Academic Rheumatology, Department of Clinical Science at North Bristol, University of Bristol, Bristol, United Kingdom
University of Bristol Academic Rheumatology, Department of Clinical Science at North Bristol, AMBI Research Laboratories, Avon Orthopaedic Centre, Southmead Hospital, Bristol BS10 5NB, United Kingdom. Telephone: 44-117-959-6171; Fax: 44-117-959-6187
The identification of stem cell–specific proteins and the elucidation of their novel regulatory pathways may help in the development of protocols for control of their self-renewal and differentiation for cell-based therapies. Nucleostemin is a recently discovered nucleolar protein predominantly associated with proliferating rat neural and embryonic stem cells, and some human cancer cell lines. A comprehensive study of nucleostemin in human adult bone marrow stem cells is lacking. The aim of the study was to determine if nucleostemin is synthesized by adult bone marrow stem cells and to analyze its expression during their expansion and differentiation. Using a multipotential adherent population of stem cells, nucleostemin was localized to the nucleoli and occurred in 43.3% of the cells. There was a high level of expression of nucleostemin mRNA in bone marrow stem cells and this remained unchanged over time during cell expansion in culture. When bone marrow stem cells were stimulated to proliferate by fibroblast growth factor (FGF)-2, nucleostemin expression increased in a dose-dependent manner. Small interfering RNA (siRNA) knockdown of nucleostemin abolished the proliferative effect of FGF-2. When bone marrow stem cells were differentiated into chondrocytes, adipocytes, or osteocytes, nucleostemin expression was 70%–90% lower than in the undifferentiated cells retained in monolayer culture. We conclude that nucleostemin is a marker of undifferentiated human adult bone marrow stem cells and that it is involved in the regulation of proliferation of these cells.
Stem cells are present throughout embryonic development as well as in several organs of the adult . They constitute a pool of undifferentiated cells with the remarkable ability to perpetuate through self-renewal while also retaining the potential to terminally differentiate into various mature cell types . Bone marrow stromal cells (BMSCs) can be easily isolated from adult marrow and contain a population of pluripotent progenitors that can give rise to mesenchymal lineages including chondrocytes, osteoblasts, fibroblasts, and adipocytes . It is probable, however, that true mesenchymal stem cells represent a rare subpopulation of BMSCs [3–5]. These cells are capable of dividing many times while retaining their ability to differentiate into various lineages with more restricted developmental potentials .
Isolation and characterization of human mesenchymal stem cells has been attempted by several investigators using a variety of techniques including plastic adherence , positive selection using antibody , selection based on adherence to fibronectin in low serum conditions , and in vitro growth under low oxygen tension in the presence of hematopoietic cells . A number of cell-surface markers have been identified on these cells, including CD105 , CD49a , and STRO-1 . However, because these markers are not unique to mesenchymal stem cells, their functional capacity to self-renew and to differentiate into multiple lineages remains the only true marker. Many studies have addressed the topic of adult stem cell differentiation; however, relatively little is known about their capacity to self-renew . Whereas these cells in vivo have a life-long ability to self-renew, during in vitro culture they eventually undergo replicative senescence after extensive expansion . This could be a result of reduced or absent telomerase activity in these cells [13–15].
Recent reports suggest that some cytokines can enhance the proliferation and the differentiation potential of BMSCs [13, 16, 17]. Exposure of human BMSCs to fibroblast growth factor (FGF)-2 during mitotic expansion increases cell yield and shortens time in culture . FGF-2 was found to enrich BMSC primary cultures for early mesenchymal progenitor cells having longer telomeres and extended life spans and differentiation potentials . The mechanism(s) by which FGF-2 modulates its proliferative activity in BMSCs remains unclear.
The identification of stem cell–specific proteins and the elucidation of their novel regulatory pathways may help in the development of protocols for control of their self-renewal and differentiation . Nucleostemin is a newly discovered nucleolar protein present in both embryonic and adult rat central nervous system stem cells, and several human cancer cell lines . This protein is abundantly expressed while the cells are proliferating in an early, multipotential state, but it abruptly and almost entirely disappears at the start of differentiation. For example, nucleostemin expression was found to be high in adult rat bone-marrow–derived hematopoietic stem cells but not in B lymphocytes or granulocytes. This phenomenon suggests a role in maintaining stem cell self-renewal. Although the exact function of nucleostemin remains to be determined, it is believed to bind to the p53 protein, thereby regulating the proliferation of both stem cells and some types of cancer cells [21, 22].
