Overproduction of thrombopoietin by BRAFV600E-mutated mouse hepatocytes and contribution of thrombopoietin to hepatocarcinogenesis.

In hepatocarcinogenesis induced by diethylnitrosamine (DEN) in B6C3F1 mice, the BrafV637E mutation, corresponding to the human BRAFV600E mutation, plays a pivotal role. The livers of transgenic mice with a hepatocyte-specific human BRAFV600E mutation weighed 4.5 times more than that of normal mice and consisted entirely of hepatocytes, resembling DEN-induced preneoplastic hepatocytes. However, these transgenic mice spontaneously died 7 wk after birth, therefore this study aimed to clarify the causes of death. In the transgenic mice, the liver showed thrombopoietin (TPO) overexpression, which is associated with eventual megakaryocytosis and thrombocytosis, and activated platelets were deposited in hepatic sinusoids. TPO was also overexpressed in the DEN-induced hepatic tumors, and sinusoidal platelet deposition was observed in the hepatic tumors of humans and mice. Podoplanin was expressed in some of the Kupffer cells in the liver of the transgenic mice, indicating that platelet activation occurred via the interaction of podoplanin with C-type lectin receptor 2 (CLEC-2) on the platelet membrane. Additionally, erythrocyte dyscrasia and glomerulonephropathy/interstitial pneumonia associated with platelet deposition were observed. In the transgenic mice, aspirin (Asp) administration prevented platelet activation, reduced the liver/body weight ratio, decreased the platelet deposition in the liver, kidney, and lung, and prevented erythrocyte dyscrasia and ameliorated the renal/pulmonary changes. Thrombopoietin overproduction by BRAFV600E-mutated hepatocytes may contribute to hepatocyte proliferation via thrombocytosis, platelet activation, and the interaction of platelets with hepatic sinusoidal cells, while hematologic, renal, and pulmonary disorders due to aberrant platelet activation may lead to spontaneous death in the transgenic mice.

Albumin‐Cre‐recombinase

aspirin

C57BL/6

C‐type lectin receptor 2

diethylnitrosamine

hepatocellular carcinoma

liver sinusoidal endothelial cells

platelet factor 4

thrombopoietin

INTRODUCTION

Hepatocarcinogenesis starts with the generation of carcinogenically altered hepatocytes. Following liver damage caused by chronic viral hepatitis, metabolic steatosis, or alcohol‐induced injury, carcinogenically altered hepatocytes are forced to expand by the cytokine, chemokine, and growth factor milieu derived from chronic hepatic injury, inflammation, and restorative proliferation, resulting in the formation of preneoplastic nodules from which early hepatocellular carcinomas (HCCs) initiate, eventually leading to progressed and advanced HCC.1

The mouse model of hepatic tumor induced by neonatal treatment with diethylnitrosamine (DEN) has been often used for investigation of hepatocarcinogenesis.2 In this model, carcinogenically altered preneoplastic hepatocytes emerged at the earliest stage and eventually progressed into preneoplastic foci, adenomas, and HCC.3 The carcinogenic process is driven by either an H‐ras or a Braf mutation.4, 5, 6 The role of the mutation is dependent on the animal's genetic background: in hepatocarcinogenesis‐sensitive C3H mice, the H‐ras codon 61 mutation is prevalent, while in hepatocarcinogenesis‐resistant C57BL/6 (B6) mice, the BrafV637E mutation is preferentially selected.7

We have previously reported that Albumin‐Cre‐recombinase (Alb‐Cre)/BRAFV600E transgenic mice that express the human BRAFV600E mutation, corresponding to the mouse BrafV637E mutation, specifically in hepatocytes from the B6 mouse genetic background showed a marked increase in liver size (four‐ to five‐fold larger liver/body weight ratio than normal mice), and the liver consisted entirely of hepatocytes that resembled DEN‐induced preneoplastic hepatocytes.8 This model therefore enabled us to investigate biological events that occurred within the early preneoplastic lesions induced by neonatal treatment with DEN. However, although these mice were healthy at birth, they spontaneously died at 7 wk of age due to unknown reason(s).

