A Novel Therapeutic Approach Using MesenchymalStem Cells to Protect Against Mycobacterium

abscessus

JONG-SEOK KIM,a SANG-HO CHA,b WOO SIK KIM,a SEUNG JUNG HAN,a SEUNG BIN CHA,a

HONG MIN KIM,a KEE WOONG KWON,a SO JEONG KIM,a HONG-HEE CHOI,a JIENNY LEE,b

SANG-NAE CHO,a WON-JUNG KOH,c YEONG-MIN PARK,d SUNG JAE SHINa

Key Words. Mesenchymal stem cells • Stem cell-microenvironment interactions • Cell interactions• Cellular therapy • Cytokines

ABSTRACT

Recent studies have demonstrated the therapeutic potential of mesenchymal stem cells (MSCs)for the treatment of acute inflammatory injury and bacterial pneumonia, but their therapeuticapplications in mycobacterial infections have not been investigated. In this study, we demon-strated the use of MSCs as a novel therapeutic strategy against Mycobacterium abscessus

(M. abscessus), which is the most drug-resistant and difficult-to-treat mycobacterial pathogen.The systemic intravenous injection of MSCs not only improved mouse survival but alsoenhanced bacterial clearance in the lungs and spleen. Additionally, MSCs enhanced IFN-c, TNF-a, IL-6, MCP-1, nitric oxide (NO) and PGE2 production and facilitated CD41/CD81 T cell,CD11bhigh macrophage, and monocyte recruitment in the lungs of M. abscessus-infected mice.To precisely elucidate the functions of MSCs in M. abscessus infection, an in vitro macrophageinfection system was used. MSCs caused markedly increased NO production via NF-jB activationin M. abscessus-infected macrophages cultured in the presence of IFN-c. Inhibiting NO or NF-jBsignaling using specific inhibitors reduced the antimycobacterial activity of MSCs. Furthermore, thecellular crosstalk between TNF-a released from IFN-c-stimulated M. abscessus-infected macro-phages and PGE2 produced by MSCs was necessary for the mycobacterial-killing activity of themacrophages. Finally, the importance of increased NO production in response to MSC administra-tion was confirmed in the mouse M. abscessus infection model. Our results suggest that MSCs mayoffer a novel therapeutic strategy for treating this drug-resistant mycobacterial infection byenhancing the bacterial-killing power of macrophages. STEM CELLS 2016; 00:000—000

SIGNIFICANCE STATEMENT

Mycobacterium abscessus is resistant to many antibiotics, thus making it difficult to obtain asuccessful treatment outcome. Here, we demonstrated that the administration of mesenchymalstem cell (MSCs) has significant potential in the treatment of M. abscessus infection. We foundthat the systemic injection of MSCs improved mouse survival and enhanced bacterial clearancein the lungs. In subsequent in vitro studies, we discovered that MSCs increased NO productionin M. abscessus-infected macrophages through NF-jB activation in the presence of IFN-c. Wefurther demonstrated that the cellular crosstalk between TNF-a and PGE2 was essential for theincreased NO production in macrophages.

INTRODUCTION

Mesenchymal stem cells (MSCs) are multipo-tent stromal cells with the capacity to differ-entiate into multiple cell types such asosteoblasts, adipocytes, and chondroblasts[1]. Previous studies have investigated thetherapeutic potential of MSCs for the treat-ment of various conditions, including bone/cartilage diseases, graft-versus-host disease,autoimmune diseases, liver diseases, and can-

cer [1–3]. These works mostly focused onthe immune-regulatory properties and tissue-reparative functions of MSCs. MSCs expressimmunosuppressive molecules and variousgrowth factors that facilitate tissue repairand maintain immune homeostasis [1, 4].The immunosuppressive properties of MSCsare well characterized, but the immune-enhancing ability of MSCs is not welldefined. In certain immune environments,particularly when low-level inflammation

aDepartment of Microbiologyand Institute for Immunologyand Immunological Diseases,Yonsei University College ofMedicine, Seoul, South Korea;bAnimal Stem Cells ResearchLab, Animal and PlantQuarantine Agency, Anyang-si,Gyeonggi-do, South Korea;cDivision of Pulmonary andCritical Care Medicine,Department of Medicine,Samsung Medical Center,Sungkyunkwan UniversitySchool of Medicine, Seoul,South Korea; dDepartment ofImmunology, Lab of DendriticCell Differentiation &Regulation, School ofMedicine, Konkuk University,Chungju, South Korea

Correspondence: Yeong-Min Park,M.D., Ph. D., Department of Immu-nology, Laboratory of Dendritic CellDifferentiation and Regulation,School of Medicine, Konkuk Univer-sity, Chungju 380-701, South Korea.Telephone: 1 82-2-2049-6158; Fax:1 82-2-2049-6192; e-mail:immun3023@kku.ac.kr; or Sung JaeShin, D.V.M., Ph. D., Department ofMicrobiology, Yonsei University Col-lege of Medicine, 50 Yonsei-ro, Seo-daemun-gu, Seoul 120-752, SouthKorea. Telephone: 1 82-2-2228-1813; Fax: 1 82-2-392-9310;e-mail: sjshin@yuhs.ac

Received June 12, 2015;accepted for publicationFebruary 12, 2016; firstpublished online in STEM CELLS

EXPRESS March 6, 2016.

VC AlphaMed Press1066-5099/2016/$30.00/0

http://dx.doi.org/10.1002/stem.2353

STEM CELLS 2016;00:00–00 www.StemCells.com VC AlphaMed Press 2016

TRANSLATIONAL ANDCLINICAL RESEARCH

exists, MSCs enhance, rather than suppress, the immuneresponse [1, 5–7].

Although the immunomodulatory properties of MSCs havebeen extensively investigated, the association of MSCs withinfectious diseases still remains unknown. Previous studieshave provided evidence of the beneficial effects of MSCs insepsis models following lipopolysaccharide (LPS) injection orcecal ligation and puncture [8–10]. In addition, other studieshave demonstrated that MSCs have the ability to clear extrac-ellular E. coli by increasing the antimicrobial peptide LL-37, byupregulating lipocalin 2 production or by enhancing the phag-ocytic activity of host immune cells [11, 12].

On the other hand, Mycobacterium tuberculosis (Mtb)uses MSCs to evade the host immune system because theyprovide a niche for maintaining dormancy [13, 14]. Mycobac-

terium leprae (ML) reprogram adult Schwann cells to becomestem-like cells and use the plasticity of the resulting cellularniche to promote the dissemination of the infection [15]. Inaddition, Chlamydia trachomatis evades host immunity byinhibiting NO production in human MSCs (hMSCs) [16].

However, whether MSCs have a beneficial effect in infec-tions with Mycobacterium species (spp.) remains unknown.Mycobacterium abscessus (M. abscessus sensu stricto) is anintracellular pathogen that belongs to the rapidly growingfamily of Mycobacterium spp. M. abscessus is an emergingrespiratory pathogen that causes pulmonary infections inpatients with cystic fibrosis and chronic lung diseases such asbronchiectasis [17–20]. Furthermore, M. abscessus is resistantto many antibiotics, thus making it difficult to obtain a favor-able treatment outcome. Unfortunately, there are no reliabletreatment regimens for M. abscessus infection, and bettertreatment options are urgently needed [17–20].

Recent studies have shown that M. abscessus can main-tain an infectious status in certain immune-deficient murinemodels, indicating that M. abscessus infection can be immu-nologically controllable if a key immune responder exists [21,22]. Thus, an alternative treatment approach to antibiotictherapy is needed for relatively low virulent but highly drug-resistant Mycobacterium spp. such as M. abscessus. BecauseMSCs have a strong potential to alter the host immuneresponse to an immunological condition, we investigated thetherapeutic potential of MSCs in a murine model and in themacrophage system of M. abscessus infection. We hypothe-sized that MSCs may control M. abscessus growth via enhanc-ing the mycobacterial killing activity of MSCs. We alsoinvestigated whether the crosstalk between MSCs and macro-phages can kill M. abscessus, and we identified the criticalfactors of MSCs that enhance the intracellular killing activityof macrophages at the molecular level. Finally, we confirmedthe therapeutic and mechanistic effects of MSC in a murineM. abscessus-infection model.

MATERIAL AND METHODS

Ethics Statement

All animal experiments were performed in accordance withthe Korea Food and Drug Administration (KFDA) guidelines.The experimental protocols used in this study were reviewedand approved by the Ethics Committee and Institutional Ani-mal Care and Use Committee (Permit Number: 2013-0421-3)

of the Laboratory Animal Research Center at Yonsei UniversityCollege of Medicine (Seoul, Korea).

