Inhibition of natural killer cells protects the liver against acute injury in the absence of glycine N-methyltransferase


  • Laura Gomez-Santos,

    1. Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    Search for more papers by this author
  • Zigmund Luka,

    1. Department of Biochemistry, Vanderbilt University, Nashville, TN
    Search for more papers by this author
  • Conrad Wagner,

    1. Department of Biochemistry, Vanderbilt University, Nashville, TN
    2. Tennessee Valley Department of Medical Affairs Medical Center, Nashville, TN
    Search for more papers by this author
  • Sara Fernandez-Alvarez,

    1. Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    Search for more papers by this author
  • Shelly C. Lu,

    1. Division of Gastrointestinal and Liver Diseases, USC Research Center for Liver Diseases, Southern California Research Center for Alcoholic Liver and Pancreatic Diseases and Cirrhosis, Keck School of Medicine, University of Southern California, Los Angeles, CA
    Search for more papers by this author
  • Jose M. Mato,

    1. Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    Search for more papers by this author
  • Maria L. Martinez-Chantar,

    1. Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    Search for more papers by this author
  • Naiara Beraza

    Corresponding author
    1. Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    • Department of Metabolomics, CIC bioGUNE, Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (Ciberedh), Technology Park of Bizkaia, 48160-Derio, Bizkaia, Spain
    Search for more papers by this author
    • fax: 0034944061301

  • Potential conflict of interest: Nothing to report.

  • Supported by grants from Instituto de Salud Carlos III (to N.B.; FIS, PS09/02010; Ministry of Health, Spain), NIH AT-1576 (to S.C.L., M.L.M.-C., and J.M.M.), Plan Nacional SAF2011-29851 to JMM, ETORTEK-2010 (to M.L.M.-C), Sanidad Gobierno Vasco 2008 (to M.L.M.-C), Educación Gobierno Vasco 2011 (to M.L.M.-C), PI11/01588 (to M.L.M.-C). Ciberehd is funded by the Instituto de Salud Carlos III. N.B. is funded by the Program Ramón y Cajal (Ministry of Science and Innovation, Spain).


Glycine N-methyltransferase (GNMT) catabolizes S-adenosylmethionine (SAMe), the main methyl donor of the body. Patients with cirrhosis show attenuated GNMT expression, which is absent in hepatocellular carcinoma (HCC) samples. GNMT−/− mice develop spontaneous steatosis that progresses to steatohepatitis, cirrhosis, and HCC. The liver is highly enriched with innate immune cells and plays a key role in the body's host defense and in the regulation of inflammation. Chronic inflammation is the major hallmark of nonalcoholic steatohepatitis (NASH) progression. The aim of our study was to uncover the molecular mechanisms leading to liver chronic inflammation in the absence of GNMT, focusing on the implication of natural killer (NK) / natural killer T (NKT) cells. We found increased expression of T helper (Th)1- over Th2-related cytokines, tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL)-R2/DR5, and several ligands of NK cells in GNMT−/− livers. Interestingly, NK cells from GNMT−/− mice were spontaneously activated, expressed more TRAIL, and had strong cytotoxic activity, suggesting their contribution to the proinflammatory environment in the liver. Accordingly, NK cells mediated hypersensitivity to concanavalin A (ConA)-mediated hepatitis in GNMT−/− mice. Moreover, GNMT−/− mice were hypersensitive to endotoxin-mediated liver injury. NK cell depletion and adoptive transfer of TRAIL−/− liver-NK cells protected the liver against lipopolysaccharide (LPS) liver damage. Conclusion: Our data allow us to conclude that TRAIL-producing NK cells actively contribute to promote a proinflammatory environment at early stages of fatty liver disease, suggesting that this cell compartment may contribute to the progression of NASH. (HEPATOLOGY 2012)

Glycine N-methyltransferase (GNMT) is the most important methyltransferase of the liver and is responsible for the catabolism of S-adenosylmethionine (SAMe), the main methyl donor of the body. GNMT is down-regulated in patients at risk of developing cirrhosis and is absent in hepatocellular carcinoma (HCC) samples.1 Accordingly, we describe that mice lacking GNMT accumulate high levels of SAMe and develop spontaneous steatosis that progresses to steatohepatitis, cirrhosis, and HCC.1-3

The importance of the liver as an immune organ is well accepted. The liver receives the majority of its blood from the gut and represents an essential line of defense against products from the digestive tract. Several studies underlined the implication of gut-derived endotoxins in the pathogenesis of nonalcoholic steatohepatitis (NASH) in mice and humans, as fatty liver disease sensitizes the liver to lipopolysaccharide (LPS) injury.4-6 In this line, several works highlighted natural killer T (NKT) cells as especially relevant to preserve a balanced T helper (Th)1/Th2 response in genetic- and diet-induced models of NASH.7, 8 More recently, the implication of tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL)-producing NK cells as mediators of liver inflammation and NASH progression has been demonstrated.9

Liver steatosis is a benign process that can be persistent, with no further consequences for the patient. However, when liver inflammation is chronic the progression of NASH might be irreversible. To better understand the molecular mechanisms triggering the progression from steatosis to chronic hepatitis will allow us to develop potential therapeutic strategies to counteract NASH progression.

