesis Supplement (Invitrogen). To assess the presence of osteoblasts, cell cultures were stained with 5-bromo-4-chloro-3- indolyl-phosphate/nitro blue tetrazolium (NBT/BCIP; Invitrogen). In all three differentiation studies, positive cells were assayed by counting 50-100 cells in multiple fields using light phase microscopy.
RNA isolation and RT-PCR
Total RNA was extracted from freshly isolated and culture-expanded undifferentiated cells as well as differentiated cells. Total RNA was purified using TRIzol® reagent according to the manufacturer’s protocol (Invitrogen). The same source of RNA was used for RT-PCR and DNA microarray analysis (see below). First-strand cDNA was obtained by reverse transcription using 3 mg total RNA according to the manufacturer’s instructions (Invitrogen). Primer sequences are shown in Table 1. PCR products were electrophoresed on 1.5% agarose gels to verify DNA fragment sizes. For DNA microarray analysis, cDNA was synthesized using the SuperScriptTM III First-Strand Synthesis System (Miltenyi Biotec, San Diego, CA). RT-PCR assays were performed using qPCR Mastermix Plus for SYBR Green (Miltenyi Biotec) according to the manufacturer’s protocol (see below). For normalization, differential levels of gene expression were calculated in relation to beta actin and expressed as a ΔCT value, as previously described .
Preparation for DNA microarray analysis
Total RNA was extracted from synovial ELA cells in a monolayer of human synovial ELA cells cultured for 3 d. Cells were lysed using the SuperAmp preparation kit and delivered to Miltenyi Biotec on dry ice. SuperAmp RNA amplification was performed according to Miltenyi Biotec’s protocol based on a global PCR protocol. mRNA was isolated using magnetic bead technology. Amplified cDNA samples were quantified using an ND-1000 Spectrophotometer (NanoDrop Technologies), and 250 ng of each cDNA were used as template for Cy3 and Cy5 labeling according to Miltenyi Biotec’s protocol. Cy3- and Cy5-labeled cDNAs were combined and hybridized for 17 h at 65°C to the Agilent Whole Human Genome Oligo Microarray 4x44K probe set using Agilent’s recommended hybridization chamber and oven. Control samples were labeled with Cy3 and experimental samples were labeled with Cy5.
Data processing and analysis
Feature Extraction Software (Agilent) was used to read and process the microarray image files and raw datasets. These datasets, together with publically available datasets from the NIH Gene Expression Omnibus (GEO), were exported to JMP software (SAS Institute Inc., Cary, NC) for further analyses. The input datasets were transformed into log base 2, and row-byrow statistics were computed. Datasets were normalized to the median global intensity [25,26].
Hierarchical clustering and functional analysis
To identify genes expressed at high levels in ELA cells, an unsupervised hierarchical clustering was performed on the normalized dataset. JMP Software was used to perform and visualize this clustering. One-way ANOVA was performed on the data obtained from the hierarchical clustering, and a volcano plot was generated to represent the intensity ratio for each gene in ELA cells and MSCs. The x-axis displays the log2 ratio of the gene intensities. A log2 ratio of 1corresponds to approximately two-fold change. The y-axis shows the –log10 (p-value) for the comparison between ELA cells and MSCs. Genes that were differentially expressed in ELA cells and MSCs were identified. These genes, along with their fold-change values, served as the input to the Ingenuity Pathway Analysis (IPA®, Qiagen) program. Differently expressed genes were uploaded into the IPA application and used as the starting point for generating biological networks . A right-tailed Fisher’s test with α=0.05 and the whole database as a reference set was used to determine if the enrichment of genes with particular biological functions or molecular processes was significant.
Immunomodulatory properties of ELA cells
The immunosuppressive properties of ELA cells were assayed by several methods. Suppression of T cell proliferation was determined by in vitro co-culture experiments performed in triplicate. ELA cells were irradiated in suspension with a dose of 30 Gy prior to coculture using the Gamma cell Elan device (Best Theratronic, Ottawa, ON, Canada). Irradiated and non-irradiated ELA cells, either freshly isolated or cryopreserved and then cultured for 24-48h, were co-cultured with freshly isolated human peripheral blood mononuclear cells (PBMCs) at a 1:10 ratio for 5 d at 37°C. ELA cell suppressive function was determined by a [3H]-Thymidine (1μCi/well; 37kBq; NEN-Du- Pont) uptake assay, as previously described . Data are expressed as counts per minute (cpm) or as a stimulation index (SI). SI was determined by calculating the ratio of experimental [3H]-Thymidine incorporation to background [3H]-Thymidine incorporation by unstimulated T cells. These experiments also assessed if allo-reactive T cells were stimulated by ELA cells, which would be indicated by higher counts in a proliferation assay.
