Isolation and characterization of MPs derived from apoptotic HPCs
During apoptosis, cells can release small MPs called apoMPs (< 1 μm in diameter) . In order to obtain a useful amount of apoMPs from apoptotic HPCs, we used different lethal doses of doxorubicin (Additional file 2: Fig. S1A) to treat HPCs (WB-F344 cell line) and the released MPs were then isolated by centrifugation steps (Additional file 2: Fig. S1B) . It has previously been shown that when tumor cells were treated with doxorubicin, the doxorubicin was packaged into the released MPs, and because of the fluorescent nature of doxorubicin, MPs encapsulating doxorubicin could be clearly observed by fluorescence microscopy . The released apoMPs were also observed by fluorescence microscopy in our study. We found that WB-F344 cells treated with 100 µg/ml of doxorubicin released a considerable amount of apoMPs (apoHPC-MPs) (Additional file 2: Fig. S1C). Flow cytometry further showed that 2 × 107 WB-F344 cells released about 1 × 106 apoHPC-MPs (Additional file 2: Fig. S2A, S2B). These apoHPC-MPs had a membrane structure and were ~800 nm in size, as detected by transmission electron microscopy (TEM; Additional file 2: Fig. S2C). The sizes were confirmed by dynamic light scattering (DLS) analysis (Additional file 2: Fig. S2D). DLS analysis also revealed that apoHPC-MPs displayed zeta potentials of ~ − 17 mV (Additional file 2: Fig. S2E). Thus, during apoptosis, HPCs can release large numbers of apoHPC-MPs.
MPs derived from apoptotic HPCs prevent hepatocarcinogenesis in a primary rat HCC model
To test whether apoMPs can inhibit hepatocarcinogenesis, it is essential to identify which cellular types are suitable sources of apoMPs. Besides the apoHPC-MPs described above, which were produced from WB-F344 cells, we also produced apoMPs from the hepatocyte cell line BRL (apoHep-MPs) and the liver tumor cell line RH35 (apoLTC-MPs) by treatment with 100 µg/ml of doxorubicin. The apo-MPs were observed by fluorescence microscopy (Additional file 2: Fig. S2F). ApoHPC-MPs, apoHep-MPs and apoLTC-MPs had a similar irregularly spherical morphology as observed by TEM (Additional file 2: Fig. S2C). They also had similar diameters of ~800 nm (Additional file 2: Fig. S2D) and similar zeta potentials of ~ − 17 mV (Additional file 2: Fig. S2E). To test whether these apoMPs have similar anticancer effects, we employed the diethylnitrosamine (DEN)-induced primary HCC model in Sprague Dawley rats. 6 weeks of oral DEN treatment showed the obvious HPC activation (Additional file 2: Fig. S3), HPCs were labeled with an antibody against CK7 , and in our previous studies, we found that the activation and malignant transformation of HPCs promote hepatocarcinogenesis and HCC recurrence [9,10,11,12]. Thus, the rats were intrasplenically administered with 40 µg of apoLTC-MPs, apoHep-MPs, apoHPC-MPs or saline after 6 weeks of oral DEN treatment. Rats were injected with MPs or saline twice per week for 7 weeks, and oral DEN treatment was continued at the same time. On week 13 of DEN treatment, the rats were sacrificed (Fig. 1A). We then assessed the therapeutic efficacy of apoMPs in the rat HCC primary model. Compared with saline control, apoLTC-MPs or apoHep-MPs, the apoHPC-MPs effectively inhibited tumorigenesis in DEN-exposed rats (Fig. 1B). This was also evidenced by the maximum tumor volume (Fig. 1C), liver-to-body weight ratio (Fig. 1D), tumor number (Fig. 1E) and tumor incidence (Fig. 1F). H&E staining showed that livers from rats treated with apoHPC-MPs had a reduced inflammatory response and a clear tissue structure (Fig. 1G). Besides, we found that apoHPC-MPs ameliorated the weight loss of rats during hepatocarcinogenesis (Fig. 1H). Because of the known toxicity of doxorubicin in heart , we also tested whether the apoMPs have side effects on heart. Notably, we did not observe toxic effects of apoHPC-MPs in heart, as shown by serological analysis (creatine kinase) and H&E staining (Fig. 1I and Additional file 2: Fig. S4). ApoHPC-MPs also had no side effects in other major organs, as evidenced by H&E staining of other tissues (Additional file 2: Fig. S4). Taken together, these results show that apoHPC-MPs have much better anticancer efficacy than apoLTC-MPs and apoHep-MPs without typical side effects in a rat primary HCC model.