Herein, we found spontaneous activation of NK cells in livers from GNMT−/− mice that play a key role in mediating concanavalin A (ConA)- and LPS-mediated liver injury, as selective depletion of NK cells significantly attenuated liver damage. Moreover, adoptive transfer of liver-derived TRAIL−/−NK cells significantly attenuated LPS/GalN (D-galactosamine)-mediated acute liver injury in GNMT−/− animals.

Overall, our data support the essential role of TRAIL-producing NK cells in mediating acute liver injury when GNMT is absent and points to this cell compartment as a potential mediator of chronic inflammation in the onset of NASH progression.


ConA, concanavalin A; GNMT, glycine N-methyltransferase; HCC, hepatocellular carcinoma; LPS, lipopolysaccharide; NASH, nonalcoholic steatohepatitis; NK, natural killer; NKT, natural killer T; SAMe, S-adenosylmethionine; TRAIL, TNF-related apoptosis-inducing ligand.

Materials and Methods

GNMT Knockout Animals and Models of Liver Injury.

GNMT−/− mice were bred as described.10 TRAIL−/− mice were kindly provided by Amgen (Seattle, WA). Male 8 to 10-week-old mice were used in our experiments. Animals were treated according to the guidelines of the National Academy of Sciences (National Institutes of Health publication 86-23, revised 1985).

Liver Injury Experimental Models.

ConA (15 mg/kg) (Sigma) was injected intravenously. LPS (16 mg/kg)/GalN (800 mg/kg) (Sigma) was administered intraperitoneally. Jo2 antibody (0.5 μg/g) (BD Bioscience) was injected intraperitoneally. TNF (6 μg/kg) (Preprotech) was administered intravenously. Nicotinamide (NAM) (50 μM; Sigma) was administered dissolved in the drinking water.

Determination of Liver Damage and Apoptosis.

Alanine aminotransferase (ALT) was determined in serum. Histological examination was performed in formalin-fixed liver sections stained with hematoxylin and eosin (H&E). Apoptotic cell death was determined on frozen livers by Transferase-Mediated dUTP Nick-End Labeling (TUNEL) assay using the In-Situ-Cell-Death Detection Kit (Roche-Diagnosis). Caspase-3 activity was quantified in snap-frozen livers.9

RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction (PCR).

RNA was isolated with TriZol Reagent (Invitrogen). Quantitative real-time PCR (qRT-PCR) was performed using SYBR Green reagent (Quantas Biosciences) in an MyiQ single color real-time PCR detection system (Biorad). Gene expression was normalized with GAPDH (glyceraldehyde-3-phosphate-dehydrogenase) and shown in times versus wildtype (WT) basal expression.

Western Blot Analysis.

Protein extracts were resolved in sodium dodecyl sulfate (SDS) polyacrylamide gels, transferred to nitrocellulose membranes (Whatman), and probed with primary antibodies: p-JNK1/2 (Thr 183/Tyr185), p-c-jun (Ser73), and poly(ADP-ribose) polymerase (PARP) (Cell Signaling); Bcl2, Bcl-xL, and NOS2 (Santa Cruz Biotechnology). GAPDH antibody was used as loading control (Abcam). Secondary antibodies: antirabbit IgG-HRP-linked (Cell Signaling) and antimouse IgG-HRP linked (Santa Cruz Biotechnology).

MNC Isolation and Flow Cytometric Analysis.

Spleen and liver mononuclear cells (MNCs)9 were stained with CD45-APC-Cy7, NK1.1-PE-Cy7, Annexin V-FITC (BD), and CD3-APC (eBioscience). Flow cytometry analysis was performed on a flow cytometric analysis (FACS) Canto II (BD). NK cells were purified from cell suspension by magnetic cell sorting (MACS; Miltenyi Biotec) using anti-DX5-conjugated magnetic beads (Miltenyi Biotec). Only cell preparations with a >90% purity were used for further experiments.

