| |
Targeting siRNA to arrest fibrosis
|
| |
| |
Scott L Friedman1
1. Scott L. Friedman is in the Division of Liver Diseases, Mount Sinai School of Medicine, 1425 Madison Ave., New York, New York 10029, USA. e-mail: scott.friedman@mssm.edu
News and Views
Nature Biotechnology 26, 399 - 400 (2008)
doi:10.1038/nbt0408-399
Vitamin A-mediated targeting of small interfering (si)RNA to stellate cells reverses liver cirrhosis.
Introduction
Hepatic fibrosis is a scarring response to injury that underlies the development of end-stage liver disease, or cirrhosis, in hundreds of millions of patients with chronic hepatitis worldwide. Despite substantial progress in uncovering the cellular and molecular mechanisms of fibrosis, to date there are no approved therapies. In this issue, Sato et al.1 serve up a technical tour de force that could bring us closer to this elusive goal. Using several rodent models of hepatic fibrosis, they demonstrate a novel system for delivering an antifibrotic siRNA to hepatic stellate cells-the primary source of extracellular matrix, or collagen scar tissue, produced in response to liver damage2 (Fig. 1). A dramatic reversal of fibrosis and cirrhosis was observed after several treatments, attesting not only to the therapeutic potential of this approach but also to the remarkable resilience and regenerative capacity of liver.
This study is among the first to successfully target stellate cells in vivo, using an elegant liposomal delivery system that exploits the uptake by these cells of vitamin A linked to liposomes. The liposomal targeting complex cleverly incorporates two critical constituents: vitamin A and an siRNA against the collagen chaperone heat shock protein 47 (HSP47).
Inclusion of vitamin A confers a high degree of specificity for hepatic stellate cells, the primary site for storage of dietary vitamin A (retinoid). Ordinarily, dietary retinoids delivered to the liver are first processed by hepatocytes, the major parenchymal cell type, and then transferred to stellate cells via a retinol-binding protein-dependent pathway3. Thus, it is interesting that the vitamin A-coupled liposomes were not first recognized by hepatocytes but instead delivered intact directly to stellate cells bound to retinol-binding protein. This was confirmed using gel filtration and by the capacity of antibodies to retinol-binding protein to inhibit uptake of the liposomes. Although some uptake by hepatic macrophages was noted, the absence of target sequence for the siRNA minimizes the impact of nonspecific uptake in these phagocytic cells. No uptake was seen in retinal cells, which are also rich in retinoids, presumably because the blood-brain barrier prevents the liposome complexes from reaching the eye.
The choice of HSP47 as the siRNA target is intriguing, as only a few studies have examined its role in liver fibrosis. HSP47 expression in the endoplasmic reticulum correlates closely with collagen production, and in liver is localized to collagen-producing activated stellate cells4, 5, 6.
Having first established the rationale for inhibiting collagen production by knocking down this chaperone in cultured cells, the authors extended their investigations to three mechanistically distinct in vivo models. These involved using bile-duct obstruction or administering dimethylnitrosamine or carbon tetrachloride, each of which mimics cirrhosis. An impressive number of control experiments were performed to confirm the specificity of gene targeting and to exclude off-target effects due to an endogenous interferon response.
The effects of the liposome complex on liver fibrosis were striking and required all components-liposome, vitamin A and siRNA-for activity. Using three models to ensure that the effect of HSP47 antagonism was relevant to all etiologies of liver fibrosis, the authors found that repeated intravenous administration of the complex dramatically reversed advanced fibrosis, eliminated features of severe liver disease and prolonged survival. Although reduced collagen secretion was a major consequence of the treatment, there was also an increase in collagenase activity, which was essential to degrade the large mass of scar that had already accumulated.
The molecular basis for the antifibrotic effect of increasing matrix degradation is obscure and among the most intriguing findings of the study. Although efforts to combat hepatic fibrosis by targeting production of collagen, a major constituent of the hepatic scar, have been considered for decades, enthusiasm had waned more recently owing to the view that targeting a single participant in the complex cellular milieu of fibrosis would be insufficient to reverse the disease. The findings of Sato et al.1 thus raise speculation whether HSP47 antagonism affects other targets besides collagen, such as elements of the immune system, or whether the downstream impact of inhibiting collagen deposition by this route is far broader than previously thought. One way to identify unanticipated targets of HSP47 knockdown would be to use microarrays to survey the molecular effects of the liposome complexes in cultured stellate cells. Perhaps reduced expression of collagen would alter the balance between matrix production and degradation by downregulating tissue inhibitors of metalloproteinases, which are a critical switch in mediating both matrix degradation and stellate cell survival7.