In order to further confirm the role of apoHPC-MPs on hepatocarcinogenesis, we isolated primary HPCs from rats via collagen IV digestion and formed liver organoids. The organoids were treated with 100 µg/ml of doxorubicin, then organoid-apoMPs were produced (Fig. 2A). In order to test the therapeutic efficacy of organoid-apoMPs, the rat primary HCC model was established and the animals were intrasplenically treated with organoid-apoMPs or saline. Organoid-apoMPs (40 µg) were administered to rats after 6 weeks of DEN treatment to induce hepatocarcinogenesis. Rats were treated with MPs or saline twice per week for 7 weeks. On week 13 of DEN treatment, the rats were sacrificed and the tumor growth was evaluated. As shown in Fig. 2B, almost no tumor nodes were observed in the organoid-apoMPs treatment group. In contrast, several tumor nodes were observed in the saline control group. Treatment with organoid-apoMPs also reduced the maximum tumor volume (Fig. 2C), tumor number (Fig. 2D), liver-to-body weight ratio (Fig. 2E) and tumor incidence (Fig. 2F). Compared with livers from the saline group, livers from the organoid-apoMPs group showed a reduced inflammatory response and a clear tissue structure, as revealed by H&E staining (Fig. 2G). Therefore, primary HPC-derived apoMPs efficiently inhibit hepatocarcinogenesis.
HPCs take up more apoHPC-MPs than apoLTC-MPs and apoHep-MPs
The above results show that apoHPC-MPs prevent hepatocarcinogenesis more efficiently than apoLTC-MPs and apoHep-MPs. To elucidate the underlying mechanism, we first asked whether HPCs preferntially take up apoHPC-MPs compared to apoLTC-MPs and apoHep-MPs. Therefore, doxorubicin (100 µg/ml) was added to cultured WB-F344 cells (1 × 107), RH35 cells (1 × 107) or BRL cells (1 × 107). After 12 h, the released apoHPC-MPs, apoLTC-MPs and apoHep-MPs were isolated from the culture medium by centrifugation. The same numbers of apoHPC-MPs (1 × 105), apoLTC-MPs (1 × 105) and apoHep-MPs (1 × 105) were incubated with WB-F344 cells (5 × 103) for 30 min. Interesting, only the apoHPC-MPs, and not the apoLTC-MPs and apoHep-MPs, were efficiently taken up by WB-F344 cells (Fig. 3A). Around 75% of WB-F344 cells took up apoHPC-MPs (Fig. 3B). Furthermore, the specific targeting of apoHPC-MPs, apoLTC-MPs and apoHep-MPs was quantitatively determined by flow cytometry analysis. Equivalent numbers of apoHPC-MPs (1 × 106), apoLTC-MPs (1 × 106) and apoHep-MPs (1 × 106) were incubated with WB-F344 cells (3 × 105) for 30 min. The results indicated that 96% of WB-F344 cells were positive for apoHPC-MPs, whereas only about 8% of WB-F344 cells were positive for apoLTC-MPs, and about 30% of WB-F344 cells were positive apoHep-MPs (Fig. 3C, D). This suggests that WB-F344 cells selectively take up apoHPC-MPs, rather than apoLTC-MPs or apoHep-MPs. Consistently, RH35 cells showed a strong preference for apoLTC-MPs, rather than apoHPC-MPs and apoHep-MPs (Additional file 2: Fig. S5A). Based on flow cytometry analysis, about 97% of RH35 cells were positive for apoLTC-MPs, while only about 12% of RH35 cells took up apoHep-MPs and about 38% of RH35 cells took up apoHPC-MPs (Additional file 2: Fig. S5B). BRL cells also indicated a consistent pattern of uptake, with a strong preference for apoHep-MPs, rather than apoHPC-MPs and apoLTC-MPs (Additional file 2: Fig. S5C, D). These results indicate that cells can efficiently take up parental cell-derived MPs. Thus, apoHPC-MPs efficiently target WB-F344 cells.
To further test the targeting of apoMPs to HPCs, we isolated primary HPCs from rat liver and generated organoids. Equivalent numbers of apoHPC-MPs (1 × 104), apoLTC-MPs (1 × 104) and apoHep-MPs (1 × 104) were incubated with primary HPCs for 30 min. Primary HPCs efficiently took up apoHPC-MPs, rather than apoHep-MPs and apoLTC-MPs, as evidenced by fluorescence microscopy (Fig. 3E). Because all these findings were based on cultured HPCs, we wanted to further clarify whether apoMPs have the same ability to target HPCs in vivo. For this experiment, 1 × 107 of apoHPC-MPs, apoHep-MPs and apoLTC-MPs were intrasplenically injected into DEN-treated rats and liver slices were acquired for fluorescence detection 30 min after injection. The expression of the HPC marker CK19 in liver sections was examined [24,25,26]. As observed in cell culture, accumulation of red fluorescent apoHPC-MPs, rather than apoHep-MPs and apoLTC-MPs, was found in CK19-positive HPCs (Fig. 3F). This demonstrates that HPCs can selectively take up HPC-derived apoHPC-MPs in vivo.
