TREM2 Regulates the Removal of Apoptotic Cells and Inflammatory Processes during the Progression of NAFLD

Nonalcoholic fatty liver disease (NAFLD) is the most common liver pathology worldwide. In mice and humans, NAFLD progression is characterized by the appearance of TREM2-expressing macrophages in the liver. However, their mechanistic contributions to disease progression have not been completely elucidated. Here, we show that TREM2+ macrophages prevent the generation of a pro-inflammatory response elicited by LPS-laden lipoproteins in vitro. Further, Trem2 expression regulates bone-marrow-derived macrophages (BMDMs) and Kupffer cell capacity to phagocyte apoptotic cells in vitro, which is dependent on CD14 activation. In line with this, loss of Trem2 resulted in an increased pro-inflammatory response, which ultimately aggravated liver fibrosis in murine models of NAFLD. Similarly, in a human NAFLD cohort, plasma levels of TREM2 were increased and hepatic TREM2 expression was correlated with higher levels of liver triglycerides and the acquisition of a fibrotic gene signature. Altogether, our results suggest that TREM2+ macrophages have a protective function during the progression of NAFLD, as they are involved in the processing of pro-inflammatory lipoproteins and phagocytosis of apoptotic cells and, thereby, are critical contributors for the re-establishment of liver homeostasis.


Introduction
NAFLDs, or, as recently proposed, "metabolic associated fatty liver diseases" (MAFLD) [1], encompass a broad spectrum of liver pathologies, ranging from simple steatosis to nonalcoholic steatohepatitis (NASH), potentially resulting in liver fibrosis and cirrhosis. NAFLD is common in obese and diabetic people and roughly 25% of the global population is affected by NAFLD [2]. However, up to now, there is no approved drug for the treatment of NAFLD. One important feature of hepatocytes is their ability to efficiently export excess triglycerides and other lipids as triglyceride-rich very-low-density lipoproteins (VLDLs). Thus, an impairment in VLDL secretion, induced by choline-deficient diets or depletion of proteins pivotal for VLDL production, such as apolipoprotein E (APOE), causes liver steatosis [3],

Animal Experiments
All animal experiments were approved by the Animal Welfare Officers of University Medical Center Hamburg-Eppendorf (UKE) and Behörde für Gesundheit und Verbraucherschutz Hamburg (G15/66). Mice were bred in-house at UKE (Hamburg, Germany) and housed at 22 • C with a day and night cycle of 12 h. Animals had ad libitum access to food and water. For cell-isolation studies, male C57BL/6J wild-type (WT) Trem2 +/+ -mice or Trem2 −/− (Trem2 tm1(KOMP)Vlcg /MbpMmucd) mice were sacrificed at age of 8-12 weeks. To deplete KC, mice were injected intravenously via the tail vein with 200 µL clodronate liposome solution (ClodronateLiposomes.org, Amsterdam, The Netherlands) two days before the experiments, as described before [16]. Triglyceride-rich lipoproteins (TRLs) were isolated from healthy and hyperlipidemic individuals; as described in, 2.15. 100 µg of TRL (10 mg protein/mL) was injected intravenously via the tail vein.

Human Data
For correlative analysis, log-transformed hepatic gene expression of TREM2 (log TREM2) and COL1A1 (log COL1A1) data as well as metabolic parameters from patients of different body-mass index (BMI) were included in a biobank study conducted at the Department of General and Visceral Surgery, University of Ulm, Germany from 2002 to 2010, previously described in detail [17]. Briefly, people (n = 120) undergoing bariatric surgery (n = 57), abdominal surgery for early colorectal cancer (n = 34), pancreatic cancer (n = 18), other cancers of the gastrointestinal tract or benign diseases, such as sigma-diverticulosis, were included in the study. Exclusion criteria were type 1 diabetes and acute inflammatory disease. None of the cancer patients exhibited signs of metastases or reported excessive, unintended recent weight loss. Liver biopsies and blood samples were immediately placed on ice and subsequently snap frozen and stored in liquid nitrogen or at −80 • C.
Plasma TREM2 levels were measured after overnight fasting in patients with biopsyproven NAFLD and liver-healthy control subjects, who were characterized by normal liver function tests, low liver elastography and controlled attenuation parameter measurements (FibroScan; Echosens, Paris, France). The study was approved by the local ethics committee in Hamburg, Germany (PV5036), and complied with the 1975 Declaration of Helsinki. Written, informed consent was obtained from each subject.

