Phagocytic macrophages and dendritic cells are desirable targets for potential RNAi (RNA interference) therapeutics because they often mediate pathogenic inflammation and autoimmune responses. We recently engineered a complex 5 component glucan-based encapsulation system for siRNA (small interfering RNA) delivery to phagocytes. In experiments designed to simplify this original formulation, we discovered that the amphipathic peptide Endo-Porter forms stable nanocomplexes with siRNA that can mediate potent gene silencing in multiple cell types. In order to restrict such gene silencing to phagocytes, a method was developed to entrap siRNA–Endo-Porter complexes in glucan shells of 2–4 μm diameter in the absence of other components. The resulting glucan particles containing fluorescently labelled siRNA were readily internalized by macrophages, but not other cell types, and released the labelled siRNA into the macrophage cytoplasm. Intraperitoneal administration of such glucan particles containing siRNA–Endo-Porter complexes to mice caused gene silencing specifically in macrophages that internalized the particles. These results from the present study indicate that specific targeting to phagocytes is mediated by the glucan, whereas Endo-Porter peptide serves both to anchor siRNA within glucan particles and to catalyse escape of siRNA from phagosomes. Thus we have developed a simplified siRNA delivery system that effectively and specifically targets phagocytes in culture or in intact mice.
Sequence-specific gene silencing by RNAi (RNA interference) is currently being explored as a novel therapeutic strategy for a variety of diseases [1–5]. However, in contrast with the robust efficacy and potency of RNAi when applied to cultured cells, silencing target genes in specific tissues of intact animals has been a challenging task. Multiple hurdles impede delivery of large macromolecules of siRNA (small interfering RNA) to the cellular machinery that catalyses mRNA degradation, including non-specific accumulation in tissues, poor cellular uptake and inefficient release of siRNA from intracellular membranes into the cytoplasm . Nonetheless, a number of approaches have shown promise in achieving delivery of functional siRNA to specific tissues in vivo. These include technologies combining siRNA with carriers such as liposomes [7,8], cationic polymers [9–11], CPPs (cell-penetrating peptides)  and covalently linked lipophilic moieties  with antibodies  or other targeting components, such as ligands for specific receptors [15,16] and aptamers . Studies have also reported cancer-targeted siRNA delivery using nanoparticles that specifically bind to cancer-associated molecules and receptors [18,19]. Also, poly(ethylene glycol)ylated PEI (polyethyleneimine) conjugated with RGD peptides was developed to selectively deliver VEGF (vascular endothelial growth factor) siRNA to tumours [16,20] and siRNA-carrying liposomes decorated with transferrin receptor-specific antibody fragments silenced this gene in xenograft tumours in mice, significantly inhibiting tumour growth . Some of these delivery systems have progressed to the stage of being tested in human subjects within clinical trials. Indeed, in the past decade, a total of 14 RNAi therapeutic programmes have entered clinical trials for diseases such as age-related macular degeneration, respiratory syncytial virus, chronic myeloid leukaemia, pachyonychia congenital, hypercholesterolaemia, transthyretin amyloidosis, glaucoma, primary and secondary liver cancer, cancer, asthma, acute kidney injury, delayed graft function and diabetic macular oedema (http://www.clinicaltrials.gov) [22,23].
Phagocytic macrophages and dendritic cells represent potentially important targets for RNAi therapeutics on the basis of their role in mediating inflammation and immune responses . Moreover, they promote pathogenic responses in such diseases as rheumatoid arthritis, atherosclerosis and inflammatory bowel disease . Delivery of siRNA to such cells has been achieved by its encapsulation within 2–4 μm spherical, hollow, porous shells extracted from Saccharomyces cerevisiae (baker's yeast) [15,25]. These shells are composed mainly of β-1,3-D-glucan, a ligand of the dectin-1 receptor and other receptors that undergo phagocytosis by macrophages and dendritic cells [26,27]. Such encapsulation of anionic siRNA was achieved using a layer-by-layer method with the cationic trapping polymer PEI . These GeRPs (β-1,3-D-glucan-encapsulated siRNA particles), containing siRNA directed against TNF-α (tumour necrosis factor α) or the pro-inflammatory MAP4K4 (mitogen-activated protein kinase kinase kinase kinase 4), depleted the respective mRNA in macrophages and lowered serum TNF-α levels in mice [15,29]. The original GeRP formulation is composed of five different components: a tRNA core, two layers of PEI, the amphipathic peptide EP (Endo-Porter), the siRNA and the glucan shell. Although the PEI used in the original formulation showed low toxicity, its inclusion in the GeRPs severely limits their clinical applications [25,30]. In addition, the five-component GeRP formulation has been difficult to produce with uniformity and with optimal release of siRNA into cells, and tends to be structurally heterogeneous and unstable. These issues represent major disadvantages from the standpoint of reproducibility, manufacturing and delivery.
On the basis of the above considerations, the goal of the present studies was to test the hypothesis that robust siRNA-mediated gene silencing in macrophages could be achieved with GeRPs composed of fewer components and therefore more reliably produced. We discovered that the peptide EP, but not the cationic PEI, was required for silencing of target gene expression in macrophages. EP is designed to be an amphipathic α-helical peptide, with one face composed predominantly of aliphatic lipophilic amino acids, and the other face composed predominantly of basic amino acids, 70% of which are histidine residues . It has been shown that amphipathic peptides electrostatically interact with the phosphate backbone of siRNA, mediate the interaction with the cell membrane and facilitate endosomal escape . We found that EP binds to siRNA to form complexes of various sizes that can silence gene expression in many cell types, including macrophages and hard-to-transfect cells, such as adipocytes . We show in the present study that phagocyte-specificity of gene silencing can be conferred by physically entrapping these siRNA–EP complexes in glucan particles to form simplified GeRPs.