We also examined the interaction of HPCs and apoHPC-MPs. WB-F344 cells (3 × 105) were incubated with apoHPC-MPs (1 × 105) for 30 min, then treated with trypsin and washed several times with citric acid to remove the surface-bound MPs. Fluorescence imaging revealed that apoHPC-MPs were not cleared from WB-F344 cells after trypsinization and acid washing (Fig. 3G). This suggests that apoHPC-MPs were internalized into WB-F344 cells, rather than being attached to the surface of the WB-F344 cells. We further studied the dynamics of apoHPC-MPs internalization into WB-F344 cells. For this experiment, 3 × 105 WB-F344 cells were incubated with 2 × 105 apoHPC-MPs. The internalization of apoHPC-MPs was detected after 0.5 h and increased markedly with time (Fig. 3H), suggesting that the uptake of apoHPC-MPs by HPCs is time-dependent.
ApoHPC-MPs are cytotoxic to HPCs
Next, we asked if apoHPC-MPs are cytotoxic to HPCs after they are internalized. ApoHPC-MPs (20 µg), apoLTC-MPs (20 µg) or apoHep-MPs (20 µg) were added to cultured WB-F344 cells (5 × 103) for 24 h. Compared with apoHep-MPs and apoLTC-MPs, apoHPC-MPs induced more killing of WB-F344 cells (Fig. 4A). Moreover, the proliferation of WB-F344 cells was efficiently inhibited by treatment with apoHPC-MPs (Fig. 4B). To test this result in vivo, 40 µg of apoHPC-MPs, apoHep-MPs or apoLTC-MPs were intrasplenically injected into rats that previously received DEN treatment for 6 weeks. After 24 h, liver slices were acquired for fluorescence detection. HPCs were labeled with an antibody against SOX9 [26, 27] and apoptotic cells were detected with an antibody against cleaved caspase 3. The highest number of apoptotic HPCs (positive for both SOX9 and cleaved caspase 3) was seen in the apoHPC-MPs treatment group (Fig. 4C, D). In the apoHep-MPs group, a small number of HPCs were positive for cleaved caspase 3 staining, while saline and apoLTC-MPs did not induce apoptosis of HPCs (Fig. 4C, D). To corroborate these observations, we also evaluated the expression of the HPC marker SOX9 and CK7 in liver sections by IHC [23, 24]. ApoLTC-MPs and the control group showed intense and diffuse positive SOX9 and CK7 staining, while apoHep-MPs caused a small decrease in the SOX9 and CK7 signal and apoHPC-MPs caused a significant decrease in SOX9 and CK7 staining (Fig. 4E, F and Additional file 2: Fig. S6). Therefore, after injection of apoMPs into rats, the number of HPCs was greatly reduced by apoHPC-MPs, partly reduced by apoHep-MPs, and unaffected by apoLTC-MPs. As a result of all these experiments, we conclude that HPCs can selectively take up cytotoxic apoHPC-MPs, inhibiting the proliferation of hepatoma-initiating cells.
MPs from non-apoptotic HPCs have no affect hepatocarcinogenesis
To further clarify the role of the death signal, we generated MPs from non-apoptotic HPCs. WB-F344 cells were treated with 100 µg/ml of DOX, or irradiated with 20 Gy, and then MPs without death signal (DOXO-MPs and RT-MPs) were collected from the cells before they underwent apoptosis. 40 µg of DOXO-MPs and RT-MPs were injected into rats, which were treated with DEN for 6 weeks (Fig. 5A). MPs or saline were injected twice every week for 7 weeks. The rats were then sacrificed to observe the tumor growth. As shown in Fig. 5B, compared with the saline control group, treatment with DOXO-MPs and RT-MPs had no effect on hepatocarcinogenesis. This result was also confirmed by the maximum tumor volume (Fig. 5C), tumor number (Fig. 5D), liver-to-body weight ratio (Fig. 5E) and tumor incidence (Fig. 5F). H&E staining showed that the inflammatory response was similar in the livers of each group (Fig. 5G). Therefore, MPs derived from non-apoptotic HPCs had no effect on hepatocarcinogenesis.
