Cytotherapy, 2015; 0: 1e10
Human mesenchymal stromal cells suppress T-cell proliferation independent of heme oxygenase-1
SEEMA R. PATEL1, IAN B. COPLAND1, MARCO A. GARCIA1, RICHARD METZ2 & JACQUES GALIPEAU1,3 1
Departments of Hematology & Medical Oncology, Emory University Winship Cancer Institute, Atlanta, Georgia, USA, 2NewLink Genetics Inc, Plymouth Meeting, Pennsylvania, USA, and 3Department of Pediatrics, Emory University, Atlanta, Georgia, USA
Abstract Mesenchymal stromal cells deploy immune suppressive properties amenable for use as cell therapy for inflammatory disorders. It is now recognized that mesenchymal stromal cells necessitate priming with an inflammatory milieu, in particular interferon-g, to exert augmented immunosuppressive effects. It has been recently suggested that the heme-catabolizing enzyme heme oxygenase-1 is an essential component of the mesenchymal stromal celledriven immune suppressive response. Because mesenchymal stromal cells upregulate indoleamine 2,3-dioxygenase expression on interferon-g priming and indoleamine 2,3-dioxygenase requires heme as a cofactor for optimal catabolic function, we investigated the potential antagonism of heme oxygenase-1 activity on indoleamine 2, 3-dioxygenase and the impact on mesenchymal stromal cell immune plasticity. We herein sought to evaluate the molecular genetic effect of cytokine priming on human mesenchymal stromal cell heme oxygenase-1 expression and its functional role in differentially primed mesenchymal stromal cells. Contrary to previous reports, messenger RNA and protein analyses demonstrated that mesenchymal stromal cells derived from normal subjects (n ¼ 6) do not express heme oxygenase-1 at steady state or after interferon-g, tumor necrosis factor-a, and/or transforming growth factor-b priming. Pharmacological inhibition of heme oxygenase-1 with the use of tin protoporphyrin did not significantly abrogate the ability of mesenchymal stromal cells to suppress T-cell proliferation in vitro. Overall, these results unequivocally demonstrate that under steady state and after cytokine priming, human mesenchymal stromal cells immunoregulate T-cell proliferation independent of heme oxygenase-1. Key Words: heme oxygenase 1, immune modulation, indoleamine 2,3-dioxygenase, inflammatory cytokine priming, mechanism of T-cell suppression, mesenchymal stromal cells
Introduction Bone marrowederived mesenchymal stromal cells (MSCs) are multipotent progenitors that have an array of immunomodulating properties [reviewed in Le Blanc et al. [1]]. The beneficial effects that MSCs exert on immune cells range from inhibiting proinflammatory polarization [2,3] and effector pathways [4] to enhancing the generation of regulatory cells [5]. Moreover, MSC infusion strongly correlates with prolonged survival of allografts [6] and alleviation of graft-versus-host disease (GvHD) [7,8] and autoimmunity [9,10]. Accordingly, MSCs have emerged as a cellular therapy for immune-mediated ailments and have been the object of numerous early-phase clinical trials that conclusively support the safety of this pharmaceutical strategy [11]. Clinical
studies for treatment of autoimmune and alloimmune ailments are encouraging in their signal for possible efficay, though the frequency of patient responders differs among lead trials [12e15]. The exact mechanism(s) responsible for the immunosuppressive effects of MSCs under pathophysiological conditions are now being rigorously defined, and the mechanistic understanding of immune plasticity function of MSC informs the field on how best to optimally use this cellular therapy. Despite differences in the immunobiology of murine and human MSCs [reviewed in Romieu-Mourez et al. [16]], it is widely accepted that the acquistion of immunoregulatory properties is dependent on MSC priming by an inflammatory milieu, in particular interferon (IFN)-g. Early neutralization of IFNg
Correspondence: Jacques Galipeau, MD, Departments of Hematology & Medical Oncology and Pediatrics, Emory University Winship Cancer Institute, 1365 Clifton Road, Emory Clinic B, Rom B5117, Atlanta, GA 30322, USA. E-mail: [email protected] (Received 10 August 2014; accepted 27 November 2014) http://dx.doi.org/10.1016/j.jcyt.2014.11.010 ISSN 1465-3249 Copyright Ó 2015, International Society for Cellular Therapy. Published by Elsevier Inc. All rights reserved.
2
S. R. Patel et al.
produced by T cells or natural killer cells inhibit the suppressive effects used by MSCs [17], and immunomodulation is augmented by priming MSCs with IFNg before co-culture [5]. Likewise, MSCs primed with IFNg are more potent than are non-treated cells at attenuating graft-versus-host disease (GvHD), and MSCs are ineffective in controlling GvHD if mice are transplanted with T cells defective for IFNg [8]. Exposure to IFNg upregulates or induces de novo expression of multiple, non-redundant pathways involved in immune modulation. Both contactdependent and contact-independent pathways have been attributed to how MSCs communicate with the immune system. Neutralization or small interfering RNA knockdown of PD-L1 (B7-H1; CD274) on MSCs diminishes T-cell suppression. Likewise, impeding the function of numerous soluble factors underscores the intrinsic immunoregulatory characteristics of MSCs. Several of these pivotal immune mediators include cyclooxygenase 2/prostaglandin E2, transforming growth factor-b (TGFb), interleukin-6, hepatocyte growth factor, human leukocyte antigen (HLA)-G5, CC chemokine ligand (CCL)-2, CCL3, CCL12, tumor necrosis factor (TNF), stimulated gene 6 and indoleamine 2, 3-dioxygenase 1 (IDO1) [reviewed in Le Blanc et al. [1], Singer et al. [18] and Uccelli et al. [19]]. It has recently been proposed that heme oxygenase1 (HO-1) can provide an additional pathway by which MSCs modulate immunity [20,21]. HO-1 is an inducible enzyme that confers homeostatic protection against various stress-related conditions through the metabolism of free heme into labile iron, carbon monoxide and biliverdin. Ultraviolet radiation, reactive oxygen species [22] and inflammatory cytokines [23,24] are among the abundant stressors known to elicit HO-1. Polymorphisms in the human HMOX1 gene correlates with increased susceptibility to a proinflammatory state [reviewed in Exner et al. [25] and Gozzelino et al. [26]], and HO-1 deficiency in mice is linked to progressive inflammatory diseases and immune dysregulation [27] [reviewed in Gozzelino et al. [26]]. Pharmacological and genetic induction of HO-1 in animal models of autoimmunity [28], graft rejection or GvHD [29e31] and arteriosclerosis [31] has been shown to associate with amelioration of disease symptoms. Two in vitro studies implicate that human MSCs utilize HO-1 to modulate cellular immunity [20,21] and that IFNg priming enhances HO-1 production. In the present study, we sought to evaluate the effects of cytokine priming on HO-1 expression in MSCs, followed by elucidating the mechanistic necessity of HO-1 in the ability of primed MSCs to modulate immunity. However, contradictory to previous reports, messenger RNA (mRNA) and protein analyses demonstrated that human bone marrowederived MSCs do not express HO-1 at steady state or after
inflammatory cytokine exposure. Moreover, catalytic inhibition of HO-1 did not abrogate the proficiency of MSCs to dampen the proliferative capacity of T cells. Combined, the findings herein oppose previous reports and unequivocally demonstrate that the suppressive effect of human MSCs on proliferation of activated T cells does not involve expression of catalytically active HO-1. Methods Ethics statement The study herein was performed under Emory Institutional Review Board (IRB)-approved protocol IRB00046063. The Emory IRB has reviewed and reapproved the aforementioned protocol. As approved by the aforementioned IRB, a written consent form was obtained from all participants before samples were obtained. Generation of bone marrowederived MSCs Human MSCs were isolated from bone marrow aspirates collected from the iliac crest of healthy consenting adult (18e40 years of age) male and female subjects. Diluted aspirates were layered over a Ficoll density gradient and cultured in MSC media (1 amodified essential medium, 16% platelet lysate [PL] media, 20 mmol/L i-glutamine and 1% penicillin/ streptomycin) at 100,000e400,000 mononuclear cells/cm2. Hematopoietic cells were removed after 72 h to generate a homogenous population that according to the International Society of Cell Therapy guidelines is defined as a CD73þ CD105þ CD90þ CD44þ HLA class I (ABC)þ population lacking hematopoietic markers CD34, CD45, CD11b and CD19. MSCs were passaged weekly at 1000 cells/ cm2 in 10% PL or fetal bovine serum (FBS) MSC media and used between passages 3e8 to eliminate residual hematopoietic cells and avoid senescence. Reagents Flow antibodies were purchased from BD Biosciences (phycoerythrin anti-human CD44, CD73 or CD105; fluorescein isothiocyanate anti-human CD34, CD45 or CD11b; Allophycocyanin anti-human CD19, CD90 or HLA-ABC). Human lung epithelial adenocarcinoma cell line A549 was obtained from American Type Culture Collection. Recombinant human IFNg, CellTrace carboxyfluorescein diacetate succinimidyl ester (CFSE) cell proliferation kit, 7-aminoactinomycin D (7-AAD) and anti-human CD3/CD28 Dynabeads were purchased from Life Technologies. Recombinant human TNFa and TGFb were obtained from R&D Systems.
