Dysregulation of the Lysophosphatidylcholine/Autotaxin/Lysophosphatidic Acid Axis in Acute‐on‐Chronic Liver Failure Is Associated With Mortality and Systemic Inflammation by Lysophosphatidic Acid–Dependent Monocyte Activation

Acute‐on‐chronic liver failure (ACLF) is characterized by systemic inflammation, monocyte dysfunction, and susceptibility to infection. Lysophosphatidylcholines (LPCs) are immune‐active lipids whose metabolic regulation and effect on monocyte function in ACLF is open for study.

(AD) to extrahepatic organ failure. This syndrome often requires intensive care unit admission (1) and is associated with a high risk of early mortality. (2,3) A common precipitant of ACLF is bacterial infection, occurring in up to 40% of cases and complicating recovery in many more. (4) The high rate of susceptibility to infection in ACLF may derive from repeated stimuli from bacterial translocation where there is failure to produce proinflammatory cytokines or respond to further microbial cues with defective phagocytic capacity. (5,6) Patients with ACLF have multiple defects/abnormalities in cellular and soluble components of the immune system, (7) impairing the host's antimicrobial responses and thus conferring an increased susceptibility to infections. (8,9) Circulating monocytes play a fundamental role in mediating antimicrobial defence and innate immune responses in liver injury. (9)(10)(11) Monocyte phenotypic changes in ACLF include increased CD163 and Mer-tyrosinekinase (MerTK) expression, indicating prorestorative yet immunosuppressive functions. (12) MerTK overexpression conferred a decreased ex vivo response to lypopolysaccharide (LPS) and closely correlated to levels of immunosuppression, systemic inflammatory response syndrome activation, and disease severity scores in ACLF. (12) MerTK antagonism is therefore a proposed therapeutic strategy to restore innate response at later stages of ACLF where prolonged compensatory anti-inflammatory response may predispose to infections. (13) Metabolic reprogramming has been proposed to explain some of these defects in innate immunity (14) however, the underlying mechanisms and immunometabolic interactions are poorly understood. Recent evidence suggests important roles of mitochondrial dysfunction, (15) biogenic amines, and lipids as important mediators of the ACLF phenotype. Patients with ACLF have lower HDL cholesterol than those with AD and this level correlates with organ failures and survival. (16) In a large cohort of >500 patients with stable and decompensated cirrhosis, HDL and related biomarker concentrations were profoundly reduced with worsening AD and had significant prognostic potential. (17) Moreover, HDL may be implicated in complications developed by patients with acute and chronic liver disease, and impaired cholesterol efflux capacity has been shown to inhibit cytokine production by monocytes and predict 1-year mortality in these patients. (18) A comprehensive metabotyping study in patients with decompensated cirrhosis determined that the lipid class, lysophosphatidylcholine (LPC), was down-regulated in patients with AD and could be used to accurately predict mortality. (19) The underlying reason for LPC down-regulation was a topic open for study, but LPCs have been widely reported as important immune regulatory molecules in patients with cardiovascular disease. LPCs mediate important aspects of myeloid cell function, including migration and activation, particularly for monocyte-derived macrophages within atherosclerotic plaques. (20) Therefore, LPC down-regulation may have important sequelae for innate immune characteristics in patients with ACLF.
LPC conversion to lysophosphatidic acid (LPA) is mediated by autotaxin (ATX/ectonucleotide pyrophosphatase/phosphodiesterase type 2 [ENPP2]), a phospholipase expressed ubiquitously in mammalian tissues. ATX is up-regulated in plasma and liver of patients with primary biliary cholangitis (where its role as a mediator of pruritus is under investigation), (21) and ATX-induced LPA production can also promote survival of HSCs and initiate fibrosis chronically. (22) However, neither the role of this enzyme in mediating LPC/LPA concentrations in liver failure syndromes nor the differential immune effects of LPC/LPA in this context are yet understood. Although previous work has shown increased ATX concentrations correlating with Child-Pugh score and Model for End-Stage Liver Disease (MELD) in the context of decompensated cirrhosis, (23)(24)(25) in patients with ACLF this remains open for study. ATX is upregulated in blood outside the context of fibrosis in patients with chronic inflammatory states, (26) infections, and cancer. (27) However, the effect of LPA on monocytes in ACLF is an interesting finding.
LPA is a ligand for six LPA receptors (LPAR 1-6 ), acting through G-protein-coupled receptors on a variety of intracellular signaling pathways. Several LPARs are implicated in inflammatory conditions, emerging as potential therapeutic targets in airway-associated inflammation, (28) cancer, (29) and cardiovascular disease. (30) To date, the role of this axis in liver failure and immune dysfunction remains open for study. We therefore sought to characterize the LPC-ATX-LPA axis and mechanistically assess its role on innate immune dysfunction in patients with ACLF.

