Biodistribution of Small Interfering RNA at the Organ and Cellular Levels after Lipid Nanoparticle-mediated Delivery

Abstract

Chemically stabilized small interfering RNA (siRNA) can be delivered systemically by intravenous injection of lipid nanoparticles (LNPs) in rodents and primates. The biodistribution and kinetics of LNP–siRNA delivery in mice at organ and cellular resolution have been studied using immunofluorescence (IF) staining and quantitative polymerase chain reaction (qPCR). At 0.5 and 2 hr post tail vein injection of Cy5-labeled siRNA encapsulated in LNP, the organ rank-order of siRNA levels is liver > spleen > kidney, with only negligible accumulation in duodenum, lung, heart, and brain. Similar conclusions were drawn by using qPCR to measure tissue siRNA levels as a secondary end point. siRNA levels in these tissues decreased by more than 10-fold after 24 hr. Within the liver, LNPs delivered siRNA to hepatocytes, Kupffer cells, and sinusoids in a time-dependent manner, as revealed by IF staining and signal quantitation methods established using OPERA/Columbus software. siRNA first accumulated in liver sinusoids and trafficked to hepatocytes by 2 hr post dose, corresponding to the onset of target mRNA silencing. Fluorescence in situ hybridization methods were used to detect both strands of siRNA in fixed tissues. Collectively, the authors have implemented a platform to evaluate biodistribution of siRNA across cell types and across tissues in vivo, with the objective of elucidating the pharmacokinetic and pharmacodynamic relationship to guide optimization of delivery vehicles.

Small interfering RNA (siRNA) has become a novel therapeutic strategy since its discovery in 1993 in Caenorhabditis elegans (Fire et al. 1998; Lee et al. 1993). siRNA therapeutics have the potential to specifically and potently knock down the expression of disease-causing genes. Rapid screening of siRNAs also offers a robust approach for high-throughput preclinical and clinical target validation (Sepp-Lorenzino and Ruddy, 2008; Shim and Kwon, 2010). Naked unmodified siRNAs delivered intravenously in vivo are rapidly degraded by circulating RNase, but therapeutic siRNAs can be chemically stabilized to achieve optimal pharmacokinetics in vivo (Abrams et al. 2010; Morrissey, Blanchard, et al. 2005; Morrissey, Lockridge, et al. 2005). Systemic delivery of siRNA has been an area of extensive investigation in recent years (Kawakami 2008; Li and Shen, 2009; Peer and Shimaoka, 2009; Tseng et al. 2009; White, 2008). Different RNA delivery vehicles (RDVs) have been studied, including lipid nanoparticles (Abrams et al. 2010; Tao et al. 2010), polymers (Kim et al. 2009; Rozema et al. 2007), cell-degradable multilayered polyelectrolyte films (Dimitrova et al. 2008), nanocages (Yavuz et al. 2009), aptamer-based approaches (Dassie et al. 2009; McNamara et al. 2006; Thiel and Giangrande, 2009), peptide-mediated delivery (Jafari and Chen 2009), glucan encapsulated siRNA particles (Aouadi et al. 2009), and other non-viral (Chen and Huang, 2008) and viral vectors (Crowther et al. 2008; Guibinga et al. 2008; Manjunath et al. 2009). Human clinical trials using synthetic siRNA started in 2004 (Soutschek et al. 2004), and a list of ongoing clinical trials applying siRNA was summarized (Morin et al. 2009). The first siRNA clinical trial that used a targeted nanoparticle delivery system was reported (Davis et al. 2010) in which a human transferrin protein targeting ligand was displayed on the exterior of a polymer nanoparticle to engage transferrin receptors of patients with melanoma after systemic administration. Evidence of siRNA specific cleavage by rapid amplification of cDNA end was demonstrated in tumor biopsies from patents, confirming the siRNA mechanism of action (Davis et al. 2010).

Systemic delivery of siRNA often results in massive deposition of siRNA along with the delivery vehicles into the liver and spleen, which belongs to the reticular– endothelial system. Targeting delivery of siRNA to the corresponding organ and cell type at the right time, right dose, and siRNA stability in circulation is required to achieve optimal efficacy and reduce unwanted side effects and toxicity. Therefore, a complete understanding of siRNA biodistribution for novel experimental delivery vehicles is important for the success of RNA interference (RNAi)-based therapeutics. Several approaches have been reported to evaluate siRNA delivery efficiency and distribution in vivo after systemic delivery, including scintillation counting and gel electrophoresis when siRNA is labeled with ³²P on antisense oligo (Gao et al. 2009), fluorescence microscopy on tissue sections when siRNA passenger strand is labeled with fluorophores (Lu et al. 2009; Rozema et al. 2007), magnetic resonance imaging (MRI) when siRNA is delivered by metal-carrying nanoparticles (Ali et al. 2009), non-invasive whole-body imaging such as positron emission tomography (PET) and bioluminescent imaging (Bartlett et al. 2007), single photon emission computed tomography (SPECT) imaging (Merkel, Librizzi, Pfestroff, Schurrat, Behe, et al. 2009; Merkel, Librizzi, Pfestroff, Schurrat, Buyens, et al. 2009), and stem-loop quantitative PCR of siRNA by Taqman real-time PCR methodology (Abrams et al. 2010; Cheng et al. 2009, Seitzer et al. 2010). Assessment and quantification of the biodistribution of siRNA in liver tissues and parenchymal liver cells are the major goals of the research presented in this report.

