Reprogramming the spleen into a functioning 'liver in vivo

Introduction

Liver transplantation is the only possible solution for end-stage liver diseases (ESLD) with significant loss of hepatocytes and organ failure. The need for alternative therapeutic approaches has become progressively urgent, due to the increasingly severe shortage in both whole organs and cells available for transplantation.1 2 Particularly, hepatocyte transplantation aimed at repopulating the damaged liver parenchyma suffer from poor engraftment within the highly damaged liver, which hampers clinical translation.3

Recently, direct reprogramming of adult fibroblasts into hepatocyte-like cells by overexpressing hepatocyte-specific transcription factors has shown positive outcomes in reversing liver fibrosis in mice.4 5 However, it remains an unmet challenge to use this technology for regenerating the liver tissue with adequate physiological functions. One fundamental problem is where to perform the reprogramming process. Anatomically, the native liver itself is the ideal site for practice. But, for most ESLD patients, the damaged liver tissue environment does not even support the native hepatocytes to survive,3 much less to serve as a bed for cell reprogramming and tissue regeneration. Ectopic regeneration is another possible way. However, the liver is too large to be ectopically, fully functionally engineered by current methods.6–8 It is especially difficult to rebuild a circulatory system as large and robust as that in the healthy liver, which is required for supporting enough liver cells to survive and function.9

An alternative approach is to exploit the native blood vessel network in an existing organ, such as the spleen, to support liver regeneration in vivo. Transforming the spleen into a liver has several advantages. First, manipulating cell functions in the spleen would have less impact on the body than in other vital organs, because the spleen is a dispensable organ—as clinically evidenced in cases of splenectomy.10 Second, as a large lymphoid organ, the spleen has space and a loose structure preferred for accommodating the massive number of liver cells.11 Third, the spleen has an extensive blood supply directly connected to the main circulation,12 which can support the growth and functionalisation of the regenerated liver. We have verified the possibility of spleen transformation in our recent study, by growing hepatocytes from different species into a remodelled mouse spleen.13

Based on the above considerations, we hypothesised to transform the spleen in mice into a functioning liver, through directly reprogramming the splenic fibroblasts into hepatocytes. We designed a three-step procedure comprising: (1) increasing fibroblasts: activating splenic fibroblasts (reticular fibroblasts and reticular fibres, ER-TR7+) to generate more fibroblasts and extracellular matrices; (2) transforming fibroblasts: expressing three transcriptional factors (3TF)—Foxa3, Gata4 and Hnf1a—through lentiviral vectors (LV) to induce fibroblasts into iHeps and (3) expanding iHeps: stimulating them to proliferate within the spleen to perform liver functions. We comprehensively evaluated the efficacy and safety of this strategy in a 90% hepatectomy model.

Materials and methodsAnimals

The 8-week-old wild-type C57BL/6J female mice were purchased from Vital River Laboratories (Beijing, China). For diet-induced obesity study, 6-week-old male mice were fed with a standard chow diet (10% calories from fat; Research Diets, New Brunswick, New Jersey, USA) or a high-fat diet (HFD) (60% calories from fat; Research Diets, New Brunswick, New Jersey, USA) for 16 weeks. They were fed sufficient food and water, maintained on a regular 12-hour day/night cycle under specific pathogen-free conditions.

Lentivirus production

To produce lentivirus, modified plasmids carrying candidate genes were transfected to 293 T cells together with packaging plasmid psPAX2 (Addgene) and envelope plasmid pMD2.G (Addgene) using lipofectamine 2000 (Thermo Fisher Scientific, Waltham, Massachusetts, USA) as previously described.14 The virus particle was dissolved in sterile phosphate buffer saline (PBS) and stored at −80°C. Virus titers were assessed by quantitative real-time PCR (qRT-PCR).

Generation of iHep in vitro

The protocol followed literature with modifications.15 Briefly, primary mouse spleen fibroblasts (MFs) at passage 3, following SiO2 remodelling, were used for lentiviral infection. To generate iHeps, MFs were cultured in type I collagen-coated 6 cm dishes. 1×105 MFs were transduced with LV encoding for 3TF at multiplicity of infection 5 supplemented with 4 µg/mL polybrene (Sigma-Aldrich, St. Louis, Missouri, USA). After 24 hours, the complete Dulbecco’s modified Eagle’s medium medium was replaced by medium containing 0.1 µM dexamethasone, 10 µg/L epidermal growth factor (EGF), 20 µg/L transforming growth factor-α, 4.2 mg/L insulin, 5 µg/L sodium selenite and 3.8 mg/L human transferrin. After 14 days, cells were digested by 0.01% trypsin and fibroblastic cells were discarded to gather the epithelial cells by differential digestion.

