CD39 expression by hepatic myeloid dendritic cells attenuates inflammation in liver transplant ischemia-reperfusion injury in mice


  • Osamu Yoshida,

    1. Department of Surgery, Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Shoko Kimura,

    1. Department of Surgery, Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Edwin K. Jackson,

    1. Department of Pharmacology and Chemical Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Simon C. Robson,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • David A. Geller,

    1. Department of Surgery, Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Noriko Murase,

    1. Department of Surgery, Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Angus W. Thomson

    Corresponding author
    1. Department of Surgery, Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA
    2. Department of Immunology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    • Address reprint requests to: Angus W. Thomson, Ph.D., D.Sc., Starzl Transplantation Institute, University of Pittsburgh School of Medicine, 200 Lothrop Street, BST W1540, Pittsburgh, PA 15261. E-mail:; fax: 412-624-1172.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • This work was supported by National Institutes of Health (NIH) grants P01AI81678 (to A.W.T.), R01 HL094400 (to S.C.R.), and P30DK079307 (to E.K.J.) and by grant no. 87429717 from the Roche Organ Transplantation Research Foundation (to A.W.T.). O.Y. is the recipient of an American Society of Transplantation Basic Science Fellowship and an NIH postdoctoral fellowship (T32 AI74490).


Hepatic innate immune cells, in particular, interstitial dendritic cells (DCs), regulate inflammatory responses and may promote inherent liver tolerogenicity. After tissue injury, adenosine triphosphate (ATP) is released and acts as a damage-associated molecular pattern that activates innate immune cells by pattern recognition receptors. CD39 (ectonucleoside triphosphate diphosphohydrolase-1) rapidly hydrolyzes extracellular ATP to maintain physiological levels. We hypothesized that CD39 expression on liver DCs might contribute to regulation of their innate immune functions. Mouse liver conventional myeloid DCs (mDCs) were hyporesponsive to ATP, compared with their splenic counterparts. This disparity was ascribed to more efficient hydrolysis of ATP by higher expression of CD39 on liver mDCs. Human liver mDCs expressed greater levels of CD39 than those from peripheral blood. The comparatively high expression of CD39 on liver mDCs correlated strongly with both ATP hydrolysis and adenosine production. Notably, CD39−/− mouse liver mDCs exhibited a more mature phenotype, greater responsiveness to Toll-like receptor 4 ligation, and stronger proinflammatory and immunostimulatory activity than wild-type (WT) liver mDCs. To investigate the role of CD39 on liver mDCs in vivo, we performed orthotopic liver transplantation with extended cold preservation using CD39−/− or WT donor mouse livers. Compared to WT liver grafts, CD39−/− grafts exhibited enhanced interstitial DC activation, elevated proinflammatory cytokine levels, and more-severe tissue injury. Moreover, portal venous delivery of WT, but not CD39−/− liver mDCs, to donor livers immediately post-transplant exerted a protective effect against graft injury in CD39−/− to CD39−/− liver transplantation. Conclusions: These data reveal that CD39 expression on conventional liver mDCs limits their proinflammatory activity and confers protective properties on these important innate immune cells against liver transplant ischemia/reperfusion injury. (Hepatology 2013; 58:2163–2175)






alanine aminotransferase


antigen-presenting cell(s)


aspartate aminotransferase


adenosine triphosphate


bone marrow


counts per second


danger-associated molecular pattern


dendritic cell(s)


enzyme-linked immunosorbent assay


ecto-nucleoside triphosphate diphosphophydrolase


flow cytometry


hematoxylin and eosin


high-mobility group box B1






ischemia reperfusion (injury)




Langerhans cells




liver transplantation


monoclonal antibody


microbe-associated molecular pattern


monocyte chemotactic protein 1


myeloid DCs


mean fluorescence intensity


major histocompatibility complex


mixed leukocyte reaction


messenger RNA


natural killer


orthotopic liver transplantation




pDC Ag


reverse-transcription polymerase chain reaction


Toll-like receptor(s)


tumor necrosis factor alpha


regulatory T cell(s)


University of Wisconsin


wild type.