The RDVs used in this report are lipid nanoparticles (LNPs) consisting of cholesterol (20–50 mol%), a PEGylated lipid (DMG-PEG_2K, 2–6 mol%), and a cationic lipid (CLinDMA, 30–50 mol%). One of them, LNP201, has been shown to be an efficacious siRNA delivery vehicle in rodents (Abrams et al. 2010). We tested the delivery of the siRNA against the ubiquitously expressed Ssb (Sjogren syndrome antigen B) mRNA. Ssb is a single nucleotide binding protein that is involved in RNA metabolism (Chan, Sullivan, Fox, et al. 1989; Chan, Sullivan, Tan 1989; Chambers and Keene, 1985). The Ssb gene is highly conserved in eukaryotes and is expressed in most tissues, making Ssb siRNA well suited to serve as a probe with which to study siRNA delivery and distribution in different tissues and cells.

In this report, we demonstrated the distribution of Cy5-labeled Ssb siRNA delivered by LNP201 in mouse at organ, suborgan, and cellular levels by histological methods and by stem-loop quantitative polymerase chain reaction (qPCR). We provided an in-depth description of some promising approaches we have developed to qualitatively and quantitatively study siRNA biodistribution on tissue sections, colocalization quantification, and siRNA in situ detection by fluorescence in situ hybridization (FISH). Our histological and imaging analytical tools provide insight into the relationship between pharmacokinetics and pharmacodynamics for experimental siRNA therapeutics.

Materials and Methods

Animals and LNP–siRNA Treatment

All animal experiments were conducted in accordance with the standards established by the United States Animal Welfare Act and approved by Merck & Co., Inc.’s Institutional Animal Care and Use Committee. Female Crl:CD-1/ICR mice (20–30 g) were obtained from Charles River (Wilmington, MA).

We performed LNP201 biodistribution study in CD-1 mice. The mice were randomized into five groups (three mice per group) according to body weight measurement to receive 0.2 ml IV LNP201–siSsb (Ssb siRNA)–Cy5 (3 mg/kg body weight) for 0.5, 2, 6, and 24 hr (Group 1, 2, 3, and 4, respectively). PBS was injected via tail vein at 0.2 ml, and mice were harvested at 2 hr post dosing (Group 5, vehicle control). Seven organs—liver, spleen, lung, kidney, heart, duodenum, and brain—were collected at necropsy. The four major lobes of liver—caudate, left, right, and medial lobes—were also collected. Some tissues were embedded in OCT (Tissue-Tek, Dublin, OH) on dry ice for histology examination; other tissues were saved in RNAlater (Qiagen, Valencia, CA) for RNA extraction.

Lipid Nanoparticle LNP201

LNP201 and Ssb–siRNA assemblies were described in detail by Abrams et al. (2010). The lipids included CLinDMA (2-{4-[(3β)-cholest-5-en-3-yloxy]butoxy}-N,N-dimethyl-3-[(9Z,12Z)-octadeca-9,12-dien-1-yloxy]propan-1-amine), cholesterol, and polyethylene glycol-dimyristoylglycerol (α-[8′-(1,2-dimyristoyl-3-propanoxy)-carboxamide-3′,6′-dioxaoctanyl]carbamoyl-ω-methyl-poly(ethylene glycol)₂₀₀₀) at a molar ratio of 50:44:6. The siRNA guide strand was as follows: 5′-ACAACAGACUUUAAUGUAA-3′. Cy5 was end-labeled at the 5′-end of the passenger strand before the mixed with the lipids.

siRNA and mRNA Assays

The concentrations of siRNA of Ssb and mRNA of Ssb were measured by stem-loop quantitative PCR and standard quantitative RT-PCR, respectively, using Applied Biosystems (Foster City, CA) reagents, using previously reported methods (Abrams et al. 2010; Seitzer et al. 2010).

Immunofluorescence Staining

Seven-µm-thick frozen tissue sections were cut using Cryostat (Leica CM3050S), mounted on Superfrost Plus slides (Fisher Scientific Co., Houston, TX), and stored in a –70C freezer. Before staining, slides were allowed to air dry at room temperature for 10 min, followed by 55C for another 10 min. Tissue sections were then fixed in freshly made 4% paraformaldehyde (from 16% paraformaldehyde, Electro Microscopy Sciences, Hatfield, PA) in 1× PBS for 10 min. After washing with 1× PBS (pH 7.5, 3 × 2 min), sections were incubated for 30 min at room temperature in a protein-blocking solution (Dako Corp., Carpentaria, CA) containing 10% serum from the species in which the secondary antibody was generated. Sections were then incubated with the primary antibodies at final concentration 10 µg/ml for 1 hr at room temperature. The antibodies, CD68, CD11b, and CD31, are from AbD SeroTec (Raleigh, NC); collagen IV from Millipore (Billerica, MA); Desmin from Epitomics (Burlingame, CA). After washing in PBS, sections were incubated with secondary antibody (goat anti-rat or rabbit IgG) conjugated with Alexa 488 (Invitrogen, Carlsbad, CA). Finally, sections were mounted with a DAPI containing aqueous mounting medium Prolong (Invitrogen). The slides were stored at 4C.