Bone-Marrow-Derived Macrophage Differentiation
For the differentiation of bone-marrow-derived macrophages (BMDMs), hematopoietic pluripotent stem cells were isolated from bone marrows of Trem +/+ or Trem −/− mice and differentiated in cRPMI (10% FCS, 15% L929 supernatant, 200 mM L-Glu and 0.5% Gentamycin). On day 3, half of the medium was replaced, while on day 5, cells were split and seeded into fresh media (10 mL). Fully differentiated macrophages were harvested on day 7, seeded into a 24-well plate (0.33 × 10 6 cells/well) and used for experiments.
BMDMs were then stimulated with LPS (100 ng/mL) for 4 h (to measure gene expression) or for 18 h (to determine cytokine secretion).

Phagocytosis Assays
Phagocytosis assay was performed using primary liver macrophages. Therefore, liver cells were isolated, as described in 2.4. Without lysating the red blood cells, the nonparenchymal fraction was diluted in 7-10 mL of pre-warmed DMEM (5% FCS, 200 mM L-Glu and 0.5% Gentamycin). Afterward, cells were plated in Cell+ 12-well plates (Sarstedt, Nümbrecht, Germany) overnight in 1.5 mL media per well. On the next day, cells were gently washed to remove non-adherent cells, examined for their morphological phenotype under a microscope and supplemented with fresh warm DMEM. After 24 h, liver macrophages were used for further analysis. First, apoptosis was induced in hepatocytes (Hepa 1-6 cell line) via heating. Briefly, hepatocytes were washed and harvested in PBS with a cell scraper. Subsequently, cells were incubated at 40 • C for 45 min while shaking. Thereafter, apoptosis was verified via Annexin V and Prodium Iodide (PI) staining, using the Apoptotic Detection Kit (BioLegend, London, UK) The apoptotic hepatocytes (aHs) were counted and stained with CellTrace™CSFE (Thermo Fisher Scientific, Massachusetts, USA) in a final concentration of 10 × 10 6 cells/mL, before performing the phagocytosis assay at 37 • C and at 4 • C or in the presence of Cytochalasin D (Cyt.D). The plated BMDMs or KCs were incubated with 0.66 × 10 6 aHs in a ratio of 1:2 for 45 min. Afterward, cells were washed 5× with PBS to remove unbound/not-phagocytosed aHs and stained with F4/80 (AF700) and CD11b (APC-Cy7). The phagocytosis rate was examined at the LSR II (BD Biosciences). The ratio between macrophages that are positive for the aH-dye (CSFE) at 37 • C and 4 • C/Cyt.D defines the phagocytic rate of the macrophages.

CD14 Assay
In some experiments, the phagocytosis assay was performed in the presence of anti-CD14 or the IgG control. KCs (isolated as described in 2.4) were incubated with 10 µg of the anti-CD14 antibody (BD Biosciences, Heidelberg, Germany) or the IgG control (BD Biosciences, Heidelberg, Germany) for 2 h. Subsequently, cells were stimulated with 10 ng/mL of LPS for 20 h. The phagocytosis assay was performed afterward, as described previously.

Liver Immunophenotyping
For the immunophenotyping, the liver was perfused manually with 10 mL of PBS. The liver was intersected and the gallbladder removed before mincing the liver. The liver suspension was incubated for 45 min at 37 • C while shaking in digestion buffer (DMEM, 1 mg/mL Collagenase IV (StemCell, Cologne, Germany), 150 U/mL DNaseI, 0.2 M MgCl 2 , 0.5 M CaCl 2 ). Subsequently, the cell suspension was filtered through a 70 µm filter and washed 3 times with PBS (300× g 5 min, 4 • C). To remove the hepatocytes and to obtain the non-parenchymal fraction, the cell suspension was centrifuged at 50xg for 4 min. The pellet was discarded and the supernatant harvested. The cells were pelleted and red blood cells lysated (3 mL, 5 min, RT). Cells were then washed, incubated with Fc block (BioLegend, London UK, anti-CD16/CD32 1:100,000) for 15 min at 4 • C prior to the staining of surface epitopes. For the staining of surface epitopes, the cells were incubated with the antibody cocktail for 35 min at 4 • C in the dark. For the analysis of intracellular epitopes, cells were fixed and permeabilized using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience, Massachusetts, USA) and stained afterward for 35 min at 4 • C in the dark. Cells were analyzed at the LSRII (BD Biosciences, Heidelberg, Germany). The data were analyzed using the FlowJo software (BD).

