Disruption of tumor suppressor gene Hint1 leads to remodeling of the lipid metabolic phenotype of mouse liver.

A lipidomic and metabolomic investigation of serum and liver from mice was performed to gain insight into the tumor suppressor gene Hint1. A major reprogramming of lipid homeostasis was found in both serum and liver of Hint1-null (Hint−/−) mice, with significant changes in the levels of many lipid molecules, as compared with gender-, age-, and strain-matched WT mice. In the Hint1−/− mice, serum total and esterified cholesterol were reduced 2.5-fold, and lysophosphatidylcholines (LPCs) and lysophosphatidic acids were 10-fold elevated in serum, with a corresponding fall in phosphatidylcholines (PCs). In the liver, MUFAs and PUFAs, including arachidonic acid (AA) and its metabolic precursors, were also raised, as was mRNA encoding enzymes involved in AA de novo synthesis. There was also a significant 50% increase in hepatic macrophages in the Hint1−/− mice. Several hepatic ceramides and acylcarnitines were decreased in the livers of Hint1−/− mice. The changes in serum LPCs and PCs were neither related to hepatic phospholipase A2 activity nor to mRNAs encoding lysophosphatidylcholine acetyltransferases 1-4. The lipidomic phenotype of the Hint1−/− mouse revealed decreased inflammatory eicosanoids with elevated proliferative mediators that, combined with decreased ceramide apoptosis signaling molecules, may contribute to the tumor suppressor activity of Hint1.

Laboratory Animals and were approved by local regulatory authorities. For metabolomic and lipidomic investigations, serum and liver tissue were obtained from 9-week-old male Hint1 +/+ (n = 6) and Hint1 Ϫ / Ϫ (n = 6) mice between 8:00 AM and 12:00 PM under fed conditions and transported frozen on dry ice to Bethesda for analysis.

Untargeted metabolomics and lipidomics and targeted quantitative methods
For serum metabolomics, serum samples were diluted 20-fold with 66% aqueous acetonitrile containing 10 M chlorpropamide (Sigma-Aldrich) as internal standard, vortexed, microfuged at 18,000 g for 20 min, and transferred to ultraperformance liquid chromatography (UPLC) injection vials. Samples were analyzed by UPLC-electrospray ionization-quadrupole time-of-fl ight mass spectrometry (UPLC-ESI-QTOFMS) using an Acquity BEH C18 1.7 m (2.1 × 50 mm) column (Waters Corp., Milford, MA) under the following conditions: solvent A, 0.1% aqueous formic acid; solvent B, 0.1% formic acid in acetonitrile; 95% solvent A for 0.5 min, to 40% solvent A at 4 min, to 1% solvent A at 8 min, fl ushed for 1 min, then equilibrated at initial conditions for 1.5 min using a Waters Acquity UPLC. Flow rate was 0.5 ml/min, and column temperature was maintained at 60°C. Mass spectrometry on UPLC eluates was conducted in positive (ESI+) and negative (ESI Ϫ ) modes using a Waters Synapt HDMS QTOFMS, scanning 50-1,000 amu at a rate of 3.3 scans/s. Capillary voltage was 3 kV, source temperature was 120°C, sampling cone was 30V, and desolvation gas (nitrogen) fl ow was 850 l/h at 400°C. Total run time was 10 min. Samples were randomized and included multiple pooled samples, blanks, and a standard mixture to monitor instrument stability.
