Intracellular Delivery of His-Tagged Genome-Editing Proteins Enabled by Nitrilotriacetic Acid-Containing Lipidoid Nanoparticles

Abstract

Protein and peptide-based therapeutics with high tolerance and specificity along with low off-target effects and genetic risks have attracted tremendous attention during the last three decades. Herein, we report a new type of non-cationic lipidoid nanoparticle (LNP) for His-tagged protein delivery. Active lipidoids are synthesized by conjugating nitrilotriacetic acid (NTA) group with hydrophobic tails (EC16, O16B and O17O) and nanoparticles are formulated in the presence of nickel ions and helper lipids (cholesterol, DOPE and PEG2k-DSPE). We demonstrated that the newly-developed LNPs are capable of delivering various His-tagged proteins including GFP, (−30)GFP-Cre recombinase, and CRISPR/Cas9 ribonucleoprotein (RNP)into mammalian cells.

Graphical Abstract

Nitrilotriacetic acid-containing non-cationic lipidoids are synthesized and formulated with helper lipids to form the lipidoid nanoparticles for intracellular delivery of the His-tagged genome-editing proteins which takes advantage of the coordination interaction between divalent nickel ions immobilized lipidoids and the imidazole groups presented on the histidine residues.

1. Introduction

Proteins are vital molecules in cells and life, playing important roles in maintaining structures and functions. Deficiency or malfunction of proteins could induce serious conditions. Oftentimes, the most straightforward strategy to deal with these conditions or diseases is delivering functional proteins to specific sites.^[1] Most of the current protein-based therapeutics, including enzymes, monoclonal antibodies, act upon secreted and surface-bound targets.^[2] We believe that intracellular delivery methods that can facilitate the cell internalization of intact peptides and proteins could further exploit the potentials of protein therapeutics and expand protein-based therapeutic modalities.^[3] In this case, both mechanical/physical and biochemical-based techniques have been developed for intracellular delivery of proteins. Microinjection, electroporation and acoustic-assisted transfection are straightforward and usually efficient for protein delivery.^[4] However, their invasive features complicated the in vivo applications. Biochemical methods like supramolecular encapsulation and covalent conjugation have also been previously demonstrated to be feasible for intracellular delivery of therapeutic proteins.^[5] In addition, lipid,^[6] polymer,^[5b] metal and metal oxide nanoparticles^[7] etc., have been reported for protein delivery purposes. Our group has developed several combinatorial libraries of lipid-like molecules (lipidoid)-based nanoparticles (LNPs) for intracellular delivery applications;^[8] the effectiveness of the LNPs have been shown by us and others in both in vitro and in vivo studies.^{[8c, 9]} These cationic lipidoids contain amine head groups and the main driving-force for cargo complexation was the electrostatic interaction between LNP carriers and anionic guest molecules (negatively charged proteins, mRNA, siRNA and pDNA etc.).^{[8e, 10]} The possibility of using non-cationic lipidoids and other types of supramolecular interactions for protein loading and delivery is explored in this study.

One of the most widely-used protein purification processes takes advantage of the coordination interaction between the nickel-immobilized beads with the imidazole groups on the His-tagged proteins. Dowdy^[11] and Zuber^[12] developed nitrilotriacetic acid (NTA) moiety-containing PTDs (peptide transduction domains) peptide and synthetic polymers complexed with divalent nickel ions for His-tagged protein/nanoparticle cargoes binding and intracellular delivery. Inspired by their designs, we report here the synthesis of three types of NTA-containing lipidoids with different hydrophobic tail structures (NTA-EC16, NTA-O16B and NTA-O17O; Figure 1) for protein delivery. As a proof-of-concept, nanoparticles were fabricated using active lipidoids containing NTA groups, together with divalent nickel ions and helper lipids (cholesterol, phospholipid, and macromolecular lipids; Figure 1B) for optimized intracellular delivery efficiencies. His-tagged protein including green fluorescent protein, Cre recombinase variant and recently developed CRISPR (clustered regularly interspaced short palindromic repeat)-associated protein 9 (Cas9) nuclease were complexed with the nanoparticle formulations and their internalization efficiencies as well as genome editing activities of Cre recombinase and Cas9 RNP were studied (Figure 1A).

Figure 1.

NTA-lipidoid nanoparticles for intracellular delivery of genome-editing proteins. A) Schematics of cargo protein loading and delivery by lipidoid nanoparticle formulations. B) Chemical structures of NTA-lipidoids and helper lipids used in this study.

2. Results and Discussion

2.1. Lipidoids Synthesis and Nanoparticles Fabrication

Nitrilotriacetic acid group-containing amphiphilic lipidoids NTA-EC16, NTA-O16B and NTA-O17O (Figure 1B) were synthesized by reacting Nα,Nα-bis(carboxymethyl)-L-lysine head group with hydrophobic tails (EC16^[8c], O16B^{[8b, 8d]} and O17O^[8a], Figure S1); the products were purified using a Teledyne Isco Chromatography purification system. The structures of NTA-lipidoids were confirmed by ESI-MS and NMR analysis and the results are summarized in Table S1 and Figure S2.