All procedures involving animals were approved by the Institutional Animal Care and Use Committee at the University of Massachusetts Medical School.
Preparation of glucan shells
Glucan shells were prepared as described previously . Briefly, β-1,3-D-glucan particles were prepared by suspending Sacharomyces cerevisiae (100 g of SAF-Mannan; SAF Agri) in 1 litre of 0.5 M NaOH and heating to 80 °C for 1 h. The insoluble material containing the yeast cell walls was collected by centrifugation at 5000 g for 10 min. This insoluble material was then suspended in 1 litre of 0.5 M NaOH, and incubated at 80 °C for 1 h. The insoluble residue was again collected by centrifugation (5000 g for 10 min) and washed three times with 1 litre of water, three times with 200 ml of propan-2-ol, and three times with 200 ml of acetone. The resulting slurry was placed in a glass tray and dried at room temperature (20 °C) to produce 16.2 g of a fine, slightly off-white powder.
Fluorescein labelling of glucan shells
Glucan shells (1 g) were washed with sodium carbonate buffer (0.1 M sodium carbonate, pH 9.2) and resuspended in 100 ml of sodium carbonate buffer. 5-(4,6-Dichlorotriazinyl) aminofluorescein (Invitrogen; 1 mg/ml in ethanol) was added to the buffered glucan shell suspension (10% v/v) and mixed at room temperature in the dark overnight. Tris buffer (2 mM Tris/HCl, pH 6.8) was added, incubated for 15 min and glucan shells were washed with sterile pyrogen-free water until the supernatant was clear. The glucan shells were then flash-frozen and freeze-dried in the dark.
Peritoneal macrophage preparation
C57BL6/J male mice (10 weeks old) were i.p. (intraperitoneally) injected with 4% thioglycollate broth (Sigma–Aldrich). At 5 days following injection, mice were killed and the peritoneal cavity was washed with 5 ml of ice-cold PBS to isolate PECs (peritoneal exudate cells). Peritoneal fluid was filtered through a 70 μm diameter pore nylon mesh and centrifuged at 270 g for 10 min. The pellet was first treated with red blood cell lysis buffer (8.3 g of NH4Cl, 1.0 g of KHCO3 and 1.8 ml of 5% EDTA) and then plated in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) FBS (fetal bovine serum), 50 μg/ml streptomycin and 50 units/ml penicillin. At 24 h after isolation, PECs were treated with siRNA.
siRNA–EP complex formation analysis
Chemically synthesized siRNA [Dharmacon; scr (scrambled) or MAP4K4 sequences] duplexes were 5′-phosphorylated using T4 polynucleotide kinase (New England BioLabs) and [γ-32P]ATP according to the manufacturer's protocol. The reaction was quenched with 20 mM EDTA and free nucleotides were removed using a NucAway spin column (Ambion). RNA was extracted using TRIzol® (Invitrogen), ethanol precipitated, and re-suspended in water. A trace amount of the radiolabelled duplex (1%) was added to phosphorylated, non-radiolabelled siRNA (40 pmol) and complexed with EP (sequence is Gene Tools proprietary information ) at molar ratios spanning from 100 to 0.125. The complexes were made in either PBS, pH 7.4, or 30 mM sodium acetate, pH 4.8, at room temperature for 1 h. After complexing, siRNA–EP was sedimented by brief centrifugation (13400 g for 1 min). The supernatant (free siRNA) was asessed using scintillation counting for the percentage of complex formation. Free siRNA was calculated relative to the no EP siRNA control (siRNA alone, 100% free).
The size of siRNA–EP complexes formed at pH 7.4 at a ratio of 1:50 was determined by DLS (dynamic light scattering) at 25 °C using a Zetasizer Nano ZS (Malvern Instruments).
siRNA–EP in vitro treatment
For in vitro treatment, siRNA (Dharmacon; sequences listed in Supplementary Table S1 at http://www.BiochemJ.org/bj/436/bj4360351add.htm) was incubated with EP in PBS for 1 h and added to media. The concentrations of siRNA and ratio of siRNA/EP are mentioned in the text and Figure legends for different experiments.
The 3T3-L1 fibroblasts and COS-7 cells were acquired from A.T.C.C. and cultured in high-glucose DMEM supplemented with 10% FBS, 50 μg/ml streptomycin and 50 units/ml penicillin. All cells were cultured at 37 °C with 5% CO2 and transfected at 80% confluence. At 2 days after 3T3-L1 fibroblasts achieved confluence, differentiation to adipocytes was induced by incubating the cells for 3 days in medium containing 4 μg/ml insulin, 0.25 μM dexamethasone and 0.5 mM IBMX (3-isobutyl-1-methylxanthine). The cells were then maintained in high-glucose DMEM with 10% FBS and treated with siRNA–EP or GeRPs for 48 h, 5 days after inducing the differentiation.
