The level of KAL is increased in HLP and NAFLD patients and positively correlated with TG

We reanalyzed a plasma proteome profiling in ProteomeXchange (PRIDE archive PXD011839, https://www.ebi.ac.uk/pride/archive/). The samples were divided into HTG and control groups based on TG levels (individuals with TG ≥ 1.7 mM are defined as HTG), excluding obese individuals and individuals over 60 years of age. With at least one valid value in the samples, we get a dataset of 593 protein groups. Using Student’s t-test to determine the significantly changed proteins with a p-value of 0.05. We found that KAL was significantly up-regulated in the HTG group (supplementary Fig. 1a).

To further investigate the potential association between KAL and HLP, we conducted a study involving 253 HLP and 221 age-matched healthy controls. The clinical and biochemical characteristics of the participants are summarized in supplementary Table 1. Our results indicated that plasma KAL levels were significantly higher in HLP subjects compared to healthy control and positively correlated with TG, free fatty acid (FFA), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and negatively correlated with high-density lipoprotein cholesterol (HDL-C) (Fig. 1a-b, supplementary Fig. 1b-e). Moreover, the plasma KAL levels in HTG subjects were even higher than those in the HLP without HTG group (Fig. 1c). We also found that the alanine transaminase (ALT) and aspartate aminotransferase (AST) levels in HTG were higher than those in the HLP without HTG group and the healthy control group and positively correlated with KAL (Fig. 1d-e, supplementary Fig. 1f-g).

Fig. 1

The content of KAL is increased in HLP and NAFLD patients, and positively correlated with TG. a The plasma KAL concentration in HLP subjects. b Correlation analysis of plasma KAL levels and TG concentrations. c The plasma KAL concentration in control and HLP subjects. d The serum ALT concentration in control and HLP subjects. e Correlation analysis of plasma KAL levels and ALT concentrations in control and HLP subjects. (a–e: Control, n = 221; HLP, n = 253; HLP w/o HTG, HLP without HTG, n = 167; HTG, n = 86). f The serum KAL concentration in NAFLD patients. g–i Correlation analysis of serum KAL levels with AST, ALT and TG. (f–i: Control, n = 62; NAFLD, n = 44). j Representative images and quantification of KAL level in the livers from NAFLD patients (Control, n = 6; NAFLD, n = 16). Scale bar: 50 μm. Data are expressed as mean ± SD. **p < 0.01, ***p < 0.001

In addition, the serum KAL levels were higher in non-obese NAFLD patients than in matched healthy controls (Fig. 1f) and positively correlated with AST, ALT, and TG but not TC, LDL-C, and HDL-C (Fig. 1g-i, supplementary Fig. 1h-j, supplementary Table 2). Similarly, the KAL staining was increased in the liver of mild NAFLD patients (Fig. 1j). These findings suggested that elevated KAL levels are involved in the accumulation of TG and subsequent liver damage.

Elevated KAL induces hepatic steatosis and NASH in chow-diet mice and aggravates hepatic steatosis to NASH in HFD mice

To investigate the role of KAL in NAFLD, we conducted experiments on KAL-transgenic (KAL-Tg) mice. And the average serum KAL concentration of KAL-Tg mice was 2.6 μg/mL (Fig. 2a), which is of the same order of magnitude as that in the human HLP population (Fig. 1a, 2-10 μg/mL). We observed that KAL-Tg mice exhibited increased serum AST and ALT levels (Fig. 2b), indicating liver damage. Moreover, KAL-Tg mice exhibited slight hepatic lipid droplet deposition at 3 months and developed severe hepatic steatosis, disordered arrangement of hepatocytes, elevated hepatic TG and fatty acid levels, and increased liver weight at 6 months (Fig. 2c-g). It suggested that elevated KAL contributes to hepatic steatosis.

