T cell subsets differently regulate osteogenicdifferentiation of human mesenchymal stromalcells in vitroFrancesco Grassi1*, Luca Cattini2, Laura Gambari2, Cristina Manferdini2, Anna Piacentini2,Elena Gabusi1, Andrea Facchini2 and Gina Lisignoli21Laboratorio RAMSES, Istituto Ortopedico Rizzoli, Bologna, Italy2S. C. Laboratorio di Immunoreumatologia e Rigenerazione Tissutale, Istituto Ortopedico Rizzoli, Bologna, Italy
Abstract
T lymphocytes play a key role in the regulation of bone homeostasis and bone healing. The inflammatoryresponse at the site of bone injury is essential to the initiation of the bone repair program; however, anuncontrolled exposure to inflammatory environment has a negative effect on tissue regeneration –
Received 11 June 2012; Revised 18 October 2012; Accepted 24 January 2013
Keywords T cells; CD4; PCR array; mesenchymal stromal cells; bone regeneration
1. Introduction
Bone is a dynamic tissue with a high capacity for self-repair. However, failure or delay in the regeneration offractured bone is still a relevant clinical problem affectingup to 10% of patients (Einhorn, 1995). Upon injury or dis-ease, bone commences a repair programme that startswith the formation of haematoma and inflammation,
followed by the development of a callus and the begin-ning of the remodelling of the newly formed bone tissue.Each of these stages is controlled by a complex network ofbiological players. Among these, the interplay betweenbone tissue and the host immune system has been increas-ingly recognized as a key regulator of the outcome of thebone regeneration process.
Lymphocytes are critical regulators of physiologicalbone turnover and peak bone mass (Li et al., 2007); inpathological conditions, they are the essential source ofthe inflammatory cytokines which induce bone erosion(Kong et al., 1999; Weitzmann and Pacifici, 2007).The role of lymphocytes in wound healing has long beenappreciated (Schaffer and Barbul, 1998), but their
*Correspondence to: F. Grassi, Laboratorio di ImmunoreumatologiaeRigenerazioneTissutale, IstitutoOrtopedicoRizzoli, Via di Barbiano1/10, 40136 Bologna, Italy. E-mail: [email protected]
JOURNAL OF TISSUE ENGINEERING AND REGENERATIVE MEDICINE RESEARCH ARTICLEJ Tissue Eng Regen Med (2013)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/term.1727
function in the bone repair process has only recently beeninvestigated.
A series of studies have clarified the spatial and tempo-ral pattern of inflammatory response at the site of boneinjury; based on the expression of some of the inflamma-tory mediators, it was established that the acute inflamma-tory phase peaks within 24 h of injury and fades within 7days (Cho et al., 2002, 2007). T lymphocytes are quicklyrecruited at the site of damage and in the early stages pre-dominantly show an activated phenotype with a high rateof cytotoxic T cells (Schmidt-Bleek et al., 2009). While theinflammatory response was shown to exert an essential roleas a primer of tissue regeneration (Gerstenfeld et al., 2003;Yang et al., 2007), an uncontrolled exposure to the inflam-matory environment has a detrimental effect on bonerepair. These observations raised the question of whetherT cells ultimately stimulate or delay the process of boneregeneration. In a recent pivotal paper, Liu et al. (2012)showed that pro-inflammatory T cells inhibit bone regener-ation byMSCs inmice by amechanism involving a synergis-tic role of TNFa and IFNg; however, the Treg subset wasfound to promote bone mineralization. In Rag1�/� mice,where both T and B cells are lacking, fracture healing wasaccelerated compared to wild-type mice, suggesting thatlymphocytes as a whole could hamper bone regeneration(Toben et al., 2011). On the other hand, evidence fromwound healing models revealed a delayed tissue repair inthe absence of T cells and a differential role for CD4 andCD8 in promoting tissue regeneration (Schaffer and Barbul,1998; Schaffer et al., 2007).
Despite recent progress, a detailed understanding ofthe potential of human T cell populations to stimulateosteogenic differentiation of osteoprogenitor cell is stillelusive and further investigation can help understandingnot only the role of these cells (either autologous or alloge-neic) in the bone repair process, but also the pathogenesisof the skeletal diseases in which homeostatic mechanismscontrolling bone turnover are impaired.
To gain further insight into the role of the two main Tcell subsets, CD4 and CD8, in the process of new boneformation, we compared these populations for theexpression of an array of osteogenic genes using a PCR-array approach, and followed-up gene expression datawith functional mineralization assays using T cell-conditioned media and human MSCs. To focus onthe post-acute inflammatory stage, we isolated restingT cell subsets expressing low or undetectable levelsof TNFa and IFNg. Our data show that CD4 and CD8T cells regulate the differentiation of osteoprogenitorcells differently.