Patients and Methods patIeNt CHaRaCteRIStICS
The study was ethically approved (12/LO/0167 and 12/LO/1417), and patients provided written informed consent. The London East Ethics Committee were the institutional review board. Assent was obtained by the patients' nominated next of kin, if they were unable to provide informed consent themselves. Patients were categorized into different groups: ACLF; acute decompensation of cirrhosis with no ACLF (AD; according to the Consortium on Chronic Liver Failure-Sequential Organ Failure Assessment [CLIF-SOFA] classification described) (3) ; stable cirrhosis (SC) and no evidence for acute decompensation; acute liver failure (ALF); and healthy controls (HCs). As per the CLIF definition, patients with decompensated cirrhosis admitted for elective procedures were not characterized as AD, but as chronic decompensated (DC). (31) Between December 2012 and July 2014, 151 (SC = 13, AD = 50, ACLF = 49, and HC = 39) subjects were recruited to the study within 24 hours after admission to the Liver Intensive Therapy Unit or liver wards (Supporting Table S1). A second validation cohort (recruited between May 2018 and January 2020) included a slightly different profile of patients, focusing mainly on DC, namely patients with cirrhosis who did not develop ACLF during the study period, which included both "unstable decompensated cirrhosis" requiring readmission and "stable decompensated cirrhosis" not readmitted (as per definition of the PREDICT study). (32) A total of 191 (AD or DC = 73, ACLF = 34, sepsis/infection = 38, and HC = 46) subjects were recruited with the same criteria (Supporting Table S2).
Cirrhosis was diagnosed by a previous liver biopsy or clinical presentation with typical ultrasound or CT imaging. Exclusion criteria were as follows: aged <18 years; malignancy other than HCC; and immunosuppressive therapy other than corticosteroids (maximal dose 40 mg of prednisolone or equivalent), which were accepted if required for the treatment of autoimmune liver disease, alcohol-associated hepatitis, or septic shock. Full blood count, international normalized ratio, liver and renal function tests, lactate, ammonia, and clinical variables, including presence of infections, were entered prospectively into a database. The following disease severity scores were calculated: Child-Pugh, MELD, (33) CLIF-SOFA, (3) North American Consortium for Study of End-Stage Liver Disease (NACSELD), (34) SOFA, (35) CLIF AD, and CLIF-C ACLF scores. Presence of infection in the 10 days of observation was recorded and defined according to the criteria already used by the NACSELD. (36) BIoSaMplINg Plasma and serum samples for metabolic or cytokine profiling were obtained within 24 hours of admission to the hospital. Blood was drawn into plain or lithium heparin-containing vacuum tubes (BD Vacutainer; BD Biosciences, Franklin Lakes, NJ). Serum and plasma fractions were collected following incubation at room temperature and centrifugation; aliquots were stored immediately at -80°C until further analysis. Separately, peripheral blood mononuclear cells (PBMCs) were isolated through density centrifugation (Ficoll-Paque) and used fresh or were stored at -80°C. Further details are given in the Supporting Information.

MetaBolIC pRoFIlINg 1 H Nuclear Magnetic Resonance Spectroscopy
All serum samples were prepared according to the published article by Dona et al. (37) 1 H NMR (nuclear magnetic resonance) spectra were acquired on a Bruker 600-MHz (Avance III) spectrometer, operating at a 1 H frequency of 600.16 MHz (14.1 Tesla) at 300 K using a 5-mm BBI probe and an automated sample Jet system (Bruker Biospin, Rheinstetten, Germany). A full explanation of processing is given in the Supporting Information.

