Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo

Abstract

Efficient intracellular delivery of proteins is needed to fully realize the potential of protein therapeutics. Current methods of protein delivery commonly suffer from low tolerance for serum, poor endosomal escape and limited in vivo efficacy. Here we report that common cationic lipid nucleic acid transfection reagents can potently deliver proteins that are fused to negatively supercharged proteins, that contain natural anionic domains or that natively bind to anionic nucleic acids. This approach mediates the potent delivery of nM concentrations of Cre recombinase, TALE- and Cas9-based transcription activators, and Cas9:sgRNA nuclease complexes into cultured human cells in media containing 10% serum. Delivery of unmodified Cas9:sgRNA complexes resulted in up to 80% genome modification with substantially higher specificity compared to DNA transfection. This approach also mediated efficient delivery of Cre recombinase and Cas9:sgRNA complexes into the mouse inner ear in vivo, achieving 90% Cre-mediated recombination and 20% Cas9-mediated genome modification in hair cells.

Figure 1

Strategy for delivering proteins into mammalian cells by fusion or non-covalent complexation with polyanionic macromolecules and complexation with cationic lipids. (a) Recombinases, transcriptional-activator-like effector (TALE) proteins, and Cas9 endonucleases bind nucleic acids and are natively cationic (net theoretical charges are shown in black) and are not efficiently complexed with cationic lipids. These proteins can be rendered highly anionic, however, by fusion to either a supernegatively charged protein such as (−30)GFP, or by complexation with polyanionic nucleic acids. (b) We envisioned that cationic lipids commonly used to transfect DNA and RNA would complex with the resulting highly anionic proteins or protein:nucleic acid complexes, mediating their delivery into mammalian cells.

Figure 2

Delivery of Cre recombinase to cultured human cells. (a) Fusion of either highly cationic (+36)GFP or highly anionic (−30)GFP to Cre recombinase. We used a HeLa reporter cell line that expresses DsRed upon Cre-mediated recombination to evaluate Cre delivery efficiency. (b) HeLa dsRed cells treated with 10 nM (−30)GFP-Cre and 1.5 µL of the cationic lipid formulation RNAiMAX. Cells were visualized after incubation for 48 hours in media containing 10% fetal bovine serum (FBS). (c) Delivery of (+36)GFP-Cre in 10% FBS media or in serum-free media, and (−30)GFP-Cre with or without the cationic lipid RNAiMAX (0.8 µL) in full-serum media. (d) Effect of cationic lipid dose on functional (−30)GFP-Cre delivery efficacy after 48 hours. (e) Comparison of several commercially available cationic lipids and polymers for functional delivery efficacy of (−30)dGFP-Cre. (f) RNAiMAX-mediated delivery of multiple anionic peptide or protein sequences fused to Cre. The net theoretical charge of the VP64 activation domain and the 3xFLAG tag is −22 and −7, respectively. All experiments were performed with 25 nM protein in 48-well plate format using 275 µL DMEM with 10% FBS and no antibiotics. Error bars reflect s.d. from three biological replicates performed on different days.

Figure 3

Delivery of TALE transcriptional activators into cultured human cells. (a) Design of an 18.5-repeat TALE activator fused C-terminally to a VP64 activation domain and N-terminally to (−30)GFP and an NLS. The overall net theoretical charge of the fusion is −43. (b) Activation of NTF3 transcription by traditional transfection of plasmids encoding TALE-VP64 activators that target sites in the NTF3 gene, or by RNAiMAX cationic lipid-mediated delivery of the corresponding NTF3-targeting (−30)GFP-TALE-VP64 proteins. For protein delivery experiments, 25 nM VEGF TALE, 25 nM NTF3 TALE 1, or 25 nM NTF3 TALEs 1–5 (5 nM each) were delivered with 1.5 µL RNAiMAX in 275 µL DMEM-FBS without antibiotics for 4 hours before being harvested. For plasmid transfections, a total of 300 ng of one or all five NTF3 TALE expression plasmids (60 ng each) were transfected with 0.8 µL Lipofectamine 2000 in 275 µL DMEM-FBS without antibiotics and harvested 48 hours later. Gene expression levels of harvested cells were measured by qRT-PCR and are normalized to GAPDH expression levels. Incubation times for TALE activators by plasmid transfection and protein delivery were those found to give maximal increases in NTF3 mRNA levels. (c) Time course of TALE activation for protein delivery and plasmid transfection by measuring NTF3 mRNA levels and then normalizing each method to the highest activation value achieved over any time point for that method. Optimal protein (25–50 nM) and lipid dosage (1.5 µL RNAiMAX) was used for each delivery technique. Error bars reflect s.d. from three biological replicates performed on different days.