Fig. 2

KAL induces hepatic steatosis and NASH in chow-diet mice and progresses hepatic steatosis to NASH in HFD mice. a The serum KAL concentration of KAL-Tg mice (n = 8). b The serum AST and ALT levels of 6-month-old mice (n = 5). c Oil red O staining and steatosis score of livers from 3-month-old mice (WT, n = 3; KAL-Tg, n = 4). Scale bar: 200 μm. d–g Weight, H&E (Scale bar: 100 μm) and Oil red O (Scale bar: 200 μm) staining, steatosis score, TG and fatty acid content of liver tissues from 6-month-old mice (n ≥ 8 per group). h–l Representative images, Oil red O and H&E staining, NAS, TNFα level of livers and serum TNFα in 10-month-old mice (n ≥ 5 per group). Scale bar: 100 μm. m–o Sirius staining and fibrosis scores, mRNA levels, and protein levels of genes related to inflammation and fibrosis in livers from 16-month-old mice (n ≥ 5 per group). Scale bar: 100 μm. p–r H&E and Sirius staining, NAS, fibrosis scores, mRNA levels and protein levels of genes related to inflammation and fibrosis of livers from mice fed an HFD for 28 weeks (n ≥ 4 per group). Scale bar: 100 μm. Data are expressed as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001

To further investigate the effect of KAL on NASH, we found that the livers of 10-month-old KAL-Tg mice exhibited yellow and granular surfaces, and developed NASH with apparent inflammation, balloon-like degeneration of hepatocytes, and NAFLD activity score (NAS) greater than 5 (Fig. 2h-i, supplementary Fig. 2a-d). The expression and secretion of the inflammatory factor tumor necrosis factor α (TNFα) were also increased (Fig. 2j-l, supplementary Fig. 2e). Since the 10-month-old mice did not exhibit extensive fibrosis with collagen deposition (supplementary Fig. 2f), we continued to raise the mice for 16-month-old and found that 16-month-old KAL-Tg mice developed not only NASH (supplementary Fig. 2g–i) but also hepatic fibrosis with elevated α-smooth muscle actin (α-SMA) and collagen I (Fig. 2m-o).

Additionally, to clarify whether KAL speeds up the development of hepatic steatosis to NASH, the classic model of hepatic steatosis, HFD mice, was used. When we fed KAL-Tg mice and wild-type (WT) mice with HFD, KAL-Tg mice exhibited extensive inflammation, injury, and fibrosis along with aggravated hepatic steatosis (Fig. 2p, supplementary Fig. 2j-m), and the expression of TNFα, interleukin-6 (IL-6), α-SMA, and collagen I was also increased (Fig. 2q-r, supplementary Fig. 2n). Our findings suggest that elevated KAL contributes to the progression of simple hepatic steatosis to NASH in HFD mice.

Knockout of KAL ameliorates hepatic steatosis and inflammation in NAFLD rats

As previously mentioned, KAL is a secreted protein. To elucidate the role of KAL in NAFLD, we generated KAL gene knockout (Serpina4−/−) rats and fed them with a HFruD for 16 weeks to induce an HTG non-obese NAFLD model (Fig. 3a), and found that the expression of KAL in the liver was higher in HFruD rats; Serpina4−/− rats showed a significant improvement in hepatic steatosis in HFruD rats (Fig. 3b-c). Furthermore, fed Serpina4−/− rats with a MCD diet for 4 weeks to induce hepatic steatosis and 10 weeks to NASH, respectively. We observed that the expression of KAL in the liver was also higher in MCD diet-induced NAFLD rats (Fig. 3d); Serpina4−/− rats showed a significant improvement in hepatic steatosis, inflammation, and collagen fiber deposition in MCD-induced NAFLD rats (Fig. 3e-f). Additionally, Serpina4−/− down-regulated the expression of TNFα, α-SMA and Collagen I in the liver tissue of MCD-induced NAFLD rats (Fig. 3g-h, supplementary Fig. 2o). These loss-of-function experiments provide evidence that high levels of KAL contribute to NAFLD.