For serum lipidomics, to serum (20 l) was added cold chloroform/methanol (2:1 v/v; 100 l) containing 5 M each of LPC(17:0) and PC(17:0/17:0) as internal standards. The solution was vortexed for 30 s and stood for 5 min at room temperature, followed by additional vortexing. Samples were microfuged at 18,000 g for 10 min to separate the phases. Organic phase was removed and evaporated to dryness under nitrogen, and the dried samples were stored at Ϫ 20°C until just before analysis, when evaporated extracts were dissolved in chloroform/methanol (1:1 v/v; 20 l), propan-2-ol/water/acetonitrile (50:25:25 v/v; 200 l) was added, and the solution was transferred to UPLC novo synthesis from serine palmitoyl transferase or ceramide synthases ( 14 ), lead to the activation of downstream kinases and phosphatases that alter the phosphorylation levels of transcription factors, such as MYC, NF-B, and RB ( 15 ). CERs can modulate apoptosis ( 15 ), cell proliferation, differentiation ( 16 ), and ER stress and senescence ( 17 ). Current opinion suggests that there are perhaps in excess of 200 CER species ( 17 ) and that CERs participate in the formation of plasma membrane rafts ( 18 ), which are involved in cell signaling and membrane traffi cking ( 16 ). Regarding mediation of TSG activity, CERs were reported to regulate the TSG phosphatase PTEN in membrane rafts ( 19 ). Moreover, p53 downregulates sphingosine kinase 1, which may compete with ceramide synthase for the metabolism of sphingosine, thus leading to increased CER synthesis ( 20 ). Nothing is known regarding the effect of the TSG HINT1 on CER and sphingolipid metabolism.
To shed further light on the functioning of Hint1 in relation to its activity as a TSG, a comprehensive metabolomic and lipidomic investigation of the Hint1 Ϫ / Ϫ mouse and gender-, age-, and strain-matched WT controls was undertaken. These studies show that a major reprograming of lipid metabolism occurred in the liver when the Hint1 gene was disrupted, an effect also detectable in serum. The net outcome was an increase in LPCs and mitogenic LPAs, together with a shift away from proapoptotic CERs and infl ammatory eicosanoids in the Hint1 Ϫ / Ϫ mouse.

Animals
All mice used in this study were of the 129SvJ strain. Hint1 -null ( Hint1 Ϫ / Ϫ ) mice were obtained from Weinstein and colleagues, who developed this mouse line ( 7 ). They were continuously bred at EPFL, Lausanne and maintained in our facility. Control WT ( Hint1 +/+ ) 129SvJ mice were also maintained in our facility and cohoused with the 129SvJ/ Hint1 Ϫ For confi rmation of arachidonic acid (AA) concentrations in liver, GC-MS was performed on organic extracts of weighed Hint1 +/+ and Hint1 Ϫ / Ϫ mouse liver containing heptadecanoic acid as internal standard. Extracts were derivatized with BSTFA/ TMCS (Sigma-Aldrich) to yield TMS esters as previously described ( 30 ). The Agilent GC-MS system consisted of an oven (6890N), a mass-selective detector (MSD) (5973N), and an autosampler (7683 series). The column was an Agilent J and W HP-5ms (30 m × 0.25 mm × 0.25 m) using helium as carrier gas at 1.0 ml/min, constant fl ow. The inlet was ran in split mode (ratio 5:1) and maintained at 240°C. The oven conditions were as follows: the initial temperature of 125°C was held for 2 min, ramped at 5°C/min to a fi nal temperature of 300°C, and held for 5 min. The MSD transfer line was held at 280°C. The MSD was run in scan mode (50-550 amu) with a 6 min solvent delay. Single ion chromatograms and peak areas were generated using ChemStation software. AA TMS ester ( m/z = 361) and heptadecanoic acid TMS ester ( m/z = 327) were eluted at 25.70 and 21.85 min, respectively. Results were expressed as peak area ratios (relative concentrations) for AA TMS ester/heptadecanoic acid TMS ester per mg wet liver.