2. Materials and methods
2.1. Isolation and characterization of MSCs
Human MSCs were obtained as previously described(Grassi et al., 2011; Lisignoli et al., 2009). Briefly, MSCs
were obtained from 5 ml bone marrow aspirate duringhip surgery of four post-traumatic patient donors. Thecells were washed twice, resuspended in Dulbecco’s mod-ified Eagle’s medium (D-MEM; Gibco, Life Technologies,Milan, Italy) with low glucose and 15% fetal bovine se-rum (FBS; Lonza, Basel, Switzerland), counted andplated at a concentration of 2�106 cells/T150 flask.After 1 week, non-adherent cells were removed andthe adherent h-MSCs expanded in vitro. MSCs wereused at passage 3 in culture. MSCs were characterizedfor the expression of common markers of the MSCphenotype (CD3, CD14, CD31, CD34, CD45, CD90, CD73,CD105, CD146), as detailed elsewhere (Torreggiani et al.,2012). MSCs from four different donors were used inthis study. A written consensus form was obtained fromthe donors according to the rules and regulation of theHospital Ethical Committee.
2.2. Isolation and characterization of T cells
T cells were isolated from the buffy coat obtained fromfour healthy donors. Peripheral Blood Mononuclear Cells(PBMCs) were obtained using Ficoll–Hypaque andwashed twice in phosphate-buffered saline (PBS); the Tcells were isolated through immunomagnetic negative se-lection for Pan T, CD4 Tand CD8 Tcells, using commercialkits (Pan T cells Isolation Kit II; CD4 T cells Isolation Kit;CD8 T cells Isolation Kit, Miltenyi Biotec), following themanufacturer’s instructions. At the end of the isolation,T cell phenotype and purity were controlled by flow cy-tometry with PE-Cy5-labelled anti-CD4 and PE-labelledanti-CD8 antibodies (BD Biosciences, San Jose, CA,USA). TNFa and IFNg production by T cells was assessedby flow cytometry and the secreted cytokines were mea-sured in the cell culture supernatants by ELISA. For flowcytometry, T cell subsets were washed once and left for5 h at 37�C in the presence of Golgi Plug (BD Bioscience).The cells were then labelled with anti-CD4 and anti-CD8(BD Bioscience) and intracellularly stained for TNFa, IFNgor isotype control; the cells were fixed and permeabilizedusing a Cytofix-Cytoperm Kit (BD Bioscience), followingthe manufacturer’s instructions. The cells were analysedby flow cytometry. For the quantification of secretedcytokines, cells were seeded at a density of 6 � 106/cm2
and the conditioned media harvested after 48 h. TNFaand IFNg were measured by Quantikine ELISA Kits(R&D Systems, Minneapolis, MN, USA), following themanufacturer’s instructions. Minimum detectable doseswere 1.6 pg/ml for TNFa and 8 pg/ml for IFNg.
2.3. RNA extraction and PCR array analysis
Total cellular RNA was isolated from T cell subsets andMSCs using the RNeasy Mini Kit (Qiagen, Milano, Italy),according to the manufacturer’s instructions. Contamina-tion of genomic DNA was removed from total RNAsamples by treating them with DNase I (DNA-free Kit,
Ambion, Austin, TX, USA); the samples were thenanalysed for purity and quantitated spectrophotometrically.
Complementary DNA (cDNA) synthesis was performedfrom 0.8 mg RNA with the RT2 PCR array First Strand Kitas a part of the Human Osteogenesis RT2 profiler PCRArray Kit (SABioscience, Qiagen, Milan, Italy). The PCRarray is a set of optimized real-time PCR primer assaysfor pathway-focused genes as well as appropriate RNAquality controls. The PCR array allows gene expressionprofiling of a large set of genes with the sensitivity of areal-time PCR analysis.
Amplification was carried out on a Rotor Gene ThermalCycler (Corbett Research, Qiagen) equipped with a 100-well rotor, under cycling and thermal conditionssuggested by the manufacturer. A set of five housekeepinggenes were included in the amplification: b2-microglobulin,hypoxanthine phosphoribosyltransferase 1, ribosomal pro-tein L13a, glyceraldehyde-3-phosphate dehydrogenaseand b-actin. The data were analysed using the web-basedRT2 PCR array Data Analysis tool provided by the manufac-turer. For each gene of interest, mRNA levels were normal-ized to the average CT values of the five housekeepinggenes. Data were either represented as a cluster analysisor expressed as fold difference values of CD4 T cells com-pared to CD8 T cells.
2.4. Conditioned media and osteogenesis
Conditioned media (CM) from Pan T, CD8 and CD4 Tcells were collected by seeding the cells at the densityof 6 � 106/cm2 in 96-well plates in RPMI 1640 mediumcontaining 10% FBS; cells were harvested after 48 h,centrifuged at 500 � g for 5 min and the supernatantwas collected and stored at �80�C prior to use in func-tional assays. Osteogenesis of MSCs was induced by incu-bating cells in CM obtained from T cell subgroups mixedwith MSCs medium at the 1:1 ratio; the resultingmedium was supplemented with 100 mM ascorbic acid,2 mM b-glycerophosphate and 100 nM dexamethasone(all from Sigma). Culture medium including CM wasreplaced twice a week. MSCs were cultured for 2 weeksprior to mRNA analysis for markers of osteogenic differ-entiation. Mineralized matrix was evaluated by alizarinred S staining (AR-S; Sigma), as previously described(Gabusi et al., 2011).