Role of HO-1 in MSC-mediated T-cell suppression All assays were performed with the use of 10 ng/mL IFNg, 3 ng/mL TNFa and 2 ng/mL TGFb unless indicated otherwise. Tin and zinc protoporphyrin (SnPP and ZnPP) reconstituted in dimethyl sulfoxide (DMSO) was provided by Dr Richard Metz at NewLink Genetics Corporation. Quantitative real-time polymerase chain reaction MSCs were cultured in the presence or absence of IFNg, TNFa and/or TGFb in 10% PL or FBS media. A549 were included as a control for IDO1 and HO-1 expression. Total RNA was extracted with the use of the Qiagen RNeasy Mini Kit (Qiagen) or Zymo Research Quick-RNA Kit. RNA was reversetranscribed with the use of the Qiagen Quantitect reverse transcription kit (Qiagen). SYBR green Mastermix (Quanta Biosciences) and the following human primers (forward, reverse) were used for quantitative real-time polymerase chain reaction (qRT-PCR): IDO1, 50 -GCCCTTCAAGTGTTTCACCAA-30 , 50 CCAGCCAGACAAATATATGCGA-30 ; HO-1, 50 CTTCTTCACCTTCCCCAACA-30 , 50 -AGCTCC TGCAACTCCTCAAA-30 ; and b-actin, 50 -GGG AAATCGTGCGTGACAT-30 , 50 -CAGGAGGAGC AATGATCTT-30 . Samples were performed as duplicates and read with the use of an Applied Biosystems 7500 Fast Real-Time PCR system thermal cycler. To discriminate primer dimerization and target gene amplification, cycle threshold values (CT) were compared with corresponding melting curves (Supplementary Table S1 and Supplementary Table S2). Relative expression was calculated as the CT normalized to the CT of the reference mRNA, bactin. Fold changes were quantified as sample relative expression normalized to relative expression of A549 (Figure 1) or resting MSCs (Supplementary Figure S3). Immunoblotting Whole-cell lysates were generated from resting or IFNg-, TNFa- and/or TGFb-primed MSCs in 10% PL media. A549 lysate was included as a control for IDO1 and HO-1 expression; 10 mg total extract was run on a 4e20% Tris-glycine gel (Thermo Scientific) and transferred onto a nitrocellulose membrane (BioRad, Hercules, CA, USA). Proteins were detected with the use of primary rabbit anti-human IDO1 (1:1,000; EMD Millipore Corporation), anti-human HO-1 (1:2,000; Abcam) or anti-human b-actin (1:1,000; Cell Signaling Technology, Inc) and corresponding secondary horseradish peroxideecoupled goat anti-rabbit immunoglobulin G h þ l (1:10,000e20,000 Bethyl Laboratories, Inc). Protein bands were revealed with the use of an echochemiluminescent system (Amersham Pharmacia Biotech).
3
Protoporphyrin inhibition of IDO1 Inhibition by SnPP or ZnPP was assessed against stably and doxycycline-inducible human IDO1 enzyme expressed in TRex cells as described in [32]. Cells were grown at 2.5 104 cells/well. ZnPP and SnPP solubilized in DMSO at concentrations indicated were serially diluted and incubated for 16 h at 37 C. Reactions were terminated with 3% trichloroacetic acid and incubated at 4 C for 1 h. To measure kynurenine, the media was heated to 65 C for 15 min and clarified by use of 4000 rpm centrifugation for 10 min. Supernatants were mixed with Ehrlich’s reagent (2% p-dimethylaminobenzaldehyde wt/vol in acetic acid) and quantified at 490 nm with the use of a Synergy HT microtiter plate reader (Bio-Tek). Samples were analyzed in triplicate with control values averaged and subtracted from experimental samples (Supplementary Table S3). In vitro cytotoxicity assay Compound cytotoxicity was assessed with the use of the sulforhodamine B (SRB) viable cell assay [33]. Fixed cells from the protoporphyrin inhibition assay (above) were stained with 0.4% (wt/vol) SRB (Sigma-Aldrich) dissolved in 1% acetic acid. Unbound dye was removed with 1% acetic acid, and protein-bound dye was extracted with the use of 10 mmol/L unbuffered Tris base (pH 10.5) for 5 min. Optical density was read at 570 nm with the use of a Synergy HT microtiter plate reader. SnPP or ZnPP toxicity was measured as a percentage of reduced protein content compared with controls without the inhibitor (data not shown). In vitro proliferation assay MSCs expanded in 10% PL media were seeded at 100,000 cells/well in R10 (10% FBS, 1 Roswell Park Memorial Institute medium, 1% penicillin/streptomycin, 20 mmol/L i-glutamine and 5 mL 4(2hydroxyethyl)-1-piperazine-ethane sulfonic acid) for 2 h. Human peripheral blood mononuclear cells (PBMCs) were isolated from whole blood of healthy consenting volunteers through the use of Ficoll density gradient (GE Healthcare). PBMCs were labeled with 5 mmol/L CFSE. CFSE-labeled PBMCs (n ¼ 500,000) were co-cultured with 100,000 MSCs/well in R10. T-cell stimulation was induced with antihuman CD3/CD28 Dynabeads (1 bead:100,000 PBMCs). DMSO-dissolved SnPP was added at 10 mmol/L. After 4 days, T-cell proliferation was assessed by flow cytometric analysis of CFSE dilution. Percentage of inhibition was calculated with percentage of proliferation measured by use of FlowJo 9.6.
