Gut 61:1330-1339 doi:10.1136/gutjnl-2011-300449
  • Hepatology
  • Original article

Hepatic cell-to-cell transmission of small silencing RNA can extend the therapeutic reach of RNA interference (RNAi)

  1. Luc J W van der Laan2
  1. 1Department of Gastroenterology and Hepatology, Erasmus MC-University Medical Center, Rotterdam, The Netherlands
  2. 2Department of Surgery, Laboratory of Experimental Transplantation and Intestinal Surgery, Erasmus MC-University Medical Center, Rotterdam, The Netherlands
  3. 3Department of Surgery, Columbia University Medical Center, Columbia University, New York, USA
  1. Correspondence to Dr Luc J W van der Laan, Laboratory of Experimental Transplantation and Intestinal Surgery (LETIS), Department of Surgery, Erasmus MC-University Medical Center, Room L458, 's Gravendijkwal 230, Rotterdam 3015 CE, The Netherlands; l.vanderlaan{at}
  1. Contributors QP designed and performed the experiments, performed data analysis and wrote the manuscript. VR performed the experiments. SH constructed the vectors and performed the experiments. SF and PdR performed the experiments. JK, HWT and HLAJ contributed to the discussion and the manuscript. LJWL conceived the idea, designed and supervised the project, performed the data analysis and wrote the manuscript.

  • Revised 28 November 2011
  • Accepted 29 November 2011
  • Published Online First 23 December 2011


Background/aims RNA interference (RNAi), a sequence-specific gene silencing technology triggered by small interfering RNA (siRNA), represents promising new avenues for treatment of various liver diseases including hepatitis C virus (HCV) infection. In plants and invertebrates, RNAi provides an important mechanism of cellular defence against viral pathogens and is dependent on the spread of siRNA to neighbouring cells. A study was undertaken to investigate whether vector-delivered RNAi can transfer between hepatic cells in vitro and in mice, and whether this exchange could extend the therapeutic effect of RNAi against HCV infection.

Methods Transmission of RNAi was investigated in culture by assessing silencing of HCV replication and expression of viral entry receptor CD81 using a human hepatic cell line and primary B lymphocytes transduced with siRNA-expressing vectors. In vivo transmission between hepatic cells was investigated in NOD/SCID mice. Involvement of exosomes was demonstrated by purification, uptake and mass spectrometric analysis.

Results Human and mouse liver cells, as well as primary human B cells, were found to have the ability to exchange small RNAs, including cellular endogenous microRNA and delivered siRNA targeting HCV or CD81. The transmission of RNAi was largely independent of cell contact and partially mediated by exosomes. Evidence of RNAi transmission in vivo was observed in NOD/SCID mice engrafted with human hepatoma cells producing CD81 siRNA, causing suppression of CD81 expression in mouse hepatocytes.

Conclusion Both human and mouse hepatic cells exchange small silencing RNAs, partially mediated by shuttling of exosomes. Transmission of siRNA potentially extends the therapeutic reach of RNAi-based therapies against HCV as well as other liver diseases.

Significance of this study

What is already known about this subject?

  • RNA interference (RNAi) represents a new therapeutic modality for the treatment of diseases.

  • Cell-to-cell transmission of small silencing RNA in plants and invertebrates is critical for defence against viral infection.

  • Transmission of small silencing RNA in mammalian cells has been demonstrated in culture.

What are the new findings?

  • In vivo transmission of small silencing RNA occurs between hepatic cells in mouse liver.

  • Shuttling of small silencing RNA is independent of cell contact and is mediated in part by exosomes.

  • Transmission of small silencing RNA can extend the reach of vector-delivered RNAi.

How might it impact on clinical practice in the foreseeable future?

  • Hepatic transmission of small silencing RNA potentially extends the therapeutic reach of RNAi-based therapies against hepatitis C and other liver diseases.


The translation of molecular biology research has recently fuelled a rapid progress in the drug development for hepatitis C virus (HCV) infection. The directly acting antivirals, including a range of protease and polymerase inhibitors, are at various stages of clinical development.1 These compounds have potent antiviral activity but also dramatically potentiate the efficacy of the current standard of care, based on pegylated interferon α combined with ribavirin.2 ,3 However, given the large infected population (approximately 170 million carriers), accumulated non-responders, poor tolerability to interferon or the directly acting antivirals and special populations (eg, HIV co-infected patients and transplanted patients), novel antivirals remain urgently required, which ideally should act on distinct mechanisms and be applicable in current non-responders and special populations with fewer side effects.

RNA interference (RNAi) is a sequence-specific inhibition of gene expression at the post-transcriptional level. It is triggered by small interfering RNA (siRNA), which can be introduced into cells as synthetic siRNA or synthesised from a transgene in the cells as the short-hairpin RNA (shRNA) precursor.4 By using the cellular gene silencing/microRNA (miRNA) biogenesis machinery, these delivered siRNA induce degradation of mRNA by targeting the complementary sequences.5 This technology has now emerged as a new avenue to combat viral infections, and recent developments in the field of gene therapy have increased the feasibility of clinical applications with numerous clinical trials of RNAi currently underway ( Both the viral genome and host cellular factors involved in the viral life cycle, such as viral receptor CD81, can be targeted by RNAi and convey protection against infection.6 ,7 In the context of treating chronic HCV or preventing recurrence in HCV-positive transplants, a single dose administration with long-lasting therapeutic effects would be ideal. Integrating lentiviral vector expressing shRNA therefore represents a suitable strategy.8