UltRapeRFoRMaNCe lIQUID CHRoMatogRapHy/MaSS SpeCtRoMetRy aNalySIS
Sample preparation and data acquisition for ultraperformance liquid chromatography/mass spectrometry (UPLC-MS) analysis of lipids from serum was performed according to the recently published National Phenome Centre (Imperial College, London, UK) blood lipids standard operating procedure. (38,39) Validation of lipid abnormalities in peripheral blood was performed using the AbsoluteIDQ p180 kit BIOCRATES (BIOCRATES Life Sciences AG, Innsbruck, Austria), using a Waters TQS Micro instrument at King's College Hospital. Full details of the UPLC-MS sample preparation and methods are given in the Supporting Information.

CD14 + Cell ISolatIoN
Monocytes were isolated using CD14 microbeads according to the manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). Purity and viability of CD14 + monocytes were assessed by flow cytometry.

NeUtRopHIl ISolatIoN
Neutrophils were isolated with Polymorphprep (Alere Technologies AS, Oslo, Norway) according to the manufacturer's procedure. Purity and viability of CD66b + , CD15 + , and CD14 − cells were assessed by flow cytometry.

Monocytes
In the last 6 hours of a 24-hour cell culture, 100 ng mL of LPS (Invivogen, San Diego, CA) was added to wells.

Neutrophils
In the last 2 hours of a 4-hour culture, cells were stimulated with LPS (100 ng/mL).

aSSeSSMeNt oF MoNoCyte pHagoCytIC CapaCIty
A total of 500,000 PBMCs were added per well and cultured for 24 hours in complete medium. Cells were incubated with various lipids for 4 hours. Harvested cells were supplemented by 10% autologous plasma, and pHrodo Escherichia coli Red BioParticles (Invitrogen, Paisley, UK) were added for 60 minutes (see Supporting Information).

IMMUNoHIStoCHeMIStRy FoR atX
Hepatic explant tissue from patients undergoing liver transplantation with AD and ACLF, in comparison with background liver parenchyma distant from the tumor in patients undergoing resection for focal liver lesions, underwent hematoxylin and eosin staining and immunohistochemistry for ATX as described in the Supporting Methods (Fig. 3).

geNe eXpReSSIoN pRoFIlINg
Public microarray data sets (Gene Expression Omnibus data set GDS4389, series GSE28619/30) were interrogated to measure selected genes of interest in liver tissue from patients with alcohol-associated hepatitis (n = 15), HBV-related ALF (n = 20), and HCs (n = 7). See Supporting Table S4 for lists of all queried/analyzed genes.

Real-tIMe qpCR oF CD14 + MoNoCyteS
To assess the expression levels of LPAR 1-6 and ENPP2 genes, monocytes were isolated from PBMCs using MACS sorting beads (Miltenyi Biotec) and cultured for 24 hours in complete medium with or without LPS stimulation. qRT-PCR was performed as triplicate using TaqMan Gene Expression Assay (FAM-MGB, Thermo Fisher) for human LPAR 1-6 , ENPP2, and glyceraldehyde-3-phosphate dehydrogenase and TaqMan Fast advanced MasterMix (ThermoFisherScientific, Waltham, MA), using a QuantStudio 5 real time-PCR system (ThermoFisherScientific; more details in the Supporting Information). The 2 −ΔΔCt method was used to assess the difference in gene expression between LPS-stimulated and unstimulated CD14 + cells; the 2 −ΔCt method was used to assess differential expression between disease groups.

StatIStICal aNalySIS
See the Supporting Information for full details.