Fluorescence Signal Detection on Liver Lobe Slices Using Ariol

Liver slices approximately 3 mm thick were collected from each of the four lobes of liver per mouse. The slices of left, right, medial, and caudate lobes were embedded in OCT, one block per mouse. Seven-µm liver sections were made on Fisher Plus slides using Cryostat (Leica S3050) and stained with phalloidin and mounted in Prolong medium containing DAPI (Invitrogen). Slides were bar coded and scanned automatically using Ariol (Genetix, San Jose, CA), equipped with an Olympus BX61 microscope with 5× and 20× objectives, a 1.4 mega-pixel progressive scan camera (jAi CV-M4+CL), and Ariol software (version 3.2). Images were captured under three channels: DAPI for nuclei, FITC for phalloidin, and Cy5 for siRNA. The exposure time for Cy5 was found using tissue containing the highest Cy5 signal (0.5 hr post dosing). The Cy5 autofluorescence signal was defined using the PBS-treated tissue, and images were normalized using this threshold.

Fluorescence Quantification Using Opera/Columbus

Image fields of stained sections were captured using an OPERA confocal microscope (Perkin Elmer, Bridgeville, PA) equipped with a 40× air objective (0.60 numerical aperture), a digital charge-coupled device video camera (SensiCam QE, PCO Imaging, Romulus, MI), and OPERA CN/QEHS software (version 1.8.1). Image fields were captured in a 3 × 3 layout, total 9 fields for each tissue section. The optimal laser power and exposure time for Cy5 were assessed using the sections containing highest Cy5 signals (0.5 hr post dosing liver section), and the background signals in Cy5 channel were defined according to PBS-treated tissue sections. Any Cy5 fluorescence signals above PBS control were counted as true Cy5–siRNA signals. FITC channel represents the cell type–specific markers (collagen outlined sinusoids and hepatocellular region; CD68-stained Kupffer cells). DAPI counter stain in the Prolong mounting medium (Invitrogen) stains all nuclei. Images were imported into Columbus (Perkin Elmer) server for image analysis using the associated Columbus analysis building blocks. Columbus scripts for defining sinusoid mask, hepatocyte mask, and Kupffer cell mask, as well as quantification of Cy5 signals in these masks, were generated according to the manufacturer’s recommendations (see Supplemental Fig. SF1; supplementary material for this article is located at http://jhc.sagepub.com/supplemental). The mean intensities of Cy5–siRNA in hepatocytes vs. sinusoids and in Kupffer cell (KC) vs. non-KC area were calculated.

Fluorescence In Situ Hybridization (FISH)

Tissues on the slides were fixed in 4% paraformaldehyde and blocked in 3% H₂O₂/PBS and 10% goat serum sequentially. To detect KCs on mouse liver sections, rat anti-mouse CD68 antibody was applied to each section for 30 min, followed by secondary antibody, goat anti-rat IgG conjugated with Alexa 555 for 30 min. KCs were visualized under the red channel in Olympus BX51 microscope, equipped with 10× and 20× objectives, a digital charge-coupled device video camera (Hamamatsu, ORCA-ER, C4742-80-12AG Bridgewater, NJ), and SlideBook (Intelligent Imaging Innovations, Inc, Denver, CO) software (version 4.2). Following staining of cell-type specific marker, siRNA was detected in situ. The guide strands of Ssb siRNAs were hybridized with the digoxigenin-labeled (DIG) locked nucleic acid probes designed and synthesized by Exiqon (Woburn, MA).

The sequences of guide (anti-sense, AS) and passenger strand (sense) of Ssb probes, as well as their melting temperature (Tm), are listed below.

$$\begin{array}{l}\text{AS}-\text{Ssb}:5^{\prime }-\text{DIG}-\text{acaacagactttaatgtaa},\text{Tm}=59.7\text{C}\\ \text{Sense}-\text{Ssb}:5^{\prime }-\text{DIG}-\text{ttacattaaagtctgttgt},\text{Tm}=60.5\text{C}\end{array}$$

In situ hybridization on cryosection was conducted as follows. Tissues were first acetylated in 500 µl of 6N HCl, 670 µl of triethanolamine, 300 µL of acetic anhydride, and 48.5 ml of DEPC-water for 5 min. Acetylation of the positively charged amino groups in the proteins of the tissue can decrease the background binding of the negatively charged probe. Then the section was hybridized with the probe in ULTRAhyb-Oligo Hybridization Buffer containing 50% deionized formamide, 5× SSC (saline sodium citrate buffer), 500 µg/µL yeast tRNA, 1× Denhardt’s solution, DEPC-water, pH of 6–6.5 (Ambion, Foster City, CA) for 2 hr in a humidity chamber. Next, the slides were washed 3 times in 0.1× SSC for 10 min/each and rinsed in 2× SSC. For antibody detection, the sections were blocked for 30 min in Dako blocking reagent. The slides were then incubated for 30 min at room temperature in sheep anti-digoxigenin conjugated with FITC (1:200, Roche, Nutley, NJ). To amplify the FISH signal, we applied tyramide signal amplification (TSA) (TSA Kit #22, Invitrogen) following the manufacturer’s instruction. Sections were mounted in Vector Laboratory’s (Burlingame, CA) mounting media containing DAPI (Vectashield, Cat# HT1200), cover-slipped, and sealed with nail polish. For image acquisition, 3 × 3 montage images were captured under 20× objective using Olympus BX51 fluorescence microscope and analyzed using SlideBook software. Green fluorescence (FL) signals represent FISH detected siRNA, and red FL signals represent CD68-stained KCs. The qualitative yellow (green + red) signals were counted as siRNA colocalized with KCs.