Remodelling of the spleen

PBS or SiO2 (25 mg/mL; 1 mg in 40 µL PBS per mice) were injected into the translocated spleen using a 31-gauge needle (BD Biosciences, San Diego, California, USA) every 4 days in the indicated times. Samples were collected and analysed on day 26.

MFs reprogramming to hepatocytes in vivo

1×108 plaque forming unit (PFU) lentiviral NC (LV-NC) or lentiviral 3TF (LV-3TF) was injected the SiO2 remodelled spleen with a 31-gauge needle. After 4 weeks, samples were collected and analysed.

Expansion of hepatocytes in vivo

In order to promote the proliferation of iHeps in vivo, we injected 0.025 mg/kg tumour necrosis factor-α (TNF-α) into the SiO2-3TF treated spleen. Afterwards, we injected 1×108 PFU lentiviral EGF (LV-Egf) and lentiviral hepatocyte growth factor (HGF) (LV-Hgf) into the SiO2-3TF treated spleen with a 31-gauge needle. After 4 weeks, samples were collected and analysed.

Histological analyses

Immunofluorescence staining, HE staining, Masson trichrome staining, Periodic acid-Schiff (PAS) staining and Oil Red O staining were performed according to the manufacturer’s instructions and the standard procedures as previously described.13

Cyp450 metabolism assay

MF, iHep and primary hepatocytes (PH) were cultured in the medium containing 50 µM rifampicin (Sigma-Aldrich) for 48 hours in a six-well plate. The substrate solutions were prepared with the medium containing 100 µM pheancetin (Sigma-Aldrich) for 3 hours. To stop the reaction, 500 µL cold methanol was added. The cell supernatants were collected to measure cytochrome P450, family 1 subfamily a polypeptide 2 (CYP1A2) metabolised products by liquid chromatography-tandem mass spectrometry (LC-MS/MS) with a QTRAP 6500 mass spectrometer (AB Sciex, Singapore). Freshly isolated primary mouse hepatocytes were used as the positive control.

For the CYP1A2 and cytochrome P450 family 2 subfamily C member 9 (CYP2C9) metabolism assay in vivo, hepatic portal vein and hepatic artery were ligated, and then 20 mg/kg phenacetin and 20 mg/kg diclofenac (Sigma-Aldrich) were respectively injected into the spleen. After 30 min, the plasma samples were collected. The metabolites in the plasma were measured by LC-MS/MS. The amount of the metabolism products was determined as analyte peak area. Data were analysed by Analyst V.1.6.3 (AB Sciex).

Indocyanine green assay

Cells were cultured in the medium containing 1 mg/mL indocyanine green (ICG) for 1 hour at 37°C. For animal experiments, mice received 10 mg/kg of body weight ICG through intravenous injection. After 15 min, spleen tissues were collected and embedded in OCT and sectioned. The cryostat sections were photographed by BX51 microscope.

For the ICG retention rate at 15 min (ICGR15) assay, mice received 10 mg/kg of body weight ICG through intravenous injection after the ligation of hepatic portal vein and hepatic artery. After 15 min, the concentration of ICG in serum was measured using Varioskan LUX (Thermo Fisher Scientific) at 805 nm.

5-ethynyl-2’-deoxyuridine cell proliferation assay

For intrasplenic hepatocytes and fibroblasts proliferation in vivo, mice were injected intraperitoneally with 5-ethynyl-2’-deoxyuridine (EdU; Ribobio, Guangzhou, China) at 50 mg/kg of body weight for seven consecutive days. Spleens were harvested a day after the last injection. Paraffin sections were first stained with the appropriate primary and secondary antibodies, and then incubated with Cell-Light Apollo Stain Kit (Ribobio) according to the manufacturer’s instructions. To detect the effect of SiO2 on the proliferation of MF in vitro, MF were seeded on 24-well plates. MF were treated with PBS, 50 µg/mL SiO2 and 5 nM TGF-β1 receptor inhibitor (SB431542; Med Chem Express, New Jersey, USA) for 6 days. Cells were fixed with 4% paraformaldehyde for 10 min and washed three times with PBS. EdU labelling was performed using Cell-Light Apollo Stain Kit.