In plants and invertebrates, RNAi naturally provides an important defence mechanism against pathogens. Pathogen-derived siRNA, formed by processing of double stranded RNA (replication) intermediates during infection, spreads to neighbouring cells and even propagates throughout the entire organism.9–11 This transmission of RNAi was shown to be of critical importance for plant and insect resistance against infections.9 ,11 ,12 RNAi transmission is also able to direct epigenetic modification in recipient cells in plants and conveys protection against pathogenic challenges.13 ,14 Mammalian cells such as mouse or human mast cell lines,15 the African green monkey kidney fibroblast-like cell line16 and human glioma, embryonic kidney, Epstein–Barr virus positive nasopharyngeal carcinoma and B lymphocyte cell lines16–19 have been shown to be able to transfer cellular or viral encoded miRNAs in culture via secreted exosomes in a cell contact-independent manner. In contrast, transmission of endogenous miRNA, viral miRNA or delivered small RNA between B and T cell lines in culture occurs in a cell contact-dependent manner.20

In this study we investigated the transmission of vector-derived RNAi in culture of human hepatic cells and primary human B cells and in mouse liver. We found that human and mouse liver cells and primary human B cells have the ability to exchange small RNAs including small silencing RNA as well as miRNA. We further demonstrated that transmission of gene silencing is independent of cell-cell contact and, as for miRNA, can be partially mediated by exchange of secreted exosomes. The property of hepatic cells to exchange small silencing RNAs can significantly extend the therapeutic reach of RNAi-based therapy against HCV infection and other liver diseases.

Materials and methods

Cell culture

Cell monolayers of the human embryonic kidney epithelial cell line 293T and human hepatoma cell lines Huh7, Huh6 and HepG2 were maintained in Dulbecco's Modified Eagle Medium (DMEM, Invitrogen-Gibco, Breda, The Netherlands) supplemented with 10% v/v fetal calf serum (Hyclone, Logan, Utah, USA), 100 IU/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine (Invitrogen-Gibco). Huh7 cells containing a subgenomic HCV bicistronic replicon (I389/NS3-3V/LucUbiNeo-ET, Huh7-ET) were maintained with 250 μg/ml G418 (Sigma, Zwijndrecht, The Netherlands). Primary human B lymphocytes were obtained from multiorgan donors and expanded from splenocytes using a mouse fibroblast cell line stably transfected with human CD40L. The detailed protocol has been described in our previous study.21 The Medical Ethical Council of the Erasmus MC approved the use of human samples.

Luciferase assay

Effects on HCV replication were determined based on luciferase activity. 100 mM luciferin potassium salt (Sigma) was added to Huh-7 ET cells and incubated for 30 min at 37°C. Luciferase activity was quantified using a LumiStar Optima luminescence counter (BMG LabTech, Offenburg, Germany).

miR-122 reporter assay

pMiR-Luc reporter vector expressing firefly luciferase gene incorporated with a unique miR-122 target site at its 3′UTR was purchased from Signosis (Sunnyvale, California, USA). 293T cells were transfected with the plasmid and treated with concentrated Huh7-CM or control medium for 24 h. Luciferase activity was measured as described above.

Lentiviral vectors, conditioned medium (CM) and RNAi transfer experiments

Lentiviral vectors LV-shCD81 and LV-shNS5b were constructed and produced as previously reported.8 LV-shNS5b contains expression cassettes of shRNA and targets the viral NS5b region (GACACUGAGACACCAAUUGAC 6367-6388). LV-shCD81 targets human and mouse CD81 mRNA (GGAUGUGAAGCAGUUCUAU). Lentiviral vector expressing miR-122 (LV-miR-122) was constructed by cloning the precursor sequence of mature miR-122 amplified by PCR from human genomic DNA. A third-generation lentiviral packaging system (pND-CAG/GFP/WPRE) was used to produce high-titre VSV-G-pseudotyped lentiviral vectors in 293T cells. Vector supernatants were removed 36 and 48 h after transfection, passed through a 0.45 μm filter and concentrated 1000-fold by ultracentrifugation. Concentrated virus stocks were titrated using 293T cells 24 h after infection, with transduction efficiency based on the number of green fluorescent protein (GFP)-positive cells as determined by flow cytometry (FACSCalibur; BD BioSciences, Mountain View, California, USA) after 72 h. The vector concentration was determined in 293T cells based on the number of GFP-positive cells as determined by flow cytometry. CD81 expression was determined using flow cytometry by staining with phycoerythrin (PE) conjugated mouse anti-human CD81 monoclonal antibody (BD Pharmingen, San Diego, California, USA). Mouse IgG1 was used as isotype-matched control antibody (BD Pharmingen). The effect of RNAi on CD81 expression was determined by flow cytometry.

Huh7 cells were cultured with normal culture medium. When cultures reached 60–70% confluence the cells were untransduced or transduced with LV-shCD81, LV-shNS5b or LV-shCon for 6 h, washed three times with PBS and subcultured in normal medium for more than 8 days. Conditioned medium (CM) was collected after the second refreshment of the culture medium. To generate CM specifically containing miR-122, 293T cells were transduced with LV-miR-122 or control lentiviral vector (LV-CTR). After overnight transduction, 293T cells were washed three times and cultured for up to 8 days. The CM from 293T was prepared using fresh culture medium and collected after 48 h. All CM were centrifuged at 4000 rpm for 30 min to remove cell contaminants. Concentrated CM (approximately 25–100-fold) was prepared using ultrafiltration units with a 3 kDa cut-off membrane (Millipore, Bedford, Massachusetts, USA). Huh7 cells were treated with CM for 48 h at 1:1 dilution.