In this study, we aimed to clarify the cause of death in the BRAFV600E transgenic mice by analyzing the blood and tissue specimens isolated from the transgenic mice. Our current findings showed that the hepatocytes in Alb‐Cre/BRAFV600E transgenic mice overproduced thrombopoietin (TPO), and this overproduction eventually resulted in megakaryocytosis and thrombocytosis; platelets were activated in the peripheral blood; and large numbers of platelets were deposited in liver sinusoids. Furthermore, podoplanin, which can activate platelets via C‐type lectin receptor 2 (CLEC‐2) on the platelet membrane,9, 10, 11 was expressed in some of the Kupffer cells in the liver of Alb‐Cre/BRAFV600E transgenic mice. Additionally, these mice showed erythrocyte dyscrasia, glomerulonephropathy, and interstitial pneumonia, which were thought to be derived from the deposition of aberrantly activated platelets. Our data suggested that the TPO‐mediated macroenvironmental changes involving the liver and bone marrow may promote proliferation of the BRAFV600E‐mutated hepatocytes, but cause hematological, renal, and pulmonary disorders that lead to death.

MATERIALS AND METHODS

Experimental animals

Male Alb‐Cre/BRAFV600E transgenic (5–13 wk of age), and age‐matched control BRAFV600E and C57BL/6 mice (Charles River Japan) were used in this study. Alb‐Cre/BRAFV600E transgenic and control BRAFV600E mice were generated by in vitro fertilization using homozygous male BRAFV600E mice [B6.129P2(Cg)‐Braftm1Mmcm>/J; Jackson] and heterozygous female Alb‐Cre mice (Jackson) as described previously.8 The male litters were genotyped to separate the Alb‐Cre/BRAFV600E transgenic and control BRAFV600E mice with and without the Alb‐Cre gene. Some of the Alb‐Cre/BRAFV600E transgenic and control BRAFV600E mice were given aspirin (Asp) at a daily dose of 5 μg/g body weight by oral intubation from 4 to 8 wk after birth. Hepatic tumors were induced by neonatal treatment with 5 μg/g body weight of DEN on day 15 after birth in male B6C3F1 mice (Charles River Japan), and the tumor samples were collected at 8‐12 mo of age. The mice were housed in plastic cages with sterilized wood chips and given a standard chow diet (CMF, Oriental Yeast) and sterilized water ad libitum. The mouse breeding room was kept at 25°C with a 12 h light and 12 h dark cycle. All procedures performed on mice were approved by the Asahikawa Medical University Animal Experiment Committee following the guidelines for the humane care and protection of animals.

Human materials

Paraffin‐embedded human hepatic tumor materials taken from patients who received hepatic biopsy or hepatectomy surgery during 2015‐2018 in the affiliated hospital of Asahikawa Medical University were collected. All patients gave informed consent, and the study was approved by the ethical committee at the hospital.

Antibodies

The antibodies used were against CD31 (Novus), CD61 (Cell Signaling), F4/80 (R&D Systems), podoplanin (R&D Systems), TPO (LSBio), Stat3 (Cell Signaling), S727 phospho‐Stat3 (Cell Signaling), Y705 phospho‐Stat3 (Cell Signaling) and α‐tubulin (Novus).

Hematological analysis

Complete blood count analyses were performed using an automated hematology analyzer (Sysmex). Peripheral blood and bone marrow smear samples were processed for May‐Giemsa staining and examined under a microscope.

Hepatocyte culture

Hepatocytes were isolated by collagenase perfusion from Alb‐Cre/BRAFV600E and normal B6 mice at the age of 8 wk, purified by low speed centrifugation and cultured in Williams’ E medium supplemented with 10% fetal bovine serum and antibiotics for 12 h, followed by incubation with serum‐free Williams’ E medium for 24 h to collect the conditioned medium.

Enzyme‐linked immunosorbent assay

Platelet factor 4 (PF4) and TPO levels were analyzed using enzyme‐linked immunosorbent assay kits (R&D Systems) according to the manufacturer's instructions.

Real‐time quantitative reverse transcription (qRT)‐polymerase chain reaction (PCR)

RNA was extracted from liver tissue using an RNeasy Mini Kit (Qiagen). cDNA was reverse‐transcribed from RNA using a high‐capacity cDNA reverse transcription kit (Thermo Fisher). The expression levels of mouse TPO mRNA were then analyzed using a real‐time PCR system (Thermo Fisher) with the TaqMan probe for mouse TPO mRNA (Thermo Fisher). mRNA expression was normalized to 18S rRNA expression.

Western blotting

Tissue samples were lysed in RIPA buffer, separated by polyacrylamide gel electrophoresis and electrotransferred to nitrocellulose membranes. The membranes were probed with the primary antibodies and then incubated with a horseradish peroxidase (HRP)‐conjugated anti‐mouse or anti‐rabbit IgG secondary antibody (R&D Systems). Antibody binding was visualized using the SuperSignal West Pico chemiluminescent substrate (Thermo Fisher).