Reagents and Antibodies

Anti-iNOS Ab, antiphosphorylated p38 Ab, antiphosphorylatedSTAT-1 Ab and anti-IRF-1 Ab were purchased from Cell SignalingTechnology, Inc. (Danvers, MA, www.cellsignal.com). Anti-GAPDH Ab, anti-Lamin B Ab and anti-Cox-2 Ab were obtainedfrom Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, www.scbt.com). HRP-conjugated anti-mouse IgG Ab and HRP-conjugatedanti-rabbit Ab were obtained from Calbiochem (San Diego, CA,www.merckmillipore.com), and anti-b-actin Ab was purchasedfrom Sigma (St. Louis, MO, www.sigmaaldrich.com). TNF-a, IL-12p70, IL-10, IL-6, IL-4, IL-17, MCP-1, and IFN-c levels were ana-lyzed using ELISA kits purchased from eBioscience (San Diego,CA, www.ebioscience.com). PGE2 levels were analyzed usingELISA kits purchased from R&D System (Minneapolis, MN,www.rndsystems.com). FITC-conjugated anti-CD11b Ab, PE-cy7-conjugated anti-CD11c Ab, Alexa Fluor 700-conjugated anti-Gr-1Ab, Pacific blue-conjugated anti-CD3 Ab, PerCP 5.5-conjugatedanti-CD4 Ab, APC-eFluor780-conjugated anti-CD8 Ab, APC-conjugated anti-F4/80 Ab and TNF-a-neutralizing Ab were pur-chased from eBioscience. NG-monomethyl-L-arginine monoace-tate salt (L-NMMA), PGE2, SB203580, Bay-11-7082, andcelecoxib were purchased from Sigma.

Mouse MSC Isolation, In Vitro Expansion

C57BL/6 and iNOS2/2 (NOS2tm1Lau) purchased from Jackson lab-oratory (Bar Harbor, ME, www.jax.org). COX-22/2 mice werekindly provided by professor Young Myeong Kim (KangwonNational University, South Korea). Bone marrow cells were col-lected from 6-week-old mice that were killed by CO2-inhalation.Their femurs and tibiae were carefully cleaned of adherent softtissue and harvested by flushing with Dulbecco’s modified Eagle’smedium (DMEM) (HyClone, Logan, UT 84321, low glucose, www.gelifesciences.com) supplemented with 10% heat-inactivatedfetal bovine serum (FBS; HyClone) and penicillin/streptomycin(100 U/ml and 100lg/ml; HyClone). After washing and centrifu-gation at 400g for 5 minutes, mononucleated cells were isolatedfrom marrow cells followed by density-gradient centrifugation(1.077 g/cm3) using HISTOPAQUE-1077 (Sigma). After then, cellswere immunodepleted of granulomonocytic cells using CD11b1

micromagnetic beads (Miltenyi Biotec, Bergisch-Gladbach, Ger-many, www.miltenyibiotec.com). The residual cells were culturedin DMEM (Hyclone) containing 10% FBS and penicillin/streptomy-cin (100 U/ml and 100lg/ml) supplemented at 378C in a 5% CO2

humidified incubator. The nonadherent cells were removed after72–96 hours, adherent cells were harvested by trypsinization(0.05% trypsin-EDTA, Gibco-Invitrogen, Carlsbad, CA, www.ther-mofisher.com) when reaching 90% confluence, and thenreplated. The medium was changed every 3–4 days.

Murine Bone Marrow-Derived Macrophage Generation

Murine bone marrow-derived macrophages (BMDMs) wereobtained as previously described [23]. Briefly, female C57BL/6Jmice were euthanized by CO2 asphyxiation. Bone marrow cellswere cultured for 6 days in DMEM (HyClone, Logan, UT) con-taining 100 U/ml penicillin, 100lg/ml streptomycin, 10% FBS(Invitrogen, Carlsbad, CA, www.thermofisher.com), and 20 ng/ml recombinant mouse macrophage colony stimulating factor(M-CSF; R&D Systems) at 378C in the presence of 5% CO2.

2 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

In Vitro Coculture Studies of MSCs and BMDMs

To directly prepare mixed coculture systems, MSCs andBMDMs were suspended in RPMI1640 supplemented with10% FBS, 100 U/ml penicillin and 100lg/ml streptomycin.BMDMs were plated in 24-well plates at a density of 33 105

cells/wells. After 24 hours, MSCs (33 104 cells) were addedto the plates (BMDM:MSC ratio of 10:1). To prepare coculturesystems using transwell plates, MSCs (33 104 cells) wereadded into transwells (0.4 lm pore size) in 24-well plates ofcultured BMDMs (BMDM:MSC ratio of 10:1). To produceMSC-conditioned medium (CM), MSCs (33 104 cells) wereplated in 24-well plates for 24 hours. The supernatants werecollected, centrifuged and stored at 2308C.

Bacterial Culture

M. abscessus culture was performed as previously described[22, 24]. The M. abscessus ATCC19977T strain was obtainedfrom the American Type Culture Collection (ATCC; Manassas,VA). M. abscessus ATCC19977T originally showed a smoothmorphotype, and an isogenic rough variant (M. abs-R) wasisolated from the lungs of a C57BL/6J mouse infected withthe smooth morphotypic strain of M. abscessus ATCC19977T

during macrolide treatment. The rough isogenic strains of M.

abscessus ATCC19977T were confirmed to have an identicalgenotype by variable number tandem repeat analysis [24]. Forred fluorescent protein (RFP) expression, M. abs-R was trans-formed with pMV261–ERFP.

Animals and Infection

In an initial animal infection study conducted to analyze thein vivo effect of MSCs on bacterial colony-forming units(CFUs) in the lungs and spleens of M. abs-R-infected mice,C57BL/6J mice were infected with aerosolized M. abs-R(4.03 105 CFUs/mouse). At specific time points post-infection,the mice were euthanized, and the lungs and spleens wereaseptically collected for bacterial counts. In some experi-ments, to analyze the in vivo effect of MSCs on the survivalrate, mice were infected via the intravenous injection of M.

abs-R (1.63 107 CFUs/mouse) through the tail vein. The num-bers of viable bacteria in the organs were assessed by platingserial dilutions of the organ homogenates onto 7H10-OADCagar plates. Colonies were counted after 3–5 days of incuba-tion at 378C. To inhibit NO production in mice, C57BL/6J micewere given a 0.25% solution of aminoguanidine hydrochloride(AG; Sigma) in sterilized drinking water ad libitum beginning 1day prior to M. abs-R infection.

Collection of Bronchoalveolar Lavage (BAL) Fluids andBAL Cells

BAL fluid samples were harvested by infusing 1ml of ice-coldPBS through a 25-gauge blunt needle into the lungs via thetrachea [22]. This procedure was followed by gently aspiratingthe fluid into a syringe at 5 and 10 days post-infection. BALfluids and BAL cells were separated using centrifugation at13,000g for 10 minutes at 48C.

Flow Cytometric Analysis

Cell surface staining was performed with specifically labeledfluorescent-conjugated Abs, as recently described [25]. Thefluorescence was measured by flow cytometry (FACSVerse, BD

Biosciences, www.bdbiosciences.com), and the data were ana-lyzed using FlowJo data analysis software (TreeStar, Inc., www.flowjo.com).

Enzyme-Linked Immunosorbent Assay (ELISA)

Cell culture supernatants or bronchoalveolar lavage fluid(BALF) samples from mice were collected, sterile-filtered, andstored at 2808C until use. Cytokine levels were measuredusing ELISAs with commercial reagent kits, following the man-ufacturers’ instructions.

Bacterial Survival in BMDMs

BMDMs were infected with M. abscessus for 4 hours at aratio of one CFU per macrophage. The infected BMDMs werewashed with DMEM to remove extracellular bacteria and cul-tivated at 378C in 5% CO2. After incubation, the cells werelysed with 0.05% Triton-X 100, and the lysates were plated in10-fold serial dilutions onto 7H10 agar (Becton Dickinson,Franklin Lakes, NJ, www.bdbiosciences.com) to quantify thenumber of viable bacteria. Colonies were counted after 3–5days of incubation at 378C.

Immunoblotting Analysis and Subcellular Fractionation

Immunoblotting analysis and subcellular fractionation wereperformed as previously described [26].