Statistical Analysis

Statistical evaluation was performed by two-way ANOVA or one-way ANOVA followed by Dunnett’s multiple comparison tests or Tukey’s all pair comparison test (GraphPad Prism 5.03 software). Differences where the p value was 0.05 were considered statistically significant.

Results

Level of siRNA and Target Ssb mRNA Knockdown Detected in Different Tissues in CD-1 Mice

In a mouse time course biodistribution study, LNP201–Ssb–siRNA–Cy5 was injected via tail vein of CD1 mice at 3 mg/kg body weight (3 mpk). Seven organs (liver, spleen, lung, kidney, heart, duodenum, and brain) were collected at 0.5, 2, 6, and 24 hr post dosing. The total RNAs were extracted and the levels of siRNA were measured at these time points (Fig. 1A). The rank of siRNA levels in these organs at 0.5 and 2 hr is liver > spleen >> kidney> lung > heart. Low levels of siRNA were detected in duodenum and brain. The majority of siRNA was delivered to liver and spleen as early as 0.5 hr, with about 2.5-fold more in liver. The siRNA levels in liver and spleen decreased 10-fold by 24 h post dose compared with early time points (0.5 and 2 hr). The target gene Ssb mRNA levels were also evaluated by real-time RT-PCR. Fig. 1B illustrates the percentage of target gene Ssb mRNA knockdown (KD) in the seven organs at 24 hr, compared with the corresponding PBS-treated tissues. Significant KD of mRNA (85%) was observed in the liver, KD was observed to a lesser extent (about 25% KD) in the spleen, and no KD was observed in lung, kidney, heart, duodenum, or brain. Because Cy5 was labeled at the 5′-end of the passenger strand of Ssb siRNA, efficient knockdown of Ssb mRNA in this study demonstrated that the 5′-phosphate of passenger strand is not a requirement for the silencing effect of double stranded-siRNA (ds-siRNA).

Figure 1.

Small interfering RNA (siRNA) and mRNA detection in CD1 mice. (A, C) Stem-loop quantitative polymerase chain reaction (qPCR) was applied to detect siRNA levels in different organs of mouse (A) and different lobes of liver (C). siRNA levels measured from medial lobes of liver were plotted in (A). siRNA delivered mainly to liver and spleen, much less in kidney and lung, and in small amounts in brain and duodenum (A). The levels of siRNA decreased in the liver from 0.5 to 24 hr post dosing (C). No significant difference was observed among different lobes of liver at each time point (C). (B, D) Sjogren syndrome antigen B (Ssb) mRNA was measured in different organs (liver, spleen, lung, kidney, heart, duodenum, and brain) and in different lobes of rat liver (caudate, right lateral, left lateral, and medial lobes). (B) Nearly 85% knockdown of mRNA seen in liver and 30% in spleen at 24 hr. No statistical significant knockdown was observed in other organs. (D) Similar degree of mRNA knockdown observed across lobes of livers, except caudate lobe, which showed similar level of knockdown but had an original increase of Ssb mRNA post LNP–siRNA treatment at 0.5 hr. **p < 0.01, ****p < 0.0001, one-way ANOVA followed by Dunnett multiple comparison test.

Level of siRNA and Target Ssb mRNA Knockdown Detected in Different Lobes of Mouse Liver

In the mouse study mentioned above, we collected liver samples from four different lobes of mouse liver, namely caudate, left lateral, right lateral, and medial lobes. We collected 3-mm punches and stored them in RNAlater for RNA extraction, and we also collected slices of livers from the four lobes for histological analysis. Fig. 1C showed the siRNA levels in the four lobes by stem-loop RT-PCR at different time points post LNP201–siRNA treatment. Similar amounts of siRNA were detected in different lobes of liver at the corresponding time points. Fig. 1D demonstrates KD levels of target Ssb mRNA in the four lobes at the different time points. There is no KD of target mRNA at 0.5 hr post dose. Knockdown of target mRNA was initially observed at 2 hr and persisted through 24 hr. A slightly higher level of mRNA of Ssb was observed in caudate lobe compared with the other lobes but was not statistically significant. For mouse studies, we standardized our procedure to always collect tissue from the medial lobe for RNA extraction and further siRNA and mRNA evaluation (Seitzer et al. 2010).

