of September 21, 2018.This information is current as

Lymphocytes and Cancer CellsExpression on Human-Activated TPropionibacteria Induces NKG2D Ligand Propionic Acid Secreted from

Jurlander and Søren SkovFalsig Pedersen, Peter Stougaard, Helle Rüsz Hansen, Jesper Lars Andresen, Karen Aagaard Hansen, Helle Jensen, Stine

http://www.jimmunol.org/content/183/2/897doi: 10.4049/jimmunol.08030142009;

2009; 183:897-906; Prepublished online 24 JuneJ Immunol 

Referenceshttp://www.jimmunol.org/content/183/2/897.full#ref-list-1

, 14 of which you can access for free at: cites 42 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2009 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Propionic Acid Secreted from Propionibacteria InducesNKG2D Ligand Expression on Human-Activated TLymphocytes and Cancer Cells1

Lars Andresen,2* Karen Aagaard Hansen,2* Helle Jensen,* Stine Falsig Pedersen,†

Peter Stougaard,‡ Helle Rusz Hansen,§ Jesper Jurlander,¶ and Søren Skov3*

We found that propionic acid secreted from propionibacteria induces expression of the NKG2D ligands MICA/B on activated Tlymphocytes and different cancer cells, without affecting MICA/B expression on resting peripheral blood cells. Growth superna-tant from propionibacteria or propionate alone could directly stimulate functional MICA/B surface expression and MICA pro-moter activity by a mechanism dependent on intracellular calcium. Deletion and point mutations further demonstrated that aGC-box motif around �110 from the MICA transcription start site is essential for propionate-mediated MICA promoter activity.Other short-chain fatty acids such as lactate, acetate, and butyrate could also induce MICA/B expression. We observed a strikingdifference in the molecular signaling pathways that regulate MICA/B. A functional glycolytic pathway was essential for MICA/Bexpression after exposure to propionate and CMV. In contrast, compounds with histone deacetylase-inhibitory activity such asbutyrate and FR901228 stimulated MICA/B expression through a pathway that was not affected by inhibition of glycolysis, clearlysuggesting that MICA/B is regulated through different molecular mechanisms. We propose that propionate, produced either bybacteria or during cellular metabolism, has significant immunoregulatory function and may be cancer prophylactic. The Journalof Immunology, 2009, 183: 897–906.

T he ability of the immune system to recognize infected ortransformed cells is dependent on the induction of immu-nostimulatory molecules by distressed cells. One of the

principal sensing mechanisms is the NKG2D/NKG2D ligandsystem.

NKG2D ligands are induced on the cell surface of numerousstressed, transformed, and infected cells, while the expression onhealthy human cells is low. The immune system recognizesNKG2D ligand-positive cells through the NKG2D receptor, a ma-jor activating receptor on CD8 T lymphocytes, NK cells, and some�/� T lymphocytes and activated CD4 T lymphocytes (1, 2).

A central step in activation of the NKG2D/NKG2D ligand path-way is the induction of NKG2D ligands on stressed cells. Severaldifferent proteins act as NKG2D ligands: MHC class I polypep-tide-related sequence A (MICA)4 and MICB belong to the MIC

gene family, and UL16 binding protein 1–4 (ULBP1–4) and reti-noic acid early transcript 1G (RAET1G) belong to the RAET1gene family (1, 3, 4). The plethora of NKG2D ligands most likelyreflects an evolutionary protection against pathogen-inflicted inhi-bition of NKG2D ligand expression. Several different forms ofcellular stress have been implicated in increased NKG2D ligandexpression, including heat shock, virus infection, inflammatory cy-tokines, histone deacetylase (HDAC) inhibitors, retinoic acid, pro-teasome inhibitors, TLR signaling, and DNA damage response (3,5–13).

We have previously shown that intracellular calcium, GSK-3activity, and Sp1 binding to the promoter region is involved inregulation of MICA expression after HDAC inhibitor treatment(12). In agreement with this, Venkataraman et al. have shown thatheat shock induces expression of MICA/B through Sp1 and Hsf-1promotor binding (14). Gasser et al. have shown that ATM/ATRactivity is involved in NKG2D ligand expression following DNAdamage response (10). Despite these studies, the molecular signal-ing pathway that regulates NKG2D ligand expression is not fullyunderstood; in particular, a potential main pathway that regulatesNKG2D ligand expression remains to be found.

Propionibacteria converts carbohydrates to short-chain fatty ac-ids (SCFA), especially propionic acid, through anaerobic metab-olism. Propionibacteria in the proximal colon makes an average of25 mM propionate and during in vitro culture propionibacteriaproduces between 26 and 40 mM propionate (15, 16). Severaltypes of propionibacteria are nontoxic and considered probiotic,meaning they positively affect the host after ingestion. It has beenshown that propionibacteria can induce apoptosis of colorectal car-cinoma cells via propionate secretion (15) and that preoperative

*Department of Veterinary Disease Biology, †Cell and Developmental Biology, ‡De-partment of Ecology, and §Department of Pharmaceutics and Analytical Chemistry,University of Copenhagen, Copenhagen, Denmark; and ¶Department of Hematology,The State Hospital, Copenhagen, Denmark

Received for publication September 11, 2008. Accepted for publication May 14,2009.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This study was supported from the Danish Medical Research Council, the NovoNordisk Foundation, and the A.P. Møller Foundation for Advancement of MedicalScience to S.S. K.A.H. was supported by the Danish Cancer Society. H.J. was sup-ported by the Novo Nordisk Scholarship program.2 L.A. and K.A.H. contributed equally to this work.3 Address correspondence and reprint requests to Dr. Søren Skov, Laboratory of Im-munology, Faculty of Life Sciences, University of Copenhagen, Stigbøjlen 7, 1870Frederiksberg C, Denmark. E-mail address: sosk@life.ku.dk4 Abbreviations used in this paper: MICA/B, MHC class I polypeptide-related se-quence A and B; 2fDG, 2-fluoro-2-deoxy-D-glucose; 6-AN, 6-aminonicotinamide;BCECF-AM, 2�,7�-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester;DiO, 3,3�-dilinoleyloxacarbocyanine perchlorate; HDAC, histone deacetylase; IAA,iodoacetic acid; pHi, intracellular pH; PI, propidium iodide; PPP, pentose phosphate

pathway; RAET1G, retinoic acid early transcript 1G; SCFA, short chain fatty acid;siRNA, small interfering RNA; TCA, tricarboxylic acid; TPI, triosephosphate isomer-ase; ULBP1–4, UL16 binding proteins 1–4; WT, wild type.

Copyright © 2009 by The American Association of Immunologists, Inc. 0022-1767/09/$2.00

The Journal of Immunology

www.jimmunol.org/cgi/doi/10.4049/jimmunol.0803014

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

treatment with Propionibacterium granulosum increased survivalof patients with colorectal carcinoma (17, 18).

Certain propionibacteria are pathogenic. The acne-causing Pro-pionibacterium acnes stimulates macrophages through TLR2 andprimes them for proinflammatory cytokine production (19, 20).

Propionate can also be produced during metabolism and can begenerated from propionyl-CoA, an end product in the metabolismof the amino acids valine, isoleucine, threonine, and methionine, aswell as odd-chain fatty acid oxidation and degradation of choles-terol. Propionyl-CoA is metabolized to succinyl-CoA and isthereby feed into the TCA cycle. Defects in propionyl-CoA me-tabolism, mainly observed in patients with propionyl-CoA carbox-ylase gene defects (21), leads to increased propionate levels in theblood (propionic acidemia). Continuous propionic acidemia leadsto neurodegeneration, mental retardation, cardiomyopathy, andhigh frequency of infections (22, 23).

Here we show that propionibacteria or purified SCFA caninduce gene activation and surface expression of the NKG2Dligands MICA/B. Of interest, we could separate the variousinducers of MICA/B by their dependency on cellular glycolysis,suggesting that different signaling pathways regulate MICA/Bexpression.