Apoptosis Detection
For the detection of Annexin V + apoptotic cells in the liver, cells were isolated as described above and stained with an antibody cocktail for labelling surface epitopes. Afterward, cells were washed twice with PBS + 2% FCS and then with Annexin V binding buffer. Thus, 200 µL of Annexin V binding buffer was added to the cells, followed by 2 µL

Plasma Analysis of Alanine Aminotransferase and LPS
Liver transaminase plasma activity was assessed by automated measurement of alanine aminotransferase (ALT) using a COBAS Mira System (Roche, Mannheim, Germany). LPS in serum and isolated TRLs was assessed using the LAL-Kit (Hycultech, Beutelsbach, Germany).

ELISA
Cytokine secretion in cell supernatants was determined using commercial kits ( TREM2 concentrations in human plasma samples were determined using DuoSet ® ELISA for Human TREM2 from R&D Systems (DY1828-05) according to manufacturer instructions.

Liver Histology
Organ samples for histology were collected in 3.7% formaldehyde solution in PBS. Organs were embedded in paraffin and slides were stained with hematoxylin and eosin (H&E), Sirius Red staining or F4/80 Ab (AbD Serotec, MCA497) staining via immunohistochemistry, according to standard protocols.

Liver Triglyceride Measurements
Liver tissue was homogenized in PBS (10 µL/mg). Liver triglycerides were determined directly from the suspension by using a commercial kit from Roche according to the manufacturer's instructions.

Isolation of Triglyceride-Rich Lipoprotein (TRLs) from Plasma
TRLs were isolated from plasma via density gradient ultracentrifugation. Blood plasma from patients was mixed with a 60% sucrose solution to yield an end concentration of 10% sucrose. Polyallomer tubes (for SW41 rotors, Beckman) were filled with NaCl (ad 12 mL (total volume of tubes)) and the plasma-sucrose mix was carefully added underneath. Centrifugation was carried out at 4 • C and 28.000 rpm for 3 h. TRLs were then taken from the upper phase.

Statistics
Data are expressed as mean ± S.E.M. As stated in the figure legends, depending on the experimental setup, statistically significant differences between experimental groups were assessed using Mann-Whitney test, student's t test and two-way ANOVA. Association of hepatic TREM2 expression with liver parameters was performed using Pearson's correlation coefficient. All statistical analyses were performed using GraphPad Prism 9 and statistically significant differences are depicted by p-values or as different letters.

Lipoproteins Evoke a Macrophage-Dependent Pro-Inflammatory Response
Adipose tissue as well as gut-derived signals are implicated in the progression of NAFLD [5]. In fact, it has been shown that the severity of NAFLD is associated with alterations in the intestinal microflora and even dysbiosis [18]. In line, higher endotoxin levels have been reported in patients suffering from NASH [19,20]. Lipopolysaccharide (LPS), the major endotoxin, induces a pro-inflammatory response via its receptor Toll-like receptor 4 (TLR4) [21]. Interestingly, mice with impaired TLR4 signaling show reduced NASH after high-fat diet (HFD) feeding and TLR4 protein levels are altered in patients with NASH, suggesting an important role for LPS in the progression of NASH. Previously, it was reported that high-fat meals increase plasma LPS levels [22][23][24] and that in mice, LPS may be transported in the circulation via Triglyceride-rich lipoproteins (TRLs) [25]. Indeed, when we measured LPS levels in serum and in isolated TRL particles from human healthy controls and obese patients, we detected substantial higher amounts of LPS in the TRL particles compared to the serum. Of note, TRL from obese patients had a higher LPS load compared to controls (Figure 1a), indicating that, during obesity, LPS-containing TRL particles may contribute to systemic and, most likely, hepatic inflammation. The uptake of lipoproteins into macrophages has been studied extensively in the context of atherosclerosis, where macrophages in the intima take up oxidized low-density lipoprotein (LDL) particles, become foam cells and promote atherogenesis [26]. However, recently, it was shown that TRL can also be internalized by BMDMs [27] and, more importantly, also by KC in the liver [28]. We hypothesized that during the progression of NASH, LPS-enriched TRLs are taken up by hepatic macrophages, thereby driving a pro-inflammatory response in the liver. To address this hypothesis, we injected TRL particles isolated from lean and obese/hyperlipidemic individuals into wild-type mice. TRL particles elicited an acute pro-inflammatory response in the liver, as indicated by higher hepatic expression of Tnfa, Il1b and Il6 (Figure 1b-d). Most importantly, the generation of a pro-inflammatory environment was mainly driven by hepatic macrophages, as in mice pre-treated with clodronate liposomes, the depletion of phagocytic cells led to a reduced pro-inflammatory response (Figure 1b-d). Of note, following both TRL injection as well as clodronate treatment, hepatic gene expression of Trem2 was reduced ( Figure 1e). Overall, these data suggest that in the context of obesity and NAFLD, hepatic macrophages induce a pro-inflammatory response via sensing LPS-containing TRL particles.