To further demonstrate their immunosuppressive properties, ELA cells were co-cultured with T cells labeled with 5-6-carboxyfluorescein diacetate succinidyl ester (CFSE; Cell Trace Cell Proliferation Kit; Molecular Probes/Invitrogen Life Technologies) in 96-well plates at 1:10, 1:20, and 1:40 ratios in triplicate, along with non-ELA cell controls. T cell proliferation was stimulated with CD3/CD28 and analyzed with flow cytometry for CFSE fluorescence after 5 d. Immunosuppressive properties of MSCs (Lonza, Walkerville, MD) were assayed in parallel with the same methods. Flow cytometry data was analyzed using FlowJo software (Ashland, OR) to obtain the Proliferation Index (PI). T cell suppression for each sample was calculated as (1- ([PI with ELA cells]/[PI without ELA cells]) x 100. As a final assay for immunosuppression, freshly isolated ELA cells and PBMCs were co-cultured 1:10 in 96-well plates for 5-7 days, harvested into 5 ml tubes, and labeled with a combination of directly conjugated antibodies as follows: CD4-FITC/ CD25-PE; CD8-FITC/CD25-PE; CD3-FITC/CD69-PE, and CD3- FITC/PD1-PE, as well as matching isotype controls. The percentages of CD4+ or CD8+ T cells expressing CD25, and CD3+ T cells expressing CD69 or PD1 were determined by bi-dimensional FACS analysis.
ELA cell suppression of NK cell activity was determined by a chromium release assay . NK cells were co-cultured with equal numbers of ELA cells in RPMI 1640 tissue culture media (Mediatech, Herndon, VA) supplemented with 10% pooled human AB serum, antibiotics, and cytokines for 24 h. Following incubation, NK cells were transferred to wells for co-culture with chromium-labeled K562 target myeloma cells (ATCC, Manassas, VA) at 10:1, 5:1, 2.5:1, and 1:1 ratios. Specific cytotoxicity was calculated as 100 x (experimental-spontaneous)/ (maximum-spontaneous).
Results are expressed as mean ± SEM. Statistical comparisons were performed using the Student’s t-test. P values <0.05 were considered statistically significant.
Cellular phenotypes of SF cells
SF contains a wide variety of mononuclear cells, some of which revealed a forward and side scatter flow cytometry pattern similar to peripheral blood. This pattern consists of four regions, three of which represent neutrophils, myeloid cells, and lymphocytes (Figure 1A). The fourth region, much smaller in size and side scatter, was previously thought to primarily represent cell debris and RBCs (Figure 1B). We found that this population had less fluorescence compared with other regions and <200 forward scatter (FSC) linear units (Figure 1A). Analysis of this cell population from 5 OA patients revealed a mean viability of 94 ± 0.65% and a mean cell size of 5.9 ± 0.31 μm (range: 4-8 μm) (Figure 1C).
Figure 1. Identification of early lineage adult (ELA) stem cells in the synovial fluid (SF) of patients with osteoarthritis (OA). (A) Representative forward- and side-scatter profiles of mononuclear cells isolated from the SF of an OA patient indicating the location of a small cell population in relation to other cell types. (B) Forward- and side-scatter profile of a gated small population of cells depicting a heterogeneous population with a varied cell size and scatter profile. (C) Viability and cell-size determination of a gated small cell population using the Roche CASY Cell Counter and Analyzer System. (D) Expression of pluripotent intracellular and surface markers, as determined by FACS analysis of a gated population of small cells in Group 1, Group 2, and Group 3. (E) RT-PCR analysis of pluripotency marker expression in a small cell population isolated from three separate samples of SF, with NTERA-2 cells (a stem cell line) as a positive control. Primers are specific for transcripts from the respective endogenous locus. GAPDH was used as a loading and internal control.