Preparation of GeRPs
To load siRNA in glucan shells, 1–5 nmol of siRNA (Dharmacon) were incubated with 50 nmol of EP (Gene Tools) in 30 mM sodium acetate, pH 4.8, for 15 min at room temperature in a final volume of 20 μl. The siRNA–EP solution was added to 1 mg (≈109) of glucan shells and then vortex mixed and incubated for 1 h. Tris/EDTA buffer (10 mM Tris and 1 mM EDTA, pH 7.4) was added to the particles and incubated for 15 min at room temperature to adjust the pH. The siRNA-loaded GeRPs were then resuspended in PBS and sonicated [15 s at 18 W at room temperature using a Sonicator 3000 (Misonix)] to ensure homogeneity of the GeRP preparation. GeRPs were aliquoted into tubes for daily dosing and either flash-frozen in liquid nitrogen and stored at −20 °C, or kept at 4 °C. siRNA–EP complexes were found to be stable in GeRPs for at least 2 h at 37 °C and 3 days at 4 °C (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/436/bj4360351add.htm).
GeRPs were loaded with EP and siRNA as described above and 1×106 PECs were treated at a ratio of 10:1 GeRPs/cell for 48–72 h.
siRNA–glucan shells binding analysis
Glucan shells [1 mg (≈109)] were loaded with different concentrations of 32P-labelled siRNA and 10 nmol of EP as described above. Glucan shells loaded with EP and siRNA were sedimented by centrifugation at 16000 g for 1 min and the supernatant was assessed for siRNA content by scintillation counting.
C57BL6/J male mice (10 weeks old) were i.p. injected once a day for 5 days with 5 mg or 5×109 GeRPs/kg of body weight, containing 125 or 625 μg of siRNA/kg of body weight. On day 6, mice were killed and PECs were isolated and filtered through a 70 μm pore nylon mesh, treated with red blood cell lysis buffer and plated in plastic dishes for 2–3 h in medium (DMEM plus 10% FBS). The cells were washed with PBS to remove non-adherent cells and adherent cells were used for real-time PCR and microscopy.
5′-RACE (rapid amplification of cDNA ends) and nested PCR
mRNA was purified from PECs from mice treated with unloaded or GeRPs loaded with scr or MAP4K4 siRNA. The GeneRacer Kit (Invitrogen) was used to produce 5′-RACE products, followed by nested PCR. GeneRacer primers and gene-specific primers were used. Products were electrophoresed in an agarose gel and cloned into the pCR4.1 TOPO vector for sequencing.
Isolation of RNA and real-time PCR
RNA isolation was performed according to the TRIzol® Reagent protocol (Invitrogen). cDNA was synthesized from 0.5–1 μg of total RNA using the iScript cDNA Synthesis Kit (Bio-Rad) according to the manufacturer's instructions. For real-time PCR, synthesized cDNA forward and reverse primers along with the iQ SYBR Green Supermix were run on the MyIQ Realtime PCR System (Bio-Rad). Sequences of the primers used were designed with Primer Bank (see Supplementary Table S2 at http://www.BiochemJ.org/bj/436/bj4360351add.htm). The ribosomal mRNA 36B4 was used as an internal loading control, as its expression did not change over a 24 h period with the addition of LPS (lipopolysaccharide) or siRNA against the genes used in the present study.
Western blot analysis
For CD40 protein level detection, 48 h after siRNA treatment, cells were treated with 10 ng/ml of LPS (Sigma–Aldrich) for 24 h. Then cells were washed with ice-cold PBS and harvested on ice as described previously . Protein samples were separated on SDS/PAGE (10% gels) and transferred on to a nitrocellulose membrane. Membranes were then analysed by Western blot analysis using anti-CD40 (Santa Cruz Biotechnology) and anti-actin (Sigma–Aldrich) antibodies.
PECs treated with fluorescein–GeRPs were incubated for 20 min in blocking buffer (PBS plus 1% BSA) and stained for 30 min with conjugated antibodies against the surface marker F4/80 (BD Biosciences). Subsequently, cells were washed and analysed by flow cytometry in a FACSCalibur cytometer (BD Biosciences). FlowJo software (Treestar) was used to identify the FITC+ macrophage population and F4/80 staining. A total of 25000 events were recorded.
Mice treated with fluorescein–GeRPs loaded with scr or F4/80 siRNA and PECs were isolated and directly analysed without staining. Cells were run through a FACS Vantage (BD Biosciences) at the University of Massachusetts Medical School Flow Cytometry Core Lab. Cells (5×106) were passed through the flow cytometers at a rate of 3000–5000 cells/s. The sort region was based on the 80% of the highest FITC signal for FITC+ cells and 80% of the lowest FITC signal for FITC− cells. Both FITC+ and FITC− populations were collected, and RNA was harvested for real-time PCR.
PECs were treated for 2 h with 1 μg/ml LPS and TNF-α levels in the media were measured using mouse ELISA kits (Pierce) as recommended by the vendor.
ALT (alanine aminotransferase) and AST (aspartate aminotransferase) activity measurement
To test for liver toxicity, levels of ALT and AST activities in serum were measured using a commercial kit (Fisher Scientific) according to the manufacturer's instructions.