NK1.1+ Cell Depletion and Adoptive Transfer Experiments.

NK cell-specific depletion was obtained with 250 μg/mouse of Asialo-GM1 mAb (Wako Chemicals) administered intraperitoneally 48 and 2 hours before ConA or LPS/GalN treatment. Adoptive transfer experiments were performed by intrahepatic injection of 2 × 106 splenic MNCs or 20 × 106 liver NK cells from TRAIL−/− or GNMT−/− mice.

Statistical Analysis.

Data are expressed as mean ± standard deviation of the mean. Statistical significance was determined by two-way analysis of variance followed by Student's t test. All data shown are representative of at least three independent experiments.


Hepatic NK1.1+ Cells Are Activated in the Absence of GNMT.

Down-regulation/deficiency of GNMT correlates with progression of steatohepatitis, cirrhosis. and HCC development in humans and mice (GNMT−/−).1, 11 Nonalcoholic fatty liver disease (NALFD) patients and GNMT−/− mice have increased free fatty acids (FFA) and bile acids (BA) in serum.12 Previous work in NASH patients correlated high presence of FFA and BA with expression of TRAIL receptor-2 (DR5) that sensitizes hepatocytes to TRAIL-mediated apoptosis.13, 14 Similarly, we found strong DR5 expression in young (8 to 10-week-old) GNMT−/− mice (Fig. 1A), showing early signs of fatty liver disease (Supporting Fig. 1A). We found higher interferon-gamma (IFN-γ) in GNMT−/− mice, whereas TNF levels were lower, but not significantly, than in WTs (Fig. 1A), suggesting the lower presence or activation of Kupffer cells. Accordingly, we observed lower F4/80, interleukin (IL)-1β, and IL-6 expression in GNMT−/− mice compared to WT (Supporting Fig. 2A). NK/NKT cell-related cytotoxic molecule perforin was significantly overexpressed in GNMT−/− mice, whereas IL-4 and IL-10 were down-regulated when compared to WT (Fig. 1A).

Figure 1.

GNMT deficiency promotes activation and increased cytotoxicity of NK/NKT cells. (A) qRT-PCR of livers from GNMT−/− and GNMT+/+ mice. (B) FACS analysis of livers. The graph represents the percentage of NK1.1+ cells versus CD45+ cells. Histogram/graph showing NK1.1+/CD69+ and NK1.1+/TRAIL+ in livers. (C) Cytotoxicity assay coculturing liver-derived MNCs with YAC-1 and liver-MACS isolated NK cells with GNMT−/− hepatocytes at the E:T ratios shown in (D) qRT-PCR of MACS-isolated NK/NKT cells. (E) qRT-PCR of hepatocytes. n = 4. *P < 0.05; **P < 0.01; ***P < 0.001 (GNMT−/− versus GNMT+/+). Error bars represent SD.

Fatty liver disease is associated with alterations of the innate immune system.7-9 We observed depletion of NK1.1+ cells in liver (Fig. 1B) and spleen (Supporting Fig. 3A) from GNMT−/− mice, accompanied by higher expression of the early activation marker, CD69, and TRAIL (Fig. 1B). Cytotoxic assays either coculturing liver (Fig. 1C) or splenic (Supporting Fig. 3A) MNCs with YAC-1 cells or MACS-isolated NK cells with primary hepatocytes from GNMT−/− mice (Fig. 1C) confirmed strong NK cell activation in GNMT−/−. Moreover, analysis of isolated NK/NKT cells showed higher TRAIL, IFN-γ, perforin, CCL5, and CCR5 but low IL-4 in GNMT−/−-cells (Fig. 1D). Interestingly, hepatocytes isolated from GNMT−/− animals showed increased expression of RAE-1 and MULT-1; both NK cells activating ligands expressed in stressed cells but not in normal hepatocytes.15-17 CCL5 and DR5 were also augmented in GNMT−/− hepatocytes (Fig. 1D).

Overall, these data show spontaneous depletion, activation, and cytotoxic activity of NK1.1+ cells and unbalanced cytokine expression shifted toward a proinflammatory Th1-response in GNMT−/− mice. This could potentially contribute to the chronic inflammation that precedes the progression of NASH in the absence of GNMT.

GNMT−/− Mice Are Sensitive to T-Cell-Mediated Acute Hepatitis.

