Introduction

Liver steatosis, defined as lipid deposition in hepatocytes exceeding 5 to 10% of the liver mass, is a common phenotype found in a group of liver complications collectively known as nonalcoholic fatty liver disease (NAFLD) [44]. NAFLD comprises simple steatosis (NAFL) to the more severe diseases steatohepatitis (NASH), cirrhosis, and hepatocellular carcinoma (HCC) [44]. NAFLD affects between 25 and 45% of people worldwide, and its incidence and prevalence are globally rising owing to the increasing rates of obesity and type 2 diabetes [44]. Noteworthy, NAFLD can be also found in lean patients with a body mass index lower than 25 due to enhanced visceral fat accumulation or lipodystrophy, among other reasons [1].

Lipid accumulation in hepatocytes, the primary triggering event in NAFLD, occurs when rates of fatty acid uptake and de novo synthesis exceed rates of fatty acid oxidation and export as very low-density lipoprotein (VLDL)-triacylglycerol [46]. Generally, liver lipid accumulation occurs as the result of impaired adipose tissue lipid storage and/or excessive lipolysis and, therefore, enhanced lipid flux to the liver, as well as enhanced hepatic de novo fatty acid synthesis (lipogenesis) and insulin resistance [46].

Along with lipid metabolism, NAFLD also impacts liver endocrine function and secretion of hepatokines, hormones that communicate liver metabolic status to the central nervous system and peripheral organs [25]. Tsukushi (TSK), for instance, a hepatokine that is highly secreted in patients with steatosis and acetaminophen-injured liver was shown to regulate cholesterol homeostasis and conversion to bile acids [38]. Whether TSK also regulates glucose homeostasis, brown fat thermogenesis, and energy expenditure is still a matter of debate [37, 55]. Furthermore, liver production and secretion of fibroblast growth factor 21 (FGF-21), a member of the FGF-19 subfamily, are increased in several mouse models of NASH and HCC induced either chemically (diethylnitrosamine) or genetically (hepatocyte deletions of Pten, LKB1, p53, etc.) [57], as well as in patients with either NAFL or NASH [14], reflecting liver steatosis score and strongly predicting hepatic steatosis stage [58]. The mechanisms driving liver FGF-21 secretion upon NAFLD were not completely defined yet, but they may involve activation of the transcription factors p53, signal transducer and activator of transcription 3 (STAT3), carbohydrate-responsive element-binding protein (ChREBP), and peroxisome proliferator-activated receptor α (PPARα) in response to stress, cytokines, sucrose, and lipids [17, 2, 57], among other factors. FGF-21 has important actions on energy balance and thermoregulation, reducing body weight and NAFLD and increasing energy expenditure and body temperature, such effects that are in part due to an upregulation of brown and white adipose tissue (BAT and WAT, respectively), uncoupling protein 1 (UCP-1) content and mediated thermogenesis [59, 56, 12]. Higher hepatic FGF-21 secretion upon NAFLD may be considered, therefore, a compensatory mechanism to counteract liver lipid deposition and steatosis.

Considering FGF-21 robust pro-thermogenic actions and enhanced secretion upon HCC, as well as the previous speculation that UCP-1-mediated thermogenesis may promote cancer-associated cachexia [50, 19, 29], we tested herein the hypothesis that an enhanced BAT and subcutaneous inguinal WAT (iWAT) UCP-1-mediated thermogenesis induced by high levels of FGF-21 are involved in the catabolic state and reduction in fat mass associated with HCC. To test this, we evaluated energy balance, serum and tissue FGF-21, BAT and iWAT UCP-1 content, and thermogenic capacity in mice with Pten deletion in hepatocytes, which develop independently of diet, chemical agents, and other external confounding factors that could independently affect BAT function, a well-defined temporal NAFL-NASH-HCC progression due to the constitutive activation of PI3K-mTORC2-Akt-mTORC1 signaling [21, 52]. Noteworthy, mutations or single nucleotide polymorphisms (SNPs) that inactivate Pten and, therefore, constitutively activate PI3K-mTORC2-Akt-mTORC1 signaling are closely associated with the development of steatosis and HCC in humans [3, 11], while whole-body Pten overexpression protects mice from NAFLD [42]. Furthermore, reduced liver Pten expression and protein content are common features found in rodent models of NAFLD (the Zucker diabetic fatty and high-fat-fed Wistar rats) and patients with steatosis [53], as well as human hepatocarcinoma cell lines and liver samples [23]. Finally, homozygous deletion of Pten is a recurrent genetic signature found in HCC [18, 47].