Materials and MethodsCells and reagents

Jurkat E6-1 (JE6-1) T cells were from American Type Culture Collection;Jurkat T cells stably transfected with large T Ag from VS40 virus (JTag-9)were provided by C. Geisler (University of Copenhagen). JTag-9 cellswere primarily used for transient transfection studies. Arh77 (humanplasma cell leukemia), DOHH-2 (human B cell lymphoma), Cem (humanT cell leukemia), Granta (human B cell lymphoma), Aml193 (human acutemyeloid leukemia), and Jurkat T cells were grown in RPMI 1640 (Sigma-Aldrich) supplemented with 10% FBS, 2 mmol/L glutamine, 2 mmol/Lpenicillin, and streptomycin. HT-29 cells (human colon adenocarcinoma)were provided by B. Deibjerg Kristensen (Herlev Hospital, Herlev, Den-mark). HT-29 cells were grown in McCoy’s 5A medium (Sigma-Aldrich)supplemented with 10% FBS, 2 mmol/L glutamine, 2 mmol/L penicillin,and streptomycin. Buffy coats from healthy human volunteer donors wereobtained from The State Hospital, Copenhagen. PBMC were obtained us-ing Lymphoprep (Axis-Shield) according to the description of the manu-facturer. Cells isolated from human blood were grown in RPMI 1640 me-dium supplemented with 150 U/ml IL-2 (Boehringer Mannheim).Polyclonal T cells were expanded according to the protocol from Dyna-beads CD3/CD28 T cell expander (Invitrogen); cells were used 3 days afterrestimulation. The HDAC inhibitor FR901228 was provided by the Na-tional Cancer Institute (Bethesda, MD). 2-Fluoro-2-deoxy-D-glucose, am-monium acetate, sodium propionate (�99%), sodium butyrate (98%), So-dium DL-lactate (�99%), BAPTA-AM, EGTA, iodoacetic acid (IAA)caproic acid, sodium succinate, and 6-aminonicotinamide (6-AN) werefrom Sigma-Aldrich. Caproic acid was neutralized with NaOH before use.Recombinant human NKG2D-Fc chimera was from BD Biosciences. Am-monium formate (�98%) was acquired from Riedel-de Haen, and metha-nol was from BDH Chemicals.

Bacterial cultures

Three propionibacterium strains, Propionibacterium freudenreichiisubsp. freudenreichii (F20271), P. freudenreichii subsp. shermanii(S4902), and Propionibacterium acidipropionici (A20272) were col-lected from Deutsche Sammlung von Mikroorganismen und Zellkul-turen. The propionibacteria were grown under anaerobic conditions inLuria-Bertani media at 24°C. Cultivation was conducted in anaerobicjars (Anaerocult; Merck) with Anaerotest strips (Merck) to control theanaerobic environment. Pseudomonas aeruginosa (PAO1) bacteriumwas provided by M. Kolpen (University of Copenhagen) and grown inox bouillon (Statens Serum Institut, Copenhagen Denmark). Esche-richia coli OneShot OmniMAX 2-T1

R cells (Invitrogen) were grown inLuria-Bertani media at 37°C.

Flow cytometry

For surface staining, cells were washed twice in cold PBS and stained withPE-coupled MICA/B Ab (558352; BD Biosciences) at a dilution of 1/100

for 30 min at 4°C, washed, resuspended, and analyzed in PBS. For GFPexpression, cells were washed, resuspended, and analyzed in PBS. Dataacquisition and flow cytometric analysis were done on a FACSCaliburusing CellQuest software (BD Biosciences). Results are always show-ing forward scatter/side scatter on a linear scale and fluorescence on alog10 scale. JE6-1 cells were stained with recombinant humanNKG2D-Fc chimera at a dilution of 1/100 for 30 min at 4°C. Secondarystaining was made with FITC-coupled Ab against human IgG1 (Dako).For annexin V staining the cells were stained with annexin V-FITC (BDBiosciences, catalog no. 556420) in combination with either control-IgG-PE (BD Biosciences, catalog no. 555574) or MICA/B-PE Ab inannexin V staining buffer (10 mM HEPES (pH 7.4), 140 mM NaCl, 2.5mM CaCl2) for 30 min at 4°C.

Cytolytic assay

The functional cytolytic assay was performed as described by Kane et al.(24). In brief, the target cells were incubated with 0.6 �M 3,3�-dilinoley-loxacarbocyanine perchlorate (DiO; Invitrogen) and allowed to rest in theincubator for 6 h. The cells were then split in two and treated with orwithout 10 mM sodium propionate. Twelve hours later the cells weremixed with effector cells. As effector cells we used PBMCs obtained fromhealthy blood donors and cultured them at a concentration of 5 � 106/mlin RPMI 1640 supplemented with 10% human serum (Lonza) and 20 ng/mlIL-15 (R&D Systems) for 3 days. Effector cells and target cells were mixedat a ratio of 10:1 in 200 �l of media with 10% human serum in a round-bottom microplate. For blocking experiments a monoclonal NKG2D block-ing Ab clone 149810 (R&D Systems) was added at a final concentration of2.5 �g/ml. After 4 h in the incubator the cells were stained with propidiumiodide (PI) and immediately analyzed by flow cytometry. To measure cy-tolytic activity we gated on DiO-positive cells and recorded the percentageof PI-positive cells.

Transient transfections

JTag-9 cells were transiently transfected with the Nucleofector kit (Amaxa)according to the protocol of the manufacture. In brief, 2 � 106 cells wereresuspended in 100 �l of Cell Line Nucleofector Solution V, mixed with2.4 �g of plasmid or 0.8 nmol of small interfering RNA (siRNA) andpulsed using the Nucleofector program G10. The siRNA sequences were:small interfering triosephosphate isomerase (siTPI)_266: CCUGGCAUGAUCAAAGACU and siTPI_1188: CCACCCAUGUGAGGGAAUA. si-Control was described previously (25).

Plasmids, generation of promoter constructs, and mutagenesis

The plasmids pEQ336 (encoding the CMV IE promoter, but without any IEcoding sequences) and pEQ326 (encoding the CMV IE2) were provided byA. Geballe (Fred Hutchinson Cancer Research Center, Seattle, WA). Thep3.2k wild-type (WT)-GFP plasmid, the p3.2k GC-mut-GFP plasmid, and5� deletion constructs has previously been described (25). Promoter activitywas calculated by multiplying percentage of GFP-expressing cells with themean fluorescence of these cells.

Real time RT-PCR analysis

RNA was isolated using TRIzol (Invitrogen) and reverse-transcribed usingSuperScript III reverse transcriptase enzyme (Invitrogen). PCR was per-formed using standard conditions. MICA primer sequences were:MICA_523F, 5�-GCCATGAACGTCAGGAATTT-3�; and MICA_760R,5�-GACGCCAGCTCAGTGTGATA-3�. The housekeeping gene ribo-somal protein, large, P0 (RPLP0) was used as a loading control. RPLP0primer sequences were: RPLP0 forward, 5�-GCTTCCTGGAGGGTGTCC-3�; and RPLP0 reverse, 5�-GGACTCGTTTGTACCCGTTG-3�. For quan-titative RT-PCR analysis the PCR was performed using the Brilliant SYBRGreen QPCR Master Mix kit (Stratagene) and run on a StratageneMx3000P thermocycler.

Measurement of intracellular pH

JE6-1 cells were resuspended in RPMI 1640 media at 2 � 106 cells/ml,and measurements of intracellular pH (pHi) were performed spectro-photometrically using the pH- sensitive probe 2�,7�-bis(2-carboxy-ethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM), es-sentially as described previously (26). An aliquot (5 ml) of cells wasincubated with BCECF-AM (3.2 mM) for 30 min at 37°C. The sampleswere centrifuged for 30 s at 700 � g, washed once in RPMI 1640medium, and resuspended in 1.25 ml of RPMI 1640 medium. The sam-ples were placed in a magnetically stirred cuvette in a fluorescencespectrophotometer model PTI RatioMaster (Photon Technology

898 PROPIONATE INDUCES MICA/B EXPRESSION

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

International), controlled by Felix software. BCECF emission was de-tected at 525 nm after excitation at 445 nm and 500 nm. Backgroundfluorescence was subtracted before calculation of the 445 nm/500 nm

fluorescence ratio, and a three-point, linear calibration of BCECF ratioto pHi was performed using the high K�/nigericin technique as de-scribed previously (26).

