CRISPR Spherical Nucleic Acids

INTRODUCTION

The CRISPR-Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9) system provides adaptive immunity in bacteria and has been repurposed into a revolutionary technology for editing and modulating genomes.^1-3 As a promising therapeutic agent, this system has the potential to treat a wide range of human diseases, including many forms of cancer, muscular dystrophy and Alzheimer’s disease.^4-7 However, the full potential of CRISPR-Cas systems has not yet been realized, in part due to the inability to get the requisite cargo to different tissues, cells, and compartments within cells required to broadly address and treat disease. A primary approach has been to create and deliver plasmids or RNAs that genetically encode the Cas9 protein and single guide RNA (sgRNA) to cells of interest.^8-10 Alternatively, direct injection or packaging of Cas9-sgRNA ribonucleoprotein (Cas9 RNP) complexes with heterogenous transfection agents enable cellular uptake and precise control over intracellular protein stoichiometry and editing timeframes.^11-13 This latter approach minimizes undesirable off-target editing that can occur due to uncontrollable protein expression levels and overdosing. Consequently multiple cell-permeable systems, such as those based on cationic liposomes or polymers^14-16 and inorganic nanoparticles,^{12, 17-19} have been explored with varying degrees of success due to inefficient cellular delivery as well as toxicity and immunogenicity issues. Direct covalent conjugation²⁰ or noncovalent mixtures^{11, 21-22} of cell penetrating peptides on Cas9 RNPs may offer promising alternatives; however, variable cell penetrating efficiency and proteolytic instability limit their use in many cases.⁵ Therefore, a cell-permeable, transfection agent-free, proteolytically-stable, and self-contained genome editing system capable of accessing many tissue types is still needed.

Spherical nucleic acids (SNAs) are formed by chemically modifying nanoparticles with dense layers of single or double stranded DNA or RNA and provide access to many different tissues and over 60 different cell types.^23-25 Protein spherical nucleic acids (ProSNAs),²⁴ formed similarly by the conjugation of a nucleic acid shell to a protein, have emerged as promising architectures for intracellular biosensing and therapeutics;^26-27 indeed, in many cases, the protein core retains its bioactivity²⁸ and is even more biocompatible and stable than the native protein.²⁶ If one could synthesize SNAs with Cas9 cores and load them with the genome editing nucleic acid cargo, one could create structures that potentially could expand the scope of use for CRISPR-Cas9 genome editing.

Herein, we report a Cas9 ProSNA-sgRNA ribonucleoprotein (Cas9 ProSNA RNP) platform and evaluate its genome editing capabilities in multiple cell lines (Scheme 1). Due to its high genome editing efficiency and specificity, the GeoCas9 (termed Cas9 herein) protein²⁹ was transformed into a ProSNA construct by modifying its surface with a shell of DNA (d(TGG)₁₀) to yield Cas9 ProSNAs for cellular uptake.²⁸ This sequence was chosen because it is known to facilitate cellular uptake in SNA form.^{24, 30} These structures were designed to maximize endosomal escape by modifying them with three successive N-terminus GALA peptides.^31-34 These peptides are glutamate-rich and adopt amphiphilic structures that do not interact with the cell membranes.^33-34 In acidic endosomes, the glutamate residues of GALA peptides are protonated to form hydrophobic helices that disrupt endosomal membranes. In addition, Cas9 proteins were modified at the C-terminus with nuclear localization signals (NLSs), which bind to the nuclear pore receptor to facilitate nuclear import. We explored the use of these constructs, in the absence of ancillary transfection agents, with four different cell lines and demonstrated their genome editing efficiencies to be between 32 and 47%.