Reverse Transcription-PCR Analysis

Reverse transcription (RT)-PCR was performed as previouslydescribed [27]. The following primer sequences were used:iNOS, 50-gtttctggcagcagcggctc-30 (sense) and 50-gctcctcgctcaagttcagc-30 (antisense); b-actin, 50-cgtgggccgccctaggcaccaggg-30

(sense) and 50-gggaggaagaggatgcggcagtgg-30 (antisense); Cox-2,50-tctccaacctctcctactac-30 (sense) and 50-gcacgtagtcttcgatcact-30

(antisense); IL-6, 50-atgaactccttctccacaagcgc-30 (sense) and 50-gaagagccctcaggctggactg-30 (antisense); and IL-10, 50-ccagttttacctggtagaagtgatg-30 (sense) and 50-tgtctaggtcctggagtccagcagctc-30

(antisense).

Immunocytochemistry

The cells were then fixed in 4% paraformaldehyde and perme-abilized in 0.1% Triton X-100. The samples were then blockedwith CAS-Block (Invitrogen) before being incubated with anti-iNOS Ab, anti-F4/80 Ab for 2 hours at room temperature (RT).The samples were washed with phosphate-buffered salinewith 0.1% Tween-20 (PBS/T), incubated with Alexa Fluor 488-conjugated secondary antibody for 1 hour and then stainedwith 1lg/ml DAPI for 10 minutes at RT. Samples wereobserved using a confocal laser scanning microscope (FV1000;Olympus, Japan, www.olympus-lifescience.com).

Immunohistochemistry

At the end of the experiments, tissues were fixed overnight informalin and embedded in paraffin. The sections were depar-affinized in xylene and washed in PBS. After washing, the sec-tions were subjected to antigen retrieval in boiling citric acidbuffer (10mM, pH 6), and nonspecific proteins were thenblocked with CAS-Block. The slides were then incubated withanti-iNOS Ab and anti-F4/80 Ab at 48C overnight. The slideswere washed with PBS/T and then incubated with anti-rabbitIgG Alexa Fluor 488-conjugated secondary antibody (Invitro-gen) for 1 hour at RT. The cells were also stained with DAPI

Kim, Cha, Kim et al. 3

www.StemCells.com VC AlphaMed Press 2016

(Sigma) and Texas Red-phalloidin (Invitrogen) to visualize cellshape and were then observed using an FV1000.

Nitric Oxide Measurement

NO production was monitored in the extracellular medium ofmacrophages by assessing the nitrite levels using commercialGriess reagent kits, following the manufacturer’s instructions.The samples were measured at 540 nm on a microplatereader (Epoch, BioTek, Winooski, VT, www.biotek.com).

Statistical Analysis

All experiments were repeated at least three times with con-sistent results. Significant differences between samples wereassessed with Tukey’s multiple comparison tests using Graph-Pad Prism software (San Diego, CA). The data in the graphsare expressed as the mean6 SEM. *, p< .05; **, p< .01 and***, p< .001 were considered statistically significant. Survivalcurves were analyzed using the Kaplan-Meyer log-rank test.

RESULTS

Generation and Characterization of MSCs

The cell morphotypes were observed to confirm that the iso-lated bone marrow-derived MSCs had a typical phenotype. Thecells were highly proliferative in in vitro culture and exhibitedarchetypal fibroblastic morphology (Supporting Information Fig.

S1A). Flow cytometric analysis demonstrated that more than90% of the cells were Sca-1-, CD29-, CD44- and CD90-positive,while less than 1% were CD117- and CD45-positive (SupportingInformation Fig. S1B). To confirm their multipotent ability, MSCswere exposed to differentiation medium. Cell type-specificstaining showed that the MSCs differentiated into adipocytesand chondrocytes (Supporting Information Fig. S1C).

MSCs Enhance M. abscessus Clearance and ImproveMouse Survival

Based on colony morphology on solid growth media, thereare two variants of M. abscessus, designated as the smooth(S) and rough (R) morphotypes [28]. The R variant ofM. abscessus is considered more pathogenic in humans [22,29, 30]. Therefore, we used a hypervirulent rough variant ofM. abscessus (M. abs-R) in this study. To test the effect ofMSCs on mycobacterial growth, we aerogenically infectedC57BL/6J mice with M. abs-R (4.03 105 CFUs/mouse), andMSCs (2.53 105 cells/mouse) were intravenously administeredone day post-infection (Fig. 1A). As shown in Fig. 1B, MSCsstrongly inhibited bacterial CFUs in the lungs at 10 days post-infection. Furthermore, MSCs also significantly repressed M.

abs-R dissemination from the primary infection site of thelungs to the spleen (Fig. 1C).

Previous studies have thoroughly documented the inabilityof M. abscessus aerosol or intratracheal infection models toinduce lethality in mice [21, 31, 32]. According to our

Figure 1. MSCs inhibit M. abs-R growth and improve survival. (A): Schematic diagram of the experiment design for M. abs-R aerosolinfection. C57BL/6J mice were aerogenically infected with 53 104 M. abs-R cells, and MSCs (2.53 105) were then injected via the tailvein 1 day after infection. The lungs (B) and spleens (C) were harvested at the indicated time points, and the mycobacterial loads weremeasuring using a standard plate count method. The results are expressed as the mean estimated CFU6 SEM from the lung samples offive to six mice in three independent experiments. Significant differences in CFUs are shown for MSC-injected mice compared with PBS-treated mice (***, p< .001). (D): Schematic diagram of the experimental design for the mouse survival test. (E): C57BL/6J mice wereintravenously infected with 1.63 107 M. abs-R cells, and MSCs (2.53 105) were then injected via the tail vein 1 day after infection. Sur-vival was monitored for 14 days. The results are representative of three independent experiments. (F): The lungs were harvested at day10 after infection, and the mycobacterial loads were measured using a standard plate count method. The results are expressed as themean estimated CFU6 SEM from the lung samples of five to six mice in three independent experiments. Significant differences in CFUsare shown for the MSC-injected group compared with the PBS-treated group (*** p< .001). Abbreviations: CFUs, colony-forming units;MSCs, mesenchymal stem cells; PBS, phosphate-buffered saline.

4 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

previous study, M. abs-R could induce mice lethality by theintravenous route [22]. Therefore, to investigate the effect ofMSCs on the survival of M. abs-R-infected mice, we intrave-nously infected C57BL/6J mice with abs-R (1.63 107 CFUs/mouse). At one day post-infection, MSCs were intravenouslyadministered with a low (13 105 cells/mouse) or high(2.53 105 cells/mouse) cell count (Fig. 1D). We then exam-ined the mouse survival rates and bacterial CFUs in the lungs.MSC-injected mice had a significantly higher rate of survivalover 15 days in a dose-dependent manner compared with thePBS-injected group (Fig. 1E). The bacterial CFUs in the lungswere decreased by more than two log scales when the micewere injected with a high dose of MSCs (Fig. 1F).

To investigate whether MSCs could affect the productionof inflammatory cytokines and the recruitment of immunecells into the bronchoalveolar space following aerogenic M.

abs-R infection, BALF was obtained from mice at day 5 post-infection. The cells within the BALF were stained and countedto determine the type of cells recruited into the bronchoal-veolar space. As shown in Fig. 2A-2D, the recruitment of F4/80-CD11bmidCD11cmid monocytes, F4/801CD11bhighCD11clow

macrophages, CD41 and CD81 T cells into the bronchoalveo-lar space was significantly elevated by MSC administration. Inaddition, IFN-c, TNF-a, IL-6, MCP-1, NO, and PGE2 levels in theBALF were also significantly increased by MSCs (Fig. 2E). How-ever, IL-12p70, IL-4, IL-10, and IL-17 levels were not changedfollowing MSC administration.

We next examined immune responses and the recruit-ment of immune cells into the bronchoalveolar space at 10days post-infection. Overall, the characteristics of the cytokineprofiles and the recruitment of immune cells were similar at5 and 10 days post-infection, but the numbers of each typeof cytokine and infiltrated immune cell were lower than 5days post-infection (Supporting Information Fig. S2A, S2B). Incontrast, IL-10 levels were higher in the BALF at day 10 thanat day 5 post-infection (Supporting Information Fig. S2A).

