An efficient method for gene silencing in human primary plasmacytoid dendritic cells: silencing of the TLR7/IRF-7 pathway as a proof of concept

Plasmacytoid dendritic cells (pDC) are specialized immune cells that produce massive levels of type I interferon in response to pathogens. Unfortunately, pDC are fragile and extremely rare, rendering their functional study a tough challenge. However, because of their central role in numerous pathologies, there is a considerable need for an efficient and reproducible protocol for gene silencing in these cells. In this report, we tested six different methods for siRNA delivery into primary human pDC including viral-based, lipid-based, electroporation, and poly-ethylenimine (PEI) technologies. We show that lipid-based reagent DOTAP was extremely efficient for siRNA delivery into pDC, and did not induce cell death or pDC activation. We successfully silenced Toll-Like Receptor 7 (TLR7), CXCR4 and IFN regulatory factor 7 (IRF-7) gene expression in pDC as assessed by RT-qPCR or cytometry. Finally, we showed that TLR7 or IRF-7 silencing in pDC specifically suppressed IFN-α production upon stimulation, providing a functional validation of our transfection protocol.

Moreover, major advances in siRNA modifications and delivery reagents have overcome initial problems of stability and cellular uptake in most cases. However, active cellular nucleases, specific membrane properties and other yet unidentified factors often make siRNA-mediated gene inactivation in human primary cells difficult 19 . Furthermore, some siRNA, also called immunostimulatory siRNA, can be endocytosed and thus trigger TLR7 pathway, leading to nonspecific pDC activation 20,21 . This phenomenon can be reduced by the incorporation of 2′ -O-methyl modifications into the sugar structure of selected nucleotides within both the sense and antisense strands 21,22 . Such 2′ -O-methyl modifications have also been shown to confer resistance to endonuclease activity 23 and to abrogate off-target effects 24 . However, the transfection of siRNA in human primary pDC, in itself, remains a major challenge.
This latter issue was addressed in this report, as the transfection method seems to be decisive for a successful gene silencing. Among the various methods to deliver siRNA into cells, liposome-based systems are widespread due to their practical mode of use in most cell types. This method is based on siRNA packaging by cationic lipids into liposomal particles that facilitate the cellular uptake through plasma membrane and siRNA protection from enzymatic degradation during cellular endocytosis. Recently, liposome-based transfection has been shown particularly efficient for gene silencing in monocytes and myeloid dendritic cells 25 . Besides liposome-based technology, poly-ethylenimine (PEI) was one of the first transfection agent discovered 26 , after poly-L-lysine. PEI condenses nucleic acids into positively charged particles, which bind to anionic cell surface residues. Such polymer-oligonucleotide complexes (polyplex) are brought into the cell via endocytosis. Once inside the endosomes, protonation of the amines results in an influx of counter-ions and a lowering of the osmotic potential. Osmotic swelling bursts the endosomal vesicle, thus releasing the polyplex into the cytoplasm. If the polyplex unpacks, then the nucleic acids are free to diffuse to their targeted compartment 27,28 . However, PEI can be extremely cytotoxic 29,30 , due to the disruption of endosomal membranes leading to cellular stress and cell death.
As an alternative to chemical agents, electroporation methods were also shown to have high transfection efficiencies in many cell lines. This method physically translocates siRNA into cells by a brief electric pulse, which induces a membrane perturbation allowing entry of nucleic acids. However, even if this technic is often used in primary human T cells, cells tend to exhibit higher levels of cell death after electroporation. Finally, gene silencing in hard-to-transfect cells can be achieved with virus-based vectors, and in particular lentiviruses, that encode short-hairpin RNA (shRNA) to induce specific mRNA degradation. However, pDC are known to be hardly infectable by Human Immunodeficiency Virus (HIV-1) 31 and transduction with HIV-derived vectors might be inefficient. Furthermore, it is likely that RNA encapsidated within viral particles will engage TLR, triggering cellular activation and induce type I IFN production.
To our knowledge, there is no precise and efficient protocol dedicated to transfection and gene inactivation in primary human pDC by siRNA. In this report, we have evaluated a panel of six different methods to achieve gene silencing in pDC. Three different liposome-based reagents, including DOTAP, HiPerFect and the new generation of Lipofectamine RNAiMAX, an optimized PEI reagent (JetPRIME) and electroporation using the Amaxa Nucleofector technology were tested. Viral-based transduction using lentiviral vectors was also tested. We show that liposome-based method using DOTAP is the only efficient method for siRNA delivery into pDC, providing a technological solution for gene silencing and functional studies in this essential cellular population.