FIGURE 1. MICA/B expression by Jurkat T cells after exposure to supernatant from anaerobic propionibacteria. A, Three propionibacteria, P. freuden-reichii subsp. freudenreichii (F20271), P. freudenreichii subsp. shermanii (S4902), and P. acidipropionici (A20272), were grown under anaerobic con-ditions. E. coli and P. aeruginosa (PAO1) were grown under aerobic conditions. JE6-1 cells were incubated with a 1/10 dilution of the collected supernatantand incubated for 18 h. As a positive control we used JE6-1 cells stimulated with 20 ng/ml HDAC inhibitor FR901228 for 18 h. Cells were then stainedwith anti-MICA/B Ab and analyzed by flow cytrometry. Data shown are representative of two experiments. The bar graph shows mean � SD from twoindependent experiments. Selected data are also shown as dot plots. B, JE6-1 cells were treated with different concentrations of propionate or 20 ng/mlFR901228 for 18 h. The cells were stained with anti-MICA/B Ab and analyzed by flow cytometry. The bar graph shows means � SD from two independentexperiments. Selected data are also shown as dot plots. C, JE6-1 cells were treated for the indicated time with 10 mM propionate, stained with anti-MICA/BAb, and analyzed by flow cytometry. D, JE6-1 cells were incubated with propionate as indicated for 18 h and then stained with annexin V-FITC incombination with either control IgG-PE or MICA/B-PE Abs followed by PI staining. Stained cells were analyzed by flow cytometry. The dot plots showPI-negative cells with annexin V-FITC fluorescence on the y-axis and MICA/B-PE fluorescence on the x-axis. Numbers indicate percentage gated cells ineach quadrant. The figure shown is representative of three independent experiments. E, JE6-1 cells were treated with 20 ng/ml FR901228, 10 mMpropionate, bacteria growth supernatant from F20271, A20272, S4902, E. coli, or PAO1 and incubated for 18 h. The cells were stained with recombinanthuman NKG2D-Fc and analyzed by flow cytrometry. The bar graph shows means � SD from two independent experiments. F, Granta cells were labeledwith 0.6 �M DiO for 6 h and divided into a control group or treated with 10 mM propionate. After another 12 h the cells were mixed with PBMCs at aneffector/target ratio of 10:1 and incubated for 4 h in the incubator. For the blocking experiment anti-NKG2D Ab was added to the wells at a finalconcentration of 2.5 �g/ml. The cells were then stained with PI and analyzed by flow cytometry. The figure consists of four dot plots showing DiO-positivecells with PI fluorescence on the y-axis and DiO fluorescence on the x-axis. The two upper plots shows the experiment without blocking Ab (�ab) andthe two lower plots show an experiment with blocking Ab (�ab). The numbers indicate percentage gated cells in each quadrant. The figure shown isrepresentative of three independent experiments.

899The Journal of Immunology

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Measurement of intracellular propionate by HPLC-electrospraymass spectrometry

JE6-1 cells (2 � 106 cells/ml) were grown in RPMI 1640 media withoutFBS and stimulated with 20 ng/ml FR901228 or 10 mM propionate for 0,2, 4, and 18 h. The cells were pelleted and resuspended in 1/10 methanol/water. The liquid chromatographic (LC) system used for separation of or-ganic acids consisted of an Agilent 1100 series HPLC pump system (Agi-lent Technologies) and a Hamilton PRP-X100 analytical column (150 �4.1 mm, � 3 �m). The mobile phase was a mixture of 50 mM ammoniumformate and 10% methanol and its flow rate was 0.7 ml/min. The injectionvolume was 50 �l and biological samples were filtered through a 0.45-�mnylon filter before injection. HPLC-electrospray mass spectrometry mea-surements were performed on a Hewlett-Packard series 1100 MSD singlequadropole mass spectrometer with an electrospray ionization interfaceoperating in the negative ionization mode. Nitrogen was used as nebulizinggas at a pressure of 40 psi at 12 L/min; the temperature was 350°C and thecapillary voltage 4000 V. The instrument was operated at cone voltage of50 V in either the full scan mode or in the selected ion monitoring mode(m/z of 73 and m/z of 89 for detecting propionic and lactic acid, respec-tively) (for further details, see Ref. 27). The system was optimized on 0.5mM propionic and L-lactic acid.

TPI assay

The activity of TPI was measured by adding 5 �l of cell lysate to 85 �l ofreaction mixture consisting of 0.025 M triethanolamine (pH 7.9), 0.1 mM

NADH (Sigma-Aldrich, catalog no. N6660), and 1 U �-glycerophosphatedehydrogenase (Sigma-Aldrich, catalog no. G6880). The mixture was in-cubated 10 min at room temperature before the addition of 10 �l of 30 mMDL-glyceraldehyde 3-phosphate solution (Sigma-Aldrich, catalog no.G5251), which starts the reaction. The decrease in A340 nm was monitoredin 10-s intervals, and specific activity in mU/mg was calculated as�A340 nm � volume � 20/(6.3 � protein concentration).

GAPDH assay

GAPDH activity was determined using a KDalert GAPDH assay kit fromAmbion. Briefly Jurkat E6-1 cells were lysed with KDalert lysis buffer.Lysate (20 �l) was added to 180 �l of reaction mix in a microplate andincubated at room temperature for 15 min. A standard curve was alsoincluded. The OD630 was measured in a microplate reader and the activityin the samples was calculated.

Immunoblotting

Immunoblotting was done as previously described (25). The TPI Ab wasfrom Santa Cruz.

FIGURE 2. Activated T cells and various cancer cells express MICA/Bafter propionate treatment. A, The indicated cells were treated with andwithout 10 mM propionate for 18 h and stained with anti-MICA/B; the gateis set arbitrarily to enlighten the difference in staining intensity. B, PBMCswere stimulated with or without CD3/CD28 beads and IL-2 as described inMaterials and Methods. The activated and naive PBMCs were exposed to10 mM propionate or 20 ng/ml FR901228 for 18 h, stained with anti-MICA/B Ab, and analyzed by flow cytometry.

FIGURE 3. Propionate activates MICA promoter activity. A, JE6-1cells were treated with 20 ng/ml FR901228 or 10 mM propionate for4 h. Total RNA was then extracted and used for quantitative RT-PCR.MICA mRNA expression was normalized to the expression of thehousekeeping gene RPLP0 and displayed as fold expression relative tocontrol. B, JTag-9 cells were transfected with 5� deletions of the MICApromoter construct as indicated. After 24 h, the cells were either leftuntreated (control) or treated with 10 mM propionate for 18 h. The cellswere analyzed for GFP fluorescence by flow cytometry. The bar graphshows means � SD from two independent experiments. C, JTag-9 cellswere transfected with p-115-GFP and p�2-GFP and incubated for 24 h.The cells were stimulated with vehicle control, 10 mM propionate, 20ng/ml FR901228, or 1:10 supernatant from F20271, A20272, S4902, E.coli, or PAO1 for 18 h. The next day, cells were analyzed for GFPfluorescence by flow cytrometry. The bar graph shows means � SDfrom two independent experiments. D, JTag-9 cells were either trans-fected with a WT MICA promoter construct (p3.2k-WT-GFP) or aMICA promoter construct with a GC-box mutation (p3.2k-GCmut-GFP). The next day, the cells were either left untreated (control) ortreated with 20 ng/ml FR901228, 10 mM propionate, and 10 mM bu-tyrate for 18 h. The cells were analyzed for GFP fluorescence by flowcytometry. The bar graph shows means � SD from two independentexperiments.