MSCs Inhibit M. abs-R Growth in BMDMs

Although the above results indicated that MSCs could sup-press M. abs-R growth in mice, we could not exclude the pos-sibility that MSCs have a direct killing effect on bacteria thatenhances their ability to control bacterial growth in macro-phages. To confirm this possibility, we infected BMDMs withM. abs-R (multiplicity of infection, MOI5 1) and measuredintracellular growth for 72 hours. We tested the effect ofMSCs on M. abs-R growth in BMDMs under three differentconditions, as follows: a direct contact coculture condition toassess the influence of cell-cell interactions, an indirect cocul-ture condition using transwell plates to identify soluble factorcrosstalk, and a culture condition with MSC CM to identifysoluble factor crosstalk. Unexpectedly, MSCs did not inhibitthe growth of M. abs-R in BMDMs in any of the three condi-tions (Fig. 3A), suggesting that additional factors wererequired to inhibit bacterial growth in these conditions. IFN-c,an important activator of macrophages, is a cytokine that iscritical for innate and adaptive immunity against bacterialinfection, including M. abscessus [21]. Thus, we examinedwhether IFN-c treatment could alter the unexpected results inthe three conditions described above. Interestingly, IFN-ctreatment resulted in suppressed M. abs-R growth in all threeconditions. The strongest antibacterial effect was found when

the MSCs were directly cultured with the M. abs-R-infectedBMDMs. Moreover, bacterial growth was also significantlyinhibited when MSC CM was added to the M. abs-R-infectedBMDMs (Fig. 3B).

Because the strongest antibacterial effect was observedwhen the MSCs were directly cultured with the infectedBMDMs, we sought to determine whether MSCs directlyattach to macrophages. As shown in Supporting InformationFig. S3B, we found that F4/80-negative MSCs were directlyattached to F4/80-positive BMDMs. In addition, MSCs stronglyinhibited M. abs-R-RFP growth in BMDMs (Supporting Infor-mation Fig. S3). However, we did not identify any specificcell-cell chemotactic interaction between MSCs and M. abs-Rinfected BMDMs.

It has been shown that murine and hMSCs are largely dif-ferent in terms of immunomodulatory processes [1]. Wetherefore examined whether hMSCs could also inhibit bacte-rial growth in the THP-1 human macrophage cell line. Asshown in Supporting Information Fig. S4A, when grown in thepresence of IFN-c, hMSCs inhibited M. abs-R growth.

Therefore, MSCs inhibit M. abs-R growth in macrophagesin the presence of IFN-c. In addition, the MSCs secrete spe-cific soluble factors that are involved in regulating the intra-cellular pathogens, which are controlled by cellular crosstalkbetween soluble factors and cell-cell interactions.

MSCs Increase NO Production in M. abs-R-InfectedBMDMs Through Upregulating NF-jB Signaling

To investigate the macrophage-killing mechanisms that arepotentially responsible for the effects of treatment with MSCs,we analyzed nitric oxide (NO) production using a transwell sys-tem. As shown in Fig. 4A, NO production was significantlyincreased when MSCs were cocultured in the presence of IFN-c.Because MSCs also produce NO, we measured NO levels inMSCs cultured without macrophages. However, low levels ofNO were detected in the media of the MSCs that were culturedalone, and NO levels were slightly increased when cells weregrown in the presence of IFN-c (Fig. 4A). Next, we examinedinducible NO synthase (iNOS) expression in BMDMs and MSCs.After BMDMs and MSCs were cocultured, we divided the cellsand subsequently measured iNOS expression. As shown in Fig.4B, the iNOS mRNA and protein levels in the BMDMs were sig-nificantly higher when the cells were cocultured with MSCs inthe presence of IFN-c. However, iNOS mRNA and protein levelswere undetectable in MSCs when we loaded the same amountsof protein and mRNA that we used to detect their expression inmacrophages. We further confirmed these results using immu-nocytochemistry (Fig. 4C).

We further investigated NO production under the threedifferent culture conditions described above. As expected, thehighest levels of NO production were observed in MSCsdirectly cocultured with M. abs-R-infected BMDMs. Noincreases in production occurred in the conditions in whichMSCs were cocultured with M. abs-R-infected BMDMs usingtranswell plates and in which MSCs were cultured with CM.Interestingly, higher NO levels were detected in MSC transwellcoculture compared with CM treatment (Supporting Informa-tion Fig. S5A). These results suggested potential soluble factorcrosstalk. To confirm this possibility, we first collected CMfrom BMDMs with or without M. abs-R infection in the pres-ence of IFN-c and then cultured MSCs with the CM for 24

Kim, Cha, Kim et al. 5

www.StemCells.com VC AlphaMed Press 2016

hours. After incubation, we collected the MSC CM and used itto treat M. abs-R-infected-BMDMs in the presence of IFN-cbefore measuring NO production. As shown in SupportingInformation Fig. S5B, the CM from MSCs stimulated with M.

abs-R-infected BMDM CM produced a higher amount of NOthan MSC CM from incubation with uninfected BMDM CM.However, a small amount of NO was detected in all of theCM samples (Supporting Information Fig. S5C).

Next, we investigated whether hMSCs could also induceNO production in M. abs-R-infected THP-1 cells. Although arelatively smaller amount (less than 5 mM) was observed inthese cultures than was observed in the mouse systems, theresults showed a trend similar to that described above (Sup-porting Information Fig. S4B).

To investigate whether the upregulation of iNOS improvedthe ability of macrophages to kill M. abs-R, we used the NO

Figure 2. Effects of MSCs on cytokine production and cellular influx into the bronchoalveolar space. C57BL/6J mice were aerogenicallyinfected with 53 104 M. abs-R cells, and MSCs (2.53 105) were then injected via the tail vein 1 day after infection. (A): Immune cellsin the lungs were harvested at day 5 post-infection, and the cells were assayed by flow cytometry. The representative dot-blot was pri-marily gated on FSCmid/high/SSCmid/high and secondarily gated on F4/80high or F4/80low. F4/80high/CD11bhigh/CD11clow/mid cells were classi-fied as CD11bhigh macrophages, and F4/80low/CD11blow/CD11chigh cells were classified as alveolar macrophages. F4/80low/CD11bmid/CD11cmid cells were classified as monocytes, and F4/80low/CD11bhigh/CD11chigh cells were classified as CD11bhigh dendritic cells. F4/80low/CD11blow/CD11chigh cells were classified as CD11blow dendritic cells. (B): CD41 T cell and CD81 T cell influx into the lungs. The rep-resentative dot-blot of bronchoalveolar lavage cells was primarily gated on FSCmid/SSClow and secondarily gated on CD3high. The numbersindicate the percentages of cells in each quadrant. (C): A representative dot-blot was primarily gated on FSCmid/high/SSCmid/high and sec-ondarily gated on CD11clowF4/80low. F4/80low/CD11clow/CD11Bhigh/Gr-1high cells were classified as neutrophils. (D): The bar graph repre-sents the absolute number of cells in each quadrant or square on day 5 post-infection. The results are presented as the means6 SEMof each group (*, p< .05; **, p< .01, or ***, p< .001). (E): BALF was harvested at the indicated time points, and IFN-c, TNF-a, IL-6,MCP-1, IL-12p70, IL-4, IL-10, and IL-17 levels were measured by ELISA. The results are presented as the means6 SEM of each group(*, p< .05 and **, p< .01). Abbreviation: MSC, mesenchymal stem cells.

6 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

synthase inhibitor L-NMMA. The addition of L-NMMA to theinfected BMDMs cocultured with MSCs caused almost com-plete inhibition of NO production and intracellular killingactivity against M. abs-R (Fig. 4A, 4D).

Interestingly, we found that adding L-NMMA to M. abs-R-infected THP-1 cells that were cocultured with hMSCs alsocompletely inhibited NO production but only partially inhib-ited intracellular killing activity against M. abs-R (SupportingInformation Fig. S4C).

Previous studies have shown that iNOS expression of macro-phage is regulated by signal integration between IFN-c and TLR,which are involved in the interactions among STAT-1, interferon-regulated factor 1 (IRF-1), p38 and NF-jB [33, 34]. Because thedata showed that MSCs induced iNOS expression in BMDMs, weexamined the associated signal transduction pathway. As shownin Fig. 4E, MSCs induced IjBa degradation and p65 translocationfrom the cytosol to the nucleus but did not alter p38 phospho-rylation, STAT-1 phosphorylation or IRF-1 expression. Further-more, MSC-induced NO production was attenuated by the NF-jB inhibitor Bay-11-7082 but not by the p38 inhibitor SB203580(Supporting Information Fig. S6A, S6B). Consistent with theabove results, NF-jB inhibitor treatment resulted in a marked

inhibition of the MSC-induced enhancement of BMDM killingactivity against M. abs-R (Fig. 4F). Taken together, these resultsindicate that MSC-induced NO production in infected BMDMs ismediated by NF-jB but not by p38 or STAT1-IRF1.