ConA induces acute hepatitis through activation of T and NKT cells.18 ConA promoted extensive liver injury in GNMT−/− mice, as evidenced by significant increase in ALT levels 6 hours after administration (Fig. 2A). Histological analysis revealed evident liver parenchyma degeneration in GNMT−/− mice (Fig. 2B), further confirmed by TUNEL (Fig. 2C) and quantification of caspase-3 activity (Fig. 2D). ConA resulted in a significant increase of IFN-γ but comparable TNF expression in GNMT−/− animals (Fig. 2E). Enzyme-linked immunosorbent assay (ELISA) on serum samples confirmed these data (Supporting Fig. 4A). Perforin and granzyme were augmented both at 3 and 6 hours after ConA (Fig. 2E). In line with the strong proinflammatory response, we found low expression of IL-10 in ConA/GNMT−/− mice (Fig. 2E). Additionally, NOS2 up-regulation observed in ConA/GNMT+/+ mice was significantly delayed in ConA/GNMT−/− (Fig. 2E). We detected strong JNK-activation in GNMT−/− mice 3 hours after ConA, whereas in WT animals this response was weaker and delayed (Fig. 2F).

Figure 2.

GNMT deficiency hypersensitizes the liver to ConA-acute hepatitis. (A) ConA increased ALT serum levels. (B) H&E staining. (C) TUNEL assay in liver sections (200×). (D) Caspase-3 activity on liver extracts. (E) qRT-PCR. (F) Western blot analysis. n = 4. *P < 0.05; **P < 0.01 (GNMT−/− versus GNMT+/+). Error bars represent SD.

NK/NKT cell activation upon ConA was confirmed by FACS analysis of spleen-derived MNCs showing stronger depletion and apoptosis of NK1.1+ cells in GNMT−/− animals after 3 hours (Supporting Fig. 5A).

GNMT Deficiency Does Not Sensitize the Liver Against Fas-Mediated Cell Death.

During ConA-acute hepatitis, NKT cells undergo depletion, activation, and apoptosis through activation of the Fas-signaling pathway.18 In order to identify if this could be the mechanism mediating ConA-liver injury in GNMT−/− mice we administered the Fas-activating antibody Jo2.

As expected, 50% of GNMT+/+ mice died within 4 hours after Jo2 (Fig. 3A). On the other hand, all GNMT−/− survived up to 6 hours after Jo2. ALT levels were elevated, although not significantly different than WTs (Fig. 3B). H&E staining on liver sections revealed severe but similar tissue damage within mouse strains 6 hours after Jo2 (Fig. 3C). This observation was confirmed by TUNEL and Caspase-3 activity assays (Fig. 3D). Accordingly, Jo2 promoted similar cleavage of PARP and down-regulation of Bcl-xL (Fig. 3E).

Figure 3.

GNMT deficiency does not hypersensitize the liver to Fas-mediated liver injury. (A) Jo2 injection was lethal for 50% WT animals after 4 hours, whereas no GNMT−/− mice died. (B) ALT levels on serum. (C) H&E staining (200×). (D) TUNEL assay (200×), quantification of TUNEL positive (% versus DAPI+ cells) and Caspase3 activity. (E) Western blot analysis. (F) qRT-PCR (times versus WT control). n = 4. *P < 0.05; **P < 0.01 (GNMT−/− versus GNMT+/+). Error bars represent SD.

Improved survival of GNMT−/− mice over GNMT+/+ after Fas engagement could be better explained by the lower expression of Fas in GNMT−/− hepatocytes and whole livers at basal conditions (Fig. 3F).

In summary, the lower presence of Fas may counteract higher sensitivity to apoptosis in GNTM−/− mice, rendering an overall comparable impact of Fas engagement when compared to WT littermates. Overall, our data suggest that Fas does not mediate the hypersensitivity of GNMT−/− livers to tissue damage.

NK Cells Mediate Acute Hepatitis After ConA in GNMT−/− Mice.

Several works highlighted the main role of NKT cells18 and excluded NK cells19 as mediators of ConA-hepatitis. Conversely, NK cells mediate ConA-liver injury in an animal model that spontaneously develops NASH.9 Because we found NK cell activation in GNMT−/− mice (Fig. 1), we hypothesize that they may mediate ConA-hepatitis. To prove this, we pretreated GNMT−/− mice with the anti-ASIALO-GM1 antibody, which selectively depletes NK but not NKT cells.20, 21 Accordingly, treatment with anti-ASIALO-GM-1 led to significantly lower ALT (Fig. 4A), substantial improvement of liver parenchyma, and fewer hepatocyte cell death in GNMT−/− mice (Fig. 4B). Perforin, granzyme, and IFN-γ were significantly down-regulated in ConA/ASIALO/GNMT−/− mice. Interestingly, despite ASIALO treatment TNF expression remained comparable to ConA/GNMT+/+ animals (Fig. 4C). ELISA on serum samples confirmed these data (Fig. 4D). ASIALO/GNMT−/− mice showed stronger phosphorylation of JNK than GNMT−/− animals 6 hours after ConA (Fig. 4E). c-jun and NOS2 are mediators of cell protection against ConA-hepatitis.22 In consonance, phosphorylation of c-jun correlated with elevated NOS2 in ConA/ASIALO/GNMT−/− mice (Fig. 4E). Finally, stronger Bcl-xL confirmed the protective effect of ASIALO in GNMT−/− mice (Fig. 4E).