900 PROPIONATE INDUCES MICA/B EXPRESSION

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

ResultsPropionic acid from propionibacteria regulates MICA/Bexpression

We were interested in identifying bacterial components that couldmodulate the NKG2D/NKG2D ligand axis of immunoactivation.Various bacteria and their growth supernatants were screened forthe ability to induce MICA/B surface expression on Jurkat T cells;we have previously shown that Jurkat T cells can be induced toexpress MICA/B after different kinds of cell stress (12, 25). Fig.1A shows surface MICA/B expression after 18 h of cocultivationof Jurkat T cells with supernatants from E. coli, P. aeruginosa, andthree different strains of propionibacteria: P. freudenreichii subsp.freudenreichii (F20271), P. freudenreichii subsp. shermanii(S4902), and P. acidipropionici (A20272). Growth supernatantsfrom the three strains of propionibacteria induced strong MICA/Bexpression, whereas E. coli and P. aeruginosa had no effect. Thepropionibacterial supernatants were potent; a 100-fold dilution stillinduced MICA/B expression (data not shown).

Propionibacteria produces propionate and other SCFAs dur-ing their anaerobic growth. To determine whether propionatecould be a causative factor in MICA/B expression, we exposedJurkat T cells to purified propionate. There was a robust induc-tion of MICA/B expression by 10 –20 mM propionate and lessbut distinct expression down to 1 mM (Fig. 1B). Labeling ofJurkat T cells with an isotype-matched control Ab was not af-fected by the propionibacteria supernatants or propionate (datanot shown). A kinetic analysis using 10 mM propionate showedthat MICA/B expression could be detected 8 h after pro-pionate exposure, with maximal expression within 14 to 24 h(Fig. 1C). The propionate-mediated MICA/B expression wascomparable to the induction mediated by HDAC inhibitors (12),with the general distinction that the intensity (mean fluores-cence intensity) of MICA/B staining was higher after treatmentwith the HDAC inhibitor FR901228, and the percentage ofMICA/B-positive cells was higher after propionate treatment(Fig. 1B).

FIGURE 4. Propionate-mediated expression of MICA/B is dependent on intracellular calcium. A, JE6-1 cells were loaded with the indicated concen-trations of BAPTA-AM for 30 min before treatment with vehicle control or 10 mM propionate for 18 h. Cells were then stained with anti-MICA/B Ab andanalyzed by flow cytometry. The bar graph shows means � SD from two independent experiments. B, JE6-1 cells were loaded with different concentrationsof EGTA as indicated for 30 min before treatment with vehicle control or 10 mM propionate for 18 h. Cells were then stained with anti-MICA/B Ab andanalyzed by flow cytometry. The bar graph shows means � SD from two independent experiments. C, JTag-9 cells were transfected with p-173-GFP orp3.2k-WT-GFP and incubated for 24 h. Cells were then loaded with 5 �M BAPTA-AM or 5 mM EGTA for 30 min and subsequently treated with vehiclecontrol (data not shown) or 10 mM propionate. The next day, cells were analyzed for GFP fluorescence by flow cytometry. The bar graph shows means �SD from two independent experiments. D, JE6-1 cells were incubated with BCECF-AM for 30 min at 37°C and pHi was measured as described in Materialsand Methods. BCECF-AM-loaded cells were stimulated with 10 mM propionate, (E) 10 mM butyrate, or (F) 20 ng/ml FR901228. G, JE6-1 cells weregrown in media without FBS and treated with vehicle (blank), 20 ng/ml FR901228, or 10 mM propionate for 0 or 4 h. The cells were harvested andresuspended in 500 �l of 10% methanol and analyzed by HPLC-electrospray mass spectrometry as described in Materials and Methods.

901The Journal of Immunology

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

In some tumor cell lines propionic acid has been shown to in-duce apoptosis at concentrations ranging from 10 to 40 mM (15).To reveal whether MICA/B-expressing cells were apoptotic cells,we double stained propionate-treated Jurkat cells with annexin V-FITC and MICA/B-PE Ab. As shown in Fig. 1D, the apoptoticannexin V-positive cells did not express MICA/B, and vice versa,suggesting that the underlying pathway from propionate toMICA/B expression does not connect with the apoptotic pathway.

NKG2D ligand expression induced by propionibacteria growthsupernatant or propionate was also detected with the native puri-fied receptor NKG2D-Fc (Fig. 1E), demonstrating that propionate-induced NKG2D ligand expression can interact with the nativereceptor NKG2D. Coincubation with anti-MICA/B Ab blockedNKG2D-Fc binding to propionate-induced NKG2D ligands by45% (data not shown). Ab blocking is often incomplete, suggest-ing that most propionate-induced NKG2D ligands stem fromMICA/B. To demonstrate the functional ability of propionate-in-duced NKG2D ligands, we made a cytotoxic assay with purifiedblood lymphocytes activated for 3 days with IL-15 to enhanceNKG2D-mediated killing by NK cells. We have previously shownthat Granta cells are efficient target cells for monitoring NKG2Dligand-dependent NK cell activity (12). Fig. 1F shows that Grantacells exposed to 10 mM propionate for 12 h, in contrast to controls,were potently killed by IL-15-activated PBLs. Inclusion of ablocking NKG2D Ab strongly inhibited killing of these cells. Thissuggests that propionate-induced NKG2D ligands are able to func-tionally activate NK cells.

Next we wanted to see if propionate also regulated MICA/Bexpression in other cancer cell types. Fig. 2A shows that propi-onate clearly induced MICA/B expression in HT-29 (colon can-cer), Granta (mantle cell lymphoma), Arh77 (multiple myeloma),DOHH2 (malignant non-Hodgkin’s lymphoma), Aml193 (acutemyelogenous leukemia), and Cem (T cell acute lymphoblastic leu-kemia) cells. We have previously shown that activated, but notresting, CD4 T lymphocytes can be induced to express MICA/B(12). In line with this, propionate enhanced MICA/B surface ex-pression on CD4 T lymphocytes previously stimulated with beadscoupled to anti-CD3/CD28 Abs. Propionate did not affect MICA/Bexpression on resting CD4 T lymphocytes (Fig. 2B).

Molecular regulation of MICA/B by propionate

Real-time PCR analysis demonstrated that propionate leads to anincrease in MICA mRNA level (Fig. 3A), suggesting that propi-onate either stabilizes MICA mRNA or directly stimulates MICApromoter activity.

Using MICA reporter constructs, consisting of 3.2 kb or subse-quent deletions of the proximal MICA promoter in front of GFP(25), we tested the ability of propionate to activate the MICApromotor. As shown in Fig. 3B, propionate stimulated MICA pro-moter activity in transiently transfected Jurkat T cells. Superna-tants from the different propionibacteria were also tested and, asexpected, they induced a robust MICA promoter activity (Fig. 3C).These different promoter activations were comparable to, or higherthan, those observed after optimal HDAC inhibitor treatment,which has previously been shown by us (25). Promotor deletionanalysis showed that the region from �115 to �97, relative to thetranscription start site, was important for propionate-mediated pro-moter activity (Fig. 3B). The interesting region from �115 to �97contains a GC-box motif and has earlier been shown to interactwith the transcription factor Sp1, which is important for MICApromoter activity after heat chock and HDAC inhibitor exposure(14, 25). In agreement, mutation of the GC-box motif from �113to �105 by shifting CCC to AAA in position �107 to �105strongly inhibited propionate-mediated MICA promoter activity

(Fig. 3D). These results strongly suggest that propionate directlystimulates MICA promoter activity through involvement of theproximal GC-box motif.