Fixed cells from in vitro and in vivo experiments were incubated with rat anti-mouse F4/80 primary antibody (AbD-Serotec) followed by goat anti-rat Alexa Fluor® 633 secondary antibody (Invitrogen). Cells were mounted in Prolong Gold anti-fade with DAPI (4′,6-diamidino-2-phenylindole) (Invitrogen). For microscopy on cells from in vivo experiments, cells were permeabilized with a blocking and permeabilization buffer containing 2% goat serum (Invitrogen), 1% BSA (Sigma), 0.1% Triton X-100 and 0.05% Tween 20 (American Bioanalytical), in 1× PBS (Invitrogen), filtered with a 0.22 μm filter. Cells were then stained with Mouse anti-(1,3)-β-glucan antibody (Biosupplies, Australia) followed by a secondary Alexa Fluor® 488-conjugated antibody (Invitrogen).
Cell images were obtained with a Solamere CSU10 Spinning Disk confocal system mounted on a Nikon TE2000-E2 inverted microscope. Images were taken with a multi-immersion 20× objective with a NA (numerical aperture)=0.75; Oil: W.D. (working distance)=0.35 mm, or a 100× Plan Apo VC objective NA=1.4, Oil:W.D.=0.13 mm.
For siRNA–EP-binding studies, 1 nmol of Dy547-labelled siRNA was incubated with 10 nmol of siRNA for 15 min in 30 mM acetic acid (pH 4.8), PBS (pH 7.4) or Tris/EDTA (pH 8.4). For siRNA–EP–glucan shells binding studies, 1×109 purified glucan shells were loaded with 10 nmol of EP and 1 nmol of Dy547-labelled siRNA, as described above. Images were obtained using a Zeiss Axiovert 200 inverted microscope equipped with a Zeiss AxioCam HR CCD (charge-coupled-device) camera with 1300×1030 pixels basic resolution and a Zeiss Plan NeoFluar 40×/0.50 Ph2 (DIC II) objective.
The statistical significance of the differences in the means of experimental groups was determined by t test, or one-way or two-way ANOVA analysis, and Bonferroni or Tukey post-tests using GraphPad Prism 5.0a Software. The data are presented as the means±S.E.M.
The amphipathic peptide EP is necessary for GeRP-mediated gene silencing in macrophages
In order to evaluate the necessity of having both cationic PEI and the histidine-rich EP in GeRPs to anchor the siRNA, we removed either the outer layer of PEI or the EP independently, and then assayed for gene silencing efficacy in primary macrophages in cell culture. Figure 1(A) shows a diagram of our original GeRP formulation, composed of five separate components arranged in six layers: a tRNA core, two layers of PEI, the peptide EP, the siRNA and the glucan shell. Consistent with previous findings, the original GeRP formulation loaded with siRNA against TNF-α could efficiently silence TNF-α at the mRNA (50%, as shown in the right histogram of Figure 1A) and protein levels (≈80%, as shown in the left histogram of Figure 1A) in primary macrophages, as measured by real-time PCR and ELISA respectively. Removing the outer layer of PEI did not affect this knockdown induced by GeRPs containing TNF-α siRNA, suggesting that the second layer of PEI was not necessary for silencing activity (Figure 1B). However, removing the peptide EP resulted in a complete loss of TNF-α knockdown (Figure 1C). These results demonstrated that EP is required for efficient GeRP-mediated gene silencing in vitro.
EP is required for GeRP-mediated gene silencing in macrophages
EP forms physical complexes with siRNA that can silence target genes in macrophages
Figure 2(A) shows a schematic representation of the α-helical peptide EP, with one lipophilic face and a weak-base face, as previously described . EP has been reported to bind oligonucleotides and mediate transfection of cells . These considerations, in combination with the data in Figure 1, suggested that association of EP peptide with siRNA might be a key feature of the gene silencing we observed in GeRPs. To analyse EP–siRNA interactions, we monitored siRNA–EP complex formation by incubating 32P-labelled siRNA and EP at different molar ratios and at different pH conditions. At pH 7.4, maximum complex formation is achieved at an siRNA/EP molar ratio of 1:50. A pH value under the pKa of histidine (6.5) increases the number of positive charges on histidine residues and the binding of negatively charged molecules such as siRNA. In concert with this concept, at pH 4.8 siRNA is fully associated with EP starting at a molar ratio of 1:5 siRNA/EP (Figure 2B). This suggests that siRNA binds to EP via electrostatic interactions, and that the positive charge of the histidine residues on the EP play an important role in forming siRNA–EP complexes.
EP and siRNA form electrostatically stable complexes that mediate gene silencing in macrophages
We also studied EP–siRNA complex formation by microscopy, combining 40 pmol of Dy547-siRNA and 1 nmol of EP (ratio siRNA/EP; 1:25). Figure 2(C) shows that Dy547-siRNA–EP complexes formed at pH 7.4 could be readily detected, whereas no complexes could be observed without EP. All images were obtained at the same brightness and contrast, but the dispersed siRNA in the absence of EP creates a haze that appears as a brighter background. In order to determine whether siRNA–EP complex formation was required for gene silencing in vitro, 40 pmol of either scr siRNA or TNF-α siRNA were mixed with EP at different ratios (1:6.25–1:200) and tested for their ability to silence TNF-α in primary macrophages in culture. Only ~30% TNF-α silencing was achieved with the ratios 1:6.25 and 1:12.5, which were previously found to be sufficient for 40% and 60% siRNA association with EP respectively. However, when 90–100% siRNA was bound to EP (ratios 1:25 and 1:200), we observed 70–90% silencing of TNF-α mRNA levels (Figure 2D). TNF-α protein levels were also strongly silenced at the molar ratio 1:50, as shown in Figure 2(E).