4
S. R. Patel et al.
Figure 1. Quantitative HO-1 mRNA levels in resting and cytokine-treated MSC. Expression levels of HO-1eencoding (A) and IDO1encoding (B) genes were measured by means of qRT-PCR in donor BMH01, BMH04 and BMH06 resting MSCs and MSCs stimulated for 24, 48 or 72 h with 10 ng/mL IFNg, 3 ng/mL TNFa, 2 ng/mL TGFb, 10 ng/mL IFNg þ 3 ng/mL TNFa or 10 ng/mL IFNg þ 2 ng/mL TGFb. (C) HO-1 mRNA transcript levels were measured at steady state or after 24-h stimulation with 10 ng/mL IFNg in donor BMH01, BMH04, BMH06, BMH12, BMH13 and BMH14 MSCs cultured in 10% PL- or FBS-containing media. Untreated A549 were included as a negative and positive control for IDO1 and constitutive HO-1 expression, respectively. When evaluating fold changes, A549 were used as the internal control and were attributed a value of 1. Fold changes were computed by normalizing the relative expression of MSCs to the relative expression of A549. Error bars represent mean standard error of the mean. Statistics were generated with the use of two-way ANOVA with Dunnett’s test (A, B) or one-way ANOVA with Tukey’s post-test (C). For (A) and (B), data shown are the combined results from two (BMH06) or three (BMH01 and BMH04) independent experiments. Data illustrated in (C) are the combined results of six donors (BMH01, BMH04, BMH06, BMH12, BMH13 and BMH14). ****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05.
Viability assay MSCs (n ¼ 100,000) grown in 10% PL media were seeded in R10 and treated with indicated concentrations of DMSO-dissolved SnPP. After 4 days, viability of MSCs was assessed by use of the uptake of nucleic acid marker 7-AAD, by means of flow cytometry. Percentage of 7-AADþ MSCs was computed with the use of FlowJo 9.6. Statistical analysis Statistical analyses were performed by means of oneor two-way analysis of variance (ANOVA) with the use of a post-Dunnett’s, Tukey’s or Sidak’s multiple comparison tests. In all statistical analyses,
significance was defined as P < 0.0001, P < 0.001, P < 0.01 and P < 0.05. Results Characterization of human bone marrowederived MSCs The MSCs used in the current study were isolated by means of plastic adherence from bone marrow aspirates collected from healthy consenting donors. Homogeneity of adhered cells was confirmed by use of phenotypic analysis of cell surface markers CD34, CD45, CD11b, CD19, CD90, CD44, CD73, CD105 and HLA class I (ABC). All donor MSCs used in the investigations herein were studied no later than passage 8 and expressed equivalent and
Role of HO-1 in MSC-mediated T-cell suppression significantly detectable levels of CD73, CD105, CD90, CD44 and HLA class I (ABC) (Supplementary Figure S1). The absence of hematopoietic cells in each donor homogenous culture was confirmed by the absence of CD34, CD45, CD11b and CD19 (Supplementary Figure S1). Human bone marrowederived MSC mRNA expression of HO-1 Several inflammatory effectors are known to influence HO-1 expression [23,24,34e36]. Anti-inflammatory TGFb [24] and pro-inflammatory TNFa or interleukin-1b [23] are a few reported to induce HO-1 in epithelial and endothelial cells, respectively. Therefore, to elucidate the expression of HO-1 in MSCs under various inflammatory states, HO-1 mRNA levels at steady state and after exposure to IFNg, TNFa and/or TGFb for 24, 48 or 72 h were measured in three MSC donors (BMH01, BMH04 and BMH06) cultured in 10% PL media with the use of qRT-PCR. A549 were included as a positive control for constitutive HO-1 expression [24,37]. b-Actin gene levels were used as a reference control for computing relative expression levels of HO-1 (Supplementary Figure S2). Fold change was calculated by normalizing the relative expression of samples to the relative expression of A549. The data from three independent experiments demonstrated that comparable to HO-1 transcript levels in A549 resting MSCs correlated with a statistically lower quantity of HO-1 (P < 0.001, P < 0.01 and P < 0.05; Figure 1A). Likewise, MSCs primed with IFNg, TNFa and/or TGFb for 24, 48 and 72 h did not elicit considerable amounts of HO-1 above that expressed at steady state (P > 0.05; Figure 1A) or A549 (P < 0.0001, P < 0.001, P < 0.01 and P < 0.05; Figure 1A). The failure to identify HO-1 upregulation was not due to insufficient priming or use of metabolically inactive MSCs because statistically significant levels of IDO1, an IFNg-inducible enzyme, were expressed in MSCs after 24-, 48- and 72-h exposure to IFNg, albeit with or without TNFa and TGFb (P < 0.0001 and P < 0.01; Figure 1B). Indeed, these findings dispute prior reports that HO-1 is constitutive in MSCs and augmented in response to a pro-inflammatory stimulus, IFNg [20,21]. However, when comparing the current findings with previous reports, it is conceivable that inconsistency in MSC media is a source of these disparate results, with HO-1 expression characterized herein through the use of MSCs expanded in PL versus FBS media. Comparative analyses of different types of media used to culture MSCs have demonstrated the effects that media have on the immune profile of MSCs primed with IFNg þ TNFa; de novo
5
IDO1 and inducible nitric oxide synthase expression varies in PL-expanded and fetal calf serum (FCS)expanded MSCs [38]. Thus, it is legitimate to investigate HO-1 expression in MSCs expanded under differing culture medium conditions. Because earlier reports observed an upregulation of constitutive HO-1 in FBS-expanded MSCs after 24-h exposure to IFNg, HO-1 transcripts were quantified at steady state or succeeding 24-h IFNg stimulation in six MSC donors cultured in 10% PL or FBS media with the use of qRTPCR. A549 were included as a positive control for constitutive HO-1 expression [24,37]. b-Actin gene levels were used as a reference control for computing relative expression levels of HO-1 (Supplementary Figure S2). Fold change was calculated by normalizing the relative expression of samples to the relative expression of A549. The combined data from six donors demonstrated that measurable levels of HO-1 did not statistically differ between MSCs cultured in PL or FBS media (P > 0.05; Figure 1C). Likewise, IFNg priming did not correlate with a statistically significant variance in HO-1 mRNA levels between differentially cultured MSCs (P > 0.05; Figure 1C). Moreover, compared with resting MSCs, IFNg exposure did not associate with a statistical upregulation of HO-1 albeit in PL- or FBS-expanded MSCs (P > 0.05; Figure 1C). The failure to detect HO-1 transcripts was not due to the inability of the primers to recognize the HO-1 gene because statistically significant levels of HO-1 were observed in A549 (P < 0.0001, P < 0.001, P < 0.01; Figure 1C). Additionally, the lack of detectable HO-1 transcripts succeeding IFNg priming is not due to the use of an insufficient concentration of IFNg, because data from five combined MSC donors grown in PL or FBS media demonstrated that detectable levels of HO-1 did not statistically diverge between resting MSCs and MSCs exposed to different concentrations of IFNg (P < 0.0001; Supplementary Figure S4A). The failure to identify HO-1 upregulation was not due to the inability of IFNg to prime MSCs or the use of metabolically inactive MSCs, because all indicated concentrations of IFNg correlated with a substantial increase in de novo IDO1 (P < 0.0001; Supplementary Figure S4B). HO-1 protein expression in resting and cytokine-treated MSCs Expression of HO-1 is regulated at the transcriptional level, with each stimuli differentially affecting mRNA stability [39,40]. Accordingly, it is equally consistent that the lack of detectable HO-1 transcripts was a direct result of mRNA instability. Thus, the protein levels of HO-1 at steady state and after 24, 48 or 72 h of IFNg, TNFa and/or TGFb stimulation were assessed by means of immunoblotting.