Cell co-culture experiments

To generate stable shRNA integrated cell lines, naïve Huh7 cells were transduced with the lentiviral vectors and expanded in culture for at least 8 days before using in experiments. Co-culture experiments were performed for 48 h in 96-well culture plates, with 20 000 Huh7-ET HCV replicon cells per well mixed with 20 000, 10 000 or 2000 control LV-shRNA or LV-shNS5b transduced Huh7 cells. HCV replication was determined by luciferase activity. Co-culture with control (parental) Huh7 cells had no effect on HCV replication/luciferase activity and did not affect Huh7-ET cell proliferation at any condition, as measured by CFSE dilution assays. Primary human B cells were also transduced with LV-shRNA or LV-shNS5b vector to generate stable shRNA donor cells. Co-culture experiments were performed by mixing with Huh7-ET cells.

RNA transfer experiments in mice

Immunodeficient NOD/SCID mice (Charles River Laboratories, Wilmington, Massachusetts, USA) aged 3–4 weeks were used. The use of animals was approved by the institutional animal ethics committee at Erasmus Medical Center Rotterdam. Mice were engrafted with 0.5×106 Huh7-shCD81 cells (four mice) or Huh7-shCon cells (seven mice) injected intrasplenically. Cell transplantations and surgical procedures were performed under 1.5% isoflurane inhalation anaesthesia and a prophylactic antibiotic was given. Two and a half weeks after engraftment the mice were killed and liver tissue was obtained for analysis. To demonstrate cell-free transfer of small RNA, NOD/SCID mice were intravenously injected with 200 μl 100-fold concentrated shCD81-CM or shCon-CM every 2 days on three occasions (four animals per group). After 6 days the mouse livers were procured, dissociated by collagenase digestion22 and analysed for CD81 expression by flow cytometry.

Exosome purification and electron microscopy imaging

Exosomes were prepared from the supernatant of Huh7 cells by differential centrifugation. Briefly, supernatant was centrifuged at 3000g for 20 min to eliminate cells and at 10 000 g for 30 min to remove cell debris. Exosomes were pelleted by ultracentrifugation (Beckman SW28) at 64 047g for 110 min followed by a sucrose gradient isolation at 100 000 g (Beckman SWTi60). For uptake experiments, 0.1% Rhodamine C18 solution was added to the sucrose before centrifugation. For electron microscopy, exosomes were visualised by negatively staining using uranyl acetate.

Exosome uptake and RNAi transfer

For visualisation of exosome uptake, Huh7 cells were seeded on glass cover slips. Rhodamine-labelled exosomes were added to live cells on coverslips in a heated chamber (37°C) and uptake was measured in real time by confocal microscopy (Zeiss LSM510META). To determine the kinetics of exosome uptake, images were taken every minute for 45 min. Paraformaldehyde (PFA)-fixed cells served as controls to exclude passive transfer of Rhodamine by exosome cell fusion. In order to specify the subcellular localisation of exosomes, nuclear staining using the Hoechst dye was performed. In these experiments, measurements were taken at only two time points (1 and 30 min after adding exosomes) to avoid cytotoxicity of Hoechst induction by the laser and decay of the nuclear staining.

RNAi transfer by purified exosomes was tested by treating Huh7-ET cells with shNS5-containing exosomes for 48 h and viral replication was measured based on luciferase activity. Similarly, Huh7 cells were treated with shCD81-containing exosomes for 48 h and CD81 cell surface expression was quantified by flow cytometry.

Mass spectrometric analysis

Two batches of purified exosomes were subjected to mass spectrometry at the Erasmus MC Proteomics Center. Briefly, 1D SDS-PAGE gel lanes were cut into 2 mm slices using an automatic gel slicer and subjected to in-gel reduction with dithiothreitol, alkylation with iodoacetamide and digestion with trypsin (Promega sequencing grade), essentially as described by Wilm et al.23 Nanoflow LC-MS/MS was performed on a 1100 series capillary LC system (Agilent Technologies) coupled to an LTQ-Orbitrap mass spectrometer (Thermo) operating in positive mode and equipped with a nanospray source. Peptide mixtures were trapped on a ReproSil C18 reversed phase column (Dr Maisch GmbH; column dimensions 1.5 cm×100 μm, packed in-house) at a flow rate of 8 μl/min. Peptide separation was performed on ReproSil C18 reversed phase column (Dr Maisch GmbH; column dimensions 15 cm×50 μm, packed in-house) using a linear gradient from 0% to 80% B (A=0.1% formic acid; B=80% (v/v) acetonitrile, 0.1% formic acid) for 70 min at a constant flow rate of 200 nl/min using a splitter. The column eluent was directly sprayed into the ESI source of the mass spectrometer. Mass spectra were acquired in continuum mode; fragmentation of the peptides was performed in data-dependent mode. Peak lists were automatically created from raw data files using the Mascot Distiller software Version 2.3 (MatrixScience). The Mascot search algorithm Version 2.2 (MatrixScience) was used for searching against a customised database containing all IPI_human protein sequences (release 2010_09). The peptide tolerance was typically set to 10 ppm and the fragment ion tolerance was set to 0.8 Da. A maximum number of two missed cleavages by trypsin was allowed and carbamidomethylated cysteine and oxidised methionine were set as fixed and variable modifications, respectively. The Mascot score cut-off value for a positive protein hit was set to 65. Individual peptide MS/MS spectra with Mascot scores below 40 were checked manually and either interpreted as valid identifications or discarded. Typical contaminants also present in immunopurifications using beads coated with pre-immune serum or antibodies directed against irrelevant proteins were omitted from the table.

