Research Article

Activated human hepatic stellate cells induce myeloidderived suppressor cells from peripheral blood monocytes

in a CD44-dependent fashion

Bastian Höchst1,⇑,�, Frank A. Schildberg1,�, Pia Sauerborn1, Yvonne A. Gäbel1, Heidrun Gevensleben2,Diane Goltz2, Lukas C. Heukamp3, Andreas Türler4, Matthias Ballmaier5, Friederike Gieseke6,

Ingo Müller7, Jörg Kalff8, Christian Kurts1, Percy A. Knolle1, Linda Diehl1,⇑

1Institutes of Molecular Medicine and Experimental Immunology, University of Bonn, Germany; 2Institute for Pathology,Universitätsklinikum Bonn, Germany; 3Institute for Pathology, University of Cologne, Germany; 4Department of General and Abdominal Surgery,

Johanniter-Krankenhaus Bonn, Germany; 5Department of Pediatric Hematology and Oncology, Medizinische Hochschule Hannover,Germany; 6Research Institute, Children’s Cancer Center, Hamburg, Germany; 7Clinic for Pediatric Hematology and Oncology,

University Medical Center Hamburg-Eppendorf, Hamburg, Germany; 8Department of Surgery, University Hospital Bonn, Germany

Background & Aims: Myeloid derived suppressor cells (MDSCs)are a heterogeneous population of cells associated with the sup-

Conclusions: Our study shows that activated human HSCs con-vert mature peripheral blood monocytes into MDSCs. As HSCs

pression of immunity. However, little is known about how orwhere MDSCs are induced and from which cells they originate.The liver is known for its immune regulatory functions. Here,we investigated the capacity of human hepatic stellate cells(HSCs) to transform peripheral blood monocytes into MDSCs.Methods: We cultured freshly isolated human monocytes fromhealthy donors on primary human HSCs or an HSC cell-lineand characterized the phenotype and function of resultingCD14+HLA-DR�/low monocytes by flow cytometry, quantitativePCR, and functional assays. We analyzed the molecularmechanisms underlying the induction and function of theCD14+HLA-DR�/low cells by using blocking antibodies or knock-down technology.Results: Mature peripheral blood monocytes co-cultured withHSCs downregulated HLA-DR and developed a phenotypic andfunctional profile similar to MDSCs. Only activated but not freshlyisolated HSCs were capable of inducing CD14+HLA-DR�/low cells.Such CD14+HLA-DR�/low monocyte-derived MDSCs suppressedT-cell proliferation in an arginase-1 dependent fashion. HSC-induced development of CD14+HLA-DR�/low monocyte-derivedMDSCs was not mediated by soluble factors, but required physi-cal interaction and was abrogated by blocking CD44.

Journal of Hepatology 20

Keywords: Hepatic stellate cell; Myeloid derived suppressor cell; CD44; Arginase.Received 15 January 2013; received in revised form 2 April 2013; accepted 20 April2013; available online 9 May 2013⇑ Corresponding authors. Address: Sigmund-Freud-Str. 25, 53105 Bonn, Ger-many. Tel.: +49 228 287 11038; fax: +49 228 287 11052.E-mail addresses: B.Hoechst@gmail.com (B. Höchst), Linda.Diehl@ukb.uni-bonn.de(L. Diehl).

� These authors contributed equally to this work.Abbreviations: HSC, hepatic stellate cell; IFN, interferon; MDSC, myeloid derivedsuppressor cell; PGE2, prostaglandin E2; SCF, stem cell factor; siRNA, small inte-rfering RNA; shRNA, small hairpin RNA; VEGF, vascular endothelial growth factor.

are activated during chronic inflammation, the subsequent localinduction of MDSCs may prevent ensuing excessive liver injury.HSC-induced MDSCs functionally and phenotypically resemblethose isolated from liver cancer patients. Thus, our data suggestthat local generation of MDSCs by liver-resident HSCs may con-tribute to immune suppression during inflammation and cancerin the liver.� 2013 European Association for the Study of the Liver. Publishedby Elsevier B.V. All rights reserved.

Introduction

The immune system is precisely balanced regarding inflamma-tion and tolerance. This balance is upheld by both effector andregulatory/suppressive cells. Myeloid derived suppressor cells(MDSCs) can exert immunosuppressive functions [1] and com-prise a group of immature myeloid cells or progenitors of maturemacrophages, granulocytes or dendritic cells. MDSCs accumulatein the blood, bone-marrow, spleen, and tumor of patients andmice under pathological conditions and can establish a localimmunosuppressive environment. In humans, MDSCs are associ-ated with the suppression of T-cell responses [1], the induction ofregulatory T cells [2] as well as poor prognosis in cancer patients[3]. MDSCs generally express CD33, however, different types oftumors harbor distinct subsets of MDSCs. They can be dividedinto three phenotypic subpopulations: Lin-HLA-DR�/low [4],CD14+HLA-DR�/low [5], and CD15+HLA-DR�/low [6] MDSCs. HowMDSCs are induced and from which cells they arise is still a mat-ter of debate. Many soluble factors are reported to be involved inthe induction of MDSCs, such as CSF-1, and -2 [7,8]. However,they are also important for physiological development orfunction of myeloid cells.