Results
Transfection and silencing efficiency in human primary pDC. Human primary pDC were purified from peripheral blood of healthy donors by centrifugation on a density gradient and negative selection using antibody-coupled magnetic beads (Fig. 1a) 32 . Purity was assessed by flow cytometry analysis and was found superior to 90% (Fig. 1b). Cells were seeded at 10 5 cells/mL in 96-well plates and treated with the different transfection methods as explained in Methods. First, we investigated whether a HIV-derived vector would be an appropriate tool to deliver a shRNA into pDC. In order to evaluate the transduction efficiency, we transduced pDC with a VSV-G-pseudotyped GFP-encoding HIV-derived vector at high MOI (200 ng/mL) and transduction efficiency was evaluated by FACS analysis. Only 17.9% of pDC expressed GFP, thus confirming that pDC are poorly transducible with HIV-derived vectors (Fig. 1c). Furthermore, we observed by confocal microscopy that pDC formed numerous giant mononuclear cells (GMC) (data not shown) at a 5-10 times higher frequency than previously described in HIV-treated pDC cultures 33 . Nevertheless, we tried to transduce pDC at a higher multiplicity of infection (MOI) in an attempt to increase the transduction efficiency. The activated pDC was defined as IKpDC, expressing TRAIL at its surface, as previously characterized 34 (Supp. Fig. 1). pDC tend to change phenotype during the different stages of activation. The activating cells are bigger. The phenotype of the resting cells characterized the cells prior to their activation or prior to their death. However, when pDC were transduced at 2000 ng/mL, cell mortality reached 30% and the percentage of GFP-positive live cells was limited to 25% (Fig. 1c,d), even if a very small portion of cells (2.4%) were efficiently transduced. Altogether, these results disqualify this virus-based method to achieve gene silencing in pDC.
Then, we evaluated different chemical transfection reagents and electroporation to achieve pDC transfection with siRNA. As control for transfection, we used a siRNA targeting cyclophilin B that is labeled with the fluorescent dye DY-547 (siGLO-DY-547). The percentage of transfected pDC was determined by FACS analysis using fluorescence as readout (Fig. 2a,b). The siGLO was mixed with transfection agent DOTAP, HiPerFect, RNAiMAx or JetPRIME, and added to cell cultures following manufacturer's instructions. For the Amaxa method, pDC were electro-stimulated following instructions adapted to primary DC cells. At 4 h post-transfection, a very high percentage of siGLO-DY-547 positive pDC was observed with DOTAP (96%), HiPerFect (98%) and JetPRIME (91%), suggesting some high transfection efficiency. In contrast, pDC transfected with Lipofectamine RNAiMAX or Amaxa showed lower percentages of transfection (35% and 30% of positive cells, respectively). We next tested the efficiency of gene silencing of the cyclophilin B gene in pDC using quantitative RT-PCR. As shown in Fig. 2c, mRNA levels of cyclophilin B where only reduced when pDC were transfected with DOTAP (more than 70%).
In contrast, none of the other methods led to an efficient inhibition of cyclophilin B mRNA levels suggesting that DOTAP was the only efficient system to transfect pDC and achieve gene silencing. As the major issue with pDC transfection is cellular activation, this parameter was determined by quantification of TNF-related Apoptosis-Inducing Ligand (TRAIL) mRNA expression levels in pDC as previously described 35,36 . The siGLO complexed with either liposome-based transfection methods or JetPRIME did not induce any increase in TRAIL mRNA expression levels (Fig. 2d). In contrast, the Amaxa method induced 6-fold increase in TRAIL mRNA expression levels, thus supporting pDC activation by this transfection method.