We next evaluated the phenotypic characteristics of these cells by staining for proteins associated with peripheral blood mononuclear cells and ASCs. Further flow cytometric analysis showed three distinct subgroups of cells (Figure 1B). CD45 and CD235a, surface markers of leukocytes and RBCs, respectively, were absent from this cell population. MHC Class I, a protein found on all cells except RBCs and immature stem cells, was observed on a subset of this group (Figure 1D). Furthermore, we observed very high expression of MUC1 and CD99 on this cell population. Further analysis revealed no expression of CD73, CD90, CD105, CD133, CXCR4, or SSEA-1 on any of the subgroups. However, intracellular staining with directly conjugated monoclonal antibodies revealed high expression of REX1 and varying degrees of OCT4, NANOG, SOX9, and RUNX2 expression (Figure 1D). RT-PCR analysis from three patients revealed that most of the tested pluripotency-associated mRNAs were expressed in this small cell population, with REX1 being the most highly expressed (Table 1, Figure 1E). Notably, we identified a splicing variant of NANOG, which might suggest a greater diversity of self renewal and pluripotency proteins. The Ntera cell line was used as a positive control in these studies , and a sample lacking reverse transcripts was used as a negative control.
ELA cell growth and self-renewal
In order to investigate and optimize cell growth, ELA cells were cultured in three different media types; standard expansion medium, chemically defined medium (CD), and CD supplemented with 1% FBS. ELA cells cultured in standard medium reached 90% confluence in 8 days for donor A and 7 days for donor B, generating 2.8 x 106 and 2.2 X 106 cells, respectively. Doubling times for both samples were greater than 120 h. In contrast, when ELA cells were cultured in CD in the absence or presence of 1% FBS, higher yields were generated with shorter doubling times. Notably, CD supplemented with 1% FBS appeared to be the optimal media for self-renewal capacity, generating yields of 31.5 x 106 (donor A) and 21.2 x 106 (donor B) ELA cells with doubling times of 19h and 27h, respectively (Figure 2A-B). We therefore chose this culture media for all future experiments. In CD media with 1% FBS, cell morphology was round immediately after plating but became elongated and spindle-like within 4 d (Figure. 2C-D).
Figure 2. Cell culture and self-renewal properties of early lineageadult (ELA) cells in vitro. (A) Bar graph depicting growth of ELA cells from two separate donors cultured in standard culture media or chemically defined (CD) culture media with or without 1% fetal bovine serum (FBS). (B) Phase micrographs demonstrating pattern and density of ELA cell growth at two different magnifications. (C) Phase micrographs exhibiting the pattern and density of ELA cell growth at days 1, 4, and 7 in CD media supplemented with 1% FBS from three separate donors. (D) Total ELA cell counts from three separate donors cultured in CD media with 1% FBS at different passage numbers. Numbers in parenthesis represent population doubling time during labeled passage growth period.
ELA cell differentiation
The differentiation potential of ELA cells was investigated by culturing cells under conditions that favored adipogenic, chondrogenic, and osteogenic differentiation. Cells cultured in Adipocyte Differentiation Media for 21 d formed vacuoles that stained positive for Oil Red O, a fat-soluble dye that stains lipids (Figure. 3A). Control cells showed no incorporation of Oil Red O. Cells cultured for 21 d in Chondrocyte Differentiation Media exhibited diffuse Alcian Blue staining, suggesting production of acid mucopolysaccharides and glycosaminoglycans normally found in cartilage (Figure 3B). Cells cultured in Osteocyte Differentiation Media and stained with NBT/BCIP revealed flat, purple cell bodies (Figure 3C). NBT/BCIP is converted into purple stain by alkaline phosphatase, an enzyme found in osteoblasts. No enzymatic activity was observed in control cells.