RESULTS AND DISCUSSION

Recombinant Cas9 protein fused with N-terminus GALA peptides and C-terminus NLSs was expressed in Escherichia coli BL21 (DE3) (Figure S1). To prepare Cas9 ProSNAs, the Cas9 protein was first labeled with Alexa Fluor 647 fluorophore (AF647), which allows for the quantification of protein concentrations and fluorescence imaging (Scheme 1, Figures S2-3). The surface lysines were next reacted with N-hydroxy succinimide (NHS) polyethylene glycol polymers (PEG) terminated with azide groups. The terminal azides were then reacted with dibenzocyclooctyne (DBCO)-terminated DNA to yield Cas9 ProSNAs (the d(TGG)₁₀ sequence was chosen due to its formation of a G-quadruplex which facilitates receptor-mediated endocytosis),^28,30 and the excess DNA was removed from solution using size-exclusion chromatography (SEC) (Figures S4-6). The Cas9 ProSNAs eluted earlier than the Cas9-AF647-azide, suggesting that Cas9 ProSNAs have larger hydrodynamic radii, which was further confirmed by dynamic light scattering (DLS) (Figure S7). In addition, Cas9 ProSNAs were much less mobile than unmodified Cas9 protein when examined via denaturing polyacrylamide gel electrophoresis (PAGE), indicating that they have a substantially higher molecular weight (Figure S8). Based on the absorbance of DNA and AF647 on Cas9 ProSNAs at 260 nm and 650 nm, respectively, approximately 14 ± 0.5 DNA strands were covalently conjugated per protein (Figure S9). DNA loading can be increased; however, this leads to a decrease in the enzymatic activity of the Cas9 ProSNAs. The experimental spectrum of Cas9 ProSNAs, as measured by circular dichroism (CD), conformed with the calculated spectrum, indicating that negligible changes in protein structure occurred post-DNA conjugation (Figure S10). Moreover, in contrast to the rapid degradation of unmodified Cas9 proteins, Cas9 ProSNAs did not significantly degrade when incubated with trypsin (an enzyme that digests proteins) over the course of 1 h, likely due to the spatial and electrostatic barrier provided by the DNA shell (Figure S11).

SNAs have been found to rapidly enter many cell types without accompanying transfection reagents.²³ The cellular uptake was investigated using HaCaT cells (human keratinocytes cells, which can be used for the study of carcinogenesis in human cells³⁵) incubated with 20 nM Cas9 RNPs or Cas9 ProSNA RNPs, and their intracellular delivery was quantified by flow cytometry (Figure 1a and S12). The uptake of Cas9 ProSNA RNPs was significantly higher than that of Cas9 RNPs alone, even after only a 0.5-h incubation period. After 4 h, the fluorescence intensity in cells treated with Cas9 ProSNA RNPs was approximately 45-fold higher than that of cells treated with Cas9 RNPs alone. It is likely that the multivalent binding between ProSNAs and scavenger receptors on the cell surface enabled efficient uptake.³⁶ Next, the cytotoxicity of the Cas9 ProSNA RNPs was assessed in multiple mammalian cell lines owing to their importance in practical applications (i.e., HaCaT; HEK293T, human embryonic kidney 293 cells, which are used widely to produce therapeutic proteins and viruses;³⁷ hBMSCs, human bone marrow stem cells, which are very important for regenerative medicine;³⁸ Raw 264.7, murine macrophage cell line, which is commonly used to study cellular uptake and immune response³⁹), and standard Cell Counting Kit-8 (CCK-8) assays were utilized (Figure 1b). The cell viabilities were not negatively impacted after 24 h of incubation with up to 60 nM Cas9 ProSNA RNPs. The HaCaT cells were further visualized by live/dead staining, and no significant difference in cell viability was observed for the different concentrations of Cas9 ProSNA RNPs that were investigated (Figure S13).