We also tested the relationship between siRNA–EP complex formation and gene silencing using siRNA directed against other target genes. Consistently, using 40 pmol of siRNA targeting the macrophage surface marker F4/80 and a molar ratio of 1:50 siRNA/EP, we observed a significant silencing of F4/80 mRNA and protein levels in macrophages (Supplementary Figures S2A and S2B at http://www.BiochemJ.org/bj/436/bj4360351add.htm). Similarly, we obtained a significant silencing of another surface marker, CD40, at both protein and mRNA levels. Upon mixing 160 pmol of siRNA with an increasing amount of EP, we observed a range of 10–80% of siRNA to be associated with the EP. We observed that the more CD40 siRNA was bound to EP, the better the siRNA-mediated gene silencing was obtained (Supplementary Figures S2C and S2D). Taken together, these results suggest that siRNA–EP complex formation is required for efficient gene silencing in macrophages when using these reagents.
To further establish that electrostatic interactions occur between EP and siRNA, we analysed EP–siRNA complex formation in the presence of PEI, a highly positively charged polymer. Microscopic analysis showed that adding PEI to the mixture reduced the formation of siRNA–EP complexes, confirming the electrostatic interaction between EP and siRNA (see Supplementary Figure S3A at http://www.BiochemJ.org/bj/436/bj4360351add.htm). Moreover, although siRNA–EP complexes could mediate silencing of multiple target genes, siRNA–EP knockdown efficacy was blocked in the presence of PEI, suggesting that siRNA–EP complex formation is a necessary step for efficient gene silencing in macrophages in vitro (see Supplementary Figures S3B–S3D). These results show that EP and siRNA bind through electrostatic interactions and that the resultant siRNA–EP complexes efficiently silence genes in vitro.
Glucan shells containing siRNA–EP complexes mediate gene silencing specifically in macrophages in vitro
Peptides such as EP containing lipophilic faces composed mostly of leucine residues and weak-base faces composed mostly of histidine residues are poorly soluble in aqueous solutions at neutral pH. On the other hand, acidic pH increases the positive charges on histidine residues and the repulsion of the EP molecules from each other. We studied EP–siRNA complex formation by microscopy after incubating 40 pmol of Dy547-siRNA with 1 nmol of EP under different pH conditions (ratio of 1:25 siRNA/EP). As expected, Dy547-siRNA–EP complexes formed aggregates as the pH was increased from 5.2 to 7.4, and then to 8.4 (Figure 3A). The mean particle size of siRNA–EP complexes formed at pH 7.4 was 473±14 nm with polydispersity of 0.29±0.04 as measured by DLS using a Malvern zetasizer (Figure 3B). However, these complexes assume a broad range of sizes from ~300 nm to ~1300 nm, consistent with microscopic analysis.
Strategy for loading glucan shells with siRNA–EP complexes
A key feature of the glucan shells is their ability to deliver siRNA specifically to phagocytes in a glucan receptor-dependent manner, making them a useful delivery system for macrophages and dendritic cells. Previous studies also showed that the hollow cavity of the glucan shells allows for efficient absorption and encapsulation of payload molecules . In order to take advantage of the cell-type selectivity of GeRPs, we developed methods whereby siRNA–EP complexes could be loaded into glucan shells. siRNA–EP complexes were formed at pH 4.8, to maximize electrostatic interactions between EP and siRNA as well as the solubility of EP, as shown in Figure 3(A). The siRNA–EP complexes were then loaded in a volume just sufficient to swell the glucan particles at pH 4.8. Next, the siRNA–EP complexes were increased in size following a pH change to pH 7.4, according to the concept that they may be trapped within the glucan shells as depicted in the representation in Figure 3(C). We performed microscopy for experiments in which Dy547-siRNA was employed and biochemical analysis in studies involving radiolabelled siRNA. Using the pH-based strategy described in Figure 3(B), Dy547-siRNA binding was found associated with the glucan shells only in presence of EP (right-hand panel, Figure 3D), whereas only free siRNA could be observed without EP (left-hand panel, Figure 3D). All images shown were acquired at the same brightness and contrast. Further analysis using 32P-labelled siRNA showed that EP was necessary for the binding of siRNA to the glucan shells (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/436/bj4360351add.htm).