As to the lipidoid nanoparticle formulations, the phospholipids like DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine) and DSPC (1,2-distearoyl-sn-glycero-3-phosphocholine) could provide support to the lipids bilayer structures and facilitate the endosomal escape, which is essential for efficient bio-active molecules delivery. Cholesterol could further stabilize the nanoparticle structures and enhance cell internalization efficacy through promoting membrane fusion process; and PEGylated lipids like PEG2k-DSPE (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000) and PEG-Ceramide, could reduce nanoparticles aggregation, non-specific adsorption, and enable longer circulation time once administrated systematically.^[13] In this study, lipidoid nanoparticles were fabricated together with nickel ions (Ni²⁺/NTA group= 1/1, molar ratio) and different amounts of helper lipids including cholesterol, DOPE and PEG2k-DSPE to optimize the intracellular delivery performance (Figure 1A). The codes and components of lipidoid nanoparticle formulations tested in this study are summarized in Table1. Briefly, nanoparticles in category A (coded as A1) consists of NTA lipidoids complexed with nickel ions (Ni-NTA lipidoids) without helper lipids (denoted as NTA-EC16-A1, etc.); category B nanoparticle formulations (coded as B1 and B2) have Ni-NTA lipidoids and cholesterol (denoted as NTA-EC16-B1, etc.); category C formulations (coded as C1, C2 and C3) have Ni-NTA lipidoids, cholesterol and DOPE (denoted as NTA-EC16-C1, etc.); category D formulations (coded as D1 and D2) have Ni-NTA lipidoids, cholesterol, DOPE and PEG2k-DSPE (denoted as NTA-EC16-D1, etc.).

Table 1.

Codes and parameters of category A, B, C and D lipidoid nanoparticle formulations used in this study

Code	Molar Composition / %
	Ni-NTA lipidoid	Cholesterol	DOPE	PEG2k-DSPE
A1	100.0	0	0	0
B1	89.3	10.7	0	0
B2	67.6	32.4	0	0
C1	73.0	8.8	18.2	0
C2	57.8	27.8	14.4	0
C3	50.5	24.3	25.2	0
D1	64.1	30.8	4.0	1.1
D2	55.7	26.7	13.9	3.7

2.2. Intracellular Delivery of GFP Protein

For the intracellular delivery study, NTA-EC16 lipidoids were formulated with nickel ions and helper lipids into nanoparticles at first. Green fluorescent protein with a 6xHis tag (His-tagged GFP; shortened as GFP in this study) was used as model cargo protein to measure the delivery efficacy of the lipidoid nanoparticles. GFP loaded lipidoid nanoparticles were first prepared by mixing the protein with nickel ions complexed lipidoid nanoparticles at room temperature (see Experimental Section). Using nanoparticle formulation NTA-EC16-B2, both the protein loading content (W_{protein loaded in nanoparticle}/(W_{protein loaded in nanoparticle} + W _nanoparticle)*100%) and protein loading efficiency (W_{protein loaded in nanoparticle}/W_{total feeding protein}*100%) were determined. As shown in Figure S3, in general, low protein/NTA-lipidoid feeding ratio resulted in low protein loading content and high loading efficiency; while high feeding ratio generated high loading content and relative low loading efficiency. For example, the GFP protein loading content and loading efficiency at protein/NTA-EC16 of 1/10 and 5/1 (weight ratios) were determined to be 7.9%/95.1% (protein loading content/protein loading efficiency) and 50.4%/23.0%, respectively. At the feeding ratio of GFP/NTA-EC16 = 1/1, the protein loading content and efficiency turned out to be 33.0% and 54.9% for the formulation of NTA-EC16-B2. SEM examination results (shown in Figure S4) revealed that both the blank and GFP protein loaded NTA-EC16-B2 complexes (NTA-EC16-B2 and GFP/NTA-EC16-B2) are spherical nanoparticles with the sizes of 150–300 nm at dry state.

All the other categories of GFP protein loaded formulations were then fabricated using he feeding ratio of GFP/NTA-lipidoid as 1/1 (weight ratio). Their averaged hydrodynamic sizes (<D_h>) and polydispersity indexes (PDIs) were determined by dynamic light scattering (DLS) measurements. As shown in Figure 2A, most (6 out of 8) of the GFP loaded nanoparticles’ sizes are between 130–300 nm, which is similar to the size range of our previously fabricated cationic combinatorial lipidoid libraries and is considered to be suitable for intracellular delivery purposes.^{[8a, 8b]} GFP loaded NTA-EC16-A1 (138.7 nm) and NTA-EC16-D2 (144.3 nm) possessed the smallest sizes while, NTA-EC16-B2 (290.5 nm) was determined as the biggest. Furthermore, except for NTA-EC16-A1 (PDI = 0.314), all other nanoparticles have PDIs smaller than 0.3, indicating the uniformity and relative homogeneity of particle sizes as well as the lack of evident aggregations. The typical hydrodynamic size distribution profiles of GFP loaded NTA-EC16-A1 (<D_h> = 138.7 nm, PDI = 0.314), NTA-EC16-B2 (<D_h> = 290.5 nm, PDI = 0.299), NTA-EC16-C2 (<D_h> = 236.4 nm, PDI = 0.267), NTA-EC16-C3 (<D_h> = 253.2 nm, PDI = 0.195) and NTA-EC16-D1 (<D_h> = 197.9 nm, PDI = 0.242) are shown in Figure 2B and S5. The stability of GFP loaded nanoparticles (NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3) stored at room temperature was measured by DLS (Figure 2C). After 24 h, all the three tested nanoparticles showed negligible size variations, less than ±7% of relative size change (−3.4% for NTA-EC16-B2, −6.4% for NTA-EC16-C2, and −3.7% for NTA-EC16-C3), compared to the sizes of freshly prepared samples which demonstrates that the nanoparticles are relatively stable under this condition. After 48 h, NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3 showed relative size change of −10.6%, −16.5% and −7.6%, respectively, which indicates the rearrangement process of the supramolecular structures of GFP protein loaded NTA-EC16-based lipidoids nanoparticles during storage under room temperature; meanwhile no evident aggregation was observed. It should be noted that in this study, freshly prepared protein loaded nanoparticles were used for the all the cell transfection applications.