6
S. R. Patel et al.
Because the level of HO-1, mRNA did not differ among MSC donors or differential culture media, protein expression and subsequent studies were performed with the use of a PL-cultured MSC donor, BMH04. A549 were included as a positive control for constitutive HO-1 expression. b-Actin protein detection was included as a loading control. The findings from two independent experiments indicated that HO-1 is undetectable at steady state and after exposure to inflammatory conditions in MSCs by means of immunoblotting (Figure 2A). The failure to observe HO-1 was not due to a deficiency in the sensitivity of the assay to detect HO-1 because constitutive HO-1 was observed in A549 (Figure 2A). Moreover, the absence of HO-1 in MSCs was not a result of metabolic or priming insufficiencies, because IFNg-primed MSCs demonstrated substantial de novo expression of IDO1 protein (Figure 2B). MSCs suppress T-cell proliferation independently of HO-1 Although mRNA and protein analysis indicate that MSCs do not express HO-1, these findings do not rule out that MSCs express HO-1 at levels below the sensitivity threshold of qRT-PCR and immunoblotting performed herein. Thus, the functional requirement of HO-1 in MSC-mediated suppression of T-cell proliferation was tested by blocking the catabolic activity of HO-1. Two pharmacological inhibitors used to inhibit HO-1 catalytic function are
SnPP and ZnPP. Protoporphyrin complexes are heme precursor compounds that regulate heme oxygenase enzymes by competing with the substrate heme and are recognized as heme oxygenase enzymeespecific inhibitors. However, ZnPP has also been shown to irreversibly block the catalytic activity of IDO1 [41], which utilizes heme as a co-factor. Because ZnPP has the capacity to block HO-1 and IDO1 and IFNg-primed MSCs upregulate IDO1, ZnPP is not an ideal utility to determine the specific requirement of HO-1 in MSC-mediated T-cell suppression. Conversely, the specificity of SnPP remains unclear. Therefore, to assess the specificity of SnPP for HO-1, the relative concentration of kynurenine was quantified in the supernatant of SnPP- or ZnPP-treated 293-T-REx cells expressing IDO1 under doxycycline regulation. The findings presented herein demonstrated that at various concentrations, SnPP does not inhibit IDO1 catalytic function (Figure 3A). The failure to regulate IDO1 was not due to the inability of the assay system to detect IDO1 inhibition, because ZnPP potently blocked degradation of i-tryptophan by IDO1 (Figure 3A). Overall, these findings indicate that SnPP does not inhibit IDO1 and thus can be used to assess the necessity of MSC-derived HO-1 in T-cell suppression. To determine the appropriate concentration of SnPP, resting and IFNg-primed MSCs were treated with 1 mmol/L, 10 mmol/L or 50 mmol/L SnPP. MSC viability was assessed 4 days later by the uptake of a nucleic acid marker, 7-AAD, by means of
Figure 2. HO-1 protein expression is undetectable in resting and cytokine-treated MSCs. HO-1 (A) and IDO1 (B) protein expression was analyzed in donor BMH04 resting MSCs and MSCs treated with 10 ng/mL IFNg, 3 ng/mL TNFa, 2 ng/mL TGFb, 10 ng/mL IFNg þ 3 ng/ mL TNFa or 10 ng/mL IFNg þ 2 ng/mL TGFb after 24, 48 or 72 h by means of immunoblot. Untreated A549 was included as a negative and positive control for IDO1 and constitutive HO-1 protein expression, respectively. b-Actin protein expression was used as a loading control.
Role of HO-1 in MSC-mediated T-cell suppression
7
Figure 3. MSC-mediated suppression of T-cell proliferation is mechanistically independent of HO-1. (A) Percentage of IDO1 inhibition was measured in 293-T-REx cells treated with various concentrations of SnPP (circle) or ZnPP (square) for 16 h at 37 C. Percentage of inhibition was calculated as 100% inhibition e [(normalized kynurenine produced with inhibitor)/(kynurenine without inhibitor) 100]. Kynurenine levels were normalized to the background absorbance at 490 nm with the use of media from cells not expressing IDO1. The value for each sample was adjusted to take into consideration changes in protein levels resulting from toxicity that was assessed by the SRB cell toxicity assay; very little toxicity was observed at the concentrations of protoporphyrin tested (data not shown). (B) Viability of resting (white bar) and IFNg-primed (black bar) BMH04 MSC treated with 1 mmol/L, 10 mmol/L or 50 mmol/L SnPP or a volume equivalent (VE) of DMSO for 4 days was assessed by use of the uptake of a nucleic acid marker, 7-AAD, by means of flow cytometry. Percentage of 7-AADþ was defined as the percentage of MSCs positive for 7-AAD. (C) Proliferation assays were performed with the use of anti-CD3/CD28 Dynabead-activated, CFSE-labeled PBMCs from healthy donors. Bead-activated, CFSE-labeled PBMCs were cultured in the presence or absence of donor BMH04 resting MSCs for 4 days at an MSC:PBMC ratio of 1:5. Where indicated, the catalytic inhibitor of HO-1, SnPP, was added to PBMC alone or PBMC:MSC co-cultures at 10 mmol/L. PBMC proliferation was assessed by means of flow cytometry and defined as CFSE dilution. Resting PBMCs were included as a negative gating control (CFSEhi) for quantifying percent proliferation, which was measured as the percentage of CFSElow leukocytes. Percent proliferation was subsequently used to calculate percent inhibition. Error bars represent mean standard error of the mean. Statistical analysis was performed with the use of one-way ANOVA with Dunnett’s post-test (B) and two-way ANOVA with Sidak’s post-test (C). Illustrated data are combined findings from three distinct experiments that used PBMCs from differing donors. ****P < 0.0001, ***P < 0.001, **P < 0.01.
