Platelet-Rich Plasma ActivatesProinflammatory Signaling Pathwaysand Induces Oxidative Stressin Tendon Fibroblasts

Joshua L. Hudgens,* MD, Kristoffer B. Sugg,*yz MD, Jeremy A. Grekin,* MS,Jonathan P. Gumucio,*y BS, Asheesh Bedi,* MD, and Christopher L. Mendias,*y§ PhD, ATCInvestigation performed at the Department of Orthopaedic Surgery,University of Michigan Medical School, Ann Arbor, Michigan, USA

Background: Tendon injuries are one of the most common musculoskeletal conditions in active patients. Platelet-rich plasma(PRP) has shown some promise in the treatment of tendon disorders, but little is known as to the mechanisms by which PRPcan improve tendon regeneration. PRP contains numerous different growth factors and cytokines that activate various cellularsignaling cascades, but it has been difficult to determine precisely which signaling pathways and cellular responses are activatedafter PRP treatment. Additionally, macrophages play an important role in modulating tendon regeneration, but the influence ofPRP on determining whether macrophages assume a proinflammatory or anti-inflammatory phenotype remains unknown.

Purpose: To use genome-wide expression profiling, bioinformatics, and protein analysis to determine the cellular pathways acti-vated in fibroblasts treated with PRP. The effect of PRP on macrophage polarization was also evaluated.

Study Design: Controlled laboratory study.

Methods: Tendon fibroblasts or macrophages from rats were cultured and treated with either platelet-poor plasma (PPP) or PRP.RNA or protein was isolated from cells and analyzed using microarrays, quantitative polymerase chain reaction, immunoblotting,or bioinformatics techniques.

Results: Pathway analysis determined that the most highly induced signaling pathways in PRP-treated tendon fibroblasts wereTNFa and NFkB pathways. PRP also downregulated the expression of extracellular matrix genes and induced the expression ofautophagy-related genes and reactive oxygen species (ROS) genes and protein markers in tendon fibroblasts. PRP failed to havea major effect on markers of macrophage polarization.

Conclusion: PRP induces an inflammatory response in tendon fibroblasts, which leads to the formation of ROS and the activationof oxidative stress pathways. PRP does not appear to significantly modulate macrophage polarization.

Clinical Relevance: PRP might act by inducing a transient inflammatory event, which could then trigger a tissue regenerationresponse.

Keywords: platelet-rich plasma; tendon; tendinopathy; oxidative stress; inflammation; autophagy

Acute and chronic tendon injuries are relatively commonproblems in the general population, secondary to sportsparticipation and other physical activity, and may be thesource of significant morbidity.8,24 Platelet-rich plasma(PRP) is a commonly used biological treatment in sportsmedicine, specifically with interstitial tendon tears andchronic tendinopathies.8,25 Initially described for use inoral and maxillofacial surgery, PRP is a refined productof autologous blood with a platelet concentration greaterthan that of whole blood, typically isolated via differential

centrifugation.1 Platelets are important in the injuryresponse, as they release growth factors that initiate andmodulate wound healing in both soft and hard tissues.The justification for the use of PRP clinically stems froman attempt to recapitulate or augment this natural biolog-ical process.1 Despite numerous clinical outcome studies onthe effects of PRP in sports medicine, there remains a pau-city of information on its mechanism of action.19,25

Two cell types, fibroblasts and macrophages, appear topredominate and coordinate the healing process in injuredand diseased tendons.9,37,39 Fibroblasts function as theprincipal cells involved in tendon maintenance and repair,while macrophages help to break down damaged tendontissue and can secrete cytokines and other signaling mole-cules that modulate the activity of fibroblasts.26,28,38,39

The American Journal of Sports Medicine, Vol. 44, No. 8DOI: 10.1177/0363546516637176! 2016 The Author(s)

Winner of the 2016 Cabaud Award

1931

Macrophages exhibit 2 phenotypes: a proinflammatory(M1) phenotype and an anti-inflammatory (M2) pheno-type.28,38 In response of tissues to injury, the M1 popula-tion of macrophages predominates initially, mediatingphagocytosis and apoptosis, while M2 macrophages appearlater and become the more prevalent population that coor-dinates the repair process and promotes fibroblast prolifer-ation.7,28,39 There have been several in vitro studies on theeffect of PRP on tendon cells, measuring the expression ofspecific genes or proteins related to tendon function,40,41

and to our knowledge, no studies of the effect of PRP onmacrophage polarization. As PRP is a dense milieu ofnumerous growth factors and other signaling molecules,and only a limited number of signaling pathways havebeen studied in tendon cells treated with PRP, we soughtto determine transcriptome-wide changes in gene expres-sion using microarrays and informative bioinformaticsanalyses to evaluate which cellular signaling pathwayswere activated by PRP in an unbiased fashion. Further-more, because macrophages appear to play an importantrole in tendon inflammation and repair, we determinedthe effect of PRP on macrophage polarization. We hypoth-esized that PRP would activate signaling pathwaysinvolved in extracellular matrix (ECM) synthesis andremodeling and would polarize macrophages to an anti-inflammatory M2 phenotype.

METHODS

Animals

This study was approved by the University of MichiganInstitutional Animal Care and Use Committee and fol-lowed United States Public Health Service guidelines forthe ethical treatment of animals. Male retired-breederinbred Lewis rats were obtained from Charles River Labo-ratories and were housed under specific pathogen-free con-ditions. The inbred Lewis strain was selected to avoidadverse immune reactions from blood pooling.29 Ratswere deeply anesthetized with sodium pentobarbital toobtain blood and tendon tissue and then humanely eutha-nized via an anesthetic overdose and the induction ofa bilateral pneumothorax.

Plasma Preparation

Blood was obtained via cardiac puncture and collected intosodium citrate Vacutainer tubes (BD). Platelet-poorplasma (PPP) and PRP were prepared from whole bloodunder sterile conditions. As no widely used commerciallyavailable clinical system is available to prepare PRP from

rats, a manual preparation approach modified from previ-ous studies1,10 was used. Briefly, blood was centrifuged at500g for 5 minutes at 4"C, followed by a 5-minute restperiod, and then another cycle again at 700g for17 minutes at 4"C. The supernatant above the packed cellscontained 2 visibly different layers: the uppermost andnearly transparent layer containing PPP and a lower par-tially flocculent layer containing PRP. A total of 3 separatebatches of PRP and PPP were prepared. A Hemavet 950system (Drew Scientific) was used to quantify platelet den-sities. The mean (6SD) concentration of platelets fromPRP was 1.4 3 106 6 0.5 3 105 platelets/mL and fromPPP was 3.0 3 104 6 0.5 3 103 platelets/mL, whichresulted in PRP having an approximately 4-fold elevationin platelet concentration compared with whole blood.PRP and PPP were frozen at 280"C until use.