Because apoHPC-MPs were produced by treating HPCs with doxorubicin, the released apoHPC-MPs contained doxorubicin in addition to the death signal (Additional file 2: Fig. S1C). To exclude any effect of the therapeutic drug doxorubicin on killing HPCs by apoHPC-MPs in the rat primary HCC model, the amount of doxorubicin in apoHPC-MPs was detected by liquid chromatography-mass spectrometry (LC-MS) analysis (Additional file 2: Fig. S7A). Free doxorubicin (~ 10 µg, which was the same amount of doxorubicin encapsulated in apoHPC-MPs for injection of each rat) was intrasplenically injected into rats, which had been treated with DEN for 6 weeks. Injections were performed twice every week for 7 weeks. The rats were then sacrificed for tumor examination. Compared with the saline control group, the free doxorubicin had no significant effect on hepatocarcinogenesis (Additional file 2: Fig. S7B). There was no significant difference in the maximum tumor volume (Additional file 2: Fig. S7C), tumor number (Additional file 2: Fig. S7D), liver-to-body weight ratio (Additional file 2: Fig. S7E) and tumor incidence (Additional file 2: Fig. S7F). H&E staining showed that the inflammatory response was similar in the livers of each group (Additional file 2: Fig. S7G). Therefore, doxorubicin encapsulated in apoHPC-MPs had no effect on hepatocarcinogenesis.
HPC-derived MPs containing death signals inhibit hepatocarcinogenesis in the absence of doxorubicin
To further clarify the role of the death signal in MPs, and exclude the effect of the encapsulated chemotherapeutic drug, we wanted to isolate MPs without chemotherapeutic drugs. To do this, we used a lethal dose of radiation to induce WB-F344 cell death. WB-F344 cells were treated with various doses of radiation to establish the lethal dose. We observed increased cell killing with an increased radiation dose until 20 Gy, which induced the maximum level of cell death (Additional file 2: Fig. S8A). The amount of MPs differed according to the subsequent culture time of cells after 20 Gy radiation: the maximal amount of MPs was isolated after culturing WB-F344 cells for another 72 h (Additional file 2: Fig. S8B). We chose these conditions for subsequent studies. The MPs were designated as RT-apoHPC-MPs (Fig. 6A). To further evaluate the therapeutic effect of RT-apoHPC-MPs on HPCs, we performed cell toxicity studies on the HPC cell line WB-F344. We found that RT-apoHPC-MPs efficiently inhibited the growth of WB-F344 cells in a dose- and time-dependent manner (Fig. 6B and Additional file 2: Fig. S8C). We then investigated the therapeutic effect of RT-apoHPC-MPs in the rat primary HCC model. RT-apoHPC-MPs (40 µg) were intrasplenically injected into rats, which had been treated with DEN for 6 weeks. Injections were performed twice every week for 7 weeks. The rats were then sacrificed for tumor examination. As shown in Fig. 6C, we observed reduced tumor growth upon treatment with RT-apoHPC-MPs, superior to that of the control treatment. This result was supported by analysis of maximum tumor volume (Fig. 6D), tumor number (Fig. 6E), liver-to-body weight ratio (Fig. 6F) and tumor incidence (Fig. 6G). H&E staining showed a reduced inflammatory response and clear tissue structure in the livers of rats treated with RT-apoHPC-MPs (Fig. 6H). Therefore, RT-apoHPC-MPs carrying a death signal can efficiently inhibit hepatocarcinogenesis in the absence of doxorubicin.
Quantitative proteomic analysis of extracellular microparticles reveals enrichment of death-related proteins in apoHPC-MPs
To further investigate the death signal encapsulated by apoHPC-MPs, quantitative proteomics analysis was performed on apoHPC-MPs from apoptotic WB-F344 cells and MPs from non-apoptotic WB-F344 cells by MS/MS. Three independent experiments reliably identified 523, 344 and 652 proteins from MPs, and 310, 769 and 978 proteins from apoHPC-MPs. Of these, 278 proteins were quantitatively identified in all replicates (Additional file 1: Table S1). The threshold for significant up-regulation was set as fold change > 1.5 and the threshold for significant down-regulation was set as less than 1/1.5. Under these criteria, there were 116 proteins significantly upregulated in apoHPC-MPs as compared to MPs (Additional file 1: Table S1). Gene ontology (GO) enrichment analysis indicated that the 116 up-regulated proteins were enriched in biological processes related to regulation of apoptotic process and regulation of programmed cell death (Fig. 7A). Further, Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis showed that the 116 up-regulated proteins were also related to the necroptosis pathway (Fig. 7B). Moreover, based on 116 up-regulated proteins, gene encoding death-related proteins are present at higher levels in the apoHPC-MPs than in MPs (Fig. 7C). Together, these results indicate that apoHPC-MPs inhibit hepatocarcinogenesis by transmitting a death signal to hepatoma-initiating cells, which results in the death of hepatoma-initiating cells.