Figure 4.

Specific NK cell depletion ameliorates acute liver injury after ConA in GNMT−/− mice. (A) ASIALO pretreatment decreased ALT levels. (B) Improved liver parenchyma and lower apoptosis (H&E and TUNEL; 200×). (C) qRT-PCR (times versus WT control). (D) ELISA on serum samples. (E) Western blot analysis of livers n = 4. *P < 0.05; **P < 0.01 (GNMT−/− versus GNMT+/+). Error bars represent SD.

Overall, our data indicate that inhibition of NK cells protects the liver against ConA-acute hepatitis in the absence of GNMT.

GNMT Deficiency Sensitizes the Liver to Endotoxin-Mediated Injury.

Gut-derived endotoxins contribute to chronification of inflammation in the onset of NASH, as fatty livers are hypersensitive to LPS.7, 23 We hypothesized that loss of GNMT may affect LPS-driven hepatic inflammation. LPS/GalN had no impact on WT animals, whereas ALT was significantly increased in GNMT−/− mice after 8 hours (Fig. 5A). H&E staining confirmed acute injury in GNMT−/− animals, as liver parenchyma destruction, abundant red-blood cells infiltration, and apoptotic bodies were observed (Fig. 5B). Extensive hepatocyte apoptosis was underlined by TUNEL assay (Fig. 5C), cleavage of PARP, and low expression of Bcl-2 in livers from GNMT−/− mice (Fig. 5D). Moreover, JNK was strongly phosphorylated in LPS/GalN/GNMT−/− mice (Fig. 5D). GNMT−/− mice had higher IFN-γ levels than GNMT+/+ littermates (Fig. 5E), but despite the strong impact of LPS on liver injury, no significant differences on TNF expression were observed (Fig. 5E). NOS2 up-regulation in GNMT−/− animals was significantly lower than that found in LPS/GalN/GNMT+/+ (Fig. 5E). Finally, strong production of perforin was detected 8 hours after LPS/GalN, whereas the significant up-regulation of IL-4 found in WTs was greatly attenuated in GNMT−/− (Fig. 5E).

Figure 5.

GNMT deficiency oversensitizes the liver to LPS/GalN-mediated liver injury. (A) Increased serum ALT levels, (B) H&E, and (C) TUNEL assay (200×) confirmed that LPS/GalN (16 μg/kg / 800 mg/kg) promotes liver injury in GNMT−/− mice. (D) Western blot analysis. (E) qRT-PCR. n = 4. *P < 0.05; **P < 0.01 (GNMT−/− versus GNMT+/+). Error bars represent SD.

Overall, these data demonstrate the hypersensitivity to endotoxin-mediated liver injury in the absence of GNMT at early stages of fatty liver disease.

Inhibition of NK Cells Protects the Liver from LPS Injury in the Absence of GNMT.

Previous work showed that endotoxin (Propionibacterium acnes) reduced liver NKT cells, promoting a polarization toward a Th1 response, a phenomenon also observed in genetically obese mice upon LPS administration7, 8 and similar to what we found in GNMT−/− mice (Fig. 5). However, NK but not NKT cells mediate ConA-hepatitis in GNMT−/− animals (Fig. 3). To better characterize the molecular mechanisms underlying the hypersensitivity of GNMT−/− mice to LPS liver injury we selectively depleted NK cells. Liver damage was greatly attenuated after LPS/GalN in ASIALO/GNMT−/− mice, as ALT levels were largely reduced (Fig. 6A). H&E staining evidenced attenuation of red blood cell infiltration, lower presence apoptotic bodies, and less parenchyma disruption in ASIALO/GNMT−/− mice (Fig. 6B). TUNEL assay confirmed the robust antiapoptotic effect of ASIALO (Fig. 6B). Accordingly, the inflammatory response was greatly weakened; IFN-γ expression was significantly lower in ASIALO/GNMT−/− mice compared to GNMT−/− (Fig. 6C). Interestingly, a decrease in TNF levels was not significant between treatment groups (Fig. 6C). NOS2 was significantly induced in ASIALO/GNMT−/− animals upon LPS/GalN, reaching comparable levels to those found in WT animals (Fig. 6C). In consonance with NK cell inactivation, perforin expression was lower in ASIALO/GNMT−/− livers after LPS/GalN (Fig. 6C). Finally, IL-4 expression was enhanced but did not change significantly when compared to GNMT−/−, suggesting that NK cell depletion does not alter NKT cells response to LPS/GalN (Fig. 6C). Phosphorylation of both JNK (Fig. 6D) and c-jun was detected in ASIALO/LPS/GalN/GNMT−/−.