PCR, ELISA and Western blotting

Experiments were performed according to the manufacturer’s instructions and the standard procedures as previously described.13 For qRT-PCR, all primers sequences were listed in online supplemental table 1.

Gene expression microarray analysis

Total RNA was extracted from MF, iHep, PH, spleen and liver samples using TRIzol. RNA samples were hybridised to the Whole Mouse Genome Oligo Microarray (4×44 K; Agilent Technologies, Santa Clara, California, USA) according to the manufacturer’s instruction. Microarray hybridisation and analysis were carried out by KangCheng Biotechnology (Shanghai, China). The data were normalised by GeneSpring GX (Agilent Technologies). Hierarchical clustering of samples was performed by Cluster 3.0. The results were viewed by using Java TreeView. The microarray data have been submitted to the Gene Expression Omnibus under accession number GSE172245.

Statistics analysis

All data were normally distributed. Data are shown as means±SE of the mean. Data were analysed using Graphpad Prism V.8.0 (Graphpad). Statistical comparisons were performed by using Student’s t-test, one-way and two-way analysis of variance followed by Tukey’s multiple comparisons test. A value of p<0.05 was considered significant.

ResultsStimulation of fibroblast proliferation in the mouse spleen

The entire animal study was performed in a spleen-translocated mouse model established in our recent study.13 Briefly, the spleen was surgically moved from the inner abdomen to a subcutaneous site (online supplemental figure 1A). According to our previous investigation, the operation causes no physiological abnormities in the mice and facilitates repeated administrations of reagents into the spleen via direct intra-splenic injections (online supplemental figure 1B).

The number of fibroblasts in the spleen is lower (<10% in one spleen) than many other organs and should be increased before being reprogrammed into hepatocytes by transgenic expression of specific genes.15–17 We previously reported that SiO2 particles with a size of 100 nm effectively induced fibrosis in the lung via hijacking TGF-β signalling,18 which could be used here to induce fibroblast proliferation. We first performed an in vitro test on fibroblasts separated from the mouse spleen, verifying that stimulation with SiO2 particles significantly enhanced the proliferative activity of these cells—as evidenced by their growth rate (online supplemental figure 2A), activation markers (alpha-smooth muscle actin (α-SMA) and desmin) (online supplemental figure 2B) and the EdU cell proliferation assay (online supplemental figure 2C). Moreover, the TGF-β1 signalling inhibitor SB431542 suppressed the effect of SiO2 (online supplemental figure 2A–C), suggesting the role of TGF-β1 as previously revealed.18

The in vivo tests were scheduled (figure 1A) that the mice were given one injection of SiO2 particles every 4 days (1 mg in 40 µL PBS for each injection), four times in a row. The spleens were harvested and analysed 19 days after the first injection. The size and morphology of the spleens significantly changed after the SiO2-remodelling (figure 1B). The spleens were enlarged that their average weight reached up to 2.5 times (figure 1B,C). Histological examination illustrated that the white and red pulps structure of the spleen significantly shrank while the presence of fibroblasts and fibrotic tissues increased (figure 1D). The splenic reticular fibroblasts specifically express ER-TR7. On receiving SiO2 particles, this group of cells actively proliferated, as evidenced by the EdU assay (figure 1E) and α-SMA staining (figure 1F). The fibroblasts’ activation was throughout the whole spleen after four doses of SiO2 (figure 1G). Detailed analysis indicated that the actively proliferating fibroblasts mainly presented in the red pulp and the marginal zone (figure 1H). The splenic fibroblasts (Vimentin+) were counted along with the SiO2 stimulation. Their number steadily increased within 2 weeks before the mice received the third SiO2 intrasplenic injection, but the additional fourth dose did not further increase the number of fibroblasts (figure 1I). Moreover, we continued measuring the fibroblast number for two more weeks after the fourth SiO2 injection and found that the fibroblasts decreased to almost the original level, suggesting that the fibrotic reaction in the spleen was reversible (online supplemental figure 3).