Fig. 3

KAL knockout improves hepatic steatosis, inflammatory infiltration and fibrosis of NAFLD rats. a The serum TG of rats fed with HFruD for 16 weeks. b, c mRNA levels of Serpina4, Oil red O staining and steatosis score of livers from the rats fed with HFruD for 16 weeks. d mRNA levels of Serpina4 in the livers from rats fed with MCD. e Oil red O staining and steatosis score of livers from the rats fed with MCD for 4 weeks. f Oil red O staining and steatosis score, H&E staining and NAS, Sirius staining and fibrosis scores of livers from the rats fed with MCD for 10 weeks. g The mRNA levels of Tnf, Acta2 and Col1a1 in the livers of rats fed with MCD for 10 weeks. h Immunoblot of TNFα, α-SMA, and Collagen I in the liver tissues from rats fed with MCD for 10 weeks, the black arrow represents the destination blots. Scale bar for Oil red O staining: 500 μm, for H&E staining and Sirius staining: 200 μm. Data are expressed as mean ± SD of no less than 4 rats per group. *p < 0.05, **p < 0.01, ***p < 0.001

KAL causes hepatic steatosis by down-regulating both CGI-58 and ATGL, and inflammation mainly by CGI-58

Our RNAseq result of primary hepatocytes from 6-month-old WT and KAL-Tg mice indicated the downregulation of ATGL (encoded by the Pnpla2) and CGI-58 (encoded by the Abhd5) (supplementary Fig. 3a-b) (GSA dataset CRA014927), which are closed related with NAFLD. Hepatic ectopic deposition of lipid droplet-like triglycerides is influenced by several factors, including increased uptake and de novo synthesis of FFA, reduced β-oxidation of FFA, reduced TG hydrolysis, and extracellular transport through VLDL.2 Our study revealed that KAL did not increase FFA synthesis and uptake, had no effect on the FFA β-oxidation and the critical molecules of VLDL synthesis (supplementary Fig. 3c-d), nor did it alter the Akt signaling pathway or hepatic glucose transporter (supplementary Fig. 3e). Consistent with RNAseq results, the expressions of ATGL, a critical enzyme for TG hydrolysis, and its co-activator CGI-58 were significantly decreased by KAL (Fig. 4a). Moreover, ATGL and CGI-58 were down-regulated in primary hepatocytes treated with Ad-KAL or primary hepatocytes from KAL-Tg mice (Fig. 4b-c, supplementary Fig. 3f-g). Overexpression of ATGL and/or CGI-58 could reduce the lipid droplet induced by KAL in hepatocytes (Fig. 4d). Similarly, decreased ATGL and CGI-58 expression in liver tissue was observed in MCD and HFruD-induced NAFLD rats, which was effectively rescued by Serpina4−/− (Fig. 4e-f). Therefore, it is suggested that KAL inhibited hepatic TG hydrolysis by suppressing CGI-58 and ATGL, leading to hepatic steatosis.

Fig. 4

KAL causes hepatic steatosis by down-regulating both CGI-58 and ATGL, and inflammation mainly by CGI-58. a Representative immunoblot and quantification of ATGL and CGI-58 in liver tissues from 6-month-old mice. b, c Representative immunoblot and quantification of protein levels, mRNA levels of ATGL and CGI-58 in primary hepatocytes cultured from WT or KAL-Tg mice. d Oil red O staining of primary hepatocytes cultured from WT and KAL-Tg mice and transfected with ATGL and/or CGI-58 plasmids for 48 h. Scale bar: 50 μm. e, f The levels of ATGL and CGI-58 in liver tissues from rats fed with MCD for 10 weeks (e) or HFruD for 16 weeks(f). g Representative immunoblot of ATGL and CGI-58 in liver tissues from 3-month-old mice. h mRNA levels of ABHD5 in liver tissue from patients with hepatic steatosis or NASH (Control, n = 24; steatosis, n = 20; NASH, n = 19 for GEO accession number GSE89632). i Correlation analysis of mRNA levels of ABHD5 and SERPINA4. j–l Representative immunoblot and quantification of TNFα, mRNA levels of TNFα, and supernatant TNFα levels in primary hepatocytes treated with Ad-KAL or Ad-RFP for 48 h. m–o Representative immunoblot and quantification of TNFα, mRNA levels of TNFα in primary hepatocytes transfected with Ad-KAL and ATGL or CGI-58 plasmid for 48 h. p, q Protein levels and mRNA levels of CGI-58 in Raw macrophagocytes. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001