Figure 2.

Intracellular delivery of GFP protein. A) Averaged hydrodynamic diameter (<D_h>) and polydispersity index (PDI) of GFP-loaded nanoparticles characterized by DLS. B) Size distribution and C) relative size change profiles of GFP-loaded nanoparticles. D) Delivery efficacy and E) cytotoxicity of GFP-loaded nanoparticles tested against HeLa cell after 8 h exposure. F) Typical flow cytometry profiles (green fluorescence histograms) of control, GFP and GFP-loaded nanoparticles treated HeLa cells. G) Intracellular delivery efficiencies of GFP-loaded nanoparticle formulations with and without Ni-NTA lipidoids after 8 h of exposure. H) HeLa cell uptake of GFP-loaded nanoparticles (NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3) was suppressed by endocytosis inhibitors (Dynasore, a dynamin II inhibitor; M-β-CD, plasma cholesterol-depleting agent; Sucrose, a clathrin-mediated endocytosis inhibitor).

Next, the GFP loaded nanoparticles were incubated with HeLa cells ([GFP] = 230 nM, [NTA-lipidoid] = 7.14 mg L⁻¹) and harvested after 8 h of exposure. The HeLa cells were then analyzed via flow cytometry to evaluate the internalization efficiencies of the nanoparticles as measured by GFP-positive (GFP⁺) cell percentages. As shown in Figure 2D, naked GFP cannot be internalized by the HeLa cells as ~1.8% of GFP⁺ cells were determined for naked GFP treated cell, which is similar to the control group (0.6% of GFP⁺ cells). In sharp contrast, NTA-EC16 lipidoid nanoparticles including NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3 are very efficient for delivering GFP protein into HeLa cells, as 93.7%, 91.9% and 96.9%, respectively, of cells were determined to be GFP⁺. The typical flow cytometry profiles (histograms of green fluorescence intensity) of the HeLa cells treated by naked GFP and GFP loaded NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3 nanoparticles are shown in Figure 2F. Meanwhile, NTA-EC16-C1, NTA-EC16-B1 and NTA-EC16-D1 also performed well, as 56.8%, 40.4% and 29.0% of the treated cells being identified as GFP⁺, respectively. Considering that NTA-EC16-A1 induced ~6.2% of GFP⁺ cells, it was established that the helper lipids play important roles in the cell transfection process, and the amount of helper lipids added are also essential for optimized performances.^[13] Typical fluorescent images of free GFP and GFP loaded NTA-EC16-B2 nanoparticles treated HeLa cells are shown in Figure S6 (BZ-X Analyzer fluorescence microscope). Free protein without carriers cannot efficiently enter into the cells, which is consistent with the flow cytometry data as shown in Figure 2D; while green fluorescence signals from GFP proteins were detected from the nanocomplexes (GFP/NTA-EC16-B2) treated cells and the cargoes seem to be distributed as bright dots over the cell after 8 h of exposure, which is similar to our previously reported results of lipidoid-based nanocarrier systems.^[8a] After the internalization, the cargo release process was expected to occur and restore the bioactivity of guest molecules. In addition to the non-specific dissociation from the nanocomplexes, the protein cargo release could also be facilitated by competitive complexation (most likely supramolecular interactions like electrostatic interaction, hydrogen-bonding and hydrophobic interactions, etc.) between the biomolecules inside the cell (lipids, carbohydrates, amino acids, peptides and proteins, etc.) and the carrier lipidoids and/or cargo proteins.^{[8a, 8b]}

Figure 2G shows that when the Ni-NTA lipidoids were removed from the formulations (NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3), the helper lipids (cholesterol and DOPE) cannot efficiently deliver His-tagged GFP protein ([GFP] = 230 nM; exposure time = 8 h) using the self-assembly and cargo loading procedures reported in this study, into HeLa cells, under the same incubation conditions. This result demonstrated the importance of Ni-NTA lipidoids-enabled protein binding capability in the intracellular delivery processes. In addition, nanoparticles with PEG2k-DSPE in category D (NTA-EC16-D1 and NTA-EC16-D2) did not perform better than lipidoid nanoparticles in category B and C. This is probably due to the steric hindrance of the hydrophilic part (i.e. PEG) of PEG2k-DSPE, and the fact that the PEG2k-DSPE in the formulation leads to lower delivery efficiency (Figure 2D; NTA-EC16-D1 and NTA-EC16-D2). Nonetheless, PEG2k-DSPE integrated lipidoid nanoparticles may have other advantages like longer blood half-life time and reduced non-specific surface absorption characteristics, especially when administered systemically in vivo, compared its counterparts. It should be noted that the delivery efficiency of macromolecular helper lipids integrated formulations could be further optimized by varying molar concentration of both NTA-lipidoid and helper lipids and/or changing the species of macromolecular lipids (hydrophilic polymer chain and/or hydrophobic tail).

The cell internalization mechanism of nanoparticle formulations was then studied, following our previously reported procedures.^[6c] As shown in Figure 2H, to NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3 nanoparticles, the dynasore (dynamin II inhibitor), M-β-CD (plasma cholesterol-depleting agent) and sucrose (clathrin-mediated endocytosis inhibitor) treated HeLa cells showed decreased GFP transfection efficiencies. Lowest transfection efficiencies were observed from M-β-CD treated cells for all three tested nanoparticles (GFP loaded NTA-EC16-B2, NTA-EC16-C2 and NTA-EC16-C3 nanoparticles). On the other hand, nystatin (a caveolin-endocytosis inhibitor) did not suppress the internalization of GFP loaded nanoparticles efficiently comparing to other three tested inhibitors. These results indicate that clathrin, plasma cholesterol and dynamin play important roles in the cellular uptake of these GFP protein complexed NTA-EC16 lipidoid nanoparticle formulations. The plasma cholesterol and caveolin-mediated pathways possessed the most and the least significant effects on the internalization process of GFP protein loaded NTA-lipidoid nanoparticles.