Materials and methods

Mice

Mice experimental procedures were conducted at the Laboratory of Molecular Physiology and Metabolism, Institute of Biomedical Sciences, University of Sao Paulo, after approval by the Animal Care Committee of the Institute of Biomedical Sciences, University of Sao Paulo (#115/2016 and 6160250820, CEUA). All mice used were on a C57BL/6J background. PtenLox/Lox mice (B6.129S4-Ptentm1Hwu/J, Jackson Laboratories) were crossed to albumin-cre mice (B6.Cg-Speer6-ps1Tg(Alb-cre)21Mgn/J, Jackson Laboratories) to obtain heterozygous PtenLox/WT; albumin-cre+/− offspring (where WT refers to wild type) in the F1 generation. These heterozygous mice were crossed with PtenLox/Lox mice to obtain mice with the genotype PtenLox/Lox; albumin cre+/− (referred henceforth as L-Pten KO mice) and their littermates PtenLox/Lox; albumin cre−/− (referred henceforth as L-Pten WT). Mice genotypes were determined by PCR analysis of tail genomic DNA. Mice were kept at 23 ± 1 °C on a 12:12 h light-dark cycle, fed a nonpurified chow diet (70% carbohydrate, 20% protein, 10% fat, in % kcal, NUVILAB CR-1®-Sogorb Inc., Paraná, Brazil) and killed at 8, 24, and 48 weeks of age after 10 h of food deprivation (from 10:00 pm to 08:00 am) by cervical dislocation for tissue and blood harvesting after anesthesia with isoflurane. Body weight and food intake were measured weekly. Subcutaneous inguinal white adipose tissue (iWAT) was evaluated herein because it shows the higher propensity and capacity for browning in mice [54, 13, 48]. BAT, herein, refers to the interscapular depot only.

Indirect calorimetry

Mice were adapted to the metabolic cages for 2 consecutive days and evaluated for oxygen consumption (VO2), carbon dioxide production (VCO2), spontaneous motor activity, and respiratory exchange ratio (RER, VCO2/VO2 ratio) during 24 h in a comprehensive laboratory monitoring system calorimeter (Columbus Instruments) as previously described [40]. The relationship between oxygen consumption and body mass in 8-, 24-, and 48-week-old mice was evaluated using ANCOVA essentially as described [26, 27].

Tissue oxygen consumption

Explants from the liver, iWAT, and BAT were incubated in DMEM containing 2% BSA (w/v) under continuous stirring at 37 °C and evaluated for oxygen consumption in an Oroboros Oxygraph-2 K as described [34].

Mice oxygen consumption and body temperature in response to CL-316,243 (thermogenic capacity)

Mice, 48 weeks old, were anesthetized with isoflurane and placed on a heat pad warmed at 37 °C for the insertion of a temperature transmitter (model TA-F10; Data Sciences International, MN) in the peritoneal cavity. The abdominal wall was sutured, and the skin was closed with surgical glue (Dermabond Topical Skin Adhesive; Johnson & Johnson, Sao Paulo, Brazil). After surgery, mice were treated with antibiotic (enrofloxacin 5 mg/kg sc) and analgesic (ketoprofen 5 mg/kg sc) and allowed to recover for 7 days. After this period, mice were transferred to metabolic cages at 30 °C at 09:00 h and food restricted. After 3 h (12:00 h), mice were intraperitoneally injected with the β-adrenergic agonist CL-316,243 (1 mg/kg in PBS) and evaluated for body temperature and oxygen consumption. Core body temperature was monitored and acquired in continuous mode by the Data Sciences telemetry system. Oxygen consumption was determined by open-flow respirometry as described elsewhere [51]. A similar procedure was used to evaluate oxygen consumption in response to CL-316,243 in 8 weeks old mice, but without the surgery and insertion of a temperature transmitter.

FGF-21, leptin, insulin, and adiponectin

Serum FGF-21, leptin, insulin, and adiponectin levels were measured with mouse ELISA kits (FGF-21 DuoSet ELISA, R&D, MN, USA; and mouse leptin, adiponectin and ultra-sensitive mouse insulin from Crystal Chem, IL, USA) following supplier recommendations. For the evaluation of tissue FGF-21 content, liver, BAT, and iWAT were homogenized in a buffer composed of 50 mM HEPES, 2 mM EDTA, 10 mM sodium pyrophosphate, 40 mM NaCl, 10 mM sodium glycerophosphate, 50 mM NaF, 2 mM sodium orthovanadate, 1% Triton-X 100, and EDTA-free protease inhibitors and centrifuged. Supernatant was used for FGF-21 content measurement by ELISA.