flow cytometry. The percentage of 7-AADþ was defined as the percentage of MSCs positive for 7-AAD; live cells do not take up 7-AAD. The data from three independent studies indicated that 50 mmol/L SnPP induced death in roughly 20e40% of resting and IFNg-primed MSCs (P < 0.0001, P < 0.001; Figure 3B). The observed cellular death was not caused by toxic effects of DMSO used to dissolve the SnPP, because volume-equivalent DMSOtreated MSCs were
Human mesenchymal stromal cells suppress T-cell proliferation independent of heme oxygenase-1
SEEMA R. PATEL1, IAN B. COPLAND1, MARCO A. GARCIA1, RICHARD METZ2 & JACQUES GALIPEAU1,3 1
Departments of Hematology & Medical Oncology, Emory University Winship Cancer Institute, Atlanta, Georgia, USA, 2NewLink Genetics Inc, Plymouth Meeting, Pennsylvania, USA, and 3Department of Pediatrics, Emory University, Atlanta, Georgia, USA
Abstract Mesenchymal stromal cells deploy immune suppressive properties amenable for use as cell therapy for inflammatory disorders. It is now recognized that mesenchymal stromal cells necessitate priming with an inflammatory milieu, in particular interferon-g, to exert augmented immunosuppressive effects. It has been recently suggested that the heme-catabolizing enzyme heme oxygenase-1 is an essential component of the mesenchymal stromal celledriven immune suppressive response. Because mesenchymal stromal cells upregulate indoleamine 2,3-dioxygenase expression on interferon-g priming and indoleamine 2,3-dioxygenase requires heme as a cofactor for optimal catabolic function, we investigated the potential antagonism of heme oxygenase-1 activity on indoleamine 2, 3-dioxygenase and the impact on mesenchymal stromal cell immune plasticity. We herein sought to evaluate the molecular genetic effect of cytokine priming on human mesenchymal stromal cell heme oxygenase-1 expression and its functional role in differentially primed mesenchymal stromal cells. Contrary to previous reports, messenger RNA and protein analyses demonstrated that mesenchymal stromal cells derived from normal subjects (n ¼ 6) do not express heme oxygenase-1 at steady state or after interferon-g, tumor necrosis factor-a, and/or transforming growth factor-b priming. Pharmacological inhibition of heme oxygenase-1 with the use of tin protoporphyrin did not significantly abrogate the ability of mesenchymal stromal cells to suppress T-cell proliferation in vitro. Overall, these results unequivocally demonstrate that under steady state and after cytokine priming, human mesenchymal stromal cells immunoregulate T-cell proliferation independent of heme oxygenase-1. Key Words: heme oxygenase 1, immune modulation, indoleamine 2,3-dioxygenase, inflammatory cytokine priming, mechanism of T-cell suppression, mesenchymal stromal cells
Introduction Bone marrowederived mesenchymal stromal cells (MSCs) are multipotent progenitors that have an array of immunomodulating properties [reviewed in Le Blanc et al. [1]]. The beneficial effects that MSCs exert on immune cells range from inhibiting proinflammatory polarization [2,3] and effector pathways [4] to enhancing the generation of regulatory cells [5]. Moreover, MSC infusion strongly correlates with prolonged survival of allografts [6] and alleviation of graft-versus-host disease (GvHD) [7,8] and autoimmunity [9,10]. Accordingly, MSCs have emerged as a cellular therapy for immune-mediated ailments and have been the object of numerous early-phase clinical trials that conclusively support the safety of this pharmaceutical strategy [11]. Clinical
studies for treatment of autoimmune and alloimmune ailments are encouraging in their signal for possible efficay, though the frequency of patient responders differs among lead trials [12e15]. The exact mechanism(s) responsible for the immunosuppressive effects of MSCs under pathophysiological conditions are now being rigorously defined, and the mechanistic understanding of immune plasticity function of MSC informs the field on how best to optimally use this cellular therapy. Despite differences in the immunobiology of murine and human MSCs [reviewed in Romieu-Mourez et al. [16]], it is widely accepted that the acquistion of immunoregulatory properties is dependent on MSC priming by an inflammatory milieu, in particular interferon (IFN)-g. Early neutralization of IFNg
Correspondence: Jacques Galipeau, MD, Departments of Hematology & Medical Oncology and Pediatrics, Emory University Winship Cancer Institute, 1365 Clifton Road, Emory Clinic B, Rom B5117, Atlanta, GA 30322, USA. E-mail: [email protected] (Received 10 August 2014; accepted 27 November 2014) http://dx.doi.org/10.1016/j.jcyt.2014.11.010 ISSN 1465-3249 Copyright Ó 2015, International Society for Cellular Therapy. Published by Elsevier Inc. All rights reserved.
2
S. R. Patel et al.
produced by T cells or natural killer cells inhibit the suppressive effects used by MSCs [17], and immunomodulation is augmented by priming MSCs with IFNg before co-culture [5]. Likewise, MSCs primed with IFNg are more potent than are non-treated cells at attenuating graft-versus-host disease (GvHD), and MSCs are ineffective in controlling GvHD if mice are transplanted with T cells defective for IFNg [8]. Exposure to IFNg upregulates or induces de novo expression of multiple, non-redundant pathways involved in immune modulation. Both contactdependent and contact-independent pathways have been attributed to how MSCs communicate with the immune system. Neutralization or small interfering RNA knockdown of PD-L1 (B7-H1; CD274) on MSCs diminishes T-cell suppression. Likewise, impeding the function of numerous soluble factors underscores the intrinsic immunoregulatory characteristics of MSCs. Several of these pivotal immune mediators include cyclooxygenase 2/prostaglandin E2, transforming growth factor-b (TGFb), interleukin-6, hepatocyte growth factor, human leukocyte antigen (HLA)-G5, CC chemokine ligand (CCL)-2, CCL3, CCL12, tumor necrosis factor (TNF), stimulated gene 6 and indoleamine 2, 3-dioxygenase 1 (IDO1) [reviewed in Le Blanc et al. [1], Singer et al. [18] and Uccelli et al. [19]]. It has recently been proposed that heme oxygenase1 (HO-1) can provide an additional pathway by which MSCs modulate immunity [20,21]. HO-1 is an inducible enzyme that confers homeostatic protection against various stress-related conditions through the metabolism of free heme into labile iron, carbon monoxide and biliverdin. Ultraviolet radiation, reactive oxygen species [22] and inflammatory cytokines [23,24] are among the abundant stressors known to elicit HO-1. Polymorphisms in the human HMOX1 gene correlates with increased susceptibility to a proinflammatory state [reviewed in Exner et al. [25] and Gozzelino et al. [26]], and HO-1 deficiency in mice is linked to progressive inflammatory diseases and immune dysregulation [27] [reviewed in Gozzelino et al. [26]]. Pharmacological and genetic induction of HO-1 in animal models of autoimmunity [28], graft rejection or GvHD [29e31] and arteriosclerosis [31] has been shown to associate with amelioration of disease symptoms. Two in vitro studies implicate that human MSCs utilize HO-1 to modulate cellular immunity [20,21] and that IFNg priming enhances HO-1 production. In the present study, we sought to evaluate the effects of cytokine priming on HO-1 expression in MSCs, followed by elucidating the mechanistic necessity of HO-1 in the ability of primed MSCs to modulate immunity. However, contradictory to previous reports, messenger RNA (mRNA) and protein analyses demonstrated that human bone marrowederived MSCs do not express HO-1 at steady state or after
inflammatory cytokine exposure. Moreover, catalytic inhibition of HO-1 did not abrogate the proficiency of MSCs to dampen the proliferative capacity of T cells. Combined, the findings herein oppose previous reports and unequivocally demonstrate that the suppressive effect of human MSCs on proliferation of activated T cells does not involve expression of catalytically active HO-1. Methods Ethics statement The study herein was performed under Emory Institutional Review Board (IRB)-approved protocol IRB00046063. The Emory IRB has reviewed and reapproved the aforementioned protocol. As approved by the aforementioned IRB, a written consent form was obtained from all participants before samples were obtained. Generation of bone marrowederived MSCs Human MSCs were isolated from bone marrow aspirates collected from the iliac crest of healthy consenting adult (18e40 years of age) male and female subjects. Diluted aspirates were layered over a Ficoll density gradient and cultured in MSC media (1 amodified essential medium, 16% platelet lysate [PL] media, 20 mmol/L i-glutamine and 1% penicillin/ streptomycin) at 100,000e400,000 mononuclear cells/cm2. Hematopoietic cells were removed after 72 h to generate a homogenous population that according to the International Society of Cell Therapy guidelines is defined as a CD73þ CD105þ CD90þ CD44þ HLA class I (ABC)þ population lacking hematopoietic markers CD34, CD45, CD11b and CD19. MSCs were passaged weekly at 1000 cells/ cm2 in 10% PL or fetal bovine serum (FBS) MSC media and used between passages 3e8 to eliminate residual hematopoietic cells and avoid senescence. Reagents Flow antibodies were purchased from BD Biosciences (phycoerythrin anti-human CD44, CD73 or CD105; fluorescein isothiocyanate anti-human CD34, CD45 or CD11b; Allophycocyanin anti-human CD19, CD90 or HLA-ABC). Human lung epithelial adenocarcinoma cell line A549 was obtained from American Type Culture Collection. Recombinant human IFNg, CellTrace carboxyfluorescein diacetate succinimidyl ester (CFSE) cell proliferation kit, 7-aminoactinomycin D (7-AAD) and anti-human CD3/CD28 Dynabeads were purchased from Life Technologies. Recombinant human TNFa and TGFb were obtained from R&D Systems.