2.5. Real-time PCR analysis ofosteogenic markers
Total RNA was extracted from human MSCs grown for 2weeks as described above, using TRIZOL reagent(Invitrogen, Carlsbad, CA, USA) according to themanufac-turer’s instructions, and then treated with DNase I(DNAfree Kit, Ambion, Austin, TX, USA). Reverse tran-scription was performed using M-MLV reverse transcrip-tase and random hexamers, following the manufacturer’sprotocol (Perkin-Elmer, Norwalk, CT, USA). Primers for
PCR amplification of alkaline phosphatase (ALP), bonesialoprotein (BSP), collagen type I (COL I), Runt-relatedtranscription factor 2 (RUNX-2), collagen type XV (COLXV) and osteocalcin (OC) were as previously reported(Lisignoli et al., 2009; Manferdini et al., 2011; Tonnarelliet al., 2009). Real-time PCR was run in a LightCyclerInstrument (Roche Molecular Biochemicals, Mannheim,Germany), using the SYBR Premix Ex Taq (TaKaRaBiomedicals, Tokyo, Japan) with the following protocol:initial activation of HotStar Taq DNA polymerase at 95�Cfor 10 min; 45 cycles of 95�C for 5 s; and 60�C for 20 s.For each target gene, mRNA levels were calculated,normalized to GAPDH according to the formula 2–ΔΔCT
and expressed as fold increase compared to the expressionat day 0 (= 1).
2.6. Statistical analysis
Data were analysed by one-way ANOVA and Tukey multi-ple comparison tests. Simple comparisons were made byusing a two-tailed unpaired Student’s t-test; p< 0.05was considered statistically significant.
3. Results
3.1. T cell isolation and characterization
T cell subsets isolated by immunomagnetic negative sec-tion were first characterized for CD4 and CD8 expressionto assess phenotype (Figure 1A). The CD4 and CD8 T cellsubsets showed, respectively, an average purity of 96.6%and 94.3%. Representative dot plots for each of the sub-sets are shown in Figure 1A. The activation state of thecells was also assessed; as T cell production of TNFa andIFNg was recently shown to be critically linked to the reg-ulation of osteogenesis, we measured intracellular pro-duction of these pro-inflammatory cytokines. As shownin Figure 1B, levels of TNFa and IFNg were low orundetectable in both subsets; in particular, on average TNFawas detected in 12.1% and 5.8%, respectively, of CD4 andCD8 T cells, while IFNg was undetectable in CD4 T cellsand was produced by an average 11% of the CD8 T cells.The concentrations of TNFa and IFNg secreted in the condi-tioned medium were also assessed by ELISA. In particular,TNFa was below the detectable level (1.6 pg/ml) for bothCD4 and CD8 T cells after 48 h in culture; the concentra-tions of IFNg were 11.23� 0.95 and 13.98� 1.05 pg/ml(mean� SD), respectively. Furthermore, none of thecell populations analysed showed detectable prolifera-tion after 48 h in culture (data not shown).
3.2. PCR array analysis of osteogenesis
Samples were screened for the expression of 84 geneslinked to osteoblast differentiation, bone metabolism orbiosynthesis of mineral matrix. Human MSCs were
CD4 T cells support osteogenic differentiation of human MSCs
included in the PCR array screening as a control for con-stitutive expression of osteogenic genes. Data analysis,as shown in Figure 2, reveals that MSCs showed a highconstitutive expression of the vast majority of the genesconsidered in our investigation. A total of 28 genes werehighly expressed in Pan T cells, most of which wereco-expressed in the CD4 T cell subset (Figure 2).
Analysis of absolute expression levels in the CD4 andCD8 subsets revealed that 51 of 84 genes were expressedabove detectable levels (CT< 33 in each of the replicates).Table 1 shows the list of these genes and reports geneexpression as ΔΔCT values. Genes consistently expressedin these groups included members of the bone mor-phogenetic protein family (BMP1, BMP4 and BMP6),metalloproteases (MMP1, MMP2, MMP8, MMP9, andMMP10), collagens (Coll I, COLL III, COLL IV, COLL V,COLL X and COLL XV); moreover, CD4 and CD8 expressedseveral growth factors implicated in the regulation ofbone growth and homeostasis, such as TGFb1, TGFb3,IGF, FGF-1, FGF-2, VEFG-A and VEFG-B. While we did notobserve any gene that was uniquely expressed in eitherof the two subsets, expression levels were considerablydifferent between CD4 and CD8 T cells. A comparativeanalysis of gene expression in the CD4 and CD8 subsets
revealed that 27 genes showed expression levels abovethe two-fold increase cut-off in CD4 T cells compared toCD8, while 12 genes showed expression levels> two-fold greater in CD8 T cells (Figure 3).