The aim of the present study was to determine if nucleostemin is synthesized by adult human BMSCs and to analyze its expression during the expansion and differentiation of the bone marrow stem cell population.
Materials and Methods
Isolation and Expansion of BMSCs
BMSCs were harvested from bone marrow plugs taken from the femoral heads of five patients (two men, three women; mean age, 53.8 years; range, 31–79) undergoing hip replacement. All patients gave their informed consent and the study was carried out according to local ethical guidelines. Cells were suspended in stem cell medium consisting of low glucose Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT, http://www.hyclone.com), 1% (vol/vol) Glutamax (1×; Invitrogen, Carlsbad, CA, http://www.invitrogen.com), and 1% (vol/vol) penicillin (100 U/ml)/streptomycin (100 μg/ml) (P/S; Invitrogen). The serum batch was selected to promote the growth and differentiation of mesenchymal stem cells . The cell suspension was separated from any bone in the sample by repeated washing with media, the suspension being drawn off using a 20-ml syringe and 19-gauge needle. The cells were centrifuged at 1,500 rpm for 5 minutes, and the supernatant/fat was removed. The resulting cell pellet was resuspended in medium and then plated at a seeding density of 5–10 million cells per 75-cm2 flask. These flasks were incubated at 37°C in a humidified atmosphere of 5% CO2 and 95% air. Four days were allowed before the first medium change, and then the medium was changed every other day until adherent cells reached 90% confluence and were ready for passaging.
All the cells used for multilineage differentiation assays were expanded as described above without FGF-2. For osteogenic differentiation , cells were plated at 1,000 cells/cm2 in 2-cm2 wells and grown to 50%–70% confluency. They were then incubated in osteogenic medium containing 10−8 M dexamethasone, 0.2 mM ascorbic acid, and 10 mM β-glycerolphosphate (Sigma-Aldrich). The medium was replaced every 3–4 days for 21 days. Cultures were washed with PBS, fixed in a solution of ice-cold 70% ethanol for 1 hour, and stained for 10 minutes with 1 ml of 40 mM Alizarin Red (pH 4.1; Sigma-Aldrich). Alternatively, fixed cells were immunostained using osteocalcin antibody (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) as described under Immunocytochemistry below. For adipogenic differentiation , 50%–70% confluent cultures were incubated in complete medium supplemented with 0.5 μM hydrocortisone, 0.5 mM isobutyl-methylxanthine, and 60 μM indomethacin, all from Sigma-Aldrich. The medium was replaced every 3–4 days for 21 days. Cells were washed with PBS, fixed in 10% formalin for 10 minutes, and stained for 15 minutes with fresh Oil Red-O solution (Sigma-Aldrich). Alternatively, fixed cells were immunostained using fatty acid binding protein-4 (FABP-4; R&D Systems) antibody as described under Immunocytochemistry below. For chondrogenic differentiation, the pellet culture system by Johnstone et al.  was used with minor modifications. Five hundred thousand cells were centrifuged in a 15-ml polypropylene tube forming a pellet that was cultured in chondrogenic media consisting of high-glucose DMEM (Sigma-Aldrich) supplemented with 10 ng/ml of transforming growth factor β3 (TGF-β3), 0.1 μM dexamethasone, 50 μg/ml ascorbate-2-phosphate, and 50 mg/ml ITS Premix (6.25 μg/ml insulin, 6.25 μg/ml transferrin, 6.25 ng/ml selenious acid, 1.24 mg/ml bovine serum albumin (BSA), 5.35 mg/ml linoleic acid; Invitrogen). The medium was replaced every 2–3 days for 21 days. Cartilage formation was quantified by measuring the major components of hyaline cartilage matrix: type II collagen and proteoglycans as glycosaminoglycans using well established assays [26, 27]. Parallel experiments using pellets of differentiated mature articular chondrocytes were treated under the same conditions acting as a control.