Previous studies by our group and others have suggested that both of the delivery efficiency and toxicity of synthetic lipidoids were largely depending on their molecular structures, i.e. head groups, linker units and tails.^{[8a-c, 10c]} We synthesized another two types of NTA-lipidoids with O16B^[8b] and O17O^[8a] tails to examine if integrating disulfide bond and chalcogen could achieve improved delivery performances. NTA-O16B and NTA-O17O lipidoids were characterized by MS and NMR, and the results are shown in Table S1 and Figure S2.

In order to make the screening process more efficient, we chose to test those formulations that have been demonstrated to be highly-active for NTA-EC16 lipidoid, with the newly synthesized NTA lipidoids (NTA-O16B and NTA-O17O). Encouraged by the results of NTA-EC16 lipidoid nanoparticles as shown in Figure 2D, NTA-O16B and NTA-O17O nanoparticles with B2, C2 and C3 formulations (Table 1) were then prepared for intracellular delivery of GFP protein. The NTA-O16B-B2 (<D_h> = 273.7 nm, PDI = 0.240), NTA-O16B-C2 (<D_h> = 284.4 nm, PDI = 0.343), NTA-O16B-C3 (<D_h> = 487.7 nm, PDI = 0.377), NTA-O17O-B2 (<D_h> = 216.8 nm, PDI = 0.233), NTA-O17O-C2 (<D_h> = 259.7 nm, PDI = 0.145) and NTA-O17O-C3 (<D_h> = 340.9 nm, PDI = 0.333) nanoparticles were characterized by DLS and results are shown in Figure 2A. Then HeLa cells were treated with formulated NTA-O16B and NTA-O17O nanoparticles loaded with GFP protein and analyzed through flow cytometry. As shown in Figure 2D, both NTA-O16B-B2 (7.6% of GFP⁺ cells) and NTA-O17O-B2 (26.4%) induced much less GFP⁺ cells comparing to NTA-EC16-B2 nanoparticles, while NTA-O16B-C2 (81.2% of GFP⁺ cells), NTA-O16B-C3 (76.4%), NTA-O17O-C2 (82.3%), and NTA-O17O-C3 (87.9%) can efficiently facilitate the internalization of GFP protein. These results reveal that the intracellular transfection efficiency of formulated lipidoid nanoparticles depends on both the species and amount of NTA-lipidoids (NTA-EC16, NTA-O176B and NTA-O17O) and helper lipids (cholesterol and DOPE) in the formulation.

Including delivery efficacy, biocompatibility is also an important issue to consider when developing nanocarriers for bio-active molecules delivery.^[14] The cell viability profiles of GFP protein loaded NTA-lipidoid nanoparticles treated cells were measured by MTT assay. HeLa cells were treated with nanoparticles under the same conditions as used in the transfection study ([GFP] = 230 nM; exposure time = 8 h) and cell viability was calculated comparing to the control groups. As shown in Figure 2E, naked GFP protein was non-toxic under the tested conditions as the cell viability was determined to be similar to the untreated control cells. In regards to the NTA-lipidoid nanoparticles treated cells, with the exception for NTA-EC16-D2 which showed a relatively lower cell viability of ~72.9%, all other groups have greater than 80% cell viabilities. The generally high percentage of cell viability indicates a good compatibility of GFP loaded NTA-lipidoid (NTA-EC16, NTA-O16B and NTA-O17O) nanoparticles against HeLa cells. We expect that top delivery carries are more likely to be identified from those formulations with relative low toxicity (i.e. high cell viability) and high delivery efficacy for further applications.

2.3. Intracellular Delivery of (−30)GFP-Cre Recombinase

Motivated by the results of intracellular transfection and biocompatibility tests of GFP loaded NTA-lipidoid nanoparticles, the possibility of using the newly developed nanoparticles to deliver functional proteins for genome editing purposes was further explored. In this context, supernegatively charged green fluorescent protein variant fused Cre recombinase with 6xHis tag (denoted as (−30)GFP-Cre) was used as the cargo and DsRed-HeLa model cells as the target cells that expresses red fluorescent protein, DsRed, only upon Cre-mediated recombination.^{[6b, 6c]} Similar to the GFP protein loading measurements (Figure S3), as shown in Figure S7, low (−30)GFP-Cre protein/NTA-lipidoid feeding ratio resulted in low protein loading content and high loading efficiency, while high feeding ratio generated high loading content and relative low loading efficiency. All the nanoparticle formulations were fabricated using 1/1 (w/w) feeding ratio of (−30)GFP-Cre protein/NTA-lipidoid. (−30)GFP-Cre protein loaded nanoparticles were then prepared and characterized by DLS measurements. As shown in Figure 3A, most of the nanoparticles have larger sizes than their counterparts that loaded with GFP protein (Figure 2A), and within the range of 150–350 nm. 10 out 14 nanoparticles showed PDI values lower than 0.3, which indicates the relative uniformity of these (−30)GFP-Cre loaded nanoparticles.

Figure 3.