Further, we observed that the expression of ATGL, but not CGI-58, had begun to decline in the liver tissues of 3-month-old KAL-Tg mice (Fig. 4g, supplementary Fig. 3h). In addition, we analyzed a human NAFLD data set (GSE89632) and found that CGI-58 expression was decreased in liver tissues of patients with hepatic steatosis and NASH and was significantly negatively correlated with KAL (Fig. 4h-i). Studies have shown that deficiency of liver ATGL causes hepatic steatosis but not NASH or fibrosis,17 whereas hepatic CGI-58 deficient mice exhibit pronounced hepatic steatosis and later develop NASH and liver fibrosis, accompanied by increases in TNFα and α-SMA.18 Individuals with CGI-58 loss-of-function mutations typically experience severe hepatic steatosis, NASH, and cirrhosis.19 Our results showed that KAL up-regulated the expression and secretion of TNFα in primary hepatocytes (Fig. 4j-l), which could be inhibited by CGI-58 but not ATGL (Fig. 4m–o). Furthermore, KAL did not affect the expression of CGI-58 in macrophages (Fig. 4p-q). Thus, our findings suggest that KAL induces the expression of TNFα by downregulating CGI-58 in hepatocytes and leading to inflammation through an ATGL-independent mechanism.

KAL induces nuclear translocation of NF-κB p65 by reducing its binding to CGI-58 in hepatocytes

In relation to the upregulation of TNFα expression resulting from the reduction of CGI-58, a question was raised regarding the mechanism involved. Our investigation first focused on NF-κB, a crucial inflammatory signaling pathway implicated in chronic inflammation associated with NAFLD11 and a known inducer of TNFα transcription.20 Previous research has indicated that the activation of the NF-κB signaling pathway by KAL results in inflammation in diabetic wound tissues.13 However, whether reducing CGI-58 would activate the NF-κB signaling pathway and lead to the upregulation of TNFα induced by KAL is unknown.

It is essential for NF-κB p65 to translocate into the nucleus to induce gene expression.21 Our findings indicate a significant increase in the nuclear translocation of NF-κB p65 in liver tissues of KAL-Tg mice and primary hepatocytes treated with Ad-KAL (Fig. 5a-b, supplementary Fig. 4a-c). Knocking down CGI-58 led to a notable increase in the nuclear translocation of NF-κB p65, along with the expression of TNFα and matrix metalloproteinase (MMP9), a well-established target of NF-κB, in primary hepatocytes (Fig. 5c-f). Furthermore, overexpressing CGI-58 could block the nuclear translocation of NF-κB p65 induced by KAL (Fig. 5g). Taken together, our results demonstrate that reducing CGI-58 leads to the activation of the NF-κB/TNFα signaling pathway.

Fig. 5

KAL induces nuclear translocation of NF-κB p65 by reducing its binding to CGI-58 in hepatocytes. a NF-κB p65 levels in cytosolic (Cyto.) and nuclear (Nuc.) extracts in liver tissues from 10-month-old mice. b NF-κB p65 (green) immunostaining in primary hepatocytes treated with Ad-KAL for 48 h. Scale bar: 100 μm. c–f Representative immunoblot and quantification of NF-κB p65 in nuclear extracts (c), mRNA levels of TNFα (d), representative immunoblot and quantification of TNFα (e), and supernatant TNFα levels (f) in hepatocytes treated with si-Abhd5 for 48 h. g Representative immunoblot and quantification of NF-κB p65 in nuclear extracts of primary hepatocytes transfected with KAL and CGI-58 plasmid for 48 h. h NF-κB p65 (green) and CGI-58 (red) immunostaining in primary hepatocytes. Images were acquired under a laser-scanning confocal microscope. Scale bar: 25 μm. i NF-κB p65 and CGI-58 levels in Cyto and Nuc extracts of primary hepatocytes. j NF-κB p65 immunoblotting after immunoprecipitation (IP) for CGI-58 in primary hepatocytes, IP for IgG as the negative control. k The co-IP blot and quantification of CGI-58 and NF-κB p65 in primary hepatocytes treated with Ad-KAL for 48 h, IP for IgG as the negative control. l, m Oil red O staining, H&E staining and nuclear NF-κB p65 levels in liver tissues of hepatic CGI-58-overexpressing KAL-Tg mice (KAL + CGI-58LSL/+:Cre), the black arrow represents the destination blots. Scale bar for Oil red O staining: 200 μm, for H&E staining: 100 μm. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01