Histopathology, immunohistochemistry, and immunofluorescence

Tissues were fixed in 10% formalin in phosphate‐buffered saline (PBS), paraffin‐embedded and stained with hematoxylin and eosin (H&E). For live mice, the liver was perfused with PBS via the portal vein at 2‐3 mL/min under ether anesthesia, followed by perfusion‐fixation with 10% formalin in PBS. For immunohistochemistry, after deparaffinization, rehydration, and antigen retrieval, the tissue sections were incubated with the primary antibodies, followed by incubation with an HRP‐conjugated anti‐mouse or anti‐rabbit IgG secondary antibody (Vector). For immunofluorescence, after quenching autofluorescence with the MaxBlock™ Autofluorescence Reducing Kit (Maxvision), the tissue sections were incubated with the primary antibodies and then with an Alexa Fluor (598/488)‐conjugated anti‐mouse or rabbit IgG secondary antibody (Thermo Fisher). Fluorescence images were analyzed using a fluorescence microscope (BZ‐X700, Keyence).

Electron microscopy

Renal tissues were fixed with a cacodylate‐buffered glutaraldehyde solution followed by osmic acid fixation, dehydration, epoxy resin embedding, and ultrathin sectioning. The sections were examined using an electron microscope (JEOL).

Statistics

The differences in the experimental values of Alb‐Cre/BRAFV600E transgenic mice and those of control BRAFV600E or normal B6 mice were statistically analyzed using Student's paired t test or the chi‐squared test.

RESULTS

Spontaneous death and hematological changes in the Alb‐Cre/BRAFV600E transgenic mice

Although the Alb‐Cre/BRAFV600E transgenic mice were healthy at birth, 8 of 18 (44.4%) mice spontaneously died between 7 and 13 wk after birth (Figure 1). To clarify the cause of death, we not only performed autopsies on the dead mice but also sacrificed the live mice to collect blood and tissue samples between 5 and 13 wk after birth. The complete blood count analysis for the live mice 8 wk after birth revealed that platelet counts were significantly increased in Alb‐Cre/BRAFV600E transgenic mice compared with the control BRAFV600E mice (Figure 2A) or normal B6 mice (data not shown). There were no significant differences in the white blood cell and red blood cell counts and the hemoglobin and hematocrit levels between the Alb‐Cre/BRAFV600E transgenic and control BRAFV600E mice (data not shown). Consistent with thrombocytosis in the peripheral blood, the number of megakaryocytes was increased 10‐fold in the bone marrow smear samples and tissue sections of the bone marrow and spleen in the Alb‐Cre/BRAFV600E transgenic mice compared with the control BRAFV600E mice (Figure 2B‐H). A few mitotic megakaryocytes were observed in the bone marrow of the Alb‐Cre/BRAFV600E transgenic mice, but not in the control or normal mice (Figure 2F inset). We also observed schizocytes (fragmented or abnormally shaped red blood cells) and enlarged platelets in the peripheral blood smears of the Alb‐Cre/BRAFV600E transgenic mice (Figure 2I). Urine analyses revealed a positive reaction for hemoglobin, presumably due to hemoglobinuria caused by red blood cell destruction as well as hematuria derived from glomerulonephropathy as described below (Figure 2J). Because schizocytes are generated in association with thrombotic microangiopathy characterized by aberrant platelet activation and formation of fibrin mesh within blood,12 we investigated the plasma levels of PF4, which is released from α‐granules in association with platelet activation.13 The PF4 levels were elevated in most Alb‐Cre/BRAFV600E transgenic mice (Figure 2K).

Figure 1

A Kaplan‐Meier plot of the survival rate in control 600E mice (control) and Alb‐Cre/ 600E transgenic mice (TG) until 13 wk after birth; *P < .01

Figure 2

Hematological changes in Ab‐Cre/ 600E transgenic mice. A‐K, control; 8‐wk‐old control 600E mice, TG; 8‐wk‐old Ab‐Cre/ 600E transgenic mice. A, Increased numbers of platelets in TG. *P < .05. B‐G, Increase in the numbers of megakaryocytes in TG, (B, E) bone marrow smears, (C, F) bone marrow and (D, G) spleen tissues. Inset in F: mitotic display of a megakaryocyte. B, E, May‐Giemsa staining; C, D, F, G, H&E staining; scale bars: 200 μm. H, Increased numbers of megakaryocytes in bone marrow in TG; *P < .01. I, Peripheral blood smear. Schizocytes (arrows) and an enlarged platelet (*) in TG. J, Positive urinary hemoglobin test in TG. K, Increased platelet factor 4 (PF4) levels in TG; *P < .05