The liver is regarded as a tolerogenic environment.[1-3] Interstitial antigen (Ag)-presenting cells (APCs) in the liver, in particular, bone marrow (BM)-derived dendritic cells (DCs), appear refractory to stimulation with microbe- or danger-associated molecular patterns (MAMPs or DAMPs), compared with their counterparts, in blood and secondary lymphoid tissues. There is also evidence that liver DCs play important roles in the regulation of hepatic injury[4-6] and innate and adaptive immunity.[3] Several mechanisms may contribute to negative regulation of liver DC maturation and their ability to suppress hepatic inflammation and immunity.[3, 4, 7]

Adenosine triphosphate (ATP) is an essential metabolic energy source in biological systems.[8] Cells undergoing apoptosis or necrosis release ATP, which acts as a DAMP, with proinflammatory and immunostimulatory capacity. Thus, ATP can activate various immune cells, including DCs.[9, 10] ATP also recruits monocytes and neutrophils.[11] The extracellular ATP concentration is strictly maintained by CD39, a member of the ecto-nucleoside triphosphate diphosphophydrolase (E-NTPDase) family that hydrolyzes ATP into adenosine monophosphate. The latter is degraded to adenosine, a potent anti-inflammatory molecule, by ecto-5'-nucleotidase (CD73), another ecto-nucleotidase.[12] CD39 is expressed on regulatory T cells (Tregs) and its hydrolysis of ATP and production of adenosine are considered mechanisms of immune regulation by Tregs.[13] Though expression of CD39 by epidermal Langerhans cells and blood monocyte-derived DCs has also been reported on and implicated in T-cell activation,[14, 15] its expression and function on other tissue-resident DCs, in particular, liver DCs, has not been investigated.

Hepatic ischemia-reperfusion injury (IRI) remains an important clinical problem.[16, 17] In transplantation, its significance is enhanced by the increased use of extended criteria donor organs. Oxygen deprivation induces death of hepatocytes, which release various DAMPs, such as high-mobility group box B1 (HMGB-1), self-DNA, self-SNA, and ATP. DAMPs stimulate innate immune mechanisms through cell-associated pattern recognition receptors, which include Toll-like receptors (TLRs), HMGB-1-like receptors, C-type lectin receptors, and nucleotide-binding domain leucine-rich repeats,[18] expressed on innate immune cells. Triggering of DCs by these receptors induces their activation and maturation.[19] DCs have been implicated in the regulation of inflammation and tissue injury after liver IR,[4, 20-22] with both inhibitory and enhancing effects being reported. Though there is evidence for a protective role of CD39 in total hepatic warm ischemia[23] and liver cold IRI[24] based on studies using CD39−/− mice and CD39-overexpressing mice, respectively, cold IRI is more clinically relevant for assessing tissue injury during liver transplantation (LT). Here, we examined the expression and function of CD39 on liver conventional myeloid DCs (mDCs) in vitro and using a cold liver IRI model in vivo. Our novel findings suggest that expression of CD39 on liver mDCs attenuates their proinflammatory activity and exerts a protective affect against extended cold liver preservation injury.

Materials and Methods


Male C57BL/6 (B6;H-2b) and BALB/c (H-2d) mice (8 to 12 weeks old) were purchased from The Jackson Laboratory (Bar Harbor, ME). CD39−/− mice (B6 background) were bred from pairs received from the Beth Israel Medical Center, Harvard University (Boston, MA). Animals were maintained in the specific pathogen-free Central Animal Facility of the University of Pittsburgh School of Medicine (Pittsburgh, PA). Experiments were conducted under an institutional animal care and use committee–approved protocol and in accord with criteria outlined in the National Institutes of Health publication, Guide for the Care and Use of Laboratory Animals. Mice were fed a diet of Purina rodent chow (Ralston Purina, St. Louis, MO) and received tap water ad libitum.


ATP was purchased from Sigma-Aldrich (St. Louis, MO) and Escherichia coli lipopolysaccharide (LPS) was from InvivoGen (San Diego, CA).

Isolation of Mouse Liver, Spleen, and Other Tissue DCs

DCs were isolated and purified as previously described.[7, 25] Thus, livers, kidneys, and spleens were harvested from mice given recombinant human fms-like tyrosine kinase 3 ligand (10 μg/day intraperitoneally for 10 days; Amgen Inc., Seattle, WA) and digested in collagenase (Sigma-Aldrich). Plasmacytoid DCs (pDCs) were positively selected from the DC-enriched fraction using pDC Ag (PDCA)-1 immunomagnetic microbeads (Miltenyi Biotec, Auburn, CA), as previously described.[26] Conventional mDCs (CD11b+CD11c+NK1.1mPDCA-1) were isolated from the pDC-depleted, DC-enriched fraction using anti-CD11c microbeads (Miltenyi Biotec).[7]

Isolation of Human Liver and Blood DCs

Human liver nonparenchymal cells were obtained from histologically normal surgical resection liver tissue as a by-product of hepatocyte isolation using a three-step collagenase perfusion technique[27] and density-gradient centrifugation. Liver and circulating mDCs were isolated using human BDCA-1+(CD1+) DC isolation kits (Miltenyi Biotec).

