Nonneoplastic liver tissues in HBsAg-seronegative HCC patients harboured HBV cccDNA with extensive mutations in PreS/S regions
We recognized 236 HBsAg-seronegative HCC patients after reviewing 1567 cases of histology-confirmed HCCs that had been diagnosed in the China NCC [31] (Additional file 1: Table S4). These HBsAg-seronegative HCC patients presented with HBV-DNA in serum at levels < 1000 IU/ml. We then sampled 35 cases to confirm the presence of replication-competent HBV in these HBsAg-seronegative HCC patients. HBV rcDNA was detectable at a concentration of 102–103 copies/ml in sera of the 35 patients. HBV cccDNA was detected in all (100%) of the nonneoplastic liver tissues, but in 6 (17%) of the paired tumour tissues. An IHC analysis showed that HBsAg was detectable in all 35 nonneoplastic liver tissues and in some HCC tissues (Fig. 1A).
We sequenced the HBV PreS/S regions that were obtained from the blood and from liver tissues, The sequences were deposited in GenBank with accession numbers of MW422170: MW422204. An alignment analysis showed that the amino acid (AA) sequences in PreS/S regions from blood and from liver tissue of each individual were identical (Additional file 1: Fig. S1). Phylogenetic analysis showed that all the isolates were HBV genotype C2. However, the PreS/S AA sequences differed between patients (Additional file 1: Fig. S2). We then compared the PreS/S AA properties of the isolates derived from the 35 HBsAg-seronegative HCC patients with the Ref-HBV, which was obtained from one HBsAg-seropositive HCC patient. The 35 isolates harboured different variations in PreS/S regions, including in-frame deletion or stop codon mutations as previously reported [10, 32]. However, most mutations altered the AA properties (Fig. 1B). Six variant types of PreS/S mutations were identified from 28 patients. The most frequently detected variants harboured mutations with altered AA properties leading to the conversion of hydrophobic properties into hydrophilic properties, and vice versa, and charge modifications (Fig. 1C).
Infection with PreS variants that altered AA properties inhibited HBsAg secretion
HBV envelope proteins perform specific roles relating to the protein transmembrane topology [8, 9]. Because many HBsAg-seronegative HCC patients were infected with HBV variants that altered the PreS AA properties (Fig. 1), we hypothesized that HBV large envelope proteins might change the transmembrane topology that leads to HBsAg retention within hepatocyte ER. Two HBsAg-seronegative HCC patients carried PreS1 E54K, V90A and G102R mutations (named mtPreS1), and one patient carried PreS2 mutations of V32L, I42T, 16-22 (RVRGLYF) deletion (named mtPreS2). We analysed the 3D structure of the large envelope protein using the RaptorX structure prediction server. The structure of the envelope protein encoded by mtPreS1 or mtPreS2 was profoundly different from the structure of the envelope protein encoded by Ref-HBV (Fig. 2A).
To confirm this finding, we constructed the plasmids of mtPreS1 and mtPreS2 based on the Ref-HBV genome. Plasmid of mtPreS1 was mutated only in the PreS1 sequence, and mtPreS2 was mutated only in the PreS2, and both plasmids shared identical S sequences and other sequences with Ref-HBV plasmid (Additional file 1: Fig. S3). Hepatocyte cell lines were transfected with these plasmids to mimic HBV infection. A confocal microscope analysis on the transfected HepG2 cells showed that HBsAg mainly localized to the ER which was positive for calnexin (Fig. 2B) and GRP78 (Additional file 1: Fig. S4). Some HBsAg also localized to the mitochondria (Fig. 2C). We collected the culture medium and harvested transfected cells respectively. ER compartment was separated from the cells by density gradient ultracentrifugation of total cellular protein. HBsAg amount in the supernatant, in the total cells and in the ER was determined using quantitative ELISA. The transfected HepG2 cells generated similar amounts of envelope proteins after transfection with Ref-HBV, or mtPreS1, or mtPreS2. However, more HBsAg was found within cells and in the ER compartment after HepG2 transfected with mtPreS1 or mtPreS2 than in those transfected with Ref-HBV (Fig. 2D). We transfected L02 and Huh7 cells and observed the same results as those obtained in HepG2 cells (Additional file 1: Fig. S5).