Figure 6.

Selective depletion of NK cells attenuates strong liver injury after LPS/GalN. (A) Lower ALT serum levels, (B) H&E and TUNEL assay on liver sections confirmed amelioration of liver damage in ASIALO/GNMT−/− mice (200×). (C) qRT-PCR detecting IFN-γ, TNF, NOS2, Perforin, and IL-4 (times versus WT control). (D) Western blot analysis showed strong p-JNK1/2 and p-c-jun in ASIALO/GNMT−/− mice 8 hours after LPS/GalN. GAPDH was used as loading control. n = 4. *P < 0.05; **P < 0.01 (GNMT−/− versus GNMT+/+). All data are representative of three independent experiments. Error bars represent SD.

Taken together, our data suggest that NK cells mediate LPS/GalN-liver injury when GNMT is absent. Moreover, NK cell inhibition seems to shift cell signaling toward the c-jun/NOS2 pathway, with cell-protective characteristics.

Lack of TRAIL Protects the GNMT-Deficient Liver Against LPS/GalN-Mediated Acute Injury.

Liver NK cells constitutively express TRAIL and are the main producers of this cytokine in the body.20, 21 To uncover the implication of TRAIL as a mediator of liver injury we performed adoptive transfer experiments of TRAIL-deficient liver NK cells into ASIALO/GNMT−/− mice, which significantly protected against LPS/GalN-liver injury. Lower ALT levels (Fig. 7A), restoration of the liver parenchyma status, and attenuation of apoptosis in GNMT−/− mice receiving TRAIL−/−NKs were patent (Fig. 7B). Adoptive transfer of liver NK cells from GNMT−/− mice restored LPS liver injury in ASIALO/GNMT−/− animals (Fig. 7A,B). qRT-PCR analysis confirmed attenuation of the inflammatory response, as IFN-γ was significantly reduced in TRAIL−/− NKs/ASIALO/GNMT−/− mice (Fig. 7C). Moreover, NOS2 expression was greatly increased in TRAIL−/− NKs/ASIALO/GNMT−/− mice, whereas perforin was significantly downregulated (Fig. 7C). Finally, strong JNK phosphorylation observed in GNMT−/− livers after LPS/GalN was blunted in the presence of TRAIL−/− NKs (Fig. 7D). The damaging impact of LPS was restored by adoptive transfer of GNMT−/− liver NKs (Fig. 7). Similar effects were observed when splenic MNCs were adoptively transferred into mice pretreated with ASIALO and further challenged with LPS (Supporting Fig. 6), confirming the activation of NK cells also in spleens in GNMT−/− mice (Supporting Fig. 3).

Figure 7.

Adoptive transfer of liver TRAIL-deficient NK cells protects the liver against LPS/GalN damage in GNMT−/− mice. ASIALO/GNMT−/− mice received TRAIL−/− or GNMT−/− NK cells isolated from liver. (A) Serum ALT. (B) H&E staining and TUNEL. (C) qRT-PCR. (D) Western blot analysis n = 4. *P < 0.05; **P < 0.01 (GNMT−/−ASIALO/GNMT−/−NKs versus GNMT−/−ASIALO/TRAIL−/−NKs); ##P < 0.01 (GNMT−/− versus GNMT−/− ASIALO/TRAIL−/− NKs). Error bars represent SD.

These data showing systemic activation of NK cells in the absence of GNMT suggest the direct impact of SAMe, which is greatly elevated in GNMT−/− mice, in the activation of NK cells.

SAMe Depletion with NAM Protects the Liver Against ConA and LPS Liver Injury.