Flow Cytometry

Mouse cell-surface molecule and intracellular cytokine and FoxP3 staining was performed as previously described.[26] Details of the monoclonal antibodies (mAbs) used are described in the Supporting Methods. Human DCs were also stained as previously described,[28] with the additional use of anti-human CD39 PE (eBioA1; eBioscience, San Diego, CA). Flow cytometry (FCM) analysis was performed using an LSR II flow cytometer (BD Biosciences, San Jose, CA), and data were analyzed using FlowJo software (version 7.6; TreeStar, Inc., Ashland, OR).

T-Cell Purification

Bulk T cells from spleens of BALB/c mice were incubated with a mAb cocktail consisting of anti-CD45R/B220 (RA3-6B2), anti-CD16/CD32 (2.4G2), anti-TER-119, anti-CD11b (M1/70), and anti-Ly6G (RB-8C5; BD PharMingen, San Diego, CA) and non-T cells eliminated by negative selection using Dynabeads (InvitroGen, Grand Island, NY). Methods use to purify Tregs and assess their function are described in the Supporting Methods.

Mixed Leukocyte Reaction

Unstimulated or ATP-conditioned B6 DCs were used as stimulators of bulk normal allogeneic BALB/c T cells (2 × 105/well) in a 72-hour mixed leukocyte reaction (MLR), as previously described.[7]

Cytokine Measurements

Cytokine levels were determined by cytometric bead array (BD Bioscience) (interleukin [IL]-6, tumor necrosis factor alpha [TNF-α] and monocyte chemotactic protein 1 [MCP-1]) or enzyme-linked immunosorbent assay (ELISA; IL-12p40; BioLegend, San Diego, CA).

Real-Time Reverse-Transcription Polymerase Chain Reaction

Total RNA was isolated and messenger RNA (mRNA) expression was quantified, as previously described,[7] by Fast SYBR Green real-time reverse-transcription polymerase chain reaction (RT-PCR) with an ABI-Prism 7000 Fast Sequence Detection System (Applied Biosystems, Foster City, CA) and with appropriate primers (all from Invitrogen, Carlsbad, CA) in triplicate. Primer sequences are provided in the Supporting Methods. Expression of each gene was normalized to β-actin mRNA content and calculated with respect to normal liver tissue.

ATP Hydrolysis Assay

DCs (1 × 105) were incubated with ATP (100 μM), and supernatants were collected at multiple time points (0, 30, 60, and 90 minutes and 2 and 3 hours). ATP concentration was determined by luminescence assay (ATPlite; PerkinElmer, Boston, MA), and the data are expressed as the frequency of luminescent events (counts per second; cps). Adenosine concentrations were measured by mass spectrometric analysis.[29]


Orthotopic LT (OLT) was performed as previously described,[30] with minor modifications.[31] Liver grafts (B6) were perfused with 5 mL of University of Wisconsin (UW) solution by the inferior vena cava, stored in UW solution for 24 hours at 4°C, and then transplanted into normal wild-type (WT) B6 or CD39−/− B6 recipients.

Adoptive Transfer of DCs to Liver Grafts

Purified liver mDCs (3 × 106) syngeneic with the (B6) liver graft were infused intraportally in 50 μL of phosphate-buffered saline using a 35-G needle, immediately after graft implantation.

Assessment of Liver Injury

Liver enzymes (aspartate aminotransferase [AST] and alanine aminotransferase [ALT]) were quantified in serum, as previously described,[31] and graft histopathology was assessed on hematoxylin and eosin (H&E)-stained paraffin sections in a “blinded” fashion. Areas of necrosis were quantified and Suzuki's scores[32] were determined.

Statistical Analyses

Statistical significance was ascertained by the unpaired Student t test using Prism software (version 5.00; Graphpad Software Inc., San Diego, CA). A probability value of P < 0.05 was considered significant.