TPO overproduction in the liver of Alb‐Cre/BRAFV600E transgenic mice

Thrombopoietin stimulates the differentiation of megakaryocytes from hematopoietic stem cells and the generation of platelets from megakaryocytes.14 TPO is mainly produced by hepatocytes and interacts with the c‐Mpl receptor expressed on megakaryocytic lineage cells in the bone marrow.14 qRT‐PCR analysis revealed that liver TPO mRNA levels were six‐fold higher in the Alb‐Cre/BRAFV600E transgenic mice than in the normal B6 mice (Figure 3A). TPO protein levels were also increased in the liver and plasma of the Alb‐Cre/BRAFV600E transgenic mice compared with the normal B6 mice (Figure 3B,C). Furthermore, higher levels of TPO were detected in conditioned medium collected from the cultured hepatocytes of the Alb‐Cre/BRAFV600E transgenic mice than in medium from cultured hepatocytes from normal B6 mice (Figure 3D). TPO production by hepatocytes is upregulated by transcriptional activation of the TPO gene via Stat3 activation.15, 16 Stat3 was hyperphosphorylated in the liver of the Alb‐Cre/BRAFV600E transgenic mice compared with the normal B6 mice (Figure 3E), similar to findings for DEN‐induced hepatic tumors, as previously reported.17

Figure 3

Thrombopoietin (TPO) overproduction and Stat3 activation in the liver of Alb‐Cre/ 600E transgenic mice. A‐E, normal; normal B6 mice, TG; Alb‐Cre/ 600E transgenic mice. A, Increased TPO mRNA levels in the TG liver; *P < .01. B, Increased TPO protein in the TG liver. TPO western blot analysis with α‐tubulin as a loading control. The numbers below represent the relative density of TPO to α‐tubulin bands. C, Increased TPO levels in TG plasma; *P < .01. D, Increased TPO levels in the hepatocyte conditioned medium (CM) in TG; *P < .05. E, Stat3 activation in the TG liver. Western blot analysis of the Stat3 phosphorylation status. The numbers below represent the relative density of phospho‐Stat3 (pSTAT3) to total Stat3

Platelet deposition in hepatic sinusoids in the Alb‐Cre/BRAFV600E transgenic mice

Because platelets can contribute to hepatocyte proliferation via interaction with hepatic sinusoidal cells,18, 19 we investigated using immunohistochemistry the interaction of platelets with sinusoidal cells in the liver, which was perfused with PBS and 10% formalin in PBS to flush out blood. Hepatocytes from Alb‐Cre/BRAFV600E transgenic mice were characteristically basophilic and small in size, similar to DEN‐induced preneoplastic hepatocytes20, in contrast with those in the normal mice (Figure 4A,B). CD61 (marker of platelets) staining revealed that many platelets were deposited, either sparsely or densely, on the sinusoidal walls in the liver of Alb‐Cre/BRAFV600E transgenic mice, in contrast with normal mice (Figure 4C,D‐F). Simultaneous staining for CD61 and CD31 (the marker for sinusoidal endothelial cell [LSEC]) revealed that platelets sparsely adhered to LSECs (Figure 4G‐J), while simultaneous CD61 and F4/80 (marker for Kupffer cells) staining showed that platelets densely adhered to and were incorporated into Kupffer cells (Figure 4K‐N).

Figure 4

Platelet deposition in the liver sinusoids of the Alb‐Cre/ 600E transgenic mice. A, C, The liver of 8‐wk‐old normal B6 mice (normal), and (B, D‐N) the liver of 8‐wk‐old Alb‐Cre/ 600E transgenic mice (TG). A, B, The TG liver consisting of small basophilic hepatocytes in contrast with the normal liver. C, D‐F, Increased platelet deposition in the liver sinusoids (D), either sparse (E) or densely (F) in the TG liver in contrast to the normal liver (C). G‐J, Sparse platelet deposition on the CD31‐positive liver sinusoidal endothelial cells (LSEC). K‐N, Dense platelet adherence/incorporation to the F4/80‐positive Kupffer cells in the TG liver. A, B, H&E staining, (C‐F) CD61 immunohistochemical staining, (G‐J) Immunofluorescence of CD31 (green) and CD61 (red). K‐N, Immunofluorescence of F4/80 (green) and CD61 (red). Scale bars: 200 μm in A‐D, and 50 μm in E, F