To further confirm this, we first depleted SAMe systemically in GNMT−/− mice and further challenged GNMT−/− mice with ConA and LPS. To do so, we fed GNMT−/− mice with nicotinamide (NAM), a substrate of the nicotinamide N-methyltransferase that leads to the normalization of SAMe content and prevents fatty liver and fibrosis formation in GNTM−/− mice.24 As shown in Fig. 8AB, ConA had only a minor impact on NAM/GNTM−/− mice, showing ALT levels, liver damage, and the presence TUNEL-positive cells comparable to those found in WT animals (Fig. 2). Moreover, JNK activation was blunted in GNTM−/− mice in the presence of NAM (Fig. 8C). Similarly, NAM/LPS GNTM−/− mice showed low ALT levels, normal histology, and very low presence of TUNEL-positive cells (Fig. 8D,E). Finally, JNK phosphorylation was almost undetectable in NAM-pretreated mice (Fig. 8F).

Figure 8.

SAMe depletion with NAM protects the liver against damage in GNMT−/− mice. GNMT−/− mice fed with NAM were protected against ConA and LPS damage. (A,D) ALT, (B,E) H&E and TUNEL, and (C,F) p-JNK western blot. **P < 0.01 (GNMT−/− versus GNMT−/−/LPS or ConA); ###P < 0.001 (GNMT−/−/LPS or ConA versus GNMT−/−/NAM/LPS or ConA). Error bars represent SD.

Overall, these data suggest that SAMe depletion successfully protects the liver against acute liver injury, which we proved to be mediated by TRAIL/NK cells in the absence of GNMT, suggesting the implication of SAMe in promoting activation of the NK cell compartment.

GNMT Deficiency Does Not Sensitize the Liver to TNF Injury.

Interestingly, despite evident tissue protection upon NK cell inhibition, TNF levels remain elevated after ConA or LPS/GalN in ASIALO/GNMT−/− mice. We further investigated the impact of TNF on the liver in the absence of GNMT. GNMT−/− primary hepatocytes showed stronger JNK phosphorylation after TNF compared to WT cells (Supporting Fig. 7A). JNK activation correlated with higher p-c-jun in GNMT−/− hepatocytes (Supporting Fig. 7A). Interestingly, TNF did not result in GNMT−/− hepatocyte apoptosis; PARP cleavage was dimly increased in TNF/GNMT−/− hepatocytes but densitometric quantification showed no significant differences with TNF/GNMT+/+ hepatocytes (Supporting Fig. 7B). The same was observed regarding the regulation of Bcl-xL (Supporting Fig. 7B). Finally, in vivo administration confirmed the overall modest effect of TNF on GNMT−/− mice, as a moderate increase in ALT levels (487.3 ± 121.1) (Supporting Fig. 7C) was observed. H&E staining did not evidence tissue damage and both GNMT+/+ and GNMT−/− livers lacked TUNEL-positive cells (Supporting Fig. 7D). In accordance with what was found in vitro, TNF led to stronger phosphorylation of JNK and c-jun phosphorylation in GNMT−/− mice (Supporting Fig. 7E).

Overall, our data indicate that in GNMT−/− mice the TNF signaling pathway activates both JNK and c-jun cascade, having a mild impact on liver homeostasis.


GNMT expression negatively correlates with progression of human liver disease and cirrhosis and it is markedly reduced in human HCC samples. GNMT-deficient mice develop spontaneous steatosis that progresses to steatohepatitis, cirrhosis, and HCC.1 Therefore, GNMT−/− mice are a valuable tool to investigate the molecular mechanisms underlying the progression of NASH.

The liver is considered an important organ of the innate immune system as it is highly enriched with lymphocytes, of which 30%-50% are NK/NKT cells characterized by NK1.1+ expression. Interestingly, in young GNMT−/− animals we found significant depletion of NK1.1+ cells, which correlated with strong cytotoxic activity and cytokine release. We hypothesize that, at early stages of fatty liver disease, activated NK1.1+ cells may contribute to chronic inflammation in the liver, which precedes the progression of NASH.

Previous work highlighted the protective role of NKT cells in genetically or diet-induced fatty livers against inflammation, whereas the implication of NK cells during fatty liver disease has been neglected. The former studies correlated steatosis with reduced NKT cells number and function that lead to a shift toward a Th1 proinflammatory response.7, 8 Accordingly, in GNMT−/− mice we observed low expression of Th2-cytokines (IL-4, IL-10), likely derived from NKT cells, but a strong presence of TRAIL, IFN-γ, and perforin, potentially due to NK cell activation. These data suggested a differential activation of NKT and NK cells in the absence of GNMT. To confirm this, we used an experimental model of fulminant hepatitis using ConA that promotes liver injury through T and NKT cells.18, 19 Herein, we show that GNMT deficiency hypersensitizes the liver to ConA-hepatitis. Notably, selective inactivation with ASIALO indicated that NK cells, but not NKT cells, are the main mediators of ConA-deleterious effects. Furthermore, Fas engagement did not hypersensitize livers from GNMT−/− mice to liver injury. These data suggest that the well-described mechanism by which NKT cells contribute to ConA-damage through FasL is not relevant in the absence of GNMT.18