Trem2 Controls TRL-Induced Pro-Inflammatory Response
To investigate how Trem2 regulates the generation of a pro-inflammatory environment upon liver exposure to lipoproteins, we first injected WT (Trem2 +/+ ) and Trem2 −/− mice with TRL particles and evaluated the activation status of the liver. Upon TRL injection, hepatic protein levels of the phosphorylated and active form of the stress kinases JNK1 and JNK2, markers of liver fibrosis [29], were elevated in Trem2 −/− mice compared to WT mice (Figure 2a,b). In the same line, expression of pro-inflammatory cytokines, such as Tnfa, Il1b and Il6, was higher in Trem2 −/− mice compared to WT mice (Figure 2c). We also confirmed these findings in vitro, where treatment of Trem2 −/− KCs with LPS resulted in higher secretion of the pro-inflammatory cytokines TNFα, IL1β and IL6 compared to the control counterpart ( Figure 2d). Overall, these data demonstrate that loss of Trem2 results in an exacerbated pro-inflammatory response to LPS-loaded TRLs, hence, indicating a liver protective role for Trem2 during the progression of NASH.
mice. TRL particles elicited an acute pro-inflammatory response in the liver, as indica by higher hepatic expression of Tnfa, Il1b and Il6 (Figure 1b-d). Most importantly, generation of a pro-inflammatory environment was mainly driven by hep macrophages, as in mice pre-treated with clodronate liposomes, the depletion phagocytic cells led to a reduced pro-inflammatory response (Figure 1b-d). Of n following both TRL injection as well as clodronate treatment, hepatic gene expressio Trem2 was reduced ( Figure 1e). Overall, these data suggest that in the context of obe and NAFLD, hepatic macrophages induce a pro-inflammatory response via sensing L containing TRL particles.

Trem2 Controls TRL-Induced Pro-Inflammatory Response
To investigate how Trem2 regulates the generation of a pro-inflammat environment upon liver exposure to lipoproteins, we first injected WT (Trem2 +/+ ) Trem2 −/− mice with TRL particles and evaluated the activation status of the liver. U TRL injection, hepatic protein levels of the phosphorylated and active form of the st kinases JNK1 and JNK2, markers of liver fibrosis [29], were elevated in Trem2 −/− m compared to WT mice (Figure 2a-b). In the same line, expression of pro-inflammat cytokines, such as Tnfa, Il1b and Il6, was higher in Trem2 −/− mice compared to WT m (Figure 2c). We also confirmed these findings in vitro, where treatment of Trem2 −/− K with LPS resulted in higher secretion of the pro-inflammatory cytokines TNFα, IL1β IL6 compared to the control counterpart ( Figure 2d). Overall, these data demonstrate loss of Trem2 results in an exacerbated pro-inflammatory response to LPS-loaded TR hence, indicating a liver protective role for Trem2 during the progression of NASH.
Next, we investigated the mechanisms of how Trem2 expression might regulate inflammatory response in the liver. As TREM2 + macrophages are well known as L [12], which arise in the liver upon HFD feeding [9,10] and have been involved in li handling [15], we first analyzed the expression of genes involved in lipid and lipopro uptake and processing [30] in WT and Trem2 −/− BMDMs. Interestingly, Trem2 −/− BMD showed reduced mRNA levels of lipoprotein lipase (Lpl) and the cluster of differentiat 36 (Cd36), a scavenging receptor involved in fatty acid uptake (Figure 2e). Simila expression of Lipa encoding lysosomal acid lipase was decreased in Trem2 −/− BMD compared to the control counterpart (Figure 2e), suggesting the contribution of Trem2 lipoprotein and fatty acid uptake subsequent processing. Next, we studied whether   Together, these data suggest that in response to TRLs, loss of Trem2 leads to the generation of a pro-inflammatory environment, without affecting the clearance of these particles by hepatic macrophages. Next, we investigated the mechanisms of how Trem2 expression might regulate the inflammatory response in the liver. As TREM2 + macrophages are well known as LAM [12], which arise in the liver upon HFD feeding [9,10] and have been involved in lipid handling [15], we first analyzed the expression of genes involved in lipid and lipoprotein uptake and processing [30] in WT and Trem2 −/− BMDMs. Interestingly, Trem2 −/− BMDMs showed reduced mRNA levels of lipoprotein lipase (Lpl) and the cluster of differentiation 36 (Cd36), a scavenging receptor involved in fatty acid uptake (Figure 2e). Similarly, expression of Lipa encoding lysosomal acid lipase was decreased in Trem2 −/− BMDMs compared to the control counterpart (Figure 2e), suggesting the contribution of Trem2 for lipoprotein and fatty acid uptake subsequent processing. Next, we studied whether the loss of Trem2 affected KC-dependent uptake of TRL particles. For this purpose, we injected WT (Trem2 +/+ ) and Trem2 −/− mice with DiD-labeled TRL particles and quantified their uptake into KCs (gated as CD45 + LY6G − CD11b + LY6C − TIM4 + ) and infiltrating macrophages (gated as CD45 + LY6G − CD11b + LY6C + ) via flow cytometry. Interestingly, in this setting, Trem2 −/− macrophages showed a similar capacity to uptake TRL particles as the control counterpart ( Figure 2f).