Another implication of the study is the potential use of this targeting complex to administer other therapies directly to hepatic stellate cells, including other siRNAs or small molecules. The latter might be particularly useful to deliver small antifibrotic molecules, such as hepatocyte growth factor (HGF), while minimizing their impact on other cell populations, as untargeted HGF carries the theoretical risk of enhancing liver carcinogenesis through stimulation of hepatocyte mitogenesis. Moreover, as our appreciation of the range of functions of the hepatic stellate cell increases1, indications other than fibrosis could emerge, such as enhancing regeneration or modulating immune responses. Finally, the complex could be used to deliver diagnostic or imaging molecules for assessment of fibrosis, currently a major unmet clinical need8.
Although the results of this study are both technically and conceptually impressive, it seems unlikely that this targeting complex will become a widely used therapy for chronic hepatic fibrosis, as a range of experimental oral or parenteral therapies currently seem more appealing. Nonetheless, one could envision its use as an initial therapy to achieve rapid, potent antifibrotic activity followed by oral therapies to sustain the therapeutic benefit. Clearly, this technology opens new doors to improved understanding, diagnosis and treatment of hepatic diseases.
Resolution of liver cirrhosis using vitamin A-coupled liposomes to deliver siRNA against a collagen-specific chaperone
Nature Biotechnology 26, 431 - 442 (2008)
Published online: 30 March 2008 | doi:10.1038/nbt1396
Yasushi Sato1,2, Kazuyuki Murase1,2, Junji Kato1,2, Masayoshi Kobune1, Tsutomu Sato1, Yutaka Kawano1, Rishu Takimoto1, Kouichi Takada1, Koji Miyanishi1, Takuya Matsunaga1, Tetsuji Takayama1 & Yoshiro Niitsu1
1. Fourth Department of Internal Medicine, Sapporo Medical University, School of Medicine, Sapporo, 060-8543, Japan.
2. These authors contributed equally to this work.
Correspondence to: Yoshiro Niitsu1 e-mail: niitsu@sapmed.ac.jp
our modality actually reverses liver cirrhosis both histologically and functionally. This underscores its promise for clinical translation to treat liver cirrhosis.
Abstract
There are currently no approved antifibrotic therapies for liver cirrhosis. We used vitamin A-coupled liposomes to deliver small interfering RNA (siRNA) against gp46, the rat homolog of human heat shock protein 47, to hepatic stellate cells. Our approach exploits the key roles of these cells in both fibrogenesis as well as uptake and storage of vitamin A. Five treatments with the siRNA-bearing vitamin A-coupled liposomes almost completely resolved liver fibrosis and prolonged survival in rats with otherwise lethal dimethylnitrosamine-induced liver cirrhosis in a dose- and duration-dependent manner. Rescue was not related to off-target effects or associated with recruitment of innate immunity. Receptor-specific siRNA delivery was similarly effective in suppressing collagen secretion and treating fibrosis induced by CCl4 or bile duct ligation. The efficacy of the approach using both acute and chronic models of liver fibrosis suggests its therapeutic potential for reversing human liver cirrhosis.
Liver cirrhosis, or fibrosis, the ultimate pathological feature of all forms of chronic hepatic damage, is responsible for much morbidity and mortality worldwide. The principal cell type responsible for liver fibrosis is the hepatic stellate (HS) cell, a resident perisinusoidal cell that takes up vitamin A from circulation and stores it. When stimulated by reactive oxygen intermediates or cytokines, HS cells become activated and are transformed to proliferative, fibrogenic and contractile myofibroblasts1, which synthesize and secrete procollagen, which accumulates as insoluble collagen after its terminal domains are cleaved by procollagen peptides, causing fibrosis. The collagen-specific chaperone, heat shock protein 47 (HSP47), facilitates collagen secretion by ensuring proper triple-helix formation of procollagen in the endoplasmic reticulum and has also been implicated in translational regulation of procollagen synthesis2, 3.
The demonstration that liver fibrosis in animals4 and humans5 can regress when collagen synthesis is inhibited suggests that fibrosis can be reversed, most likely by the activity of matrix metalloproteinases. Various therapeutic approaches to inhibit collagen synthesis or activate matrix metalloproteinases have been investigated in animal models6, 7. However, none have yet been applied clinically, mainly because of side effects resulting from an inability to specifically target particular molecules and/or cells.
The specific association of HSP47 with a diverse range of collagen types3, 8, 9 makes it an excellent candidate for targeting HS cell-mediated collagen secretion using siRNA. To enhance the specificity of such a strategy, we reasoned that encapsulating the siRNA in vitamin A-coupled liposomes should target it preferentially to HS cells, which have a remarkable capacity for vitamin A uptake, most likely through receptors for retinol binding protein (RBP).