Effect of MSCs on Cytokine Productionof M. abs-R-Infected BMDMs

We hypothesized that MSCs might alter the cytokine produc-tion of M. abs-R-infected BMDMs. Therefore, we assessed thelevels of TNF-a, IL-6, IL-12p70, IL-10, and PGE2 produced byBMDMs cocultured with MSCs in transwell plates. When theBMDMs were infected with M. abs-R at an MOI of 1 in theabsence of IFN-c, low levels of TNF-a, IL-6, IL-12p70, and PGE2were detected in the culture medium. However, high levels ofTNF-a, IL-6, and IL-12p70 were produced from M. abs-R-infected BMDMs in the presence of IFN-c. IL-10 was alsodetected in the culture medium of infected BMDMs but wasnot affected by IFN-c. Interestingly, the coculture of infectedBMDMs with MSCs strongly increased the production levels ofIL-6 and PGE2, but the levels of TNF-a and IL-12p70 in the pres-ence of IFN-c and IL-10 remained unchanged in this condition(Supporting Information Fig. S7A–S7E). As previous studies

Figure 3. Effects of MSCs on the intracellular growth of M. abs-R. BMDMs were infected with an multiplicity of infection of 1 bacte-rium per cell, and the unbound bacteria were removed with PBS after 4 hours. The cells were incubated with MSCs using three differentconditions in the absence (A) or presence (B) of IFN-c (10 ng/ml), as described in the “Materials and Methods.” The number of bacterialcolony forming units (CFUs) was determined at the indicated time points after removing the supernatant. The data are presented as themeans6 SEM of three replicates of two experiments. Significant differences in CFUs are shown for the MSC cocultured group comparedwith the PBS-treated group (**, p< .01; ***, p< .001). Abbreviations: BMDMs, bone marrow-derived macrophages; CFUs, colony-forming units; MSCs, mesenchymal stem cells; PBS, phosphate-buffered saline.

Kim, Cha, Kim et al. 7

www.StemCells.com VC AlphaMed Press 2016

have shown that MSCs produce several types of cytokines [35],we measured the production of the above-mentioned cytokinesfrom MSCs. As shown in Supporting Information Fig. S7A-S7E,MSCs secreted high levels of IL-6, IL-10 and PGE2 and low levelsof TNF-a and IL-12p70. Because IL-6, IL-10, and PGE2 are pro-duced by both cell types, we examined IL-6 and Cox-2 mRNAlevels in BMDMs and MSCs, respectively, in our coculture sys-tem. As shown in Supporting Information Fig. S7F, IL-6 expres-sion levels were increased in both cell types and were higher incells grown with IFN-c. IL-10 was expressed mainly in BMDMs,whereas Cox-2 was detected mainly in MSCs that were cocul-tured with M. abs-R-infected BMDMs in the presence of IFN-c.

MSC-Derived PGE2 is a Key Regulatorof NO Production by BMDMs

Because IL-6 levels were synergistically increased in our cocul-ture system, we tested whether IL-6 was important for the

action of MSCs by treating M. abs-R-infected BMDMs withrecombinant IL-6 protein or a monoclonal antibody againstthe a-chain of the IL-6 receptor to block IL-6 signal transduc-tion. Unexpectedly, neither recombinant IL-6 nor the mono-clonal antibody against the IL-6 receptor affected theproduction of NO in infected BMDMs (Supporting InformationFig. S8A, S8B). These results suggested that IL-6 is not a majormediator of NO production in this system.

Thus, we focused on PGE2, which was also synergisticallyincreased in our coculture system. We examined the effect ofcelecoxib, a specific PGE2 inhibitor, on NO production. Asexpected, PGE2 inhibition resulted in a significant reduction ofNO production in our coculture system. In addition, thisdecrease was recovered by adding exogenous PGE2 (Fig. 5A).The anti-bacterial effects of MSCs were completely attenuatedby treatment with celecoxib, and this attenuation recoveredwhen exogenous PGE2 was added (Fig. 5B). To further examine

Figure 4. MSCs enhance NO production in M. abs-R-infected BMDMs via the upregulation of NF-jB signaling. BMDMs were infectedwith an multiplicity of infection (MOI) of 1 bacterium per cell, and the cells were cocultured with MSCs using transwell (BMDM:MSCratio of 10:1) in the presence of IFN-c (10 ng/ml) or L-NMMA (300 lM). (A): NO was measured in the extracellular medium of cells cul-tured for 24 hours by assessing nitrite levels using the Griess reagent. (B): The iNOS mRNA and protein levels were measured using RT-PCR and immunoblotting assays, respectively, in the BMDMs and MSCs. (C): Representative immunofluorescence staining of iNOS (scalebar5 10 lm). (D): BMDMs were infected with an MOI of 1 bacterium per cell and cocultured with MSCs using transwell plates in thepresence of IFN-c or L-NMMA (300lM). The CFUs were counted 72 hours post-infection. (E): BMDMs were infected with an MOI of 1bacterium per cell and cocultured with MSCs using transwell in the absence or presence of IFN-c for 24 hours. The levels of phospho-p38, phospho-STAT-1, IRF-1, IjBa, GAPDH and translocation of p65 were assessed by immunoblotting. Lamin B and b-actin were used asmarkers for the nuclear and cytosolic fractions, respectively. (F): M. abs-R-infected BMDMs were cocultured with MSCs in the presenceof a Bay-11-7082 at concentrations of 5, 10, and 20lM, respectively. All cells were treated with IFN-c, and the CFUs within BMDMswere measured at 72 hours. The data are presented as the means6 SEM of three replicates of two experiments (**, p< .01). Abbrevia-tions: BMDMs, bone marrow-derived macrophages; CFUs, colony-forming units; L-NMMA, NG-monomethyl-L-arginine monoacetate salt;MSCs, mesenchymal stem cells.

8 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

the effect of PGE2 on the production of NO and on thebacterial CFUs in the presence of IFN-c, we infected BMDMswith M. abs-R and added exogenous PGE2 without MSCs. Withthe addition of PGE2 in the presence of IFN-c, NO productionsignificantly increased, and bacterial CFUs gradually decreased,as shown in Fig. 5C, 5D. Therefore, these results suggestedthat PGE2 is a major mediator of NO production in thissystem.

The Cellular Basis for BMDM-MSC Interactions

To understand the molecular mechanism of the increase inNO production caused by MSCs through PGE2 in M. abs-R-infected BMDMs, we focused on TNF-a because it was greatlychanged by IFN-c (Supporting Information Fig. S7A). To investi-gate whether TNF-a plays a key role in BMDM-MSC crosstalk,we measured the level of NO production in the coculturemedium from cells treated with or without a TNF-a neutraliz-ing antibody. The amount of NO was significantly decreased inthe TNF-a-neutralized group, and this effect was recoveredwith the addition of exogenous PGE2 (Fig. 6A). Similarly, whenM. abs-R-infected BMDMs were cocultured with MSCs in thepresence of the TNF-a neutralizing antibody, the CFUs werenot reduced. Moreover, the CFUs were reduced in this condi-

tion when exogenous PGE2 was added (Fig. 6B). These resultssuggest that the IFN-c-stimulated upregulation of TNF-a in thecocultures plays a key role in BMDM-MSC crosstalk.

Therefore, to investigate whether PGE2 secretion fromMSCs is regulated by TNF-a levels, we prepared CM fromBMDMs with or without M. abs-R infection and cultured inthe presence of IFN-c. After 24 hours, we collected the CMfrom the BMDMs and used it to treat MSCs with or withoutTNF-a neutralizing antibody. After 24 hours, we measuredPGE2 levels in the MSC culture medium. As shown in Fig. 6C,MSCs treated with CM from M. abs-R-infected BMDMs pro-duced higher levels of PGE2 than those treated with CM fromuninfected BMDMs. Surprisingly, these effects were com-pletely blocked by incubation with the TNF-a neutralizing anti-body. We also found significantly increased Cox-2 expressionin MSCs at 6 hours after stimulation with CM, which was col-lected from BMDMs with or without M. abs-R infection in thepresence of IFN-c. However, when the MSCs were treatedwith a TNF-a neutralizing antibody, Cox-2 expression did notchange (Fig. 6D).