Intracellular delivery of (−30)GFP-Cre protein. A) Averaged hydrodynamic diameter and polydispersity index of (−30)GFP-Cre-loaded nanoparticles characterized by DLS. B) Delivery efficacy of (−30)GFP-Cre-loaded nanoparticles tested against HeLa and DsRed-HeLa cells after 8 h of exposure. C) DsRed protein expression and D) cell viability of (−30)GFP-Cre-loaded nanoparticles treated DsRed-HeLa cells after 24 h of exposure.

As illustrated before, NTA-lipidoid nanoparticles could be readily internalized by HeLa cells; nevertheless, the delivery efficiency of (−30)GFP-Cre protein loaded formulations were first tested against HeLa cells. After 8 h of exposure ([(−30)GFP-Cre] = 109 nM, [NTA-lipidoid] = 7.14 mg L⁻¹), HeLa cells were harvested and analyzed by flow cytometry. As shown in Figure 3B, similar to GFP protein, the naked (−30)GFP-Cre protein cannot get into HeLa cells on its own. NTA-EC16-C2, NTA-EC16-C3 and NTA-EC16-C1 were identified to be most efficient for (−30)GFP-Cre protein delivery, as 96.3%, 95.5% and 69.8%, respectively, of the treated HeLa cells were GFP-positive; this result is similar to the results of GFP protein delivery shown in Figure 2D. NTA-EC16-B2 also performed well, despite, the GFP⁺ cell percentage being determined to be lower than that of its GFP protein loaded counterparts where the phenomenon was also observed for (−30)GFP-Cre loaded NTA-O16B-C2, NTA-O16B-C3, NTA-O17O-C2 and NTA-O17O-C3 treated cells. It should be noted that lower protein molar concentrations were used in (−30)GFP-Cre delivery study ([(−30)GFP-Cre] = 109 nM) comparing to the GFP protein delivery experiments ([GFP] = 230 nM; Figure 2D); the low protein concentration is due to consideration that the size of (−30)GFP-Cre recombinase is much larger than that of GFP protein. Nevertheless, these results illustrate that most of the NTA-lipidoid nanoparticles that are effective for GFP protein delivery are also capable of delivering (−30)GFP-Cre protein into HeLa cells, and that the physicochemical properties of the cargo proteins may also influence the intracellular delivery efficiencies.

Next, the (−30)GFP-Cre protein loaded NTA-lipidoid nanoparticles were tested against DsRed-HeLa cells. The cells were treated under the same conditions used for HeLa cells and the internalization efficacies were analyzed after 8 h of exposure ([(−30)GFP-Cre] = 109 nM, [NTA-lipidoid] = 7.14 mg L⁻¹). As shown in Figure 3B, naked (−30)GFP-Cre cannot be internalized by DsRed-HeLa cells which is consistent with our previous results. While both NTA-O16B-B2 and NTA-O16B-C2 induced less than 10% of GFP⁺ cells, all other nanoparticles treated samples showed around or greater than 15% of GFP⁺ cells. NTA-EC16-C2, NTA-O17O-C3 and NTA-EC16-D1 are proved to be most efficient for delivering (−30)GFP-Cre into DsRed-HeLa cells, with 43.2%, 33.0% and 30.5%, respectively, of the cells being identified as GFP⁺ after 8 h of exposure. It was determined that most of the active nanoparticles for transfecting HeLa cells are also efficient against DsRed-HeLa cells, however, the delivery efficacies are dependent on the cell models tested. Similar to the GFP delivery results shown in Figure 2G, the nanoparticle formulations without Ni-NTA lipidoids cannot efficiently deliver (−30)GFP-Cre protein into DsRed-HeLa cells (Figure S8), using the self-assembly and cargo loading procedures reported here.

Then, the gene recombination efficiency of (−30)GFP-Cre loaded nanoparticles treated DsRed-HeLa cells were evaluated after 24 h of exposure ([(−30)GFP-Cre] = 109 nM, [NTA-lipidoid] = 7.14 mg L⁻¹). The successful genetically edited cells are presented as DsRed-positive (DsRed⁺) cells in Figure 3C. It was obvious that naked (−30)GFP-Cre protein induced negligible DsRed⁺ cells as a result of its poor internalization efficacy as revealed in Figure 3B. On the other hand, NTA-EC16-C2, NTA-EC16-C3 and NTA-EC16-C1 are demonstrated to be most efficient for delivering (−30)GFP-Cre into DsRed-HeLa cells and inducing gene recombination, as 63.0%, 44.2% and 32.9%, respectively, of the cells were determined to be DsRed⁺; these values are comparable to those of our previously reported cationic lipidoid nanoparticles thus indicating the effectiveness of the newly developed delivery system.^{[6c, 8a]} Moreover, other NTA-EC16 based lipidoid nanoparticles including NTA-EC16-A1 (14.7% of DsRed⁺ cells), NTA-EC16-B1 (17.0%), NTA-EC16-B2 (19.6%), NTA-EC16-D1 (16.3%) and NTA-EC16-D2 (14.6%) also performed well, while NTA-O16B and NTA-O17O originated formulations showed lower (less than 10%) genome editing efficiencies against DsRed-HeLa cells under the same tested conditions. Upon examination of the results of internalization (8 h of exposure; Figure 3B) and genome editing (24 h of exposure; Figure 3C), it was concluded that the efficient internalization of the cargo protein loaded nanoparticles is a necessary but not the sufficient condition for successful genome editing events; this is reasonable considering that aside from the internalization process, effective cargo protection and release, nuclei localization and gene positioning, etc. are all needed for the successful final genome editing step. One other point to note is that different formulations may possess different internalization kinetics. Nonetheless, it can be expected that with further formulation optimization, the enhancement of genome editing efficiency using Cre recombinase or other types of genome-engineering proteins against a wide range of target cells could be achieved.