RNA extraction and qPCR

Total RNA was extracted from the liver (30 mg), iWAT (100 mg), and BAT (40 mg), reverse transcribed, and destined for quantitative PCR analysis as previously described [34]. Primer nucleotide sequences are depicted in Table 1. Analysis of real-time PCR data was performed using the 2-ΔΔCT method. Data are expressed as the ratio between the expression of the target gene and the housekeeping genes (36B4 and RPL3), the expression of which was not significantly affected by the mouse genotype.

Table 1 Primers used in qPCR
Full size table

Liver histology

Liver, BAT, and iWAT samples were fixed in 4% paraformaldehyde and embedded in paraffin. Sections 5.0 μm-thick were obtained and stained with hematoxylin and eosin (H/E) to assess general liver morphology. Liver histopathological analysis was carried out using a previously defined NAFLD scoring system for rodents [32].

Western blot analysis

BAT (30 mg) and iWAT (50 mg) were homogenized in a buffer composed of 50 mM HEPES, 2 mM EDTA, 10 mM sodium pyrophosphate, 40 mM NaCl, 10 mM sodium glycerophosphate, 50 mM NaF, 2 mM sodium orthovanadate, 1% Triton-X 100, and EDTA-free protease inhibitors, centrifuged, and analyzed for protein content as previously described [9]. Primary antibody against UCP-1 (#14670) (Cell Signaling Technology, Beverly, MA, USA) was diluted at 1:1000 in TBS-T 5% milk. Membranes were incubated with peroxidase-conjugated secondary antibody (1:5000) and revealed with ECL-enhanced chemiluminescence substrate (GE Healthcare).

Statistical analysis

Data are expressed as mean ± SEM. Student’s t-test was used to analyze the effects of hepatocyte Pten deletion (same age L-Pten WT vs L-Pten KO) using GraphPad Prism. The significance level was set at p ≤ 0.05.

Results

As illustrated in Fig. 1, mice with Pten deletion exclusively in hepatocytes displayed a gradual and robust increase in liver mass, lipid accumulation, and steatosis upon aging, as evaluated by H/E staining, culminating at 24 weeks in NASH and 48 weeks in hepatomegaly and development of liver tumors (Fig. 1A, B). Indeed, all 48 weeks old L-Pten KO evaluated displayed several tumors homogenously distributed all over the liver (Fig. 1A). Histopathologic analysis of L-Pten KO mouse liver revealed a clear NAFL-NASH-HCC progression upon aging characterized by increased steatosis score (sum of micro and macrovesicular steatosis and hepatocyte hypertrophy) and fibrosis (Fig. 1A and Table 2). Of note, inflammatory foci were found exclusively in the NASH (24 weeks) and HCC (48 weeks), but not NAFL displayed by L-Pten KO mice (Table 2).

Fig. 1
figure 1

Pten deletion in hepatocytes promotes liver cancer and reduces iWAT and gastrocnemius masses in 48-week-old mice. Liver mass (A) and microphotography and histology by H/E staining (B), body weight (C), inguinal white adipose tissue (iWAT) (D), gastrocnemius (E), and brown adipose tissue (BAT) (F) masses and food intake (G) in 8-, 24-, and 48-week-old Pten floxed (L-Pten WT) and Pten floxed albumin cre+/− (L-Pten KO) mice bearing Pten deletion in hepatocytes. Results are expressed as mean ± SEM. n = 6–16 mice per group. Student’s t-test was used to analyze the effects of Pten deletion (same age L-Pten WT vs L-Pten KO). *p ≤ 0.05

Full size image
Table 2 Histopathological features of NAFLD in 8, 24, and 48 weeks-old mice with Pten deletion in hepatocytes (L-Pten KO) and littermate controls (L-Pten WT)
Full size table

Despite the marked increase in L-Pten KO liver mass, there were no differences in mice’s body weight between groups (Fig. 1C). Indeed, L-Pten KO displayed reduced masses of iWAT (Fig. 1D) at 24 and 48 weeks of age and of gastrocnemius skeletal muscle at 48 weeks of age only (Fig. 1E). In contrast to iWAT and gastrocnemius, BAT mass was significantly increased in 8, 24, and 48 weeks old L-Pten KO (Fig. 1F). There were no differences in food intake between groups (Fig. 1G). Altogether, these findings indicate that mice bearing HCC induced by hepatocyte Pten deletion show increased liver and BAT but reduced iWAT and skeletal muscle masses, ensuring no changes in body weight.