Figure 1Figure 1Figure 1

SiO2 particle promoted mouse spleen fibroblasts proliferation. (A) Scheme of SiO2 remodelling spleen. (B) The morphology of translocated spleen injected with PBS or SiO2 four times on day 26. (C) The weight of PBS spleen and SiO2-remodelled spleen (n=5). (D) HE staining of the PBS spleen and SiO2-remodelled spleen. The proliferation activity of splenic reticular fibroblasts was determined by (E) EdU staining and (F) α-SMA staining. (G) Expression of α-SMA and Vimentin in the whole spleen sections. (H) Representative fluorescent images of α-SMA showing the distribution of activated fibroblasts in the spleen. (I) The proportion of mouse spleen fibroblasts and representative sample data was detected by flow cytometry in the spleens treated with PBS or different SiO2 concentrations at the indicated time points (n=3). (J) Gene associated with TGF-β1 signalling pathway in transcript level by RNASeq analysis. Five tissue samples from independent experiments were pooled. (K) The mRNA expression levels of fibroblasts markers in spleen were measured by qRT-PCR (n=5). (L) Western blotting analysis of the expression p-Smad2 and Cyclin D1. Statistics: data are shown as means±SEM. (C, K) Student’s t-test. (I) Two way ANOVA followed by Tukey’s multiple comparisons test. ANOVA, analysis of variance; α-SMA, alpha-smooth muscle actin; EdU, 5-ethynyl-2’-deoxyuridine; PBS, phosphate buffer saline; qRT-PCR, quantitative real-time PCR.

The SiO2 stimulation activated the TGF-β1 signalling, as revealed by the RNASeq analysis of splenic mRNA (figure 1J). The results were verified by qRT-PCR analysis of key genes (figure 1K), as well as western blotting analysis of p-Smad2 and Cyclin D1 (figure 1L), the signals for proliferative activation. However, overactivation of TGF-β1 may induce excess fibrosis in the affected tissue, which is undesirable for subsequent cell reprogramming and liver regeneration. Therefore, we examined the content of collagen with Masson’s staining (online supplemental figure 4A) and hydroxyproline quantification (online supplemental figure 4B) between healthy and fibrotic livers (the latter harvested from an 8-week carbon tetrachloride (CCl4)-induced murine liver fibrosis). The outcomes suggested that the collagen content in SiO2-treated spleens reached the level of a normal liver and was far below that of a pathologically fibrotic liver.

As TGF-β1 is also a key cytokine for immune suppression, we analysed the impact of SiO2 treatment on the lymph cells and macrophages in the spleens. The results in online supplemental figure 5A–D demonstrated that the higher dose (1 mg per injection per mouse) significantly reduced the number of both T (CD4+ or CD8+) and B cells (B220+), while the number of macrophages remained constant during the treatment. Consistently, the levels of key pathways related to the immune rejection were lower in SiO2-treated spleens than in PBS-treated ones (online supplemental figure 5E), according to RNASeq data.

We assessed the safety of SiO2 treatment by analysing different tissue from the treated mice 4 weeks after the final dose of SiO2 administration. First, we found no obvious change in vital organs' histology (online supplemental figure 6A) or damage to the liver that was possibly the most influenced organ (online supplemental figure 6B–E). The number of lymph cells returned to the normal level (online supplemental figure 6F–I). These findings suggested that the treatment would not damage the long-term health of the animals.

Together, our investigation demonstrated that SiO2 particle treatment was efficient to increase fibroblasts in the spleen. The expanded fibroblasts would facilitate the lentivirus-mediated gene transfection and the subsequent reprogramming of the fibroblasts into hepatocytes.

Characterisation of iHeps derived from SiO2-remodeled murine spleen fibroblasts in vitro

Based on previous research,15 we selected Foxa3, Gata4 and Hnf1a—three key transcription factors (online supplemental table 2) to reprogram fibroblasts into hepatocytes. To overexpress 3TF, we cloned foxa3, gata4 and hnf1a cDNA into the LV, respectively. We first examined whether overexpression of 3TF could induce hepatocytic phenotype in the fibroblasts from the remodelled spleens. The expression of the 3TF in MFs was verified on day 4 after the LV-3TF transduction (online supplemental figure 7). After a 14-day culture post-transduction, the induced iHeps displayed a typical epithelial morphology (figure 2A). We gathered the epithelial cells for further characterisation. The results of qRT-PCR and reverse transcription PCR (RT-PCR) showed that hepatocyte-specific genes were remarkably upregulated in iHeps (figure 2B, online supplemental figure 8), whereas fibroblast-specific genes were significantly down-regulated (figure 2C, online supplemental figure 8).