Podoplanin‐positive Kupffer cells in the liver of Alb‐Cre/BRAFV600E transgenic mice

The interaction of podoplanin with CLEC‐2 on platelet membranes is an important mechanism to reciprocally activate platelets and the podoplanin‐expressing cells.9, 10, 11 Furthermore, podoplanin is expressed in activated macrophages.21, 22, 23, 24 Simultaneous staining for podoplanin and F4/80 revealed that approximately 5% of F4/80‐positive Kupffer cells in the liver sections of the Alb‐Cre/BRAFV600E transgenic mice were positive for podoplanin, while no such cells were observed in the liver sections of the normal B6 mice (Figure 5A‐H).

Figure 5

Expression of podoplanin in some of the Kupffer cells in the liver of Alb‐Cre/ 600E transgenic mice. A, C, E, G, normal; normal B6 mice and B, D, F, H, TG; Alb‐Cre/ 600E transgenic mice. A‐H, Immunofluorescence of podoplanin (green) and F4/80 (red). Asterisks and insets in B, D, F, H, F4/80 and podoplanin +/+ Kupffer cells in TG liver

Other histopathological changes in the Alb‐Cre/BRAFV600E transgenic mice

Histopathological analysis was performed on whole organs, including central nervous system (brain), respiratory (lung), digestive (alimentary tract, liver, pancreas and biliary tract), urinary (kidney and urinary bladder), endocrine (thyroid and adrenal gland), salivary (submandibular gland) and sexual organs (testis), in all of the Alb‐Cre/BRAFV600E transgenic mice between 5 and 13 wk after birth, including those that died spontaneously. The most conspicuous changes were noted in the kidney, lung, and liver. At 5‐8 wk after birth, the renal glomerular capillaries were congested with red blood cells in the Alb‐Cre/BRAFV600E transgenic mice, in contrast with age‐matched normal mice (Figure 6A,B). Immunohistochemistry for CD61 showed that many platelets were deposited in the glomeruli in the Alb‐Cre/BRAFV600E transgenic mice, in contrast with the normal mice (Figure 6C,D). Electron microscopy revealed that platelets adhered to the glomerular endothelial cells, and podocyte foot processes were swollen and effaced in the Alb‐Cre/BRAFV600E transgenic mice, in contrast with the normal mice (Figure 6E,F). At 13 wk after birth, cellular and extracellular matrix components were increased in Bowman's capsules, forming a crescent‐shaped deposit in some of the glomeruli (Figure 6G). In lung tissue, the alveolar space was narrowed by the thickening of the alveolar septa due to congestion with red blood cells and infiltration with mononuclear cells at 5‐8 wk in the Alb‐Cre/BRAFV600E transgenic mice, in contrast with the normal B6 mice (Figure 6H,I). CD61 staining revealed that large numbers of platelets were deposited in the alveolar septa in Alb‐Cre/BRAFV600E transgenic mice compared with normal B6 mice (Figure 6J,K). At 13 wk after birth, the alveolar septa were thickened even more, and the alveolar space disappeared in a large area of the lungs in the Alb‐Cre/BRAFV600E transgenic mice (Figure 6L). In addition, necrotic foci of 1‐2 mm in diameter were observed in the liver of some Alb‐Cre/BRAFV600E transgenic mice, presumably due to thrombotic embolization in the hepatic vessels (Figure 6M,N), but not in normal mice (data not shown). Although the degree of histopathological change in the kidney, lung, and liver was diverse in individual Alb‐Cre/BRAFV600E transgenic mice, and changes tended to become more severe with age. These histopathological changes, however, were not observed in the control BRAFV600E mice or the normal B6 mice. There were no remarkable histological changes in other organs or tissues in the Alb‐Cre/BRAFV600E transgenic mice (data not shown).