To further confirm the presence of cellular cross-talk, weused Cox-22/2 and iNOS2/2 MSCs. As shown in Fig. 6E, NOproduction was observed when wild-type MSCs and iNOS2/2

Figure 5. Involvement of PGE2 in the MSC-induced upregulation of NO production. (A, B): The effects of celecoxib on NO productionand bacterial growth in BMDMs. (A): BMDMs were infected with an multiplicity of infection (MOI) of 1 bacterium per cell, and the cellswere then cocultured with MSCs using transwell in the presence of IFN-c. In some cases, cells were treated with celecoxib (10lM) orPGE2 (2 ng/ml). After 24 hours, the supernatants were collected to measure NO concentrations. (B): The numbers of bacterial CFUswithin the BMDMs were counted 72 hours post-infection. The data are presented as the means6 SEM of three replicates of two experi-ments (**, p< .01 and ***, p< .001). (C, D): The effects of adding exogenous PGE2 on NO production and bacterial growth in BMDMs.BMDMs were infected with an MOI of 1 bacterium per cell, and the cells were then treated with different concentrations of PGE2 (0.5,1, and 2 ng/ml) in the presence of IFN-c (10 ng/ml). NO was measured 24 hours post-infection (C), and bacterial CFUs were counted 72hours post-infection (D). The data are presented as the means6 SEM of three replicates of two experiments (**, p< .01 and***, p< .001). Abbreviations: BMDMs, bone marrow-derived macrophages; CFUs, colony-forming units; MSC, mesenchymal stem cell.

Kim, Cha, Kim et al. 9

www.StemCells.com VC AlphaMed Press 2016

MSCs, but not Cox-22/2 MSCs, were cocultured with M. abs-R-infected BMDMs. Likewise, Cox-22/2 MSCs did not inhibitthe growth of M. abs-R in BMDMs (Fig. 6F).

Taken together, these results suggested that the secretionof high levels of TNF-a from M. abs-R-infected BMDMsinduced by IFN-c led to PGE2 secretion by MSCs.

Effects of NO Inhibition on MSC-Induced M. abscessus

Clearance in Mice

Although our in vitro data strongly suggest that NO is themain effector of MSC-induced M. abs-R clearance, whetherNO is also an effecter of MSCs in mice is unknown. To extend

our in vitro studies, we used aminoguanidine, a selectiveiNOS inhibitor, in the M. abs-R mouse infection model.C57BL/6J mice were infected with an aerosolized strain of M.

abs-R and were then intravenously injected with MSCs at oneday post-infection with or without aminoguanidine (Fig. 7A).The iNOS inhibition resulted in a significant reduction of MSC-induced M. abs-R clearance (Fig. 7B). Furthermore, iNOSexpression in the lungs of MSC-injected mice was markedlyincreased compared with control mice (Fig. 7C).

We next examined iNOS expression in F4/801 macro-phages using immunohistochemical staining of lung tissue sec-tions. When we merged the captured images, the majority of

Figure 6. PGE2 secretion by MSCs is TNF-a dependent. (A, B): The effects of a monoclonal anti-TNF-a neutralizing antibody on NO pro-duction and bacterial growth in BMDMs. (A): BMDMs were infected with an multiplicity of infection (MOI) of 1 bacterium per cell, andthe cells were then cocultured with MSCs using transwell plates in the presence of IFN-c. In some cases, cells were treated with ananti-TNF-a neutralizing antibody (10lg/ml) or PGE2 (2 ng/ml). After 24 hours, the supernatants were collected to measure NO concen-trations. (B): The numbers of bacterial CFUs within the BMDMs were counted 72 hours post-infection. The data are presented as themeans6 SEM of three replicates of two experiments (*, p< .05 and **, p< .01). (C, D): PGE2 secretion from MSCs was regulated byTNF-a signaling. (C): Conditioned medium (CM) was collected from BMDMs with or without M. abs-R infection in the presence of IFN-cand transferred to MSCs (BMDM-CM: MSCs culture media5 1:10 ratio) in the absence or presence of an anti-TNF-a neutralizing anti-body (10lg/ml). After 24 hours, CM from MSCs was collected to measure PGE2 concentrations. The data are presented as the mean-s6 SEM of three replicates of two experiments (***, p< .001). (D): Immunoblotting analysis of Cox-2 expression in MSCs stimulatingwith CM from BMDMs in the absence or presence of an anti-TNF-a neutralizing antibody. Cox-2 and b-actin levels were measured after6 hours of incubation using specific antibodies. (E, F): BMDMs were infected with an MOI of 1 bacterium per cell, and the cells werethen cocultured with wild type, Cox-2-/- or iNOS-/- MSCs in transwell plates in the presence or absence of IFN-c. NO levels were meas-ured at 24 hours post-infection (E), and bacterial CFUs (F) were counted at 72 hours post-infection. Abbreviations: BMDMs, bonemarrow-derived macrophages; CFUs, colony-forming units; MSC, mesenchymal stem cell.

10 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

the iNOS expression was observed to colocalize with F4/801

cells when treated with wild type and iNOS2/2 MSCs(Fig. 7D).

Finally, to further confirm the results of our in vitro stud-ies, we used Cox-22/2 and iNOS2/2 MSCs in the abs-R mouseinfection model. When mice were injected with wild type andiNOS2/2 MSCs, we observed M. abs-R clearance in lung andspleens, but we did not observe clearance when mice weresimilarly treated with Cox-22/2 MSCs (Fig. 7E). These datasuggest that the NO that is produced by macrophages is amajor effector molecule that is used by MSCs to controlM. abs-R in mice.

DISCUSSION

Although NO has been described as an essential effector mol-ecule for microbial killing within macrophages, the role of NOin mycobacteria remains controversial. NO has been shown tobe essential for host defense against Mtb in murine models,

but the role of NO has not been fully established in humans[36]. For example, failure to express iNOS resulted in therapid proliferation of Mtb and early death in murine models[37]. However, NO was not found to be involved in thebacteria-killing mechanisms of M. avium-infected macro-phages [38]. Regardless of this, NO still has an important rolein defense against nontuberculous mycobacterial (NTM) dis-eases [39].

In this study, our data clearly demonstrate that NO is spe-cifically associated with host defenses that are aimed at con-trolling M. abscessus. Previous studies have shown that MSCscan generate NO, which exerts highly immunosuppressiveeffects at high concentrations via undefined mechanisms inmurine models [40]. Although IFN-c can stimulate MSCs torelease high levels of immunosuppressive factors [41, 42], wedetected less than 2mM NO in the culture medium of MSCsthat were treated with or without IFN-c. Therefore, in thisstudy, the main source of NO in the coculture medium wasmacrophages, and IFN-c did not have any effect on the pro-duction of NO by MSCs. In addition, iNOS mRNA expression

Figure 7. Effects of aminoguanidine on MSC-induced M. abs-R clearance in mice. (A): Schematic diagram of the experimental design.(B): C57BL/6J mice were infected with an aerosolized strain of M. abs-R (1.63 107 CFUs/mouse), and MSCs (2.53 105 cells/mouse)were administered by intravenous injection at 1 day post-infection. Aminoguanidine was administered at a dose of 0.25% ad libitum inthe drinking water each day, starting 1 day post-infection. The data are presented as the means6 SEM of two experiments (**, p< .01and ***, p< .001). (C): The effects of MSCs on iNOS expression in the lungs of M. abs-R-infected mice. Immunohistochemistry was per-formed as described in “Materials and Methods” (scale bar5 100 lm). (E): C57BL/6J mice were infected with an aerosolized strain ofM. abs-R (1.63 107 CFUs/mouse), and Cox-22/2 and iNOS2/2 MSCs (2.53 105 cells/mouse) were administered by intravenous injectionat 1 day post-infection. After 10 day after infection, bacterial CFUs were measured at lung and spleen as described in “Materials andMethods.” The results are expressed as the mean estimated CFU6 SEM from the lung samples of five to six mice in two independentexperiments. (D): Representative immunohistochemistry staining of iNOS and F4/80 (scale bar5 10 lm). Abbreviations: CFU, colony-forming unit; MSC, mesenchymal stem cell.

Kim, Cha, Kim et al. 11

www.StemCells.com VC AlphaMed Press 2016

increased in macrophages but not in MSCs. In this study, ourdata in hMSCs and THP-1 cells demonstrated that NO levelsare associated in part with host defenses against M. abscessus

in humans because adding NMMA only partially abrogatedthe anti-bacterial abilities of hMSCs. This may be becausehMSCs use indoleamine 2,3-dioxygenase instead of NO toexert their immunomodulatory functions [1, 4]. Therefore,hMSCs might employ another antibacterial mechanism tocombat M. abscessus, and further studies are required to bet-ter define these molecular mechanisms in humans.

According to a previous study, the systemic transplanta-tion of autologous MSCs into multidrug-resistant pulmonarytuberculosis patients induced positive clinical effects, such asstopping bacterial discharge [43]. A clinical trial of an autolo-gous MSC therapy for multidrug-resistant pulmonary tubercu-losis patients has been ongoing since 2014 [44]. In contrast,recent studies have clearly shown that CD2711 MSCs providea long-term protective intracellular niche in the host in whichdormant Mtb can reside, and this provides an antibiotic-protective niche for Mtb [13, 14]. Therefore, endogenouslyadministered MSCs and exogenously administered MSCs maybehave differently in response to Mycobacterium spp.depending on individual immune conditions.