Transfection of HBV PreS variants augmented hepatocyte generation of abnormal ceramides
Due to HBsAg retention within the hepatocyte ER, we examined the expression levels of genes related to ER stress. Protein kinase R-like ER kinase (PERK), inositol-requiring enzyme 1α (IRE1α) and activating transcription factor (ATF)-6 are anchored to the ER membrane and initiate three classical pathways to restore cellular protein homeostasis under ER stress conditions [33]. Compared with the results obtained after empty-vector transfection, transfection with mtPreS1, mtPreS2 or Ref-HBV significantly enhanced the gene expression levels of IRE1α in HepG2 cells. The expression of the ER chaperone protein GRP78 and transducer molecules in the IRE1α pathway, including XBP1 and CHOP, was upregulated in cells at both the transcriptional and translational levels after HBV transfection. Notably, compared with Ref-HBV-transfected HepG2 cells, mtPreS1- or mtPreS2-transfected cells expressed significantly higher levels of ER stress-related genes (Fig. 3A). However, no significant alteration in the transcriptional levels of PERK and ATF6 was detected after the HBV transfection (Additional file 1: Fig. S6). These results indicated that the IRE1α-XBP1 pathway, which is key to lipid metabolic homeostasis [33], responded mainly to HBsAg retention in the hepatocytes.
Abnormal accumulation of ceramides is a hallmark of metabolism-related disorders. Ceramides exert profoundly different effects depending on amide-linked fatty acid chain lengths which are regulated by CerS1–CerS6 and DES1. We then quantified the transcriptional levels of CERS1–CERS6, which encode six ceramide synthases, and DEGS1, which encodes DES1. After transfection with different HBV plasmids, HepG2 cells showed significantly elevated transcription of CERS2, CERS6 and DEGS1, particularly in cells transfected with the PreS variants (Additional file 1: Fig. S7A). Using HPLC-MS/MS, we determined the amounts and species of ceramides with various lengths of fatty acyl chains. Compared with empty-vector-transfected cells, HepG2 cells generated significantly more ceramides after HBV transfection, particularly after transfection with mtPreS1 or mtPreS2. The level of the C16:0 ceramide species, the biosynthesis of which is mediated by CERS6 and CERS5, increased 3-fold after Ref-HBV transfection and 5-fold after transfection with either mtPreS1 or mtPreS2. The levels of C22-C24 ceramides, which are the major membrane components of hepatocytes and are regulated by CERS2, increased no more than 3-fold (Fig. 3B, Additional file 1: Table S5). CerS5 is predominantly expressed in the organs of the testis, epididymal white adipose tissue, lung, spleen and thymus [34]. Because the CerS5 transcriptional level was not significantly increased after HBV transfection compared with that in empty-vector-transfected cells (Additional file 1: Fig. S7A), CERS6 and CERS2 might be the main mediators of increases in C16:0 and C22-C24 ceramide species.
Free fatty acids aggravated the generation of abnormal ceramides in HBV-transfected cells
Modern diets are rich in unhealthy fats, and excess lipid exposure imposes significant challenges on the ER action [33]. We treated HBV-transfected HepG2 cells with variated PA concentrations to mimic the overload of saturated fatty acids. Compared with empty-vector-transfected HepG2 cells, the PA treatment led to significant elevations in the transcriptional levels of CERS2, CERS6 and DEGS1 in HBV-transfected cells. In particular, HepG2 cells transfected with either mtPreS1 or mtPreS2 exhibited higher expression of CERS2, CERS6 and DEGS1 than Ref-HBV-transfected cells. The effect depended on the concentration and time period of PA treatment (Fig. 3C). The transfection of Huh7 cell line and L02 cell line with three HBV plasmids resulted in the same outcomes (Additional file 1: Fig. S7B and S7C). The increase in CERS6 and CERS2 protein levels was confirmed by immunoblotting (Fig. 3C).
An HPLC-MS/MS analysis showed that the abundance of certain ceramide species in HepG2 cells without HBV transfection increased after PA treatment. Notably, significantly higher levels of ceramides were generated in HBV-transfected cells than in empty-vector-transfected cells that were treated with the same concentrations of PA, especially the cells that were transfected with PreS variants. Among that of the different ceramide species, the level of C16:0 ceramide was elevated the highest (Fig. 3D). The addition of Myr, a specific inhibitor of de novo ceramide synthesis, partially reversed the increase in abnormal ceramides (Fig. 3D and Additional file 1: Table S5). All these results indicated that the AA property alterations in HBV envelop proteins due to PreS mutations can inhibit HBsAg secretion from infected hepatocytes. HBsAg retention within the ER disturbed hepatocyte lipid metabolism homeostasis to generate abnormal amounts and abnormal species of ceramides.