Determination of hepatic eicosanoids by fast liquid chromatography-quadrupole linear ion-trap mass spectrometry
Frozen liver specimens from Hint1 +/+ (n = 6) and Hint1 Ϫ / Ϫ (n = 6) mice were shipped on dry ice to Leipzig for analysis. Weighed samples were kept on ice during the extraction. Extraction solution (n-hexane/propan-2-ol, 3/2 v/v; 600 µl), internal standard solution (5 ng/ml, 50 l) ( 31 ), and ceramic beads were added to murine liver tissue (10 mg) or calibrators (200 µl) and quality controls in Nalgene vials (Thermo) and homogenized with a Mikro-Dismembrator (Sartorius AG, Göttingen, Germany) for 1 min and 2,500 rpm. Samples were transferred into 2 ml vials, and extraction solution (600 µl) was added. Samples were incubated for 15 h at 4°C with vortexing and then centrifuged for 5 min at 10,000 g and 4°C. The complete supernatant was transferred into 1.5 ml vials and evaporated under a stream of nitrogen. All dried extracts were stored at Ϫ 80°C. Before LC/MS/MS, the liver extracts were reconstituted in methanol/H 2 O (62:38, v/v; 500 µl), vortexed for 10 min at 4°C, and centrifuged for 1 min at 10,000 g and 4°C. The supernatant was used for LC/MS/MS as previously described ( 31 ) using an AB Sciex 5500 QTrap mass spectrometer (Darmstadt, Germany) with ESI in negative ion mode.

Hepatic phospholipase A2 activity
Liver homogenates were analyzed for phospholipase A2 (PLA2) activity using the EnzChk ® Phospholipase A2 assay kit (Life Technologies, Grand Island, NY) according to the manufacturer's instructions. Fluorescence determinations were made using a Tecan Infi nite M200 multimode microplate reader (Tecan US, Inc., Morrisville, NC).

Histology and immunohistochemistry
Liver sections from all six Hint1 +/+ and six Hint1 Ϫ / Ϫ mice were stained in duplicate with hematoxylin and eosin and scored by a histopathologist for infl ammatory infi ltrates, and lobular foci of more than 10 infl ammatory cells (mainly lymphocytes) were scored. The presence of sinusoidal lymphocytes and hepatocel-injection vials. UPLC-ESI-QTOFMS analysis was conducted using a Waters Acquity CSH 1.7 m C18 column (2.1 × 100 mm) under the following conditions: solvent A, acetonitrile/water (60:40 v/v) containing 10 mM ammonium acetate and 0.1% formic acid; solvent B, propan-2-ol/acetonitrile (90:10 v/v) containing 10 mM ammonium acetate and 0.1% formic acid. Gradient was 60% A to 57% A at 2 min, to 50% A at 2.1 min (ballistic gradient), to 46% A at 12 min, to 30% A at 12.1 min (ballistic gradient), to 1% A at 18 min before returning to initial conditions at 18.5 min and equilibration for 2 min. Flow rate was 0.4 ml/min, and the column was maintained at 55°C. Conditions of the mass spectrometer were as above but with a run time of 18 min. For liver lipidomics, the protocol was adapted in part from a published method ( 29 ). Accurately weighed liver samples ( ‫ف‬ 50 mg) were homogenized in water-methanol (43:57 v/v) in a Precellys ® 24 bead homogenizer (Bertin Corp., Rockville, MD). Chloroform [800 l; containing 5 M each of LPC(17:0) and PC(17:0/17:0) as internal standards] was added, and the mixture was shaken at 37°C for 30 min. Samples were microfuged at 18,000 g for 20 min to separate the phases. Organic phases were removed and evaporated to dryness under nitrogen, and the dried samples were stored at Ϫ 20°C until just before analysis, when evaporated extracts were dissolved in chloroform/methanol (1:1 v/v; 20 l), and propan-2-ol/ water/acetonitrile (50:25:25 v/v; 200 l) was added and transferred to UPLC injection vials. UPLC-ESI-QTOFMS was conducted as described above for serum lipidomics.