The question whether EP–siRNA complexes are able to silence genes in non-phagocytic cells was next addressed. We transfected a number of different cell types, primary macrophages, 3T3-L1 fibroblasts, 3T3-L1 adipocytes and COS-7 cells, with a siRNA targeting a gene ubiquitously expressed in all of these cell types, MAP4K4 (Figure 4, left-hand panel). In these experiments, primary macrophages were treated with 10 pmol of siRNA mixed with 500 nmol of EP (siRNA/EP; 1:50). We observed a 50% silencing of Map4k4 mRNA in macrophages treated with siRNA–EP complexes (Figure 4A, left-hand panel). 3T3-L1 fibroblasts and adipocytes were treated with complexes prepared by mixing 1 nmol of siRNA targeting MAP4K4 or scr siRNA with 25 nmol of EP. We observed an 80% knockdown of MAP4K4 in 3T3-L1 fibroblasts treated with MAP4K4-siRNA–EP (Figure 4B, left-hand panel). Similarly, MAP4K4 expression was significantly silenced when adipocytes were treated with EP–MAP4K4-siRNA compared with scr siRNA (Figure 4C, left-hand panel). Finally, we used COS-7 cells for similar experiments and found significant knockdown of MAP4K4 with 40 pmol of MAP4K4-siRNA and 1 nmol of EP (Figure 4D, left-hand panel). These data show that EP is a potent transfection reagent that delivers functional siRNA to phagocytic, as well as non-phagocytic, cells under tissue culture conditions. We then compared treatment of phagocytic macrophages with other cell types, including 3T3-L1 fibrobasts, 3T3-L1 adipocytes and COS-7 cells with 1×107 glucan shells loaded with 500 pmol of EP and 10 pmol of scr or MAP4K4 siRNA following the protocol described in Figure 3(C). Cells were treated with these new three-component GeRPs at the ratio previously showed to be most effective, 10 GeRPs/cell. Silencing of Map4k4 mRNA was observed only in macrophages (Figure 4A, right-hand panel), whereas MAP4K4 expression was unchanged in 3T3-L1 fibrobasts, adipocytes and COS-7 cells treated with MAP4K4-GeRPs compared with scr GeRPs (Figures 4B–4D, right-hand panels). Importantly, such specificity is not obtained when the GeRPs are overloaded with very high concentrations of siRNA and EP, such that siRNA–EP complexes reside outside of the glucan shells (results not shown). These results suggest that siRNA–EP complexes carried by the glucan shells provide targeted delivery to phagocytic cells, such as macrophages in vitro.
GeRPs mediate gene silencing in macrophages, but not other cell types, in cell culture
Gene silencing is limited to macrophages that have phagocytosed GeRPs in vitro
Next, we carefully analysed the efficiency of GeRP-mediated silencing in macrophages to determine whether silencing might be mediated by siRNA leaking out from GeRPs prior to their internalization into the cells. Fluorescein-labelled glucan shells (green) were loaded with Dy547-siRNA (red) following the protocol described in Figure 4(B). Microscopic analysis showed that GeRPs were efficiently phagocytosed by macrophages and that intracellular siRNA was released from the glucan shells (Figure 5A). Importantly, upon loading glucan shells with EP plus siRNA directed against the surface marker CD45 (CD45-GeRPs), a significant silencing of CD45 mRNA and protein levels was observed without effect on the expression of other surface markers such as CD68 or F4/80 (Figure 5B). In order to rule out the possibility of knockdown by free siRNA–EP complexes unbound to glucan shells, we performed FACS analysis on macrophages following their treatment with increasing amounts of FITC–GeRPs per cell; 1 GeRP/cell, 5 GeRPs/cell and 10 GeRPs/cell. In this set of experiments, we loaded GeRPs with EP plus siRNA directed against F4/80 (F4/80-GeRPs). Figure 5(C) shows dot-plot graphs of (1) untreated and unstained cells (left-hand panel), (2) untreated and F4/80-stained cells (middle panel) and (3) FITC–GeRP-treated and F4/80-stained cells (right-hand panel). Most of the cells were F4/80-positive (Figure 5C, panel 2) and approximately 85% of the cells contained FITC–GeRPs (Figure 5C, panel 3). Silencing of F4/80 mRNA was observed only in macrophages treated with 10 GeRPs per cell (Figure 5D), whereas CD45 expression was unchanged in cells treated with F4/80-GeRPs compared with scr GeRPs (Figure 5E). Moreover, FACS analysis showed that F4/80 staining was lower only in cells containing the F4/80-GeRPs (FITC+ cells) when treated with 10 GeRPs per cell. Importantly, no gene silencing was observed in the macrophages that did not contain detectable GeRPs (FITC− cells) or cells treated with fewer than 10 GeRPs per cell (Figures 5E and 5F). These data indicate that the F4/80 knockdown observed in primary macrophages specifically results from phagocytosis of the three-component GeRPs, composed of glucan shell, EP and siRNA.
Gene silencing is limited to the macrophage population that has phagocytosed GeRPs
Glucan shells containing EP and siRNA specifically silence genes in macrophages in mice
To confirm the phagocytosis of the simplified three-component GeRP formulation by macrophages in vivo, we treated mice with GeRPs i.p. Briefly, 10-week-old C57Bl6/J male mice were i.p. injected once with GeRPs containing 125 μg of siRNA/kg of body weight. The day after the injection, PECs, which contain a high percentage of macrophages (results not shown), were isolated. Microscopic analysis was performed on PECs isolated from mice treated with fluorescein–glucan shells loaded with Dy547-siRNA in the absence or presence of EP (Figure 6A). PECs were stained with F4/80 antibody prior to confocal microscopy. The results showed that glucan shells were significantly phagocytosed by F4/80-positive macrophages and that the siRNA was delivered to the cytoplasm of the cells only when EP was included in the GeRPs (Figure 6A).