IMpRoVeD StRatIFICatIoN oF patIeNtS WItH lIVeR FaIlURe USINg lIpID-optIMIZeD UplC-MS CoMpaReD WItH 1 H NMR aND tHe pRogNoStIC IMpoRtaNCe oF lpCS
To validate, extend, and improve our previous observations, (19) we used a combination of 1 H NMR spectroscopy ( Fig. 1A-I,II) and lipid-optimized UPLC-MS ( Fig. 1A-III, IV and B) on serum samples. Principal component analysis (PCA) revealed that the UPLC-MS lipid-optimized method more accurately captured the phenotype and AD-to-ACLF progression in patients with cirrhosis (Fig. 1A) and is the method used in patients with ACLF. We noted also the prognostic ability of the UPLC-MS method to predict 30-day mortality with an area under the receiver operating characteristic curve (AUROC) for the entire profile of 0.94, P < 0.001 (Fig. 1B,C). A number of LPCs were associated with increased mortality by univariate analysis and the class and prognostic distinction confirmed by cross-validated ANOVA and subsequent false discovery rate correction P values (Supporting Table S3). The predominant lipid subclass associated with this progression was the LPC; a number of LPCs were down-regulated in AD/ACLF and associated with mortality (see Fig. 1E).
Targeted metabonomics analysis using the Biocrates p180 kit in a separate validation sample set confirmed the prognostic ability of both LPC 16.0 and LPC 18.1 to predict 90-day mortality, with an AUROC, respectively, of 0.8529, P < 0.0001 and 0.8223, P = 0.0006, in a population of patients with cirrhosis (see Supporting Fig. S1). LPC concentrations dynamically rose over the first 3 days of admission with ACLF and had reached levels associated with AD by days 7-10 of admission (Supporting Figs. S2 and S3). LPA species were sought within the mass spectrometry data of the first cohort and LPAs identified by mass accuracy and tandem mass spectrometry. Their concentrations mirrored the parent LPC molecule and were negatively correlated across patient classes ( Fig. 2A; Supporting Table S3). Given that the UPLC-MS method was not optimized for LPA quantification, a total LPA ELISA was performed demonstrating similar changes in concentration and correlation with ATX levels (see Supporting Fig. S4).

atX aND pla 2 aRe DyNaMICally Up-RegUlateD IN lIVeR FaIlURe SyNDRoMeS aND aSSoCIateD WItH lpa pRoDUCtIoN
Having identified the importance of the LPC-LPA axis, we next measured relevant regulatory lipases in plasma. We found that ATX levels increased in both stable and acutely decompensated cirrhosis (AD and ACLF) together with the markers of cell death, M65 and M30 ( Fig. 2A-E). Given the higher concentrations in patients with ACLF and AD compared to SC, it is likely that this is acutely up-regulated, and, in fact, ATX concentrations rose over the course of evolving ACLF. We confirmed that this was a liverspecific process given that peripheral ATX concentrations in patients with sepsis are similar to HCs (Supporting Fig. S4).
Conversely, PLA 1 concentrations were not modulated in any of the liver disease presentations; however, those of sPLA 2 in plasma behaved in a similar manner to those of ATX in terms of both acute and chronic disease (Fig. 2B). Moreover, both ATX and PLA 2 were dynamically up-regulated during the early evolution of ACLF.
Moreover, both ATX and total LPA were increased in ACLF grade 3 compared to lower grade, even if not statistically significant, given that most of patients were included in the most severe group. Conversely, LPC 16.0 and LPC 18.1 were significantly reduced in ACLF grade 3 compared to other groups (Supporting Fig. S4).
Although other investigators showed higher ATX levels in women, (23,24) our results show no sex difference in HC (n = 12) and patients with cirrhosis (n = 25); on the contrary, men affected by ACLF (n = 21) have higher ATX concentrations than women, suggesting that severity of illness is a primary driver of ATX up-regulation. Renal dysfunction is frequently associated with ACLF, and other investigators showed increased serum ATX and LPA levels in renal failure. (41,42) In our cohort, creatinine did not correlate with ATX, but total LPA levels increased with worsening renal failure (Spearman's r, 0.348; P < 0.05) whereas LPC 16:0, 18:0 and 18:1 were significantly reduced as creatinine increased (Spearman's r, −0.412, −0.457, and −0.351; P < 0.05). Moreover, ATX was increased in patients with renal failure, requiring renal replacement therapy (Supporting Fig. S4B).
Respiratory dysfunction, expressed as reduced P/F (partial oxygen pressure/fraction of inspired oxygen) ratio, was correlated with a decrease of LPC 16:0 (Spearman's r = 0.4977; P = 0.0255), whereas other variables of the LPC-ATX-LPA axis appeared to not be related with respiratory function (ATX: r = −0.3805, LPA: r = 0.1930, and LPC 18.1: r = 0.4211; P > 0.05). Moreover, patients with respiratory failure, expressed as P/F < 300, showed reduced levels of LPCs compared to patients with normal respiratory function (Supporting Fig. S4B).
In two thirds of cases, ACLF patients had been diagnosed with a concurrent infection during the hospital stay, mainly chest (43.4%), followed by positive blood cultures (21.7%), spontaneous bacterial peritonitis (17.4%), and other causes (17.4%). There was no difference in ATX, LPA, or LPC 18.1 concentrations in patients affected by ACLF with or without a detectable infection. However, LPC 16.0 was found to be reduced in the group of patients with detectable infections (Mann-Whitney U test, P < 0.05; Supporting Fig. S1).