Cy5 siRNA Detected in Mouse Liver and Spleen Tissue Sections

In our studies, Cy5 is covalently attached to the 5′ end of passenger strand of ds-siRNA. Cy5 signal observed on frozen tissue sections may represent the siRNA location, especially at early time points. Liver and spleen frozen sections were stained with phalloidin to visualize cell membranes of all cell types (Fig. 2). At 0.5 hr post LNP201–siSsb–Cy5 injection, Cy5–siRNA is primarily localized in the sinusoidal area, which is characterized as the area between the liver parenchymal cells (also called hepatocytes, which contain large DAPI-stained blue nuclei) (Fig. 2A and E). By 2 hr, the intense Cy5 signals in the sinusoids decreased, but diffuse Cy5 signals remained in the hepatocytes (Fig. 2B and F). At 24 hr post dosing, the overall Cy5 intensity decreased in liver sections (Fig. 2C and G), indicating clearance of Cy5–siRNA from liver, consistent with the stem-loop qPCR result (Fig. 1C). In mouse spleen, at 0.5 and 2 hr post dose, substantial Cy5 signals were found localized in the red pulp region (Fig. 2I and J). By 24 hr post dosing, the Cy5 intensity in spleen was significantly reduced (Fig. 2K). PBS-treated animal tissues served as controls (Fig. 2D, H, and L). In addition to being found in liver and spleen, Cy5-labeled siRNA was observed in kidney but not in lung, duodenum, heart, and brain (data not shown). These data correlated well to siRNA levels as determined by qPCR (Fig. 1A).

Figure 2.

Cy5–small interfering RNA (siRNA) biodistribution in mouse liver and in spleen. Sections were stained with phalloidin (green) and counterstained with DAPI (blue). (A–D) Cy5–siRNA distributed in the liver is time dependent. (A) 30 min, (B) 2 hr, (C) 24 hr, and (D) PBS control. (E–H) Overlaid images of Cy5 (purple) and phalloidin outlined liver cells (green) and nuclei (DAPI blue) for the time points corresponding to A–D. Abundant Cy5 signals detected in the sinusoids of liver; diffuse Cy5 siRNA signals detected in liver parenchyma starting at 30 min post dose, lasting at least 24 hr. (I–L) Cy5–siRNA detected in the red pulp of spleen at 30 min (I), 2 hr (J), and 24 hr (K); PBS control in (L). V for central veins or portal veins of the liver. W for white pulp and R for red pulp of the spleen.

Similar Cy5–siRNA Distribution in the Four Lobes of Mouse Liver

Liver slices 3-mm thick of each of caudate, left, right, and medial lobes were embedded in OCT, sectioned, and stained with phalloidin. The slides were scanned using Ariol and images were captured under three channels: DAPI for nuclei, FITC for phalloidin-stained cell membranes, and Cy5 for siRNA. After reviewing the entire slice of each lobe of each mouse, we concluded that the distributions of Cy5 signals at early time points (0.5 and 2 hr) were similar among all four lobes (Fig. 3). By 2 hr, Cy5 signals were less concentrated in sinusoids compared with 0.5 hr, as the siRNA appeared to have migrated from sinusoids to the hepatocytes. At zone I (portal triad) and III (central vein) and II (in between), the Cy5–siRNA signals showed similar pattern. Taken together, from stem-loop qPCR (Fig. 1C) and from image analysis here, we confirmed that LNP201 delivered siRNA is similarly and evenly distributed in the four lobes of CD1 mouse.

Figure 3.

Cy5–small interfering RNA (siRNA) biodistribution in different lobes of mouse livers: Four slices of liver from caudate, left, right, and medial lobes are illustrated in A, which is stained with phalloidin (green) and DAPI (blue). LNP201–siRNA–Cy5 was dosed for 0.5 hr (D, E) and 2 hr (A, B, C). Enlarged images from medial lobe were illustrated in B and C. At the same scale, images from a medial lobe of 0.5 hr dosing group were illustrated in D and E. Cy5–siRNA is purple in B and D. Overlaid images of Cy5, phalloidin (green), and DAPI nuclei (blue) showed in C and E. Cy5–siRNA signals relatively evenly distributed in the entire liver and no Cy5 signals inside the veins. Arrow heads indicate zone I, portal triad areas. Arrow indicates zone III, central vein area. Zone II is the area between I and III.

Cell-type Localization within the Liver

The four major compartments of the liver are hepatocytes, sinusoids (which contain endothelial linings), KCs, and stellate cells (also called Ito cells) (Malarkey et al. 2005; Ogawara et al. 2002). We first applied immunofluorescent staining of CD31 to outline sinusoidal endothelial cells and to separate sinusoidal and hepatocellular areas. We later found that collagen IV outlined sinusoids more precisely than did CD31 on mouse liver sections (Fig. 4B). Using collagen IV as a marker, we applied Columbus software to generate masks for sinusoidal and hepatocellular areas and analyzed the Cy5–siRNA intensity in these two compartments (Fig. 4J; also see Supplemental Fig. SF1). We also stained KCs and Ito cells with CD68 (Fig. 4E) and Desmin (Fig. 4H), respectively. In the overlaid images, the majority of Cy5–siRNA signals colocalized with collagen stained sinusoidal lining (arrows in Fig. 4C). Additional punctate and diffuse Cy5–siRNA signals were located inside hepatocytes (cells with large DAPI-stained nuclei). By 2 hr, much fewer Cy5 signals localized in the sinusoids (Fig. 4A) compared with 0.5 hr (Supplemental Figure SF1D), again suggesting a shift in biodistribution of siRNA from vasculature sinusoids toward hepatocytes in this time frame. Fig. 4F also showed some intense Cy5 signals colocalized with CD68-positive KCs (arrows). In Fig. 4I, the arrows showed that Cy5–siRNA has presented in the cell body of Ito cells, but siRNA (red) was not much colocalized with Ito cell protrusions (green).

Figure 4.