Quantitation of lipid species was performed using either multiple reactions monitoring (MRM) or parent ion scanning using a Waters Acquity UPLC coupled to a Waters Xevo TQ mass spectrometer (UPLC-ESI-TQMS). Samples were prepared as described above for serum metabolomics; however, only 5 M LPA(17:0) was added as internal standard. A Waters Acquity BEH C18 column (2.1 × 100 mm) was used under the following conditions: solvent A, water containing 10 mM ammonium acetate and 0.1% formic acid; solvent B, acetonitrile/propan-2-ol (5:2 v:v) containing 10 mM ammonium acetate and 0.1% formic acid. Gradient was 70% A for 1 min, then linear gradient to 50% A at 3 min, to 1% A at 8 min, held until 15 min, returning to initial conditions over 1 min, and held for an additional 2 min for column equilibration. Flow rate was 0.4 ml/min, and the column was maintained at 50°C. The mass spectrometer was operated in MRM and parent ion scanning modes and, as appropriate, in ESI+ and ESI Ϫ modes. Capillary voltage was 2.2 kV, source temperature was 150°C, and desolvation gas (nitrogen) fl ow rate was 850 l/h at 450°C. The cone voltage and collision energy for the MRM transitions were optimally determined by the instrument's Intellistart TM software. Details of the MRM transitions are given in supplementary Table I. Calibration curves were constructed using authentic standards and the internal standards as described above. Serum concentrations of lipid species were expressed in M and liver concentrations in mmol/mg wet liver.
To determine relative concentrations of FAs, CERs, and acylcarnitines (ACs) in liver, extracted ion chromatograms were generated in MassLynx (Waters  Fig. 1A ). Further analysis of the data by OPLS-DA revealed loadings S-plots with positive and negative ions that were signifi cantly elevated and depressed in the Hint1 Ϫ / Ϫ mice relative to the Hint1 +/+ mice ( Fig. 1A ). Assignment of these ions is given in supplementary Table II. The overall metabolic pattern in Hint1 Ϫ / Ϫ mice was of upregulated LPCs.
Sixteen ions corresponding to 15 lipids were found to be lower in Hint1 Ϫ / Ϫ mice in the serum metabolomics experiment, as well as 12 in ESI+ and four in ESI Ϫ mode. Assignment of these ions is given in supplementary Table III. The overall pattern in Hint1 Ϫ / Ϫ mice was of downregulated phosphatidylcholine (PC) and SMs in the metabolomic ( Fig. 1A ) and lipidomic ( Fig. 1B ) investigations.
Only lipidomic analysis was performed on mouse liver. In these experiments, a clear separation of WT and Hint1 Ϫ / Ϫ liver extracts was found in the PCA scores plots in ESI+ and ESI Ϫ modes ( Fig. 1C ). A total of 44 ions (37 in ESI+, 7 in ESI Ϫ ) were annotated as 43 upregulated lipid constituents of the Hint1 Ϫ / Ϫ liver (supplementary Table V). These comprised one LPC, three monoacylglycerols, four FAs (linoleic, arachidonic, docosapentaenoic, and docosahexaenoic acids), 11 PCs (red symbols in  Table V). These comprised one LPC, three PCs, four SMs (green symbols in Fig. 1C ), 12 phosphatidylethanolamine species, and four ceramides (blue symbols in Fig. 1C ). A summary of the lipidomic and metabolomic changes in the Hint1 Ϫ / Ϫ mouse is given in Table 1 .
First, the hypothesis was tested that elevated serum LPC arose from increased PLA2 activity in the liver. Because LPCs are released from membrane PCs by PLA2, the total PLA2 activity of Hint1 +/+ and Hint1 Ϫ / Ϫ liver was determined in liver homogenates. There was no statistically signifi cant difference in PLA2 activities in the two Hint1 genotypes. Therefore, formation of LPCs from PCs by PLA2 did not explain the large differences in serum LPCs and PCs ( Fig. 2A ) between the genotypes; nor did this explain the patterns of hepatic LPCs and PCs ( Fig. 2B ) observed.
Second, the hypothesis was tested that HINT1 might increase the synthesis of PC from LPC by lysophosphatidylcholine acyl transferase (LPCAT) enzymes by upregulating expression of one or more Lpcat genes. Quantitative PCR (QPCR) determinations of Lpcat1 , Lpcat2 , Lpcat3 , and Lpcat4 mRNAs in WT and Hint1 Ϫ / Ϫ livers revealed no statistically signifi cant differences in expression of any of these four Lpcat genes between Hint1 genotypes.