Figure 4

Delivery of Cas9:sgRNA, Cas9 D10A nickase, and dCas9-VP64 transcriptional activators to cultured human cells. (a) Green entries: U2OS EGFP reporter cells were treated with 100 nM of the Cas9 protein variant shown, 0.8 µL of the cationic lipid shown, and either 50 nM of the sgRNA shown for Cas9 protein treatment, or 125 nM of the sgRNA shown for (+36)dGFP-NLS-Cas9 and (−30)dGFP-NLS-Cas9 treatment. The fraction of cells lacking EGFP expression was measured by flow cytometry. Blue entries: plasmid DNA transfection of 750 ng Cas9 and 250 ng sgRNA expression plasmids using 0.8 µL Lipofectamine 2000. (b) T7 endonuclease I (T7EI) assay to measure modification of EGFP from no treatment (lane 1), treatment with EGFP-targeting sgRNA alone (lane 2), Cas9 protein alone (lane 3), Cas9 protein + VEGF-targeting sgRNA + RNAiMAX (lane 4), DNA transfection of plasmids expressing Cas9 and EGFP-targeting sgRNA (lane 5), or Cas9 protein + EGFP-targeting sgRNA + RNAiMAX (lane 6). (c) T7EI assay of simultaneous genome modification at EGFP and three endogenous genes in U2OS cells 48 hours after a single treatment of 100 nM Cas9 protein, 25 nM of each of the four sgRNAs shown (100 nM total sgRNA), and 0.8 µL RNAiMAX. (d) Delivery of Cas9 D10A nickase and pairs of sgRNAs either by plasmid transfection or by RNAiMAX-mediated protein:sgRNA complex delivery under conditions described in (a) with 50 nM EGFP-disrupting sgRNAs (25 nM each) for protein delivery, and 250 ng sgRNA-expressing plasmids (125 ng each) for DNA delivery. EGFP-disrupting sgRNAs g1 + g5, or g3 + g7, are expected to result in gene disruption, while g5 + g7 target the same strand and are expected to be non-functional. (e) Delivery of dCas9-VP64 transcriptional activators that target NTF3 either by DNA transfection or RNAiMAX-mediated protein delivery. Error bars reflect s.d. from six biological replicates performed on different days.

Figure 5

DNA sequence specificity of Cas9-mediated endogenous gene cleavage in cultured human cells by plasmid transfection or by cationic lipid-mediated protein:sgRNA delivery using 1.6 µL RNAiMAX complexed with 100 nM Cas9 and 100 nM sgRNA. (a) T7EI assay was performed for on-target modification of endogenous CLTA, EMX, and VEGF genes in HEK293T cells. (b–d) On-target:off-target DNA modification ratio resulting from Cas9:sgRNA for plasmid transfection or cationic lipid-mediated protein:sgRNA delivery. The conditions for each treatment were adjusted to result in ~10% on-target cleavage, enabling a comparison of DNA cleavage specificity between the two delivery methods under conditions in which on-target gene modification efficiencies are similar. P values for a single biological replicate are listed in Supplementary Table 2. Each on- and off-target sample was sequenced once with > 10,000 sequences analyzed per on-target sample and an average of > 111,000 sequences analyzed per off-target sample (Supplementary Table 2). All protein:sgRNA deliveries and plasmid transfections were performed in 24-well format using 1.6 µL RNAiMAX in 550 µL DMEM-FBS without antibiotics. Error bars reflect s.d. from three biological replicates performed on different days.

Figure 6

In vivo delivery of Cre recombinase and Cas9:sgRNA complexes to hair cells in the mouse inner ear. (a) The scala media (cochlear duct) of P0 floxP-tdTomato mice (n = 4) were injected with 0.3 µL of 23 µM (−30)GFP-Cre in 50% RNAiMAX or with RNAiMAX alone (control). After 5 days, tdTomato expression indicative of Cre-mediated recombination was visualized using immunohistology. Red = tdTomato; green = Myo7a; white = Sox2; blue = DAPI. Yellow brackets indicate the outer hair cell (OHC) region. (b) Ten days after (−30)GFP-Cre delivery, intact espin (Esp)-expressing stereocilia of tdTomato-positive outer hair cells were present (arrow), similar to stereocilia in control cochlea. Red = tdTomato; green = Esp; white = Sox2; blue = DAPI. (c) Identical to (a) except using Lipofectamine 2000 instead of RNAiMAX. (n = 4). The upper and lower panels are images of mice cochlea at low and high magnification, respectively, detailing the efficiency of delivery and the effect on cochlear architecture and hair cell loss. (d) The scala media of P2 Atoh1-GFP mice (n = 3) were injected with 0.3 µL of 33 µM Cas9, 16.5 µM EGFP sgRNA in 50% RNAiMAX or Lipofectamine 2000 commercial solutions. Cas9-mediated gene disruption results in the loss of GFP expression when visualized 10 days later. The upper panels show GFP signal only, while lower panels include additional immunohistological markers. Yellow boxes in the lower panels highlight hair cells that have lost GFP expression. No obvious OHC loss was observed in the Cas9 + RNAiMAX or Cas9 + Lipofectamine 2000 groups. Red = tdTomato; green = Myo7a; white/light blue = Sox2; blue = DAPI. All scale bars (white) are 10 µm.