13 vol. 59 j 528–535

JOURNAL OF HEPATOLOGY

Under steady state conditions, hepatic stellate cells (HSCs) are

vitamin A storing pericytes situated in the space of Dissé. Chronicinflammation leads to HSC activation, proliferation, and transdif-ferentiation into myofibroblasts resulting in fibrotic remodelingof the liver. In a murine model of islet-transplantation in vivo,transfer of HSCs increased MDSC numbers [9], however, the exactmechanism by which this occurs could not be identified. As thefrequency of HSCs increases during persistent inflammation andMDSCs are especially increased in liver diseases [10], we investi-gated whether human HSCs are involved in the induction ofMDSCs.

Here, we show that hepatic stellate cells can induceCD14+HLA-DR�/low myeloid derived suppressor cells from matureperipheral blood monocytes. Furthermore, we provide evidencethat this MDSC induction requires cell-cell contact and is medi-ated via CD44.

Materials and methods

Suppression assay

CD14+ monocytes were cultured with HSCs or control cell-lines for 3 days or wereleft untreated. CD14+HLA-DR�/low MDSCs were isolated ex vivo as describedbefore [2]. T cells were labeled with 0.1 lM carboxyfluorescein-succinimidyl-ester (CFSE) and were stimulated with T-cell activation and expansion kit(Miltenyi, Bergisch Gladbach, Germany). CD14+ cells were titrated as indicated.Proliferation of T cells was measured after 4 days based on the dilution of CFSE.Where indicated, L-N-monomethyl-arginine citrate (L-NMMA), N-hydroxyl-L-arginine (L-NOHA) (10 lM each) (all Sigma-Aldrich, Seelze, Germany) or anti-TGF-b blocking antibody (Clone 1D11, 40 lg/ml) (R&D Systems, Minneapolis,MN) were added.

Gene silencing using siRNA

LX2 cells were trypsinized and 106 cells were resuspended in 200 ll NucleofectorSolution (Lonza, Verviers, Belgium) containing 300 nmol/L siRNA. Cells weretransfected using Nucleofector II according to the manufacturers’ protocol usingthe protocol for human monocytes.

Statistical analysis

Statistical analysis was done using Student’s t test. Data are depicted, asmean ± SEM. p values 60.05 were considered significant. ⁄p 60.05, ⁄⁄p 60.01,⁄⁄⁄p 60.001.

Results

Primary human hepatic stellate cells downregulate HLA-DRexpression on peripheral blood monocytes

Given the complex regulatory functions of HSCs described in theliterature [9,11,12], we investigated whether human HSCs werecapable of generating MDSCs. It is known that hematopoieticstem cells or early myeloid progenitors have the capacity todevelop into MDSCs [13], but some MDSC subsets in humansexpress the monocyte marker CD14 [5]. Therefore, we investi-gated whether circulating CD14+ monocytes, which pass throughhepatic sinusoids within the blood stream, would develop intoMDSCs after contact with HSCs. Therefore, we cultured freshlyisolated peripheral blood CD14+HLA-DR+ monocytes on primaryhuman HSCs (pHSCs) or the HSC cell-line LX2. Co-culture withHSCs resulted in marked downregulation of HLA-DR expression

Journal of Hepatology 201

on monocytes (Fig. 1A). This effect was HSC-specific as we foundno such effect upon co-culture with hepatocyte cell-lines (HepG2,HuH-7), a liver sinusoidal endothelial cell-line (SK-Hep1), pri-mary B-cells or the colon carcinoma cell-line CCL253 (Fig. 1Aand B). Moreover, activated pHSCs were more potent inducersof CD14+HLA-DR�/low monocytes compared to freshly isolatedpHSCs (Fig. 1A) and 7 days after isolation, primary HSCs wereas efficient in HLA-DR downregulation as the HSC cell-line LX2(Fig. 1A), indicating that only activated HSCs could functionallymodulate circulating monocytes. The number of HLA-DR�/low

cells among CD14+ monocytes increased over time in co-culturewith HSCs (Fig. 1C). These CD14+ cells did not proliferate (Supple-mentary Fig. 1A) or undergo increased apoptosis (data notshown), indicating that these cells trans-differentiated frommature monocytes and did not arise from immature progenitorsthat may have been present in the culture. Furthermore, weexcluded that HSCs in general downregulate HLA-DR expressionin immune cells, as co-culture with LX2 did not downregulateHLA-DR on B-cells or myeloid dendritic cells (mDC) (Fig. 1D). Inaddition, we did not find HSCs to be able to induce other pheno-typical MDSC populations, such as Lin-CD11b�CD33+ cells, fromPBMC (Supplementary Fig. 1B). MHC-class-II downregulation byHSCs required cell-cell contact, because LX2 culture supernatantalone or culturing LX2 separately from monocytes in a transwell-system did not affect HLA-DR (Fig. 1E). Together, these data indi-cate that human HSCs induce CD14+HLA-DR�/low monocytes fromcirculating HLA-DRhi monocytes in a cell-cell-contact dependentfashion.

CD14+HLA-DR�/low monocytes induced by hepatic stellate cellsphenotypically resemble MDSCs

To further characterize HSC-induced CD14+HLA-DR�/low mono-cytes, we analyzed their surface marker expression profile andcompared it to monocytes cultured with other cell lines or freshlyisolated CD14+HLA-DR+ monocytes. CD14+HLA-DR�/low MDSCsisolated from healthy individuals served as control. There wereno differences with respect to CD15, CD40, and CD80 expressionlevels on freshly isolated monocytes and HSC-co-cultured mono-cytes, also ex vivo MDSCs did not express these molecules (Fig. 2).While co-culture with hepatocyte cell lines led to upregulation ofco-stimulatory molecules like CD40 and CD80 on peripheralblood monocytes, such regulation was not observed when mono-cytes were co-cultured with HSCs (Fig. 2). Upon contact withHSCs, however, monocytes showed increased levels of IL-4Ra(Fig. 2), a phenotype also observed in MDSCs isolated frompatients suffering from melanoma and colorectal carcinoma[14]. Interestingly, monocytes cultured on HSCs downregulatedCD11b (Fig. 2). Thus, phenotypic characterization of surface mol-ecules did not allow us to identify a profile typical for MDSCs inmonocytes co-cultured with HSCs.