Analysis of Surface Epitope by Fluorescence-Activated Cell Sorting
The cells were suspended in PBS at a concentration of about 100,000 cells/ml, fixed in 4% (w/v) paraformaldehyde at 4°C for 10 minutes, and washed with PBS. Nonspecific antigens were blocked by incubating the cells at room temperature for 1 hour in a blocking solution containing 1% (wt/vol) BSA, 5% (vol/vol) FCS (Sigma-Aldrich), and 10% (vol/vol) human serum (Sigma-Aldrich). The cells were washed by centrifugation in three volumes of PBS, and the cell pellet was suspended in 100 μl of a primary antibody solution containing 20–100 μg/ml of antibody in blocking solution. After incubation for 40 minutes at 4°C, the cells were washed in PBS. All the primary antibodies were mouse anti-human, obtained from R&D Systems. For an isotype control, non-specific mouse IgG (Sigma-Aldrich) was substituted for the primary antibody. For antibodies that required a second antibody for detection, the cell pellet was incubated under the same conditions for 20 minutes with anti-mouse IgG labeled with fluorescein isothiocyanate (FITC; DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com). The cells were then washed in PBS and suspended in 1 ml of PBS for analysis on a FACSCalibur flow cytometer (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). Positive expression was defined as the level of fluorescence greater than 95% of corresponding isotype-matched control antibodies.
Cells from passages 2 or 3 in culture were used for cytospins. The slides were left to air-dry then were washed three times in wash solution consisting of 0.5% (vol/vol) Tween and 0.1% (wt/vol) BSA in PBS. Cells were fixed with ice-cold methanol for 10 minutes. The cells were blocked and permeablized for 45 minutes at room temperature using a blocking solution consisting of 1% (w/v) BSA, 10% (vol/vol) donkey serum, and 0.2% (vol/vol) Saponin (Sigma-Aldrich) in PBS. The slides were washed three times then incubated with the primary antibody solution overnight at 4°C. The antibodies used were goat anti-human for nucleostemin (R&D Systems), mouse anti-human for nucleoli (Chemicon, Temecula, CA, http://www.chemicon.com), and goat or mouse IgG for isotype control (Sigma-Aldrich). The slides were washed three times then incubated with Rhodamine Red-conjugated secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com) for 1 hour at 37°C. The cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) stain (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com) then mounted. Nucleostemin-positive cells were counted in samples from five patients by two independent investigators spanning fields of more than 100 cells for each sample.
Cell Proliferation Analysis
Cells from passages 2 or 3 were seeded at a density of 1,000 cells/well into 96-well plates (in triplicate) in expansion medium containing 2% (vol/vol) FBS. Cells were treated with various concentrations of FGF-2 as indicated in Results for 72 hours. Control wells containing 2% or 10% (vol/vol) FBS were used. Cell proliferation was assessed using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (inner salt) (MTS) assays performed according to the manufacturer's instructions (CellTiter 96 AQueous One Solution Cell Proliferation Assay; Promega, Madison, WI, http://www.promega.com). A linear standard curve with a range of 325–12,500 cells was included on every plate. Cells from parallel plate cultures were harvested for nucleostemin mRNA analysis as explained under RNA Isolation and Reverse Transcription below. Each experiment was performed in triplicate and repeated at least three times.
siRNA Knockdown Analysis
siRNA (50 pmol) was introduced into cells by complexing with Lipofectamine 2000 transfection reagent (Invitrogen). The transfected cultures were incubated with FGF-2 (10 ng/ml) for 72 hours then the cells were analyzed for proliferation using the MTS assay (Promega) or harvested for nucleostemin mRNA analysis using real-time reverse transcription-polymerase chain reaction (RT-PCR) analysis. Transfection efficiency and cell viability were optimized using BLOCK-iT Fluorescent Oligos (Invitrogen). The resultant transfection efficiency and cell viability were both > 90%. Stealth Select RNAi sequence (Invitrogen) specific for the three variants of human nucleostemin was used in the experiments. The sequence was as follows: human nucleostemin-specific siRNA, ACAACUUGCAUGCUCCUUG-UAAGCC; Stealth RNAi Med GC sequence was used as negative control.