According to recent studies, a strong inflammatoryresponse is important for prompting the immunosuppressivefunctions of MSCs [1, 4]. However, insufficient inflammatorycytokines can stimulate MSCs to produce chemokines andtrophic factors, which enhance immune responses [1]. In addi-tion, IL-10 has been shown to suppress the effects of MSCactivities [1]. It is noteworthy that the secretion of highamounts of IL-10 was detected in M. abscessus-infected mac-rophages that were cultured with or without IFN-c. Thus, theinflammatory status of the host may determine the immuno-modulatory properties of MSCs. In this study, M. abs-R-infected (MOI5 1) macrophages induced low levels of proin-flammatory cytokines. When M. abs-R-infected macrophageswere treated with IFN-c, the production of proinflammatorycytokines was slightly increased. Accumulating evidence hassuggested that a low Th1 cell response, low levels of Th1-associated cytokines such as TNF-a and IFN-c, and a delayedinfiltration of innate immune cells such as monocytes andmacrophages into the lungs could be associated with hostsusceptibility to the development of NTM disease [22, 45]. Inthis study, we found that MSCs lead to the increased recruit-ment of inflammatory cells such as monocytes, CD11bhigh

macrophages, and CD41 and CD81 T cells to the lungs inmice infected with M. abs-R compared with M. abs-R-infectedmice not injected with MSCs. The production of proinflamma-tory cytokines, including IFN-c, IL-6, TNF-a and MCP-1, in thelungs of M. abs-R-infected mice was also increased by MSCinjection. Consequently, our data show that MSCs may per-form opposite functions depending on the inflammatory con-dition, which is in agreement with previous reports [1, 5, 7].

MSCs constitutively produce detectable levels of PGE2.Under inflammatory conditions, PGE2 production is induced,causing a marked increase in the amount of PGE2 secreted byMSCs [40]. Proinflammatory cytokines such as TNF-a, IL-1b,and IFN-c as well as LPS are direct regulators of PGE2 produc-tion by MSCs [46]. PGE2 released by MSCs can reprogrammacrophages to produce additional IL-10, thus shifting thebalance between Th1 and Th2 cells [8, 46]. In this study, we

also showed that MSCs produced higher levels of PGE2 in thepresence of TNF-a, which is upregulated by IFN-c. However,we did not observe an effect on IL-10 secretion in M. abs-R-infected macrophages in the presence or absence of IFN-c byMSCs. These results were partially in agreement with a previ-ous report [8]. Although many studies have reportedenhanced IL-10 production by PGE2, our study suggests thatan inflammatory condition such as infection affects the regu-lation of IL-10 expression by PGE2. However, further studiesare required to better understand the specific nature of therelationship between PGE2 and IL-10 secretion in M. absces-

sus-infected macrophages.Soluble factors produced by MSCs have been suggested to

possess immune-regulatory properties [8]. In this study, wecompared the NO production levels of macrophages in threeconditions (cell-to-cell contact, transwell plates, and CM) totest whether soluble factors produced by MSCs are critical forNO production in macrophages. According to our data, thehighest NO production was found when there was direct con-tact between MSCs and macrophages, indicating that cell-to-cell contact is necessary for the production of NO in macro-phages in this context. The molecules produced by MSCs thatare involved in the direct interaction with macrophages havenot been reported, and the mechanisms that significantlyamplify NO production when macrophages are in direct con-tact with MSCs are also unknown. Thus, the identification ofmolecules involved in the direct interaction between MSCsand macrophages will help to verify the mechanism ofincreased NO production in macrophages and will determinethe applicability of MSCs for therapeutic use in infectiousdiseases.

Recent studies have begun to shed light on the roles ofeicosanoids in mycobacterial infection [47–49]. PGE2, anessential factor in cell membrane protection, is a lipid media-tor derived from arachidonic acid. The infection of macro-phages with virulent H37Rv Mtb induces lipoxin A4production and blocks Cox-2 and PGE2 production, leading tothe necrotic death of Mtb-infected macrophages [47–49].Because these previous studies have focused on the role ofPGE2 in determining the type of cell death that occurs ininfected macrophages, the effector role of PGE2 and the kill-ing mechanisms against mycobacterial spp. remain poorlyunderstood. Therefore, our results provide important evidencethat PGE2 elevates IFN-c-induced NO production, resulting inincreased intracellular pathogen clearance. Further studies areneeded to explore the molecular mechanisms of PGE2-inducedNO production in macrophages.

SUMMARY

In summary, MSCs enhance NO production in M. abs-R-infected macrophages by inducing the secretion of PGE2,which results in the inhibition of M. abs-R growth. TNF-a,which is upregulated by IFN-c in macrophages, may play animportant role the production of PGE2 in MSCs. Therefore,MSCs may facilitate IFN-c-mediated immunity by increasingIFN-c signaling in macrophages. MSCs contribute to the abilityof macrophages to protect against mycobacteria, and theymay therefore represent a novel therapeutic alternative toantibiotics. Because MSCs can be successfully used in human

12 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS

applications, further studies may lead to the development ofnew approaches or interventions that use MSCs to treat pre-viously untreatable mycobacterial infections.

ACKNOWLEDGMENTS

This study was supported by grants from the Basic ScienceResearch Program through the Ministry of Science, ICT,and Future Planning (NRF-2013R1A2A1A01009932, NRF-2014R1A2A1A11053330) and from the Animal and Plant Quar-antine Agency-supported research project (Z-1541780-2012-13-10) and the Korean Health Technology R&D Project, Minis-try for Health & Welfare, Republic of Korea (HI15C2778).

AUTHOR CONTRIBUTIONS

J.S.K., S.H.C., W.S.K., S.J.H., S.B.C.: conceived and designed theexperiments; J.S.K., H.M.K., K.W.K., S.J.K., H.H.C., J.L.: per-formed the experiments; J.S.K., S.N.C., W.J.K.: analyzed thedata; J.S.K., Y.M.P.: contributed reagents/materials/analysistools; J.S.K., Y.M.P., S.J.S.: wrote the paper.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors have declared no competing financial interests.

REFERENCES

1 Ma S, Xie N, Li W et al. Immunobiologyof mesenchymal stem cells. Cell Death Differ2014;21:216–225.

2 Trounson A, Thakar RG, Lomax G et al.Clinical trials for stem cell therapies. BMCMed 2011;9:52.

3 Meier RP, Muller YD, Morel P et al.Transplantation of mesenchymal stem cellsfor the treatment of liver diseases, is thereenough evidence? Stem Cell Res 2013;11:1348–1364.

4 Nauta AJ, Fibbe WE. Immunomodulatoryproperties of mesenchymal stromal cells.Blood 2007;110:3499–3506.

5 Chan JL, Tang KC, Patel AP et al. Anti-gen-presenting property of mesenchymalstem cells occurs during a narrow window atlow levels of interferon-gamma. Blood 2006;107:4817–4824.

6 Francois M, Romieu-Mourez R, Stock-Martineau S et al. Mesenchymal stromal cellscross-present soluble exogenous antigens aspart of their antigen-presenting cell proper-ties. Blood 2009;114:2632–2638.

7 Renner P, Eggenhofer E, Rosenauer Aet al. Mesenchymal stem cells require a suffi-cient, ongoing immune response to exerttheir immunosuppressive function. TransplantProc 2009;41:2607–2611.

8 Nemeth K, Leelahavanichkul A, Yuen PSet al. Bone marrow stromal cells attenuatesepsis via prostaglandin E(2)-dependentreprogramming of host macrophages toincrease their interleukin-10 production. NatMed 2009;15:42–49.

9 Hall SR, Tsoyi K, Ith B et al. Mesenchy-mal stromal cells improve survival duringsepsis in the absence of heme oxygenase-1:The importance of neutrophils. STEM CELLS2013;31:397–407.10 Yagi H, Soto-Gutierrez A, Kitagawa Yet al. Bone marrow mesenchymal stromalcells attenuate organ injury induced by LPSand burn. Cell Transplant 2010;19:823–830.11 Krasnodembskaya A, Song Y, Fang Xet al. Antibacterial effect of human mesen-chymal stem cells is mediated in part fromsecretion of the antimicrobial peptide LL-37.STEM CELLS 2010;28:2229–2238.12 Gupta N, Krasnodembskaya A,Kapetanaki M et al. Mesenchymal stem cellsenhance survival and bacterial clearance inmurine Escherichia coli pneumonia. Thorax2012;67:533–539.