atX and pla 2 expression Within the liver is Up-regulated During aClF and alF
Given the increased peripheral ATX concentrations observed here, we next explored whether hepatic expression was also up-regulated and contributed to this phenomenon. Intrahepatic expression of ATX in publicly available microarray data sets in alcoholassociated disease and hepatitis B-related ALF was mined for ENPP2/LPAR1-6 expression levels as discussed in the Supporting Information. This analysis showed an increased ATX expression in both chronic and acute liver disease.
Using single-stain tissue immunohistochemistry, we found ATX expression in hepatocytes within centrilobular areas of pathological control tissue (Fig. 3B-D). As with the gene data sets, compared to pathological controls, ATX expression was markedly up-regulated in ACLF patients (Fig. 3D). ATX expression was detected in all hepatic zones and was specifically found in hepatocytes (Fig. 3D).
To further evaluate the liver specificity of ATX production, serum samples were taken during liver transplantation from the portal vein, hepatic vein, and peripheral artery in patients with decompensated cirrhosis or ACLF (Supporting Fig. S4E). We defined the transpulmonary gradient as the comparison between hepatic vein and arterial blood levels. ATX was increased in arterial blood, suggesting a role of the pulmonary epithelium in ATX production. The transintestinal gradient, that is, the comparison between arterial and portal levels, shows no change in ATX. Moreover, considering the double hepatic inflow (portal vein 80% and hepatic artery 20%), ATX concentrations have been weighted accordingly and thus we compared the hepatic inflow with outflow (hepatic

Inflammatory Cytokines Correlated With lpC levels in plasma of aClF patients
We wished to delineate the correlation between circulating lipids and their regulatory lipases and peripheral cytokines in patients with ACLF. We demonstrated a classical proinflammatory state in patients with ACLF with higher IL-1β, IL-6, IL-8, and TNFα (Fig. 4A) and intermediate levels of regulatory cytokines (IL-10 and IL12β) compared with AD patients. Correlation analysis demonstrated positive significant correlations between LPA, ATX and IL-6, IL-8 and TNFα, and MELD and CLIF SOFA. Peripheral LPC concentrations were negatively correlated with ATX, severity scores, and proinflammatory cytokines (Fig. 4B).

plaSMa lpC leVelS CoRRelate WItH EX VIVO MoNoCyte pHeNotype IN aClF patIeNtS
To further explore the relationship between circulating lipids and immune phenotype, we correlated circulating LPC concentrations with innate and adaptive cell populations in 36 patients (23 ACLF and 13 with decompensated cirrhosis) with PBMC phenotyping. Total lymphocytes and CD4 + /CD8 + /regulatory T cell subsets did not correlate with liver severity scores or outcome. Whereas total lymphocytes were directly and significantly related to LPC 18:1, no correlation was observed in the T-cell subsets. However, LPC 16.0 and LPC 18.1 were negatively correlated to monocyte MerTK and CD163 expression (Fig. 5A). Neutrophils count correlates inversely with total plasma LPA, but not with LPCs. However, the neutrophil/lymphocyte ratio increased with the decrease of LPCs (Supporting Fig. S8).
Analysis of monocyte subsets showed that MerTK and CD163 expression was increased in ACLF in the classical (CD14 + CD16 − ) and intermediate (CD14 + CD16 + ) populations whereas MerTK, but not CD163, was higher in ACLF in nonclassical monocytes. MerTK and CD163 expression correlated negatively with lipid concentrations in the classical and intermediate populations only (Supporting Fig. S5). Therefore, neutrophil and CD14 high isolation experiments were performed. We chose to further investigate the effects of LPC 16.0 and 18.1, given that they are the most biologically active, with the highest concentration in human plasma, and show clear correlation with immunophenotyping.