Trem2 Regulates KC-Dependent Uptake of Apoptotic Cells
Together, these data suggest that in response to TRLs, loss of Trem2 leads to the generation of a pro-inflammatory environment, without affecting the clearance of these particles by hepatic macrophages.

Trem2 Regulates KC-Dependent Uptake of Apoptotic Cells
Accumulation of apoptotic cells due to impairment in their clearance by efferocytic macrophages is one of the processes leading to the generation of a pro-inflammatory environment. Trem2 expression by microglial cells has been described to be implicated in the phagocytosis of apoptotic neurons [31]. However, whether TREM2 engagement regulates efferocytosis in the damaged liver has not been completely dissected so far. Thus, we isolated KCs from the livers of WT (Trem2 +/+ ) and Trem2 −/− mice and exposed them to aHs. Similarly, BMDMs, which resemble the cells infiltrating the liver upon damage, were incubated with dying cells, and their phagocytic capacity was assessed. In both scenarios, Trem2 regulated the capacity of resident and infiltrating phagocytes to clear aHs (Figure 3a,b). These data highlight the critical role of TREM2 as a phagocytic receptor in bone-marrow-derived and hepatic macrophages.
The synergistic interaction of receptors involved in phagocytosis has been described as a strategy to ensure efficient and prompt clearance of pathogens and potentially harmful dying cells in various experimental settings [32,33]. Given the impact of Trem2 on LPSmediated response (Figure 2d) and the role of CD14, an LPS-binding protein, in regulating phagocytosis directly or in cooperation with other receptors [34,35], we tested whether synergy between TREM2 and the LPS co-receptor CD14 [36] occurs and whether it contributes to the regulation of phagocytosis. We thus treated WT (Trem2 +/+ ) and Trem2 −/− hepatic macrophages with anti-CD14 or isotype control (IgG) antibodies, prior to their stimulation with LPS. Afterwards, these LPS-activated macrophages were exposed to apoptotic cells and their phagocytic capacity was evaluated. Similar to what is described in human monocyte-derived macrophages [34], treatment of WT KCs with anti-CD14 diminished their capacity to phagocyte apoptotic cells. In contrast, no further reduction in the uptake of apoptotic cells was observed in Trem2 −/− KCs pre-treated with anti-CD14 antibody compared to IgG (Figures 3c and S1). These data suggest that TREM2 and CD14 act in concert to mediate LPS-induced phagocytosis. The synergistic interaction of receptors involved in phagocytosis has been described as a strategy to ensure efficient and prompt clearance of pathogens and potentially harmful dying cells in various experimental settings [32,33]. Given the impact of Trem2 on LPS-mediated response (Figure 2d) and the role of CD14, an LPS-binding protein, in regulating phagocytosis directly or in cooperation with other receptors [34,35], we tested whether synergy between TREM2 and the LPS co-receptor CD14 [36] occurs and whether it contributes to the regulation of phagocytosis. We thus treated WT (Trem2 +/+ ) and Trem2 −/− hepatic macrophages with anti-CD14 or isotype control (IgG) antibodies, prior to their stimulation with LPS. Afterwards, these LPS-activated macrophages were exposed to apoptotic cells and their phagocytic capacity was evaluated. Similar to what is described in human monocyte-derived macrophages [34], treatment of WT KCs with anti-CD14 diminished their capacity to phagocyte apoptotic cells. In contrast, no further reduction in the uptake of apoptotic cells was observed in Trem2 −/− KCs pre-treated with anti-CD14 antibody compared to IgG (Figures 3c and S1). These data suggest that TREM2 and CD14 act in concert to mediate LPS-induced phagocytosis. (c) Hepatic macrophages were isolated from WT (Trem2 +/+ ) and Trem2 −/− mice, pre-treated with control (IgG) or anti-CD14 antibody (aCD14), stimulated with LPS for 20 h and incubated with labeled aHs. Phagocyting macrophages were quantified (n = 8). Data are shown as mean + SEM. Statistically significant differences between groups (p < 0.05) were determined by Mann-Whitney test (a,b) and Student's t-test (c) and are indicated as p-value.