RNA isolation and real-time RT-PCR analysis

Total RNA was extracted using the miRNeasy mini kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). Mouse liver tissues were mechanically disrupted and lysed using Trizol (Invitrogen-Gibco). RNA was quantified using a Nanodrop ND-1000 (Wilmington, Delaware, USA). cDNA was prepared from 1 μg total RNA using a iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Stanford, California, USA). The cDNA of mouse CD81, TBP, CyB and GAPDH was quantified using real-time PCR (MJ Research Opticon, Hercules, California, USA) performed with SybrGreen (Sigma-Aldrich) according to the manufacturer's instructions. CD81 mRNA levels were normalised to the average level of the three independent reference genes using the ddCT method. TaqMan-based real-time PCR kit for detection of miR-122 was purchased from Applied Biosystems and analysis was performed according to the manufacturer's instructions. A customised kit for quantification of small silencing RNA was designed by amplification of the antisense sequence of shCD81 (UAGAACUGCUUCACAUCC) using TaqMan-based real-time PCR from Applied Biosystems. The assay is supposed to amplify the mature miR-122 or siCD81 preferentially but may also detect the precursors.

Fluorescent immunohistochemistry

Mouse liver tissue was dissected and cryoprotected in 30% sucrose for generation of frozen sections. Serial 6 μm cryosections were air-dried for 48 h at room temperature followed by a washing step with PBS. Sections were fixed with 50% acetone in PBS for 10 min on ice and blocked in PBS containing 4% fat-free milk for 1 h at room temperature. The sections were incubated with Alexa Fluor647 labelled anti-mouse CD81 antibody (AbD Serotec, Oxford, UK) at a dilution of 1:100 for 30 min. After three washes, nuclear staining was achieved by incubating with DAPI (Sigma-Aldrich) at a dilution of 1:50 for 5 min. Multiple areas from the mouse liver tissue surrounding nodules of engrafted Huh7 cells were analysed by confocal microscopy. The Huh7 nodules were distinguished from liver parenchyma on the basis of GFP positivity and tumour morphology.

Statistical analysis

Statistical analysis was performed with either a matched-pair non-parametric test (Wilcoxon signed-rank test) or a non-paired non-parametric test (Mann–Whitney test) using GraphPad Prism software. p Values <0.05 were considered as statistically significant.


Transmission of lentiviral vector-delivered RNAi targeting HCV receptor or viral genome

We have constructed a lentiviral vector, LV-shNS5b, which contains both the GFP reporter gene and a shRNA targeting the HCV NS5b region, which encodes the viral RNA-dependent RNA polymerase. We used a subgenomic HCV replication model based on a Huh7 hepatoma cell line containing the non-structural sequence of the HCV genome with a luciferase reporter gene (Huh7-ET), mimicking viral replication without virus particle production.24 As reported previously, LV-shNS5b resulted in a mean±SD maximum inhibition of HCV replication of 98±0.5% (n=8, p<0.001) at the highest transduction efficiency.8 ,25 However, at suboptimal transduction efficiency, the percentage inhibition of viral replication as measured by luciferase activity significantly exceeded the percentage of transduced cells as measured by GFP expression (figure S1 in online supplement). For example, with a transduction efficiency of 45% GFP, the observed inhibition of HCV replication was 58%, suggesting possible extension of RNAi to non-transduced cells. Similar results were observed with the LV-shCD81, a vector containing GFP and shRNA targeting the HCV receptor CD81. LV-shCD81 significantly reduced CD81 cell surface expression in transduced Huh7 cells (mean±SD inhibition 92.9±5.9%, n=8, p<0.001), but also significantly reduced CD81 expression in non-transduced GFP-negative cells (30.1±12.9% inhibition, p<0.001; figure 1A). CD81 reduction was not related to loss of cell viability as dead/permeable cells were excluded from analysis.

Figure 1

Evidence for intercellular functional transmission of small silencing RNAs. (A) Silencing of CD81 expression by the lentiviral vector LV-shCD81 extended to green fluorescent protein (GFP)-negative non-transduced cells. Huh7 cells were transduced by LV-GFP containing either a CD81 targeting shRNA (LV-shCD81) or a scrambled control shRNA (LV-shCon). The upper panel shows a representative histogram of GFP fluorescence intensity. The lower panels show flowcytometric analysis of CD81 staining ingated GFP-negative (left panel) and GFP-positive (right panel) cellstransduced with LV-shCon (see arrow) orLV-shCD81 (see arrow). The left lines show isotype-matched control staining. (B) Hepatitis C virus (HCV) replicon cells (Huh7-ET) were directly co-cultured with cells stably expressing shNS5b (Huh7-shNS5b) or control shRNA (Huh7-shCon) or treated with conditioned culture medium (CM) of these cells. (C) Significant inhibition of HCV replication was observed at ratios of 1:1 and 1:0.5 Huh7-ET with Huh7-shNS5b compared with co-cultures with Huh7-shCon or untreated cells. Mean±SD values of six independent experiments are shown (**p<0.01). (D) Huh7-ET replicon cells treated with shNS5b-CM (at final concentration 50%) but not shCon-CM showed significantly reduced HCV replication of 39±12% (n=9, **p<0.01) compared with untreated controls. (E) Huh7-ET cells were co-cultured with primary human B cells stably expressing shNS5b or shCon. (F) Significant reduction of viral replication was observed when co-cultured with B cells expressing shNS5b at a ratio of 1:1 (n=4, **p<0.01). Such an effect was also confirmed at a ratio of 1:5 (n=3, *p<0.05), although a high density of B cells appeared to cause some non-specific effects. This figure is produced in colour in the online journal–please visit the website to view the colour figure.