Viability and activation of pDC. In parallel to transfection efficiency, we determined the impact of the different protocols on pDC viability. Cells were subjected to the different methods of transfection in the absence or presence of control siRNA. Dead cells were stained specifically using the live-dead technology, and their percentage determined by flow cytometry analysis at 24 h post-treatment. As shown in Fig. 3b (upper panels), the percentage of living cells was similar in non-transfected and DOTAP-treated cells (around 85% of living cells). In contrast, the percentage of dead pDC massively increased when cultured in presence of HiPerFect or Lipofectamine RNAiMAX. Amaxa technology induced around 50% of cell death. However, the number of cells in the well was drastically reduced after the electroporation (loss of 70% of the cells) (Supp Fig. 1). Cell cultures treated with JetPRIME showed an intermediate percentage of dead cells, with approximately a two-fold increase compared to control. As shown in Fig. 3b   We also evaluated whether transfection induced pDC activation, which was determined by quantification of TRAIL mRNA expression levels. Liposome-based transfection methods or JetPRIME did not induce any increase in TRAIL mRNA expression levels, whether control siRNA was present or not (Fig. 3d). In contrast, the Amaxa method induced 6-fold increase in TRAIL mRNA expression levels, as observed previously.
Study of Transfection efficiency of labeled siRNA by microscopy. In order to better understand our observations, especially the high percentage of siGLO-DY-547 positive cells when transfected with JetPRIME contrasting with the very low level of gene silencing, we performed a microscopic study of pDC transfected with the various transfection reagents and especially, DOTAP and JetPRIME. Cells transfected with the siGLO-DY-547 were cultured for 4 h at 37 °C, then stained with both DAPI for nucleus labeling and the lysosomal-associated membrane protein 1, LAMP-1. In order to be efficient, siRNA should be localized in the cytoplasm but not into lysosomal compartments. As previously shown, pDC harbored a very large nucleus and a small cytoplasm  characteristic of this cell subtype (Fig. 4a). When human pDC were incubated with the siGLO-DY-547 in the absence of any transfection reagent, we did not observe any siRNA associated to the cells, confirming that siRNA must be actively transfected to reach the intracellular compartment. In contrast, we found high levels of siRNA within pDC transfected with DOTAP ( Fig. 4a,b). Furthermore, we showed that the labeled siRNA was not localized in lysosome, as we did not find any colocalization with the LAMP-1 marker. In fact, the vast majority of the siRNA was close to the nucleus. In contrast to results obtained with DOTAP, the vast majority of labeled siRNA transfected with JetPRIME was localized outside the cells (defined by localized in the medium or bound to the cell surface). We observed large aggregates of siRNA, forming large clusters of 0.5 to 2 μ m diameter at the periphery of the cells. We thus quantified the "outside" siRNA versus the intracellular siRNA in cells transfected by DOTAP or JetPRIME. As shown in Fig. 4c, the vast majority of labeled siRNA was localized intracellularly when DOTAP was used, in contrast to JetPRIME transfected cells in which the siRNA was found outside the cells as large aggregates, either bound or not to cellular surface. This latest result may provide explanation for our previous flow cytometry and gene silencing efficiency data.

Functional validation of gene silencing into human primary pDC.
As key effector cells of the innate immune response, pDC express TRAIL and produce massive amounts of type I IFN after recognition by TLR7 of single-stranded RNA from pathogens such as HIV or Influenza A. This TLR7 activation induces the phosphorylation and nuclear translocation of IRF-7, which is the main transcription factor regulating type I IFN production in pDC 14,37 . We thus attempted to silence TLR7 and IRF-7 in human primary pDC by siRNA transfection with DOTAP. As a negative control, we decided to silence the C-X-C chemokine receptor type 4 (CXCR4), a highly expressed chemokine receptor that is not implicated in TRAIL or type I IFN production in HIV-stimulated pDC 36,38 . We first tested several concentrations of TLR7, IRF-7, or CXCR4 siRNA, and showed that 160 nM was the optimal concentration for an efficient silencing of targeted mRNA (Fig. 5a). This was confirmed at protein levels, as assessed by IRF-7 or CXCR4 immunostaining and FACS analysis (TLR7 protein could not be determined as we are missing a specific antibody) (Fig. 5b-e). We then verified that at this concentration, TLR7, IRF-7 and CXCR4 siRNAs were not inducing pDC activation. For this, we cultured pDC for 24 h with one of the four siRNA (CTL, TLR7, IRF-7 or CXCR4) or with a synthetic non-modified viral RNA (35 nucleotides long), all complexed with DOTAP at the concentration 160 nM and we measured TRAIL mRNA production by quantitative RT-PCR. As shown in Fig. 5f, none of the siRNA induced pDC activation at 160 nM unlike the control RNA, which induced a dramatic increase in TRAIL mRNA expression. Finally, we showed that TLR7 or IRF-7 siRNA drastically reduced IFN-α induction in pDC stimulated with TLR7 activators, including HIV-1, influenza virus and Gardiquimod (Fig. 6a,b). In contrast and as expected, IRF7, but not TLR7 silencing inhibited IFN-α when pDC were activated with the TLR9 ligand CpG A (Fig. 6a,b). We also showed by using a functional cellular assay that type I IFN secretion was strongly inhibited in pDC treated with TLR7 or IRF-7 siRNA (Fig. 6c). In contrast, CXCR4 silencing had not effect on IFN-α transcription, thus demonstrating that DOTAP transfected pDC were perfectly functional and produced similar levels of cytokine compared to untransfected cells (Fig. 6d). Altogether, these results demonstrate that gene silencing can be achieved in pDC by siRNA transfection with DOTAP, while preserving cellular viability and functional properties (Table 1).