Next, we investigated the impact of NAFLD progression on serum and tissue FGF-21 content. As illustrated in Fig. 2A, serum FGF-21 concentration was significantly increased in L-Pten KO upon aging and progression of liver disease, attaining its highest levels in 48-week-old HCC-bearing mice. A similar pattern was seen in liver FGF-21 content, namely, an increase upon aging and progression of liver disease (Fig. 2B). In contrast, FGF-21 content was either not altered in BAT or increased in iWAT, but only in 48-week-old L-Pten KO (Fig. 2C, D, respectively). Serum insulin and leptin were significantly reduced in L-Pten KO in all ages investigated (Fig. 2E, F), while serum adiponectin was reduced in 24- and 48-week-old L-Pten KOs only (Fig. 2G). Liver mRNA content of Tsukushi, a hepatokine secreted upon liver disease, was significantly increased in 24- and 48-week-old L-Pten KOs (Fig. 2H), while mRNA content of the transcription factors and regulators of FGF-21 expression, namely, ChREBPβ (all ages) and PPARα (8 and 48 weeks old), but not ChREBPα, was significantly increased in L-Pten KO mice (Fig. 2I–K). These findings indicate that NAFLD progression induced by hepatocyte Pten deletion increases serum FGF-21 levels mainly by increasing liver production and secretion of this hepatokine through a mechanism that may involve activation of ChREBPβ and PPARα.

Fig. 2
figure 2

NAFLD induced by Pten deletion in hepatocytes is associated with increased serum and liver FGF-21 content and Tsukushi, ChREBPβ, and PPARα mRNA but reduced serum insulin, leptin, and adiponectin levels. Serum (A), liver (B), brown adipose tissue (BAT) (C), and inguinal white adipose tissue (iWAT) (D) FGF-21 content, serum insulin (E), leptin (F), adiponectin (G) levels and liver Tsukushi (H), ChREBPα (I), ChREBPβ (J), and PPARα (K) mRNA content in 8-, 24-, and 48-week-old Pten floxed (L-Pten WT) and Pten floxed albumin cre+/- (L-Pten KO) mice bearing Pten deletion in hepatocytes. Results are expressed as mean ± SEM. n = 5-8 mice per group. Student’s t-test was used to analyze the effects of Pten deletion (same age L-Pten WT vs L-Pten KO). *p ≤ 0.05

Full size image

Our findings that BAT mass and serum FGF-21 are increased in L-Pten KO mice motivate us to investigate the impact of NAFLD progression on BAT and iWAT thermogenic function. As illustrated in Fig. 3, fatty liver disease was associated with major changes in BAT and iWAT gene expression profile, uncoupling protein 1 (UCP-1) content, and tissue morphology. Indeed, L-Pten KO displayed in BAT reduced mRNA content of UCP-1 (Fig. 3A), glycerol kinase (GyK) (Fig. 3B), lipoprotein lipase (LPL) (Fig. 3C), and fatty acid transporter protein 1 (FATP-1) (Fig. 3D) at 8, 24, and 48 weeks of age; reduced protein content of UCP-1 at 24 and 48 weeks of age (Fig. 3E); and increased lipid deposition and percentage of unilocular cells at 8, 24, and 48 weeks of age (Fig. 3F). In contrast, L-Pten KO displayed in iWAT increased UCP-1 (Fig. 3G) and GyK (Fig. 3H) mRNA content at 24 and 48 weeks of age, but no changes in LPL (Fig. 3I) and FATP-1 (Fig. 3J) mRNA. Consistently with UCP-1 and GyK mRNA, histological analysis of iWAT revealed the appearance of multilocular adipocytes mainly in 48-week-old L-Pten KO (Fig. 3K). Altogether, these findings suggest that liver cancer induced by Pten deletion in hepatocytes impacts BAT and iWAT differently, with the former showing clear signs of inactivation and the latter featuring thermogenic recruitment and browning.