Figure 3. Differentiation of early lineage adult (ELA) cells to adipocytes, chondrocytes, and osteocytes. The differentiation potential of ELA cells was investigated by culturing cells for 21 d under conditions that favored adipogenic, chondrogenic, or osteogenic differentiation. (A) Adipogenic differentiation was indicated by accumulation of neutral lipid vacuoles that stained with Oil Red O. (B) Chondrogenic differentiation was assayed with Alcain Blue, which labels the acid mucopolysaccharides and glycosaminoglycans of cartilage. Diffuse blue staining was observed throughout the slide. (C) Osteogenic differentiation was assayed with BCIP/NBT, a substrate that turns purple in the presence of alkaline phosphatase. Uninduced cells were used as negative controls in the differentiation experiments. Total RNA was extracted from these differentiated cells, and cDNA derived from mRNA was amplified based on the global PCR protocol described in Materials and Methods. (D-F) Real time RT-PCR analysis of selected specific genes expressed in differentiated ELA cells. The expression of genes was compared to the expression of beta-actin as an internal control and the values expressed as ΔCT. Negative bars indicate a decrease in expression of that particular gene. (D) ELA cells differentiated into adipocytes. (E) ELA cells differentiated into chondrocytes. (F) ELA cells differentiated into osteocytes.
In order to further investigate the extent of induced ELA cell differentiation, cells were harvested to investigate the presence of adipogenic, chondrogenic, and osteogenic genes by qPCR. Cells from the adipogenesis conditions showed high expression of the adipocyte lineage genes PPARG-tv1, PPARGtv2, LPL, FABP4, ADIPOQ, LEP, PLIN, and CFD (Figure 3D). Adipogenesis- specific genes were also detected in chondrogenic and osteogenic conditions. Cells from the chondrogenic conditions showed high expression of the chondrocyte lineage genes BGN, DCN-tvA2, ANXA6-tv2, MMP13, SRY, and COMP and low/absent expression of MATN1 and COL2A1 (Figure. 3E). These genes were also detected at similar levels in cells undergoing adipogenesis or osteogenesis. Cells from the osteogenic conditions showed high expression of the osteocyte lineage genes RUNX2-tv3, RUNX2-tv1, RUNX2-tv2, and PHEX, similar to the chondrogenic conditions. RUNX2.2 and PHEX were also detected in adipogenesis conditions. Low/absent expression of BGLAP, SPP1, SPP2, and SPP3 was also observed in the osteogenic conditions (Figure 3F). There was low/absent expression of all lineage-specific genes in control cells. Although cells were cultured in specialized media for differentiation, the differentiation process was not absolute because cells also expressed genes from other lineages (Figure 3D-F). These observations suggest that detailed studies are necessary to determine optimum culture conditions for stem cell differentiation, and that molecular signatures may be necessary to define the differentiation process.
Large-scale gene expression profiling of freshly isolated, undifferentiated ELA cells showed expression of differentially expressed progenitor and tissue-specific genes with diverse functions (Table 2). These profiling studies suggested the potential of ELA cells to differentiate into other lineages, such as neuronal, cardiac, pancreatic, and liver cells (Tables 2 and 3). Moreover, expression of genes with specific cellular functions, such as mucins, ICAM, tetraspans, and collagens (Table 3), suggests that ELA cell phenotype may not be tissue specific. Instead, ELA cells might be a heterogeneous population of multipotent cells with the capacity to differentiate into endodermal, mesodermal, and ectodermal cells.
Table 2. Selected panel of differentially expressed progenitor-‐ and tissue-‐speciDic genes from freshly-‐isolated ELA cell population
Freshly harvested ELA cells (small population) was isolated from SF of OA patient. Total mRNA from the sample was used for
Table 3. Expression of selected group of genes for secreted proteins (collagens/mucins) and surface-‐expressed proteins (mucins, ICAMS, and tetraspans) in freshly-‐isolated ELA cells
ELA cells were isolated from SF of OA paEents and the total RNA was extracted for microarrary analysis. (Chr. Chromosome)
DNA microarray analysis
To compare ELA cells and MSCs, we examined their respective gene expression profiles in triplicate by microarray. The correlation coefficient between these microarray datasets obtained from repeated experiments was greater than 0.98, indicating high reproducibility. The entire set of expressed protein- coding genes was used for a non-supervised hierarchical clustering analysis. The dendrogram in Figure 4A shows that freshly isolated and frozen/expanded ELA cells isolated from the same tissue were found in different clusters, whereas the three categories of MSCs (BM-derived, CD105+, and CD133+) clustered together. These results suggested that ELA cells have a gene expression profile distinct from MSCs.