Among these, 9 genes were expressed at statisticallysignificant higher levels in CD4 T cells as comparedto CD8, while only 3 genes were significantly moreexpressed in the CD8 group (Figure 4). Of interest, genesupregulated in CD4 T cells included BMP-4, the keyosteoinductive protein; another member of the BMP fam-ily, BMP-6, although expressed at higher levels in CD4 Tcells (Table 1), did not achieve statistical significancecompared to CD8 in our analysis.
Collectively, the PCR array dataset suggested that CD4and CD8 T cells hold a distinct osteogenic ‘signature’ sug-gestive of a greater potential for CD4 T cells to stimulateMSCs differentiation.
3.3. CD4 and CD8 T cells differently regulatedin vitro osteogenic differentiation of MSCs
To investigate whether the profile of expression of osteo-genic genes in T cell subsets corresponds to a functional
Figure 1. T cell phenotype: (A) phenotypical analysis of the T cells subsets obtained from negative immunomagnetic selection, show-ing percentage of positive cells for each group, Pan T, CD4 and CD8; (B) flow-cytometry analysis of TNFa and IFNg expression by Tcells in CD4 and CD8 T cells. Plots are representative of three independent experiments
capacity to stimulate differentiation of osteoprogenitorcells, we used conditioned media (CM) obtained fromeach of the three groups of T cells after 48 h in cul-ture. CM were combined with osteogenic medium ina 1:1 ratio and added to human MSCs in in vitromineralization assays.
As shown in Figure 5, when CM from the PanT subsetwas added to osteogenic medium, mineralization by MSCssignificantly increased; however, when CM obtained fromthe CD4 or CD8 T cell subset were employed in the assay,CD4-derived CM strongly upregulated mineralization,while CM from CD8 cultures did not significantly changemineralization obtained with control osteogenic medium.We further analysed osteogenic differentiation of MSCs byevaluating typical osteogenic markers in MSCs by real-time PCR after 2 weeks of culture. In keeping withevidence from mineralization assays, PanT-derived CMinduce a significant upregulation of ALP and BSP expres-sion in MSCs, as compared to control osteogenic medium(OST) (Figure 6); CD4-derived CM induced an overallstronger upregulation of osteogenic markers, as revealedby a significant increase in mRNA expression of RUNX-2,ALP, OC and BSP as compared to control medium. On theother hand, CD8-derived CM only induced a significantupregulation of COL XV in MSCs. COL I expression, whileincreased in the OST+CD4 group, did not achievestatistical significance compared to the control group.
Therefore, whole T cells not only express a set of osteo-genic genes involved at multiple levels in the regulation ofbone homeostasis, but also secrete soluble factors suffi-cient to upregulate MSCs differentiation toward the oste-oblastic phenotype in vitro. However, when the specificcontribution of CD4 and CD8 was investigated, we foundthat the CD4 subset induced a stronger andmore significant
upregulation of MSCs differentiation as measured by bothmRNA expression of osteogenic markers and functionalmineralization assay.
4. Discussion
To better define the mechanisms of cellular and molecularcrosstalk between immune cells and bone tissue is the keyobjective of the emerging field of osteoimmunology(Takayanagi, 2007). Based on previous findings in animalmodels, here we hypothesized that human T cells regulateosteogenic differentiation of human osteoprogenitor cellsand that different T cell subsets might play different rolesin this process. By using a PCR array approach, the pres-ent study demonstrates that the pattern of expression ofgenes involved in MSCs differentiation and mineralizationdiffers in CD4 and CD8 T cells. In keeping with geneexpression data, functional assays of MSCs differentiationconfirmed that soluble factors from CD4 T cells, but notCD8 T cells, stimulate osteogenic differentiation and min-eral deposition by MSCs.
T cells substantially contribute to the process of bonerepair (Mountziaris et al., 2011; Schmidt-Bleek et al.,2012); however, studies focusing on the inflammatorystage of fracture healing showed an overall inhibitory rolefor activated T cells in the osteogenic differentiation ofMSCs. In this study we explored the function of resting,unstimulated T cell subsets in an effort to further under-stand the mechanisms of the late-stage repair process. Tcell phenotype in our experiments is exemplified by a lowor undetectable intracellular protein level of TNFa and IFNg,the two cytokines that were reported to synergistically
Figure 2. Cluster analysis of PCR array data for osteogenic genes expression. Overview of 84 osteogenic gene expressions by Pan T,CD4 and CD8 T cells. Human MSCs are shown as a control. Each cell subset represents the average of three independent experiments
CD4 T cells support osteogenic differentiation of human MSCs
inhibit bone formation (Liu et al., 2012). Furthermore,when measured in the conditioned media of T cell sub-sets, these cytokines were either undetectable (TNFa)or found at concentrations about 1000-fold lower(IFNg) than those shown to inhibit MSCs differentia-tion in vitro (Liu et al., 2012).