Role of HO-1 in MSC-mediated T-cell suppression All assays were performed with the use of 10 ng/mL IFNg, 3 ng/mL TNFa and 2 ng/mL TGFb unless indicated otherwise. Tin and zinc protoporphyrin (SnPP and ZnPP) reconstituted in dimethyl sulfoxide (DMSO) was provided by Dr Richard Metz at NewLink Genetics Corporation. Quantitative real-time polymerase chain reaction MSCs were cultured in the presence or absence of IFNg, TNFa and/or TGFb in 10% PL or FBS media. A549 were included as a control for IDO1 and HO-1 expression. Total RNA was extracted with the use of the Qiagen RNeasy Mini Kit (Qiagen) or Zymo Research Quick-RNA Kit. RNA was reversetranscribed with the use of the Qiagen Quantitect reverse transcription kit (Qiagen). SYBR green Mastermix (Quanta Biosciences) and the following human primers (forward, reverse) were used for quantitative real-time polymerase chain reaction (qRT-PCR): IDO1, 50 -GCCCTTCAAGTGTTTCACCAA-30 , 50 CCAGCCAGACAAATATATGCGA-30 ; HO-1, 50 CTTCTTCACCTTCCCCAACA-30 , 50 -AGCTCC TGCAACTCCTCAAA-30 ; and b-actin, 50 -GGG AAATCGTGCGTGACAT-30 , 50 -CAGGAGGAGC AATGATCTT-30 . Samples were performed as duplicates and read with the use of an Applied Biosystems 7500 Fast Real-Time PCR system thermal cycler. To discriminate primer dimerization and target gene amplification, cycle threshold values (CT) were compared with corresponding melting curves (Supplementary Table S1 and Supplementary Table S2). Relative expression was calculated as the CT normalized to the CT of the reference mRNA, bactin. Fold changes were quantified as sample relative expression normalized to relative expression of A549 (Figure 1) or resting MSCs (Supplementary Figure S3). Immunoblotting Whole-cell lysates were generated from resting or IFNg-, TNFa- and/or TGFb-primed MSCs in 10% PL media. A549 lysate was included as a control for IDO1 and HO-1 expression; 10 mg total extract was run on a 4e20% Tris-glycine gel (Thermo Scientific) and transferred onto a nitrocellulose membrane (BioRad, Hercules, CA, USA). Proteins were detected with the use of primary rabbit anti-human IDO1 (1:1,000; EMD Millipore Corporation), anti-human HO-1 (1:2,000; Abcam) or anti-human b-actin (1:1,000; Cell Signaling Technology, Inc) and corresponding secondary horseradish peroxideecoupled goat anti-rabbit immunoglobulin G h þ l (1:10,000e20,000 Bethyl Laboratories, Inc). Protein bands were revealed with the use of an echochemiluminescent system (Amersham Pharmacia Biotech).
3
Protoporphyrin inhibition of IDO1 Inhibition by SnPP or ZnPP was assessed against stably and doxycycline-inducible human IDO1 enzyme expressed in TRex cells as described in [32]. Cells were grown at 2.5 104 cells/well. ZnPP and SnPP solubilized in DMSO at concentrations indicated were serially diluted and incubated for 16 h at 37 C. Reactions were terminated with 3% trichloroacetic acid and incubated at 4 C for 1 h. To measure kynurenine, the media was heated to 65 C for 15 min and clarified by use of 4000 rpm centrifugation for 10 min. Supernatants were mixed with Ehrlich’s reagent (2% p-dimethylaminobenzaldehyde wt/vol in acetic acid) and quantified at 490 nm with the use of a Synergy HT microtiter plate reader (Bio-Tek). Samples were analyzed in triplicate with control values averaged and subtracted from experimental samples (Supplementary Table S3). In vitro cytotoxicity assay Compound cytotoxicity was assessed with the use of the sulforhodamine B (SRB) viable cell assay [33]. Fixed cells from the protoporphyrin inhibition assay (above) were stained with 0.4% (wt/vol) SRB (Sigma-Aldrich) dissolved in 1% acetic acid. Unbound dye was removed with 1% acetic acid, and protein-bound dye was extracted with the use of 10 mmol/L unbuffered Tris base (pH 10.5) for 5 min. Optical density was read at 570 nm with the use of a Synergy HT microtiter plate reader. SnPP or ZnPP toxicity was measured as a percentage of reduced protein content compared with controls without the inhibitor (data not shown). In vitro proliferation assay MSCs expanded in 10% PL media were seeded at 100,000 cells/well in R10 (10% FBS, 1 Roswell Park Memorial Institute medium, 1% penicillin/streptomycin, 20 mmol/L i-glutamine and 5 mL 4(2hydroxyethyl)-1-piperazine-ethane sulfonic acid) for 2 h. Human peripheral blood mononuclear cells (PBMCs) were isolated from whole blood of healthy consenting volunteers through the use of Ficoll density gradient (GE Healthcare). PBMCs were labeled with 5 mmol/L CFSE. CFSE-labeled PBMCs (n ¼ 500,000) were co-cultured with 100,000 MSCs/well in R10. T-cell stimulation was induced with antihuman CD3/CD28 Dynabeads (1 bead:100,000 PBMCs). DMSO-dissolved SnPP was added at 10 mmol/L. After 4 days, T-cell proliferation was assessed by flow cytometric analysis of CFSE dilution. Percentage of inhibition was calculated with percentage of proliferation measured by use of FlowJo 9.6.