Tendon Fibroblast Culture

Fibroblasts were isolated and from tail tendons as previ-ously described.31 Tail tendons develop from the same pop-ulation of somitic progenitor cells as limb tendons33 andare useful in obtaining a large number of low passage cells.Tendon fascicles were finely minced and placed in Dulbec-co’s Modified Eagle Medium (DMEM) containing 0.2% type2 collagenase (Life Technologies) and vigorously agitatedfor 3 hours at 37"C. An equal volume of growth medium(GM) composed of DMEM containing 10% fetal bovineserum (FBS) and 1% antibiotic/antimycotic (Life Technolo-gies) was added to the digested tissue, which was then fil-tered through a 100-mm strainer. Cells were centrifuged at1000g for 10 minutes, resuspended in GM, and plated on100-mm type 1 collagen–coated dishes (BD). All cellswere cultured in humidified incubators maintained at37"C and 5% CO2. Fibroblasts were grown to 70% conflu-ence, collected from dishes using TrypLE (Life Technolo-gies), and resuspended in 3-dimensional (3D) type 1collagen gels. The collagen for these gels was preparedand extracted as described previously.18 Briefly, rat tailtendons were excised and placed in 0.2% acetic acid at4"C. After 5 days, the collagen solution was centrifugedat 24,000g for 30 minutes, and the supernatant was col-lected, lyophilized, and dissolved again in 0.2% aceticacid to a final concentration of 2.7 mg/mL. To preparethe collagen gel, the collagen solution was combined with103 minimum essential medium (Life Technologies) and0.34 N NaOH in an 8:1:1 ratio at 4"C. Tendon fibroblastswere resuspended in this mixture, and 300 mL containing2 3 105 cells was added to each well of a 24-well plate(BD). The plate was then placed in the humidified incuba-tor at 37"C for 45 minutes for gelling to occur.

§Address correspondence to Christopher L. Mendias, PhD, ATC, Department of Orthopaedic Surgery, University of Michigan Medical School, 109 ZinaPitcher Place, BSRB 2017, Ann Arbor, MI 48109-2200, USA (email: cmendias@umich.edu).

*Department of Orthopaedic Surgery, University of Michigan Medical School, Ann Arbor, Michigan, USA.yDepartment of Molecular & Integrative Physiology, University of Michigan Medical School, Ann Arbor, Michigan, USA.zSection of Plastic Surgery, Department of Surgery, University of Michigan Medical School, Ann Arbor, Michigan, USA.Presented at the annual meeting of the AOSSM, Colorado Springs, Colorado, July 2016.One or more of the authors has declared the following potential conflict of interest or source of funding: This work was supported by National Institutes

of Health grants R01-AR063649, F31-AR065931, and F32-AR067086.

1932 Hudgens et al The American Journal of Sports Medicine

After being embedded in 3D collagen gels, fibroblasts werecultured in GM for 3 days. The medium was then changed tocontain DMEM plus 1% antibiotic/antimycotic and 10% PPPor PRP clot releasate. The clot releasate was prepared bytreating PPP or PRP with 30 mM CaCl2 to activate the coag-ulation cascade. Activated PPP and PRP were vigorously agi-tated for 1 hour at 4"C and then spun at 12,000g for10 minutes at 4"C. The supernatant containing the clot relea-sate was collected, added to DMEM plus 1% antibiotic/antimycotic, and passed through a 0.22-mm filter to removeany small fibrin clumps. The resulting PPP- or PRP-contain-ing medium was added to wells containing tendon fibroblastsand changed every 2 days.

Macrophage Culture

Rat resident peritoneal macrophages were purchased fromCell Biologics and were cultured in macrophage medium con-taining basal medium supplemented with granulocyte mac-rophage colony-stimulating factor, 10% FBS, and 1%antibiotic/antimycotic (Cell Biologics). Cells were thawedand cultured on plasma-treated dishes (BD) for 3 days, afterwhich 10% FBS in the medium was substituted for either10% PPP or PRP for 2 days before RNA isolation.

RNA Isolation and Gene Expression

Tendon fibroblasts or macrophages were treated with PPPor PRP for 24 hours, and RNA was isolated as previouslydescribed16,36 using an miRNeasy Micro Kit (Qiagen). AllRNA had an A260/A280 ratio .1.8 (NanoDrop; ThermoFisher) and RNA integrity number (RIN) values .8.0 mea-sured (Bioanalyzer; Agilent). After reverse transcription ofRNA with iScript Supermix (Bio-Rad), quantitative poly-merase chain reaction (qPCR) was conducted in a CFX96real-time thermal cycler using iTaq SYBR green supermixreagents (Bio-Rad). The 2–DCt technique was used to nor-malize the expression of mRNA transcripts to the stablehousekeeping gene b-actin. A listing of RNA transcriptsand primer sequences is provided in Appendix Table A1(available online at http://ajsm.sagepub.com/supplemental).

Microarray Analysis

Microarray measurements of PPP- or PRP-treated tendonfibroblasts were performed by the University of MichiganDNA Sequencing Core following manufacturer recommen-dations. Equal amounts of RNA isolated from 3 individualwells were pooled into a single sample for microarray anal-ysis, and 2 pooled samples of PPP and 2 pooled samples ofPRP were analyzed. RNA was pooled because gene expres-sion from a pooled RNA sample is similar to the averagefrom the individual samples composing the pooled sam-ple.5,23 RNA was prepared for microarray analysis usinga GeneChip WT Pico Kit (Affymetrix) and hybridized toRat Gene ST 2.1 strips (Affymetrix). Raw microarraydata were loaded into ArrayStar version 12.1 (DNASTAR)to calculate fold changes in gene expression data. Themicroarray dataset is available through the National Insti-tutes of Health Gene Expression Omnibus database

(ascension No. GSE70918). The Upstream Regulator mod-ule of Ingenuity Pathway Analysis (IPA) software (Qiagen)was used to determine the transcriptional regulators thatcould explain the observed change in gene expressionmeasurements obtained from microarrays. This moduleexamines the number of known targets of each transcrip-tion regulator that are present in the microarray dataset,along with the direction of change to predict likely relevanttranscriptional regulators. If the observed direction of thefold change in expression is consistent with a particularactivation state of the transcriptional regulator, then a pre-diction is made about whether the pathway is activated orinhibited. A full listing of the IPA Upstream Regulatorresults is listed in Appendix Table A2 (available online).