Mouse Liver Conventional mDCs Are Less Responsive to ATP Than Lymphoid Tissue mDCs and Undergo Less ATP-Induced Maturation

To address its influence on liver and spleen conventional mDCs purified from normal B6 mice, we stimulated freshly isolated cells with ATP overnight. Expression of major histocompatibility complex (MHC) II, CD80, CD86, and B7-H1 increased significantly on a subpopulation of spleen mDCs after ATP stimulation (Fig. 1A). Under identical culture conditions, the influence of ATP on liver mDCs was minimal, and the relative expression of these molecules after ATP stimulation (compared to unstimulated cells) was significantly less on liver mDCs, compared with spleen mDCs (Fig. 1B). The extent of activation of the spleen and liver DC subpopulation by ATP was dose dependent (Supporting Fig. 1A). Next, to test the functional maturation of DCs, we set up MLR, using ATP-stimulated B6 (H-2b) DCs as stimulators and normal BALB/c (H-2d) bulk CD3+ T cells as responders. Although both ATP-stimulated spleen and liver DCs acquired increased ability to induce T-cell proliferation (Fig. 1C), the influence of ATP on spleen DCs was significantly greater (Fig. 1D). This was in keeping with the ability of ATP to enhance T-cell costimulatory molecule expression on these APCs (Fig. 1B). Moreover, whereas both spleen and liver DCs secreted greater levels of proinflammatory cytokines after ATP stimulation, splenic DCs produced greater amounts of IL-1β, IL-6, and IL-12p40 (Fig. 1E). Taken together, these findings indicate that liver mDCs are comparatively resistant to ATP stimulation.

Figure 1.

Mouse (B6) conventional liver mDCs are hyporesponsiveness to ATP stimulation. Freshly isolated spleen or liver mDCs were stimulated with ATP for 18 hours. (A) Cell-surface expression (MFI) of CD40, CD80, CD86, MHC class II (I-Ab), and B7-H1 was measured by FCM. The relative increase in MFI from baseline (without ATP) for each molecule across multiple experiments (n = 5) is shown in (B). †P < 0.05, compared to unstimulated condition; *P < 0.05, comparing spleen and liver mDCs. (C) Mouse (B6) spleen or liver mDCs were prestimulated with ATP for 3 hours, then cocultured with normal allogeneic BALB/c (I-Ad) splenic T cells for 72 hours. Radioisotope (3H) incorporation during the final 18 hours of culture was determined using a beta scintillation counter. *P < 0.05. (D) Relative increase in counts per minute (cpm) from baseline (without ATP stimulation). *P < 0.05. (E) Liver mDCs secreted lower levels of proinflammatory cytokines in response to ATP stimulation than spleen mDCs (n = 3 experiments). *P < 0.05.

Mouse Liver mDCs Express P2X7 and P2Y14 and Higher Levels of CD39 Compared With DCs From Other Tissues

To explore the basis of ATP resistance of liver mDCs, we examined expression of extracellular nucleotide plasma membrane P2 purinergic receptors for ATP on freshly isolated cells by RT-PCR. Though liver mDCs expressed several P2 receptors at the mRNA level, P2X7 and P2Y14 were the most highly expressed (Supporting Fig. 1B). We confirmed the expression of P2X7 and P2Y14 on liver mDCs by FCM. Expression of both P2X7 (especially) and P2Y14 was enhanced after 18-hour ATP stimulation. However, mature liver mDCs (CD86hi) did not express P2X7 or P2Y14 (Supporting Fig. 1C), suggesting that these receptors were negatively regulated upon cell activation. Expression of P2X7[34],[34] was similar on freshly isolated mDCs from spleen, bone marrow, blood, liver, and kidney (Fig. 2A). Because CD39 is the key molecule that hydrolyzes ATP and regulates ATP concentration,[34] we considered that the resistance of liver DCs to ATP might be the result of ATP hydrolysis by cell-surface–expressed CD39. However, whereas >95% of mDCs from each tissue expressed CD39 (data not shown), liver mDCs displayed significantly higher levels (mean fluorescence intensity; MFI) than mDCs from lymphoid and other nonlymphoid tissues, including kidney mDCs (Fig. 2B). CD39 was not detected on mouse hepatocytes (Fig. 2C). Interestingly, liver mDCs, but not liver pDCs (which represent a comparatively high proportion of liver DCs, compared with spleen DC[35]), expressed greater levels of CD39 than other liver and spleen innate and adaptive immune cells (Fig. 2D). Liver mDCs also expressed CD73 (Fig. 2E,F), which contributes to adenosine generation.