Propionate regulation of MICA/B through calcium and pH

Intracellular calcium has been implicated in HDAC inhibitor- andCMV-mediated MICA/B expression (25). To assess the role ofcalcium in propionate-induced MICA/B expression, we loaded Ju-rkat T cells with BAPTA-AM (an intracellular calcium chelator) orEGTA (an extracellular calcium chelator). Cells were analyzed forMICA/B surface expression after treatment with propionate. Asshown in Fig. 4, A and B, depletion of calcium either byBAPTA-AM or EGTA abrogated MICA/B expression in a dose-dependent manner. BAPTA-AM also inhibited propionate-medi-ated MICA mRNA production measured by quantitative PCR(data not shown). Both BAPTA-AM and EGTA blocked the pro-pionate-induced promoter activity, measured using the p3.2k WT-GFP and 173-MICA-GFP promoter constructs (Fig. 4C), indicat-ing that intracellular calcium affects transcription downstreamfrom bp �173 relative to the MICA transcription start site. Theseresults demonstrate that intracellular calcium is a critical compo-nent in the propionate-induced signaling pathway leading to MICAgene activation and surface expression. The blocking by EGTAfurther suggests that the rise in intracellular calcium is dependenton extracellular influx.

Propionate causes rapid intracellular acidification, and the de-crease in pHi can, in some circumstances, be regulated by intra-cellular calcium (28–30). To determine whether intracellular acid-ification generally was implicated in MICA/B expression, werecorded pHi changes after exposure to propionate, butyrate, and

FIGURE 5. SCFAs induce MICA/B expression. JE6-1 cells weretreated with the indicated concentrations of (A) acetate, (B) butyrate, or (C)lactate or 20 ng/ml FR901228 for 18 h. The cells were stained with anti-MICA/B Ab and analyzed by flow cytometry. The bar graph showsmeans � SD from two independent experiments. D, JE6-1 cells weretreated with 1 or 10 mM propionate, succinate, or caproate for 18 h andstained with anti-MICA/B Ab and analyzed by flow cytometry. The bargraph shows mean percentage MICA/B-positive cells � SD from threeindependent experiments made in duplicates.

902 PROPIONATE INDUCES MICA/B EXPRESSION

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

FR901228. Jurkat T cells were loaded with the pH-sensitiveBCECF-AM and the pHi was recorded to �7.6 (Fig. 4, D and E).As expected, addition of either 10 mM propionate or 10 mM bu-tyrate caused a rapid intracellular acidification followed by a grad-ual recovery. However, we did not observe any immediate changesin pHi after treatment with FR901228 (Fig. 4F). Likewise, pro-longed exposure to FR901228 (up to 4 h) did not result in pHi

changes (data not shown). These results suggest that intracellularacidification is not essential for MICA/B expression.

Propionate can be generated by cell metabolism, mainlythrough propionyl-CoA generation after catabolism of theamino acids valine, isoleucine, threonine, and methionine, aswell as odd-chain fatty acid oxidation and degradation of cho-lesterol. We initially hypothesized that the various signals thatlead to MICA/B expression might all interfere with metabolismand cause high levels of intracellular propionate, meaning thatpropionate would be a general causative factor regulatingMICA/B expression.

FIGURE 6. Involvement of glycolysis in MICA/B expression. A, JE6-1 cells were loaded with different concentrations of 2fDG as indicated for 1 hbefore treatment with vehicle control or 10 mM propionate for 18 h. Cells were then stained with anti-MICA/B Ab and analyzed by flow cytometry. Thebar graph shows means � SD from two independent experiments. Insert, JTag-9 cells transfected with CMV-GFP vector and stimulated with or without1 mM 2fDG after 24 h. B, Schematic illustration of the glycolysis and pentose phosphate pathways. The target enzymes for inhibitors used are indicatedas: Glu-6P, glucose-6-phosphate; Fru-1:6P, fructose-1,6-bisphosphate; Gly-3P, glyceraldehyde-3-phosphate, DHAP, dihydroxyacetone phosphate; G6PDH,glucose-6-phosphate-dehydrogenase; as well as TPI and GAPDH. C, JTag-9 cells were transfected with siControl or siTPI_1188 siRNAs. Three hours aftertransfection cells were treated with vehicle control, 20 ng/ml FR901228, or 10 mM propionate for 18 h. The cells were stained with anti-MICA/B Ab andanalyzed by flow cytometry. The bar graph shows means � SD from two independent experiments. Selected data are also shown as dot plots. D, Westernblot from JTag-9 cells transfected with siControl, siTPI_266, or siTPI_1188 siRNA. Blots were probed with either anti-TPI Ab or anti-ERK1/2 Ab asloading control. The activity of TPI in cells transfected with siControl or siTPI_1188 was also measured, as described in Materials and Methods. E, JE6-1cells were treated with 2 �M IAA in combination with vehicle control, 10 mM propionate, or 20 ng/ml FR901228 for 18 h. Cells were then stained withanti-MICA/B Ab and analyzed by flow cytometry. The bar graph shows means � SD from two independent experiments. F, JE6-1 cells were treated with0 or 2 �M IAA for 2 h and GAPDH activity was determined as described in Materials and Methods. The bar graph shows mean GAPDH activity aspercentage of untreated cells � SD from three independent experiments. G, JE6-1 cells were treated with 100 �M 6-AN in combination with vehiclecontrol, 10 mM propionate, or 20 ng/ml FR901228 for 18 h. Cells were then stained with anti-MICA/B Ab and analyzed by flow cytometry. The bar graphshows mean � SD from two independent experiments. H, JE6-1 cells were treated with or without 1 mM 2fDG for 1 h before treatment with vehicle control,20 ng/ml FR901228, 10 mM propionate, 10 mM butyrate, or transfection with pEQ326 vector (IE2) for 18 h. Then cells were stained with anti-MICA/BAb and analyzed by flow cytometry. The bar graph shows means � SD from two independent experiments.

903The Journal of Immunology

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Intracellular propionate was measured by mass spectrometryand, as shown in Fig. 4G, treatment with FR901228 did not resultin increased intracellular levels of propionate in Jurkat T cells. Theassay was fully functional as intracellular propionate was readilyobserved after treatment with propionate, and increased propionatewas also detected in the FR901228-treated cells pulsed with pro-pionate before the assay (Fig. 4G). These results strongly suggestthat metabolic buildup of propionate is not an essential integratingcomponent shared by other MICA/B stimulators.

Different SCFAs induce MICA/B expression

Propionibacteria can produce different kinds of SCFAs in additionto propionate, especially acetate, butyrate, and lactate. It is obviousthat the induction of MICA/B by propionibacteria may not besolely related to propionate; it is more likely a combined effect ofseveral different compounds, given the potent effect of the propi-onibacteria supernatant (see Fig. 1A).

Jurkat T cells were treated with different concentrations of ac-etate, lactate, and butyrate for 18 h and analyzed for MICA/Bsurface expression (Fig. 5A–C). Acetate and lactate induced a ro-bust increase in MICA/B expression, although the effect was mostpronounced at relatively high concentrations from 30 to 50 mM.These concentrations are, however, achievable during bacterialmetabolism in the intestine (16, 28, 30). The effect of butyrate wasmore prominent, and sizable MICA/B expression could be de-tected down to 1 mM. Butyrate is a direct HDAC inhibitor (31), sothe observed effect on MICA/B is not surprising. To test whetherthe structure of the SCFA had any importance for MICA/B in-duction we tested succinate, which is a four-carbon SCFA-likebutyrate but with two carboxylic acids instead of one, and ca-proate, which is a six-carbon SCFA. As shown in Fig. 5D, succi-nate could not induce MICA/B expression and caproate was ap-proximately half as potent as propionate to induce MICA/Bexpression, indicating that optimal MICA/B expression is inducedby SCFAs with three- to four-carbon length and only one carbox-ylic acid group.

HDAC inhibitors differ from propionate and CMV with regardto glycolytic dependency of MICA/B expression

MICA/B is primarily expressed in activated/proliferating cells. Ac-tivated T lymphocytes and cancer cells are characterized by a highlevel of metabolism, especially glycolysis, when compared withnaive cells (32, 33). Based on these considerations, we wanted toelucidate the importance of glycolysis for MICA/B expression.