Using three animal models of liver cirrhosis-involving induction by dimethylnitrosamine (DMN), CCl4 or bile duct ligation-our histological analysis shows that intravenous (i.v.) injection of vitamin A-coupled liposomes carrying siRNA against mRNA encoding rat gp46, a homolog of HSP47, (VA-lip-siRNAgp46), rapidly resolves liver fibrosis. As cells analogous to HS cells apparently play an essential role in causing fibrosis associated with chronic pancreatitis10 and laryngeal fibrosis11, our approach may find value in treating fibrotic conditions in organs besides the liver.
Discussion
The inadequate target specificity of most approaches to treat liver fibrosis has limited their suitability for clinical use17. Our two-pronged strategy to ensure specificity involves, first, targeting a collagen-specific chaperone molecule (gp46) with siRNA and, second, delivery of siRNA specifically to collagen-producing liver cells, using vitamin A-coupled liposomes. This enabled us to resolve hepatic collagen deposition in rat models involving induction of cirrhosis by either DMN or CCl4 treatments or bile duct ligation. Survival of DMN-treated rats was prolonged in a dose-and duration-dependent manner, indicating a biologically specific effect of siRNAgp46 treatment.
Off-target effects18 and immune responses, such as induction of interferon (IFN-) driven by the interaction with Toll-like receptor (TLR)3 or TLR 7/8 (ref. 15), are two issues often associated with the use of siRNA that can lead to misinterpretation of siRNA experiments. We used three independent siRNAs against the same target (gp46) mRNA and found comparable gene silencing efficacy and antifibrotic effects in vivo (Fig. 1a and Supplementary Fig. 7), suggesting that the phenotype observed with downregulation of gp46 was indeed related to gp46 knockdown and not a bystander effect of the siRNA sequence. To circumvent immune responses such as induction of IFN-, we used siRNA with a 2-nucleotide 3' overhang, shown to impair activation of the transcription factor IRF3 (ref. 19). In addition, our siRNA did not contain the 5' triphosphate of the T7-transcript (manufacturer's information), which reportedly plays a role in IFN induction20. In fact, there was no elevation of either IFN- mRNA in the liver or TNF- and IL-12 in the circulation of rats treated with VA-lip-siRNAgp46. However, as the modifications of siRNA described above may not prevent the triggering of all immune responses, it is possible that the low immune activation in the present investigation was relevant to the RBP receptor-mediated uptake of siRNA, in addition to the modifications of siRNA structure.
Although specificity is important, another critical factor in the use of siRNAs as therapeutic agents is the efficacy of suppression of the target molecule. As synthetic RNA duplexes 25-30 nucleotides in length are up to 100-fold more potent than corresponding conventional 21-mer siRNA21, we used a 27-nucleotide RNA duplex with 2-nucleotide 3' overhangs.
Another factor influencing the efficacy of siRNA treatment is the duration of gene silencing, which is mainly governed by dilution resulting from cell division. In rapidly dividing tumor cells, recovery of protein levels suppressed by siRNAs to pretreatment levels occurs within <1 week, whereas in slow-dividing fibroblasts or in nondividing hepatocytes, it takes >3 weeks22. We confirmed that levels of gp46 in HS cells treated with siRNA remained suppressed for at least 72 h in vitro and in vivo and used 48-72 h as the siRNA injection interval to demonstrate in vivo efficacy.
The in vivo siRNA dose used in our study (0.75 mg/kg per single injection) was less than doses previously shown to have in vivo therapeutic effects23, 24, 25. This may be related to the use of vitamin A-coupled liposomes to preferentially deliver siRNAgp46 to HS cells. It has long been debated whether transfer of retinol from plasma RBP/retinol complex to cellular RBP in HS cells is receptor driven or proceeds by passive diffusion26. Findings from most recent studies are consistent with receptor-driven uptake of retinol by HS cells27, 28, 29, 30, although coexistence of direct positioning into the hydrophobic part of cell membranes and receptor-mediated uptake has not been entirely ruled out31.
Consistent with the receptor-based theory, specific uptake of vitamin A-coupled liposomes by HS cells after binding to plasma RBP was shown in in vitro studies assessing RBP concentration-dependence, the effect of RBP antibody (Fig. 1e,f), and subcellular localization (Fig. 1h). Notably, siRNA-FAM fluorescence in the cirrhotic liver of rats injected intravenously with vitamin A-coupled liposomes was mainly observed in areas with HS cells (identified by -SMA staining) but not in parenchymal areas. Along with the finding that the area where FAM-florescence and HS cells were merged was markedly greater than in rats treated with vitamin A-free liposomes, this is also compatible with the notion that vitamin A-coupled liposomes are specifically taken up into HS cells by RBP receptors. The observation that liver fibrosis underwent regression and that survival time was significantly prolonged only in animals treated with vitamin A-coupled liposomes carrying siRNAgp46 further supports the vitamin A-receptor theory (Breslow-Gehan Wilcoxon test, P < 0.0001).