Figure 2.

Mouse (B6) liver mDCs express the highest level of cell-surface CD39 among DCs from selected nonlymphoid and lymphoid tissues. Expression of CD39 and its principal receptor, P2X7, on freshly isolated liver, spleen, peripheral blood, BM, and kidney mDCs, as well as hepatocytes was examined by FCM. (A) Expression of P2X7 on each DC population determined by FCM (n = 4 independent experiments). (B) Intensity (MFI) of CD39 expression on each DC population; n = 4 independent experiments. (C) Histogram showing representative data comparing CD39 expression on spleen and liver mDCs and the absence of CD39 on hepatocytes. Bars indicate means plus 1 standard deviation (n = 4 independent experiments). ns, not significant. *P < 0.05. (D) CD39 expression on liver and spleen leukocytes; n = 6 independent experiments. *Significantly higher than all other liver and spleen cell populations. (E) Expression of CD73 on DCs isolated from various tissues; n = 4 independent experiments. (F) Representative data (n = 4 experiments) of CD73 expression on spleen and liver mDCs.

CD39 on Liver DCs Enhances ATP Hydrolysis and Adenosine Production

Freshly isolated DCs were cultured in ATP-containing medium, and ATP concentration was determined at various times by luminescence assay. ATP concentration decreased progressively (approximately 80%) over 120 minutes in the presence of liver mDCs from WT B6 mice. Initially (first 30 minutes), liver and spleen mDCs from WT mice hydrolyzed ATP at similar rates, but only liver mDCs continued to reduce ATP levels over the ensuing 120 minutes (Fig. 3A). By contrast, an equivalent number of DCs from CD39−/− mice failed to hydrolyze ATP. As expected, the extent of ATP hydrolysis mediated by liver versus splenic mDCs was consistent with their different levels of CD39 expression (Fig. 2B,C). However, expression levels of other ectoenzymes, such as CD39L1 and CD39L3, were similar on spleen and liver mDCs (Fig. 3C). ATP stimulation (120 minutes) did not alter CD39 expression on spleen or liver mDCs (Fig. 3C). Production of adenosine (Fig. 3B) also correlated with the differential levels of CD39 and CD73 expression on liver and spleen mDCs (Fig. 2E,F). These data indicate that the superior ability of liver mDCs to hydrolyze ATP results from their comparatively high CD39 expression. To confirm the processing of ATP by CD39 on liver mDCs, we precultured WT or CD39−/− liver mDCs in ATP-containing medium for 3 hours, then applied the cell-free culture supernatant to WT liver mDCs for 18 hours, together with LPS stimulation. As expected, the medium from WT, compared with CD39−/− DCs cultured with ATP, induced less IAb, costimulatory molecule, and B7-H1 expression (Fig. 3D).

Figure 3.

WT, but not CD39−/−, B6 mouse liver mDCs hydrolyze ATP to a greater extent than WT spleen DCs. WT or CD39−/− liver or spleen mDC (1 × 105) were cultured in ATP-containing medium (100 nM) for the specified time. Supernatants were collected at each time point. (A) ATP concentration was determined by luminescence assay; cps, counts per second. (B) Adenosine concentration was measured by mass spectrometric analysis; n = 3 experiments. *P < 0.05, comparing spleen and liver WT mDCs. (C) Expression of CD39, CD39L1, and CD39L3 was determined by FCM on liver and spleen mDCs with or without ATP stimulation (100 μM) for 120 minutes. (D) WT or CD39−/− liver DCs were cultured in ATP-containing medium for 3 hours. Cell-free culture medium was then transferred to WT liver mDC cultures that were stimulated with LPS for 18 hours. Liver mDC phenotype was determined by FCM. Data in (C) and (D) are representative of two independent experiments.

Human Liver mDCs Also Express Comparatively High Levels of CD39 and Hydrolyze ATP Faster Than Blood-Borne mDCs

We assessed CD39 expression on liver DCs freshly isolated from histologically normal surgical resection tissue. Human liver and circulating mDCs were gated on CD45+, lineage (CD3, CD14, CD19, and CD20), and BDCA-1+ cells, as previously described.[28, 36] Similarly to mice (Fig. 2B), human liver mDCs expressed significantly higher levels of CD39 than blood-borne mDCs (Fig. 4A,B) and hydrolyzed ATP faster and produced more adenosine than circulating mDCs (Fig. 4C,D).