Protein Isolation and Measurements

Fibroblasts in 3D collagen gels were treated with mediacontaining PPP or PRP for 5 days. Collagen gels werehomogenized in ice-cold radioimmunoprecipitation assaybuffer (Sigma-Aldrich) supplemented with 1:100 proteaseand phosphatase inhibitor cocktail (Life Technologies)and 1% NP-40 (Sigma-Aldrich). After vigorous homogeni-zation, samples were vortexed for 10 minutes at 4"C andthen spun at 12,000g for 10 minutes at 4"C. The superna-tant was collected, and the protein concentration was mea-sured using a bicinchoninic acid assay (Life Technologies).Proteins were diluted in Laemmli sample buffer (Bio-Rad),and 10 mg of total protein was loaded into Any kD gels (Bio-Rad). Proteins were separated with electrophoresis, andthe gels were either stained with Coomassie BrilliantBlue (Bio-Rad) or transferred to membranes for immuno-blotting (Bio-Rad). For NFkB immunoblots, nitrocellulosemembranes were blocked in 2% goat serum and incubatedwith rabbit anti–phospho-NFkB antibodies (S536, 1:1000dilution) and goat anti-rabbit horseradish peroxidase(HRPO)–conjugated secondary antibodies (1:2000 dilu-tion). For the detection of carbonylated proteins, an OxiSe-lect Protein Carbonyl Immunoblot Kit (Cell Biolabs) wasused following manufacturer directions. Briefly, polyviny-lidene difluoride membranes were blocked in 5% powderedmilk and treated with dinitrophenylhydrazine, whichreacts with carbonylated amino acid residues in proteinsto produce dinitrophenol residues. Membranes were thenincubated with anti-dinitrophenol residue antibodies(1:1000 dilution) and HRPO-conjugated secondary antibod-ies (1:1000 dilution). Membranes were treated withenhanced chemiluminescence solution (Clarity ECL; Bio-Rad) to activate HRPO. Gels and membranes were visual-ized in a ChemiDoc XRS system (Bio-Rad), and densitome-try analysis was performed using Image Lab 5.2 software(Bio-Rad).

Protein Array

A rat cytokine antibody array (C2; RayBiotech) was used tomeasure the abundance of 34 proteins in PPP and PRP sam-ples. A total of 100 mL of PPP or PRP was used per assay, whichfollowed the instructions of the manufacturer. Membranes were

AJSM Vol. 44, No. 8, 2016 PRP Induces Oxidative Stress in Tendon Fibroblasts 1933

developed with the Clarity ECL solution and quantified asdescribed above.

Statistical Analysis

Data are presented as mean 6 SD. Differences betweenPPP and PRP groups were assessed using t tests (a = .05)in GraphPad Prism 6.0.

RESULTS

Protein Abundance in PPP Versus PRP

The relative difference in cytokines, growth factors, andother proteins important in tissue inflammation and macro-phage activity in PPP and PRP was determined. Abbreviatedterms used in this article are shown in Table 1. Of the 34 pro-teins analyzed, 26 were significantly higher in PRP com-pared with PPP, including CCL2, CCL20, CXCL5, IL1a,IL1b, IL6, IL10, PDGF-AA, and TNFa (Table 2).

Microarray and BioinformaticsAnalysis of PRP Treatment

We analyzed the effect of PRP treatment on global changesin gene expression patterns in tendon fibroblasts. Usingmicroarrays, we determined that PRP treatment resultedin an upregulation greater than 1.5-fold of 315 genes anda downregulation greater than 1.5-fold of 460 genes (Figure1A). To gain more information about the biological signifi-cance of the microarray results and identify the signalingpathways predicted to be activated or inhibited by PRP,we analyzed fold changes in global gene expression withthe Upstream Regulator module of the IPA software. Thisanalysis identified the TNFa pathway (P = 6.6 3 1026)and the NFkB pathway (P = 9.5 3 10219) as the 2 pathwaysactivated in fibroblasts treated with PRP (Figure 1B).

PRP Effects on Tendon Fibroblast Gene Expression

After performing microarrays, we sought to validate thefold change values of specific genes of relevance to tendonbiology and inflammation with qPCR. For genes related totendon growth, PRP upregulated BMP7 but did not changethe expression of TGFb and downregulated CTGF andIGF1 (Figure 2). The inflammation- and immune-modulat-ing cytokines CCL2, CCL7, IL1a, IL6, IL10, and TNFawere upregulated in response to PRP, while no differencein IL1b or VEGF was observed, and IL15 was downregu-lated (Figure 2).

The expression of genes involved with ECM synthesisand remodeling was also quantified. PRP had no effect onthe expression of the hyaluronic acid (HA) synthaseenzymes HAS1 and HAS2 (Figure 3). Elastin expressionwas downregulated along with a slight elevation in thecross-linking enzyme LOX (Figure 3). The major fibrillarcollagens, type 1 and type 3 collagen, along with genes asso-ciated with collagen fibril assembly, CILP, fibromodulin,and collagen types 12 and 14, were downregulated in

PRP-treated fibroblasts, while the basement membranetype 8 collagen and the proteoglycan lubricin were upregu-lated (Figure 3). PRP induced the expression of the majorcollagenase MMP13, along with the stromelysins MMP3and MMP10 and the gelatinase MMP9, with no differenceobserved in the expression of the collagenase MMP8 andthe gelatinase MMP2 nor the TIMP genes TIMP1 orTIMP2 (Figure 3).

Subsequently, the expression of genes involved in vari-ous cell functions including fibroblast proliferation, differ-entiation, autophagy, and inflammation was assessed.