13 Raghuvanshi S, Sharma P, Singh S et al.Mycobacterium tuberculosis. evades hostimmunity by recruiting mesenchymal stemcells. Proc Natl Acad Sci USA 2010;107:21653–21658.14 Das B, Kashino SS, Pulu I et al.CD271(1) bone marrow mesenchymal stemcells may provide a niche for dormant Myco-

bacterium tuberculosis. Sci Transl Med 2013;5:170ra113.15 Masaki T, Qu J, Cholewa-Waclaw J et al.Reprogramming adult Schwann cells to stemcell-like cells by leprosy bacilli promotes dis-semination of infection. Cell 2013;152:51–67.16 Abu-Lubad M, Meyer TF, Al-Zeer MA.Chlamydia trachomatis. inhibits inducible NOsynthase in human mesenchymal stem cellsby stimulating polyamine synthesis.J Immunol 2014;193:2941–2951.17 Nessar R, Cambau E, Reyrat JM et al.Mycobacterium abscessus: A new antibioticnightmare. J Antimicrob Chemother 2012;67:810–818.18 Bryant JM, Grogono DM, Greaves Det al. Whole-genome sequencing to identifytransmission of Mycobacterium abscessus

between patients with cystic fibrosis: A retro-spective cohort study. Lancet 2013;381:1551–1560.19 Koh WJ, Stout JE, Yew WW. Advances inthe management of pulmonary disease dueto Mycobacterium abscessus complex. Int JTuberc Lung Dis 2014;18:1141–1148.20 Benwill JL, Wallace RJ Jr. Mycobacterium

abscessus: Challenges in diagnosis and treat-ment. Curr Opin Infect Dis 2014;27:506–510.21 Ordway D, Henao-Tamayo M, Smith Eet al. Animal model of Mycobacterium

abscessus lung infection. J Leukoc Biol 2008;83:1502–1511.22 Kim JS, Kang MJ, Kim WS et al. Essentialengagement of toll-like receptor 2 in the ini-tiation of early protective Th1 responseagainst rough variants of Mycobacterium

abscessus. Infect Immun 2015;83:1556–1567.23 Shin DM, Yang CS, Yuk JM et al. Myco-bacterium abscessus. activates the macro-phage innate immune response via a physicaland functional interaction between TLR2 anddectin-1. Cell Microbiol 2008;10:1608–1621.24 Choi GE, Chang CL, Whang J et al. Effi-cient differentiation of Mycobacterium

abscessus complex isolates to the specieslevel by a novel PCR-based variable-numbertandem-repeat assay. J Clin Microbiol 2011;49:1107–1109.

25 Kim JS, Kim WS, Lee K et al. Differentialimmune responses to Segniliparus rotundus

and Segniliparus rugosus infection and analy-sis of their comparative virulence profiles.PLoS One 2013;8:e59646.26 Kim JS, Lee YC, Nam HT et al. ApicularenA induces cell death through Fas ligand up-regulation and microtubule disruption bytubulin down-regulation in HM7 humancolon cancer cells. Clin Cancer Res 2007;13:6509–6517.27 Nahrevanian H, Dascombe MJ. Expres-sion of inducible nitric oxide synthase (iNOS)mRNA in target organs of lethal and non-lethal strains of murine malaria. ParasiteImmunol 2002;24:471–478.28 Byrd TF, Lyons CR. Preliminary character-ization of a Mycobacterium abscessus mutantin human and murine models of infection.Infect Immun 1999;67:4700–4707.29 Howard ST, Rhoades E, Recht J et al.Spontaneous reversion of Mycobacterium

abscessus from a smooth to a rough mor-photype is associated with reduced expres-sion of glycopeptidolipid and reacquisition ofan invasive phenotype. Microbiology 2006;152:1581–1590.30 Catherinot E, Clarissou J, Etienne Get al. Hypervirulence of a rough variant ofthe Mycobacterium abscessus type strain.Infect Immun 2007;75:1055–1058.31 De Groote MA, Johnson L, Podell B et al.GM-CSF knockout mice for preclinical testingof agents with antimicrobial activity againstMycobacterium abscessus. J Antimicrob Che-mother 2014;69:1057–1064.32 Choi GE, Shin SJ, Won CJ et al. Macro-lide treatment for Mycobacterium abscessus

and Mycobacterium massiliense infection andinducible resistance. Am J Respir Crit CareMed 2012;186:917–925.33 Davis RL, Sanchez AC, Lindley DJ et al.Effects of mechanistically distinct NF-kappaBinhibitors on glial inducible nitric-oxide syn-thase expression. Nitric Oxide 2005;12:200–209.34 Schroder K, Hertzog PJ, Ravasi T et al.Interferon-gamma: An overview of signals,mechanisms and functions. J Leukoc Biol2004;75:163–189.35 Hoogduijn MJ, Popp F, Verbeek R et al.The immunomodulatory properties of mesen-chymal stem cells and their use for immuno-therapy. Int Immunopharmacol 2010;10:1496–1500.

Kim, Cha, Kim et al. 13

www.StemCells.com VC AlphaMed Press 2016

36 Chan ED, Chan J, Schluger NW. What isthe role of nitric oxide in murine and humanhost defense against tuberculosis. Currentknowledge. Am J Respir Cell Mol Biol 2001;25:606–612.37 Scanga CA, Mohan VP, Yu K et al. Deple-tion of CD4(1) T cells causes reactivation ofmurine persistent tuberculosis despite con-tinued expression of interferon gamma andnitric oxide synthase 2. J Exp Med 2000;192:347–358.38 Lousada S, Florido M, Appelberg R. Viru-lence of Mycobacterium avium in mice doesnot correlate with resistance to nitric oxide.Microb Pathog 2007;43:243–248.39 Fowler CJ, Olivier KN, Leung JM et al.Abnormal nasal nitric oxide production, cili-ary beat frequency, and Toll-like receptorresponse in pulmonary nontuberculousmycobacterial disease epithelium. Am JRespir Crit Care Med 2013;187:1374–1381.40 Ren G, Zhang L, Zhao X et al. Mesenchy-mal stem cell-mediated immunosuppressionoccurs via concerted action of chemokines

and nitric oxide. Cell Stem Cell 2008;2:141–150.41 Sivanathan KN, Gronthos S, Rojas-Canales D et al. Interferon-gamma modifica-tion of mesenchymal stem cells: Implicationsof autologous and allogeneic mesenchymalstem cell therapy in allotransplantation. StemCell Rev 2014;10:351–375.42 Krampera M, Cosmi L, Angeli R et al.Role for interferon-gamma in the immuno-modulatory activity of human bone marrowmesenchymal stem cells. STEM CELLS 2006;24:386–398.43 Erokhin VV, Vasil0eva IA, KonopliannikovAG et al. Systemic transplantation of autolo-gous mesenchymal stem cells of the bonemarrow in the treatment of patients withmultidrug-resistant pulmonary tuberculosis.Problemy tuberkuleza i boleznei legkikh.2008;10:3–6.44 Skrahin A, Ahmed RK, Ferrara G et al.Autologous mesenchymal stromal cell infu-sion as adjunct treatment in patients withmultidrug and extensively drug-resistant

tuberculosis: An open-label phase 1 safetytrial. Lancet Respir Med 2014;2:108–122.45 Kim SY, Koh WJ, Kim YH et al. Impor-tance of reciprocal balance of T cell immunityin Mycobacterium abscessus complex lungdisease. PLoS One 2014;9:e109941.46 Kyurkchiev D, Bochev I, Ivanova-Todorova E et al. Secretion of immunoregula-tory cytokines by mesenchymal stem cells.World J STEM CELLS 2014;6:552–570.47 Mayer-Barber KD, Andrade BB, Oland SDet al. Host-directed therapy of tuberculosisbased on interleukin-1 and type I interferoncrosstalk. Nature 2014;511:99–103.48 Chen M, Divangahi M, Gan H et al. Lipidmediators in innate immunity against tuber-culosis: Opposing roles of PGE2 and LXA4 inthe induction of macrophage death. J ExpMed 2008;205:2791–2801.49 Divangahi M, Chen M, Gan H et al.Mycobacterium tuberculosis evades macro-phage defenses by inhibiting plasmamembrane repair. Nat Immunol 2009;10:899–906.

See www.StemCells.com for supporting information available online.

14 MSCs Enhance the Antimycobacterial Activity

VC AlphaMed Press 2016 STEM CELLS