Endosomal entrapment remains a major obstacle to cytosolic protein delivery via receptor-mediated endocytosis.¹¹ HaCaT cells were incubated with Cas9 ProSNA RNPs or Cas9 RNPs and visualized using confocal microscopy (Figure 2a and Figure S14). To obtain detailed information regarding the colocalization of Cas9 ProSNA RNPs and endosomes, Z-stack images were taken to calculate the Mander’s overlap coefficient (r). The r score decreased over time (i.e., 2 h (r = 0.73 ± 0.05), 4 h (r = 0.46 ± 0.03), and 8 h (r = 0.27 ± 0.02)), confirming the effective endosomal escape of Cas9 ProSNA RNPs induced by GALA peptides (Figure 2b). We next calculated the nuclear import efficiency as the percentages of nuclei colocalized by Cas9 ProSNA RNPs. The nuclear import efficiency increased over time (i.e., 2 h (13.5 ± 2%), 4 h (44.7 ± 2.4%), 8 h (80.2 ± 3.7%)), highlighting NLS-directed nuclear entry (Figure 2c). As expected, in the absence of DNA functionalization, we did not observe the colocalization of Cas9 RNPs in the nucleus (Figure S14). Taken together, these results showcase the efficient cellular uptake, endosomal escape, and nuclear import of Cas9 ProSNA RNPs.

Cas9 ProSNA RNP-mediated genomic indel (insertion/deletion) mutations, derived from sgRNA-guided double strand breaks and endogenous non-homologous end joining, were investigated in different cell lines and with different targets. First, GRIN2B was selected as a model target due to its significance in neurodevelopmental disorders and intellectual disabilities.⁴⁰ Specifically, Cas9 ProSNA RNPs targeting the GRIN2B gene were incubated with HaCaT cells for 3 d, and then a surveyor assay with T7 endonuclease I (T7EI) was performed to detect and quantify the indel efficiency within the target (Figure S15). The surveyor assay revealed an indel frequency of approximately 34.9%, providing further evidence of endosomal escape and subsequent nuclear delivery of Cas9 ProSNA RNPs (Figure 3a). In contrast, the assay did not detect GRIN2B cleavage when Cas9 RNPs were used likely because of their inefficient nuclear delivery (Figure S16). Next, to demonstrate their generalizability to other targets, Cas9 ProSNA RNPs targeting DNase I hypersensitive site were applied to HaCaT cells, resulting in an indel frequency of approximately 38.3% (Figure 3a). hBMSCs cells were also used, and relatively high indel efficiencies were achieved (approximately 39.5% in GRIN2B and 32.6% in DNase I). Furthermore, the gene editing efficiency of Cas9 ProSNA RNPs in difficult-to-transfect cells (RAW 264.7 macrophages) was studied. Compared to the commercial Lipofectamine CRISPRMAX^™ Cas9 RNP transfection reagent, Cas9 ProSNA RNPs exhibited a 2.4-fold increase in indel frequency (Figure 3b).

To evaluate whether the efficient gene editing activity of Cas9 ProSNA RNPs could further induce gene silencing, EGFP sgRNA was used to target the coding region of the expressed EGFP reporter in HEK293T cells (HEK293T/EGFP). After treating HEK293T/EGFP cells with the Cas9 ProSNA RNPs, the corresponding surveyor assay revealed an indel efficiency of 31.6%, indicating that the coding region was successfully targeted and resulted in a frameshift mutation (Figure 3c). The loss of EGFP fluorescence was further confirmed by flow cytometry (Figure 3d and Figure S17) and confocal microscopy (Figure S18). Importantly, the Cas9 RNPs alone did not show detectable genome editing activity in all cell lines with their respective target(s) (Figure S16).

In conclusion, a new chemical construct that can be used for genome editing and gene silencing based on Cas9 ProSNAs was synthesized and developed. Compared to Cas9 RNPs (fused with GALA peptides and NLSs), Cas9 ProSNA RNPs exhibit significantly enhanced cellular internalization, GALA peptide-induced endosomal escape, and NLS-directed nuclear delivery. Importantly, Cas9 ProSNAs are stable against protease digestion, exhibit negligible cellular toxicity, and showed an indel efficiency of up to 47%. In principle, this design can be generalized to other (bio)macromolecular cargos to manipulate cellular functions. The simple and versatile structure of Cas9 ProSNAs provides a direct platform for gene therapy and may inform the future application of a limitless library of protein-based therapies.

Supplementary Material