Figure 2Figure 2Figure 2

Characterisation of iHeps derived from SiO2-remodelled mouse spleen fibroblasts in vitro. (A) Representative morphologies of MFS and iHeps. The mRNA expression levels of (B) hepatocytes markers and (C) fibroblasts markers were measured by qRT-PCR in MFS and iHeps. PH was used as a positive control. (D) Immunofluorescence staining of HNF4α, ALB, GS and E-cadherin for hepatocyte markers. (E) Global gene expression analysis of MFS, iHeps and pH by cDNA microarray demonstrated that iHeps had a similar gene expression profile with pH. Three cell samples from three independent experiments were pooled. Hepatic functions in iHeps: (F) PAS staining, (G) oil red O staining and (H) ICG uptake. (I) ALB secretion and (J) urea production were measured by ELISA (n=5). (K) CYP1A2 metabolic activity of iHeps was measured by LC-MS/MS (n=5). Statistics: data are shown as means±SEM. (I, J, K) One way ANOVA followed by Tukey’s multiple comparisons test. ALB, albumin; ANOVA, analysis of variance; GS, glutamine synthetase; ICG, indocyanine green; LC-MS/MS, liquid chromatography-tandem mass spectrometry; MFS, mouse spleen fibroblasts; PAS, periodic acid-Schiff; PH, primary hepatocytes; qRT-PCR, quantitative real-time PCR.

Meanwhile, iHeps expressed liver-specific protein markers: hepatocyte nuclear factor 4α (HNF4α), albumin (ALB), glutamine synthetase (GS) and E-Cadherin, as indicated by immunofluorescence staining (figure 2D). On day 14, the percentage of ALB-positive cells was 13%, indicating effective hepatic transdifferentiation (online supplemental figure 9). Moreover, microarray analysis indicated that iHeps were clustered with adult mouse PH but different from MFs (figure 2E). Indeed, iHeps carried the gene expression pattern of mature hepatocytes, with a notable expression of genes involved in the metabolism of drugs, glucose, fatty acids and cholesterol (online supplemental figure 10).

Next, we confirmed that iHeps exerted typical functions of mature hepatocytes. First, iHeps were capable of storing glycogens and fatty droplets, as verified by PAS and Oil Red O staining (figure 2F,G). Second, iHeps could uptake ICG (figure 2H). Third, iHeps secreted ALB and urea, as an important sign that iHeps possess the capability of hepatic synthesis (figure 2I,J). Also, iHeps exerted the cytochrome P450 activity in converting phenacetin into the metabolic product acetaminophen (APAP) (figure 2K). Taken together, overexpression of Foxa3, Gata4 and Hnf1a successfully converted the fibroblasts from the remodelled spleens into hepatocyte-like cells in vitro and these iHeps had typical morphological and functional features of the mature hepatocytes.

Induction of fibroblasts into hepatocytes in the remodeled spleens in vivo

To measure the efficiency of lentivirus transduction, we injected a LV that carried a GFP reporter gene (LV-GFP), instead of 3TF, into the SiO2-remodelled spleen and analysed spleens 7 days post-transduction. This LV-GFP transduced approximately 45% of ER-TR7+ fibroblasts (online supplemental figure 11). To examine whether 3TF was capable of inducing fibroblasts into iHeps in vivo, the mice were injected with SiO2 followed by the administration of LV-3TF. The control mice (the SiO2-NC group) were treated in the same way, except for injection of an empty LV instead of LV-3TF (figure 3A). According to the immunofluorescence co-staining of 3TF and ER-TR7, the transduction efficiency of 3TF to splenic fibroblasts was about 20% (figure 3B,C).