Fig. 3
figure 3

Liver cancer reduces BAT UCP-1 content but promotes iWAT browning. Brown adipose tissue (BAT) uncoupling protein 1 (UCP-1) (A), glycerol kinase (GyK) (B), lipoprotein lipase (LPL) (C), fatty acid transporter 1 (FATP-1) (D) mRNA levels, UCP-1 protein content (E), histology by H/E staining (F), and inguinal white adipose tissue (iWAT) UCP-1 (G), GyK (H), LPL (I), and FATP-1 (J) mRNA levels and histology by H/E staining (K) in 8, 24 and 48-week-old Pten floxed (L-Pten WT) and Pten floxed albumin cre+/− (L-Pten KO) mice bearing Pten deletion in hepatocytes. Results are expressed as mean ± SEM. n = 6-10 mice per group. Student’s t-test was used to analyze the effects of Pten deletion (same age L-Pten WT vs L-Pten KO). *p ≤ 0.05

Full size image

In the face of the opposite response of both BAT and iWAT seen in 48-week-old L-Pten KO, we next compared total UCP-1 content in both tissues to evaluate their individual impact on whole-body energy expenditure. As illustrated in Fig. 4A, applying the same amount of protein, we did not detect UCP-1 in iWAT of L-Pten KO in the immunoblotting, while we found detectable UCP-1 in BAT, such content that decreased upon aging and progression of liver disease. Next, we evaluated UCP-1-dependent thermogenic capacity in 8- and 48-week-old L-Pten WT and KO by evaluating mouse oxygen consumption at thermoneutral conditions (30 °C) in response to a challenge with the β3 adrenergic receptor agonist CL-316,243. As depicted in Fig. 4B, 8-week-old L-Pten WT and L-Pten KO displayed similar oxygen consumption rates in response to the CL-316,243 injection, indicating similar thermogenic capacity between groups. In contrast, 48-week-old L-Pten KO featured a severely impaired thermogenic capacity, as evidenced by the smaller increase in whole-body oxygen consumption and core body temperature in response to CL-316,243 than those displayed by littermates L-Pten WT (Fig. 4C, D, respectively). Altogether, these findings indicate that UCP-1-mediated thermogenic capacity is markedly reduced in 48-week-old L-Pten KO mice.

Fig. 4
figure 4

Liver cancer reduces UCP-1-mediated thermogenic capacity. Inguinal white adipose tissue (iWAT) and brown adipose tissue (BAT) uncoupling protein 1 (UCP-1) content in 8-, 24-, and 48-week-old Pten floxed albumin cre+/- (L-Pten KO) mice (A), delta of whole-body oxygen consumption in 8- and 48-week-old Pten floxed (L-Pten WT) and L-Pten KO (B, C), and delta of core temperature (D) in 48-week-old L-Pten WT and L-Pten KO mice evaluated at thermoneutral conditions (30 °C) in response to CL316,243 (1 mg/ kg of body weight). Results are expressed as mean ± SEM. n = 4-8 mice per group. Student’s t-test was used to analyze the effects of Pten deletion (same age L-Pten WT vs L-Pten KO). *p ≤ 0.05

Full size image

Next, we investigated the impact of those changes in thermogenic capacity on whole-body and tissue energy expenditure through indirect calorimetry (oxygen consumption). As illustrated in Fig. 5A, 8-week-old L-Pten KO displayed increased oxygen consumption and respiratory exchange ratio (RER), but no changes in locomotor activity, indicating increased energy expenditure and utilization of carbohydrates as a substrate for energy production. There were no significant changes in oxygen consumption, locomotor activity, and RER in 24- and 48-week-old L-Pten KO (Fig. 5B, C). The relationship between oxygen consumption and body mass in 8- and 48-week-old mice was evaluated using ANCOVA and was found not significant (Supplementary Files 1 and 2). We also examined the individual contribution of tissues to whole-body energy expenditure. As illustrated in Fig. 5D, L-Pten KO displayed reduced liver rates of oxygen consumption expressed per mg of tissue at 8 and 48 weeks of age but increased whole-liver oxygen consumption at 24 and 48 weeks of age, which reflects organ contribution to whole-body energy expenditure. Similarly to the liver, L-Pten KO displayed reduced BAT rates of oxygen consumption expressed per mg of tissue at 24 and 48 weeks of age but increased whole-BAT oxygen consumption at 8, but not at 24 and 48 weeks of age (Fig. 5E). Regarding iWAT, on the other hand, L-Pten KO showed increased rates of oxygen consumption expressed per mg of tissue at 48 weeks but reduced whole-iWAT oxygen consumption at all ages investigated (Fig. 5F). Altogether, these findings indicate that HCC induced by Pten deletion in hepatocytes is associated with reduced BAT UCP-1 and thermogenic capacity, but no changes in energy expenditure.