lpa DIMINISHeS pRoRegUlatoRy MoNoCyte pHeNotypeS
We then wished to determine the ex vivo effect of lipid incubation on CD14 + monocyte phenotype. Initial experiments suggested that LPA was the dominant lipid compared to LPC in modulating monocyte phenotype (Supporting Fig. S6), predominantly through reduction in MerTK and CD163 expression, in both healthy volunteers and patients with ACLF. This was consistently demonstrated at the 30-µM concentration for both LPA 16:0 and LPA 18:1. However, the effect appeared more pronounced for LPA 16:0. HLA-DR appeared to be up-regulated in patients with decompensated cirrhosis (Supporting Fig. S7B), but not those with ACLF, where it did not affect HLA-DR expression.

lpa tReatMeNt INCReaSeD MoNoCyte tNFα pRoDUCtIoN
Patients with acutely decompensated cirrhosis and ACLF show an increased production of proinflammatory cytokines compared to controls (Fig. 4). Given the reduction in MerTK and CD163 expression with LPA, we hypothesized that LPA would generate increased production of proinflammatory cytokines following incubation. This was indeed observed for MACS sorted CD14 + cells, demonstrating that LPA 16:0 increases TNFα and IL-6 production in HC cells, and this effect was muted, but persistent, in ACLF (Fig. 6). The effect on neutrophils, instead, was not significant, with no difference in IL-6 production and a small reduction in TNFα production of 20% after LPA 16.0 treatment, statistically significant only in patients, but not in HCs (Supporting Fig. S8).

lpa tReatMeNt DID Not aFFeCt MoNoCyte pHagoCytIC CapaCIty
Having demonstrated the importance of LPAs as regulatory molecules for proresolution phenotypes in CD14 + monocytes, we explored whether they also regulated the phagocytic abilities of these cells. Exposure to LPA decreased phagocytic capacity in HCs, but had no observable effect in monocytes from patients with ACLF (Fig. 6).

lpa tReatMeNt ReDUCeS NeUtRopHIl oXIDatIVe BURSt
Neutrophil oxidative burst was increased in patients with DC/ACLF in both resting phase and after E. coli stimulation compared to HCs. LPA 16:0 stimulation reduced E. coli-induced oxidative burst in HCs and similarly in patients, although not statistically significant (Supporting Fig. S8C), with no other effect on neutrophil function observed with LPA.

patHWay aNalySIS SUggeStS aN lpaR2-DepeNDeNt aCtIVatIoN MeCHaNISM IN CD14 + CellS
In order to confirm and extend the mechanistic understanding of lipid-monocyte interaction, we used data accrued up to this point to model potential pathways of interest for further exploration. Clinical, lipidomics, (43) flow cytometry, and cytokine data were entered into Ingenuity Pathway Analysis (IPA), as shown in the Supporting Results. According to these data, the LPAR2 receptor appeared to be the most relevant LPA receptor involved in monocyte activation in ACLF and also seemed to be implicated in modulation of both MerTK and in IL-2, IL-4, IL-10, and TNFα expression (Supporting Fig. S8).