Glycolysis was inhibited using the specific inhibitor 2-fluoro-2-deoxy-D-glucose (2fDG), which is a nonmetabolizing mimetic ofglucose that is more specific than the more widely used 2-deoxy-D-glucose (34). We noted an interesting difference in MICA/B reg-ulation when glycolysis was blocked: propionate-induced MICA/Bsurface expression was strongly inhibited by 2fDG in a dose-de-pendent manner, which on the other hand did not affect MICA/Bexpression after FR901228 treatment (Fig. 6A). The lack of effectby 2fDG on FR901228-mediated MICA/B expression excludesthat the inhibition of propionate-mediated MICA/B is caused bysevere toxicity or depletion of cellular energy. This is further sub-stantiated by Jurkat T cells transiently expressing GFP; here, 2fDGdid not affect GFP expression within 24 h (Fig. 6A, insert). How-ever, as expected, 2fDG inhibited glycolysis as measured by de-creased cellular lactate production (data not shown).

Next we inhibited glycolysis farther downstream (see Fig. 6Bfor illustration). The enzyme TPI, which catalyzes the intercon-version between dihydroxyacetone phosphate (DHAP) and glyc-eraldehyde-3-phosphate (Gly-3P), was targeted by siRNA-medi-ated knockdown. Interestingly, MICA/B expression was actually

superinduced after both propionate and FR901228 exposure (Fig.6C). The siRNA targeting of TPI was effective as shown by West-ern blot analysis and measurement of TPI activity (Fig. 6D); more-over, transiently expressed GFP was not affected by the TPIknockdown (data not shown), excluding gross cell toxicity. Tosubstantiate these data we blocked GAPDH by the potent, althoughmore unspecific, inhibitor IAA (Fig. 6F); again, there was no de-crease of either propionate- or FR901228-mediated MICA/B ex-pression (Fig. 6E).

A possible explanation for these findings could be that MICA/Bexpression is dependent on a signaling pathway linked to proximalglycolytic products. The pentose phosphate pathway (PPP) is de-rived from glucose-6-phosphate (Glu-6P) by the enzyme glucose-6-phosphate dehydrogenase (G6PDH) (see Fig. 6B), and the oxi-dative PPP can be very specifically inhibited by 6-AN (35, 36). Inagreement with this, pretreatment with 6-AN completely inhibitedpropionate-induced MICA/B expression without affecting expres-sion caused by FR901228 exposure (Fig. 6G).

When combining these results with the siTPI/IAA data, ourfindings suggest that proximal glycolysis through the oxidativePPP is essential for propionate-mediated MICA/B surface expres-sion. The HDAC inhibitor FR901228 can bypass this requirementthrough an unknown mechanism.

We decided to look at the glycolytic dependency of a variety ofstimuli that lead to MICA/B activity. An interesting patternemerged: all HDAC inhibitors tested (trichostatin A, FR901228,and butyrate) were not dependent on proximal glycolysis forMICA/B expression. This was in contrast to acetate, lactate, pro-pionate, and CMV, which could not induce MICA/B expression inthe presence of 2fDG (Fig. 6H and data not shown).

DiscussionIn this report we describe that propionic acid and other SCFAs,such as acetate, butyrate, and lactate, secreted from propionibac-teria induce surface expression of the NKG2D ligands MICA/B onseveral cancer cells and activated T lymphocytes, while leavingnaive blood cells unaffected. NKG2D ligands induced by propi-onibacteria bound specifically to the chimeric NKG2D-Fc recep-tor. Approximately half of the propionate-mediated NKG2D-Fcbinding was blocked by incubation with a MICA/B Ab, suggestingthat MICA/B are the main NKG2D ligands expressed. Moreover,we observed a strong NKG2D-dependent NK cell killing of pro-pionate-exposed cancer cells. We anticipate that other ligands arelikely also induced; in this regard, Vales-Gomez et al. have re-cently shown that HDAC inhibitor exposure leads to ULBP2 sur-face expression, in addition to MICA/B (11).

The molecular signaling pathway that mediated propionate-in-duced MICA/B expression was shown to be dependent on intra-cellular calcium. Chelating calcium with BAPTA-AM or EGTAstrongly suppressed propionate-mediated MICA/B surface expres-sion and MICA promotor activity. Additionally, deletion and pointmutation of the MICA promoter construct showed that the GC-boxbetween �113 to �105 relative to the transcription start site isimportant for propionate-induced MICA promotor activity. Heatshock and HDAC inhibitors induce MICA expression through thesame GC-box (14, 25), clearly suggesting that they are generallyimportant for regulation of MICA expression.

Propionibacteria mainly reside in the intestine where they pro-duce SCFAs during fermentation of plant-derived dietary fiber.SCFAs are considered protective against colon carcinogenesis(16), one of the leading causes of cancer deaths in the Western partof the world. Propionate derived from different stains of propi-onibacteria can directly kill different human colorectal carcinoma

904 PROPIONATE INDUCES MICA/B EXPRESSION

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

cell lines (15). Preoperative treatment with different strains of pro-pionibacteria caused beneficial immunostimulation and increasedsurvival of patients with colorectal carcinoma (17, 18). It is tempt-ing to speculate that a part of these effects is mediated by increasedMICA/B expression. In this regard it is interesting that the gastro-intestinal epithelium, which is in direct contact with intraepitheliallamina propria lymphocytes, is one of the only sites in the bodywith continuous elevated MICA/B expression (5, 37).

Propionyl-CoA, which is readily transformed to propionate, isproduced in most cells by catabolism of the amino acids valine,isoleucine, methionine, and threonine, odd-chain fatty acids, andcholesterol. Patients with defects in the catabolism of propionyl-CoA have a severe and harmful buildup of propionic acid in theblood, a state termed propionic acidemia (22). From our studies itis evident that patients with propionic acidemia can suffer fromincreased MICA/B expression. In this regard, note that these pa-tients often have recurrent infections (23), although this is likelyalso influenced by other factors than MICA/B.

During propionic acidemia the accumulated propionyl-CoA cancombine with oxaloacetate leading to methylcitric acid, which canaffect cellular metabolism (38). However, addition of purified2-methylcitric acid did not affect MICA/B induction by propionateor FR901228 (data not shown), suggesting that propionate-inducedMICA/B expression is not caused by methylcitric acid production.

We wondered if other activators of MICA/B expression couldfunction through a metabolic buildup of propionic acid. This didnot seem to be the case, as cellular levels of propionic acid was notchanged by the HDAC inhibitor FR901228 as judged by massspectrometry analysis. This suggests that increased levels of pro-pionate are not a universal factor used by other stimuli that regu-late MICA/B expression.

It is well described that propionate leads to a decrease in pHi

(29), which in some cases can be linked to increases in intracellularcalcium concentration (28, 30). Since we have found that calciumis involved in MICA/B expression after propionate exposure (Fig.4, A and B), which is similar to our previous results with HDACinhibitors and CMV IE2 transfection (25), we hypothesized that adecrease in pHi could be generally involved in MICA/B expres-sion. This was however not the case: FR901228 treatment did notaffect pHi, whereas both propionate and butyrate led to an imme-diate pHi drop. The metabolic difference (described below), wherethe HDAC inhibitors butyrate and FR901228, in contrast to pro-pionate, induced MICA/B independently of glycolysis, further im-plies that a decrease in pHi is not generally coupled to MICA/Bregulation; however, we cannot completely rule out that the de-crease in pHi affects MICA/B expression after propionate or bu-tyrate exposure.

Cancer cells have a high degree of aerobic glycolysis, named theWarburg effect (33), and enhanced glycolysis is also seen in acti-vated T cells (32). Thus, we wondered if glycolysis was linked tothe ability to express MICA/B. The specific inhibitor of glycolysis,2fDG, had a striking impact: propionate-induced MICA/B expres-sion was nearly completely blocked, whereas the HDAC inhibitorFR901228 led to a normal MICA/B increase. We also tried toblock glycolysis farther downstream by siRNA-mediated knock-down of TPI or inhibition of GADPH with IAA. Interestingly, thisdid not inhibit MICA/B expression after propionate exposure; infact, both propionate and FR901228 actually superinducedMICA/B after TPI knockdown. These results suggest that only thebeginning of the glycolytic pathway is critical for MICA/B ex-pression. Since the PPP diverges early from the glycolytic pathwayfrom glucose-6-phosphate, we tried to target this step. Indeedblocking the oxidative PPP with the specific inhibitor 6-AN po-tently inhibited propionate-mediated MICA/B expression without

affecting the expression caused by FR901228. These results clearlysuggest that there are specific differences in the regulation ofMICA/B by propionate and the HDAC inhibitor FR901228.