This led us to perform quantitative PCR in order to determinethe expression levels of genes specifically associated with MDSCs[15]. S100A12, that is overexpressed in MDSCs from hepatocellu-lar and colorectal carcinoma patients [16], was selectivelyexpressed in CD14+HLA-DR�/low monocytes cultured on LX2and CD14+HLA-DR�/low MDSCs ex vivo (Supplementary Fig. 2).Also arginase-I, which is involved in the MDSC-mediated sup-pression of T-cell function, [5] was expressed selectively inHSC-induced CD14+HLA-DR�/low monocytes and ex vivoCD14+HLA-DR�/low MDSCs. In contrast, the expression of iNOS,

3 vol. 59 j 528–535 529

80

80 100

60

60

40

40

20

200

0

5

10

0Time (h)

80 1000 02 04 06Time (h)

0 02 04 06 08Time (h)

0 01 02 03 04

w/o

LX2

LX2Transwell

Supernatant

MDSC (%)(HLA-DR-/CD14+)

**

ex vivo w/o pHSC(ex vivo)

pHSC(d3)

pHSC(d7)

LX2

CD14

HLA

-DR

Collagen B cells CCL253 HuH7 HepG2 SK-Hep1

CD

14+ H

LA-D

R-/l

ow/C

D14

+ (%

)

CD

14+ H

LA-D

R-/l

ow/C

D14

+ (10

4 )(%

)

HLA

-DR

(Geo

Mea

n)B cells w/oB cells + LX2mDCs w/omDCs + LX2 CD14+ w/oCD14+ + LX2

A B

C D E

*****

*

*

LX2pHSCw/oCollagenB cellsHepG2HuH7SKHep-1CCL253

LX2pHSCw/oCollagenB cellsHepG2HuH7SKHep-1CCL253

***

Fig. 1. Hepatic stellate cells induce an MDSC-like phenotype in peripheral blood monocytes. (A–C) Monocytes were cultured alone, on primary human HSCs, the HSCcell-lines LX2, HepG2, HuH7, SK-Hep-1 or CCL253 for 3 days, and the expression of HLA-DR was measured by flow cytometry. Representative contour plots (A), %CD14+HLA-DR+ expression from 5 independent experiments (B) or absolute number (C) are shown. (D) Isolated monocytes, B-cells or myeloid dendritic cells were added toa confluent layer of LX2 in a 24-well plate. The expression of HLA-DR was analyzed on CD14+, CD19+ or CD1c+ cells by flow cytometry. Cumulative data from 3 independentexperiments are shown. (E) Monocytes were cultured alone, on LX2, with LX2 conditioned media or with LX2 cells separated by a transwell-insert for 3 days. The frequencyof CD14+HLA-DR�/low cells is plotted cumulative for 3 independent experiments. ⁄p 60.05, ⁄⁄p 60.01, ⁄⁄⁄p 60.001.

CD11b CD15 CD33 CD40 CD80 CD86 IL-4Rα MHC Class I

SK-Hep1

HepG2

HuH7

pHSC

LX2

w/o

CD14+ HLA-DR-/low (ex vivo)

CD14+ HLA-DR+ (ex vivo)

CTRL-IgG

CD

14+ cells cultured

Fig. 2. Monocytes cultured on hepatic stellate cells upregulate IL-4Ra. CD14+ cells were cultured as described before, alone, co-cultured with different cell lines for3 days or analyzed directly ex vivo. The expression of surface markers was analyzed by flow cytometry gating on CD14+, or CD14+HLA-DR+ or CD14+HLA-DR�/low cells. Thecorresponding MFI is shown. Representative histograms from 5 experiments are shown.

Research Article

which has also been described to mediate T-cell suppression [17],was not restricted to HSC-cultured monocytes or ex vivo MDSCs

530 Journal of Hepatology 201

(Supplementary Fig. 2). Taken together, CD14+ monocytes cul-tured on HSCs downregulate HLA-DR, express IL-4Ra and

3 vol. 59 j 528–535

JOURNAL OF HEPATOLOGY

S100A12, and upregulate Arg-I mRNA, which indicates that thesecells also may exert suppressive function.

HSC-induced CD14+HLA-DR�/low monocytes suppress T-cellproliferation in an arginase-I-dependent manner

We next investigated whether HSC-induced CD14+HLA-DR�/low

monocytes could suppress T-cell proliferation. To this end,human CD8+ T cells were stimulated using anti-CD2/CD3/CD28-coated beads, and CD14+HLA-DR�/low monocytes previously cul-tured with LX2 were added. Addition of increasing numbers ofHSC-induced CD14+HLA-DR�/low monocytes to a fixed numberof CD8 T cells progressively reduced T-cell proliferation(Fig. 3A). This effect was only observed with CD14+ cells culturedon LX2, pHSCs and freshly isolated MDSCs (Fig. 3B), but not withthose monocytes cultured together with LSECs (SK-Hep1) or thehepatocyte (HepG2, HuH-7) cell lines. The same inhibitory effecton proliferation was also observed for CD4 T cells (data notshown). In addition to T-cell proliferation, IFN-c productionwas also inhibited by LX2-induced CD14+HLA-DR�/low monocytesand correlated inversely with the addition of increasing numbersof LX2-induced CD14+HLA-DR�/low monocytes into the culture(Fig. 3C).