RNA Isolation and RT
RNA was extracted from cell cultures using the SV Total RNA Purification kit (Promega), according to the manufacturers instructions. RT was carried out using the Superscript II system (Invitrogen). Total RNA (2 μg) was reverse transcribed in a 20-μl reaction volume containing Superscript II (0.2 U), random primer (225 ng/μl), dNTP (0.5 mM each), and RNasin ribonuclease inhibitor (2 U/μl) at 37°C for 42 minutes.
The coding sequence for human nucleostemin gene (GenBank accession no. NM_014366) was used to design nucleostemin primers using the online software, Primer3 (Whitehead Institute for Biomedical Research, Massachusetts Institute of Technology, Cambridge, MA). The primers span intronic junctions to avoid the amplification of genomic sequences. They were also checked for the amplification of potential pseudogenes. A BLAST search against all known sequences confirmed specificity. Published primers for the housekeeping gene β-actin  were used as a reference in all RT-PCR reactions and as a means of quantifying levels of nucleostemin expression by normalization to β-actin. These primers had been specifically designed to not coamplify processed β-actin pseudogenes in contaminating genomic DNA. All the primers generated the correct sizes of the PCR fragments with no nonspecific products, confirming the specificity of the real time RT-PCR (data not shown). The primer sequences for nucleostemin were as follows: forward primer, 5′-GGGAAGATAACCAAGCGTGTG-3′; reverse primer, 5′-CCTCCAAGAAGTTTCCAAAGG-3′; product size, 98 bp. The primer sequences for β-actin were as follows: forward primer, 5′-GACAGGATGCAGAAGGAGATTACT-3′; reverse primer, 5′-TGATCCACATCTGC TGGAAGGT-3′; product size, 141 bp.
Quantitative Real-Time RT-PCR
Quantitative real-time RT-PCR was performed in a 25-μl reaction volume containing 12.5 μl of the SYBR Green PCR master mix (Takara, Otsu, Japan, http://www.takara.co.jp), 5 μl of the RT reaction mixture, and 300 nM primers using the Smart Cycler II System (Cepheid, Sunnyvale, CA, http://www.cepheid.com). For the β-actin gene, the RT reaction mixture was diluted 100 times. The amplification program consisted of initial denaturation of 95°C (2 minutes) followed by 40 cycles of 95°C (15 seconds), annealing at 58°C (30 seconds), and extension at 72°C (15 seconds). After amplification, melt analysis was performed by heating the reaction mixture from 60–95°C at a rate of 0.2°C per second. The cycle threshold (Ct) value for each gene (X) of interest was measured for each RT sample. The Ct value for β-actin was used as an endogenous reference for normalization. The relative transcript level of a given gene (X) at a given treatment time point (T) over its value at the initial treatment time (0) was calculated as 2−ΔΔCt, in which ΔΔCt = CtT − Ct0; CtT = CtT,X − CtT and Ct0 = Ct0,X − Ct0,β. Real time RT-PCR assays were done in duplicate or triplicate and repeated two to four times.
Phenotype and Multilineage Potential of Isolated BMSCs
Cells were positively selected for their ability to adhere to plastic substrate. Adherent cells from passages 2 or 3 were analyzed by fluorescence-activated cell sorting (FACS) to assess the preparation quality and the phenotypic cell surface markers associated with stem cells. The cells resulting after plating were positive for CD105, CD49a, CD117, bone morphogenetic protein receptor (BMPR)-1A, STRO-1, and vascular cell adhesion molecule (VCAM)-1A. The population was negative for CD34. The median value and range of each marker are shown in Table 1. To assess the multipotentiality of the prepared BMSCs, they were assayed for their adipogenic, chondrogenic, and osteogenic capacities. All the samples differentiated into each of the three phenotypes (Fig. 1, n = 5). The addition of chondrogenic medium to pellet cultures resulted in tissue rich in proteoglycans (Fig. 1A) and type II collagen (Fig. 1B). The quality of the resulting pellets was comparable with that of control pellets grown using mature articular chondrocytes. When the undifferentiated BMSCs were incubated in adipogenic medium, the cells contained Oil Red-O–staining vacuoles and expressed the fatty acid binding protein typical of adipocytes (Fig. 1C, 1D). Similarly, the incubation of undifferentiated BMSCs with osteogenic medium resulted in cultures rich in mineral and expressing osteocalcin, typical of osteoblasts (Fig. 1E, 1F). There was no significant variation in the extent of differentiation among the patient samples, indicating that the BMSC population used was consistently multipotent.