Figure 4.

Freshly isolated human liver mDCs express higher levels of CD39 and hydrolyze ATP faster than circulating mDC. (A) Expression of CD39 by seven individual human purified liver and peripheral blood mDC preparations was measured by FCM, as described in Materials and Methods. Liver and circulating mDCs were defined by gating on CD45+, lineage (CD3, CD14, CD19, and CD20), and BDCA-1+cells. (B) Intensity (MFI) of CD39 expression was compared between human liver interstitial and blood-borne mDCs. Each dot represents 1 individual (n = 8 for circulating DCs; n = 10 for liver mDCs). *P < 0.05. (C) Human liver mDCs and circulating mDCs were also purified using immunomagnetic beads, and ATP hydrolysis assay was conducted. (D) Adenosine concentrations were measured at each time point by mass spectrometric analysis. *P < 0.05, comparing circulating and liver mDCs.

CD39 Regulates Liver mDC Responses to ATP After TLR4 Ligation

We next tested the responses of liver mDCs from WT or CD39−/− B6 mice to ATP, in the absence or presence of the TLR4 ligand, LPS, MAMP to which liver-resident APCs are exposed continually under steady-state conditions. LPS stimulation and combined ATP plus LPS stimulation modestly up-regulated MHC II and coregulatory molecule expression on liver mDCs from WT and, especially, those from CD39−/− mice (Fig. 5A). Moreover, CD39−/− liver mDCs secreted significantly greater quantities of proinflammatory cytokines in response to LPS ± ATP stimulation, compared to WT liver DCs (Fig. 5B). CD39−/− liver mDCs also exhibited stronger naïve T-cell allostimulatory ability and induced more interferon-gamma (IFN-γ)+CD8+ T cells in MLR (Fig. 5C,D). These data suggest that CD39 contributes to the immune regulatory function of liver mDCs.

Figure 5.

CD39 regulates the proinflammatory phenotype of liver mDCs. Liver mDCs were isolated from WT or CD39−/− B6 mice. (A) Expression of CD40, CD80, CD86, MHC class II (I-Ab), and B7-H1 on liver mDCs, with or without ATP ± LPS stimulation for 18 hours (ATP, 100 nM; LPS, 1 μg/mL) was determined by FCM. (B) Proinflammatory cytokine concentrations in culture medium were measured using cytometric bead array (IL-6, TNF-α, and MCP-1) or by ELISA (IL-12p40). (C and D) WT or CD39−/− liver DCs were cultured with normal allogeneic (BALB/c) T cells for 72 hours. (C) Radioisotope (3H) incorporation during the last 18 hours of culture was measured by a scintillation counter; n = 3 experiments. *P < 0.05. (D) Incidences of CD8+IFN-γ+ T cells in MLR were determined by FCM. Data are representative of three independent experiments.

Leukocyte Populations in CD39−/− Livers

Absolute numbers of liver mDCs and all other liver and spleen leukocyte populations examined were preserved in CD39−/− mice (Supporting Table 1). There was also no significant difference between WT and CD39−/− CD4 and CD8 T cells in their expression of cell-surface activation markers (Supporting Fig. 2A) or their proliferative capacity after anti-CD3/CD28 bead or allogeneic DC stimulation (Supporting Fig. 2B). However, compared to those from WT mice, splenic Tregs from CD39−/− mice exhibited a reduced suppressive function on effector T-cell proliferation (Supporting Fig. 2C).

CD39−/− Liver Grafts Exhibit Enhanced Cold IRI That Is Attenuated by WT Liver mDCs

