Comparative nucleic acid transfection efficacy in primary hepatocytes for gene silencing and functional studies

Background Primary hepatocytes are the best resource for in vitro studies directed at understanding hepatic processes at the cellular and molecular levels, necessary for novel drug development to treat highly prevalent diseases such as non-alcoholic steatohepatitis, cardiovascular disease and type 2 diabetes. There is a need to identify simple methods to genetically manipulate primary hepatocytes and conduct functional studies with plasmids, small interfering RNA (siRNA) or microRNA (miRNA). New lipofection reagents are available that have the potential to yield higher levels of transfection with reduced toxicity. Findings We have tested several liposome-based transfection reagents used in molecular biology research. We show that transfection efficiency with one of the most recently developed formulations, Metafectene Pro, is high with plasmid DNA (>45% cells) as well as double stranded RNA (>90% with siRNA or microRNA). In addition, negligible cytotoxicity was present with all of these nucleic acids, even if cells were incubated with the DNA:lipid complex for 16 hours. To provide the proof of concept that these conditions can be used not only for overexpression of a gene of interest, but also in RNA interference applications, we targeted two liver expressed genes, Sterol Regulatory Element-Binding Protein-1 and Fatty Acid Binding Protein 5 using plasmid-mediated short hairpin RNA expression. In addition, similar transfection conditions were used to optimally deliver siRNA and microRNA. Conclusions We have identified a lipid-based reagent for primary hepatocyte transfection of nucleic acids currently used in molecular biology laboratories. The conditions described here can be used to expedite a large variety of research applications, from gene function studies to microRNA target identification.


Background
The liver is a critical tissue that controls carbohydrate and lipid homeostasis. Highly prevalent human conditions, including obesity, cardiovascular disease, the metabolic syndrome, non-alcoholic steatohepatitis (NASH), or type 2 diabetes, are characterized by alterations in hepatic glucose and/or lipid metabolism. Deciphering the molecular mechanisms that mediate metabolic responses will open the possibility for the development of more effective therapies.
Establishing the function of genes and/or validating their role in disease is commonly attained through generation of null alleles. The discovery of RNA interference (RNAi) has provided novel avenues to accelerate gene function studies as well as drug target discovery [1][2][3][4][5]. RNAi is triggered by double stranded RNA (dsRNA) of variable lengths, typically 21 to 28nucleotides (nt). These can be chemically synthesized molecules (known as small interfering RNA, or siRNA) that upon delivery into cells or tissues, block gene expression for a few days. Re-administration is necessary for a sustained silencing effect. Alternatively, short hairpin RNA (shRNA) can be used to provide a continuous source of silencing molecules. To generate shRNA, expression cassettes are engineered using RNA polymerase II or RNA polymerase III promoters. The shRNA is transcribed in the nucleus as a linear molecule that folds to generate a stem of approximately 21-nt, connected with a loop sequence of variable length. The shRNA hairpin structure is cleaved into a siRNA [6][7][8].
Once an effective silencing shRNA construct is identified, the expression cassette can be incorporated into viral vectors for delivery to tissues in vivo [9].
In recent years it has become clear that microRNA (miRNA) are crucial modulators of biological functions, affecting processes as important as development and cell cycle stage. It has been estimated that more than 30% of human genes are regulated by miRNA at an average of 200 genes per miRNA [10][11][12][13]. MicroRNA exert their action by binding to the 3' UTR of mRNA and inhibiting protein synthesis or inducing mRNA degradation [14,15]. Gene target prediction coupled to experimental confirmation is beginning to yield information on the role of miRNA in normal cellular pathways as well as in disease. MicroRNA target identification is typically attained by transfection using chemically synthesized miRNA mimics (acting as mature miRNA that decrease levels of target mRNA and/or protein).
The primary culture of hepatocytes is regarded as the cellular model with highest similarity to liver physiology, and is the preferred approach to perform functional studies. Primary hepatocytes can be easily obtained by enzymatic digestion with collagenase [16][17][18]. Compared to other methods, transfection with cationic lipids offers the advantage of being a simple method of gene transfer, shortening gene function studies considerably. In recent years a variety of novel lipofection reagents have been developed with the potential to improve nucleic acid delivery in the absence of toxic effects. Optimization experiments in primary cells are highly time-consuming and expensive to perform. Here, we show transfection conditions with DNA and RNA molecules to knockdown two hepatic genes, Sterol Regulatory Element-Binding Protein-1 (SREBP-1) and Fatty Acid Binding Protein 5 (FABP5), using shRNA-expressing plasmids or siRNA. Similar conditions were successfully used to transfect primary hepatocytes with miRNA.