Samples were screened for the expression of 84 genesinvolved in various stages of bone development and min-eralization. As expected, the vast majority of these genesare constitutively expressed at high levels in humanMSCs. The analysis of Pan T cell samples revealed that50 genes were expressed at detectable levels in each of
the three replicates we used in this study (Table 1). Whenthese genes were analysed separately in CD4 and CD8 Tcells, we found that CD4 T cells express higher mRNAlevels for most genes, displaying a stronger co-expressionprofile with the Pan T population as compared to theCD8 subset.
Collectively, gene expression data showed that unstimulatedT cells hold a strong potential to stimulate osteogenesis atmultiple levels and that CD4 and CD8 subsets are charac-terized by a very different osteogenic ‘signature’.
Among members of the BMP family proteins, BMP-4and BMP-6 showed the highest expression levels in
Table 1. List of genes expressed above detectable levels in CD4 and CD8 T cell subsets. Data represent the average of three inde-pendent experiments. Only genes consistently expressed in the three biological replicates were included in the table. Data areexpressed as ΔΔCT values calculted against the average CT of five different housekeeping genes.
DDCT
CD8 CD4
Annexin A5 5,521 4,60067Bone gamma-carboxyglutamate (gla) protein 8,271 7,39733Bone morphogenetic protein 1 9,566 9,52733Bone morphogenetic protein 4 12,951 10,6373Bone morphogenetic protein 6 11,686 10,4773Collagen, type X, alpha 1 11,286 7,824Collagen, type XV, alpha 1 13,386 15,7307Collagen, type I, alpha 1 14,676 13,549Collagen, type II, alpha 1 13,406 10,854Collagen, type IV, alpha 3 (Goodpasture antigen) 12,641 12,184Collagen, type V, alpha 1 11,921 11,619Cartilage oligomeric matrix protein 13,216 13,034Colony stimulating factor 2 (granulocyte-macrophage) 14,186 12,514Cathepsin K 7,591 6,879Dentin matrix acidic phosphoprotein 1 13,326 14,7773Fibroblast growth factor 2 (basic) 12,246 11,729Fibroblast growth factor receptor 1 10,956 9,884Fibroblast growth factor receptor 2 10,921 10,329Fms-related tyrosine kinase 1 13,161 11,689Growth differentiation factor 10 12,656 11,899Intercellular adhesion molecule 1 8,541 8,124Insulin-like growth factor 1 receptor 5,426 4,409Integrin, alpha 1 9,576 7,849Integrin, alpha 2 (CD49B, alpha 2 subunit of VLA-2 receptor) 14,6727 8,609Integrin, alpha 3 (antigen CD49C, alpha 3 subunit of VLA-3 receptor) 12,5627 7,959Integrin, beta 1 (fibronectin receptor, beta polypeptide, antigen CD29 includes MDF2, MSK12) 5,861 6,864Multiple inositol-polyphosphate phosphatase 1 10,9627 5,894Matrix metallopeptidase 10 (stromelysin 2) 19,696 12,284Matrix metallopeptidase 2 (gelatinase A, 72kDa gelatinase, 72kDa type IV collagenase) 19,696 13,009Matrix metallopeptidase 8 (neutrophil collagenase) 10,756 7,944Matrix metallopeptidase 9 (gelatinase B, 92kDa gelatinase, 92kDa type IV collagenase) 13,7627 8,799Nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 4,481 3,989Phosphate regulating endopeptidase homolog, X-linked 13,931 11,999Runt-related transcription factor 2 6,886 6,639Scavenger receptor class B, member 1 10,446 10,569Serpin peptidase inhibitor, clade H (heat shock protein 47), member 1, (collagen binding protein 1) 7,211 6,224SMAD family member 1 12,171 10,499SMAD family member 2 6,681 5,844SMAD family member 3 4,116 3,174SMAD family member 4 4,466 3,339Transforming growth factor, beta 1 4,131 3,334Transforming growth factor, beta 3 8,806 8,209Transforming growth factor, beta receptor 1 7,396 7,549Transforming growth factor, beta receptor II (70/80kDa) 7,92933 �2,281Tumor necrosis factor 7,396 5,819Tuftelin 1 10,721 8,714Twist homolog 1 (Drosophila) 12,856 18,464Vascular cell adhesion molecule 1 10,356 13,809Vitamin D (1,25- dihydroxyvitamin D3) receptor 16,096 8,389Vascular endothelial growth factor A 4,741 9,784Vascular endothelial growth factor B 3,276 2,459
T cell subsets. BMPs are among the key factors drivingosteogenic differentiation of MSCs (Deschaseaux et al.,
2009) and BMP-4 has been characterized as one ofthe most osteoinductive member of the BMP family(Cheng et al., 2003). In a mouse model of fracture repair,BMP-4 showed higher expression early after fracture, whileBMP-6 is believed to be highly expressed later in the repairprocess (Dimitriou et al., 2005).