Loss of Trem2 Aggravates Liver Damage in a Murine Model of NASH
As our data suggest an important role for Trem2 in the KC-mediated clearance of dying cells as well as for the development of a pro-inflammatory response, we next aimed to study the consequences of Trem2 deficiency during the progression of NAFLD. In a first step, we evaluated the levels of apoptotic-cell-associated proteins in livers derived from WT (Trem2 +/+ ) and Trem2 −/− mice in steady-state conditions. In line with the reduced phagocytic capacity of Trem2 −/− KC, a higher amount of cleaved CASPASE 3 protein and a higher gene expression of pro-apoptotic Bax were detected in mice lacking Trem2 compared to controls (Figure 4a-c). Next, to assess the hepatic accumulation of apoptotic cells during the progression of NAFLD, we fed WT (Trem2 +/+ ) and Trem2 −/− mice with an HFD for 2 weeks. In this experimental setting, the amount of apoptotic cells, as measured by Annexin V staining (AV + cells), remained similar in WT vs. Trem2 −/− mice (Figure 4d), thus suggesting that compensatory mechanisms leading to the uptake of dying cells occur in the absence of Trem2 during a short-term HFD. Additionally, although no statistically significant differences were observed in the frequency of macrophages expressing ARG1 or iNOS, prototypical anti-and pro-inflammatory markers respectively, isolated from the liver of HFD-fed WT (Trem2 +/+ ) and Trem2 −/− mice, a reduction in the percentage of a "metabolically active" CD206 + macrophage population [37] was detected in Trem2 −/− mice. In parallel, reduced expression of the costimulatory molecule CD80 and increased expression of the anti-inflammatory but also pro-fibrotic protein YM1 was detected in Trem2 −/− macrophages compared to controls (Figure 4e).  Additionally, we analyzed the role of TREM2 in long-term models of NALFD induced either by an HFD, which causes accumulation of F4/80 + macrophages in the damaged liver ( Figure S2), or by the combination of Apoe deletion with HFD. Here, we observed that mice lacking Trem2 had higher body and liver weight (Figure 4f,g), which is in line with increased body-weight gain described previously [12]. Further, when fed a HFD, Trem2 −/− mice had increased liver triglyceride levels (Figure 4h), suggesting hepatic steatosis as also visible in H&E-stained liver sections (Figure 4i). In line with this, plasma ALT levels, as a measure of liver damage, were increased by trend in Trem2 −/− mice (Figure 4j). Similarly, hepatic expression of pro-fibrotic genes Tgfb1 and Col1a1 was increased in Trem2 −/− mice compared to WT (Trem2 +/+ ) (Figure 4k,l) and higher hepatic fibrosis was also revealed by Sirius Red staining (Figure 4m), which is in line with previous studies [15]. In sum, these data suggest that loss of Trem2 results in an altered immune response, most likely contributing to higher liver fibrosis during the progression of NAFLD.