MDSCs can mediate the suppressive function via severalmechanisms [18]. By adding specific inhibitors or blockingantibodies into the co-culture of T cells with HSC-inducedCD14+HLA-DR�/low monocytes, we found that only L-NOHA, aspecific arginase-I inhibitor, prevented the suppression of T-cellproliferation. Neither the pan-inhibitor of NO-synthase (L-NMMA) nor blocking antibodies against TGF-b was able to block

1500

1000

500

0

IFN

-γ (p

g/m

l)

CD14+

ex vivoCD14+

cultured alone

CD14+

HLA-DR-/low

ex vivo

CD14+

cultured on LX2

- +

*** ***

Ratio CD8+ T cells: CD14+ 1:0.11:0.31:11:3

1:3 1:1 1:0.3Negative

Ratio CD8+ T cells: CD14+ (LX2)A

C D

Fig. 3. Hepatic stellate cell-modified monocytes suppress T cells in an arginase-I-deCSFE labeled CD8 T cells were stimulated using CD2/CD3/CD28-coated beads and CD14+

four days. The percentage proliferation compared to maximal proliferation (T cells culturCD14+HLA-DR�/low cells 1:0.1/0.3/1/3) (C). Blocking antibodies or inhibitors were addedhistograms (A) or cumulative results from 6 (B and C) or 3 (D) independent experiment

Journal of Hepatology 201

MDSCs mediated suppression of T cells (Fig. 3D). Moreover,neither the addition of blocking antibodies against IL-10 or IL-6R, nor co-culture with an IDO inhibitor (1-MT) or the ROS inhib-itors (YCG063 or MnTBAP) prevented suppression of T-cell prolif-eration (Supplementary Fig. 3). Together, both phenotypicallyand functionally HSC-induced CD14+HLA-DR�/low monocytes rep-resent a myeloid derived suppressor cell population that exertthe suppressive function through arginase-I.

The induction of CD14+HLA-DR�/low monocyte-derived MDSCs byHSCs depends on the expression of CD44

CD44 expression has been shown to be increased on hepatic stel-late cells from injured rat livers and is associated with theirincreased motility [19]. Moreover, in toxic liver injury, CD44was shown to prevent exacerbation of inflammation via controlof macrophage function infiltrating the liver [20]. In humans,hepatic CD44 expression is increased in patients with liver dis-ease [21] and is involved in progression of various cancer typesincluding hepatocellular carcinoma [22,23], in which MDSCs areincreased and involved in immune-suppression [5,16]. To test ifCD44 is involved in the induction of MDSCs, we added increasingconcentrations of anti-CD44 blocking antibodies into the co-cul-ture of LX2 and monocytes. Because we have previously shown,that CD54 on HSCs is important for the veto-function towardsCD8 T cells [12], we also included blocking antibodies for CD54.In contrast to HSC-mediated CD8 T-cell inhibition, CD54 wasnot involved in induction of HLA-DR�/low monocytes by HSCs,but blocking CD44 reduced the induction of MDSCs from mono-cytes dose dependently (Fig. 4A). Moreover, blocking CD44 inter-

α-TGFβL-NMMAL-NOHA

w/o

+ ++ ++ ++ +- -- +

n.s.

n.s.

n.s.

***

**

αCD

3/C

D28

CD

14+ (

LX2) 0 20 40 60 80 100

Proliferation (%)

Positive

B125

0.8 1.6 3.2

100

0.4

75

0.2

50

25

0.10

0Ratio T cells: CD14

Prol

ifera

tion

(%)

**

*

w/oEx vivoSK-Hep1HepG2LX2pHSCpMDSC

pendent mechanism. Monocytes were cultured for 3 days as indicated. Isolatedcells were added at different ratios. Proliferation was analyzed (A, B, and D) after

ed alone) is depicted (B). IFN-c production was quantified by ELISA (ratio T cells toduring the co-culture of monocytes with T cells as indicated (D). Representatives are shown. ⁄p 60.05, ⁄⁄p 60.01, ⁄⁄⁄p 60.001, n.s., not significant.

3 vol. 59 j 528–535 531

* *

***CD

14+ H

LA-D

R-/l

ow/C

D14

+

0

20

40

60

80

5 µg/ml10 µg/ml20 µg/ml40 µg/ml

LX2 + - + + + + + + + + + +αCD54 αCD44IgG2b IgG1

**

Prol

ifera

ted

cells

(%)

0

20

40

60

CD14+ (LX2)αCD3/CD28

CD44 Autofluorescence

--

-+

++

++

++

++

++

w/o

αCD

44

IgG

2b

αCD

54

IgG

1

CD44

wtCTRL-siRNAsiRNA IsiRNA II

wt

CTR

L I IIsiRNA

CD14

HLA

-DR

wtCTRL-siRNAsiRNA I siRNA II

0123456

Day

LX2w/o pHSC (d7)

w/o

+ αC

D44

HLA

-DR

CD14

CD

14+ H

L-D

R-/l

ow/

CD

14+ (%

)