Identification of cell markers of mouse liver (A-I). Quantification of Cy5–small interfering RNA (siRNA) signals in the sinusoids and hepatocytes (J) and in Kupffer cells (KCs) and non-KC areas (K). Images were taken from liver sections 2 hr post lipid nanoparticle 201 (LNP201)–siRNA–Cy5 3 mpk dose. (A, D, G) Cy5–siRNA signals. (B) Collagen IV outlined sinusoids. (E) CD68-stained Kupffer cells. (H) Desmin-stained Ito cells and their protrusions. (C, F, I) Overlay images for Cy5–siRNA (red), cell markers (green), and nuclei (DAPI, blue). (J) Columbus quantification of Cy5–siRNA intensity in hepatocytes and sinusoids from collagen-stained liver sections. (K) Columbus quantification of Cy5–siRNA intensity in KCs and non-KC areas from CD68-stained liver sections. Two-way ANOVA was performed for J and K, where p < 0.001 for time and location.

To quantify Cy5–siRNA signals in the different compartments of the liver, we generated Columbus software scripts using collagen IV–stained liver sections for hepatocytes and sinusoids, using CD68-stained sections for KC and non-KC areas. The mean intensity of Cy5 signals was measured in hepatocytes and sinusoids (Fig. 4J; also Supplemental Figure SF1) and KCs and non-KC areas (Fig. 4K) at different time points post LNP201–siRNA–Cy5 dosing. The dashed lines indicated the background Cy5 signals on liver sections. Overall, we delivered Cy5–siRNA to hepatocytes and sinusoidal areas including KCs, with Cy5–siRNA intensity slightly increased and retained in hepatocytes from 0.5 to 2 hr and decreased in sinusoids in this time frame.

Taken together, we identified cell markers to separate hepatocytes, sinusoidal endothelial cells, KCs, and Ito cells and established methods to analyze Cy5–siRNA signals in the major compartments of liver both qualitatively and quantitatively. This platform could benefit the evaluation of specific liver cell targeting delivery in the future.

Fluorescence In Situ Hybridization to Detect Guide Strand of siRNA

From the above approaches described, we detected siRNA biodistribution using Cy5-labeled passenger strand of siRNA on frozen tissue sections. The key challenge to this approach is the accuracy of Cy5 signal localization to represent the ds-siRNA, especially the functional guide strand siRNA. The Cy5 fluorophore has to be labeled on the 5′-end of the passenger strand in order to leave the 5′-end of guide strand intact for incorporation into RNA interfering silencing complex (RISC). It has been suggested that the passenger strands are cleaved when the guide strand loads into RISC and are degraded faster than the guide strand in cells (Leuschner et al. 2006; Matranga et al. 2005). To circumvent this problem, we developed a FISH method to detect and localize the guide strand of siRNA. Furthermore, we optimized the conditions in order to combine FISH with cell-type specific immunofluorescence staining to study the colocalization of siRNA guide strand in specific cell types, for example, CD68-stained KCs (Fig. 5). We dosed LNP201–siSSB (unlabeled) in CD1 mice via tail vein at two doses, 3 mpk and 9 mpk. One hr later, we collected liver tissues for FISH analysis. PBS-treated mice served as the control. Guide strand of SSB siRNA was detected and amplified using the TSA system (Pena et al. 2009; Silahtaroglu 2007) (Fig. 5A, D, G); KCs were costained on the same sections after in situ hybridization (Fig. 5B, E, H), and the overlaid images (Fig. 4C, F, I) showed the colocalization of siRNA (green FISH signals) with KCs (red) in most cases. Occasionally there were KCs with no FISH signals (arrows in Fig. 5C). We qualitatively observed dose-dependent FISH signals (Fig. 5A and D, corresponding to dose 9 mpk vs. 3 mpk) but we do not recommend using FISH as a quantitative measure of siRNA (see discussion).

Figure 5.

Fluorescence in situ hybridization (FISH) to detect small interfering RNA (siRNA) distribution on frozen mouse liver sections. Mice were dosed with lipid nanoparticle 201–Ssb siRNA without any labels for 2 hr at 3 and 9 mpk. The frozen liver sections were stained by FISH (green signals in A, D, C, and F). The same tissue sections were also stained with anti-CD68 (red) for Kupffer cells (B, E, and H). Mouse dosed with PBS served as control where no FISH signals were detected (G and I). FTIC (Fluorescein isothiocyanate) green channel represents tyramide amplificed FISH signals.

Discussion

We established several methods to study the biodistribution of siRNA delivered by Merck’s proprietary lipid nanoparticles, including 1) siRNA analyzed by Cy5 labeling at the 5′-end of passenger strand; 2) siRNA analyzed by FISH; and 3) colocalization analysis of siRNA with cell-type specific markers (sinusoids, hepatocytes, KCs, and Ito cells). All these approaches could be used to detect siRNA carried by other RDVs. The major contributions of this report are 1) identification of the best cell-type specific markers on frozen mouse liver tissues; 2) quantification of Cy5-labeled siRNA on tissue sections and colocalization analysis of Cy5–siRNA inside hepatocytes, KCs, and sinusoids; 3) evaluation of siRNA distribution and target gene knockdown across all lobes of the mouse liver; and 4) detection of exogenous siRNA by FISH and combining FISH with cell-type specific marker immunofluorescent staining. In the follow-up experiments (data not shown), we examined the siRNA biodistribution using the above tools and correlated the distribution of different RDVs with their efficacy in mouse, rat, and rhesus models. Our platform proved useful in gaining insights into biodistribution, pharmacokinetic, and pharmacodynamic relationship for new RDVs.