Figure 4. Comparison of gene expression profiles of early lineage adult (ELA) cells and mesenchymal stem/progenitor cells (MSCs). Hierarchical cluster analysis of qRT-PCR data was performed on freshly isolated and cultured/expanded ELA cells and bone marrow (BM)-derived, CD105+, and CD133+ MSCs. Expression levels were normalized to β-actin. Black represents 1, red represents >1, green represents <1, and grey represents below detection limits. (A) Dendrogram comparing gene expression in primary ELA cells, BM-derived MSCs, CD105+ MSCs, and CD133+ MSCs. (B) Dendrogram of a hierarchical cluster analysis comparing genes expressed in expanded ELA cells against MSC gene data sets available from the NIH gene Expression Omnibus in addition to those generated in this study. (C) Volcano plots comparing specific genes upregulated in primary ELA cells and expanded ELA cells (upper left), BM-derived MSCs (upper right), CD105+ MSCs (bottom left), and CD133+ MSCs. Genes expressed above the broken red line represent genes specific to primary ELA cells on the left and genes specific to MSCs on the right. (D) Venn diagram comparing upregulated genes in expanded ELA cells, BM-derived MSCs, CD105+ MSCs, and CD133+ MSCs. (E) Venn diagram depicting the percentage of genes expressed in expanded ELA cells that overlap with other categories of stem cells, utilizing published gene datasets available from the NIH Gene Expression Omnibus.
To verify that ELA cells are a distinct population of cells compared with MSCs, we utilized high-density oligonucleotide microarrays and functional network analysis. DNA microarray analysis was used to identify genes expressed in ELA cells, and the results were compared with datasets from the NIH GEO and our laboratory (Figure 4B). This analysis revealed that 25% of the genes expressed by ELA cells were shared by BM-derived MSCs (Figure 4B). The results were visually represented by a volcano plot to compare specific genes upregulated in ELA cells and MSCs (Figure. 4C). Additionally, we generated Venn diagrams to compare gene expression in ELA cells and BM-derived, CD105+, and CD133+ MSCs (Figure 4D). From this analysis, we concluded that ELA cells are a distinct population of ASCs.
DNA microarray analysis identified genes specific to MSCs, ESCs, and induced pluripotent stem cells (iPSCs) in the ELA cell population (Figure 4E). For example, ESCs expressed 1460 genes that were not expressed by the other stem cell types. ELA cells expressed unique genes as well as genes in common with ESCs, MSCs, and iPSCs (Table 4). Furthermore, ELA cells had 616 genes in common with ESCs, signifying a 42% overlap in the genetic profile of the cells. This finding is significant, as these genes play a role in both cellular function, such as cell cycle, RNA post-translation modification, cell death and DNA replication, and stem cell identity, such as self-renewal and pluripotency. Additionally, IPA determined that shared signaling pathways for DNA replication, recombination, and repair were significantly enriched between ELA cells, ESCs and iPSCs (right-tailed Fisher’s test, α=0.05), as shown in Table 4. Collectively, our data suggests that ELA cells are functional ASCs with a unique set of expressed genes that are not shared with other categories of stem cells.