When broadening our studies to other stimuli causing MICA/Bexpression, an interesting pattern emerged: lactate and CMV ex-posure behaved like propionate, whereas established HDAC inhib-itors such as trichostatin A and butyrate behaved like FR901228.

Propionate can also lead to HDAC inhibitory activity (39); how-ever, it is unclear if propionate functions as a direct HDAC inhib-itor or whether a secondary effect causes elevated cellular HDACinhibitory activity. Cousens et al. have shown that propionate leadsto HDAC inhibitory activity, although the direct in vitro HDACinhibitory activity was rather weak compared with butyrate (31).This could imply the rather trivial explanation that propionatemerely functions as a weak HDAC inhibitor. Several of our find-ings argue against this explanation: 1) propionate and butyrate in-duce MICA/B expression in similar concentrations, which is notconsistent with butyrate being a superior HDAC inhibitor; 2) ac-etate and lactate also lead to MICA/B induction; however, thesecompounds possess limited, if any, direct HDAC inhibitory activ-ity (31); and 3) there is a clear metabolic difference between es-tablished HDAC inhibitors (FR901228, trishostatin A, and bu-tyrate) and propionate with regard to MICA/B induction.

What is the basis for the dissimilar MICA/B regulation mediatedby direct enzymatic HDAC inhibitors and other inducers? We havea number of hypothesis and speculations that may enlighten thisarea. SCFA could cause cellular HDAC inhibitory activity throughproximal glycolysis and the PPP. HDAC inhibitors can directlytarget HDAC enzymes and would therefore not be dependent onintact metabolism. However, butyrate and propionate induceMICA/B expression with similar concentration kinetics (compareFigs. 1 and 5), which is not consistent with butyrate causing su-perior overall HDAC inhibition. This suggests that propionate canalso regulate MICA/B through a pathway not shared by HDACinhibitors. Eventual different signaling pathways may, however,converge at a distal point, as there is no additive effect on MICA/Bexpression by combined exposure to FR901228 and propionate(data not shown). Future studies will hopefully provide more con-clusive and general understanding of the complex MICA/Bregulation.

MICA is strongly expressed at the surface of intestinal epithelialcells from patients with autoimmune celiac disease, and this acti-vates intraepithelial T lymphocytes through NKG2D, leading toenterocyte damage (40, 41). Celiac disease can result in bacterialovergrowth of the small intestine (42), and we think that it is aninteresting hypothesis that the increased MICA expression can becaused by bacterial-produced SCFAs. If this is the case, artificiallowering of SCFA production may be useful in the future treat-ment of celiac disease.

DisclosuresThe authors have no financial conflicts of interest.

References1. Bauer, S., V. Groh, J. Wu, A. Steinle, J. H. Phillips, L. L. Lanier, and T. Spies.

1999. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 285: 727–729.

2. Groh, V., A. Bruhl, H. El-Gabalawy, J. L. Nelson, and T. Spies. 2003. Stimula-tion of T cell autoreactivity by anomalous expression of NKG2D and its MICligands in rheumatoid arthritis. Proc. Natl. Acad. Sci. USA 100: 9452–9457.

3. Cosman, D., J. Mullberg, C. L. Sutherland, W. Chin, R. Armitage, W. Fanslow,M. Kubin, and N. J. Chalupny. 2001. ULBPs, novel MHC class I-related mole-cules, bind to CMV glycoprotein UL16 and stimulate NK cytotoxicity throughthe NKG2D receptor. Immunity 14: 123–133.

4. Bacon, L., R. A. Eagle, M. Meyer, N. Easom, N. T. Young, and J. Trowsdale.2004. Two human ULBP/RAET1 molecules with transmembrane regions areligands for NKG2D. J. Immunol. 173: 1078–1084.

905The Journal of Immunology

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

5. Groh, V., S. Bahram, S. Bauer, A. Herman, M. Beauchamp, and T. Spies. 1996.Cell stress-regulated human major histocompatibility complex class I gene ex-pressed in gastrointestinal epithelium. Proc. Natl. Acad. Sci. USA 93:12445–12450.

6. Jinushi, M., T. Takehara, T. Tatsumi, T. Kanto, V. Groh, T. Spies, T. Suzuki,T. Miyagi, and N. Hayashi. 2003. Autocrine/paracrine IL-15 that is required fortype I IFN-mediated dendritic cell expression of MHC class I-related chain A andB is impaired in hepatitis C virus infection. J. Immunol. 171: 5423–5429.

7. Jinushi, M., T. Takehara, T. Kanto, T. Tatsumi, V. Groh, T. Spies, T. Miyagi,T. Suzuki, Y. Sasaki, and N. Hayashi. 2003. Critical role of MHC class I-relatedchain A and B expression on IFN-�-stimulated dendritic cells in NK cell acti-vation: impairment in chronic hepatitis C virus infection. J. Immunol. 170:1249–1256.

8. Jinushi, M., T. Takehara, T. Tatsumi, T. Kanto, V. Groh, T. Spies, R. Kimura,T. Miyagi, K. Mochizuki, Y. Sasaki, and N. Hayashi. 2003. Expression and roleof MICA and MICB in human hepatocellular carcinomas and their regulation byretinoic acid. Int. J. Cancer. 104: 354–361.

9. Schreiner, B., J. Voss, J. Wischhusen, Y. Dombrowski, A. Steinle,H. Lochmuller, M. Dalakas, A. Melms, and H. Wiendl. 2006. Expression ofToll-like receptors by human muscle cells in vitro and in vivo: TLR3 is highlyexpressed in inflammatory and HIV myopathies, mediates IL-8 release and up-regulation of NKG2D-ligands. FASEB J. 20: 118–120.

10. Gasser, S., S. Orsulic, E. J. Brown, and D. H. Raulet. 2005. The DNA damagepathway regulates innate immune system ligands of the NKG2D receptor. Nature436: 1186–1190.

11. Vales-Gomez, M., S. E. Chisholm, R. L. Cassady-Cain, P. Roda-Navarro, andH. T. Reyburn. 2008. Selective induction of expression of a ligand for theNKG2D receptor by proteasome inhibitors. Cancer Res. 68: 1546–1554.

12. Skov, S., M. T. Pedersen, L. Andresen, P. T. Straten, A. Woetmann, andN. Odum. 2005. Cancer cells become susceptible to natural killer cell killing afterexposure to histone deacetylase inhibitors due to glycogen synthase kinase-3-dependent expression of MHC class I-related chain A and B. Cancer Res. 65:11136–11145.

13. Armeanu, S., M. Bitzer, U. M. Lauer, S. Venturelli, A. Pathil, M. Krusch,S. Kaiser, J. Jobst, I. Smirnow, A. Wagner, et al. 2005. Natural killer cell-me-diated lysis of hepatoma cells via specific induction of NKG2D ligands by thehistone deacetylase inhibitor sodium valproate. Cancer Res. 65: 6321–6329.

14. Venkataraman, G. M., D. Suciu, V. Groh, J. M. Boss, and T. Spies. 2007. Pro-moter region architecture and transcriptional regulation of the genes for the MHCclass I-related chain A and B ligands of NKG2D. J. Immunol. 178: 961–969.

15. Jan, G., A. S. Belzacq, D. Haouzi, A. Rouault, D. Metivier, G. Kroemer, andC. Brenner. 2002. Propionibacteria induce apoptosis of colorectal carcinoma cellsvia short-chain fatty acids acting on mitochondria. Cell Death Differ. 9: 179–188.