About 80% of liver cells are hepatocytes, which constitute the parenchyma of the liver tissues, and the others are non-parenchymal cells, including KCs, endothelial cells, neutrophils, extrathymic T cells localized in sinusoids, and fat-storing pit (Ito cells, also called stellate cells) localized in the space of Disse (area between sinusoids and hepatocytes) (Baratta et al. 2009; Ogawara et al. 2002). Identification and selection of the best markers to differentiate liver parenchymal and non-parenchymal regions are crucial for studying the LNP–siRNA biodistribution because the hepatocytes are often the cell type of interest for therapeutic application. siRNAs delivered in sinusoids or engulfed by KCs are generally inactive for knockdown (unpublished data). The blood flow after tail vein injection of LNP201–siRNA–Cy5 is from portal vein to sinusoids and to the central veins. The LNP particles delivered by this method could potentially circulate from sinusoids to space of Disse via the fenestrae with size of about 150 nm in mice (Snoeys et al. 2007) on liver sinusoidal endothelial lining, where the LNPs come into direct contact with hepatocytes and may further diffuse or be actively taken up by hepatocytes. We first tested a series of hepatocyte markers (data not shown), including cytokeratin (Galarneau et al. 2007), albumin, α-trypsin (Lian et al. 2006), occludin and another cytoplasmic junction protein ZO-1 (Fallon et al. 1995), α-fetoprotein (Chu et al. 2002; Lian et al. 2006), organic anion transporting polypeptide 4 (Cattori et al. 2001), annexin A3 (Harashima et al. 2008), annexin I and II (Masaki et al. 1994), and C-reactive protein (Hirschfield, 2003). Cytokeratin was found to be the best for identification of non-human primate hepatocytes (Ji et al. 2010) and is illustrated in Supplemental Figure SF2, and Cell Signaling Technology’s E-cadherin (Sekine et al. 2009) is good for rat hepatic cords. We could not find a reliable marker for mouse hepatocytes. Instead, we tested CD31 and collagen IV, both of which can outline the sinusoids and hepatic veins in mouse liver. CD31 stains the liver sinusoidal endothelial cells, which showed blurred signals. Collagen demonstrates stronger and sharper signals than CD31 and thereafter was selected as the standard marker to outline sinusoids and to visualize parenchymal regions.

We also stained liver sections with phalloidin, which is a small molecule toxin that binds actin, preventing its depolymerization and poisoning the cell (Cooper, 1987). Phalloidin has been used to outline hepatocytes on frozen tissues (Rozema et al. 2007). Because phalloidin is not suitable for quantitative analysis and it non-specifically stains cell membranes of all types of cells in liver, we used it as a method for quick qualitative imaging assessment of the Cy5–siRNA location relative to hepatocytes. This was determined according to the characteristic feature of hepatocytes, which contain large single or double nuclei with approximately 7 µm in diameter.

KCs sit on sinusoidal lining and potentially interact with LNP–siRNA–Cy5 before the siRNA–Cy5 reaches the hepatocytes. KCs are the resident macrophage in the liver, and they are the first-line defense against invading foreign substrates in the liver. KCs and endothelial cells in the liver, along with other macrophage/monocytes and endothelia in the spleen, form the reticular–endothelial system, which plays an important role in the first stage of the systemically delivered exogenous substances, either pathogens or therapeutic particles. Any materials that can be phagocytosed by macrophages may serve as a measurement for KC phagocytotic activity. We tested the KC activity in vivo using India ink and microbeads (Supplemental Figures SF3 and SF4). Helmy et al. (2006) identified a receptor uniquely present in KCs, the complement receptor of the immunoglobulin family (CRIg), and demonstrated colocalization of CRIg and CD68, which has been widely used as KC marker (Holness and Simmons, 1993; Lapis et al. 1995; Ramprasad et al. 1995). Other KC markers have been reported, including F4/80 (Dambach et al. 2002; Holub et al. 2009), CD163 (Hiraoka et al. 2005), CD115 (Holub et al. 2009), and CD11b (Kinoshita et al. 2010). We tested all these KC markers in mouse, rat, and monkey liver tissues and found that CD68 is representative and consistently expressed in the livers of all three species.

Nanoparticles after systemic injection are vulnerable for opsonization by blood opsonins (Gaucher et al. 2009; Nagayama et al. 2007; Owens and Peppas, 2006), such as fibrinogen, fibronectin, complement C3b, and other serum IgGs. Opsonization by activated complement C3b facilitates the macrophage phagocytotic activity by type 3 complement receptor localized on the surface of KCs (Baldwin et al. 2005; Shan et al. 2009). Smith et al. (2008) addressed a quantitative assay to measure the clearance of adenovirus (Ad) vectors by KCs, and they detected a linear dose response of accumulation of Ad by KCs over at least two orders of magnitude of vector doses. Studies to determine the uptake of LNP by KCs might provide some evidence to address LNP–siRNA clearance in liver tissues, and saturation of KCs could indirectly explain the efficacy of LNP–siRNA reciprocally; that is, more KC uptake and saturation of KCs could potentially leave more LNP–siRNA delivered to hepatocytes. Evidence of saturation or depletion of KCs to facilitate target gene delivery has been reported by several groups (Luster et al. 1994; Schiedner et al. 2003) and also in our dose response study where we dosed LNP201 at 0.3, 1, 3, and 9 mg/kg (Supplemental Figure SF5). A potential increase of siRNA delivery into hepatocytes was observed in which increasing Cy5–siRNA localized in hepatocytes (50%, 74%, and 83% at 1, 3, and 9 mg/kg). Interestingly, at 9 mg/kg, about 16.7% of Cy5–siRNA signals were inside KCs, which reflects 18% overall population of KCs in liver (Baratta et al. 2009).