Figure 6

Histopathological changes in kidney, lung and liver tissues in Alb‐Cre/ 600E transgenic mice. Normal: 8‐wk‐old normal B6 mice and TG; 8‐wk‐old or 12‐wk‐old Alb‐Cre/ 600E transgenic mice. A‐G, The kidney, (H‐L) lung and (M, N) liver tissues. Opened glomerular capillary space (A), few platelet stain (C) and thin endothelial cells and podocyte processes (arrows) (E) in normal renal glomeruli. Congested capillary (B), large numbers of platelet stain (D) and swollen/effaced podocyte processes (arrows) and adherence of platelets to the capillary walls (asterisks) (F) in TG renal glomeruli 8 wk after birth. Crescent formation in Bowman's capsule in TG 12 wk after birth (G). Narrow alveolar septa (H) and few platelet stain in normal lung (J). Thickened alveolar septa (I) and increased platelet stain in the TG lung 8 wk after birth (K). Disappearance of alveolar space in the TG lung 12 wk after birth (L). Focal necrosis in the TG liver 8 wk after birth (M, N). A, B, G, H, I, L, N, H&E staining, C, D, J, K, CD61 immunohistochemistry and (E, F) electron microscopy. Scale bar: 100 μm in A, B, C, D, G, J and K, 200 μm in H, I and L, and 400 μm in N

Effect of Asp on the platelet activation/deposition and histopathological changes

We then investigated whether Asp, an inhibitor of platelet activation via cyclooxygenase inhibition,25 can reduce platelet activation/deposition and ameliorate the histopathological changes in Alb‐Cre/BRAFV600E transgenic mice. A daily dose of 5 μg/g body weight Asp from 4 to 8 wk after birth decreased the PF4 levels to 35% compared with the Asp‐untreated Alb‐Cre/BRAFV600E transgenic mice (Figure 7A). Although the liver/body ratio in the Asp‐untreated Alb‐Cre/BRAFV600E transgenic mice was 4.5‐fold greater than that of the control BRAFV600E mice, as reported previously,8 Asp treatment reduced the ratio to 70% in Alb‐Cre/BRAFV600E transgenic mice, but not in the control BRAFV600E mice (Figure 7B). Histopathological analysis revealed that hepatocytes in the periportal areas of the hepatic lobes in the Asp‐treated Alb‐Cre/BRAFV600E transgenic mice morphologically resembled normal hepatocytes (Figure 7C‐E), but such normal‐looking hepatocytes were not observed in the Asp‐untreated Alb‐Cre/BRAFV600E transgenic mice. CD61 immunostaining revealed that the degree of platelet deposition was lower in the normal‐looking hepatocyte areas than in the small basophilic hepatocyte areas (Figure 7F,G). Asp administration ameliorated the renal/pulmonary changes, decreased platelet deposition in renal glomeruli and pulmonary septa (Figure 7H‐K) and prevented erythrocyte dyscrasia in the Alb‐Cre/BRAFV600E transgenic mice (Figure 7L).

Figure 7

Effect of Asp on platelet activation/deposition and pathological changes in the Alb‐Cre/ 600E transgenic mice. A, B, control; control 600E mice and TG; Alb‐Cre/ 600E transgenic mice. A, Decreased platelet factor 4 (PF4) levels in Asp‐treated TG in comparison with Asp‐untreated TG. *P < .05. B, Decreased liver/body weight ratio in Asp‐treated TG in comparison with Asp‐untreated TG. *P < .05. C‐E, Normal‐looking hepatocytes (C arrowheads and D) together with small basophilic hepatocytes (E) in the liver of Asp‐treated TG. F, G, Reduced platelet deposition in the normal‐looking hepatocyte area (F) compared with the small basophilic hepatocyte area (G) in the liver of Asp‐treated TG. H‐K, Amelioration of the renal glomerular (H) and pulmonary interstitial changes (J) associated with reduced platelet deposition (I, K) in Asp‐treated TG. C‐E, H, J: H&E staining. F, G, I, K: CD61 immunostaining. Scale bars: 200 μm in C, F‐K and 100 μm in D, E. L, No erythrocyte dyscrasia in Asp‐treated TG

Overexpression of TPO in DEN‐induced hepatic tumors and platelet deposition in sinusoids in hepatic tumors of human and mice

We lastly investigated whether TPO overproduction and platelet deposition in the sinusoids are also observed in DEN‐induced hepatic tumors collected 6‐12 mo after neonatal DEN treatment. As shown in Figure 8A, TPO protein levels were higher in the DEN‐induced hepatic tumors compared with the surrounding normal hepatic tissues. Additionally, platelets were adhered to the sinusoidal cells either sparsely or densely in DEN‐induced hepatic tumors (Figure 8B‐E). The numbers of platelets in sinusoids were 0.37 ± 0.08/μm2 in surrounding normal liver (n = 3), 1.32 ± 0.06/μm2 in foci/adenoma (n = 3) and 1.23 ± 0.35/μm2 in HCC (n = 3). We further investigated whether sinusoidal platelet deposition was also seen in dysplastic/neoplastic hepatic lesions in humans. CD61 immunostaining revealed that sinusoidal platelet deposition was increased in dysplastic nodules (1.05 ± 0.2/μm2, n = 5) and well differentiated HCC (1.13 ± 0.31/μm2, n = 5) compared with surrounding normal hepatic tissues (0.25 ± 0.06/μm2, n = 5) and cirrhotic nodules (0.27 ± 0.07/μm2, n = 5), but not in moderately/poorly differentiated HCC (0.51 ± 0.1/μm2, n = 5) (Figure 8F‐O).