Figure 3Figure 3Figure 3

Induction of hepatocytes by Foxa3, GATA4 and Hnf1a in SiO2-remodelled spleen in vivo. (A) Schematic diagram of 3TF induced SiO2-remodelled spleen fibroblasts into hepatocytes in vivo. (B) The transduction efficiency of LV-3TF was measured by immunofluorescence co-staining of 3TF and ER-TR7. (C) Quantification of LV-3TF transduced in vivo. Image J was used to perform quantification analysis (n=5). (D) The morphology of the spleens treated with PBS, SiO2-NC or SiO2-3TF. (E) HE staining of spleen treated with SiO2-NC, SiO2-3TF spleen. (F) Hepatic gene expression induced by LV-3TF was analysed by qRT-PCR (n=5). (G) Immunofluorescence staining for hepatic specific markers, ALB, CYP1A2, HNF4α, CYP2C9, GS, prothrombin. (H) Glycogen synthesis was assessed by PAS staining. (I) Lipid accumulation was detected by oil red O staining. (J) ICG uptake was determined by ICG uptake assay. (K) Quantification of iHeps in total splenocytes by flow cytometry from mice treated with SiO2-NC, SiO2-3TF and representative sample data (n=5). Statistics: data are shown as means±SEM. (C, K) Student’s t-test. (F) One way ANOVA followed by Tukey’s multiple comparisons test. ALB, albumin; ANOVA, analysis of variance; GS, glutamine synthetase; ICG, indocyanine green; LV-3TF, lentiviral vectors-three transcriptional factors; PAS, periodic acid-Schiff; PBS, phosphate buffer saline; qRT-PCR, quantitative real-time PCR.

Splenic tissue samples were collected and analysed 28 days after the LV-3TF injection. LV-3TF further altered the morphology of the SiO2-remoulded spleens (figure 3D), and iHeps appeared in the spleen tissue with typical epithelial morphology and polygonal shape (figure 3E). Typical liver-specific genes were upregulated in LV-3TF treated spleens (figure 3F). We further stained the spleen tissues for ALB, CYP1A2, HNF4α, CYP2C9, GS and prothrombin—all characteristic hepatocyte markers—and confirmed the presence of iHeps in the spleens with LV-3TF treatment (figure 3G). More importantly, the LV-3TF-treated spleens exhibited capabilities of glycogen synthesis (figure 3H), lipid accumulation (figure 3I) and ICG uptaking (figure 3J).

Previous studies indicated that direct conversion of fibroblasts into hepatocytes was a stepwise transition, including the elimination of somatic memory, mesenchymal-to-epithelial transition and acquisition of hepatocyte.19 20 To determine the source of iHeps, we assessed a time-course analysis of protein expression during the initial (day 26), middle (day 40) and final phase (day 54) (online supplemental figure 12A). Immunofluorescence staining showed that ER-TR7, the splenic reticular fibroblast maker, was gradually downregulated and CYP1A2, the hepatic marker, was upregulated (online supplemental figure 12B). Also, cells expressing both ER-TR7 and CYP1A2 were observed on day 40, suggesting a conversion of fibroblasts to functioning hepatocytes in the spleen (online supplemental figure 12B). Finally, quantitative analysis of in vivo-generated iHeps by flow cytometry indicated that there were 2.06% ALB-positive iHeps among the total spleocytes from the LV-3TF treated spleens (figure 3K). Together, these results demonstrated that the fibroblasts in SiO2-remoulded spleens could be transformed into iHeps by in vivo induction of 3TF, though the total number of iHeps needed further increase to perform the physiological functions of a real liver—which was scheduled to be accomplished below.

Expansion of iHeps in vivo

TNF-α is an important factor to initiate liver regeneration and hepatocyte proliferation, by priming hepatocytes to respond to other growth factors.21–23 EGF and HGF are potent mitogens for hepatocytes and crucial for liver regeneration.24–27 We found that the levels of Egf and Hgf gene expression in iHeps were much lower than that of the PH (online supplemental figure 13). In addition, injecting EGF and HGF did not significantly increase the number of iHeps in vivo (online supplemental figure 14). Therefore, we employed TNF-α protein injection followed by transgenic expression of EGF and HGF (termed as 3C) mediated by LV (LV-Egf and LV-Hgf) to promote the proliferation of iHeps in the spleen after 3TF treatment (figure 4A). The overexpression capacity of LV-Egf and LV-Hgf was validated in AML12 and L929 cells by qRT-PCR and ELISA (online supplemental figure 15). The control mice (the SiO2-3TF-NC group) were treated in the same way, except for injection with PBS and empty LV instead of 3C.