To examine the in vivo functional significance of CD39 in LT-associated cold IRI, CD39−/− or WT livers were transplanted into syngeneic (B6) WT recipients with 24-hour cold preservation, as previously described.[37] CD39−/− liver grafts elicited significantly higher levels of serum ALT and AST than WT grafts after 6-hour reperfusion (Fig. 6A). Histological analysis confirmed more-extensive areas of necrosis and elevated Suzuki scores in CD39−/− liver grafts (Fig. 6B,C). Circulating IL-6, IL-12p40, and TNF-α levels were all significantly higher in mice with CD39−/− grafts (Fig. 6D), correlating with higher levels of production of these cytokines by CD39−/− liver mDCs in vitro (Fig. 5B). Freshly isolated mDCs from CD39−/− grafts (6 hours post-transplant) expressed higher levels of cell-surface maturation markers and lower levels of coinhibitory B7-H1 (PD-L1), compared to DC from WT liver grafts (Fig. 6E). Moreover, increased levels of proinflammatory cytokines were observed in grafts from CD39−/− donors (Fig. 6F). These results suggest that, as a result of the absence of CD39, unhydrolyzed ATP activated liver mDCs and exacerbated cold I/R injury. To verify a protective role of CD39 on liver mDCs in vivo, we also examined cold IRI in CD39−/− recipients of CD39−/− liver grafts that received WT or CD39−/− liver mDCs intraportally, immediately after liver implantation. In this experiment, only the adoptively transferred WT liver mDCs could serve as the source of CD39 for ATP hydrolysis. When 3 × 106 WT (CD39+/+), but not CD39−/− liver mDCs, were infused into the CD39−/− liver grafts, the extent of liver IRI was reduced significantly (Fig. 7). These data demonstrate that CD39 on liver mDCs can protect against LT I/R injury.

Figure 6.

CD39 deficiency exacerbates cold liver IRI and the systemic inflammatory response associated with enhanced intragraft DC maturation after OLT (CD39KO → WT B6 versus WT B6 → WT B6) with 24-hour graft preservation. (A) Serum ALT and AST levels were measured 6 hours after LT. (B) Histological assessment of liver graft injury 6 hours after LT (H&E staining; ×40). (C) Liver damage was assessed by Suzuki's score (congestion, vacuolization, and necrosis), and necrotic areas were quantified. (D) Systemic (serum) IL-6, TNF-α, and MCP-1 levels were measured by cytokine bead array and serum IL-12p40 levels by ELISA. (E) Liver graft mDC cell-surface phenotype was determined by FCM 6 hours after LT. (F) Cytokine expression by liver graft mDCs 6 hours after LT. Liver DCs were gated on CD45+CD11c+B220 cells; n = 5 transplants per group. *P < 0.05.

Figure 7.

CD39 expression by liver mDC is important in regulation of extended cold IRI associated with mouse LT. Freshly isolated WT or CD39KO liver mDCs (3 × 106) were adoptively transferred by the portal vein immediately after OLT (CD39KO→CD39KO). (A) Serum ALT and AST levels were measured 6 hours after LT. (B) Representative graft histology showing reduced necrotic damage in graft recipients given WT CD39+/+ liver mDCs. (C) Liver damage was assessed by Suzuki's score and (D) quantitation of area of necrosis; n = 3 transplants per group. *P < 0.05.


Given their high rate of constitutive exposure to dietary antigens and MAMPs, it is important that liver DCs maintain a tolerogenic state under normal physiological conditions to avoid inflammation.[3] Several mechanisms have been proposed to restrain conventional liver mDC activation and maturation that may also contribute to their inherent tolerogenicity. These include expression of negative regulators of TLR signaling[7] as well as production of IL-10.[4, 7] Here, we show, for the first time, that resistance of mouse liver mDCs to maturation induced by ATP is associated with significantly higher constitutive levels of CD39 on these cells, compared with mDCs from secondary lymphoid tissue or kidney. To what extent expression of CD39 in the cis or trans position might govern the responsiveness of the entire DC population in these tissues was not investigated. The higher levels of cell-surface CD39 on liver mDCs correlated with the superior ability of these DCs to hydrolyze ATP, a property that was absent from CD39−/− liver mDCs. Our findings also show that the enhanced cold I/R injury and systemic inflammation observed after OLT from CD39−/−, compared to WT donors, is associated with increased activation and maturation of liver interstitial mDCs. Moreover, WT, but not CD39−/−, liver mDCs exerted a protective effect against transplant-induced liver IRI when adoptively transferred to CD39−/− liver grafts, implicating CD39 expression on liver mDCs in the regulation of liver transplant IRI.