Discussion
In this report, we have established a protocol for gene silencing in human primary pDC, a key effector cell type involved in immune response to pathogens and autoimmunity 34,39-41 . The method described herein opens new perspectives for the functional study of these cells. Primary pDC are extremely rare and fragile cells that only  survive a few days in cell culture and do not proliferate, making their transfection with siRNA a real challenge. Some reports have previously described gene silencing in bone marrow-derived mouse pDC or pDC-related cells. The GeneSilencer siRNA Transfection Reagent (a cationic liposome reagent) or the Amaxa technology were previously used to achieve gene silencing in murine pDC derived in vitro from proliferating bone marrow precursors 23 . It was also shown that a lentiviral vector coding for a specific shRNA could silence TLR7 expression in Gen2.2 cells, a human cell line that is often used as a surrogate of human pDC. As a result, IFN response to HIV, foamy viruses or influenza A virus was impaired 42,43 . Zhang & al. developed siRNA linked to CpG oligonucleotides to target TLR9 and facilitate siRNA internalization. Such siRNA-CpG conjugates were successfully used to silence STAT3 in monocyte-derived CD303 + (BDCA-2) cells that exhibit phenotypic similarities with pDC 44 . However, and to our knowledge, our report is the first to achieve gene silencing in primary human pDC.
We established that most transfection methods tested, including virus-like particles, HiPerFect, Lipofectamine RNAiMAX and Amaxa, induced massive cell death of primary pDC. JetPRIME was not highly cytotoxic, but formed large aggregates that remained in the extracellular compartment and failed to deliver siRNA into the cells. Only DOTAP was found efficient to deliver siRNA in the cytoplasm where gene silencing is achieved. Reasons why DOTAP was efficient and not cytotoxic, compared to other liposome-based reagents like HiPerFect or RNAiMAX, remain elusive. We can only assume that compared to HiPerFect or Lipofectamine RNAiMAX, DOTAP transfection reagent, which is based on 1,2-dioleoyl-3-trimethylammonium-propane monocationic lipid, is the best compromise in terms of positive charges and diameter of liposome particles to transfect pDC. We also showed that none of the siRNA tested induced pDC activation by TLR7 and this probably relates to siRNA modifications. In this report we used the ON-TARGETplus siRNA reagents from Dharmacon, which uses the SMARTpool technology and 3 of 4 individual siRNA to target the mRNA. Furthermore, these siRNA have been 2′ -O-methyl modified, which was previously reported to decrease off-target activity and cell activation 24,45 . Finally, we determined that the best siRNA concentration for gene silencing in pDC (160 nM) is very close to the one successfully used in primary human monocytes and conventional dendritic cells (200 nM) 25 .
This transfection protocol allowed cyclophilin B, CXCR4, IRF-7 and TLR7 efficient silencing in primary pDC. Cyclophilin B or CXCR4 silencing had no impact on the cellular response to TLR7 ligands, thus demonstrating that siRNA transfection with DOTAP did not alter IFN-mediated pDC production. In contrast, TLR7 or IRF-7 silencing impaired pDC response to TLR7 ligands, whereas only IRF-7 silencing also inhibited pDC activation by a TLR9 ligand. It was previously shown in TLR7 knock-out mice 8 , that pDC response to single-stranded RNA depends on TLR7. Here, we provide some direct demonstration that TLR7 is required for the sensing of HIV or influenza virus particles by primary human pDC. We also confirmed that IRF-7 is needed, something previously established in primary human pDC from patients with mutated forms of this transcription factor 37 . Altogether, this provides a robust validation of our transfection protocol at the mRNA and protein levels that should greatly facilitate functional studies of this unique cell type.