Discussion
This study sought to identify and comprehensively characterize lipid-specific immunometabolic control of innate immune responses in patients with ACLF. We have demonstrated that the perturbation of LPC and LPA is mediated by ATX and inhibits expression of proregulatory monocyte phenotypes. This may explain the early proinflammatory phenotype of ACLF and supports the role of phospholipids as important molecules in the control of functional reprogramming of monocytes which underpin immune paresis and susceptibility to infection in this syndrome. Therapeutic modulation of immune responses by targeting LPA receptors could become an important future clinical strategy.
Previous large-scale metabolomics analysis of plasma in patients with ACLF did not focus on lipids. (15) Although previous investigations (19) relied on 1 H NMR spectroscopy alone or in combination with general reversed-phased liquid chromatography methods, these were not optimized for lipids. Though suggestive of the importance of LPC, these investigations did not specifically target this and other important lipid classes. Here, we used UPLC-MSbased lipidomic methods to determine the identities These data are presented as the proportion of live cells, given that LPA incubation was associated with a 5%-10% decrease in cell viability. (E) RT-PCR gene expression for CD14 + monocytes regarding genes LPAR1-6 and ENPP2 expressed as 2 -∆∆Ct on a logarithmic scale with relation to HC in the HC versus ACLF comparison and to unstimulated in the LPS experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Abbreviation: arb units, arbitrary units; Ct, cycle threshold; FMO, flavin-dependent monooxygenase; UTR, untranslated region.
FIg. 6. LPA treatment increases monocyte TNF-a production, but does not affect monocyte phagocytic capacity. (A,B) TNFα and IL-6 cytokine secretion levels in supernatants of bead-isolated CD14 + cells cultured with lipids with LPS stimulation and with and without incubation with LPA in patients with ACLF (n = 5) and HC (n = 5). (As expected, non-LPS-stimulated CD14 + cells did not produce a cytokine response [data not shown].) (C) LPA decreased phagocytic capacity in HC (n = 7), but had no observable effect in monocytes from patients with ACLF (n = 5). *P < 0.05; **P < 0.01; ***P < 0.001. A B C of the LPCs that are correlated with the presence and grade of ACLF and for mortality prediction. By using a method targeted and optimized for lipids, it is now clear that LPCs are a dominant lipid class for stratifying patients with ACLF and could form the basis of a future prognostic assay.
Whereas previous work focused on the potential mechanism of LPC depletion to be related to hepatocyte apoptosis, our finding in this work is not only that both ATX and PLA 2 appear to be up-regulated in liver failure syndromes, but also the production of the bioactive lipid, LPA, from LPC under the action of ATX has profound sequelae for innate immune function (Fig. 7). Our initial experiments suggested a primary immune-modulatory role of LPA more than LPC. Given that there is abundant evidence of LPA receptors modulating a wide range of cellular function, this is in keeping with our present knowledge of LPAmonocyte interaction, but extends and develops our understanding of this relationship in ACLF. Recent reports suggest that monocytes can be modulated by LPC directly, although whether this is, in part, mediated by LPA (given that monocytes express ATX) is in doubt, particularly given the lack of clear understanding of how LPC interfaces cellular function. (44) In particular, the control of HLA-DR, MerTK, CD163, and CCR2 expression by LPA has crucial relevance for patients with ACLF. Recent interest in immunotherapy in cirrhosis and ACLF has focused on control of these regulatory responses. Given that increased MerTK expression is associated with hypoactive responses to bacterial challenge, (12) MerTK inhibitors may offer therapeutic potential. (45) In peripheral monocytes, LPA appears to dampen proregulatory responses and promotes inflammation (thereby acting as an MerTK inhibitor), albeit without a beneficial effect on phagocytic capacity. Promotion of inflammation in the absence of bacterial killing is a common aspect of ACLF, where persistence or repeated bouts of sepsis are common in those who do not survive.