80

60

40

20

0

- αCD44+ αCD44

*** ***

2

1

0CD

44 (M

FI)

x100

pHSCLX2

*** **

CBA

F

G

ED

Fig. 4. Induction of CD14+HLA-DR�/low cells by hepatic stellate cells is mediated by CD44. CD14+ cells were cultured on LX2 as described before. Blocking antibodyagainst CD54, CD44 or appropriate isotype controls were added as indicated (A, B, F, and G). The frequency of CD14+HLA-DR�/low cells is plotted from 4 independentexperiments (A) or 2 experiments pHSCs (F and G). CD14+ cells were isolated after culture with LX2 in the absence or presence of blocking CD44 antibody and were co-cultured with CD8 T cells. Proliferation of T cells was measured on day 4. Cumulative results of 4 independent experiments are shown (B). CD44 was silenced in LX2 usingsiRNA as described. As control, scrambled siRNA or untreated cells were used as controls. After 24 h, expression of CD44 on LX2 cells was analyzed by flow cytometry.Histograms represent mean fluorescence intensity of CD44 (C). Monocytes were co-cultured with different siRNA-transfected LX2 cells and the expression of HLA-DR wasanalyzed by flow cytometry. Dot plots show percentages of HLA-DR�/low cells of 3 independent experiments (D). Primary isolated human hepatic stellate cells were isolatedand cultured for 6 days. Cells were analyzed every day for the mean fluorescence intensity of CD44 and GeoMean of autofluorescence by flow cytometry. Representativeresults from 2 independent experiments are shown (E). ⁄p 60.05, ⁄⁄p 60.01, ⁄⁄⁄p 60.001.

Research Article

actions between monocytes and HSCs prevented acquisition ofthe suppressive capacity by those monocytes (Fig. 4B). To inves-tigate whether HSC-expressed CD44 was important for inductionof MDSCs, we efficiently knocked down the expression of CD44on LX2 cells using two different siRNAs (Fig. 4C). In absence ofCD44 expression, LX2 cells were not able to downregulate HLA-DR on monocytes anymore (Fig. 4D), indicating that HSC-depen-dent MDSCs induction is CD44-mediated. As we found primaryhuman HSCs to become more efficient in inducing MDSCs frommonocytes after activation by culture on plastic, we investigatedexpression kinetics of CD44 on primary HSCs. As HSCs becomeactivated, they lose vitamin A containing lipid droplets andthereby autofluorescence (Fig. 4E). Primary HSCs did not expressCD44, but markedly upregulated CD44 during activation in vitro(Fig. 4E), achieving levels comparable to LX2. More importantly,the induction of HLA-DR�/low monocytes by primary HSCs

532 Journal of Hepatology 201

depended on CD44 expression, as it was abrogated in the pres-ence of anti-CD44 blocking antibodies (Fig. 4F and G). Together,these data suggest that activated hepatic stellate cells can effi-ciently induce MDSCs in a CD44-dependent manner from maturemonocytes.

Discussion

The liver has an extraordinary immunological function deter-mined by the unique micro-environment and its organ-residentcells that function as antigen-presenting cells to regulate anti-gen-specific immune responses [24]. Antigen-specific immunitycan eradicate infections with hepatotropic pathogens from theliver [25]. Nevertheless, persistence of pathogens in hepatocytesor development of hepatocellular carcinoma in the context of

3 vol. 59 j 528–535

JOURNAL OF HEPATOLOGY

chronic inflammation demonstrates that the tolerogenic functionof hepatic cell populations may impede efficient immune surveil-lance under certain circumstances. Here, we report that activatedHSCs, which arise as a consequence of hepatic inflammation,induce myeloid derived suppressor cells (MDSCs) in a cell-con-tact and CD44-dependent manner from peripheral bloodmonocytes.

HSCs can exert regulatory immune function via inhibition ofDC-mediated CD8 T-cell activation [12,26], induction of T regula-tory cells [11,26] or PDL1-mediated induced apoptosis of T cells[27]. Furthermore, adoptive transfer of murine HSCs leads tothe induction of MDSCs [9], which are prominent regulatory cellsknown to locally regulate adaptive immunity [28]. Data fromin vitro studies showed that MDSCs can be generated frombone-marrow cells in the presence of various cytokine cocktails[29]. Therefore, it is thought that MDSCs in vivo originate fromprecursors in the bone-marrow and represent a population ofimmature myeloid cells that are recruited to sites of inflamma-tion. Our data provide evidence that CD14+HLA-DR�/low MDSCscan also be generated from circulating CD14+ monocytes in aperipheral organ such as the liver. Interestingly, MDSCs inducedby HSCs are similar to those found in patients suffering fromhepatocellular carcinoma [5] or HCV infection [30], indicatingthat also in humans HSCs may induce MDSCs in vivo. Thus, thelocal generation of MDSCs by HSCs may provide a novel mecha-nism as to how systemic adaptive immunity is locally downreg-ulated in the liver.