There occurred stark differences in hepatic fatty acid concentrations for PUFAs, for AA, and for its precursors in the de novo synthesis pathway, linoleic acid (18:2n-6) and dihomo-␥ -linolenic acid (20:3n-6) ( Fig. 3 ) . AA, 18:2n-6, and 20:3n-6 were highly statistically signifi cantly elevated in the Hint1 Ϫ / Ϫ livers compared with Hint1 +/+ livers. The 2-fold elevated concentration of AA was confi rmed by an independent GCMS analysis, which yielded a signifi cant 1.5-fold increase of AA in the Hint1 Ϫ / Ϫ mice, with values lular giant cells was recorded. Immunohistochemistry was performed for detection of the murine F4/80 160 kDa cell surface glycoprotein that is expressed predominantly on murine macrophages. Paraffi n-embedded, formalin-fi xed sections were deparaffi nized and hydrated with xylol, ethanol, and then distilled water. Antigen retrieval was performed with proteinase K, followed by washing with distilled water and endogenous peroxidase quenching with 3% methanolic hydrogen peroxide, followed by washing with distilled water and blocking with 10% rabbit serum in PBS. Primary antibody incubation was conducted with F4/80 monoclonal antibody to mouse macrophages (clone BM8; BMA Biomedicals, Augst, Switzerland). Secondary antibody incubation used a polyclonal rabbit anti-rat immunoglobulin-biotinylated secondary antibody followed by incubation using the Vectastain Universal Elite ABC kit (Adipogen, Liestal, Switzerland). DAB was used as peroxidase substrate, and counterstaining was with hematoxylin.

Real-time quantitative PCR
Total RNA was extracted from liver tissues (30 mg) preserved with RNA later and using RNeasy columns (QIAGEN AG, Hombrechtikon, Switzerland) with an on-column digestion step as described in the manufacturer's protocol. Total RNA (2 g) was reverse transcribed with Superscript III reverse transcriptase (Invitrogen AG, Basel, Switzerland). cDNA was added to a fast universal PCR master mix (#4352042; Applied Biosystems, Zug, Switzerland) and amplifi ed using an Applied Biosystems 7500 Real-Time PCR system, with primers (Applied Biosystems) specifi c for mouse genes Fads1 (Mm00507605_m1), and Fads2 (Mm00517221_m1). The data were analyzed according to the 2 ⌬ ⌬ Ct method, and Ct values were normalized to 18S rRNA (Hs99999901_s1). All reactions were performed in duplicate.

Multivariate data analysis
For the UPLC-ESI-QTOFMS metabolomics/lipidomics experiments, mass spectral data were centroided and deconvoluted to generate a multivariate data matrix using MarkerLynx (Waters). Peak picking, alignment, deisotoping, and integration were performed automatically by the software and transformed into the multivariate data matrix containing aligned peak areas with matched m/z and Rt values. For the serum metabolomic data, the peak areas were normalized to the peak area of the internal standard chlorpropamide (ESI+ = 277.041 and ESI Ϫ = 275.024 m/z ; Rt = 5.3 min). Data were Pareto-scaled and analyzed by unsupervised principal components analysis (PCA), followed by orthogonal projection to latent structures-discriminant analysis (OPLS-DA) using SIMCA 13 (Umetrics AB, Malmö, Sweden) comparing data from Hint1 +/+ and Hint1 Ϫ / Ϫ samples.

Univariate statistics
Statistical analysis of groups was made using GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA). In all cases, nonparametric Mann-Whitney U tests were used. All means are expressed ± standard deviation, including in the fi gures.