Nucleolar Localization of Nucleostemin in BMSCs
We investigated if the human BMSC population contains nucleostemin-positive cells. Using immunofluorescence localization, there was a strong signal for nucleostemin protein in the nucleoli of adult human BMSCs (Fig. 2). Some cells showed a more diffuse nucleoplasmic distribution. The mean frequency of BMSCs expressing nucleostemin was 43.4% (range, 26%–55%; n = 5). The analysis was carried out on passage 2 or later of these cultures to avoid any potential contamination by hematopoietic stem cells.
Expression of Nucleostemin mRNA in Expanded Adult Human BMSCs
In order to analyze nucleostemin mRNA expression in BMSCs, we have devised a sensitive real-time RT-PCR method for quantifying the nucleostemin message in human cells. The lowest detectable template concentration was 8.5 pg/ml corresponding to a threshold cycle Ct of 36.6, and the highest detectable template concentration was 85 ng/ml corresponding to a Ct of 23.5 (data not shown). In order to estimate the stability of nucleostemin expression in BMSCs over time, RNA was quantified in expanded cells from passages 1–3 over a period of 5 weeks. Relative quantitative RT-PCR at each passage revealed no significant change in the expression of nucleostemin over time (Table 2).
Regulation of Nucleostemin Expression in BMSCs by FGF-2
FGF-2 increases the proliferation of BMSCs in vitro . To determine whether nucleostemin expression is influenced by FGF-2 stimulation, BMSCs were cultured in the presence of varying concentrations of FGF-2. The results show a dose-response relationship between nucleostemin expression and FGF-2 concentration (Fig. 3). The use of FGF-2 at 10 ng/ml led to a high and significant greater nucleostemin mRNA expression than with lower concentrations of the growth factor. To assess the relationship of nucleostemin expression with FGF-2, cells were transfected with a small inhibitory RNA sequence specific to nucleostemin prior to FGF-2 stimulation. This sequence was shown to downregulate nucleostemin expression in FGF-2–stimulated and unstimulated control cells by 73% and 62%, respectively, relative to cells transfected with scrambled siRNA sequence (Fig. 4A). This tool allowed for assessing the functional relationship of FGF-2–stimulated proliferation and nucleostemin expression in BMSCs. The introduction of siRNA in cultures of FGF-2–stimulated proliferating cells resulted in a significant lower cell number compared with control cells transfected with the scrambled siRNA sequence (Fig. 4B). On the other hand, nucleostemin-specific siRNA introduced into control cultures incubated with 2% (vol/vol) FBS without FGF-2 did not influence their level of proliferation and was comparable with the scrambled siRNA control sequence. These experiments demonstrate that the specific knockdown of nucleostemin mRNA results in lower FGF-2-driven proliferation of BMSCs.
Differential Expression of Nucleostemin Before and After BMSC Differentiation
Nucleostemin has been shown to be downregulated in mature and terminally differentiated cells compared with their precursor central nervous system stem cells . To investigate the expression profile of nucleostemin before and after differentiation in BMSCs, these cells were differentiated into chondrocytes, adipocytes, and osteocytes using well established methods. RNA extracted from the differentiated cells was used to quantify levels of nucleostemin using real-time RT-PCR. The results show a significantly lower nucleostemin expression level in the differentiating cells than in undifferentiated BMSCs expanded in monolayer (Fig. 5).
We have shown, for the first time, that nucleostemin is persistently expressed in adult human BMSCs, that this expression is lost upon differentiation in vitro, and that nucleostemin upregulation by FGF-2 is required for this growth factor to increase stem cell proliferation.
We used a population of BMSCs positively selected for its adherence to plastic and its growth in a selected batch of FBS known to enrich and maintain multipotential stem cells in the bone marrow . That population was relatively rich in CD105 and CD49a and contained variable levels of other cell surface markers associated with multipotential stem cell populations. More importantly, the population had the capacity to differentiate, in all samples studied, into three mesenchymal lineages with high efficiency. Taken together, these results demonstrate the enrichment for a self-renewing, functional stem cell population in the bone marrow with minimal specific lineage commitment.