To ensure that the gene silencing effect on GFP-negative cells was not due to insensitivity of GFP detection or silencing of transgenic expression, additional co-culture experiments were performed (figure 1B). A significant inhibition of HCV replication was observed when Huh7-ET HCV replicon cells were co-cultured with naïve Huh7 cells stably expressing shNS5b at a 1:1 ratio (51±12%, n=6, p<0.01) compared with Huh7-shCon co-cultures and untreated Huh7 cells (figure 1C). A similar effect was observed at a lower ratio of 1:0.5 Huh7-ET and Huh7-shNS5b cells, but lost significance when co-culturing at very low ratios (figure 1C). To confirm that primary human cells and other cell types also transmit RNAi, we tested primary B lymphocytes isolated from human spleen stably transduced with shRNA vectors (figure 1E). Similar to the co-culture with transduced Huh7 cells, a significant inhibition of viral replication was observed in replicon cells co-cultured with B cells expressing shNS5b at a ratio of 1:1 (p<0.01) or 1:5 (p<0.05) compared with co-cultures of B cells with shCon (figure 1F).

RNAi transmission has been reported to be dependent20 or independent15 on cell contact depending on the model used, although the exact mechanisms remain largely elusive. Using immunofluorescence microscopy, we observed that LV-shCD81-dependent knockdown of CD81 expression in GFP-negative cells was not restricted to cells in direct contact with GFP-positive cells but, rather, a general pattern of CD81 reduction was seen (data not shown). To further investigate whether RNAi can be transmitted in the absence of direct cell-cell contact, CM was prepared from stably transduced Huh7 cells expressing shCon, shCD81 or shNS5b (figure 1B). As shown in figure 1D, exposure of Huh7-ET cells to shNS5b-CM (at a final concentration of 50%) specifically reduced HCV replication by 39±12% (n=9, p<0.01) without transfer of GFP positivity. Treatment with shCD81-CM also significantly reduced CD81 expression in Huh7 cells (23.5±5.1% inhibition, n=7, p<0.01). These results suggest that transmission of RNAi is independent of cell contact but rather seems to involve the uptake of released silencing RNA components.

Functional transmission of liver abundant miRNA

We further investigated whether the cell contact-independent manner of small RNA transmission also exists for endogenous miRNA. Huh7 cells highly express miR-122, a liver abundant miRNA that has been reported to be a crucial positive regulator of HCV replication and translation.26 We found that cell-free CM of Huh7 cells (Huh7-CM) contained high levels of miR-122 (figure S2 in online supplement). Concentration of Huh7-CM (Huh7-C-CM) using ultrafiltration resulted in a 10-fold increase in miR-122 levels. The miR-122 level of Huh7 cells is more than 200-fold higher than another hepatoma cell line HepG2 and over 50 000-fold higher than the embryonic kidney epithelial cell line 293T (figure S2 in online supplement). Treatment of HepG2 cells with Huh7-CM or Huh7-C-CM significantly increased intracellular miR-122 levels by 3–4-fold (p<0.01), indicating uptake of miR-122 from the medium. An even more pronounced miRNA uptake was observed in 293T cells, leading to an increase in cellular miR-122 levels of about 20-fold or 1750-fold after exposure to Huh7-CM and Huh7-C-CM, respectively (figure 2A). miRNA transfer was also observed in freshly isolated human peripheral blood mononuclear cells, and incubation with Huh7-CM resulted in an increase in cellular miR-122 levels of approximately 100-fold (figure 2B). To demonstrate the transfer of miRNA more specifically and to exclude possible induction of miRNA gene expression by other factors present in CM, we generated a lentiviral vector specifically expressing the precursor of miR-122 (LV-miR-122). CM was produced from LV-miR-122 or control vector (LV-shCon) transduced 293T cells. 293T cells naturally expressed very low levels of miR-122 and transduction with LV-miR-122 (∼5% transduction efficiency) resulted in an increase in cellular miR-122 levels of approximately 10-fold. As shown in figure 2C, miR-122-CM but not shCon-CM specifically increased the cellular miR-122 levels in 293T cells by approximately fivefold. Similarly, incubation with miR-122-CM increased the cellular miR-122 levels of the T cell line (SupT1 cells) by approximately 15-fold (figure 2D). To investigate the functionality of miRNA transmission, a reporter plasmid-expressing luciferase gene coupled with miR-122 complementary sequence was used to transfect 293T cells. As shown in figure 2E, treatment of concentrated Huh7-CM significantly reduced the miR-122-targeted luciferase activity compared with untreated or control medium-treated conditions (p<0.01). This confirms functional regulation of target reporter gene expression by transferred miRNA.