We discovered that activated, but not freshly isolated HSCs,had the capacity to generate CD14+HLA-DR�/low monocyte-derived MDSCs. During chronic inflammatory conditions, HSCsare activated in vivo and trans-differentiate into pro-fibrogenicmyofibroblasts [31]. We, therefore, investigated the role ofmolecules known to be upregulated upon activation of HSCs.A number of soluble factors, like GM-CSF, M-CSF, and VEGF,have been implicated to play a role in the induction of MDSCs[32]. Recently, murine HSC were found to mediate MDSCinduction via the complement factor C3 upon adoptive transfer[33]. However, we found that soluble factors do not play a rolein the induction of CD14+HLA-DR�/low MDSCs from monocytesby HSCs. This could be due to species differences or the originof the MDSC, as murine HSC were used to generate MDSC frombone-marrow precursors and not mature monocytes [29]. Sub-sequently, we focused on cell surface molecules expressed byHSCs. CD54 expression on HSCs is critically important for directinhibitory effects on naïve T-cell activation and proliferation[12]. In the generation of CD14+HLA-DR�/low monocyte-derivedMDSCs by HSCs, CD54 was not involved. HSC activation alsoleads to increased expression levels of CD44 [34,35]. We foundevidence for the involvement of CD44 in generation ofCD14+HLA-DR�/low monocyte-derived MDSCs by using blockingantibodies to CD44 or siRNA-mediated CD44 knockdown inHSCs. It remains unclear how CD44 induced generation ofCD14+HLA-DR�/low monocyte-derived MDSCs from circulatingmonocytes because CD44 is not selectively expressed by acti-vated HSCs. CD44 is involved in cell recruitment to tissues suchas liver or bone marrow [36,37] and locomotion of T cellswithin tissues or tumors [38]. CD44 has several interactionpartners including extracellular matrix proteins such as hyalu-ronic acid, mediators such as osteopontin, and the VLA-4 mol-ecule to establish cell-cell interaction [39]. Although VLA-4 hasmainly been found to mediate cell adhesion, a signaling func-

Journal of Hepatology 201

tion for this molecule has been reported for T cells [40], indi-cating that potentially VLA-4 could be involved in signalingprocesses in CD14+ monocytes leading to MDSC differentiation.Moreover, specificity of CD44 in the induction of MDSCs couldalso be mediated via its differential splicing that can lead toaltered functionality [41]. Indeed, rat hepatic stellate cellsexpress a distinct set of CD44 splice variants necessary formigration during liver injury [19]. Thus, it is possible that aparticular splice variant of CD44 upregulated on HSCs isinvolved in induction of MDSCs.

Our findings raise the question whether monocytes or mono-cyte-derived cells present in the liver all function as MDSCs. Inthe hepatic sinusoid, fenestrated LSECs without a basementmembrane, do not provide a physical barrier and allow directphysical contact between HSCs in the space of Dissé and Kupffercells or circulating monocytes [42]. However, only activated HSCshave the capacity to generate monocyte-derived MDSCs. There-fore, it is unlikely that, under steady state non-inflammatory con-ditions, MDSCs are generated from monocytes or Kupffer cells.Consistently, Kupffer cells isolated from the normal liver functionas antigen-presenting cells, but not as suppressive MDSCs [43].However, under inflammatory conditions, Kupffer cells may alsobe modulated, as we found LX2 cells to be able to downregulateHLA-DR on LPS-stimulated Kupffer cells (data not shown). Theinduction of MDSCs by activated HSCs in the chronically inflamedliver may thus represent a negative feedback loop in the attemptto dampen or prevent immunopathology in the liver. As the acti-vation of HSCs is accompanied by increased expression of chemo-kines and adhesion molecules such as CD54 and CD44 that canlead to attraction of circulating monocytes [35], chronic inflam-mation and the consequent HSC activation may thus trigger bothincreased recruitment of monocytes and differentiation intoMDSCs. It is unclear whether activated HSCs also bear the capac-ity to modulate Kupffer-cell function and cause their differentia-tion into MDSCs.

The induction of CD14+HLA-DR�/low monocyte-derived MDSCsby HSCs did not result from proliferation of precursor cells, butrepresents a differentiation process that was so far unknown tobe involved in MDSC generation [44]. It is important to note thatHSC-induced CD14+HLA-DR�/low monocyte-derived MDSCs didnot maintain their surface and functional phenotype for morethan 2 days after losing HSC contact (data not shown), indicatingthat the increase in numbers of MDSCs in the liver will only betransient and strictly linked to persistent hepatic inflammation.It also suggests that MDSCs generated from monocytes requirecontinuous contact with activated HSCs to maintain their func-tion as suppressor cells. The dynamic induction of MDSCs frommonocytes by activated mesenchymal cells, such as alpha-smooth muscle antigen (aSMA) positive hepatic HSCs, indicatesthat aSMA+ tumor stroma cells, whose appearance in tumor tis-sue is often directly correlated with patient survival [45], mayalso locally induce MDSCs thereby creating a regulatory tumormicro-environment that impairs immune clearance of cancercells. It also provides an interesting link to the observation thattumors mostly arise in organs during chronic inflammatory con-ditions [46].

Taken together, these results demonstrate that activatedhuman HSCs can induce differentiation of monocytes intoMDSCs, which provides evidence for a so far unrecognizedmechanism in the generation of MDSCs via a re-differentiationstep of already differentiated monocytes. The HSC-mediated

3 vol. 59 j 528–535 533

Research Article

induction of CD14+HLA-DR�/low monocyte-derived MDSCs alsoattributes further immune competence to peripheral organslike the liver because attenuation of local inflammation maynot only occur as a consequence of increased chemokine-med-iated recruitment of circulating MDSCs, but also rely on a localand dynamic differentiation process initiated by organ-residentHSCs.

Financial support

This work was supported by a BONFOR grant of the University ofBonn medical faculty to BH.

Conflict of interest

The authors who have taken part in this study do not have a rela-tionship with the manufacturers of the drugs involved either inthe past or present and did not receive funding from the manu-facturers to carry out their research. BH was supported by a grantfrom the University of Bonn Medical Faculty (BONFOR). LD wassupported by a grant from the german research council (SFB704).

Authors’ contributions

LD and BH were responsible for the study concept, interpretationof data and drafting of the manuscript. BH, FAS, LCH, YAG, and PSperformed experiments and BH and FAS analyzed the data. MB,HG, DG, LCH, FG, AT, JCK and IM provided the material. LD, CK,and PK provided important intellectual content as well as criticalrevision of the manuscript.

Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jhep.2013.04.033.

References

[1] Bronte V, Apolloni E, Cabrelle A, Ronca R, Serafini P, Zamboni P, et al.Identification of a CD11b(+)/Gr-1(+)/CD31(+) myeloid progenitor capable ofactivating or suppressing CD8(+) T cells. Blood 2000;96:3838–3846.

[2] Hoechst B, Gamrekelashvili J, Manns MP, Greten TF, Korangy F. Plasticity ofhuman Th17 cells and iTregs is orchestrated by different subsets of myeloidcells. Blood 2011;117:6532–6541.

[3] Allavena P, Mantovani A. Immunology in the clinic review series; focus oncancer: tumour-associated macrophages: undisputed stars of the inflam-matory tumour microenvironment. Clin Exp Immunol 2012;167:195–205.

[4] Almand B, Clark JI, Nikitina E, van Beynen J, English NR, Knight SC, et al.Increased production of immature myeloid cells in cancer patients: amechanism of immunosuppression in cancer. J Immunol 2001;166:678–689.

[5] Hoechst B, Ormandy LA, Ballmaier M, Lehner F, Kruger C, Manns MP, et al. Anew population of myeloid-derived suppressor cells in hepatocellularcarcinoma patients induces CD4(+)CD25(+)Foxp3(+) T cells. Gastroenterol-ogy 2008;135:234–243.

[6] Zea AH, Rodriguez PC, Atkins MB, Hernandez C, Signoretti S, Zabaleta J, et al.Arginase-producing myeloid suppressor cells in renal cell carcinomapatients: a mechanism of tumor evasion. Cancer Res 2005;65:3044–3048.

[7] Dolcetti L, Peranzoni E, Ugel S, Marigo I, Fernandez Gomez A, Mesa C, et al.Hierarchy of immunosuppressive strength among myeloid-derived suppres-sor cell subsets is determined by GM-CSF. Eur J Immunol 2010;40:22–35.

534 Journal of Hepatology 201

[8] Kwan WH, Boix C, Gougelet N, Fridman WH, Mueller CG. LPS induces rapidIL-10 release by M-CSF-conditioned tolerogenic dendritic cell precursors. JLeukoc Biol 2007;82:133–141.

[9] Chou HS, Hsieh CC, Yang HR, Wang L, Arakawa Y, Brown K, et al. Hepaticstellate cells regulate immune response by way of induction of myeloidsuppressor cells in mice. Hepatology 2011;53:1007–1019.

[10] Ilkovitch D, Lopez DM. The liver is a site for tumor-induced myeloid-derivedsuppressor cell accumulation and immunosuppression. Cancer Res2009;69:5514–5521.

[11] Dangi A, Sumpter TL, Kimura S, Stolz DB, Murase N, Raimondi G, et al.Selective expansion of allogeneic regulatory T cells by hepatic stellate cells:role of endotoxin and implications for allograft tolerance. J Immunol2012;188:3667–3677.

[12] Schildberg FA, Wojtalla A, Siegmund SV, Endl E, Diehl L, Abdullah Z, et al.Murine hepatic stellate cells veto CD8 T cell activation by a CD54-dependentmechanism. Hepatology 2011;54:262–272.

[13] Marigo I, Bosio E, Solito S, Mesa C, Fernandez A, Dolcetti L, et al. Tumor-induced tolerance and immune suppression depend on the C/EBPbetatranscription factor. Immunity 2010;32:790–802.

[14] Mandruzzato S, Solito S, Falisi E, Francescato S, Chiarion-Sileni V, Mocellin S,et al. IL4Ralpha+ myeloid-derived suppressor cell expansion in cancerpatients. J Immunol 2009;182:6562–6568.

[15] Corzo CA, Condamine T, Lu L, Cotter MJ, Youn JI, Cheng P, et al. HIF-1{alpha}regulates function and differentiation of myeloid-derived suppressor cells inthe tumor microenvironment. J Exp Med. 2010;207:2439–2453.

[16] Zhao F, Hoechst B, Duffy A, Gamrekelashvili J, Fioravanti S, Manns MP, et al.S100A9 a new marker for monocytic human myeloid derived suppressorcells. Immunology 2012.

[17] Movahedi K, Guilliams M, Van den Bossche J, Van den Bergh R, Gysemans C,Beschin A, et al. Identification of discrete tumor-induced myeloid-derivedsuppressor cell subpopulations with distinct T cell-suppressive activity.Blood 2008;111:4233–4244.

[18] Marigo I, Dolcetti L, Serafini P, Zanovello P, Bronte V. Tumor-inducedtolerance and immune suppression by myeloid derived suppressor cells.Immunol Rev 2008;222:162–179.

[19] Kikuchi S, Griffin CT, Wang SS, Bissell DM. Role of CD44 in epithelial woundrepair: migration of rat hepatic stellate cells utilizes hyaluronic acid andCD44v6. J Biol Chem 2005;280:15398–15404.

[20] Kimura K, Nagaki M, Kakimi K, Saio M, Saeki T, Okuda Y, et al. Criticalrole of CD44 in hepatotoxin-mediated liver injury. J Hepatol2008;48:952–961.

[21] Urashima S, Tsutsumi M, Ozaki K, Tsuchishima M, Shimanaka K, Ueshima Y,et al. Immunohistochemical study of hyaluronate receptor (CD44) inalcoholic liver disease. Alcohol Clin Exp Res 2000;24:34S–38S.