Baddoo et al.  recently demonstrated nucleostemin expression in murine mesenchymal stem cells; however, there have been no previous studies of adult human BMSCs. Our results clearly demonstrate, for the first time, the presence of nucleostemin in the nucleoli of expanding BMSCs (Fig. 2), consistent with reports of nucleostemin as a nucleolar protein [20, 30]. The localization varied from being nucleolar to being diffuse nucleoplasmic, possibly reflecting the stage of the cell cycle [20, 30]. The frequency of cells expressing nucleostemin protein was relatively high compared with stem cell surface markers investigated in the cell population studied. Published reports estimate that the most primitive stem cell occurs at a frequency of 1:10,000 [24, 31] to 1:100,000  in whole bone marrow stromal cell preparations. Whereas it is likely that these rare primitive cells express nucleostemin at high levels , it is clear that nucleostemin expression in the population studied is associated with a more abundant cell population, more likely to be committed progenitors with multipotential.
The expression of nucleostemin in human BMSCs was persistent over a long period of time in culture. Whether this level of expression continues throughout the late life span of these cells remains to be investigated. Previous studies suggested that controlled high levels of this protein would contribute to the continued proliferation and self-renewal of stem cells and cancer cells [20, 32, 33]. The relatively high expression of nucleostemin in human BMSCs may reflect the presence of an actively proliferating, self-renewing subpopulation in the bone marrow.
Our results demonstrate that the effect of FGF-2 on the expression of nucleostemin in BMSCs is dose dependent. siRNA studies confirmed that this relationship is direct and functional because inhibiting nucleostemin expression results in neutralizing the proliferative effect of FGF-2. A recent study revealed the lack of a clear gene-expression pattern that correlates with the marked mitogenic effect of FGF-2 in BMSCs . In that study, a large percentage of known regulators, both positive and negative, of cell proliferation were downregulated despite an overall effect of an increased proliferation rate. Although the mechanism by which nucleostemin itself controls the cell cycle is unclear, its regulation by FGF-2 is extremely interesting, and the mechanism involving this pathway requires further elucidation. FGF-2 is known to enrich BMSC primary cultures for primitive mesenchymal progenitor cells that have longer telomeres and an extended life span and differentiation potential . Given the vital role nucleostemin plays in stem cell proliferation induced by FGF-2, it will be important to determine if this growth factor increases the frequency of nucleostemin-positive cells.
The use of well established differentiation methods provided us with the means to determine if nucleostemin is specifically associated with the undifferentiated phenotype of BMSCs. The results clearly show the significant loss of nucleostemin expression upon differentiation in vitro. This is in agreement with studies of rodent stem cells, showing the downregulation of nucleostemin in mature and terminally differentiated cells compared with their precursor neural stem cells . We have used quantitative mRNA analysis to establish an accurate evaluation of nucleostemin expression during differentiation. In the future, a more detailed study of nucleostemin protein in differentiated cells may be required to compare its stability and function with that in undifferentiated stem cells.
The molecular program(s) regulating the transition of stem cells between the actively dividing and differentiated states remain unclear. A step toward elucidating this program involves the understanding of signals that control the activity of nucleostemin. TGF-β, which we used to drive the differentiation of stem cells into chondrocytes in our chondrogenesis model, has been previously shown to selectively inhibit the growth of stem cells . Our data therefore suggest that the growth-arresting properties of TGF-β may be mediated through down-regulation of nucleostemin either directly or through downstream molecules in the TGF-β signaling pathway. Although it is not clear how adipogenic and osteogenic differentiation leads to nucleostemin inhibition, this multi-tissue loss of expression suggests an integral role for nucleostemin in BMSC proliferation.
Table Table 1.. Phenotypic cell surface markers of BMSCs
Table Table 2.. Expression of nucleostemin mRNA in bone marrow stromal cells
This work was funded by grants from the UK Biotechnology and Biological Sciences Research Council and the UK Arthritis Research Campaign (ARC). A.P.H. is funded in part by an endowed chair from the ARC.
The authors indicate no potential conflicts of interest.