Innate immune cells, such as natural killer (NK) cells or DCs, respond acutely in injury models by virtue of inherent cytotoxic properties and/or release of cytokines. Recently, deletion of CD39, the dominant ectonucleotidase on NK cells, has been shown to attenuate partial warm liver IRI,[38] whereas adoptive cell transfer studies have supported a role of CD39 on NK cells in liver injury.[38] Studies on NKT cells (that express both CD39 and CD73) have suggested a role for purinergic signaling in NKT cell-mediated mechanisms that result in liver immune injury.[39] Activated NKT cells appear to be important in warm hepatic IRI,[40] although less so in cold liver IRI.[24] Thus, it seems unlikely that NKT cells contributed in any major way to the enhanced liver damage associated with CD39 deficiency in the present LT IRI studies. Pommey et al.[24] have shown that mice that overexpress CD39 exhibit CD4+ T-cell lymphopenia and impaired CD4+ T-cell function that afford protection against liver IRI. Here, we found that CD4+ and CD8+ T-cell number and function were preserved in CD39−/− mice. CD39 deficiency promoted liver mDC maturation and their ability to induce T-cell proliferation and IFN-γ production. Tregs also express CD39 and CD73, and CD39−/− Tregs have impaired suppressive function,[13] as we have confirmed (for splenic Tregs) in this study. Tregs were more rare in liver than in spleen in both WT and CD39−/− mice, and whereas a role of Tregs in cold liver IRI has not been established, we cannot completely discount a possible contribution of impaired Treg function to enhanced injury in CD39−/− livers.

Only limited studies have addressed the role of CD39 on DCs. DCs are a heterogenous population of innate immune cells comprising multiple subsets that exhibit considerable phenotypic diversity and functional plasticity. CD39 on mouse epidermal Langerhans cells (LCs; a distinct DC lineage from conventional, tissue-resident mDCs) is the dominant LC-associated E-NTPDase, with diverse modulatory roles in cutaneous inflammation and immunity.[14] In these studies, epidermal CD39−/− DCs showed impaired Ag-presenting capacity, whereas in the present report, CD39−/− liver mDCs displayed enhanced proinflammatory and T-cell stimulatory ability. It has also been reported that CD39 is highly expressed on human immune-regulatory DCs generated with IL-10/TGF-β,[41] both anti-inflammatory cytokines produced by several liver cell populations in the steady state, and in response to inflammation. This property of the liver microenvironment may serve to up-regulate CD39 expression on liver DCs.

Our liver cold I/R data show that mDCs in CD39−/− liver grafts exhibit a more mature phenotype and that these grafts express more proinflammatory cytokines. This suggests that activation of liver mDCs by unhydrolyzed ATP resulting from CD39 deficiency elicits enhanced production of proinflammatory cytokines, induces stronger T-cell responses, and exacerbates CD39−/− liver damage after cold I/R. Our data further show that CD39−/− to CD39−/− LT results in less IRI, compared with CD39−/− to WT cold I/R. High concentrations of ATP induce apoptosis. Thus, CD39−/− T cells, NKT cells, macrophages, and mDCs are more susceptible to apoptosis induced by ATP. Because of the lack of ATP hydrolysis, ATP concentrations remain high, and immune cells, including liver mDCs, may undergo apoptosis in the CD39 knockout (KO) to CD39 KO liver transplant cold I/R model. On the other hand, not only is ATP hydrolyzed by recipient immune cells, but these cells are also more resistant to ATP-related apoptosis as a result of their expression of CD39 in the CD39−/− to WT LT cold I/R model.

ATP usually activates DCs through P2X7.[33, 42] Here, we show that liver mDCs express comparatively high levels of P2X7 and P2Y14. Compared with other P2 receptors, P2X7 requires high levels of ATP (>100 μM) for activation and thus plays an important role under pathological conditions. The expression level of P2X7 was similar between mouse spleen and liver mDCs and increased markedly after 18-hour ATP stimulation, but that the effect of ATP was less on liver mDCs. Interestingly, in the present study, mDCs expressed the highest level of CD39 among liver immune cells, and a higher level of CD39 than on mDCs from other hematopoietic or parenchymal organs, again suggesting that local microenvironmental factors may affect CD39 expression on mDCs. Our confirmation, in humans, that CD39 is expressed at greater levels on liver mDCs, compared with circulating mDCs, and that, as in mice, human liver mDCs hydrolyze ATP much more effectively than blood mDCs, underscores the physiological relevance of our findings.

In conclusion, our data demonstrate that CD39 is a key molecule in the regulation of liver mDC responses to the danger signal, ATP, with the ability to attenuate proinflammatory cytokine production and extended hepatic cold IRI associated with mouse LT. Improved understanding of how CD39 influences DC regulatory functions may promote the development of novel therapeutic strategies to impact inflammatory and also immune-mediated disorders, including those affecting LT.