TABLE 1Abbreviated Terms

Abbreviation Expansion

Arg ArginaseAtg Autophagy-related proteinBMP Bone morphogenetic proteinBnip B-cell lymphoma/adenovirus E1B

interacting proteinCCL Chemokine (C-C motif) ligandCCR C-C chemokine receptorCD Cluster of differentiationCILP Cartilage intermediate layer proteinCNTF Ciliary neurotrophic factorCol CollagenCox CyclooxygenaseCSF Colony-stimulating factorCTGF Connective tissue growth factorCXCL Chemokine (C-X-C motif) ligandEGR Early growth response proteinFACIT Fibril-associated collagens with interrupted

triple helicesFGF Fibroblast growth factorFmod FibromodulinGABARAPL Gamma-aminobutyric acid receptor-associated

protein–likeHAS Hyaluronan synthaseIFN InterferonIGF Insulin-like growth factorIKK IkB kinaseIL InterleukiniNOS Inducible nitric oxide synthaseLOX LipoxygenaseMMP Matrix metalloproteinaseNFE2L Nuclear factor (erythroid-derived 2)–likeNFkB Nuclear factor kappa light chain enhancer of

activated B cellsPDGF Platelet-derived growth factorPLD1 Phospholipase D1Prdx PeroxiredoxinPTGES Prostaglandin E synthaseScx ScleraxisSIRT SirtuinSOD Superoxide dismutaseTGF Transforming growth factorTIMP Tissue inhibitor of metalloproteinasesTNFa Tumor necrosis factor alphaTNFR Tumor necrosis factor receptorTnmd TenomodulinTrim Tripartite motif containingVEGF Vascular endothelial growth factor

1934 Hudgens et al The American Journal of Sports Medicine

The cell proliferation marker Ki67 was slightly elevated inresponse to PRP treatment, but a downregulation in theexpression of genes involved in tendon fibroblast specifica-tion and differentiation including EGR1, EGR2, scleraxis,and tenomodulin was observed (Figure 4). For genesinvolved with autophagy, there was an induction in theexpression of Atg10, Bnip1, and GABARAPL2, with no dif-ference in beclin 1 or Trim13 levels (Figure 4). Transcrip-tion factors involved with inflammation, Fosb, Fosl1, andc-Jun, were induced by PRP, but no difference in theexpression of the deacetylase SIRT1 or the nitric oxide–producing gene iNOS was observed (Figure 4). Genesinvolved with prostaglandin production were upregulatedby PRP treatment, including PLD1, PTGES, Cox1, andCox2, while no difference in the leukotriene synthesisenzyme 5-LOX was observed (Figure 4). PRP also inducedthe expression of markers of elevated reactive oxygen spe-cies (ROS) production, including SOD1, SOD2, NFE2L2,and Prdx1 (Figure 4).

PRP Effects on NFkB Activationand Protein Carbonylation

To verify that an elevation in ROS was indeed present andthat the predicted elevation of NFkB did occur, we mea-sured the levels of carbonylated proteins and the abun-dance of phospho-NFkB using immunoblots and observedan induction in both the amount of carbonylated proteinsand in activated NFkB protein levels (Figure 5).

PRP and Macrophage Polarization

Finally, the expression of transcripts involved in the polar-ization of macrophages to a proinflammatory or anti-inflammatory phenotype was assessed. PRP resulted inan induction in the expression of the M1 proinflammatorymarkers iNOS, IL1b, and VEGF, with no changes in theexpression of CCR7, CD11b, CD68, IL15, or TNFa (Figure6A). However, there was also a modest induction in several

TABLE 2Relative Abundance of 34 Different Proteins From Platelet-Poor Plasma (PPP) and Platelet-Rich Plasma (PRP)a

Protein PPP PRP

Activin A 1.00 6 0.51 2.07 6 0.21b

Advanced glycosylation end product (AGE) 1.00 6 0.56 2.09 6 0.22b

Agrin 1.00 6 0.32 1.95 6 0.04b

Brain-derived neurotrophic factor (BDNF) 1.00 6 0.54 2.01 6 0.12b

Chemokine (C-C motif) ligand 2 (CCL2) 1.00 6 0.48 2.87 6 0.08b

Chemokine (C-C motif) ligand 20 (CCL20) 1.00 6 0.30 2.04 6 0.11b

Chemokine (C-X-C motif) ligand 1 (CXCL1) 1.00 6 0.51 1.49 6 0.19Chemokine (C-X-C motif) ligand 2 (CXCL2) 1.00 6 0.35 1.31 6 0.22Chemokine (C-X-C motif) ligand 3 (CXCL3) 1.00 6 0.45 1.31 6 0.25Chemokine (C-X-C motif) ligand 5 (CXCL5) 1.00 6 0.48 6.32 6 0.05b

Chemokine (C-X-C motif) ligand 7 (CXCL7) 1.00 6 0.29 1.36 6 0.11Ciliary neurotrophic factor (CNTF) 1.00 6 0.53 2.00 6 0.26b

Cluster of differentiation 86 (CD86) 1.00 6 0.39 1.79 6 0.07b

Colony-stimulating factor 2 (CSF2) 1.00 6 0.52 2.29 6 0.06b

Fas ligand 1.00 6 0.79 2.83 6 0.36b

Fractalkine 1.00 6 0.57 3.04 6 0.17b

Intercellular adhesion molecule 1 (ICAM1) 1.00 6 0.49 2.74 6 0.01b

Interferon gamma (IFNg) 1.00 6 0.45 2.22 6 0.01b

Interleukin 1 receptor–like 2 (IL1RL2) 1.00 6 0.56 2.12 6 0.08b

Interleukin 1 alpha (IL1a) 1.00 6 0.32 1.90 6 0.05b

Interleukin 1 beta (IL1b) 1.00 6 0.43 2.17 6 0.07b

Interleukin 2 (IL2) 1.00 6 0.42 1.69 6 0.11Interleukin 4 (IL4) 1.00 6 0.46 1.94 6 0.19b

Interleukin 6 (IL6) 1.00 6 0.40 2.01 6 0.17b

Interleukin 10 (IL10) 1.00 6 0.40 2.02 6 0.21b

Interleukin 13 (IL13) 1.00 6 0.56 1.45 6 0.32L-selectin 1.00 6 0.46 2.83 6 0.05b

Leptin 1.00 6 0.46 2.43 6 0.03b

Matrix metalloproteinase 8 (MMP8) 1.00 6 0.27 4.17 6 0.01b

Platelet-derived growth factor AA (PDGF-AA) 1.00 6 0.32 1.73 6 0.01b

Prolactin receptor 1.00 6 0.50 1.53 6 0.05Tissue inhibitor of metalloproteinases 1 (TIMP1) 1.00 6 0.40 3.01 6 0.26b

Tumor necrosis factor alpha (TNFa) 1.00 6 0.52 2.98 6 0.24b

Vascular endothelial growth factor A (VEGF-A) 1.00 6 0.52 1.57 6 0.01

aValues are relative intensities normalized to the PPP group and are presented as mean 6 SD; n = 3 replicates from each group.bSignificantly different from the PPP group (P \ .05).