16. Scheppach, W., H. P. Bartram, and F. Richter. 1995. Role of short-chain fattyacids in the prevention of colorectal cancer. Eur. J. Cancer 31A: 1077–1080.

17. Isenberg, J., B. Stoffel, U. Wolters, J. Beuth, H. Stutzer, H. L. Ko, andH. Pichlmaier. 1995. Immunostimulation by propionibacteria: effects on immunestatus and antineoplastic treatment. Anticancer Res. 15: 2363–2368.

18. Isenberg, J., H. Ko, G. Pulverer, R. Grundmann, H. Stutzer, and H. Pichlmaier.1994. Preoperative immunostimulation by Propionibacterium granulosum KP-45in colorectal cancer. Anticancer Res. 14: 1399–1404.

19. Kim, J., M. T. Ochoa, S. R. Krutzik, O. Takeuchi, S. Uematsu, A. J. Legaspi,H. D. Brightbill, D. Holland, W. J. Cunliffe, S. Akira, et al. 2002. Activation ofToll-like receptor 2 in acne triggers inflammatory cytokine responses. J. Immu-nol. 169: 1535–1541.

20. Kim, J. 2005. Review of the innate immune response in acne vulgaris: activationof Toll-like receptor 2 in acne triggers inflammatory cytokine responses. Der-matology 211: 193–198.

21. Desviat, L. R., B. Perez, C. Perez-Cerda, P. Rodriguez-Pombo, S. Clavero, andM. Ugarte. 2004. Propionic acidemia: mutation update and functional and struc-tural effects of the variant alleles. Mol. Genet. Metab. 83: 28–37.

22. Deodato, F., S. Boenzi, F. M. Santorelli, and C. Dionisi-Vici. 2006. Methylma-lonic and propionic aciduria. Am. J. Med. Genet. C. Semin. Med. Genet. 142C:104–112.

23. Al Essa, M., Z. Rahbeeni, S. Jumaah, S. Joshi, J. E. Al, M. S. Rashed, A. M. Al,and P. T. Ozand. 1998. Infectious complications of propionic acidemia in SaudiaArabia. Clin. Genet. 54: 90–94.

24. Kane, K. L., F. A. Ashton, J. L. Schmitz, and J. D. Folds. 1996. Determinationof natural killer cell function by flow cytometry. Clin. Diagn. Lab. Immunol. 3:295–300.

25. Andresen, L., H. Jensen, M. T. Pedersen, K. A. Hansen, and S. Skov. 2007.Molecular regulation of MHC class I chain-related protein A expression afterHDAC-inhibitor treatment of Jurkat T cells. J. Immunol. 179: 8235–8242.

26. Pedersen, S. F., B. Kramhoft, N. K. Jorgensen, and E. K. Hoffmann. 1996.Shrinkage-induced activation of the Na�/H� exchanger in Ehrlich ascites tumorcells: mechanisms involved in the activation and a role for the exchanger in cellvolume regulation. J. Membr. Biol. 149: 141–159.

27. Hansen, H. R., C. Hansen, K. P. Jensen, S. H. Hansen, S. Sturup, andB. Gammelgaard. 2008. Characterization of sodium stibogluconate by online liq-uid separation cell technology monitored by ICPMS and ESMS and computa-tional chemistry. Anal. Chem. 80: 5993–6000.

28. Yonezawa, T., Y. Kobayashi, and Y. Obara. 2007. Short-chain fatty acids induceacute phosphorylation of the p38 mitogen-activated protein kinase/heat shockprotein 27 pathway via GPR43 in the MCF-7 human breast cancer cell line. Cell.Signal. 19: 185–193.

29. Sandoval, A., F. Trivinos, A. Sanhueza, D. Carretta, M. A. Hidalgo, J. L. Hancke,and R. A. Burgos. 2007. Propionate induces pH(i) changes through calcium flux,ERK1/2, p38, and PKC in bovine neutrophils. Vet. Immunol. Immunopathol. 115:286–298.

30. Wang, W. Z., X. P. Chu, M. H. Li, J. Seeds, R. P. Simon, and Z. G. Xiong. 2006.Modulation of acid-sensing ion channel currents, acid-induced increase of intra-cellular Ca2�, and acidosis-mediated neuronal injury by intracellular pH. J. Biol.Chem. 281: 29369–29378.

31. Cousens, L. S., D. Gallwitz, and B. M. Alberts. 1979. Different accessibilities inchromatin to histone acetylase. J. Biol. Chem. 254: 1716–1723.

32. Maciver, N. J., S. R. Jacobs, H. L. Wieman, J. A. Wofford, J. L. Coloff, andJ. C. Rathmell. 2008. Glucose metabolism in lymphocytes is a regulated processwith significant effects on immune cell function and survival. J. Leukocyte Biol.84: 949–957.

33. Lopez-Lazaro, M. 2008. The Warburg effect: why and how do cancer cells ac-tivate glycolysis in the presence of oxygen? Anticancer Agents Med. Chem. 8:305–312.

34. Kurtoglu, M., J. C. Maher, and T. J. Lampidis. 2007. Differential toxic mecha-nisms of 2-deoxy-D-glucose versus 2-fluorodeoxy-D-glucose in hypoxic and nor-moxic tumor cells. Antioxid. Redox Signal. 9: 1383–1390.

35. Herken, H., and K. Lange. 1969. Blocking of pentose phosphate pathway in thebrain of rats by 6-aminonicotinamide. Naunyn Schmiedebergs Arch. Exp. Pathol.Pharmakol. 263: 496–499.

36. Gupte, S. A., R. J. Levine, R. S. Gupte, M. E. Young, V. Lionetti, V. Labinskyy,B. C. Floyd, C. Ojaimi, M. Bellomo, M. S. Wolin, and F. A. Recchia. 2006.Glucose-6-phosphate dehydrogenase-derived NADPH fuels superoxide produc-tion in the failing heart. J. Mol. Cell. Cardiol. 41: 340–349.

37. Perera, L., L. Shao, A. Patel, K. Evans, B. Meresse, R. Blumberg, D. Geraghty,V. Groh, T. Spies, B. Jabri, and L. Mayer. 2007. Expression of nonclassical classI molecules by intestinal epithelial cells. Inflamm. Bowel Dis. 13: 298–307.

38. Filipowicz, H. R., S. L. Ernst, C. L. Ashurst, M. Pasquali, and N. Longo. 2006.Metabolic changes associated with hyperammonemia in patients with propionicacidemia. Mol. Genet. Metab. 88: 123–130.

39. Nguyen, N. H., C. Morland, S. V. Gonzalez, F. Rise, J. Storm-Mathisen,V. Gundersen, and B. Hassel. 2007. Propionate increases neuronal histone acet-ylation, but is metabolized oxidatively by glia: relevance for propionic acidemia.J. Neurochem. 101: 806–814.

40. Hue, S., J. J. Mention, R. C. Monteiro, S. Zhang, C. Cellier, J. Schmitz,V. Verkarre, N. Fodil, S. Bahram, N. Cerf-Bensussan, and S. Caillat-Zucman.2004. A direct role for NKG2D/MICA interaction in villous atrophy during celiacdisease. Immunity 21: 367–377.

41. Meresse, B., Z. Chen, C. Ciszewski, M. Tretiakova, G. Bhagat, T. N. Krausz,D. H. Raulet, L. L. Lanier, V. Groh, T. Spies, et al. 2004. Coordinated inductionby IL15 of a TCR-independent NKG2D signaling pathway converts CTL intolymphokine-activated killer cells in celiac disease. Immunity 21: 357–366.

42. Tursi, A., G. Brandimarte, and G. Giorgetti. 2003. High prevalence of smallintestinal bacterial overgrowth in celiac patients with persistence of gastrointes-tinal symptoms after gluten withdrawal. Am. J. Gastroenterol. 98: 839–843.

906 PROPIONATE INDUCES MICA/B EXPRESSION

by guest on September 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from