Increased ATX expression was observed in hepatic tissue, demonstrating the central role of the liver in increasing peripheral activity of this phospholipase and generating LPA. Previous animal models have demonstrated that a number of stimuli promote ATX expression in hepatocytes, and we have demonstrated its increase in the peripheral blood and liver of humans with ACLF. ATX is associated with hepatic fibrosis and has previously been associated with reduced survival. (23) Whereas most evidence in the field of hepatitis C (27) suggests that ATX is upregulated in chronic fibrotic liver diseases, we note that its expression is higher when patients with cirrhosis are developing ACLF. Furthermore, it appears that hepatic expression of ATX is augmented by peripheral monocyte coexpression, but only under LPS stimulation. Its dynamic increase demonstrated in plasma from patients with ACLF over the first 5 days of evolution is also an interesting finding. It is likely that the dynamic up-regulation of LPC pathways during liver failure syndromes is under-recognized and may explain the switch in monocyte phenotype during ACLF development. Respiratory and renal failure, together with the occurrence of sepsis, could be confounding factors. It may be that some alteration of specific LPCs (e.g., LPC 16:0) is driven by sepsis, (46) in addition to hepatic and nonhepatic modulation of the LPC-ATX-LPA axis. Our data demonstrate that monocytes themselves up-regulate ATX/ENPP2 expression following LPS stimulation, which could alter the lipid profile of their immediate microenvironment. Furthermore, conversion of LPC 16:0 to LPA 16:0 only has minor functional significance for neutrophil TNFα production and oxidative burst, although this effect appears secondary to their overall functional importance in monocytes. TNFα monocyte production after LPA 16:0 stimulation increases 9.5-fold in HCs and >3-fold in ACLF, compared to the neutrophil TNFα production that is reduced by 20% only in both HCs and DC/ACLF. This study has a number of limitations. Not all patients were fasted in terms of dietary effects on lipid concentrations, nor was dietary recall or assessment of sarcopenia undertaken. Nevertheless, others have found dietary factors less relevant for nonacyl glyceride lipids and also that the perturbations in patients with critical illness are likely to be far larger than dietary effects. We did not explore the intracellular pathway following activation of the LPA receptor to increase MerTK expression, and this is a priority area of future work. The intrahepatic assessment of ATX expression in the present study is based on a small number of immune-histochemical analyses and RNA expression extracted from publicly available databases, but requires validation in a larger sample set. Moreover, it would be interesting to evaluate the hepatic macrophage phenotype in tandem to the circulating monocytes we studied. Furthermore, HSCs are known to proliferate under the action of LPA. (47) LPA also regulates the expression profiles of cytokines and angiogenic factors in liver sinusoidal endothelial cells, which may prove useful for manipulating LPA effects on liver regeneration (this may be more relevant in ALF rather than ACLF). (48) Assessment of the ATX-LPA axis in a wide set of intrahepatic immune cell populations is an interesting topic for future studies, but is limited by the low numbers of patients who undergo transplantation during ACLF episodes as a source of ACLF tissue. As transplantation becomes more common in ACLF, explant tissue will become more widely available for intrahepatic ATX expression in ACLF.
Genetic deletion models of ATX were not undertaken. This was, in part, because of previous work already characterizing the ATX axis in murine models of fibrosis, but also the lack of an accepted ACLF model. The translational focus of this work was not in the prevention of fibrosis (which is not relevant in ACLF), but in the prevention or amelioration of sepsis or ACLF progression, hence our focus on monocytes. Finally, we did not explore a wide range of LPA lipids quantitatively in plasma and this requires further methodological development.
In summary, we demonstrate that LPC concentrations in plasma correlate closely with development of ACLF and predict mortality. Their metabolism is regulated by ATX, producing LPA, which has a highly active and relevant role in the regulatory phenotype and function of monocytes. This pathway is now open to future therapeutic exploration to modulate the risk of sepsis in this important clinical condition. Proposed mechanism of LPC-ATX-LPA axis' dysregulation and its effects on peripheral monocytes in ACLF. Gut and other derived moieties promote liver cell death, inflammation, and ATX up-regulation in liver and lung. This results in conversion of LPC to LPA in a manner dependent on disease severity while this process continues in the periphery. Monocytes respond to increased concentrations of LPA through a potential LPAR2-dependent mechanism and modulate their phenotype to express less MerTK and CD163. In conjunction with this, monocytes produce more IL-6 and TNFα, thereby perpetuating systemic inflammation. (*Nevertheless, ex vivo ACLF monocytes are less effective to produce inflammatory cytokines after LPS stimulation.) Crucially, monocytes' phagocytic ability is not altered and so this systemic inflammation lacks key antimicrobial properties, and thus a state of persistent inflammation without enhanced bacterial clearance is engendered. Abbreviations: DAMPs, damage-associated molecular patterns; PAMPs, pathogenassociated molecular patterns.