Finally, the cytotoxicity of (−30)GFP-Cre protein loaded lipidoid nanoparticles after 24 h of exposure was measured through the MTT assay. As shown in Figure 3D, the cargo protein (i.e. (−30)GFP-Cre) was non-toxic under the tested conditions. Except for NTA-EC16-D2 and NTA-O16B originated nanoparticles which have the cell viability of 52–64%, all other formulation nanoparticles are well-tolerated by DsRed-HeLa cells for higher than 74% of cell viabilities were determined. In general, the cell viabilities after NTA-lipidoid nanoparticles treatment are comparable to previously reported cationic lipidoids-treated cells.^[8a] However, considering that higher concentrations of both active lipidoid ([NTA-lipidoid] = 7.14 mg L⁻¹) and cargo protein ([(−30)GFP-Cre] = 109 nM) are tested here, the (−30)GFP-Cre protein loaded NTA-lipidoid nanoparticles are more compatible than our previously reported cationic ones.^[8a] Overall, comparing the above cytotoxicity test results (Figure 3D) to that of the GFP loaded nanoparticles treated HeLa cells (Figure 2E), as well as the internalization study (shown in Figure 3B) fully demonstrates that the toxicity of the protein loaded nanoparticles are reliant on the physicochemical properties of both the composition of carriers and cargos, experimental conditions that have been tested (exposure time, concentration, etc.), and the cell model that is being tested against.

2.4. Intracellular Delivery of CRISPR/Cas9 Ribonucleoprotein Complex

As the NTA-lipidoid nanoparticles were demonstrated to be efficient for facilitating the intracellular delivery of Cre recombinase and subsequent gene recombination, the nanoparticles were further challenged with the recently developed CRISPR/Cas9 genome-editing platform.^[15] For this purpose, S. pyogenes Cas9 protein with nuclear localization sequence (NLS) and 6xHis tags (denoted as Cas9) was synthesized and combined with guide RNA (gRNA) to form the RNP complexes (Cas9/gRNA = 1/1, molar ratio). The hydrodynamic sizes and PDIs of Cas9 complexed single-guide RNA (Cas9:sgRNA) targeting GFP gene loaded NTA-lipidoid nanoparticles were characterized by DLS measurements. As shown in Figure 4A, most of the nanoparticles (12 out of 14) have the sizes around 150–350 nm, which is comparable to the sizes of (−30)GFP-Cre loaded nanoparticles and are slightly larger than the sizes of GFP protein loaded ones. Similarly, as the PDI values of 12 out of 14 nanoparticles are lower than 0.3, these Cas9:sgRNA loaded nanoparticles are considered to be well-dispersed and suitable for intracellular delivery.

Figure 4.

Intracellular delivery of CRISPR/Cas9 ribonucleoprotein complex. A) Averaged hydrodynamic diameter and polydispersity index of Cas9:sgRNA-loaded nanoparticles characterized by DLS. B) Delivery efficacy of ATTO 550 labeled RNP complex (ATTO 550-RNP)-loaded nanoparticles tested against HEK cells after 8 h of exposure. C) GFP knockout efficacy and D) cell viability of Cas9:sgRNA-loaded nanoparticles treated GFP-HEK cells after 48 h of exposure.

The GFP-expressing HEK cells (GFP-HEK) were used to evaluate the genome editing efficiency of Cas9:sgRNA (targeting GFP gene) loaded nanoparticles. At first, the internalization study was conducted using HEK cells and Cas9:gRNA RNP complexed with fluorescent dye ATTO 550 labeled two-components guide RNA system (ATTO 550-tracrRNA and crRNA), which is denoted as ATTO 550-RNP. The HEK cells were exposed to ATTO 550-RNP loaded NTA-lipidoid nanoparticles for 8 h ([ATTO 550-RNP] = 88 nM, [NTA-lipidoid] = 7.14 mg L⁻¹) and analyzed by flow cytometry. As shown in Figure 4B, naked ATTO 550-RNP complex induced slightly higher ATTO 550-positive (ATTO 550⁺) cell portion compared to the control cells, which may be mainly a result of non-specific adsorption process. Enhanced cell internalization of ATTO 550-RNP was recorded in the presence of NTA-nanoparticles. NTA-EC16-C2 showed highest internalization efficacy as 29.3% of HEK cells were determined to be ATTO 550⁺. Meanwhile, NTA-EC16-A1, NTA-EC16-B2 and NTA-O17O-B2 also performed well as 15–20% of treated cell were found to be ATTO 550⁺. Around 10% of transfected cells were observed for the remaining nanoparticles in the NTA-lipidoid formulations. Then, Cas9 combined with sgRNA targeting GFP gene were complexed with the NTA-lipioid nanoparticles and tested against GFP-HEK cells. In this case, the GFP-HEK cells were treated with Cas9:sgRNA loaded nanocomplexes for 48 h ([Cas9:sgRNA] = 88 nM, [NTA-lipidoid] = 7.14 mg L⁻¹), harvested and the GFP gene knockout efficacy (presented as GFP-negative or GFP^- cell portions) was further analyzed using flow cytometry. As shown in Figure 4C, naked Cas9:sgRNA RNP complex induced negligible GFP knockout as the percentage of GFP^- cells were determined to be similar to the untreated control group, which is consistent with our previously reported results;^[8a] the results show that the relative weak ATTO 550 signals after 8 h of incubation as shown in Figure 4B is either a result of non-specific adsorption or the partially internalized complexes that were trapped and/or digested in the end. In comparison, NTA-EC16 lipidoids based nanoparticles showed some level of GFP knockout as 11–15% of GFP^- cells were recorded for these formulations. Nanoparticles that were previously demonstrated to be effective for GFP and (−30)GFP-Cre delivery, including NTA-EC16-B2, NTA-EC16-C1, NTA-EC16-C2 and NTA-EC16-C3 induced 11.8%, 13.4%, 14.4% and 12.2%, respectively, of GFP^- cells. Similar to the (−30)GFP-Cre internalization and Cre-mediated recombination results shown in Figure 3B and 3C, both NTA-O16B and NTA-O17O originated nanoparticle formulations induced very small amount to none GFP knockout. As shown in Figure 3C and 4C, the NTA-EC16 based lipidoid nanoparticles performed better than its counterparts, NTA-O16B and NTA-O17O, with the same formulations, for genome editing proteins delivery. The tail structure of lipidoid molecule has huge impact on both of the transfection efficiency and toxicity profile of the lipidoid nanoparticles, which is consistent with our previous results, in which one single atom variation may affect the performance of the nanoparticle.^{[8a, 8b]} The reason of the difference observed in CRISPR/Cas9 RNP delivery is complicated, including the capability of nanoparticles on self-organization, cargo protein encapsulation, protection, on-time release and localization etc. Moreover, it was noticed that the genome editing efficiencies of CRISPR/Cas9 RNP complexes delivered by the NTA-lipidoid nanoparticles were relatively lower than the cationic lipidoid systems we developed before.^[6c] Nevertheless, the newly developed NTA-lipidoid system demonstrates the possibility of using other types of supramolecular interactions, other than electrostatic interaction—which was involved in almost all the nanoparticle-based RNP delivery systems reported so far—to load and deliver CRISPR/Cas9 RNP complexes for genome editing.^{[6–7, 16]} It is believed that with further formulation optimization—which includes changing both of the species and molar compositions of active and helper lipids, exploring advanced self-assembly and cargo encapsulation procedures, optimizing incubation conditions etc.—the genome editing efficacy mediated by RNP complexes could be further improved.