Immunomodulatory capacity of ELA cells
There was no significant difference between irradiated and non-irradiated ELA cells co-cultured with fresh PBMCs in various ratios. Increasing the number of ELA cells did not influence the proliferative capacity of T cells in the co-culture (Figure 5A). In additional experiments, fresh and cryopreserved ELA cells were co-cultured with freshly isolated PBMCs from three healthy donors at a 1:10 ratio for 5 days (Figure 5B). No significant proliferation of allo-reactive T cells was observed. The SI was 1.72 (± 0.19; n=3) for fresh ELA cells and 1.29 (± 0.36; n=3) for cryopreserved cells. Treatment of ELA cells with the mitogen phytohemagglutinin showed no significant proliferation compared with the very active proliferation with freshly isolated PBMCs (Figure 5B). ELA cells at a ratio of 1:10 elicited a moderate suppressive effect on CD3/CD28-stimulated, CFSE-labeled T cells (15.6 ± 2.7%; n=3) compared with a relatively low suppression effect by MSCs (2.8% ± 0.25; n=3)(Figure 5C). ELA cells at passages 3, 5 and 6 suppressed CD3/ CD28-stimulated T lymphocytes, indicating that suppression properties are maintained when ELA cells are expanded in culture, although cells at passage 3 were most effective (Figure 5D)
Table 4. Common genes expressed in ELA, uESC, IPS, and MSCs together with enriched canonical pathways
Next, we wanted to determine if ELA cells promoted the expansion of CD4+/CD25+ regulatory T cells. ELA cells were co-cultured with freshly isolated PBMCs for a period of 5 d, stained with appropriate antibodies, and analyzed by bi-dimensional FACS analysis. The percentage of CD4+ T cells expressing CD25 expanded 3-fold (10 ± 0.36%; n=3) when co-cultured with ELA cells compared with PBMCs (1.6 ± 0.36%; n=3) (p=0.01) (Figure 5E-F). The percentage of CD8+ T cells expressing CD25was 3.6 ± 0.36% when cultured with ELA cells (n=3) compared with 1.6 ± 0.36% with PBMCs (n=3). Furthermore, a small population (< 1%) of T cells co-cultured with ELA cells expressed PD1 (Figure 5E-F). PD1 is generally associated with exhaustion of T cells. We also found a modest increase in CD69, a surrogate marker of T cell responsiveness to mitogen and antigen stimuli, in CD3+ T cells cultured with ELA cells (Figure 5E-F). Taken together, these studies indicate that ELA cells might perform their immunosuppressive functions by inhibiting T cell proliferation and expanding CD4+/CD25+ regulatory T cells.
MSCs are known to inhibit the expression of activating receptors on the surface of NK cells and potentially impair their cytotoxic activity. To evaluate potential ELA cell-mediated inhibition of NK cell lytic potential, we performed cytolytic assays. Purified populations of NK cells were co-cultured overnight with and without ELA cells at a 1:1 ratio and than exposed to  Cr-labeled K562 target cells at various ratios. Cytolytic activity was measured by  Cr release. NK cells pre-incubated with ELA cells demonstrated >60% reduction in cytotoxic effects. This reduction was consistent across decreasing concentrations of NK cells (Figure 5G). These data suggested that ELA cells have a suppressive effect on NK cells.
Figure 5. Immunomodulatory potential of early lineage adult (ELA) cells. (A) Stimulation of allo-reactive T cells was determined by co-culturing irradiated and non-irradiated ELA cells with freshly isolated peripheral blood mononuclear cells (PBMCs) at ratios of 1:10, 1:100 and 1:1000 in triplicate for 5 d. PBMCs were isolated from healthy donors, and previously expanded ELA cells were irradiated. To determine the proliferation of allo-reactive T cells, cultures were pulsed with 3[H]-Thymidine (1 μCi/well) 18 h prior to harvesting. Control background Thymidine uptake in PBMCs alone was (487 ± 62.9; n=5). Bar graphs represent the mean ± SEM of 5 separate experiments. (B) The effectiveness of cryopreserved ELA cells to stimulate allo-reactive T cell proliferation was determined as in (A). Bar graphs represent the mean of 3 separate experiments ± SEM. (C) Immunosuppressive
properties were further investigated by culturing ELA cells or MSCs with 5-6-carboxyfluorescein diacetate succinidyl ester (CFSE)-labeled allo-reactive T cells at various ratios in triplicate. After 5 d culture, T cells were analyzed by flow cytometry to determine CFSE fluorescence. T cell suppression was expressed as the Proliferation Index. (D) The immunosuppressive effectiveness of each ELA cell passage was determined by using different passage numbers of cells co-cultured with CFSE-labeled T cells. (E) A bivariate dot plot analysis of a representative experiment of expanded ELA cell/freshly isolated PBMC co-culture (1:10 ratio) to determine the expression of CD25 on CD4+ and CD8+ T cells and the expression of CD69 or PD1 on CD3+ T cells (* p < 0.001; ** p < 0.01 as compared with MSCs). (F) Bar graphs represent the mean of 3 separate experiments ± SEM (* p < 0.001 as compared with PBMCs alone). (G) The inhibitory effect of ELA cells on NK cell cytotoxicity was demonstrated by co-culturing NK cells with or without ELA cells prior to incubation with target cells at various ratios.