RESULTS
Untargeted metabolomic analysis of mouse sera was carried out to discover biomarkers associated with HINT1 expression. A clear separation of Hint1 +/+ and Hint1 Ϫ / Ϫ sera was found in the PCA scores plots in ESI+ and ESI Ϫ consistent with the observations that LPC(18:2), LPC(20:4), and PC(20:4) were also signifi cantly higher in livers of Hint1 Ϫ / Ϫ mice. It therefore appeared that one phenotype of the Hint1 Ϫ / Ϫ mouse was a small (1.5-to 2-fold) accumulation of AA, together with AA metabolic precursors ( Fig. 3B ), and their phospholipid derivatives in the liver ( Fig. 2B ).
Third, because of the signifi cantly elevated hepatic levels of AA and its phospholipid derivatives, the eicosanoid profi le of Hint1 +/+ and Hint1 Ϫ / Ϫ livers was determined using LC-ESI-TQMS.  ( Fig. 3B ). Because of these metabolic fi ndings, QPCR was conducted to evaluate mRNA expression for the three enzymes involved in the conversion of linoleic acid (18:2) to arachidonic acid (20:4). The fi ndings for Fads1 , Elvol5 , and Fads2 are given in Fig. 3B . The expression of mRNA for these three genes was 1.5-to 2-fold higher in the null mice, consistent with the observed increased hepatic AA levels. The increased hepatic levels of 18:2 and 20:4 were from the UPLC-ESI-QTOFMS data and expressed in arbitrary units/mg wet liver. All four CERs were statistically signifi cantly decreased in the Hint1 Ϫ / Ϫ livers, by 35% to 60% ( Fig. 2C ). Regarding LPAs, up to 10-fold elevated serum concentrations of LPA(16:0), LPA(18:0), LPA(18:1), and LPA(20:4) were found in Hint1 Ϫ / Ϫ mice as compared with their Hint1 +/+ counterparts ( Fig. 2A ). Sixth, because of the lipid changes observed in the livers of the null mice and the observation that C4-and C18carnitines were depressed in the null livers (supplementary Table V), hepatic concentrations of C2-, C4-, and C18-carnitines were determined from the UPLC-ESI-QTOFMS dataset using extracted ion chromatograms for each liver sample. The results showed a highly statistically signifi cant decrease in all three acyl carnitines, but not in C16-carnitine, in the null livers (supplementary Fig. III). This pattern is suggestive of increased fatty acid oxidation in the null mouse, but this is not supported by the hepatic fatty acid data in Fig. 3A .
Finally, because of the increase in AA, its metabolic precursors, and phospholipid metabolites in the Hint1 Ϫ / Ϫ livers, we decided to examine formalin-fi xed sections of WT and Hint1 Ϫ / Ϫ livers for evidence of infl ammation. Histological examination of liver sections from six Hint1 +/+ mice showed one liver with a single infi ltrate with a lobular focus of 10 or more lymphocytes. The other fi ve mice were negative in this regard. For six Hint1 Ϫ / Ϫ mice, three livers were negative, but three showed 2-3 foci of lymphocytic infi ltration. One of these livers displayed hepatocellular giant cells ( Fig. 4 ). However, these differences were not statistically signifi cant. In contrast, immunohistochemistry for the mouse macrophage surface glycoprotein F4/80 demonstrated a larger number of macrophages resident in the Hint1 Ϫ / Ϫ liver than the Hint1 +/+ liver ( Fig. 4A,B ). The macrophage cell counts were highly statistically signifi cantly that represent 44 distinct prostaglandin and leukotriene metabolites. Contrary to expectations, prostaglandins and their metabolites were signifi cantly lower in the null mice. Of particular note was the 4-fold decrease in hepatic PGF2 ␣ from 409 ± 142 to 101 ± 58 pg/mg liver ( P = 0.002) and the almost 100-fold decrease in 6-keto-PGE1 ␣ from 92 ± 104 to 1 ± 0 pg/mg liver ( P = 0.01). The only statistically signifi cant increase observed for the null livers was the level of 20-hydroxy-PGE2 from 14 ± 4 to 114 ± 63 pg/mg liver ( P = 0.026). This metabolite arises by thehydroxylation of prostaglandin E2 (PGE2) by murine CYP4A14 ( 32 ). The downregulated eicosanoids in null livers all arose from COX1/COX2 metabolism of AA (supplementary Fig. II). Eicosanoid metabolites arising from cytochrome P450 or lipoxygenase metabolism of AA were unaltered in the null livers ( Table 2 ), showing a degree of selectivity for the COX1/COX2 pathway. Fourth, increased FFA uptake by the liver might explain the differences observed in hepatic fatty acids in Hint1 Ϫ / Ϫ mice ( 33 ). Increased FFA hepatic uptake has been proposed as a mechanism of liver infl ammation ( 34 ). However, mRNA quantitation by QPCR could not detect a difference in expression of Slc27a1 or Fabp1 , two genes involved in the uptake of long-chain FFA.