Fig. 5
figure 5

Liver cancer does not affect whole-body energy expenditure. Whole-body oxygen consumption (VO2), locomotor activity, respiratory exchange ratio (RER) (A–C), liver (D), brown adipose tissue (BAT) (E), and inguinal white adipose tissue (iWAT) (F) oxygen consumption expressed by mg of tissue and total tissue in 8-, 24-, and 48-week-old Pten floxed (L-Pten WT) and Pten floxed albumin cre+/- (L-Pten KO) mice bearing Pten deletion in hepatocytes. Results are expressed as mean ± SEM. n = 6–10 mice per group. Student’s t-test was used to analyze the effects of Pten deletion (same age L-Pten WT vs L-Pten KO). *p ≤ 0.05

Full size image

Discussion

We tested herein the hypothesis that an enhanced BAT and iWAT UCP-1-mediated thermogenesis induced by high levels of FGF-21 are involved in the catabolic state and reduction in fat mass found upon HCC. Our main findings indicate that, despite high serum FGF-21 levels and iWAT browning, HCC induced by Pten deletion in hepatocytes is associated with reduced BAT UCP-1 content and expression of sympathetically-regulated genes GyK, LPL, and FATP, which altogether resulted in an impaired whole-body thermogenic capacity in response to CL-316,243. These findings strongly suggest that FGF-21 pro-thermogenic actions in BAT are context-dependent, not occurring upon NASH or HCC induced by hepatocyte Pten deletion, perhaps due to the reduced serum leptin, insulin, and adiponectin levels found in these conditions. Additionally, it can be concluded that UCP-1-mediated nonshivering thermogenesis is not a major energy-expending component of the catabolic process associated with HCC induced by Pten deletion in hepatocytes, such findings that are in accordance with studies in humans showing that brown fat thermogenesis is not involved with cancer-associated cachexia and does not affect survival in patients suffering with cachexia [15, 4].

Confirming and extending previous findings [21, 52], Pten deletion in hepatocytes was associated with a marked and progressive increase in liver mass and lipid accumulation culminating with NASH and HCC appearance in 24- and 48-week-old mice, respectively. Interestingly, the marked increase in liver mass in HCC was not associated with major changes in mouse body weight, likely due to the compensatory reductions in adipose tissue and skeletal muscle masses, which are common manifestations of a cancer-related catabolic state. Of note, HCC-related reduction in muscle and adipose tissue masses, which cannot be attributed to changes in food intake, may be related to catabolic signals secreted by the liver along the reduced serum levels of insulin, a growth factor that displays anabolic actions on muscle mass.

One liver-derived putative candidate to mediate HCC actions is FGF-21, a hepatokine whose liver and serum contents were markedly increased during the NAFLD progression induced by Pten deletion in the hepatocyte, reaching its highest levels upon HCC. Indeed, FGF-21 has been shown to act as a robust marker of NAFLD in both rodents and humans [14, 58, 57], to promote adipose tissue lipolysis in vivo under specific conditions [22, 49], and to mediate both the muscle atrophy induced by fasting [41] and the improvement in insulin sensitivity displayed by L-Pten KO mice [5]. Similarly to lipolysis, FGF-21 was also shown in specific conditions to promote BAT and iWAT UCP-1-mediated thermogenesis increasing energy expenditure and thus counteracting obesity and NAFLD development in mice [45, 59, 12, 56]. Indeed, we found that upon HCC, elevated serum FGF-21 was associated with increased UCP-1 and GyK mRNA levels and the appearance of multilocular adipocytes in iWAT, indicating enhanced tissue browning. But surprisingly, and in contrast to iWAT, high FGF-21 levels were also associated with increased lipid deposition and reduced UCP-1 content and expression of the sympathetic-regulated proteins GyK, LPL, and FATP-1 in BAT [16, 28, 6], all suggestive of reduced BAT sympathetic activity and thermogenesis. One interesting question that arises from these findings is whether iWAT browning could compensate for the reduced BAT UCP-1 content seen upon HCC. The markedly lower, non-detected UCP-1 content in iWAT in comparison to readily detectable in BAT in the immunoblotting along with the impaired thermogenic capacity of 48-week-old L-Pten KO mice evaluated in response to CL-316,243 strongly indicate that an enhanced iWAT browning cannot compensate for the reduced BAT UCP-1 content and thermogenesis in this condition. These findings are in agreement with several studies showing a minimal contribution of iWAT browning to mice thermogenic capacity and energy expenditure in vivo [7, 8, 30, 31, 48] and exclude a possible implication of UCP-1-mediated thermogenesis in HCC-associated catabolic state and cachexia. Corroborating our findings, two recent studies found through fluoro-deoxyglucose (18F-FDG) positron emission tomography (PET) scans of large human cohorts that BAT is not associated with cancer-associated cachexia and does not worsen overall survival in patients with cachexia [15, 4].