To further explore how CGI-58 regulates the nuclear translocation of NF-κB p65, we performed immunofluorescence staining with NF-κB p65 and CGI-58 and found their co-localization in the cytoplasm; CGI-58 was only localized in the cytoplasm (Fig. 5h, i, supplementary Fig. 4d). Immunoprecipitation (IP) for CGI-58 in hepatocyte cytoplasm fractions identified an interaction between NF-κB p65 and p-NF-κB p65 with CGI-58, which was impaired by Ad-KAL or si-Abhd5 (Fig. 5j, k, supplementary Fig. 4e). We also performed co-IP to identify the NF-κB subunits that mediate NF-κB activation in the absence of CGI-58 and found that the p65-p50 interactions were potentiated by reduced CGI-58 in hepatocytes (Supplementary Fig. 4f). Thus, our findings suggest that KAL down-regulates CGI-58, which interacts with NF-κB p65 and sequesters p65 in the cytoplasm, subsequently releasing p65 to facilitate its nuclear translocation.

To further investigate the key role of CGI-58 in hepatocytes regarding hepatic steatosis and inflammation induced by KAL, we constructed hepatic-specific CGI-58-transgenic mice (CGI-58LSL/+:Cre), which were bred with KAL-Tg mice to construct the hepatic CGI-58-overexpressing KAL-Tg mouse model (KAL + CGI-58LSL/+:Cre). Our results indicate that hepatic CGI-58 overexpression significantly reversed hepatic steatosis, inflammation, and the content of nuclear NF-κB p65 in KAL-Tg mice (Fig. 5l, m). It suggests that KAL induces NASH primarily through CGI-58.

KAL down-regulates ATGL and CGI-58 by inhibiting PPARγ or inducing KLF4, respectively

The underlying mechanism by which KAL down-regulated ATGL and CGI-58 requires further clarification. As shown in Fig. 4c and supplementary Fig. 3g, KAL inhibited the transcription of ATGL and CGI-58 in hepatocytes. It was reported that PPARγ could directly bind to the promoter of ATGL and enhance its transcription;22 Bioinformatics prediction indicated that PPARγ may also bind to the CGI-58 promoter (supplementary Fig. 5a); Moreover, PPARγ is closely related to NAFLD.23 However, in hepatocytes, PPARγ promoted the expression of ATGL but not CGI-58 (Fig. 6a). KAL suppressed the expression of PPARγ in liver tissues, and Serpina4-/- significantly improved the downregulation of PPARγ in MCD-induced NAFLD rats (Fig. 6b, supplementary Fig. 5b). Overexpression of PPARγ could block the downregulation of ATGL induced by KAL in hepatocytes (Fig. 6c). Therefore, it is suggested that KAL down-regulated ATGL by inhibiting PPARγ.

Fig. 6

KAL down-regulates ATGL and CGI-58 by inhibiting PPARγ and up-regulating KLF4, respectively. a PPARγ, ATGL and CGI-58 levels in primary hepatocytes transfected with PPARγ plasmid for 48 h. b PPARγ levels in liver tissues of 6-month-old mice. c Representative immunoblot and quantification of ATGL in primary hepatocytes transfected with Ad-KAL and PPARγ plasmid for 48 h. d, e Representative immunoblot and quantification of protein levels, and mRNA levels of KLF4 in primary hepatocytes transfected with Ad-KAL for 48 h. f Representative immunoblot and quantification of KLF4 and Sp1 in liver tissues of 16-month-old mice. g, h Representative immunoblot and quantification of CGI-58, mRNA levels of CGI-58 in primary hepatocytes transfected with KLF4 plasmid for 24 h. i Luciferase reporter assays of primary hepatocytes transfected with CGI-58 promoter reporters and KLF4 plasmids for 24 h. j Representative immunoblot and quantification of CGI-58 in primary hepatocytes treated with Ad-KAL and si-KLF4 for 48 h. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001