GeRPs mediate gene silencing specifically in macrophages in mice
To test whether the siRNA delivered to macrophages within GeRPs containing EP was functional, we treated mice with fluorescein-glucan shells loaded with F4/80-siRNA following the protocol outlined in Figure 6(B). Mice were treated with GeRPs containing 625 μg of siRNA/kg of body weight by injection i.p. once a day for 5 days. The day after the last injection, PECs were isolated and F4/80 expression in FITC+ and FITC− sorted cells was determined (Figure 6C). FACS analysis showed that approximately 50% of the cells were F4/80+ and 25% were F4/80+/FITC+ (Figure 6C, left-hand panels and results not shown). F4/80 silencing was observed only in PECs containing GeRPS (FITC+ cells) and not in cells with no detectable GeRPS (FITC− cells) (Figure 6D). These results confirm GeRP-specific gene silencing in macrophages in vivo using the three-component GeRP system (Figure 6D). Other surface markers, such as CD45 or CD68, were unchanged in PECs from mice treated with F4/80-GeRPs compared with scr GeRPs in FITC+ or FITC− cells (Figures 6E and 6F). We also treated mice with GeRPs loaded with siRNA targeting MAP4K4 and observed MAP4K4 knockdown in PECs from mice treated with MAP4K4-GeRPs compared with scr or unloaded GeRPs (see Supplementary Figure S5A at http://www.BiochemJ.org/bj/436/bj4360351add.htm). Moreover, using a second native or stable siRNA species targeting a different sequence of MAP4K4, we showed that MAP4K4-GeRPs could specifically silence MAP4K4 in macrophages in vivo (results not shown). This suggests that a native siRNA is as efficient as a stabilized siRNA in knocking down genes when bound to EP and glucan shells. Finally, in order to determine whether depletion of Map4k4 mRNA was mediated by an RNAi-dependent gene silencing mechanism, we performed 5′-RACE analysis, which identifies the cleavage sites of Map4k4 mRNA. Following treatment of mice with GeRPs containing MAP4K4 siRNA, but not scr siRNA or unloaded GeRPs, a unique 5′-RACE product could be detected (Supplementary Figure S5B). No liver toxicity was detected, as assessed in mice treated with PBS, scr GeRPs or MAP4K4-GeRPs by measuring levels of ALT and AST activities in serum (see Supplementary Figure S6 at http://www.BiochemJ.org/bj/436/bj4360351add.htm). These results demonstrate that the three-component GeRP formulation developed here specifically silences genes in macrophages in vivo, through an RNAi mechanism, without inducing toxic liver effects.
This key advance reported in the present paper is the development of a simplified, two-component vehicle to encapsulate and deliver functional siRNA specifically to phagocytes, either in culture or in intact mice (Figures 4–6). This delivery system employs our previously described preparation  of glucan shells with diameters of 2–4 μm to confer target selectivity to macrophages and dendritic cells on the basis of their ability to tightly bind and phagocytose such particles [26,27]. We have devised a simple method to trap siRNA within the glucan shells using the amphipathic peptide EP, which contains polymeric and leucine residues designed to form an α-helical structure. We show that EP can bind siRNA to form nanoparticles of varying sizes that increase with increasing pH (Figure 3). Such nanoparticles can themselves mediate significant gene silencing in many cell types under tissue culture conditions (Figure 4), consistent with a recent report  that was published as we were preparing the present manuscript. Interestingly, another study showed that EP could mediate gene silencing in rat hepatoma cells only when combined with poly-L-lysine . By forming very small siRNA–EP nanoparticles in the presence of glucan shells and then raising the pH, we show association of siRNA and EP with the glucan shells (Figure 3). This method defines a much simplified formulation of GeRPs compared with the original five-component system reported previously . Using this new GeRP formulation containing only glucan shells, EP and siRNA, we demonstrate effective gene silencing specifically in phagocytes in vitro and in vivo (Figures 4–6).
In the context of the new formulation of GeRPs described in the present paper, the peptide EP serves multiple functions in the overall process of anchoring the siRNA within glucan shells and in RNAi-mediated gene silencing. This allowed simplifying the original formulation that comprised five different chemical species. The first generation of GeRPs contained a cationic core composed of tRNA surrounded by PEI within the glucan shells . In this original GeRP formulation, the peptide EP and the siRNA were loaded on to this cationic core and trapped with a second layer of cationic PEI (Figure 1). In the simplified GeRP formulation described in the present paper, the peptide EP itself serves as an entrapment vehicle or core for the siRNA in the glucan shells when EP, siRNA and shells are treated at staged pH conditions, as described in Figure 3. We also determined that the siRNA within the nanoparticulate siRNA–EP complexes in such GeRPs was more readily released at endosomal-like conditions of pH 5 in vivo compared with the original GeRP formulation (results not shown). Thus the peptide EP serves as an effective entrapment material for the siRNA, but also confers more favourable release kinetics than the PEI present in the original GeRP formulation. In these experiments, we discovered that EP was a required component of the GeRPs for efficient silencing in macrophages (Figure 1), and that the EP itself was a potent transfection reagent, which binds and delivers siRNA to multiple cell lines or primary cells (Figure 4).
A major limitation of the original GeRP system was the potential toxicity derived from the PEI layers within the glucan shells that retained the siRNA. We (results not shown) and others  have noted that high concentrations of PEI are toxic to cultured cells and there is also toxicity associated with this cationic material when administered in vivo . The original GeRP formulation, which contained significant amounts of PEI, represented a significant limitation to the potential clinical applications of the GeRP technology. Thus elimination of PEI in the present GeRP formulation is highly desirable both for research purposes and for potential therapeutic development.