Finally, the cytotoxicity of CRISPR/Cas9 RNP loaded nanoparticles against GFP-HEK cells after 48 h of exposure was measured. As shown in Figure 4D, most of the nanoparticles (12 out of 14) showed cell viabilities between 55–85%; the cell viability values indicate that the NTA-lipidoid nanoparticles possess acceptable biocompatibility as higher concentrations of both cargo proteins ([Cas9:sgRNA] = 88 nM) and lipidoids ([NTA-lipidoid] = 7.14 mg L⁻¹) were used in this study comparing to our previously reported cationic lipidoids-based delivery systems.^{[8a, 8b]} NTA-EC16-D2 showed the lowest cell viability after 48 h of incubation, which is also true in the cases of both GFP (Figure 2E) and (−30)GFP-Cre (Figure 3D) proteins delivery study after 8 h and 24 h, respectively. NTA-O17O-C2 and NTA-O17O-C3 are well-tolerated by GFP-HEK cells with cell viabilities comparable to cell viability of naked Cas9:sgRNA RNP complex. As most of the NTA-EC16 based nanoparticles (7 out of 8) have higher cell viabilities than the NTA-O16B series, and taking their gene knockout efficacies into account, the NTA-EC16 nanoparticles are more favorable for intracellular delivery of CRISPR/Cas9 RNP complexes. Overall, similar to the cell internalization (Figure 4B) and transfection (Figure 4C) results, the cytotoxicities (Figure 4D) of NTA-lipidoid nanoparticles are also reliant on the physicochemical properties and components of both cargo proteins and carrier lipids, cell lines that was tested against, concentrations and exposure times, etc. It should be noticed that, including the safety and efficacy perspectives of nanocarrier-based delivery systems, other important issues and potential obstacles should also be carefully considered, for example, the pre-existing adaptive immunity^[17] and the p53-mediated damage response,^[18] when developing CRISPR/Cas9 RNP platforms for genome editing.

3. Conclusion

In summary, as a proof-of concept, we reported here the synthesis of nitrilotriacetic acid-containing lipidoids, fabrication of series of nanoparticle formulations with various species and molar compositions of both active and helper lipids, and the application of the newly developed nanoparticles for intracellular delivery of (−30)GFP-Cre and CRISPR/Cas9 RNP complexes for genome editing. The GFP, (−30)GFP-Cre, and Cas9:sgRNA loaded nanoparticles were characterized by DLS measurements and the sizes of most of the loaded nanoparticles were determined to be in the range of 130–350 nm. Internalization studies using HeLa, DsRed-HeLa and HEK cells revealed that the newly-developed nanoparticles are effective for transferring GFP, (−30)GFP-Cre and AATO 550-RNP into the cells. Subsequent genome editing efficacy of Cre-mediated gene recombination and Cas9:sgRNA induced gene knockout were determined using DsRed-HeLa and GFP-HEK cell models, and NTA-EC16 originated nanoparticles out-performed their counterparts. Cytotoxicity test indicated that most of the nanoparticles formulations have acceptable biocompatibilities under the tested conditions. It was demonstrated that, the efficiencies of cell internalization, genome editing, along with cytotoxicity profiles of the proteins loaded NTA-lipidoid nanoparticles are dependent on the species as well as compositions of active and helper lipids in the formulations, the physicochemical properties of the cargo proteins, and the experimental conditions that were tested (e.g. cell models and exposure time) for cell transfection. Over all, those nanoparticle formulations that have higher genome editing efficiencies and lower cytotoxicities will be of primary focus in our future studies, in particular for in vivo genome editing applications.