Infection with PreS variants and free fatty acids synergistically activated the liver macrophage NLRP3 through the abnormal ceramides
Because of their interface in the liver, macrophages constantly receive the signals from hepatocytes. We tested the effects of hepatocyte-derived ceramides on the activation of the macrophage NLRP3 inflammasome. After stimulation with either HBV proteins or ceramides alone, the inflammatory macrophages secreted IL-1β, IL-18 and IL-23. However, the production of IL-1β and IL-18 was synergistically enhanced in the macrophages stimulated by HBV proteins in the presence of ceramides. This synergistic effect depended on the ceramide concentration (Fig. 4A). An immunoblot analysis showed that NLRP3 generation and activation, which is indicated by production of cleaved caspase-1 (p20), were augmented with the dual stimulation of HBV protein and ceramides (Fig. 4B).
To confirm the effects of abnormal ceramides derived from HBV-infected hepatocytes, we treated macrophages with conditioned medium (CM) from HepG2 cells that had been transfected with either an empty-vector or different HBV plasmids (depicted in Additional file 1: Fig. S8). Compared with those cultured in the CM of empty-vector-transfected HepG2 cells, the macrophages treated with the CM of HBV-transfected cells generated significantly higher levels of IL-1β and IL-18. Particularly, the CM of the PreS variant-transfected cells augmented the cytokine production in macrophages to a greater extent than the CM of the Ref-HBV-transfected cells (Fig. 4C). In addition, an immunoblot analysis showed that the CM of PreS variant-transfected cells enhanced the NLRP3 generation and p20 production in macrophages. When de novo ceramide synthesis in HBV-transfected cells was inhibited by myriocin, the stimulatory effect of CM on macrophage NLRP3 was attenuated (Fig. 4D).
Infection with HBV PreS variants induced hepatic steatosis related to abnormal ceramide generation in the liver
To investigate the impacts of the HBV PreS mutations that cause AA property alterations of envelop proteins in vivo, we infected the mouse livers through administering same amounts of empty-vector, Ref-HBV, mtPreS1 or mtPreS1 plasmids to male C57BL/6J mice by intravenous hydrodynamic injection. One week after plasmid injection, we removed 3 mice from each group to confirm the HBV transfection in the mouse livers. The mice receiving same HBV plasmid injection were allocated to two subgroups and fed NC or HF diets for 5 weeks (Fig. 5A). Compared with Ref-HBV-injected mice, mtPreS1- or mtPreS2-injected mice displayed relatively low serum HBsAg levels but higher HBsAg levels in the liver (Fig. 5B).
Beginning with different diet types, two subgroups of Ref-HBV-injected mice showed similar serum HBsAg levels. The two subgroups of mice that were injected with mtPreS1 or mtPreS2 also showed similar serum HBsAg levels (Additional file 1: Fig. S9). All mice were sacrificed at 14 weeks old, and the body weights of HF diet mice were greater than NC diet mice. With the same diet type, the body weights did not differ between the mice that were injected with empty-vector and those that were injected with Ref-HBV or mtPreS1 or mtPreS2 (Additional file 1: Fig. S10). However, compared with empty-vector-injected mice, all the mice injected with HBV plasmids presented with a greater ratio of liver weight to body weight, particularly those injected with PreS variants, including the NC diet mice (Fig. 5C).
In the H&E-stained liver sections prepared from mice injected with HBV plasmids and fed the NC diet, some lipid deposition was observed. We then performed Oil red O staining to detect the evidence of hepatic steatosis. Among the NC diet mice, the lipid deposition was detectable in the HBV plasmid-injected mouse livers, but did not significantly differ in the livers of mice that were injected with Ref-HBV and those that were injected with mtPreS1 or with mtPreS2. However, among the HF diet mice, the livers of mice that were injected with HBV plasmids showed aggravated steatosis compared to those with empty-vector, as indicated by larger areas stained with Oil red O. The numbers of infiltrated inflammatory cells, which were stained positively for leukocyte common antigen CD45, were higher in the HBV-injected mice than in the empty-vector-injected mice. Notably, the mice injected with either mtPreS1 or mtPreS2 presented more severe steatosis and inflammation than those injected with Ref-HBV (Fig. 5D and Additional file 1: Fig. S11). We then analysed the expression of the CerS2, CerS6 and Degs1 genes, which regulate the generation of different ceramide species, in the mouse livers. Compared with those in the empty-vector-injected mouse livers, the transcriptional levels of these genes were significantly elevated in the HBV plasmid-injected mice, particularly in the mice injected with PreS mutants, regardless of diet type (Fig. 5E). An HPLC-MS/MS analysis showed that the total ceramide amounts increased and that the C16:0 ceramide level was the most highly elevated among different species of ceramides, although the levels of C22-24 ceramides also increased. When the mice were fed the HF diet, CerS2, CerS6 and Degs1 transcription and ceramide generation were further enhanced. The elevation of C16:0 ceramide abundance was the most significant (Fig. 5E and Additional file 1: Fig. S12).