Furthermore, we found that the CGI-58 promotor contains a G/C-rich element. The classical transcription factors that could bind to this element were KLF4 and Sp1.24 KAL was found to up-regulate KLF4 but not Sp1 in hepatocytes (Fig. 6d-e); KLF4, not Sp1, was significantly elevated in liver tissues of KAL-Tg mice (Fig. 6f). Furthermore, KLF4 inhibited the expression of CGI-58 and the activity of its promoter in hepatocytes (Fig. 6g-i). Silencing of KLF4 reversed the downregulation of CGI-58 induced by KAL in hepatocytes (Fig. 6j). Hence, it is suggested that KAL down-regulated CGI-58 by inducing KLF4.

KAL regulates KLF4 and PPARγ by LRP6/Gαs/PKA/GSK3β signals

KAL is a well-known suppressor of the GSK3β/β-catenin signaling pathway by binding with low-density lipoprotein receptor-related protein 6 (LRP6).25 Surprisingly, our findings suggest that KAL does not affect the expression of Wnt ligands, the composition of DVL/GSK3β/β-catenin, and β-catenin activity, but significantly down-regulates the phosphorylation of both LRP6 and glycogen synthase kinase-3 (GSK3β) markedly (Fig. 7a, supplementary Fig. 6a-b). Hepatocytes express various G-protein-coupled receptors (GPCRs) such as the glucagon receptor and thyrotropin-releasing hormone receptor;26 LRP6, a coreceptor of GPCRs, is required for the activation of protein kinase A (PKA) under the stimulation of different GPCR ligands through the membrane G protein α(s) subunit (Gαs);27 What is more, PKA was the classical kinases that could phosphorylate GSK3β and leading to its inactivation.28 Our results indicate that KAL can bind to LRP6 in hepatocytes, and this binding is enhanced upon KAL overexpression (Fig. 7b); KAL also disrupts the localization of Gαs to the plasma membrane (supplementary Fig. 6c) and inhibits the phosphorylation of PKA in hepatocytes, treatment with the PKA agonist Dibutyryl-cAMP (db-cAMP) blocks the downregulation of p-GSK3β induced by KAL (Fig. 7c, supplementary Fig. 6d); Moreover, the phosphorylation of LRP6, PKA, and GSK3β is reduced in the livers of KAL-Tg mice (Fig. 7d, supplementary Fig. 6e). Therefore, we propose that KAL may inhibit LRP6/Gαs/PKA/GSK3β signaling.

Fig. 7

KAL regulates KLF4 and PPARγ by activating GSK3β. a–c Protein levels of non-p-β-catenin, p-GSK3β, p-LRP6 (a), p-PKA (c), and the IP blot of KAL and LRP6 (b) in hepatocytes treated with Ad-KAL for 48 h. d Protein levels of p-LRP6, p-PKA, p-GSK3β in liver tissues of 16-month-old mice. e, f Representative immunoblot and quantification of protein levels, mRNA levels of ATGL and PPARγ in hepatocytes treated with Ad-KAL and LiCl (20 mM) for 48 h. g, h Protein levels and mRNA levels of CGI-58 in hepatocytes treated with LiCl (20 mM) for 48 h. i Protein level of CGI-58 in hepatocytes treated with LiCl (20 mM) and si-β-catenin for 48 h. j, k Representative immunoblot and quantification of protein levels, mRNA levels of CGI-58 and KLF4 in hepatocytes treated with Ad-KAL and LiCl (20 mM) for 48 h. l Representative immunoblot and quantification of KLF4 in nuclear extracts of hepatocytes treated with Ad-KAL and LiCl (20 mM) for 48 h. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001