The EP in GeRPs also replaces the additional intended function of PEI in promoting escape of siRNA from phagosome membranes. Different classes of peptides have been successfully employed as siRNA delivery carriers for crossing membranes, including protein-derived CPPs, cationic peptides and amphipathic peptides [12,34]. CPPs can be used as potential siRNA carriers and their conjugation to siRNA can improve their delivery into cultured cells, although non-covalent complexes with siRNA appear more appropriate for gene silencing in vivo [12,34,38–40]. MPG [derived from the fusion peptide domain of HIV-1 gp41 protein and the nuclear localization sequence of SV40 (simian virus 40) large T antigen] was the first amphipathic peptide shown to efficiently deliver siRNA in a biologically active form into cells [40,41]. Our present study suggests that EP, a α-helical secondary amphipathic peptide with hydrophobic and hydrophilic faces, was able to provide the necessary endosomal escape for GeRP-delivered siRNA to mediate gene silencing [39,42]. EP action has been shown to entail rapid adsorption to cell surfaces, with no detectable damage to the plasma membrane . Once internalized, the EP in the acidified endosome is converted into its polycationic form, which apparently acts to permeabilize the endosomal membrane. This acid-induced permeabilization of the endosomal membrane (denoted the proton sponge effect ) allows co-internalized cargo to pass from the endosome into the cytosol of the cell [31,35]. Further studies are needed to provide more detailed information on the kinetics and mechanisms of phagosome escape and intracellular trafficking of siRNA or siRNA–EP complexes that escape the glucan shells within the phagosome. Another secondary amphipathic peptide, CADY, has been shown to promote target knockdown without inducing toxic side effects . It will be interesting in future studies to evaluate the entrapment of other such peptide and non-peptide transfection reagents into the glucan shells for specific gene silencing in macrophages.
A key objective of siRNA delivery strategies for potential therapeutics is to generate vehicles that target specific cell types while leaving other tissues unperturbed. Such target cell specificity should reduce toxic effects derived from silencing genes that are not targeted as well as from silencing appropriately targeted genes in inappropriate tissues. Macrophages have been involved in the pathogenesis of many inflammatory diseases such as asthma, rheumatoid arthritis and inflammatory bowel disease . In addition, macrophage dysfunction has been implicated in other diseases such as atherosclerosis , diabetes [47,48] and Alzheimer's disease , in which inflammatory responses may be involved in the pathogenesis. Finally, abnormalities in macrophage function are also frequently seen in a variety of human cancers . The delivery of siRNA to macrophages without targeting other cell types is therefore an important objective in the clinical application of siRNA. Gene silencing in macrophages in vivo has been achieved using various carriers such as cationic liposomes  and immunoliposomes conjugated to surface-bound cell-specific antibodies . However, although these siRNA delivery vehicles administered in vivo led to siRNA distributions in spleen, delivery to liver at early time points was also observed . In the present study, we observed gene silencing only in cells that had actually internalized the GeRPs, as detected by FACS analysis (Figures 5 and 6). Thus the GeRP technology appears to restrict gene silencing to phagocytic cells even when incubated in the presence of other cells that do not take up the GeRPs.
Other siRNA delivery materials have recently been constructed that restrict targeting to phagocytic cell types [54–57]. Biodegradable poly-D,L-lactide microspheres and ketal-based siRNA particles, for example, also effectively mediate gene silencing in macrophages. These systems employ the principle of size exclusion to limit non-specific uptake by non-phagocytic cells [54–57]. Inflammatory macrophages produce copious amounts of reactive oxygen and inflammatory cytokines and mediate resistance against intracellular parasites and certain tumours . In general, anti-inflammatory macrophages have immunoregulatory functions to orchestrate encapsulation and containment of parasites, and promote tissue repair, remodelling and tumour progression . ROS (reactive oxygen species) that are released by macrophages during inflammation induce the release of siRNA from the ROS-sensitive thioketal-based delivery vehicle, thus localizing the release of siRNA to inflamed tissues . In contrast, GeRPs are able to target non-inflammatory as well as inflammatory macrophages based on their glucan component. Indeed, our preparations of glucan shells are specifically recognized and phagocytosed by macrophages and dendritic cells, independently of their activation state [15,25]. This is an important feature of GeRPs, considering the pathological role of anti-inflammatory macrophages in tumour progression [59,60]. The results shown in the present study reveal a simple method to produce GeRPs with only glucan and EP, components necessary for retaining and delivering functional siRNA, and should facilitate studies designed to evaluate the role of specific phagocyte genes in many biological processes.
dynamic light scattering
Dulbecco's modified Eagle's medium
fetal bovine serum
β-1,3-D-glucan-encapsulated siRNA particle
mitogen-activated protein kinase kinase kinase kinase 4
peritoneal exudate cell
rapid amplification of cDNA ends
reactive oxygen species
small interfering RNA
tumour necrosis factor α
Gregory Tesz, Myriam Aouadi, Mattieu Prot, Sarah Nicoloro, Emile Boutet, Shinya Amano, Anca Goller, Mengxi Wang, Chang-An Guo, William Salomon, Rebecca Baum and Mark O'Connor performed the experiments. All the authors participated in designing the experiments, and analysing and interpreting the results. Myriam Aouadi and Michael Czech wrote the manuscript.
We thank members of our laboratory group for excellent discussion of the data in this paper. We also appreciate the help of Paul Furcinitti for the confocal microscopy and the staff of the Core Flow Cytometry Lab.
This work was supported by grants from the National Institutes of Health [grant numbers DK085753 and AI046629 (to M.P.C.)], the Juvenile Diabetes Research Foundation [grant number 17-2009-546], and from Core Facilities in the University of Massachusetts Diabetes and Endocrinology Research Center, also funded by the National Institutes of Health [grant number DK325220].
These authors contributed equally to this work.