Trem2 in Human NAFLD
As we and others have shown an important regulatory role for Trem2 and TREM2 + macrophages in the progression of NAFLD in mice [15], we next wanted to translate our findings into humans. For this purpose, we compared levels of TREM2 in plasma from healthy controls as well as NALFD patients. In accordance with our findings, plasma levels of TREM2 were elevated in NALFD patients (Figure 5a). In a separate cohort, we correlated hepatic gene expression levels of TREM2 with plasma ALT levels, as a measure of liver damage (Figure 5b), hepatic gene expression levels of COL1A1, as a liver fibrosis marker (Figure 5c), and with liver triglyceride (TG) levels, as a marker for hepatic steatosis (Figure 5d). Hepatic gene expression of TREM2 correlated with all the investigated features of NAFLD, overall suggesting that also in humans, TREM2 might be implicated in the development of NAFLD and its progression into NASH. liver sections (scale bar = 200 µm). Data are shown as mean + SEM. Statistically significant differences between groups (p < 0.05) were determined by Student's t-test (b,c,f-h,j-l) and Mann-Whitney test (d,e) and are indicated as p-value.
Additionally, we analyzed the role of TREM2 in long-term models of NALFD induced either by an HFD, which causes accumulation of F4/80 + macrophages in the damaged liver ( Figure S2), or by the combination of Apoe deletion with HFD. Here, we observed that mice lacking Trem2 had higher body and liver weight (Figure 4f-g), which is in line with increased body-weight gain described previously [12]. Further, when fed a HFD, Trem2 −/− mice had increased liver triglyceride levels (Figure 4h), suggesting hepatic steatosis as also visible in H&E-stained liver sections (Figure 4i). In line with this, plasma ALT levels, as a measure of liver damage, were increased by trend in Trem2 −/− mice ( Figure  4j). Similarly, hepatic expression of pro-fibrotic genes Tgfb1 and Col1a1 was increased in Trem2 −/− mice compared to WT (Trem2 +/+ ) (Figure 4k-l) and higher hepatic fibrosis was also revealed by Sirius Red staining (Figure 4m), which is in line with previous studies [15]. In sum, these data suggest that loss of Trem2 results in an altered immune response, most likely contributing to higher liver fibrosis during the progression of NAFLD.

Trem2 in Human NAFLD
As we and others have shown an important regulatory role for Trem2 and TREM2 + macrophages in the progression of NAFLD in mice [15], we next wanted to translate our findings into humans. For this purpose, we compared levels of TREM2 in plasma from healthy controls as well as NALFD patients. In accordance with our findings, plasma levels of TREM2 were elevated in NALFD patients (Figure 5a). In a separate cohort, we correlated hepatic gene expression levels of TREM2 with plasma ALT levels, as a measure of liver damage (Figure 5b), hepatic gene expression levels of COL1A1, as a liver fibrosis marker (Figure 5c), and with liver triglyceride (TG) levels, as a marker for hepatic steatosis (Figure 5d). Hepatic gene expression of TREM2 correlated with all the investigated features of NAFLD, overall suggesting that also in humans, TREM2 might be implicated in the development of NAFLD and its progression into NASH.

Discussion
While the origin and transcriptional profile of macrophage populations emerging in the liver upon NAFLD have been investigated thoroughly [7][8][9][10] and resulted in the identification of a TREM2 + subset, which is either termed LAM- [9,12] or NASH-associated macrophages (NAMs) [13] and Scar-associated macrophages (SAMs) [14], respectively, little is known about the mechanisms promoting the protective role of these TREM2 +