AJSM Vol. 44, No. 8, 2016 PRP Induces Oxidative Stress in Tendon Fibroblasts 1935

M2 anti-inflammatory markers including arginase 1, IL10,CD163, and CD14, with no change in FGF2, CD206,CD168, TGFb, or IGF1 expression (Figure 6B).

DISCUSSION

PRP is commonly used in the treatment of acute and chronictendon injuries and diseases,8,25 and to our knowledge, this

is the first study that investigated global transcriptomicchanges in tendon fibroblasts after PRP administration.We hypothesized that PRP would activate signaling path-ways involved in ECM synthesis and remodeling. Surpris-ingly, the only 2 pathways predicted to be activated werethe proinflammatory TNFa and NFkB pathways. Genesrelated to cellular proliferation and tendon collagen remod-eling were also increased, suggesting that PRP may activatefibroblast activity and collagen remodeling but not collagen

315

460

2521

>1.5-fold ! for PRP

>1.5-fold " for PRP

33,389

3296 P < .05

P ≥ .05

UpregulatedDownregulatedPredicted activationPredicted inhibitionLeads to activationLeads to inhibition

A

B

Figure 1. Microarray and bioinformatics analysis of tendon fibroblasts treated with platelet-rich plasma (PRP) or platelet-poorplasma (PPP). (A) Microarray analysis identified 3296 genes of 36,685 that were significantly different (P \ .05) between PRP-and PPP-treated cells. Of the 3296 genes that were significantly different, 315 genes were .1.5-fold upregulated, and 460 geneswere .1.5-fold downregulated in PRP-treated cells compared to PPP-treated cells. (B) Ingenuity Pathway Analysis identified 2pathways that were predicted to be highly activated in cells treated with PRP compared to PPP: the TNFa pathway (P = 6.6 31026) and the NFkB pathway (P = 9.5 3 10219). The merged pathways are presented.

1936 Hudgens et al The American Journal of Sports Medicine

synthesis. We also hypothesized that PRP would polarizemacrophages to an anti-inflammatory M2 phenotype butunexpectedly failed to observe any clear effect of PRP

treatment on macrophage polarization. While more studiesare necessary, these results provide important insight intothe regulation of tendon cell activity by PRP treatmentand suggest that PRP might act by inducing an intermittentbout of inflammation, which may then trigger a tissueregeneration response.

PRP contains numerous growth factors and cytokinesthat can activate various signaling pathways in cells.Some of the signaling components that are downstreamof individual receptors can act to inhibit or further enhancethe activation of other pathways regulated by differentreceptors. For example, PRP contains both IL1b andIL10. While IL1b is a well-known activator of proinflam-matory intracellular signaling cascades, IL10 signalingpathways are able to inhibit the pathways activated byIL1b and reduce the expression of proinflammatorygenes.22 With the use of expression data from the entiretranscriptome, the IPA-based bioinformatics approachemployed in this study allowed us to evaluate the complexrelationship between various individual signaling cascadesto identify which pathways were most highly enrichedafter PRP treatment. Interestingly, the highly related

Ki67EGR1

EGR2Scx

Tnmd

Atg10

Beclin

1Bnip

1

Trim13

GABARAPL2Fos

bFos

l1

0.1

1

10

40

Rela

tive

Expr

essio

n

*

*

*

*

*

* *

**

*

c-Jun

SIRT1

iNOSPLD

1

PTGESCox

1Cox

25-L

oxSOD1

SOD2

NFE2L2

Prdx1

0.1

1

10

40

Rela

tive

Expr

essio

n

PPPPRP

*

*

*

*

*

*

*

* *

Figure 4. Gene expression of cell proliferation, differentia-tion, autophagy, and inflammatory transcripts from tendonfibroblasts treated with platelet-rich plasma (PRP) or plate-let-poor plasma (PPP). Target gene expression was normal-ized to b-actin. Values are mean 6 SD; n = 6 replicatesfrom each group. *Significantly different from the PPP group(P \ .05).

BMP7CCL2

CCL7CTGF

IGF1

IL1 IL1b IL6IL10

IL15TGF

TNFVEGF0.1

1

10

40

Rela

tive

Expr

essio

n

PPPPRP

**

*

*

*

*

**

*

*

Figure 2. Gene expression of growth factor and cytokine tran-scripts from tendon fibroblasts treated with platelet-rich plasma(PRP) or platelet-poor plasma (PPP). Target gene expressionwas normalized to b-actin. Values are mean 6 SD; n = 6 repli-cates from each group. *Significantly different from the PPPgroup (P \ .05).

Col11

Col31

Col81

Col12

1

Col14

1HAS1

HAS2

Elastin

LOX

Lubri

cin

0.1

1

10

40

Rela

tive

Expr

essio

n

* *

*

**

*

*

*

MMP2MMP3

MMP8MMP9

MMP10

MMP13TIM

P1TIM

P2CILP

Fmod

0.1

1

10

40

Rela

tive

Expr

essio

n

PPPPRP

* *

*

**

*

Figure 3. Gene expression of extracellular matrix structuraland remodeling transcripts from tendon fibroblasts treatedwith platelet-rich plasma (PRP) or platelet-poor plasma(PPP). Target gene expression was normalized to b-actin.Values are mean 6 SD; n = 6 replicates from each group.*Significantly different from the PPP group (P \ .05).

AJSM Vol. 44, No. 8, 2016 PRP Induces Oxidative Stress in Tendon Fibroblasts 1937

TNFa and NFkB pathways were the only 2 pathways thatwere predicted to be functionally activated in response toPRP. Based on the role that the TNFa and NFkB pathwaysplay in regulating ECM remodeling, oxidative stress, andinflammation, we then chose to further evaluate and char-acterize these responses in greater detail.