Figure 2

Evidence for intercellular functional transmission of liver abundant microRNA (miRNA). (A) Uptake of miR-122 by 293T cells after exposure to Huh7-CM or Huh7-C-CM. (B) Peripheral blood mononuclear cells from healthy controls showing an increase in the cellular miR-122 level of about 100-fold after 6 h of incubation with Huh7-CM. (C) To confirm miRNA transfer and rule out the induction of miRNA gene expression by other factors present in conditioned medium (CM), we generated CM of 293T cells transduced by the lentiviral vectors LV-miR-122 or LV-shCon. Treatment of naïve 293T cells with miR-122-CM but not shCon-CM increased the cellular miR-122 level by approximately fivefold. (D) Incubation with miR-122-CM resulted in about 15-fold increase of cellular miR-122 levels in the T cell line, SupT1 cells. Data shown are mean±SD of three or four independent experiments. (E) Treatment of concentrated Huh7-CM resulted in a significant reduction in miR-122-related luciferase activity in 293T cells transfected with miR-122 reporter plasmid compared with the group treated with control medium or the untreated group. Data shown are mean±SD of three independent experiments (n=11 replicates in total, **p<0.01).

Secreted exosomes contain small RNAs and RNA binding proteins

Previous studies have shown that cellular miRNA can be released from cells by secretion of microvesicles/exosomes.15 ,17 To further investigate whether exosomes are involved in the transfer of small silencing RNA, we purified secreted exosomes from Huh7-CM or CM of stably transduced Huh7 cells expressing shCon, shCD81 or shNS5b using density gradient ultracentrifugation. Figure 3A shows an electron micrograph of a purified exosome. RT-PCR analysis of shCD81-CM exosomes revealed the presence of both miRNA (miR-122) and shCD81 (figure 3B). Huh7-CM-derived exosomes were analysed by mass spectrometry to characterise the protein content. From two independent preparations of exosomes, over 600 common proteins were detected including the established exosome markers Tsg101, CD63, CD9, Alix, Flotillin and RAB5.27 These proteins were further categorised according to their location and function (see figure S3A and B in the online supplement). Importantly, 56 distinct RNA binding proteins were present, including ribosomal proteins, serine/arginine-rich splicing factors, heterogeneous nuclear ribonucleoproteins, eukaryotic translation initiation factors and proteasome subunits (see figure 3C and online table). Relevant to the content of miRNA and siRNA, we identified four proteins in exosomes which are known to be important for the miRNA pathway and which are potential binding partners of the small silencing RNA cargo in hepatic exosomes (figure 3D). Of particular interest is the nucleolar phosphoprotein B23 (NPM1) which has recently been shown specifically to protect the degradation of miRNAs.28 RAN, the Ras-related nuclear protein, is known to be involved in nucleocytoplasmic transport. Interestingly, recent studies have shown that Exportin-5-mediated nuclear export of pre-miRNA or shRNA acts in a Ran-GTP-dependent manner.29–31 Further studies will be required to identify the exact molecular machinery involved in the sorting and packaging of small silencing RNA into exosomes.

Figure 3

Exosomes contain small RNAs and RNA binding proteins. Secreted exosomes were purified from conditioned medium (CM) from Huh7 cells using density gradient ultracentrifugation. (A) Electron micrograph showing the presence of exosomes in the purified fraction. (B) RT-PCR analysis of purified exosomes from shCD81-CM showing the presence of both miRNA and shCD81. Markers indicate the anticipated amplicon size for miR-122 and shCD81. No-template (H2O) and purified exosomes from shNS5b-CM served as negative controls. This assay is supposed to amplify the mature miR-122 or siCD81 preferentially but may also detect the precursors. (C) Mass spectrometry was performed to analyse the protein content of two independent batches of Huh7-CM-derived exosomes. Using a Mascot cut-off for specificity (Mascot >40), a total of >600 common proteins were identified including many exosome-specific proteins; 56 proteins are known RNA-binding proteins, including 32 ribosomal proteins. (D) Of the RNA-binding proteins, four are known to be involved in the miRNA pathway and are potentially involved in the selection, sorting and packaging of small silencing RNA in hepatic exosomes. The protein name, main function and relative abundance in exosomes indicated by the amPAI value (mean of two samples) are shown.

Transmission of gene silencing is partially mediated by exosomes

To investigate the involvement of exosomes in small RNA transfer, real-time live cell imaging was performed with Huh7 cells exposed to fluorescent-labelled exosomes using confocal microscopy. Real-time analysis showed that exosome uptake is rapid and occurs within 45 min (data not shown). As shown in figure 4A, ingested exosomes predominantly accumulate in the cytoplasm or other intracellular compartments but not in the nucleus. Exosome uptake was observed in most of the living cells (>80%) but little uptake occurred in PFA-fixed cells (figure 4B), confirming that uptake is an active process. Treatment of HCV replicon cells with purified exosomes derived from shNS5b-CM resulted in a significant reduction of viral replication (mean±SD inhibition 21.6±6.4%, n=4, p<0.01; figure 4C). Similarly, treatment of Huh7 cells with exosomes derived from shCD81-CM resulted in a significant downregulation of CD81 cell surface expression (reduction of 24.5±3.1%, n=4, p<0.05; figure 4D). These findings confirm that secreted exosomes containing small RNAs, including miRNA and small silencing RNA, can mediate transmission of functional gene silencing. In addition, recent studies have suggested the co-existence of exosome-dependent and exosome-independent pathways of small RNA release and transfer.20 ,32 ,33

Figure 4

Exosome-mediated functional transmission of small silencing RNAs. (A) Dynamic visualisation of rhodamine-labelled exosome uptake by live Huh7 cells showing intracellular accumulation in viable cells but not in paraformaldehyde (PFA)-fixed cells (B). Red staining represents exosomes and blue staining indicates the nucleus. Shown is one of three independent experiments (magnification 800×). (C) Treatment of Huh7-ET replicon cells with purified exosomes derived from shNS5b-CM significantly reduced viral replication by 21.6±6.4%. (D) Treatment of normal Huh7 cells with purified exosomes derived from shCD81-CM resulted in a significant downregulation of CD81 cell surface expression by 24.5±3.1%. Mean±SD inhibition of four independent experiments is shown (*p<0.05). This figure is produced in colour in the online journal–please visit the website to view the colour figure.