Many groups have studied human BM stromal cells and demonstrated phenotypic and functional heterogeneity [22,31,32]. It has been hypothesized that naïve stem cells have a greater capacity to differentiate into functional cells and tissue . Stem
cell nomenclature used to refer to a heterogeneous population of marrow stromal cells, but it now just refers to MSCs, a single type of ASCs. During early bone marrow studies, cells with self-renewing capacity were discovered and referred to as stem cells. These stem cells were later subdivided into hematopoietic and non-hematopoietic populations [34,35]. HSCs, one of earliest and best characterized BM stromal stem cell types, allowed the first truly successful cellular therapies. To date, over 50,000 HSC transplants have been performed worldwide . Non-hematopoietic MSCs were commonly identified by their ability to differentiate into mesodermal tissues . However, not all marrow stromal cells are phenotypically or functionally identical, and therefore not all are capable of differentiating into similar tissues [38-40]. Most of our understanding of marrow stromal cells comes from studies of MSCs, leading many to believe that there is one category of ASC derived from the bone marrow stroma that has a hierarchical relationship with other ASCs . However, within the bone marrow stroma there exists a heterogeneous population of ASCs, such as mesenchymal precursor cells (MPCs), marrow-isolated adult multilineage inducible (MIAMI) cells , multipotent adult progenitor cells (MAPCs) , and very small embryonic-like (VSEL) stem cells . Recently, our group isolated ELA cells, a population of ASCs that contribute to the heterogeneity of the marrow stromal cell population. Despite standardization of protocols for stem cell biomanufacturing, many protocols start with a heterogeneous population that might not represent the appropriate stem cell for a particular treatment. To address this concern, we plan to focus our future studies on identifying subsets of the primitive stem cells based on molecular profiles with the intention of clonally expanding them.
ELA cells represent a CD99+/MUC1+/CD235a-/MHC class I population from the SF of osteoarthritic patients. The properties of these cells were determined by protein and mRNA analysis for pluripotency genes and proteins, such as OCT4 (embryonic form), REX1, and NANOG. Our isolation method was less cumbersome than those required for the enrichment of stem cells from other tissues. The primary reason for this was the absence of RBCs in the SF. Accurate measurements of ELA cell size, volume, and viability were performed with the CASY Cell Counter and Analyzer system. ELA diameter was 4-6 μm, in contrast to RBCs, which have a diameter of 6-8 μm. Notably, the majority of previous stem cell studies have utilized forward and side scatter perimeters on flow cytometry, which excludes cells < 6 μm. This gating strategy may explain why the ELA cell population has been overlooked until now.
Our studies have shown that the genes ELA cells share with MSCs are not the same as those shared with IPS cells or ESCs. ELA cells are similar in size to VSEL cells, which express OCT-4 and can differentiate into the three germ layers . However, the ELA cell population does not express CXCR4, SSEA-4, CD34, or CD133, which are additional markers used to identify the VSEL cell population . Moreover, ELA cells proliferate in the absence of feeder cells, making them functionally distinct from VSEL cells. Taken together, these findings suggest that the ELA cell population is distinct from the VSEL cell population. This raises the question of whether ELA cells represent a heterogeneous population of primitive stem cells and whether these subsets can be isolated and clonally expanded for therapeutic strategies for tissue repair.
In addition to the expression of the pluripotency gene transcripts NANOG, OCT4, REX-2, and DPPA3, ELA cells also express high levels of MUC1. Our transcriptome studies confirmed high expression of mucin genes. Recent studies have shown a relationship between MUC1 and ESC differentiation . There is a need to distinguish the most primitive ASCs, in particular ELA cells, from more mature forms, as primitive ASCs might possess broader differentiation capacity  and function as a better starting population for stem cell therapy. Future studies will involve the use of cell surface proteins such as MUC1 to distinguish between primitive and mature ASCs. Collectively, our findings suggest that ELA cells might represent a primitive form of ASCs, making them useful for stemcell- based treatment.