4
S. R. Patel et al.
Figure 1. Quantitative HO-1 mRNA levels in resting and cytokine-treated MSC. Expression levels of HO-1eencoding (A) and IDO1encoding (B) genes were measured by means of qRT-PCR in donor BMH01, BMH04 and BMH06 resting MSCs and MSCs stimulated for 24, 48 or 72 h with 10 ng/mL IFNg, 3 ng/mL TNFa, 2 ng/mL TGFb, 10 ng/mL IFNg þ 3 ng/mL TNFa or 10 ng/mL IFNg þ 2 ng/mL TGFb. (C) HO-1 mRNA transcript levels were measured at steady state or after 24-h stimulation with 10 ng/mL IFNg in donor BMH01, BMH04, BMH06, BMH12, BMH13 and BMH14 MSCs cultured in 10% PL- or FBS-containing media. Untreated A549 were included as a negative and positive control for IDO1 and constitutive HO-1 expression, respectively. When evaluating fold changes, A549 were used as the internal control and were attributed a value of 1. Fold changes were computed by normalizing the relative expression of MSCs to the relative expression of A549. Error bars represent mean standard error of the mean. Statistics were generated with the use of two-way ANOVA with Dunnett’s test (A, B) or one-way ANOVA with Tukey’s post-test (C). For (A) and (B), data shown are the combined results from two (BMH06) or three (BMH01 and BMH04) independent experiments. Data illustrated in (C) are the combined results of six donors (BMH01, BMH04, BMH06, BMH12, BMH13 and BMH14). ****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05.
Viability assay MSCs (n ¼ 100,000) grown in 10% PL media were seeded in R10 and treated with indicated concentrations of DMSO-dissolved SnPP. After 4 days, viability of MSCs was assessed by use of the uptake of nucleic acid marker 7-AAD, by means of flow cytometry. Percentage of 7-AADþ MSCs was computed with the use of FlowJo 9.6. Statistical analysis Statistical analyses were performed by means of oneor two-way analysis of variance (ANOVA) with the use of a post-Dunnett’s, Tukey’s or Sidak’s multiple comparison tests. In all statistical analyses,
significance was defined as P < 0.0001, P < 0.001, P < 0.01 and P < 0.05. Results Characterization of human bone marrowederived MSCs The MSCs used in the current study were isolated by means of plastic adherence from bone marrow aspirates collected from healthy consenting donors. Homogeneity of adhered cells was confirmed by use of phenotypic analysis of cell surface markers CD34, CD45, CD11b, CD19, CD90, CD44, CD73, CD105 and HLA class I (ABC). All donor MSCs used in the investigations herein were studied no later than passage 8 and expressed equivalent and
Role of HO-1 in MSC-mediated T-cell suppression significantly detectable levels of CD73, CD105, CD90, CD44 and HLA class I (ABC) (Supplementary Figure S1). The absence of hematopoietic cells in each donor homogenous culture was confirmed by the absence of CD34, CD45, CD11b and CD19 (Supplementary Figure S1). Human bone marrowederived MSC mRNA expression of HO-1 Several inflammatory effectors are known to influence HO-1 expression [23,24,34e36]. Anti-inflammatory TGFb [24] and pro-inflammatory TNFa or interleukin-1b [23] are a few reported to induce HO-1 in epithelial and endothelial cells, respectively. Therefore, to elucidate the expression of HO-1 in MSCs under various inflammatory states, HO-1 mRNA levels at steady state and after exposure to IFNg, TNFa and/or TGFb for 24, 48 or 72 h were measured in three MSC donors (BMH01, BMH04 and BMH06) cultured in 10% PL media with the use of qRT-PCR. A549 were included as a positive control for constitutive HO-1 expression [24,37]. b-Actin gene levels were used as a reference control for computing relative expression levels of HO-1 (Supplementary Figure S2). Fold change was calculated by normalizing the relative expression of samples to the relative expression of A549. The data from three independent experiments demonstrated that comparable to HO-1 transcript levels in A549 resting MSCs correlated with a statistically lower quantity of HO-1 (P < 0.001, P < 0.01 and P < 0.05; Figure 1A). Likewise, MSCs primed with IFNg, TNFa and/or TGFb for 24, 48 and 72 h did not elicit considerable amounts of HO-1 above that expressed at steady state (P > 0.05; Figure 1A) or A549 (P < 0.0001, P < 0.001, P < 0.01 and P < 0.05; Figure 1A). The failure to identify HO-1 upregulation was not due to insufficient priming or use of metabolically inactive MSCs because statistically significant levels of IDO1, an IFNg-inducible enzyme, were expressed in MSCs after 24-, 48- and 72-h exposure to IFNg, albeit with or without TNFa and TGFb (P < 0.0001 and P < 0.01; Figure 1B). Indeed, these findings dispute prior reports that HO-1 is constitutive in MSCs and augmented in response to a pro-inflammatory stimulus, IFNg [20,21]. However, when comparing the current findings with previous reports, it is conceivable that inconsistency in MSC media is a source of these disparate results, with HO-1 expression characterized herein through the use of MSCs expanded in PL versus FBS media. Comparative analyses of different types of media used to culture MSCs have demonstrated the effects that media have on the immune profile of MSCs primed with IFNg þ TNFa; de novo
5
IDO1 and inducible nitric oxide synthase expression varies in PL-expanded and fetal calf serum (FCS)expanded MSCs [38]. Thus, it is legitimate to investigate HO-1 expression in MSCs expanded under differing culture medium conditions. Because earlier reports observed an upregulation of constitutive HO-1 in FBS-expanded MSCs after 24-h exposure to IFNg, HO-1 transcripts were quantified at steady state or succeeding 24-h IFNg stimulation in six MSC donors cultured in 10% PL or FBS media with the use of qRTPCR. A549 were included as a positive control for constitutive HO-1 expression [24,37]. b-Actin gene levels were used as a reference control for computing relative expression levels of HO-1 (Supplementary Figure S2). Fold change was calculated by normalizing the relative expression of samples to the relative expression of A549. The combined data from six donors demonstrated that measurable levels of HO-1 did not statistically differ between MSCs cultured in PL or FBS media (P > 0.05; Figure 1C). Likewise, IFNg priming did not correlate with a statistically significant variance in HO-1 mRNA levels between differentially cultured MSCs (P > 0.05; Figure 1C). Moreover, compared with resting MSCs, IFNg exposure did not associate with a statistical upregulation of HO-1 albeit in PL- or FBS-expanded MSCs (P > 0.05; Figure 1C). The failure to detect HO-1 transcripts was not due to the inability of the primers to recognize the HO-1 gene because statistically significant levels of HO-1 were observed in A549 (P < 0.0001, P < 0.001, P < 0.01; Figure 1C). Additionally, the lack of detectable HO-1 transcripts succeeding IFNg priming is not due to the use of an insufficient concentration of IFNg, because data from five combined MSC donors grown in PL or FBS media demonstrated that detectable levels of HO-1 did not statistically diverge between resting MSCs and MSCs exposed to different concentrations of IFNg (P < 0.0001; Supplementary Figure S4A). The failure to identify HO-1 upregulation was not due to the inability of IFNg to prime MSCs or the use of metabolically inactive MSCs, because all indicated concentrations of IFNg correlated with a substantial increase in de novo IDO1 (P < 0.0001; Supplementary Figure S4B). HO-1 protein expression in resting and cytokine-treated MSCs Expression of HO-1 is regulated at the transcriptional level, with each stimuli differentially affecting mRNA stability [39,40]. Accordingly, it is equally consistent that the lack of detectable HO-1 transcripts was a direct result of mRNA instability. Thus, the protein levels of HO-1 at steady state and after 24, 48 or 72 h of IFNg, TNFa and/or TGFb stimulation were assessed by means of immunoblotting.