It has been reported that PPARγ is downstream of GSK3β.29 Inhibition of GSK3β with lithium chloride (LiCl) can block the downregulation of PPARγ and ATGL induced by KAL (Fig. 7e, f). Additionally, LiCl induces the expression of CGI-58, which is not improved by β-catenin interference (Fig. 7g–i). Furthermore, LiCl significantly disrupts the regulatory effects of KAL on CGI-58 and KLF4, as well as the nuclear translocation of KLF4 (Fig. 7j–l). Therefore, we suggest that KAL regulates KLF4 and PPARγ by activating GSK3β.

High FFA reverses the downregulation of KAL induced by T3, and fenofibrate down-regulates KAL in hepatocytes

The regulation of KAL gene expression remains poorly understood, leading to the question of how KAL is up-regulated. It has been reported that HTG is accompanied by higher levels of FFA, a product of lipid metabolic dysfunction,30 and NAFLD patients had significantly higher serum FFA levels than controls.31 FFA examined in our HLP subjects was positively correlated with plasma KAL levels (supplementary Fig. 1e). Our findings indicate that high levels of FFA alone do not affect KAL expression (Fig. 8a). A previous study has shown that Serpina3c, a member of the same family as KAL, is negatively regulated by thyroid hormone T3: T3 binds to thyroid hormone receptor (TR), then liganded TR recruits nuclear receptor corepressor and represses the expression of Serpina3c.32 Our results show that T3 also down-regulates KAL expression in hepatocytes (Fig. 8b). Additionally, T3 binding to TR is significantly inhibited when plasma and cell FFA are increased.33 Furthermore, our results demonstrate that high FFA can counteract the down-regulation of KAL expression and secretion induced by T3 in hepatocytes (Fig. 8c-h). Most T3 is produced from the conversion of thyroid hormone T4 in peripheral tissues, especially the liver and kidney.34 Thus, these results suggest that high FFA, a product of lipid metabolic dysfunction, can elevate KAL and indicate that KAL is a linker between HLP and NAFLD.

Fig. 8

High FFA reverses the downregulation of KAL induced by T3 and Fenofibrate down-regulates KAL in hepatocytes. a Protein levels of KAL in L-02 cells treated with OP (250 μM Oleic acid add 250 μM Palmitic acid) or BSA (Bovine serum albumin) for 24 h. b Protein levels of KAL in L-02 cells treated with T3 (25 nM) for 24 h. c–f Representative immunoblot and quantification of KAL, mRNA levels of KAL, and supernatant KAL levels in L-02 cells treated with T3 (25 nM) with/without OP for 24 h. g–h mRNA levels of KAL, supernatant KAL levels in MIHA cells treated with T3 (25 nM) with/without OP for 24 h. i–k Representative immunoblot of KAL, ATGL, CGI-58 in L-02 cells treated with Metformin (Met), Berberine (BBR), or Fenofibrate (Feno). l Supernatant KAL levels in L-02 cells treated with Feno. m–p Oil red O staining, serum TG level, steatosis score, and the mRNA level of KAL in livers from the rats fed with HFruD for 12 weeks, followed by the treatment of 100 mg/kg Feno and HFruD for 4 weeks. Scale bar: 500 μm. Data represent the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001

We also investigated whether compound drugs with improved hepatic steatosis and/or lipid-lowering effects, such as Metformin, Berberine, and Fenofibrate, could decrease KAL expression. Our findings show that only Fenofibrate, a kind of drug for severe HTG, but not Metformin and Berberine, can down-regulate the expression and secretion of KAL and subsequently improve the expression of ATGL and CGI-58 in hepatocytes (Fig. 8i–l). Further, intragastric administration of 100 mg/kg Fenofibrate daily reversed serum TG level, hepatic steatosis, and upregulation of KAL in the livers of HfruD rats (Fig. 8m–p). These results suggest that triglyceride-lowering drugs may benefit NAFLD by decreasing KAL.