Figure 8

Thrombopoietin (TPO) overproduction in diethylnitrosamine (DEN)‐induced hepatic tumors and platelet deposition in sinusoids in hepatic tumors in human and mice. A, Western blot analysis of TPO in the normal hepatic tissue (normal) and DEN‐induced hepatic tumors collected 8‐12 mo of age (DEN tumors). The numbers below represent the relative density of TPO to α‐tubulin bands. Increased numbers of platelets deposited in the sinusoids of DEN tumors (C), either sparsely (D) or densely (E). Increased platelet deposition in sinusoids in the human dysplastic nodules (H, M) and well differentiated HCC (I, N) compared with normal liver (F, K) and regenerating cirrhotic nodules (G, L), but not in poorly differentiated HCC (J, O). B‐E, K‐O: Immunohistochemical staining of CD61. F‐J, H&E staining. Scale bars: 100 μm in B, C, F –O and 25 μm in D, E

DISCUSSION

In the present study, we found that mouse hepatocytes expressing the human BRAFV600E mutation, corresponding to the mouse BrafV637E mutation that is highly prevalent in DEN‐induced hepatic tumors,8 overproduced TPO, which led to megakaryocytosis and thrombocytosis, and that large numbers of platelets were deposited on hepatic sinusoids. TPO overproduction and sinusoidal platelet deposition were also observed in DEN‐induced hepatic tumors, indicating that the observed phenomena in the Alb‐Cre/BRAFV600E transgenic mice reflected the properties of DEN‐induced hepatic tumors. Furthermore, sinusoidal platelet deposition was commonly seen in human dysplastic nodules and well differentiated HCC. This study suggested that thrombocytosis, platelet activation, and interaction of platelets to sinusoidal cells promotes hepatocarcinogenesis from an early stage via TPO overproduction by preneoplastic/neoplastic hepatocytes. These results are consistent with previous reports that HCC patients occasionally present thrombocytosis, which is correlated with an unfavorable prognosis, when TPO is overproduced by HCC cells.26, 27, 28

Thrombopoietin production by hepatocytes is enhanced by transcriptional activation of the TPO gene via the Stat3 pathway.15, 16 In our present study, we found that Stat3 was hyperphosphorylated, as observed in the DEN‐induced hepatic tumors.17 Although the exact reason(s) for Stat3 activation is not known at present, it is possible that factors derived from activated platelets and/or sinusoidal cells might activate Stat3 in hepatocytes. It is also possible that, because DEN‐induced hepatic tumor cells in which the Braf mutation is highly prevalent express various growth factors, cytokines, chemokines and their receptors,8, 29, 30, 31, 32, 33 these autocrine factors and their receptors may contribute to Stat3 activation.

Platelets adhered sparsely to LSECs and densely adhered to and were incorporated into Kupffer cells in the livers of Alb‐Cre/BRAFV600E transgenic mice, which has also been reported in the regenerating liver.18, 19 Because the liver was perfused with PBS and formalin solution to flush out blood cells, platelets were tightly adhered to the sinusoidal cells in the liver of Alb‐Cre/BRAFV600E transgenic mice. Compelling evidence indicates that platelets play an important role in liver homeostasis and pathobiology by interacting with hepatic cells, including LSECs, Kupffer cells, hepatic stellate cells, and hepatocytes, and by releasing bioactive substances.18, 19 The factors derived from both platelets and platelet‐activated sinusoidal cells can stimulate various intracellular signaling pathways in hepatocytes, including the NF‐κB, Stat3, MAPK/ERK and PI3K/Akt pathways.18, 19 The sinusoidal platelet deposition indicated that platelets contribute to hepatocyte proliferation in the Alb‐Cre/BRAFV600E transgenic mice as well as in hepatic tumor cells.