Both acute tendon tears and chronic degenerative tendi-nopathies involve a damaged or disordered ECM that mustbe remodeled and repaired by tendon fibroblasts.8,17 HA isa glycosaminoglycan that serves as a template for newECM synthesis,3,36 and PRP treatment had no effect onthe expression of the major HA synthase enzymes HAS1and HAS2. PRP downregulated the expression of the majorcollagens in tendon, collagen 1 and 3, as well as elastin,which has an important role in restoring ECM organiza-tion after being stretched. In addition to these genes, sev-eral transcripts that help to assemble mature collagenfibrils, including the FACIT collagens 12 and 14, as wellas CILP and fibromodulin, were downregulated afterPRP treatment. These results are in general agreementwith collagen expression reported in previous work11,40

and indicate that PRP treatment reduces the expressionof major ECM components in tendon fibroblasts. PRPalso induced the expression of several major MMPs, includ-ing MMP3, MMP9, MMP10, and MMP13, which togetherdegrade the major fibrillar collagens, minor collagens,and other ECM structural proteins.8 While we are still inthe early stages of understanding the networks of tran-scription factors and signaling pathways that regulate ten-don fibroblast specification and proliferation, EGR1,

EGR2, and scleraxis are transcription factors known toplay crucial roles in tendon development, growth, andremodeling,16,20,36 and in the current study, PRP downre-gulated the expression of all 3 of these genes. Tenomodu-lin, which is a marker of differentiated fibroblasts,20 wasalso downregulated by PRP. Autophagy is a catabolic cellu-lar process that is important in tissue remodeling,14,34 andmultiple genes that are important in the initiation andmaturation of autophagosomes were upregulated by PRP,including Atg10, Bnip1, and GABARAPL2. The combinedchanges in ECM, MMP, tenogenesis, and autophagy geneexpression suggest that PRP treatment likely actuallyresults in an atrophied ECM and reduced fibroblast activ-ity, which is largely consistent with a state of elevatedacute tissue inflammation.8

The TNFa/NFkB signaling cascade is a well-knownmediator of inflammation. Binding of TNFa to its receptorTNFR1 triggers activation of the IKK complex, which isresponsible for activating the p65 transcription factor sub-unit of NFkB through a combination of degradation of theinhibitor IkB complex and phosphorylation of NFkB.2,35

Once activated, NFkB translocates to the nucleus andinduces the expression of numerous genes, many of whichare associated with inflammation. Oxidative stress is alsoable to activate NFkB, often having an additive effect toTNFa signaling.27 In the current study, the treatment oftendon fibroblasts with PRP resulted in an elevation ofgenetic markers of oxidative stress, including SOD1,

A PPP PRP

Car

bony

latio

nTo

tal P

rote

in

B

PPP PRP0.0

0.5

1.0

1.5

2.0

*

PPP PRP0

2

4

6

8

10

Car

bony

latio

n R

el U

nits

pNFκ

B R

el U

nits

*C

pNFκ

B

250

10075

50

250

10075

50

kDa

75

50

Figure 5. Immunoblots for carbonylation and NFkB activa-tion. (A) Immunoblots for total carbonylated proteins andphospho-NFkB activation from tendon fibroblasts treatedwith platelet-rich plasma (PRP) or platelet-poor plasma(PPP). A Coomassie-stained gel of the same samples isshown for validation of equal protein loading. Band densi-tometry analysis for (B) total carbonylated proteins and (C)phospho-NFkB blots. Values are mean 6 SD; n = 4 replicatesfrom each group. *Significantly different from the PPP group(P \ .05).

F4/80

CCR7

CD11b

CD68iNOS

IL1 IL15TNF

VEGF

0.1

1

10

Rela

tive

Expr

essio

n

**

*

Arg1FGF2

CD14

CD163

CD168

CD206

IGF1

IL10TGF

0.1

1

10

Rela

tive

Expr

essio

n

PPPPRP

* * **

A

B

Figure 6. Gene expression of markers of (A) proinflammatoryand (B) anti-inflammatory polarization from macrophagestreated with platelet-rich plasma (PRP) or platelet-poorplasma (PPP). Target gene expression was normalized tob-actin. Values are mean 6 SD; n = 6 replicates from eachgroup. *Significantly different from the PPP group (P \ .05).

1938 Hudgens et al The American Journal of Sports Medicine

SOD2, NFE2L2, and Prdx1, as well as the chronic phos-phorylation and activation of NFkB. Consistent with this,we observed a marked increase in carbonylated proteins,which are sensitive markers of elevated oxidative stress.30

While TNFa is elevated in PRP and is able to induce oxida-tive stress through the induction of proinflammatory path-ways,21 because platelets can produce and releasehydrogen peroxide,13 it is possible that PRP also containsendogenous peroxides that can produce ROS and furtherenhance oxidative stress in tendon fibroblasts. No changein iNOS expression was observed, and combined with ele-vations in SOD1, SOD2, NFE2L2, and Prdx1, this suggeststhat the elevated oxidative stress was likely caused by per-oxide-mediated processes instead of nitric oxide. Amongthe more potent proinflammatory genes that are inducedin response to NFkB activation are the prostaglandin syn-thesis enzymes PTGES, Cox1, and Cox2.27 PRP treatmentpotently induced the expression of these 3 enzymes but hadno effect on 5-LOX expression, suggesting a role for prosta-glandins but not leukotrienes in PRP-mediated inflamma-tion. PRP did not change the expression of SIRT1, which isan important and potent inhibitor of NFkB activity.21 Sev-eral other proinflammatory transcription factors were alsoupregulated by PRP treatment, including Fosb, Fosl1, andc-Jun. These results together suggest that PRP treatmentinduces a robust and heady induction of inflammatory andoxidative stress pathways in tendon fibroblasts.

Macrophages appear to play an important role in therepair and regeneration of both acute tendon injuries andchronic degenerative tendinopathies.7,9,39 Dragoo and col-leagues12 also observed that a PRP injection into otherwisehealthy tendons resulted in an acute inflammatory responseand infiltration of macrophages into the injected tissue. Thisis consistent with findings in the current study, as PRP con-tains elevated levels of several chemokine ligand proteinsthat are involved in the recruitment of macrophages to tis-sue. CCL2 and CCL7 expression were also highly induced infibroblasts treated with PRP. Many of the individual compo-nents of PRP are also able to polarize cultured macrophagesinto a specific phenotype in isolation. For example, IFNgand TNFa can prime macrophages to an M1 phenotype,while IL4 and IL10 are able to polarize macrophages to anM2 phenotype.28 There are no widely accepted and specificand definitive binary markers of the M1 or M2 phenotype,and a panel of various markers and the fold change in thesemarkers are typically used to assess phenotypic changes inmacrophage polarization.28,32,38 Within the M2 phenotype,there are subphenotypes, including M2a macrophages,which are generally regarded as anti-inflammatory macro-phages that function to resolve the M1 response, and M2cmacrophages, which function to promote tissue repair andregeneration.28 The specific markers for these subpheno-types are less well defined than those that define M1 versusM2 macrophages; however, Arg1 and FGF2 are generallyconsidered M2a markers, while CD14, CD163, CD168,CD206, IGF1, IL10, and TGFb can mark M2c macro-phages.28,42 We observed that PRP treatment modestlyincreased the expression of the M1 markers iNOS andIL1b along with a robust increase in VEGF. However, therewere also modest increases in the M2a marker Arg1 and