6
S. R. Patel et al.
Because the level of HO-1, mRNA did not differ among MSC donors or differential culture media, protein expression and subsequent studies were performed with the use of a PL-cultured MSC donor, BMH04. A549 were included as a positive control for constitutive HO-1 expression. b-Actin protein detection was included as a loading control. The findings from two independent experiments indicated that HO-1 is undetectable at steady state and after exposure to inflammatory conditions in MSCs by means of immunoblotting (Figure 2A). The failure to observe HO-1 was not due to a deficiency in the sensitivity of the assay to detect HO-1 because constitutive HO-1 was observed in A549 (Figure 2A). Moreover, the absence of HO-1 in MSCs was not a result of metabolic or priming insufficiencies, because IFNg-primed MSCs demonstrated substantial de novo expression of IDO1 protein (Figure 2B). MSCs suppress T-cell proliferation independently of HO-1 Although mRNA and protein analysis indicate that MSCs do not express HO-1, these findings do not rule out that MSCs express HO-1 at levels below the sensitivity threshold of qRT-PCR and immunoblotting performed herein. Thus, the functional requirement of HO-1 in MSC-mediated suppression of T-cell proliferation was tested by blocking the catabolic activity of HO-1. Two pharmacological inhibitors used to inhibit HO-1 catalytic function are
SnPP and ZnPP. Protoporphyrin complexes are heme precursor compounds that regulate heme oxygenase enzymes by competing with the substrate heme and are recognized as heme oxygenase enzymeespecific inhibitors. However, ZnPP has also been shown to irreversibly block the catalytic activity of IDO1 [41], which utilizes heme as a co-factor. Because ZnPP has the capacity to block HO-1 and IDO1 and IFNg-primed MSCs upregulate IDO1, ZnPP is not an ideal utility to determine the specific requirement of HO-1 in MSC-mediated T-cell suppression. Conversely, the specificity of SnPP remains unclear. Therefore, to assess the specificity of SnPP for HO-1, the relative concentration of kynurenine was quantified in the supernatant of SnPP- or ZnPP-treated 293-T-REx cells expressing IDO1 under doxycycline regulation. The findings presented herein demonstrated that at various concentrations, SnPP does not inhibit IDO1 catalytic function (Figure 3A). The failure to regulate IDO1 was not due to the inability of the assay system to detect IDO1 inhibition, because ZnPP potently blocked degradation of i-tryptophan by IDO1 (Figure 3A). Overall, these findings indicate that SnPP does not inhibit IDO1 and thus can be used to assess the necessity of MSC-derived HO-1 in T-cell suppression. To determine the appropriate concentration of SnPP, resting and IFNg-primed MSCs were treated with 1 mmol/L, 10 mmol/L or 50 mmol/L SnPP. MSC viability was assessed 4 days later by the uptake of a nucleic acid marker, 7-AAD, by means of
Figure 2. HO-1 protein expression is undetectable in resting and cytokine-treated MSCs. HO-1 (A) and IDO1 (B) protein expression was analyzed in donor BMH04 resting MSCs and MSCs treated with 10 ng/mL IFNg, 3 ng/mL TNFa, 2 ng/mL TGFb, 10 ng/mL IFNg þ 3 ng/ mL TNFa or 10 ng/mL IFNg þ 2 ng/mL TGFb after 24, 48 or 72 h by means of immunoblot. Untreated A549 was included as a negative and positive control for IDO1 and constitutive HO-1 protein expression, respectively. b-Actin protein expression was used as a loading control.
Role of HO-1 in MSC-mediated T-cell suppression
7
Figure 3. MSC-mediated suppression of T-cell proliferation is mechanistically independent of HO-1. (A) Percentage of IDO1 inhibition was measured in 293-T-REx cells treated with various concentrations of SnPP (circle) or ZnPP (square) for 16 h at 37 C. Percentage of inhibition was calculated as 100% inhibition e [(normalized kynurenine produced with inhibitor)/(kynurenine without inhibitor) 100]. Kynurenine levels were normalized to the background absorbance at 490 nm with the use of media from cells not expressing IDO1. The value for each sample was adjusted to take into consideration changes in protein levels resulting from toxicity that was assessed by the SRB cell toxicity assay; very little toxicity was observed at the concentrations of protoporphyrin tested (data not shown). (B) Viability of resting (white bar) and IFNg-primed (black bar) BMH04 MSC treated with 1 mmol/L, 10 mmol/L or 50 mmol/L SnPP or a volume equivalent (VE) of DMSO for 4 days was assessed by use of the uptake of a nucleic acid marker, 7-AAD, by means of flow cytometry. Percentage of 7-AADþ was defined as the percentage of MSCs positive for 7-AAD. (C) Proliferation assays were performed with the use of anti-CD3/CD28 Dynabead-activated, CFSE-labeled PBMCs from healthy donors. Bead-activated, CFSE-labeled PBMCs were cultured in the presence or absence of donor BMH04 resting MSCs for 4 days at an MSC:PBMC ratio of 1:5. Where indicated, the catalytic inhibitor of HO-1, SnPP, was added to PBMC alone or PBMC:MSC co-cultures at 10 mmol/L. PBMC proliferation was assessed by means of flow cytometry and defined as CFSE dilution. Resting PBMCs were included as a negative gating control (CFSEhi) for quantifying percent proliferation, which was measured as the percentage of CFSElow leukocytes. Percent proliferation was subsequently used to calculate percent inhibition. Error bars represent mean standard error of the mean. Statistical analysis was performed with the use of one-way ANOVA with Dunnett’s post-test (B) and two-way ANOVA with Sidak’s post-test (C). Illustrated data are combined findings from three distinct experiments that used PBMCs from differing donors. ****P < 0.0001, ***P < 0.001, **P < 0.01.
flow cytometry. The percentage of 7-AADþ was defined as the percentage of MSCs positive for 7-AAD; live cells do not take up 7-AAD. The data from three independent studies indicated that 50 mmol/L SnPP induced death in roughly 20e40% of resting and IFNg-primed MSCs (P < 0.0001, P < 0.001; Figure 3B). The observed cellular death was not caused by toxic effects of DMSO used to dissolve the SnPP, because volume-equivalent DMSOtreated MSCs were