Methods
Blood samples, isolation and culture of blood leukocytes. Blood from healthy HIV-1-seronegative blood bank donors was obtained from "Etablissement Français du Sang" (convention # 07/CABANEL/106), Paris, France. A written informed consent was obtained from all subjects involved. Experimental procedures with human blood have been approved by Necker Hospital Ethical Committees for human research and were done according to the European Union guidelines and the Declaration of Helsinki. In vitro experiments were performed using human peripheral blood mononuclear cells (PBMC) isolated by density centrifugation with Lymphoprep medium (StemCell Technologies). pDC were purified by negative selection with the Human plasmacytoid DC enrichment kit (StemCell Technologies). Cells were cultured in RPMI 1640 (Invitrogen, Gaithersburg, MD) containing 10% fetal bovine serum (Hyclone, Logan, UT). After purification, we obtained purity higher than 90% for pDC.
Small interference RNA experiments. pDCs were seeded at 10 5 cells/100 μ L in 96-well plates and incubated at 37 °C. The ON-TARGETplus SMARTpool siRNA targeting TLR7, IRF-7 or CXCR4 as well as the ON-TARGETplus Non-targeting siRNA (siCTL) and the siGLO Cyclophilin B Control siRNA were purchased from Thermo Scientific, Dharmacon (Illkirch, France). Upon arrival, siRNAs came in dried pellets and were stored as such at − 20 °C. The resuspension of siRNA pellets was done in 1X siRNA Buffer (Thermo Scientific, Dharmacon). The working aliquots (20 μ M) were then stored at − 20 °C. In this report, we used a final siRNA concentration of 160 nM: 0.88 μ l (= 17.6 pmol or ≈ 0.23 μ g) for the six-well plates from the 20 μ M siRNA working stocks. Each transfection agent was used according to manufacturer's protocol. Briefly: For DOTAP, the right volume of siRNA for a final concentration of 160 nM was diluted in PBS (1:5) then DOTAP (Roche Applied Sciences) was added v/v. The mix was incubated at room temperature during 15 minutes and then was added to cells in culture (RPMI-10% FBS).
For Hiperfect, the right volume of siRNA for a final concentration of 160 nM was mix with 6 μ L of Hiperfect (Qiagen) and 85μ L of X-vivo 15 media (Lonza) for 5 to 10 minutes at room temperature before the cells were added to the mix and put in culture.
For Lipofectamine RNAiMAX, the right volume of siRNA for a final concentration of 160 nM was diluted in 50 μ L of X-vivo 15 media. The Lipofectamine RNAiMAX reagent (1 μ L) (Thermo Fisher Scientific) was diluted in 50 μ L of X-vivo 15 medium for 5 minutes at room temperature. The two solutions were mixed and incubated for 20 minutes at room temperature before the cells were added to the mix and put in culture.
For JetPrime, the right volume of siRNA for a final concentration of 160 nM was diluted in 12,5 μ L of JetPrime Buffer and 2 μ L of JetPrime (Polyplus) were added. The mix was incubated at room temperature during 10 minutes and then was added to cells in culture (RPMI-10% FBS).
For Amaxa, we used the Amaxa Human Dendritic Cell Nucleofector Kit and Amaxa II Nucleofactor (Lonza). The supplement was added to the nucleofector solution. The cell's pellet was resuspend in 100 μ L of mix and the right volume of siRNA for a final concentration of 160 nM was added. We used the Nucleofector ® Program U-002 program. Once the electroporation over, the cells were centrifuged to get rid of all supernatant and resuspend in 100 μ L of media (RPMI-10% FBS).
A synthetic non-modified viral RNA (35 nucleotides long) was used as a control RNA and complexed with DOTAP with the same method describe previously.
Finally, cells were incubated at 37 °C for 24 hours before adding the different virus overnight.  Statistical analysis. P values (P) were determined using Student's t test, one-way ANOVA or two-way ANOVA and stated in each figure legend. P < 0.05 was considered statistically significant. *P < 0.05; * *P < 0.01 and * * *P < 0.001. Univariate distributions of flow cytometry data were performed by probability binning, in 300 bins using FlowJo software.  Table 2. Primer's sequences. 1 Primers amplify both IFN-α 1 and IFN-α 13 transcripts.