Liver NK cells get activated at early stages of infections and express TRAIL and IFN-γ in response to endotoxins.20, 25 The vast majority of the blood supply that the liver receives comes from the gut through the portal vein and it is enriched with endotoxins and toxic products derived from food intake. Small intestine bacterial overgrowth and disruption of the intestinal barrier (leaky gut) are features of steatohepatitis and potentially contribute to endotoxemia in patients with NASH.5, 26 Moreover, fat accumulation sensitizes the liver to endotoxin-mediated tissue damage.4, 7, 8 In this line, herein we show a shift toward a Th1-(IFN-γ)-response in LPS/GalN-treated GNMT−/− mice that are strongly susceptible to endotoxin injury. Selective depletion confirmed that NK cells but not NKT mainly mediate LPS damage in the absence of GNMT.

Previous work suggested that the higher susceptibility to LPS damage in obese individuals is mediated by liver-cell hypersensitivity to TNF.4 Interestingly, attenuation of ConA and LPS liver injury in ASIALO/GNMT−/− was accompanied by sustained TNF expression and c-jun. In the liver, c-jun strongly protects hepatocytes,22 and thus we hypothesize that TNF may have a dual role: promoting cell damage in GNMT−/− mice through JNK activation but counteracting it by activation of c-jun/NOS2. Our data showing that TNF administration promoted JNK, c-jun activation but had a minor impact on GNMT−/− hepatocyte death supports this hypothesis.

Based on our data, it is tempting to speculate that in the context of GNMT deficiency the ability of TNF to activate a protective signaling could be overshadowed by the activation of a proapoptotic pathway, potentially mediated by TRAIL. Moreover, the lower presence of Kupffer cell activation markers could explain the lower TNF expression in untreated GNMT−/− mice and argue in favor of the implication of other cell compartments in mediating the proinflammatory status found in these knockout animals.

Liver NK cells express TNF, IFN-γ, cytotoxic molecules like perforin, and granzyme and mainly produce TRAIL,20, 21 which promote cell death through activation of JNK upon binding to DR5 in transformed cells but not in healthy hepatocytes.27, 28 Interestingly, GNMT−/− mice show the strong presence of DR5 in hepatocytes along with activation of NK cells. Our data suggest that two potential mechanisms may underlie the implication of NK cell in mediating chronic liver inflammation in GNMT−/− mice: an indirect one, implicating BAs, and a direct one, involving the impact of SAMe on NK cell biology. First, BAs and fatty acids promote DR5 expression, which is increased in NASH patients, contributing to apoptosis, which is directly linked to inflammation.14 Thus, accumulating BAs and increased DR5 levels found in GNMT−/− mice would sensitize hepatocytes to apoptosis, contributing to chronic inflammation in the onset of NASH development. Activation of TRAIL-producing NK cells found and the strong liver protection elicited by selective depletion and adoptive transfer of liver-derived TRAIL−/−NK cells further supports the deleterious effect of TRAIL-producing NK cells in the absence of GNMT. Recent work describing the implication of TRAIL/NK cells as contributors of chronic inflammation at early stages of NASH supports this mechanism.9

Finally, our data showing that systemic SAMe depletion by NAM elicits a similar degree of protection than NK cell inactivation against ConA and LPS-injury in GNMT−/− mice suggests a direct impact of SAMe in NK cell biology. This is further supported by the depletion and activation of splenic NK cells observed in GNMT−/− mice, which would argue in favor of a systemic effect of SAMe over NK cell activation. Moreover, our work highlights the importance of maintaining the fine balance of SAMe in the body, as lack of SAMe promotes liver injury and HCC development,29 whereas an excess leads to chronic inflammation, fibrosis, and HCC development.1

Overall, the present study provides new insights on the implication of TRAIL-producing NK cells in mediating liver injury at early stages of fatty liver disease when GNMT is absent. This knowledge may facilitate development of potential therapeutic strategies based on NK cell inactivation to counteract hepatocyte injury, which contributes to the chronification of the inflammatory response, a scenario that is critical for the progression of NASH.


We thank Virginia Gutierrez, Begoña Rodriguez Iruretagoyena and Karoline Fechter for technical assistance.