Fifth, representatives of two other groups of phospholipid were determined in serum of the two mouse genotypes, SMs and LPAs. There were no statistically signifi cant differences in the serum M concentrations of SM(d18:1/16:0) and SM(d18:1/18:1) as determined by UPLC-ESI-TQMS ( Fig. 2A ). Six ions corresponding to four hepatic CERs had already been observed in the lipidomic experiment to be downregulated in Hint1 Ϫ / Ϫ livers (supplementary Table V). Because SMs are metabolized to proapoptotic CERs, the hepatic concentrations of these four ceramides was determined  a ↑ and ↓ represent increases and decreases, respectively, in molecules determined by the multivariate data analysis of UPLC-ESI-QTOFMS data. The exact fold changes were not determined, except in cases where quantitation by UPLC-ESI-TQMS or extracted ion chromatograms from the UPLC-ESI-QTOFMS data were performed. Such data appear in Figs. 2, 3, and supplementary Fig. III. b n is the number of molecules of a particular class that were found to be either increased or decreased in the Hint1 Ϫ / Ϫ mice. c Glycosylceramide (see supplementary Table IV).  ( Fig. 4D ). Therefore, a previously undisclosed phenotype of the Hint1 Ϫ / Ϫ mouse liver is an enhanced population of hepatic macrophages.

DISCUSSION
The current study revealed that the Hint1 Ϫ / Ϫ mouse is characterized by a phenotype involving elevated serum Analyses were performed in duplicate for WT (n = 6) and KO (n = 6) mice. * P < 0.05; ** P < 0.001; *** P < 0.0001. ( 39 ), is activated by phosphorylation by either p42 MAPK or PKC ( 40 ). Therefore, increased PKC activity in Hint1 Ϫ / Ϫ liver due to the absence of a HINT1-PKC inhibitory interaction, as reported for Hint1 Ϫ / Ϫ brain ( 38 ), would be expected to lead to increased cytosolic PLA2 phosphorylation, metabolic activity, and AA release. This may in part explain the increase in the AA pool observed in Hint1 Ϫ / Ϫ mice. The enriched hepatic AA pool was expected to drive prostaglandin E2 (PGE2) synthesis via COX2. However, contrary to expectation, it was found that infl ammatory eicosanoids were depressed in the null mouse livers. Decreased COX1/COX2 activity might lead to a backlog of AA and possibly explain the observed slight increase in hepatic AA. Another observation of relevance was the downregulation of all PCs studied and a concomitant upregulation of all LPCs that were quantitated because several studies have reported that LPCs activate PKC ( 28 ). These observations did not appear to be due upregulated PLA2. Nor were differences detected in the expression of LPCAT genes, Lpcat1 , Lpcat2 , Lpcat3 , and Lpcat4 . Thus, an attenuated conversion of LPC to PC by LPCAT enzymes did not explain the pattern of LPCs and PCs observed. The origins of elevated serum LPCs in the Hint1 KO mouse remain unknown.
Disruption of the Hint1 gene may not only remove the inhibition of PKC by HINT1 but also may cause PKC to become hepatic macrophage population, presumably mostly Kupffer cells. The possible signifi cance of these fi ndings regarding the loss of tumor suppressor function is discussed below.