Incidentally, our results also confirm that activated HS cells, which become free of vitamin A deposition during the activation process1, take up vitamin A as effectively as resting HS cells, which store vitamin A32. Both an activated HS cell line (LI90) and primary rat liver HS cells activated by in vitro cultivation33 expressed markedly enhanced FAM fluorescence when they were incubated with vitamin A-coupled liposomes instead of vitamin A-free liposomes (Fig. 1e,g).
VA-lip-siRNAgp46-FAM distribution was negligible in retina, presumably because the eye is an isolated system with a strong blood-retinal barrier34. It was slightly evident in spleen and liver, which showed uptake of VA-lip-siRNAgp46 in a vitamin A receptor-independent manner to the reticuloendothelial system (macrophages), although these cells are not primarily collagen-producing cells and therefore should not be affected by nonspecific uptake of siRNAgp46.
When organ distribution of radiolabeled vitamin A-coupled liposomes was examined, prominent uptake of radioactivity was seen only in cirrhotic livers. The radioactivity in other organs was essentially the same in both DMN-treated and normal rats, which indicated that the delivery of vitamin A-coupled liposomes was indeed specific to HS cells possessing RBP receptors, and that delivery to other tissues occurred in a nonspecific manner, probably through nonspecific engulfment by macrophages.
It is also noteworthy that our vitamin A-coupled liposome system for the delivery of siRNA appeared to be quite efficient compared with previously reported methods using conventional liposomes. Our vitamin A-coupled liposomes produced biological effects at a dose of 0.75 mg/kg liposome, whereas in previous reports23, 24, 35, 36, doses of 8-160 mg/kg were necessary. Furthermore, most other studies37 used approximately a 1:10 ratio of solution to total blood volume (that is, 200 l/2 ml blood in mice; 2,000 l/20 ml blood volume in rats) to dissolve liposomes containing siRNA for a single bolus injection. However, we used only 200 l per injection for 200-g rats (1:100 blood volume). This suggests that in our study hydrodynamic pressure did not contribute to in vivo transduction of siRNA, whereas in previous studies, siRNAs might have been forcibly transduced by hydrodynamic pressure, at least to some extent. This may explain why background fluorescence for FAM was weak and showed low nonspecific distribution of radioactive vitamin A-coupled liposomes to tissues other than the liver.
Various animal models of liver fibrosis have been explored4, 6, 16, 38, 39. We chose to focus primarily on the DMN model because it causes progressive and lethal fibrosis, which enabled us to demonstrate prolonged survival resolution of fibrosis. We confirmed these results using both the CCl4 model, which is nonlethal and has milder features than the DMN model, and bile duct ligation models, which induces chronic cirrhosis by continuous stimulation of bile regurgitation. The consistency of the findings across all three models indicates the applicability of our approach for various types of liver cirrhosis.
In these models, regression of fibrosis occurred after five injections of siRNAgp46, whereas the animals were still being exposed to DMN, CCl4 or BDL stimulation. The primary mechanism underlying the therapeutic effect was surmised to be the inhibition of collagen secretion by siRNAgp46 (Fig. 1c,d,i) and concomitant degradation of predeposited collagen by collagenase activity, which remained as high as in normal liver until fibrosis was almost completely resolved by the five treatments with VA-lip-siRNAgp46 (Supplementary Table 3). This finding was consistent with the notion that matrix metalloproteinases, once they are secreted from cells, form a persistent extracellular matrix-associated pool by binding to type 1 collagen40, 41. In addition, HS cell apoptosis, possibly resulting from loss of anchorage sites (collagen) is considered to be a secondary mechanism for the therapeutic effect (Supplementary Fig. 5). Incidentally, apparent suppression of procollagen I mRNA in siRNAgp46-treated liver (Fig. 4d), which should not be caused by siRNAgp46 per se, may also be ascribed to the apoptosis of collagen-producing HS cells.
The improvement of serum bilirubin and hyaluronate levels in all three cirrhosis models further substantiates resolution of fibrosis in the portal area. Serum albumin and ALT levels, however, were not significantly improved by VA-lip-siRNAgp46 treatment in DMN-treated rats. This may result from the toxic effect of ongoing DMN treatment, as DMN-treated rats at day 70 showed restoration of normal hepatic architecture (Fig. 4i) and nearly complete normalization of serum albumin and ALT levels in addition to serum bilirubin and hyaluronate (Supplementary Fig. 6). Therefore, our modality actually reverses liver cirrhosis both histologically and functionally. This underscores its promise for clinical translation to treat liver cirrhosis.
|
|
| |
| |
|
|
|