Discussion
While the origin and transcriptional profile of macrophage populations emerging in the liver upon NAFLD have been investigated thoroughly [7][8][9][10] and resulted in the identification of a TREM2 + subset, which is either termed LAM- [9,12] or NASH-associated macrophages (NAMs) [13] and Scar-associated macrophages (SAMs) [14], respectively, little is known about the mechanisms promoting the protective role of these TREM2 + macrophages. Only very recently, Hendrikx and colleagues were able to show that TREM2 + macrophages are implicated in hepatic fibrosis and that loss of Trem2 aggravated murine MCD-induced NASH [15]. The here-presented data provide evidence that, mechanistically, during the progression of NASH, Trem2 expression, in addition to regulating the pro-inflammatory responses to TRL particles by liver macrophages, also affects their capacity to clear apoptotic cells in the damaged liver.
As NASH progression is believed to be linked to gut dysbiosis and gut-derived signals, and especially LPS and its receptor TLR4 [38][39][40], we first investigated if gut-derived TRL particles might be containing LPS in obesity and NAFLD. Indeed, in line with other studies highlighting the role of high-fat-meal-induced endotoxemia [22][23][24], we found increased levels of LPS in TRL particles isolated from obese patients. Of note, upon injection of these particles into mice, they were able to induce a KC-dependent pro-inflammatory response in the liver. Thus, the lipoprotein-mediated delivery of bacterial LPS might be a novel mechanism of how gut-derived signals promote liver inflammatory processes. Indeed, in other tissues, the uptake of low-density lipoproteins by macrophages has been shown to modulate inflammatory responses and contribute to overall disease progression [26] and, only very recently, it has also been shown that the uptake of TRL particles by BMDMs might influence pro-inflammatory gene expression [27].
Next, we were able to show that taming the inflammatory response upon injection of LPS-containing TRLs was attenuated by TREM2, as mice lacking Trem2 showed higher expression and secretion of pro-inflammatory cytokines.
Preventing the generation of a pro-inflammatory environment, on the one hand, may be explained by a direct effect of LPS on TREM2 [41], which, in turn, signals through the adaptor protein DAP12 (TYROBP) to restrain pro-inflammatory cytokine production [42]. In parallel to this, in our experimental settings, TREM2 expressed by hepatic macrophages may bind to TRLs and inhibit the secretion of pro-inflammatory cytokines, such as IL-6. This binding may be similar to what has been described in the context of Alzheimer's disease, where TREM2 in microglia binds to amyloid-β-lipoprotein complexes via APOE, thereby preventing amyloid-β accumulation [43].
On the other hand, an anti-inflammatory effect of TREM2 can be linked to TREM2 + -BMDMs and KC capacity to clear dying hepatocytes in the liver. The phagocytic process occurring upon engagement of TREM2, similar to what is described for the uptake of apoptotic neurons [31,44] and bacterial products [45], leads to the elimination of the pathogen and of potentially harmful dying cells. This clearance is associated with a reduced proinflammatory response [46] and with the induction of a tissue remodeling program essential for the re-establishment of homeostasis [47]. In line with these findings, we and others provide indications for higher apoptosis or decreased macrophage phagocytic capacity in the livers of mice lacking Trem2 [15,48], factors which drive the progression of NASH.
Moreover, the deletion of Trem2 resulted in increased weight gain, liver damage and hepatic fibrosis in two different models of NAFLD. This is in line with a previous work reporting that loss of Trem2 causes an exacerbated steatohepatitis, cell death and fibrosis in an MCD-induced NASH model [15]. Similarly, in the context of other liver pathologies, such as acute liver injury, primary sclerosing cholangitis and hepatocellular carcinoma, TREM2 has been shown to exert a protective function [49][50][51]. Hence, there is an urgent need for future studies to clarify if and how TREM2 + macrophages might be used therapeutically to prevent the progression of NAFLD and other liver pathologies. In this scenario, a deeper understanding of how the cooperation between TREM2 and CD14 occurs, as also confirmed in a recent preprint [52], might help to design better future therapies with TREM2-expressing macrophages. In particular, whether a physical interaction between the two receptors occurs and its meaning in physiological conditions still needs to be evaluated.
Finally, in addition to generating mechanistic insights into the protective function of TREM2 + macrophages in mice, we also provide human data on TREM2 in the context of NALFD. We confirm previously published data [15] on increased levels of soluble TREM2 in the plasma of human NASH patients and show a strong positive correlation of hepatic TREM2 expression with markers of liver damage. Whether higher TREM2 might arise due to the cleavage of TREM2 by ADAM17 [53] or is the result of an increased amount of circulating TREM2 + macrophages, remains to be clarified. Additionally, whether this potentially soluble TREM2 exerts a protective function during NASH has to be addressed in further studies. Nevertheless, based on our and previous findings [15], plasma levels of TREM2 seem to be a valid marker for human NAFLD, likely correlated with disease severity. Plasma TREM2 should be validated in the future as an easy non-invasive biomarker to assess NAFLD severity.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cells12030341/s1, Figure S1: Synergy between TREM2 and CD14 regulates phagocytosis of apoptotic hepatocytes by liver macrophages. Figure S2: Hepatic macrophage infiltration after HFD feeding.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study will be made openly available via MendeleyXXXX.