We detected Cy5–siRNA punctate and diffuse signals inside hepatocytes where the target gene knockdown was detected after LNP201–siSsb treatment. We observed colocalization of Cy5–siRNA signals with endothelial lining in sinusoids at an early time point (0.5 hr post dose, Supplemental Figure SF1) but little colocalization at a later time point (2 hr) (Fig. 4A). The translocation of Cy5–siRNA signals from sinusoids (0.5 hr) to hepatocytes (2 hr) indicated the time-dependent siRNA distribution in mouse liver. Most importantly, we detected a significant amount of siRNA delivered into KCs, but no target Ssb mRNA knockdown was observed in isolated KC (unpublished data). The mechanism is still unclear, but it might relate to the quick transition from endosome to lysosome leading to clearance of the siRNA in KCs before siRNAs are able to escape from endosome and be released to the cytosol for RISC loading and silencing effect. This hypothesis could be further investigated using leupeptin, an inhibitor of lysosomal proteases, or treatment of the cells with chloroquine, a weak base that inhibits proteolysis by raising the pH in endosomes and lysosomes (Smit et al. 1987).

Silahtaroglu et al. (2007) developed a state-of-the-art technology for micro-RNA (miRNA) in situ hybridization (ISH) using frozen tissue sections. However, their detection could not distinguish primary, precursor, and mature miRNAs. Tuschl’s group (Pena et al. 2009) successfully detected miRNA on formaldehyde-fixed tissue sections, and they also preserved the tissues in 1-ethyl-3(3-dimethly-aminopropyl) carbodiimide to avoid small RNA loss during the long hybridization procedure. However, neither method could combine ISH with immunostaining of cell-type specific markers. Our FISH protocol is the first to demonstrate the ability to detect siRNA in situ and to further combine FISH and immunofluorescence staining of cell-type specific marker to qualitatively demonstrate siRNA location relative to certain cell types. This application has been tested in mouse, rat, and monkey liver tissues (images for the later two species not shown).

To our knowledge, we are the first to apply FISH technology to detect exogenous delivery of siRNA on tissue sections. This approach provides a new platform to evaluate targeted delivery of therapeutic siRNA to different cells on tissue sections. However, we only present our approach for qualitative observation, rather than quantitation. In FISH protocol, the last several steps are TSA of positive signals. Without TSA, the FISH signals are too weak to be detected when combined with immunofluorescence of cell-type specific markers (Silahtaroglu et al, 2007). Tyramide signal amplification is an enzyme-mediated detection method that uses the catalytic activity of horseradish peroxidase (HRP) to generate high-density labeling of a target protein or nuclei acid sequence in situ (TSA amplification kit from Roche). The main probe to detect the siRNA was 5′-end labeled with DIG, and after hybridization, the secondary antibody, sheep anti-DIG-FITC, was applied to detect the hybridization signal. These green signals are usually very weak and hard to detect. To strengthen the siRNA signal, we incubated the section with goat anti-FITC–HRP, followed by enzyme (HRP) substrate (tyramide) reaction, and this substrate tyramide conjugated with Alexa 488 resulted in an exceptionally bright, photo-stable, green fluorescent signal that is easily detected. Multiple deposition of green substrate onto one site of hybridization leads to amplification of the FISH signals. This enzyme–substrate reaction causes unregulated deposition of green fluorescent signals, which makes the quantitation of the FISH signals inaccurate (Liu et al. 2006). Rather we used the method to qualitatively observe the location of FISH signals at organ, suborgan, and cellular levels. We detected similar sinusoidal and hepatocellular FISH signals at early time points as observed using Cy5–labeled siRNA, which proved the use of end-labeled passenger strand of siRNA for biodistribution study. Another reason for not quantifying the FISH signal relates to hybridization itself. For each siRNA, we can design different probes to base pair with the guide strand of siRNA. Different AT and CG content renders different melting Tm for probe hybridization and therefore different FISH signal intensities. So when testing two different probes for one siRNA or for two different siRNAs, the absolute signal intensities do not directly correlate with the abundance of siRNAs, likely because of the differences in kinetics of probe hybridization (Pena et al. 2009). If we only compare the same siRNA administrated at different doses, for a given probe, the FISH intensities might relatively reflect siRNA differential accumulation on tissue sections when the FISH procedure is run at the same time by the same personnel and without the TSA amplification step. Nevertheless, the FISH protocol we developed provides a clear, robust, and qualitative method to visualize the distribution of guide stand siRNA on tissue sections.

In summary, we have set up a platform to evaluate biodistribution of siRNA across cell types and across tissues in vivo. Our approach will lead to further understanding of siRNA therapeutics targeting and delivery in the preclinical and clinical studies in the future.