Figure 4Figure 4Figure 4

Expansion of iHeps in vivo. (A) Schematic of the whole experimental operation. (B) The morphology and (C) weight (n=5) of the spleens. The proliferation of hepatocytes in the spleen was examined by (D) EdU staining and (E) EdU fluorescence intensity quantitative analysis. Image J was used to perform quantitative analysis (n=5). (F) Quantitative analysis of hepatocytes in total splenocytes by flow cytometry and a fluorescence-activated cell sorting analysis from a representative sample (n=5). (G) Representative histological images of He staining. (H) Representative images of immunofluorescence staining for ALB in the whole spleens. (I) Representative fluorescence Photographs of specific hepatocyte markers. (J) Immunofluorescence staining of hepatic microstructure for CD31 to detect blood vessels, vimentin to detect fibroblasts, F4/80 to detect macrophages. (K) Immunofluorescence staining of GS showing the liver zonation. Statistics: data are shown as means±SEM. (C, E, F) Student’s t-test. ALB, albumin; EdU, 5-ethynyl-2’-deoxyuridine; GS, glutamine synthetase.

After these treatments, we observed notable changes in the morphology and weight of the spleens with the 3C treatment (named as hepatised spleens), larger than the ones in the SiO2-3TF-NC group, implicating a successful iHeps amplification (figure 4B,C). EdU tests revealed that numerous cells expressing both CYP1A2 and EdU, further verifying the proliferation of iHeps after 3C stimulation (figure 4D,E). Meanwhile, microarray analysis demonstrated that the expression of cytokines, growth factors and related receptors associated with liver regeneration were up-regulated in hepatised spleens (online supplemental figure 16). Also, the proportion of iHeps in the hepatised spleen reached 7.87%, significantly greater than that with only 3TF treatments (figure 4F). When calculated based on the fluorescence-activated cell sorting (FACS) analysis of ALB-positive cells in total spleen cells, the number of iHeps in one spleen increased to approximately 8×106 (online supplemental figure 17), which would be sufficient to perform key physiological functions of the liver. In addition, histological characterisation revealed that the spleens in 3C-treated mice contained densely arrayed polygonal cells, resembling hepatocytes, and displayed inferential central vein histological structure (figure 4G). The immunostaining analysis illustrated numerous ALB positive iHeps in the whole sectioning area of the hepatised spleens (figure 4H) and revealed the expression of ALB, CYP1A2, HNF4α, CYP2C9, GS and prothrombin in the hepatised spleens (figure 4I). Notably, we observed the hepatic microstructure including blood vessels, fibroblasts and macrophages (figure 4J); the hepatised spleens also contained a liver zonation-like structure as in the native liver, as demonstrated by GS immunostaining (figure 4K).

To evaluate the biosecurity of lentiviral, we traced the region of packaging signalling sequence HIV type 1 Ψ (HIV-1 Ψ), lentiviral specifical DNA sequence, in different organs. The signal was chiefly presented in the spleen, but rarely in the liver (online supplemental figure 18A). In addition, to assess the safety of continuous expression of Egf and Hgf genes, we detected the expression levels of serum tumour markers alpha-fetoprotein, carcinoembroyonic antigen, carbohydrate antigen 125 (CA125) and carbohydrate antigen 19-9 (CA19-9). The data showed that there was no significant change in mice with the hepatised spleen (online supplemental figure 18B–E). These findings were critical to underline the safety of our protocol. The use of TNF-α, EGF and HGF effectively promoted iHeps proliferation to exert physiological functions.

Physiological functions of the hepatised spleens

We evaluated whether the hepatised spleens could perform liver functions at the physiological level. First, from microarray analysis, we observed the expression of genes representing hepatic functions as in the normal liver (figure 5A). Second, the hepatised spleens could synthesise glucose and lipids, as demonstrated by PAS and Oil Red O staining (figure 5B,C). Third, a clearance test for the ICGR15 assay is widely used to reflect the hepatic uptake functions.28 29 To exclude the interference of the native liver, hepatic portal vein and hepatic artery were ligated before intravenous injection of ICG (figure 5D). Observation of ICG in frozen sections and the ICGR15 data illustrated that the hepatised spleens gained the uptaking capacity of the liver (figure 5E,F). Fourth, we tested the enzymatic activities of cytochrome P450 in the transformed group. We injected phenacetin and diclofenac into the mice with the liver ligated in the same way (figure 5G) and perceived a high serum level of APAP and 4’-OH-diclofenac, the specific liver metabolites, in mice with hepatised spleens (figure 5H,I), suggesting that the hepatised spleens possessed these important metabolic functions.