In the Alb‐Cre/BRAFV600E transgenic mice, platelets were thought to be activated in the peripheral blood because plasma PF4 levels were elevated and because some of the platelets were increased in size, which is the characteristic feature of activated platelets.34 Conversely, schizocytes were observed in the peripheral blood smear samples, indicating that erythrocytes were damaged. Aberrant platelet activation causes thrombotic microangiopathy, which leads to erythrocyte dyscrasia.12 However, the platelet numbers are usually decreased in thrombotic microangiopathy,12 and the TPO levels are commonly decreased in association with thrombocytosis because TPO is bound to the platelet c‐Mpl receptor, internalized, and destroyed.35 In the Alb‐Cre/BRAFV600E transgenic mice, however, continuous TPO overproduction by BRAFV600E‐mutated hepatocytes might maintain the high circulating TPO levels and increase the number of platelets.

In the liver of Alb‐Cre/BRAFV600E transgenic mice, some Kupffer cells were positive for podoplanin. In these physiological conditions, although podoplanin is expressed in some kinds of cells such as kidney podocytes, lung alveolar type I cells and lymphatic endothelial cells, these cells do not directly interact with platelets.9, 10, 11 In the Cre‐Alb/BRAFV600E transgenic mice, however, podoplanin‐positive Kupffer cells might trigger aberrant platelet activation via the interaction of podoplanin with platelet CLEC‐2, which facilitates platelets to interact with Kupffer cells and LSECs in the liver and to be deposited in the renal glomeruli and the pulmonary alveoli. Platelet activation through the interaction of CLEC‐2 with podoplanin expressed in sinusoidal cells was also shown in regenerating liver tissue after a two‐thirds hepatectomy in mice.36 There are two subsets of hepatic macrophages: yolk sac/fetal liver‐derived and bone marrow/monocyte‐derived macrophages.37, 38, 39 Conversely, although podoplanin is generally not expressed in macrophages, it is expressed in activated macrophages such as those treated with thioglycollate or lipopolysaccharide in vitro,21 in those found in the liver of Salmonella‐infected mice,22 in those found in the spleen and peritoneal cavity in zymosan‐treated mice 23 and in severe atherosclerotic lesions in humans.24 It is thought that, in the hepatic microenvironment of Alb‐Cre/BRAFV600E transgenic mice, a subset of resident macrophages or those recruited from outside of the liver might be influenced to express podoplanin.

In the Alb‐Cre/BRAFV600E transgenic mice, platelets were deposited in the kidney and lung, as well as in the liver. Because platelet‐derived factors promote the recruitment/activation of inflammatory cells and the migration/proliferation of mesenchymal cells,40 platelet deposition in renal glomeruli and pulmonary alveoli may be related to the pathogenesis of glomerulonephropathy and interstitial pneumonia in Alb‐Cre/BRAFV600E transgenic mice. Interestingly, the tissues where platelets were deposited contained podoplanin‐expressing cells: podocytes in the renal glomeruli and type 1 alveolar cells in the lung.9, 10, 11 Thus, tissue destruction due to the deposition of activated platelets might have enabled an interaction between platelet CLEC‐2 and podoplanin‐expressing cells, further aggravating inflammation and tissue destruction. However, it remains to be investigated whether platelets directly interact with podoplanin‐expressing cells in the kidney and lung.

In the Alb‐Cre/BRAFV600E transgenic mice, Asp, a platelet activation inhibitor via cyclooxygenase inhibition,25 efficiently prevented platelet activation and ameliorated the histopathological changes in association with decreased platelet deposition in the liver, kidney, and lung. These observations indicate that the pathological changes in these organs are dependent on the deposition of aberrantly activated platelets. Interestingly, the liver/body weight ratio was decreased to 70% in the Asp‐treated Alb‐Cre/BRAFV600E transgenic mice, and normal‐looking hepatocyte areas were specifically observed together with small basophilic hepatocytes. Furthermore, the degree of sinusoidal platelet deposition was lower in the normal‐looking hepatocyte areas than in the small basophilic hepatocyte areas. These observations suggested that the properties of BRAFV600E‐mutated hepatocytes were regulated by sinusoidal platelet deposition.

In conclusion, the macroenvironment that results from TPO overproduction by the preneoplastic/neoplastic hepatocytes, eventual megakaryocytosis/thrombocytosis, and platelet activation and interaction with hepatic sinusoidal cells may contribute to hepatocarcinogenesis from a very early stage. However, erythrocyte dyscrasia, presumably due to thrombotic microangiopathy, and glomerulonephropathy/interstitial pneumonia due to platelet deposition caused spontaneous death.

DISCLOSURE

No conflict of interest.