Although BMP-4 is expressed in the stromal compart-ment of the thymus and plays a role in the early T celldevelopment (Tsai et al., 2003), to our knowledge this isthe first report to show BMP-4 expression by mature Tcells. In particular, CD4 T cells express significantly higherlevels of BMP-4 compared to CD8. One important linkbetween T cells, BMP-4 and bone healing is provided bythe pivotal work of Peng et al. (2002), who reported thatVEGF and BMP-4 synergistically increase bone healing inmice, and that the beneficial effects of VEGF on boneregeneration are strictly dependent upon BMP-4 concen-trations. Based on our results, it is conceivable that T cells,and particularly CD4 cells, contribute to the VEFG-mediatedincrease in bone healing, which includes enhanced angio-genesis and the recruitment of circulating cells at the siteof injury.
Not surprisingly, among genes highly expressed in Tcells are MMPs; we found MMP-2, MMP-8 and MMP-9to be predominantly expressed by CD4 T cells, butonly MMP-2 reached a statistically significant increasecompared to CD8. MMPs are involved in the remodellingphase of tissue repair, and MMP-2 and MMP-9 wererecently suggested to be early biomarkers of bonehealing in humans after surgical bone repair (Gallieraet al., 2010).
We found several members of the collagen protein fam-ily to be abundantly expressed in T cell subsets; these find-ings further support the role of T lymphocytes in tissueregeneration, as collagens were widely shown to play a
Figure 3. RT2 profiler PCR array of osteogenesis. Overview of scatter plot on expression of 84 genes. Green dots are genesupregulated and red dots are genes downregulated in CD4 T cells (group 1), as compared to control groups (CD8 T cells). The centralline indicates unchanged gene expression; boundaries represents the two-fold regulation cut-off
Figure 4. Semi-quantitative analysis of gene expression in T cellssubsets: (upper panel) genes significantly upregulated in CD4 Tcells are expressed as fold increase as compared to CD8 group(= 1); (lower panel) genes significantly upregulated in CD8 Tcells are expressed as fold increase as compared to CD4 group(= 1). For all genes shown, p<0.05
CD4 T cells support osteogenic differentiation of human MSCs
relevant role in haemostasis and platelet activation. Recep-tors for collagens include discoidin domain receptors,integrins, mannose receptor family members and glycopro-tein VI (Gp VI). Beyond their renowned role inmediating celladhesion, several collagen receptors are expressed in plate-lets (Clemetson and Clemetson, 2001), where they inducethe release of a broad repertoire of growth factors involvedin angiogenesis, recruitment of osteoprogenitor cells andosteoclast and osteoblast formation in bone (Nurden,2011). Overall, our data show that Tcells are an unexpectedsource of collagens and can contribute to the collagen-richenvironment of the bone marrow; however, further investi-gation will be needed to better clarify the physiological roleof T cell-derived collagens and the predominant expressionof Coll IA2, Coll II and Coll X in the CD4 subset.
Interestingly, the single gene we found most upregulatedin CD8 T cells compared to CD4 is TWIST-1, a member of
the basic helix–loop–helix (bHLH) family of proteins;expression of TWIST-1 has been described in MSCs ofseveral lineages, including bone cells (Lee et al., 1999;Spicer et al., 1996), where it was found to inhibitdifferentiation. In particular, in osteoblastic cells TWIST-1was shown to inhibit BMP signalling by interfering withtranscriptional activities mediated by Smad proteins,thereby inhibiting BMP-induced osteoblast differentiation(Hayashi et al., 2007). In T cells, expression of TWIST-1was demonstrated to decrease the expression of IFNg andTNFa, ameliorating inflammatory response and improvingimmunopathological diseases (Niesner et al., 2008).
Contrary to other tissues, the bone marrow microenvi-ronment harbours a higher percentage of CD8 T cells thanCD4 cells in physiological conditions (Di Rosa and Pabst,2005); however, in a sheep model of bone injury it hasbeen reported that a higher number of CD4+ T cells
Figure 5. Alizarin red S staining in human MSCs. MSCs were cultured for 14 days in the presence of osteogenic medium (OST) or amix of OST and PanT, CD4 or CD8 conditioned medium. (A) Representative pictures of each sample. (B) Alizarin red S staining wasanalysed by spectrophotometer at 510 nm wavelength. Results represent the average of three independent experiments; *p<0.05;**p<0.01
Figure 6. mRNA expression of the osteogenic proteins, RUNX-2, ALP, OC, BSP, COL XV and COL I. MSCs were cultured for 14 days in thepresence of osteogenic medium (OST) or a mix of OST and PanT, CD4 or CD8 conditioned medium. Data were obtained by real-timePCR and are expressed as fold increase compared to mRNA expression at day 0. Results represent the average of 3 independentexperiments; *p<0.05 compared to OST
accumulate at the fracture site 4 h after injury as com-pared to areas of muscle–haematoma (Schmidt-Bleeket al., 2009). The differential composition in immunecell subsets in the two models of injury could accountfor the ‘priming’ effect of inflammation on the bone regen-eration process observed at the fracture site; in the light ofour data, CD4+ T cells are likely the key effectors of thisbiological function.