M2c markers CD14, IL10, and CD163. With the exceptionof VEGF, no macrophage phenotype marker that was eval-uated showed tremendous changes in expression. In theabsence of robust changes in more specific markers, we donot believe that VEGF alone can sufficiently mark the M1phenotype and conclude that PRP did not have a markedeffect on macrophage polarization. Interestingly, despitea substantial change in VEGF observed in macrophages,PRP treatment did not change VEGF expression in tendonfibroblasts. Neovascularization is often observed in acuteand chronic tendon disorders,37,39 and it is possible thatmacrophages are the cells responsible for signaling of bloodvessel ingrowth into damaged areas of tendons.

This work has several limitations. These studies were con-ducted in cultured rat cells, and while we attempted to sim-ulate a native environment as much as possible in vitro, it ispossible that cells would respond to PRP differently in vivo.We also used a single dose of PRP added to culture mediaand did not determine if there were dose-dependent effectsof PRP on cell behavior. For most measures, we evaluatedchanges in gene expression and not protein, and it is possiblethat changes in gene expression do not result in changes inprotein levels. There are no commercially available PRPkits that have been validated for rats. Further, growth factorand cytokine levels in PRP can vary widely depending on thespecific kit that is used.4 Despite these limitations, this studyprovided important information related to the mechanism ofaction of PRP in tendon fibroblasts and macrophages. Futurestudies that use human cells or in vivo animal studies, pre-pare PRP from commercial kits, use multiple doses andtime points, and analyze more changes at the protein levelwill provide further insight into the biology of PRP and hope-fully further refine its clinical use.

CONCLUSION

PRP has been used as a therapy for the treatment of ten-don injuries and chronic diseases, but meta-analyses ofnumerous clinical trials do not indicate a clear benefit forthe use of PRP in treating tendon disorders.25 A major rea-son for this is the scarcity of trials that have enrolled largecohorts of patients as well as the substantial variation inPRP preparation and delivery, patient demographics, andchronicity and site of injuries.25 Another limitation to thewidespread acceptance of PRP for use in clinical practiceis an inadequate understanding of its biological mecha-nism of action. This study provided cellular and biochemi-cal data on the mechanism of action for PRP and reportedthat it appears to work by inducing a massive inflamma-tory reaction in tendon fibroblast cells. Inflammation isgenerally thought of in a negative fashion, but it also playsan important role in triggering a regeneration and repairresponse.15 While the exacerbation of inflammation in anacute tendon injury may not be beneficial, inducing anacute bout of inflammation in chronic tendinopathiesmay end up initiating a subsequent regenerative response.PRP is not unique in this manner; prolotherapy and needlefenestration are used in the treatment of chronic tendino-pathies and have been proposed to work by a similar

AJSM Vol. 44, No. 8, 2016 PRP Induces Oxidative Stress in Tendon Fibroblasts 1939

mechanism of action.6 Considering the time and expenserequired in preparing PRP, future large clinical trialsthat evaluate the ability of PRP to treat chronic tendon dis-orders in comparison to prolotherapy or needle fenestra-tion are warranted.

REFERENCES

1. Amable PR, Carias RBV, Teixeira MVT, et al. Platelet-rich plasma prep-aration for regenerative medicine: optimization and quantification ofcytokines and growth factors. Stem Cell Res Ther. 2013;4(3):67.

2. Buss H, Handschick K, Jurrmann N, et al. Cyclin-dependent kinase 6phosphorylates NF-kB P65 at serine 536 and contributes to the regu-lation of inflammatory gene expression. PLoS One. 2012;7(12):e51847.

3. Calve S, Isaac J, Gumucio JP, Mendias CL. Hyaluronic acid, HAS1,and HAS2 are significantly upregulated during muscle hypertrophy.Am J Physiol Cell Physiol. 2012;303(5):C577-C588.

4. Castillo TN, Pouliot MA, Kim H-J, Dragoo JL. Comparison of growthfactor and platelet concentration from commercial platelet-richplasma separation systems. Am J Sports Med. 2011;39(2):266-271.

5. Chaillou T, Jackson JR, England JH, et al. Identification of a con-served set of upregulated genes in mouse skeletal muscle hypertro-phy and regrowth. J Appl Physiol. 2015;118(1):86-97.

6. Chiavaras MM, Jacobson JA. Ultrasound-guided tendon fenestra-tion. Semin Musculoskelet Radiol. 2013;17(1):85-90.

7. Dakin SG, Werling D, Hibbert A, et al. Macrophage sub-populationsand the lipoxin A4 receptor implicate active inflammation duringequine tendon repair. PLoS One. 2012;7(2):e32333.

8. Davis ME, Gumucio JP, Sugg KB, Bedi A, Mendias CL. MMP inhibi-tion as a potential method to augment the healing of skeletal muscleand tendon extracellular matrix. J Appl Physiol. 2013;115(6):884-891.

9. Dean BJF, Snelling SJB, Dakin SG, Murphy RJ, Javaid MK, Carr AJ.Differences in glutamate receptors and inflammatory cell numbersare associated with the resolution of pain in human rotator cuff ten-dinopathy. Arthritis Res Ther. 2015;17(1):176.

10. Delos D, Leineweber MJ, Chaudhury S, Alzoobaee S, Gao Y, RodeoSA. The effect of platelet-rich plasma on muscle contusion healing ina rat model. Am J Sports Med. 2014;42(9):2067-2074.

11. de Mos M, van der Windt AE, Jahr H, et al. Can platelet-rich plasmaenhance tendon repair? A cell culture study. Am J Sports Med.2008;36(6):1171-1178.

12. Dragoo JL, Braun HJ, Durham JL, et al. Comparison of the acute inflam-matory response of two commercial platelet-rich plasma systems inhealthy rabbit tendons. Am J Sports Med. 2012;40(6):1274-1281.