Disruption of Hint1 increased mediators proliferative signaling in mouse liver while decreasing infl ammatory eicosanoids. The prominent elevation of serum LPA concentrations into the M range can be expected to have a signifi cant mitogenic effect. LPAs stimulate cell surface Gprotein-coupled receptors that activate multiple cell signal transduction pathways, including those involving RAS and RHO family GTPases, which regulate cell proliferation ( 35 ). LPAs are produced extracellularly from LPC by autotaxin and intracellularly from PA by PLA2 ( 35,36 ). However, total PLA2 activity did not differ between Hint1 Ϫ / Ϫ and Hint1 +/+ livers. Studies in hepatoma cell lines Hep3B and Huh7 have indicated that the proinfl ammatory cytokine TNF ␣ promoted autotaxin expression and secretion, leading to an increase in LPA ( 37 ). Thus, infl ammatory signaling within the liver, but not involving eicosanoids, might be expected to lead to LPA-directed proliferation.
Other investigations have established that the ␥ isotype of protein kinase C (PKC ␥ ) was elevated in the brains of Hint1 Ϫ / Ϫ relative to WT mice, with a concomitant 2-fold increase in PKC activity in Hint1 Ϫ / Ϫ brain lysates ( 38 ). In addition, cytosolic PLA2, the principal form of PLA2 responsible for AA release from membrane phospholipids The aforementioned fi ndings are consistent with a phenotype in the Hint1 Ϫ / Ϫ liver of increased proliferative signaling with reduced proapoptotic signaling. Although infl ammation was shown to be required for diethylnitrosamine-induced liver cancer ( 45 ) and is thought to play a role in human HCC ( 46 ), no increase in prostaglandin infl ammatory mediators was found in this study; in fact, a decrease in these molecules was observed in the livers of the null mice. It is not known which components of the lipidomic phenotype of the Hint1 Ϫ / Ϫ mouse are key to cancer susceptibility in these animals. Future investigations will consolidate these fi ndings and lead to a fuller understanding of the role played by hepatic LPC, LPA, AA, CER, DG, and macrophages in hepatocarcinogenesis in the Hint1 Ϫ / Ϫ mouse. Moreover, the role played by HINT1 in the cellular physiology of the liver remains to be elucidated, but lipidomics has provided relevant new insights. As noted by Weinstein and colleagues ( 6 ), there is no other TSG that has a prokaryotic ortholog. HINT1 , on the other hand, is highly conserved, and in the smallest genome of any cellular organism, Mycoplasma genitalium (580 kb), there is a homolog that shares >30% identity with the full human HINT1 sequence ( 6 ). From the current investigations, it is clear that the HINT1 protein has a activated. It is reported here that 16 discrete DGs were upregulated in the livers of the null mice. DGs are physiological activators of PKC and have long been considered as important second messengers in carcinogenesis ( 41 ).
CERs play an important role in apoptosis signaling. CERs also play a role in fatty acid synthesis by upregulating SREBP1c expression. In a recent report, CER(d18:1/22:0) and CER(d18:1/24:0) were two of the three most abundant CERs in human plasma ( 42 ). Both of these CERs were downregulated in serum and liver tissue of Hint1 Ϫ / Ϫ mice. The principal ceramide synthase in mouse liver is CerS2, which can synthesize these two CERs ( 43 ). There are many transcriptional stimuli of ceramide synthase genes, including stress, DNA damage, and hypoxia ( 43 ). Use of the ceramide synthase inhibitor fumonisin B1 in a variety of cultured cells led to decreased CER concentrations and the inhibition of apoptosis in all cases ( 43 ). Moreover, Cers2 -defi cient mice developed hepatic adenomas, many of which progressed to HCC ( 44 ), showing that impaired specifi c CER synthesis can lead to HCC. Thus, the downregulation of CER serum and liver concentrations, including for CER(d18:1/22:0) and CER(d18:1/24:0), in Hint1 Ϫ / Ϫ mice may partly explain the action of Hint1 as a TSG.