A possible explanation for the pronounced osteogenicrole of CD4 T cells is regulatory T cells (Treg). The Tregsubset, which includes CD4+ Foxp3+ CD25+ cells, was re-cently reported to potently promote bone formationin vivo, with a mechanism implying downregulation ofTNFa and IFNg (Liu et al., 2012). The frequency of Tregsin peripheral blood has been reported to be anywherefrom 1% to> 8%, depending on the pool of markers uti-lized to assess their phenotype (Liyanage et al., 2002;Tokuno et al., 2009); while we did not assess the fre-quency of Treg in our subsets, Tregs are necessarilyenriched within the CD4 subset and are likely to have con-tributed to the increased osteogenicity of this subset inour functional assays. However, further investigation willbe needed to assess the expression profile of Tregs withregards to osteogenic protein and growth factors.
One possible limitation of our study is that we analyseda limited set of genes, selected based on a proven involve-ment in MSCs differentiation or mineral deposition inbone. While a whole-genome gene array analysis wouldhave included novel and unpredictable targets in our anal-ysis, with our PCR-array approach we focused on a groupof well-established targets that had never been exploredbefore in T lymphocytes.
In conclusion, establishing an osteogenic milieu in thelocal microenvironment is a fundament step for a success-ful repair of bone. In particular, our findings suggest that,while activated T cells recruited to the site of injury play apredominant inhibitory role in osteogenic differentiation,in the post-inflammatory (remodelling) phase they cancontribute to new bone formation by providing the osteo-genic growth factors necessary for the bone regenerationprocess; in particular, the CD4 and CD8 subsets playdistinct biological roles in the regulation of the healingcascade and the ability of quiescent T cells to stimulateosteogenic differentiation of MSCs is mainly, if not exclu-sively, due to the contribution of gene expression and sol-uble factors secreted by CD4 T cells. Dissecting geneexpression profile and functional roles of CD4 and CD8cells can potentially help to create new therapeutic strate-gies for enhancing skeletal repair.
Acknowledgements
Funding for the study was provided by Rizzoli Orthopaedic Insti-tute (Ricerca Corrente and ’Cinque per mille’), MIUR, and RFOUniversity of Bologna.
Author contributions
FG and CM, cell culture experiments and data acquisition;LC, flow cytometry; LG, EG and AP, PCR array and real-time analyses; FG, AF and GL, study conception, interpre-tation of data and paper writing.
References
Cheng H, Jiang W, Phillips FM, et al. 2003;Osteogenic activity of the 14 types of humanbone morphogenetic proteins (BMPs).J Bone Joint Surg Am 85A: 1544–1552.
ChoTJ, Gerstenfeld LC, EinhornTA. 2002; Dif-ferential temporal expression of members ofthe transforming growth factor-b superfamilyduring murine fracture healing. J Bone MinerRes 17: 513–520.
Cho TJ, Kim JA, Chung CY, et al. 2007;Expression and role of interleukin-6 in dis-traction osteogenesis. Calcif Tissue Int 80:192–200.
Clemetson KJ, and Clemetson JM. 2001;Platelet collagen receptors. Thromb Haemost86: 189–197.
Deschaseaux F, Sensebe L, Heymann D.2009; Mechanisms of bone repair andregeneration. Trends Mol Med 15: 417–429.
Di Rosa F, Pabst R. 2005; The bone marrow:a nest for migratory memory Tcells. TrendsImmunol 26: 360–366.
Dimitriou R, Tsiridis E, Giannoudis PV. 2005;Current concepts of molecular aspects ofbone healing. Injury 36: 1392–1404.
Einhorn TA. 1995; Enhancement of fracturehealing. J Bone Joint Surg Am 77: 940–956.
Gabusi E, Manferdini C, Grassi F, et al. 2011;Extracellular calcium chronically inducedhuman osteoblasts effects: specificmodulation
of osteocalcin and collagen type XV. J CellPhysiol 227: 3151–3161.
Galliera E, Randelli P, Dogliotti G, et al. 2010;Matrix metalloproteases MMP-2 andMMP-9: are they early biomarkers of boneremodelling and healing after arthroscopicacromioplasty? Injury 41: 1204–1207.
Gerstenfeld LC, Cho TJ, Kon T, et al. 2003;Impaired fracture healing in the absenceof TNFa signaling: the role of TNFa inendochondral cartilage resorption. J BoneMiner Res 18: 1584–1592.