Noteworthy, these changes in iWAT and BAT UCP-1 content and thermogenic capacity seen upon HCC also did not impact energy expenditure in mice, as evidenced by the similar rates of oxygen consumption displayed by 48-week-old L-Pten KO and WT; such findings are surprising, considering the important contribution of BAT thermogenesis to energy expenditure in mice acclimated to 23 °C, a temperature below the thermoneutral zone. One possible reason to explain this may be the marked increase in liver oxygen consumption found upon HCC that may compensate for the reduced BAT thermogenesis and oxygen consumption. In humans, cirrhosis and HCC have been associated with mixed results regarding energy expenditure with reports of reductions, no changes, or increases, such discrepancies that seem to be related to the cause and stage of the disease, patient sex, and lean mass, among other factors [36, 39]. Importantly, the contribution of UCP-1-mediated nonshivering thermogenesis to energy expenditure in HCC patients may be lower than that seen in rodents considering their smaller relative amounts of brown fat and that humans spend most of their lives under thermoneutral conditions.

Another interesting question that emerges from our findings is why high FGF-21 levels promote UCP-1-mediated thermogenesis and energy expenditure in some conditions such as diet-induced obesity [12, 56], but not others such as fasting [2, 24] and NAFLD induced by Pten deletion in hepatocytes shown here. One possibility is that these FGF-21 actions are context-dependent requiring other signaling molecules enriched upon diet-induced obesity, but not fasting or NAFLD induced by Pten deletion in hepatocytes. In this sense, a recent study has shown that FGF-21 signals through leptin receptor-expressing cells to regulate body weight, and that central leptin signaling is required for FGF-21 full effects to increase energy expenditure and reduce body weight [10]. Indeed, we found in our study that serum leptin levels are significantly reduced in L-Pten KO mice in all ages investigated. In addition to leptin, serum levels of insulin and adiponectin, hormones previously shown to regulate energy balance and sympathetic activity to BAT and iWAT [43, 35], were also markedly reduced in L-Pten KO mice upon NASH and HCC. While this reduction in serum insulin is expected considering the increase in insulin sensitivity induced by FGF-21 in L-Pten KO mice [5], that of adiponectin is not in view of the previous findings that FGF-21 strongly induces the secretion of this adipokine [33, 20]. Indeed, FGF-21 not only stimulates adiponectin secretion, but also requires this adipokine to exert its actions on energy expenditure and glucose homeostasis [33, 20]. Whether leptin, insulin, and/or adiponectin are the missing mediators required for the translation of high serum FGF-21 levels into higher rates of UCP-1-mediated thermogenesis and energy expenditure in NAFLD induced by Pten deletion in hepatocytes will be addressed in future studies.

This study has important limitations. Despite being a recurrent signature in NASH and HCC [18, 47, 53], Pten loss of function in hepatocytes promotes a NAFL-NASH-HCC progression in mice that does not reproduce some frequent phenotypes seen in the NAFLD-associated with obesity, namely, insulin resistance, hyperglycemia, and hyperinsulinemia. Therefore, generalizations of our findings to other types and models of NAFLD including those associated with obesity should be done carefully.

In conclusion, our main findings indicate that FGF-21 pro-thermogenic actions in BAT are context-dependent not occurring upon HCC induced by Pten deletion in hepatocytes. Furthermore, our findings indicate that UCP-1-mediated nonshivering thermogenesis is not an important energy-expending component of the catabolic state associated with HCC induced by Pten deletion in hepatocytes. Future studies will address whether these phenotypes apply to different models of NAFLD including those associated with obesity in which reduced serum adiponectin and central leptin and insulin resistance may also affect FGF-21 pro-thermogenic actions.