Plasmid transfection optimization
Critical aspects of mouse primary hepatocyte isolation by collagenase digestion are the strain and age of the animal, which affect cell viability, even if using the same collagenase concentration and perfusion time (data not shown). Using the protocol described in the Methods section, we consistently obtained high cell yield and viability (85-90%). High transfection efficiency can be obtained in cell lines with essentially all commercial transfection reagents, whereas primary cells in general have proven much more difficult to transfect [19,20]. Plasmid transfection, in particular, is more challenging, as these are large molecules and more difficult to be taken up by cells. To analyze the potential of novel lipofection methods to improve primary hepatocyte gene delivery, a 3 kb plasmid expressing green fluorescent protein (GFP) was complexed to four reagents: Metafectene, Metafectene Pro, Lipofectamine 2000, and Targefect-Hepatocyte. These reagents were selected based on information available from the respective company indicating high transfection efficiency in multiple cell lines and/or primary cells (including primary hepatocytes), in addition to low toxicity. To develop an optimal protocol for transfecting plasmid DNA into primary hepatocytes, DNA:lipid ratio, DNA concentration, culture medium and transfection time were considered. As shown in Figure 1A, we found that transfection was most efficient with all reagents when using a DNA:lipid ratio of 1:4. Targefect yielded the highest transfection efficiency (45.8 ± 4.2%), followed by Metafectene Pro (40.3 ± 2.6%), Lipofectamine 2000 (37.3 ± 3.8%), and Metafectene (31.5 ± 2.9%) ( Figure 1A). We then tested the effect of total amount of DNA, by transfecting cells with 1 to 5 μg, and maintaining the 1:4 (DNA:lipid) ratio. The highest transfection efficiency in the absence  High efficiency of transfection is often associated with toxicity. Ideally, a transfection reagent should yield high transfection efficiency with minimal or no cytotoxicity. To examine this, primary hepatocytes were cultured in the presence of DMEM or Opti-MEM, using Metafectene Pro, Lipofectamine 2000, and Targefect and cytotoxicity was evaluated after 24, 48, and 72 hr. As shown in Table 1, there was barely any difference in cytotoxicity among transfection reagents at 24 hours posttransfection; however Metafectene Pro resulted in much less cytotoxicity than Lipofectamine 2000 and Targefect at 48 and 72 hours post-transfection. Targefect yielded the highest level of transfection, but was by far the most  toxic of the three reagents. Lipofectamine 2000 gave slightly higher levels of transfection than Metafectene Pro and similar levels of toxicity when cells were cultured in the presence of Opti-MEM. However, transfection using Lipofectamine 2000 resulted in lower reproducibility than the other reagents, which was affected by elapsed time between cell plating and transfection (data not shown). Overall, Metafectene Pro gave the most consistent results, with least toxicity and shortest transfection time. This reagent has been previously shown to have superior performance in prostate cancer cells and the human embryonic carcinoma (EC) stem cell line NTERA2 [21]. Our data suggest that Metafectene Pro is a suitable reagent for use in primary hepatocytes, either for short-term expression (< 24 hours) or longer expression (>24 hours).

SREBP-1 and FABP5 silencing in primary hepatocytes
To provide the proof of concept that the transfection conditions described above are sufficient to silence an endogenous gene, primary hepatocytes were transfected with plasmids expressing shRNA to knock-down the transcription factor SREBP-1 [9] and the fatty acid binding protein FABP5 [22]. As shown in Figure 3, SREBP-1 mRNA expression was significantly reduced to 58% and 18% (at 24 and 48 hours, respectively) compared to the levels observed in cells treated with the shSCR control plasmid [expressing a scrambled sequence that does not bind to an mRNA, based on Basic Local Alignment Search Tool (BLAST) analysis (National Center for Biotechnology Information, NIH]. Western blot analysis revealed that SREBP-1 protein levels decreased in parallel to mRNA levels. FABP5 mRNA was reduced to 44% and 12% at 24 and 48 hours, respectively, compared to samples treated with shSCR plasmid. FABP5 protein reduction was evident only in samples harvested 48 hours after transfection, suggesting this protein has a long half-life. Thus, when testing the efficacy of customdesigned shRNA sequences, measuring the mRNA of the target gene is a more reliable method to assess the silencing efficacy of the constructs than measuring protein levels. In conclusion, the conditions reported here allow gene silencing of two independent genes by transfection of plasmids expressing shRNA. When conducting studies to address the function of a gene, using a chemically synthesized siRNA may be the preferred approach. In these experiments as well as in experiments directed at miRNA target identification, it is important to use conditions that result in highest transduction efficiency, to maximize the level of inhibition on mRNA and/or protein expression. To optimize the transfection conditions using a chemically synthesized double stranded RNA molecule, a fluorescently labeled miRNA, was used. Cells were transfected with 0.5, 1, or 2.5 μg of Dy547-labeled miRIDIAN microRNA Mimic Transfection Control. All conditions resulted in the presence of a fluorescent signal in the cytoplasm in a high percentage of cells (>90%). Using 2.5 μg of miRNA the intensity in the cytoplasm was slightly stronger than in cells transfected with 1 μg, but there were large clumps outside of the cells, suggesting that not all nucleic acid was taken by cells. Experiments were repeated in 6-cm dishes using the same dsRNA: lipid ratio and 1.5, 3.0 and 4.5 μg miRNA. Similar results were observed to those seen in 6-well plates (Figure 4A). No cytotoxicity was observed in any of these conditions for 48 hours.
To test the feasibility of using these conditions for transfection using siRNA, primary hepatocytes were incubated with siRNA to target FABP5, a non-targeting siRNA or with FABP1 siRNA (another member of the FABP family expressed in liver) ( Figure 4B). A significant FABP5 knock-down was observed after 72 and 96 hours with the siRNA specific to FABP5. Thus, similar transfection conditions can be used for delivery of large plasmids, miRNA and siRNA in primary hepatocytes.
In addition to be important for gene function studies, plasmid-mediated expression of shRNA in primary hepatocytes has other applications. In vivo RNAi offers the possibility to overcome many of the limitations presented by knock-out animal models. For example, absence of a gene may be incompatible with development, leading to death in utero. Using viral vector delivery of shRNA-expression cassettes it is possible to induce gene silencing in an adult animal. For the liver, in particular, adenovirus-mediated shRNA expression represents an excellent alternative, as the adenovirus has high tropism for hepatocytes [9]. Using the approach described here, several shRNA-expressing plasmids can be tested for their efficacy at silencing the target gene in primary hepatocytes, prior to transferring the most efficient shRNA expression cassette to an adenoviral vector for in vivo studies.