13. Finazzi-Agro A, Menichelli A, Persiani M, Biancini G, Del Principe D.Hydrogen peroxide release from human blood platelets. Biochim Bio-phys Acta. 1982;718(1):21-25.

14. Gumucio JP, Davis ME, Bradley JR, et al. Rotator cuff tear reducesmuscle fiber specific force production and induces macrophageaccumulation and autophagy. J Orthop Res. 2012;30(12):1963-1970.

15. Gumucio JP, Flood MD, Phan AC, Brooks SV, Mendias CL. Targetedinhibition of TGF-b results in an initial improvement but long-termdeficit in force production after contraction-induced skeletal muscleinjury. J Appl Physiol. 2013;115(4):539-545.

16. Gumucio JP, Phan AC, Ruehlmann DG, Noah AC, Mendias CL. Syn-ergist ablation induces rapid tendon growth through the synthesis ofa neotendon matrix. J Appl Physiol. 2014;117(11):1287-1291.

17. Gumucio JP, Sugg KB, Mendias CL. TGF-b superfamily signaling inmuscle and tendon adaptation to resistance exercise. Exerc SportSci Rev. 2015;43(2):93-99.

18. Hotary K, Allen E, Punturieri A, Yana I, Weiss SJ. Regulation of cellinvasion and morphogenesis in a three-dimensional type I collagenmatrix by membrane-type matrix metalloproteinases 1, 2, and 3.J Cell Biol. 2000;149(6):1309-1323.

19. Hsu WK, Mishra A, Rodeo SR, et al. Platelet-rich plasma in orthopae-dic applications: evidence-based recommendations for treatment.J Am Acad Orthop Surg. 2013;21(12):739-748.

20. Huang AH, Lu HH, Schweitzer R. Molecular regulation of tendon cellfate during development. J Orthop Res. 2015;33(6):800-812.

21. Kauppinen A, Suuronen T, Ojala J, Kaarniranta K, Salminen A. Antag-onistic crosstalk between NF-kB and SIRT1 in the regulation of inflam-mation and metabolic disorders. Cell Signal. 2013;25(10):1939-1948.

22. Kelly A, Lynch A, Vereker E, et al. The anti-inflammatory cytokine,interleukin (IL)-10, blocks the inhibitory effect of IL-1 beta on longterm potentiation: a role for JNK. J Biol Chem. 2001;276(49):45564-45572.

23. Kendziorski C, Irizarry RA, Chen KS, Haag JD, Gould MN. On the util-ity of pooling biological samples in microarray experiments. Proc NatlAcad Sci USA. 2005;102(12):4252-4257.

24. Khan K, Cook J. The painful nonruptured tendon: clinical aspects.Clin Sports Med. 2003;22(4):711-725.

25. Khan M, Bedi A. Cochrane in CORR (#): platelet-rich therapies formusculoskeletal soft tissue injuries (review). Clin Orthop Relat Res.2015;473(7):2207-2213.

26. Kjaer M. Role of extracellular matrix in adaptation of tendon and skel-etal muscle to mechanical loading. Physiol Rev. 2004;84(2):649-698.

27. Korbecki J, Baranowska-Bosiacka I, Gutowska I, Chlubek D. Theeffect of reactive oxygen species on the synthesis of prostanoidsfrom arachidonic acid. J Physiol Pharmacol. 2013;64(4):409-421.

28. Laskin DL. Macrophages and inflammatory mediators in chemicaltoxicity: a battle of forces. Chem Res Toxicol. 2009;22(8):1376-1385.

29. Marco ML. The Fischer-Lewis model of chronic allograft rejection:a summary. Nephrol Dial Transplant. 2006;21(11):3082-3086.

30. McDonagh B, Sakellariou GK, Smith NT, Brownridge P, Jackson MJ.Differential cysteine labeling and global label-free proteomics revealsan altered metabolic state in skeletal muscle aging. J Proteome Res.2014;13(11):5008-5021.

31. Mendias CL, Gumucio JP, Lynch EB. Mechanical loading and TGF-bchange the expression of multiple miRNAs in tendon fibroblasts. JAppl Physiol. 2012;113(1):56-62.

32. Mosser DM, Edwards JP. Exploring the full spectrum of macrophageactivation. Nat Rev Immunol. 2008;8(12):958-969.

33. Murchison ND, Price BA, Conner DA, et al. Regulation of tendon dif-ferentiation by scleraxis distinguishes force-transmitting tendonsfrom muscle-anchoring tendons. Development. 2007;134(14):2697-2708.

34. Neel BA, Lin Y, Pessin JE. Skeletal muscle autophagy: a new meta-bolic regulator. Trends Endocrinol Metab. 2013;24(12):635-643.

35. Pelzer C, Thome M. IKKa takes control of canonical NF-kB activa-tion. Nat Immunol. 2011;12(9):815-816.

36. Schwartz AJ, Sarver DC, Sugg KB, Dzierzawski JT, Gumucio JP,Mendias CL. p38 MAPK signaling in postnatal tendon growth andremodeling. PLoS One. 2015;10(3):e0120044.

37. Sharma P, Maffulli N. Biology of tendon injury: healing, modeling andremodeling. J Musculoskelet Neuronal Interact. 2006;6(2):181-190.

38. Sica A, Mantovani A. Macrophage plasticity and polarization: in vivoveritas. J Clin Invest. 2012;122(3):787-795.

39. Sugg KB, Lubardic J, Gumucio JP, Mendias CL. Changes in macro-phage phenotype and induction of epithelial-to-mesenchymal transi-tion genes following acute Achilles tenotomy and repair. J OrthopRes. 2014;32(7):944-951.

40. Wang X, Qiu Y, Triffitt J, Carr A, Xia Z, Sabokbar A. Proliferation anddifferentiation of human tenocytes in response to platelet richplasma: an in vitro and in vivo study. J Orthop Res. 2012;30(6):982-990.

41. Zhang J, Wang JH-C. Platelet-rich plasma releasate promotes differ-entiation of tendon stem cells into active tenocytes. Am J SportsMed. 2010;38(12):2477-2486.

42. Zizzo G, Hilliard BA, Monestier M, Cohen PL. Efficient clearance ofearly apoptotic cells by human macrophages requires M2c polariza-tion and MerTK induction. J Immunol. 2012;189(7):3508-3520.

For reprints and permission queries, please visit SAGE’s Web site at http://www.sagepub.com/journalsPermissions.nav.

1940 Hudgens et al The American Journal of Sports Medicine