Grassi F, Manferdini C, Cattini L, et al. 2011;T cell suppression by osteoclasts in vitro. JCell Physiol 226: 982–990.
Hayashi M, Nimura K, Kashiwagi K, et al.2007; Comparative roles of Twist-1 andId1 in transcriptional regulation by BMPsignaling. J Cell Sci 120: 1350–1357.
Kong YY, Feige U, Sarosi I, et al. 1999; Acti-vated T cells regulate bone loss andjoint destruction in adjuvant arthritisthrough osteoprotegerin ligand. Nature402: 304–309.
Lee MS, Lowe GN, Strong DD, et al. 1999;TWIST, a basic helix–loop–helix transcrip-tion factor, can regulate the human osteo-genic lineage. J Cell Biochem 75: 566–577.
Li Y, Toraldo G, Li A, et al. 2007; B cells and Tcells are critical for the preservation of
bone homeostasis and attainment of peakbone mass in vivo. Blood 109: 3839–3848.
Lisignoli G, Codeluppi K, Todoerti K, et al.2009; Gene array profile identifies colla-gen type XV as a novel human osteoblast-secreted matrix protein. J Cell Physiol220: 401–409.
Liu Y, Wang L, Kikuiri T, et al. 2012; Mes-enchymal stem cell-based tissue regener-ation is governed by recipient Tlymphocytes via IFNg and TNFa. NatMed 17: 1594–1601.
Liyanage UK, Moore TT, Joo HG, et al. 2002;Prevalence of regulatory T cells is in-creased in peripheral blood and tumor mi-croenvironment of patients with pancreasor breast adenocarcinoma. J Immunol169: 2756–2761.
Manferdini C, Gabusi E, Grassi F, et al. 2011;Evidence of specific characteristics and os-teogenic potentiality in bone cells fromtibia. J Cell Physiol 226: 2675–2682.
Mountziaris PM, Spicer PP, Kasper FK, et al.2011; Harnessing and modulating inflam-mation in strategies for bone regeneration.Tissue Eng Part B Rev 17: 393–402.
Niesner U, Albrecht I, Janke M, et al.2008; Autoregulation of Th1-mediatedinflammation by twist1. J Exp Med205: 1889–1901.
CD4 T cells support osteogenic differentiation of human MSCs
Peng H, Wright V, Usas A, et al. 2002; Syner-gistic enhancement of bone formation andhealing by stem cell-expressed VEGF andbone morphogenetic protein-4. J Clin Invest110: 751–759.
Schaffer M, Barbul A. 1998; Lymphocytefunction in wound healing and followinginjury. Br J Surg 85: 444–460.
Schaffer M, Bongartz M, Hoffmann W, et al.2007; MHC-class-II-deficiency impairswound healing. J Surg Res 138: 100–105.
Schmidt-Bleek K, Schell H, Kolar P, et al.2009; Cellular composition of the initialfracture hematoma compared to a musclehematoma: a study in sheep. J Orthop Res27: 1147–1151.
Schmidt-Bleek K, Schell H, Schulz N, et al.2012; Inflammatory phase of bone healing
initiates the regenerative healing cascade.Cell Tissue Res 347: 567–573.
Spicer DB, Rhee J, Cheung WL, et al. 1996;Inhibition of myogenic bHLH and MEF2transcription factors by the bHLH proteinTwist. Science 272: 1476–1480.
Takayanagi H 2007; Osteoimmunology:shared mechanisms and crosstalk betweenthe immune and bone systems. Nat RevImmunol 7: 292–304.
Toben D, Schroeder I, El Khassawna T, et al.2011; Fracture healing is accelerated inthe absence of the adaptive immune sys-tem. J Bone Miner Res 26: 113–124.
Tokuno K, Hazama S, Yoshino S, et al. 2009;Increased prevalence of regulatory T-cellsin the peripheral blood of patients withgastrointestinal cancer. Anticancer Res29: 1527–1532.
Tonnarelli B, Manferdini C, Piacentini A, et al.2009; Surface-dependent modulation of
proliferation, bone matrix molecules, andinflammatory factors in human osteoblasts.J Biomed Mater Res A 89: 687–696.
Torreggiani E, Lisignoli G, Manferdini C,et al. 2012; Role of Slug transcription fac-tor in human mesenchymal stem cells. JCell Mol Med 16: 740–751.
Tsai PT, Lee RA, Wu H. 2003; BMP4 acts up-stream of FGF in modulating thymicstroma and regulating thymopoiesis. Blood102: 3947–3953.
Weitzmann MN, Pacifici R. 2007; T cells: un-expected players in the bone loss inducedby estrogen deficiency and in basalbone homeostasis. Ann N Y Acad Sci1116: 360–375.
Yang X, Ricciardi BF, Hernandez-Soria A,et al. 2007; Callus mineralization and mat-uration are delayed during fracturehealing in interleukin-6 knockout mice.Bone 41: 928–936.