Although ELA cells reside in a dormant state in the SF (unpublished data), their origin is unclear. It is unlikely that SF converts MSCs into ELA cells. We hypothesize that the origin of ELA cells is the BM and not the systemic circulation. We base this hypothesis on elegant studies performed by Nakagawa and colleagues, who demonstrated that BM stromal cells migrate directly from the BM to the joint space in a collagen-induced arthritis model [46,47]. We propose that the enhanced migration of ELA cells into the joint space (synovial cavity) increases as a result of inflammation that accompanies osteoarthritis. This inflammatory environment is known to increase the number and size of bone canals, which allow communication between the BM and the synovial cavity [46,48]. This increase in bone canal size may allow more ELA cells to migrate into the joint space. Under non-inflammatory conditions, these canals are smaller in size and number, thus limiting the number of ELA cells in the joint space.
Our histochemical data suggests that ELA cells can differentiate into various tissues. However, molecular profiling of these tissues showed that transcripts specific for other tissues are also expressed, some of them at high levels. This property of ASCs differentiated in vitro might represent an intermediate phenotype that possesses the ability to transdifferentiate into other tissue types. It is also possible that in vitro culture conditions are not representative of in vivo conditions that induce terminal differentiation. These findings support a concept described by Quensenberry and colleagues suggesting that expansion and differentiation of cells and changes in gene expression are continuous and reversible, and that sorting stem cells by static cell surface proteins may leave behind a large percentage of potential stem cells .
It was determined that MSCs possess immunomodulatory properties [50,51] and might play specific roles in the maintenance of peripheral tolerance, transplantation tolerance, autoimmunity, tumor evasion, and fetal-maternal tolerance . We focused on the role of ELA cells in modulating the immune response by activating T cells. We demonstrated that ELA cells do not induce an allo-immune response, suggesting that they primarily immunomodulate suppression by affecting the effector arm of the immune response. Notably, our in vitro data suggests that ELA cells suppress T cell activation and induce regulatory T cells. Moreover, ELA cells did not upregulate a surrogate marker of T-cell responsiveness (CD69) in CD3+ T cells . ELA cells might interfere with T cell function in a PD-1 independent pathway . Although PD-1 was not upregulated in either CD4+ or CD8+ T cells, this does not preclude the possibility that ELA cells secrete factors or express cell surface proteins that may modulate T cell function. In addition, ELA cells were shown to inhibit the cytolytic capacity of NK cells. Taken together, these findings suggest that ELA cells evade the immune system by interfering with adaptive and innate immunity.
Our observations of the ELA cell population are of fundamental importance to the field of regenerative medicine and the development of cell therapeutics. There is an ever-growing need for stem cells that replace, regenerate, and modulate immune function. However, relying on cell and tissue donation is unreliable and cannot address the need for ASCs. Biomanufacturing of ASC therapies is the most logical option . Although ASCs can be efficiently expanded in the laboratory, this expan-sion cannot be easily translated to large-scale production for therapeutic purposes due to technical issues. To address these concerns, further studies are underway to assess whether prolonged culture periods affect the expression of ELA cell-specific genes and ELA cell function.
We have demonstrated the existence of a population of ASCs, referred to as ELA stem cells, which reside within the SF of osteoarthritic joints. These cells are phenotypically distinguishable from other ASCs. ELA cells are also molecularly distinguishable from ESCs, IPS cells, and MSCs by their unique gene expression patterns. ELA cells represent a new form of primitive ASCs with in vitro tissue regenerative capability that might benefit regenerative therapies. As with MSCs, which are a promising tool for therapeutic approaches aimed at inhibition of the immune response, ELA cells might also be useful for preventing or suppressing graft-versus-host disease in patients undergoing allogeneic HSC transplantation or for the treatment of certain autoimmune diseases. In conclusion, ELA cells have far reaching implications in both basic research and regenerative medicine.
The authors gratefully acknowledge the helpful advice provided by Drs. Andrew Makarovskiy and John Garvey. In addition, we gratefully acknowledge the help and assistance provided by the logistics staff for collecting and preparing synovial fluid samples
for research purposes. Finally, we would like to acknowledge the patients who participated in providing samples for the studies.
Research reported in this publication was supported by the Department of Orthopedic Surgery at Brigham and Women’s Hospital and Asclepius Laboratories. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.