[22] Endo K, Terada T. Protein expression of CD44 (standard and variantisoforms) in hepatocellular carcinoma: relationships with tumor grade,clinicopathologic parameters, p53 expression, and patient survival. J Hepatol2000;32:78–84.

[23] Parker D. Colorectal cancer prognosis and expression of exon-v6-containingCD44 proteins. Lancet 1995;345:583–584.

[24] Thomson AW, Knolle PA. Antigen-presenting cell function in the tolerogenicliver environment. Nat Rev Immunol 2010;10:753–766.

[25] Protzer U, Maini MK, Knolle PA. Living in the liver: hepatic infections. NatRev Immunol 2012;12:201–213.

[26] Ichikawa S, Mucida D, Tyznik AJ, Kronenberg M, Cheroutre H. Hepaticstellate cells function as regulatory bystanders. J Immunol2011;186:5549–5555.

[27] Yu MC, Chen CH, Liang X, Wang L, Gandhi CR, Fung JJ, et al. Inhibition of T-cell responses by hepatic stellate cells via B7–H1-mediated T-cell apoptosisin mice. Hepatology 2004;40:1312–1321.

[28] Bronte V. Myeloid-derived suppressor cells in inflammation: uncovering cellsubsets with enhanced immunosuppressive functions. Eur J Immunol2009;39:2670–2672.

[29] Highfill SL, Rodriguez PC, Zhou Q, Goetz CA, Koehn BH, Veenstra R,et al. Bone marrow myeloid-derived suppressor cells (MDSCs) inhibitgraft-versus-host disease (GVHD) via an arginase-1-dependent mecha-nism that is up-regulated by interleukin-13. Blood 2010;116:5738–5747.

[30] Tacke RS, Lee HC, Goh C, Courtney J, Polyak SJ, Rosen HR, et al. Myeloidsuppressor cells induced by hepatitis C virus suppress T-cell responsesthrough the production of reactive oxygen species. Hepatology2012;55:343–353.

[31] Novo E, di Bonzo LV, Cannito S, Colombatto S, Parola M. Hepatic myofibro-blasts: a heterogeneous population of multifunctional cells in liver fibro-genesis. Int J Biochem Cell Biol 2009;41:2089–2093.

3 vol. 59 j 528–535

JOURNAL OF HEPATOLOGY

[32] Pylayeva-Gupta Y, Lee KE, Hajdu CH, Miller G, Bar-Sagi D. Oncogenic Kras-

induced GM-CSF production promotes the development of pancreaticneoplasia. Cancer Cell 2012;21:836–847.

[33] Hsieh CC, Chou HS, Yang HR, Lin F, Bhatt S, Qin J, et al. The role ofcomplement component 3 (C3) in differentiation of myeloid-derivedsuppressor cells. Blood 2013;121:1760–1768.

[34] Safadi R, Friedman SL. Hepatic fibrosis-role of hepatic stellate cell activation.MedGenMed 2002;4:27.

[35] Friedman SL. Hepatic stellate cells: protean, multifunctional, and enigmaticcells of the liver. Physiol Rev 2008;88:125–172.

[36] Lapidot T, Dar A, Kollet O. How do stem cells find their way home? Blood2005;106:1901–1910.

[37] Ohata S, Nawa M, Kasama T, Yamasaki T, Sawanobori K, Hata S,et al. Hematopoiesis-dependent expression of CD44 in murinehepatic progenitor cells. Biochem Biophys Res Commun 2009;379:817–823.

[38] Ng LG, Mrass P, Kinjyo I, Reiner SL, Weninger W. Two-photon imaging ofeffector T-cell behavior: lessons from a tumor model. Immunol Rev2008;221:147–162.

[39] Hertweck M, Erdfelder F, Kreuzer K-A. CD44 in hematological neoplasias.Ann Hematol 2011;90:493–508.

Journal of Hepatology 201

[40] Nojima Y, Rothstein DM, Sugita K, Schlossman SF, Morimoto C. Ligation ofVLA-4 on T cells stimulates tyrosine phosphorylation of a 105-kD protein. JExp Med 1992;175:1045–1053.

[41] Brown R, Reinke L, Damerow M, Perez D, Chodosh L, Yang J, et al. CD44 spliceisoform switching in human and mouse epithelium is essential for epithe-lial-mesenchymal transition and breast cancer progression. J Clin Invest2011;121:1064–1074.

[42] Braet F, Riches J, Geerts W, Jahn KA, Wisse E, Frederik P. Three-dimensionalorganization of fenestrae labyrinths in liver sinusoidal endothelial cells.Liver Int 2009;29:603–613.

[43] Wiegard C, Frenzel C, Herkel J, Kallen KJ, Schmitt E, Lohse AW. Murine liverantigen presenting cells control suppressor activity of CD4+CD25+ regula-tory T cells. Hepatology 2005;42:193–199.

[44] Greten TF, Manns MP, Korangy F. Myeloid derived suppressor cells in humandiseases. Int Immunopharmacol 2011;11:802–807.

[45] Chuaysri C, Thuwajit P, Paupairoj A, Chau-In S, Suthiphongchai T, Thuwajit C.Alpha-smooth muscle actin-positive fibroblasts promote biliary cell prolif-eration and correlate with poor survival in cholangiocarcinoma. Oncol Rep2009;21:957–969.

[46] Nakagawa H, Maeda S. Inflammation- and stress-related signaling pathwaysin hepatocarcinogenesis. World J Gastroenterol 2012;18:4071–4081.

3 vol. 59 j 528–535 535