Figure 5Figure 5Figure 5

Functional characterisation of the hepatised spleens. (A) Expression heatmaps of functional hepatocyte genes participated in drug, glucose, fatty acid and cholesterol metabolism extracted from cDNA microarray assay in the SiO2 spleen, SiO2-3TF-3C spleen and normal liver. Five tissue samples from independent experiments were pooled. (B) PAS and (C) Oil Red O staining in the spleens. (D) The experiment diagram of ICGR15 assay. After hepatic portal vein and hepatic artery were ligated, (E) ICG uptake and (F) ICGR15 test (n=5) were performed. (G) The experiment diagram of CYP450 metabolic activity test. (H) CYP1A2 and (I) CYP2C9 metabolic activity (n=5) were measured. (J) Schematic representation of murine 90% hepatectomy. (K) Survival curve, (L) ammonia, (M) glucose levels and (N) hepatic encephalopathy score in 90% hepatectomised mice were detected (SiO2 n=10, SiO2-3TF-3C n=10, normal n=5). Statistics: data are shown as means±SEM. (F, H, I) One way ANOVA followed by Tukey’s multiple comparisons test. The survival curve significance was determined by log-rank (Mantel-Cox) test. 3TF, ANOVA, analysis of variance; ICG, indocyanine green; PAS, periodic acid-Schiff.

To evaluate whether the hepatised spleens could compensate key hepatic functions in the case of liver deficiency, we performed 90% hepatectomy (figure 5J), which could cause hyperammonia, hypoglycaemic and hepatic encephalopathy.30 31 After the surgery, 5 of the 10 mice with the hepatised spleen survived, but all SiO2-treated mice died within 48 hours (figure 5K). Second, the hepatised spleens were much larger than those in the mice without 90% hepatectomy, a sign that the hepatised spleens responded to the surgery (online supplemental figure 19A,B). We further confirmed this response with EdU staining (online supplemental figure 19C) and FACS, finding that the proportion of iHeps reached 11.5% among the total splenocytes (online supplemental figure 19D). Third, we monitored the blood levels of ammonia and glucose for 14 consecutive days. The hepatectomy triggered a sharp increase in blood ammonia and a decrease in blood glucose. However, their levels were quickly controlled by day 2 and returned to normal by day 8 in the mice with hepatised spleens. This recovery did not occur in the control mice that died within 48 hours (figure 5L,M). Finally, we evaluated the hepatic encephalopathy score, an indicator of overall health,32 33 and found that the hepatised spleens helped the mice recover from the lethal pathological state (figure 5N).

We next evaluated how long the iHeps in the hepatised spleens could remain viable and functional after the final fibroblasts conversion. Through a 6-month observation, these cells survived and functioned well, as evidenced by HE staining (online supplemental figure 20A), FACS for ALB (online supplemental figure 20B), immunofluorescence co-staining of HNF4α/CYP1A2 (online supplemental figure 20C), PAS staining, Oil Red O staining and ICG uptake tests (online supplemental figure 20D–F). In addition, the iHeps in 90% hepatectomised mice increasingly improved liver functions (online supplemental figure 20G–J).

Since this approach might be applied to individuals with pre-existing chronic liver disorders, we tested the efficacy of spleen transformation in mouse models with: (1) CCl4-induced liver cirrhosis and (2) HFD-fed non-alcohol fatty liver disease (NAFLD), respectively. After the liver cirrhosis model was established (online supplemental figure 21A–C), the spleens were transformed through the same process as in the above experiments. The number of splenic fibroblasts was higher than that in normal mice and could be increased under the stimulation of SiO2 in the cirrhotic mice (online supplemental figure 21D,E). As expected, these activated fibroblasts could be converted and expanded by the 3TF-3C treatment (

留言 (0)

沒有登入
gif