4. Experimental Section

General:

Nα,Nα-Bis(carboxymethyl)-L-lysine and EC16 were purchased from Sigma-Aldrich. O16B and O17O were synthesized following our previously reported procedures.^{[8a, 8b, 8d]} All other chemicals used for lipidoids synthesis were purchased from Sigma-Aldrich and used without further purification unless otherwise noted. His-tagged proteins including green fluorescent protein (GFP), negatively supercharged Cre recombinase ((−30)GFP-Cre), and S. pyogenes CRISPR-associated protein 9 nuclease (Cas9) were produced according to previously reported protocols.^[6c] ATTO 550 labeled tracrRNA and crRNA were purchased from Integrated DNA Technologies (IDT), and single-guide RNA targeting GFP gene was synthesized as previously reported.^[6c] Cas9 and guide RNAs were mixed in PBS at 1/1 molar ratio and incubated for 30 min before use. Hydrodynamic size and polydispersity index of the nanoparticles were measured by a Zeta-PALS particle size analyzer (Brookhaven Instruments). HeLa, HEK, DsRed-HeLa and GFP-HEK cells were cultured in Dulbecco’s modified eagle’s medium (DMEM, Sigma-Aldrich) with 10% fetal bovine serum (FBS, Sigma-Aldrich) and 1% penicillin-streptomycin (Gibco). Fluorescence profiles of the cells were analyzed using a flow cytometer (BD FACS Calibur, BD Science, CA) and FlowJo software. Nanoparticles were imaged by scanning electronic microscopy (SEM), Ultra 55 field-emission SEM, Carl Zeiss AG, at an acceleration voltage of 5 kV. All SEM samples were dried under room temperature and coated with 5–10 nm thick Pt/Pd (80:20).

Synthesis of Lipidoids Synthesis:

The hydrophobic tails (EC16, O16B and O17O) were reacted with head group (Nα,Nα-Bis(carboxymethyl)-L-lysine) at a 2.5/1 molar ratio in Teflon-lined glass screw-top vials in the presence of isopropanol and triethylamine for 72 h at 70 °C. The crude products were purified by a Teledyne Isco Chromotography system and characterized by ESI-MS.

Preparation of Nanoparticles:

NTA-lipidoid was dissolved in pure ethanol and combined in a clean glass vial with precalculated amount of cholesterol (Sigma-Aldrich), DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine, Avanti Polar Lipids) and PEG2k-DSPE (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000], Avanti Polar Lipids) at different molar ratios. PBS buffer (pH = 7.2) was then added into the ethanol solution dropwise with continuous stirring. The mixture solution was transferred in to a dialysis cassette (MWCO 3.5 kDa, Slide-A-Lyzer, ThermoFisher Scientific) and dialyzed against pure water to remove the ethanol. After the dialysis, precalculated amount of nickel chloride in distilled water was added (nickel/NTA lipidoid = 1/1, molar ratio). After brief sonication, the solution was left undisturbed at room temperature for 2 h. The His-tagged protein solution was then mixed with lipidoid nanoparticles and incubated for another 2 h before use.

Protein Loading Content and Efficiency:

After incubating cargo proteins (GFP and (−30)GFP-Cre) with lipidoid nanoparticles with different feeding ratios, the solutions were centrifuged under 13400 rpm for 20 min at room temperature. The supernatant was carefully transferred into another vial. Standard curves were generated by using serial dilutions of protein solutions and the protein concentrations in the supernatant were determined through measuring the fluorescent intensity. The protein loading content was calculated as W_{protein loaded in nanoparticle}/(W_{protein loaded in nanoparticle} + W _nanoparticle)*100% and the protein loading efficiency was calculated as W_{protein loaded in nanoparticle}/W_{total feeding protein}*100%. The weight of protein loaded in nanoparticle (W_{protein loaded in nanoparticle}) = weight of total feeding protein (W_{total feeding protein}) – weight of protein in the supernatant; weight of nanoparticle (W _nanoparticle) = weight NTA-lipidoid + weight of helper lipids.

Intracellular Delivery of His-tagged Proteins:

Typically, a 48-well cell culture plate was seeded with HeLa or HEK cells at an initial concentration of 20000 cells per well dispersed in 250 uL of DMEM media and incubated for 24h before transfection. Protein loaded nanoparticles solution in 30 uL volume was then added to the plate and the plate was incubated for another 8 h, 24h or 48 h before flow cytometry analysis.

Evaluation of Cytotoxicity:

The cytotoxicity of protein loaded nanoparticles against cells were measured via the standard MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay. Typically, a 96-well plate was seeded with HeLa or HEK cells at an initial concentration of 8000 cells per well dispersed in 100 μL of DMEM media and incubated for 24 h. Protein loaded nanoparticles solution was added into the cell culture media and the plate was incubated for another 8 h, 24 h or 48 h. MTT solution (5 mg mL⁻¹ in PBS) was then added into each well and the cells were incubated for another 4 h at 37 °C. The cell culture media was then removed carefully, and 200 μL of dimethyl sulfoxide (DMSO) was added to each well. The plate was gently agitated by an orbital shaker for 10 min at room temperature and absorbance measurements were immediately read at a wavelength of 570 nm by a microplate reader (Molecular Devices, SpectraMax M2) while cell viability was calculated accordingly.

Statistical Analysis:

Data were reported as mean ± SD. Experiments were repeated at least three times. Student’s t-tests were performed to determine the significance of differences between groups. P values less than 0.05 were considered to be statistically significant.