We intraperitoneally injected myriocin to inhibit de novo ceramide synthesis in the HF-fed mice. The liver weight/body weight rate showed no significant reduction (Fig. 5C). However, the gene expression of CerS2, CerS6 and Degs1 was decreased, and the increase in C16:0 ceramide and C22-24 ceramide levels was partially decreased (Fig. 5E). The infiltration of inflammatory cells was attenuated (Fig. 5D).
HBV PreS variant injection and a HF diet synergistically promoted autochthonous HCC development associated with abnormal ceramide generation
Before different HBV plasmids were injected, some mice were given DEN by intraperitoneal injection to induce HCC progenitor cells [15] (Fig. 5A). We detected liver tumour nodules only in the mice that were injected with HBV plasmids and fed the HF diet. The tumour loads were higher in the mice injected with PreS mutants than in those injected with Ref-HBV (Fig. 6A). When the mice were injected with HBV plasmids and fed the NC diet, no mouse developed tumour. Moreover, no tumour was observed in the empty-vector-injected mice fed the HF diet (Additional file 1: Figs. S11B and S13). Additionally, as determined upon sacrifice at 14 weeks old, none of the mice developed liver tumour regardless of their diet types after HBV plasmid injection without DEN injection (Additional file 1: Fig. S14). To validate the effect of the abnormal ceramides on HCC development, a subset of mice fed the HF diet received myriocin treatment (Fig. 5A). No tumours were detected in mice after ceramide de novo synthesis was inhibited (Fig. 6A). These results indicated that hepatocyte infection with HBV, particularly with PreS variants, synergistically promoted HCC development in the context of HF diet. The synergistic effect of HF diet and HBV infection on HCC promotion was associated with abnormal ceramide generation.
HF diet promoted inflammatory macrophage infiltration and activation in livers of the mice infected with HBV PreS variants
Because of the evidence of inflammatory cell infiltration (Fig. 5D), we analysed the effect of different diet types on the infiltration of inflammatory macrophages in the mouse livers. Intrahepatic infiltrating cells were prepared from tumour-free liver tissues of mtPreS2-injected mice that were fed different diets. A FCM analysis showed that the HF diet tripled the infiltration of CD11b+F4/80+ inflammatory macrophages compared with the number of infiltrating macrophages in mice fed the NC diet (Fig. 6B). Moreover, we detected significantly higher amounts of IL-1β and IL-18 in the livers of mice fed the HF diet than in those fed the NC diet (Fig. 6C). Myriocin administration in mice fed the HF diet alleviated macrophage infiltration and inflammatory cytokine generation in the liver (Fig. 6B, C).
We then purified the liver macrophages through 25%/50% Percoll gradient centrifugation. An immunoblot analysis showed that liver macrophages in the PreS variant-injected mice activated macrophage NLRP3, including in the NC diet fed mice. Notably, the macrophages in the HBV-injected mice generated more NLRP3 and caused higher levels of NLRP3 activation that was indicated by more amounts of p20 production after HF diet, especially the macrophages in the mice injected with the PreS variants. Myriocin administration attenuated NLRP3 activation in mice fed the HF diet (Fig. 6D).
Blocking NLRP3 activation of inflammatory macrophages significantly reduced autochthonous HCC development
Hepatocytes also express NLRP3. To confirm the effects of NLRP3 in liver inflammatory macrophages on autochthonous HCC development, we transferred bone marrow cells obtained from Nlrp3−/− mice into mtPreS2-injected mice. The mice were fed different diet types for 20 weeks after bone marrow cell transfer. The liver tumour loads were significantly reduced in the mice that received the Nlrp3¬/− bone marrow compared with those in the mice that received C57BL/6J bone marrow cells. The reduction of liver tumour loads in the Nlrp3−/− bone marrow receipt mice was more profound in those that were fed the HF diet than in those that were fed the NC diet (Fig. 6E). These results indicated the importance of NLRP3 inflammasome in liver macrophages for promoting autochthonous HCC development.