Transmission of gene silencing in mice liver

To explore the evidence for small RNA exchange in vivo, we engrafted Huh7-shCD81 cells, stably expressing shRNA targeting mouse CD81, or Huh7-shCon cells, containing irrelevant shRNA, in the liver of immunodeficient NOD/SCID mice by intrasplenic injection (figure 5A). Human hepatomas in the mouse liver tissue were visualised based on GFP positivity. Mouse liver tissue surrounding nodules of Huh7-shCon cells showed comparable CD81 expression (figure 5B) to that of untreated mice (figure 5C). On the contrary, liver tissue adjacent to Huh7-shCD81 nodules showed a marked reduction in CD81 expression (figure 5D). Flow cytometric quantification of mouse-specific CD81 expression on dissociated liver cells showed a mean reduction of 71.3% on both hepatocytes and non-parenchymal cells in Huh7-shCD81 versus Huh7-shCon engrafted mice (p=0.002, figure 5E). This finding suggests transfer of RNAi from the human cells to the primary mouse cells in vivo. In order to determine whether RNAi transfer in vivo is dependent on cell contact, NOD/SCID mice were intravenously treated with shCD81-C-CM or shCon-C-CM (figure 5A). At day 6 a significant reduction in the CD81 mRNA level was observed in mouse livers by shCD81-CM treatment (mean±SD reduction of 31.6±15.6%, n=4) compared with the shCon-CM controls (n=4, p<0.05; figure 5F). Consistent with the gene expression levels, an approximate reduction of 20% in CD81 cell surface expression was observed in both hepatocytes and non-parenchymal cell populations by flow cytometry (figure 5G). The gene silencing by shCD81-CM was comparable to that of liposome or nanoparticle delivery of siRNA observed in a transgenic mouse model of HCV or in human tumours.34 ,35 Despite earlier reports of hepatotoxicity by adeno-associated vector-mediated RNAi,36 we observed no evidence of liver injury by histology or serum transaminases as a result of shCD81-CM treatment.

Figure 5

In vivo evidence for transmission of RNA interference (RNAi) in mice. (A) Schematic representation of in vivo experiments with immunodeficient mice. (1) NOD/SCID mice were either engrafted with control Huh7 cells expressing irrelevant shRNA targeting NS5b (Huh7-shCon) or Huh7 cells expressing shRNA targeting murine CD81 mRNA (Huh7-shCD81) in the liver. (2) Alternatively, NOD/SCID mice were injected intravenously with 200 μl of 100-fold concentrated cell-free conditioned medium (CM) from Huh7-shCD81 or Huh7-shCon cells three times at 48 h intervals (four animals in all groups). Confocal immunofluorescence staining using an anti-mouse CD81-specific antibody showed normal CD81 expression (red fluorescence) in the mouse liver tissue (M) surrounding nodules of Huh7-shCon cells (H) (B), comparable to expression in untreated mice (C). (D) CD81 expression in mouse liver tissue (M) surrounding Huh7-shCD81 cells (H) was markedly reduced. (E) Flow cytometric quantification of dissociated liver cells showing a significant reduction in CD81 expression in mouse hepatocytes and non-parenchymal cells (mean reduction of 71.3%, p=0.002) in mice engrafted with Huh7-shCD81 (bottom panels) compared with mice engrafted with Huh7-shCon (top panels). Green fluorescent protein (GFP)-positive human cells were gated out and a mouse specific anti-CD81 antibody was used to specifically determine mouse CD81 expression. The data shown are the mean geometric mean fluorescence intensity. (F) Analysis of liver mRNA showed a significant knockdown of CD81 expression in mice treated with shCD81-CM as compared with shCon-CM treatment. (G) Knockdown of CD81 surface expression was confirmed by flow cytometry in both hepatocyte and non-parenchymal cell populations (approximately 20%; *p<0.05). This figure is produced in colour in the online journal–please visit the website to view the colour figure.


From the discovery of RNAi in 199837 to the approval of RNAi therapeutics or RNAi-based gene therapy by the FDA, its application has been dramatically changing. Much attention has been given to developing antiviral RNAi against, for example, HIV,38 HBV39 or HCV6 infection. If RNAi therapies are to be used as an effective treatment or prevention of HCV infection, long-term stable siRNA expression needs to be achieved. Raw synthetic siRNA or plasmid-encoded shRNA transfections elicit only short-term silencing, whereas viral vectors that encode for shRNA can potentially induce long-term and continuous gene silencing.6 Adeno-associated viral (AAV) vectors are currently considered the prime candidate for clinical gene therapy applications, including the treatment of various liver diseases. Biotech companies such as Tacere Therapeutics have pioneered the development of an AAV-based anti-HCV RNAi regimen termed ‘TT-033’ ( However, AAV-mediated expression of shRNA was shown to evoke liver toxicity in mice, ultimately causing death.36 It was suggested that the saturation of the endogenous miRNA processing machinery by overexpressed shRNA is the potential cause,40 but the exact mechanism remains unclear. Lentiviral vector represents another promising candidate for clinical RNAi delivery. Although certain lentiviral RNAi systems—such as some commercial RNAi libraries—express high levels of shRNA and cause disturbance of cellular miRNA machinery, no significant cell toxicity was observed.41 The lethal toxicity observed by Grimm et al36 could be caused by the combination of AAV vector and overexpressed shRNA. Of note, the lentiviral RNAi vectors used in this study express moderate levels of shRNA without a clear effect on the miRNA pathway.41 To overcome the potential toxicity and off-target issues, liver-specific promoters42 or miRNA-based RNAi constructs43 have been used to generate safer vectors.