Conclusion
This study shows efficient transfection of primary hepatocytes and negligible cytotoxicity using plasmid DNA and dsRNA complexed to Metafectene Pro. These conditions can be applied to (i) functional studies where a gene of interest needs to be silenced, (ii) gene overexpression studies, (iii) testing the efficacy of shRNA constructs to target hepatic genes, and (iv) delivery of small RNA molecules, such as siRNA or miRNA mimics, which resulted in >90% transfection efficiency. The primary hepatocyte transfection conditions described here can be used to accelerate gene discovery, drug target validation and miRNA target identification studies, that are urgently needed to treat prevalent human diseases  A and B, top). The level of silencing is expressed relative to medium-treated cells. Proteins were quantified by Western blot (A and B, bottom). β-actin was used as a normalizer. Data are representative of two independent experiments (n = 3); *p < 0.05; **p < 0.01 shSCR compared to shSREBP-1 or shFABP5.

Animals
Male C57BL/6J mice (11 to 12 wks old, 26 to 30 g) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were fed rodent chow ad libitum and allowed free access of water. A standard 12 h light/12 h dark cycle (7 AM/7 PM) was maintained throughout the experiments. Animal studies were performed in compliance with Indiana University School of Medicine Institutional Animal Care and Use Committee guidelines.

Isolation of mouse primary hepatocytes
Primary hepatocytes were isolated by collagenase digestion, following a previously described protocol [16,17] The liver was transferred to 10 mm dishes with 15 ml DMEM containing 0.05% collagenase and was mechanically dissociated into single cells. Ten ml DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) were added to cells to reduce collagenase activity. Cells were filtered through a 70 μm pore size strainer (BD, San Jose, CA) and centrifuged at 100 × g for 5 min at 10°C. The cell pellets were gently washed with 20 ml DMEM/10% FBS, and centrifuged at 100 × g for 5 min at 10°C. This low centrifugation step allows pelleting hepatocytes while significantly reducing the number of other smaller liver cells [17]. Cells were resuspended in DMEM/10% FBS with antibiotics. Cell viability was assessed by trypan blue dye exclusion (85-90%). Primary hepatocytes were plated at a density of 4 × 10 5 cells in 2 ml of DMEM/10% FBS with antibiotics on 6-well plates, which resulted in 90% confluency the day after (Corning, Lowell, MA). Medium was replaced 2 hours after plating.
The construction of shRNA-expressing plasmids to knock-down SREBP-1 and FABP5 has been previously described [9,22]. For microRNA transfection optimization in 6-well plates, 0.5, 1, or 2.5 μg of Dy547-labeled miRIDIAN microRNA Mimic Transfection Control (Dharmacon, Lafayette, CO) was mixed with Metafectene Pro at a 1:3 ratio, following the same protocol described above for plasmids. The experiments were scaled-up to 6-cm dishes using 1 × 10 6 cells and 1.5, 3.0 and 4.5 μg miRNA. For FABP5 silencing using siRNA, siGenome SMARTpool siRNA (Dharmacon) were used following the conditions described in the text.

Statistical analysis
All experimental conditions were done in duplicate and repeated in at least two separate hepatocyte isolations. Data are presented as the arithmetic mean ± standard deviation. Statistical differences were calculated with a two-tailed, unpaired Student's t-test. Data were considered significant at p < 0.05.