Although studies have demonstrated the feasibility of combating HIV infection by the ex vivo delivery of lentivral RNAi,44 it remains a challenge to produce sufficient vectors targeting the entire liver organ. For almost any type of vector, it is not possible to achieve 100% transduction efficacy in patients. The finding in the current study that gene silencing can transfer to neighbouring non-transduced cells could potentially overcome this issue of suboptimal vector transduction to a certain extent. Whether it would be sufficient to silence the virus in the non-transduced cells solely via the RNAi transmission route remains questionable. Like HIV,45 HCV is prone to develop resistant mutants if the antiviral potency is suboptimal. Vector simultaneous delivery of multiple shRNAs targeting different regions of the virus or a combination of targeting host factors could be one solution to prevent mutagenesis,46 since the non-transduced cells could receive multiple antiviral shRNAs even though the levels are less abundant. As with other new antivirals,2 ,3 combination with interferon is likely to be required for RNAi-based therapy to achieve ultimate success in patients chronically infected with HCV.47

The mechanism of RNAi transmission in plants and invertebrates has been proposed to be via direct cell-to-cell contact or systemic spreading, although the exact mechanism remains unclear. Rechavi et al reported that transmission of small RNA between B and T cell lines in culture is dependent on cell contact,20 whereas many others15–19 have described a secretory transmission pathway involving exosomes in different mammalian cell culture systems. In this study we also observed the release and uptake of small RNA-packed exosomes by hepato-like cells. We further performed mass spectrometric analysis to characterise the protein content of these exosomes. Along with previous studies characterising exosomes derived from monocytes48 ,49 or hepatocytes,50 there appears to be some cell type specificity. For instance, AGO2, a protein involved in the RNAi machinery, is detectable in monocytes48 ,49 but not in hepatocyte-derived50 exosomes. The differential enrichment of nucleotide and nucleic acid binding proteins has been observed between Huh7 and primary hepatocyte-derived exosomes.50 Interestingly, we found several proteins present in our exosomes that potentially contribute to the functional transmission of small RNAs. RAN, which is known to have a role in nucleocytoplasmic transport, was demonstrated to be involved in Exportin-5-mediated nuclear export of pre-miRNA or shRNA.29–31 The co-presence of NPM1, which can specifically protect the degradation of miRNAs,28 and TUTase, which can potentially edit miRNAs or shRNAs,51 suggests that the process of degradation and modification of small RNAs can be potentially regulated within the exosomes. It is possible that both the protein and RNA content deters the function of transferred exosomes.

In the in vivo experiments in mice, we observed a robust transmission of RNAi in the liver (figure 5). The different potency of CD81 knockdown between animals engrafted with Huh7 cells and those treated with shRNA-CM is probably due to differences in the efficacy of RNAi delivery. Most efficient RNAi transfer was observed with engrafted Huh7 cells which provided a continuous source of siRNA, whereas treatment with CM only provided a transient delivery. Moreover, the proximity to the RNAi sources could be a factor. We assumed that exosomes only partially mediated the transmission of gene silencing and the other part would be contributed by the secreted small RNAs independent of exosomes. This is highlighted by the fact that RNAi transfer was more effective with CM than with purified exosomes (figures 1 and 4). The discrepancy between the exosome uptake efficiency (figure 4A, >80% positive cells) and RNAi transfer efficiency (figure 4C,D) can be explained by the cargo. Not all exosomes derived from donor cells may contain appropriate levels of shNS5b or shCD81 required for effective transfer of RNAi. A recent study showed that substantial amounts of extracellular miRNAs are associated with Argonautes,32 which could represent an exosome-independent pathway of RNAi exchange. Further studies are required to identify the exact molecular machinery used in regulating the release and uptake of functional small RNAs.

In summary, this study has provided in vitro and in vivo evidence that small RNA could be exchanged between hepatic cells and that this property extends RNAi-mediated gene silencing against HCV receptor or viral genome. Exchange of small RNAs in our models was independent of direct cell-to-cell contact and appeared to be mediated by the secretory pathway partially involving exosomes. Cells that stably express shRNA such as stem cells may represent an effective method for the therapeutic delivery of RNAi in vivo. These findings may be relevant for the clinical application of RNAi-based therapy in the treatment of chronic hepatitis C as well as in metabolic and immunomediated liver diseases.52


The authors would like to thank Dr Jeroen Demmers for helping with mass spectrometry (Erasmus MC Proteomics Center), Dr Pascal van der Wegen, Dr Bob Scholte, Dr Rob Willemsen and Dr Guido Jenster (Erasmus MC, Rotterdam) for technical support, and Professor Ralf Bartenschlager and Dr Volker Lohmann (University of Heidelberg, Germany) for generously providing the Huh7 and Huh6 subgenomic HCV replicon cells. We also thank the Erasmus MC Translational Research Fund and the Liver Research Foundation (SLO) Rotterdam for financial support.


  • Funding Financial support was provided by the Erasmus MC Translational Research Fund and the Liver Research Foundation (SLO) Rotterdam.

  • Competing interest None.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Relevant Article

Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of Gut.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.