Delivering Gene-Editing Tools and Long DNA into Primary Blood Cells Using Gold Nanoparticles Without Viruses

We first attempted to add a gene-length HDT to our third-generation CRISPR-AuNP via electrostatics (Figure 1A). We compared several concentrations of HDT per CRISPR-AuNP and assessed nanoparticle characteristics by dynamic light scattering (DLS). While each of the dsDNA titrations chosen resulted in acceptably sized nanoparticles (i.e., <200 nm hydrodynamic diameter), none demonstrated clinically acceptable polydispersity index (PDI) values of less than 0.3 [19] (Figure 1B). High PDI values suggest that AuNP core surface charges remain unsatisfied, resulting in aggregation of particles to achieve stability.

We previously demonstrated that binding of gRNA could satisfy AuNP core surface charges [16] in our first-generation CRISPR-AUNP. We hypothesized that formation of an inner layer of HDT covering the AuNP surface would permit a polyplex layer to bind and form covalent linkages to the gold core, trapping any loose DNA in the polymer with charge left unsatisfied by RNP.

A 2.1-kb dsDNA HDT encoding an enhanced green fluorescent protein (GFP) transgene flanked by homology arms to the human β−2-microglobulin locus (B2M) was synthesized by polymerase chain reaction (PCR) using a forward primer that contained a 12-carbon oligoethylene glycol (OEG) spacer with a 5′ terminal thiol. This 5′-SH-OEG-HDT was used to coat the gold core at different ratios of HDT molecules per individual AuNP (Figure 1C). We found that while higher ratios of HDT to AuNP resulted in monostable particles, the hydrodynamic diameters of these particles exceeded 200 nm, which is expected to reduce cellular entry [20, 21] (Figure 1D).

Since the stabilizing gRNA in our first-generation CRISPR-AuNP was short (~23 nt), we tried stabilizing the AuNP surface with a combination of short SH-OEG-ssDNA (25 nt), in addition 5′-SH-OEG-HDT (Figure 1E). This mixed-size DNA monolayering resulted in monostable nanoparticles with four HDT molecules and at least 200 ssDNA molecules per AuNP core (Figure 1F). The ratio of 4HDT/200ssDNA was chosen for further studies in order to maximize HDT loading.

DLS analysis of nanoparticles formulated with 4 HDT and 200 ssDNA per core revealed monostability defined as mean hydrodynamic diameter of 97.2 ± 18 nm, PDI of 0.245 ± 0.018, and zeta potential of −18.1 ± 0.31 mV (Figure 2A).

We next evaluated stability after centrifugation to remove excipient cargo. The optimized nanoformulation maintained monostability after two rounds of spin purification, with a diameter increase to 120.5 ± 14.72 nm, PDI of 0.287 ± 0.046, and zeta potential of −20.71 ± 0.29 mV (Figure 2A). Transmission electron microscopy (TEM) images of HDT-CRISPR-AuNP stained with uranyl acetate show distinct attachment of long dsDNA to gold cores at the HDT-AuNP stage. Following complete assembly, we can also visualize polyplex corona and excess polymer in the unpurified stage, and a combination of long dsDNA and polyplex attachment to the gold core in the purified stage (Figure 2B).

For unpurified nanoparticles, Cas9 nuclease loading was determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) to be 47.12 ± 9.93 pmol Cas9 and HDT loading was determined via agarose gel electrophoresis to be 105.1 ± 10.86 ng (81.84 ± 8.456 fmol) HDT per 20 μg AuNP dose (Figure 2C). Despite the increased hydrodynamic diameter observed following centrifugation, purified particles demonstrated 16.84 ± 7.48 pmol RNP with 75.58 ± 15.46 ng (58.85 ± 12.04 fmol) HDT. HDT-AuNP alone carried nearly 400 ng HDT indicating either significant displacement of the HDT from the gold surface by competing thiolated moieties or sequestration of the HDT in the polymer matrix rendering it unavailable for charge-based gel imaging (Figure 2C). The molar ratio of Cas9 RNP per HDT molecule decreased more than twofold after centrifuge purification, with unpurified HDT-CRISPR-AuNP releasing more than 575 RNP per each HDT, and purified nanoparticles releasing slightly over 286 RNP per HDT.

To establish base efficacy of HDT-CRISPR-AuNP, unpurified HDT-CRISPR-AuNP carrying a GFP reporter transgene regulated by a minimal cytomegalovirus (CMV) promoter with homology arms to the B2M target locus cut site (Figure 3E) was tested in Jurkat cells. Jurkat cells are a human T cell line derived from acute T cell leukemia and are commonly used to evaluate CRISPR delivery prior to experiments in primary human T cells [22]. We and others have shown that CRISPR activity at the B2M locus, particularly in exon 2, results in mutations which in turn cause protein misfolding and loss of expression of B2M at the cell surface [18, 23]. A panel of B2M gRNAs were electroporated into cell lines and assessed for protein knockdown. The gRNA that resulted in the highest degree of B2M surface expression loss as measured by flow cytometry was carried forward for additional studies [18, 24]. We treated Jurkat cells with HDT-CRISPR-AuNP targeting B2M at a dose of 20 μg AuNP per million cells and analyzed resulting loss of B2M surface protein (B2Mlow/−) and gain of GFP protein expression (GFP+) by flow cytometry after 72 h in culture. Flow analysis showed significant B2M protein knockdown in cells treated with both CRISPR-AuNP and HDT-CRISPR-AuNP at 0.2% and 0.4% of live cells, respectively (Figure 3A), as well as statistically significant GFP transgene expression from both HDT-polyplex-treated and HDT-CRISPR-AuNP-treated cells at 0.012% and 0.013% of live cells, respectively (Figure 3B). Sequencing analysis revealed insertion and deletion (indel) mutations at the target B2M locus at 4.7% for cells treated with HDT-CRISPR-AuNP (Figure 3C), though not to a statistically significant extent. Importantly, nanoparticle treatment had no discernable effect on cell viability in this assay (Figure 3D).

Following successful editing and transgene expression in Jurkat cells, the same B2M exon 2-targeted nanoparticle carrying a CMV-GFP reporter transgene was tested in primary human naïve T cells (CD3 + CD45RA + CD45RO-CD197+). Here, we tested both unpurified and purified HDT-CRISPR-AuNP to determine whether editing was a result of free polyplex in solution rather than assembled HDT-CRISPR-AuNP. Flow cytometry demonstrated evident, but not statistically significant B2M knockdown and GFP expression (Figure 4A,B), while sequencing results indicated statistically significant indel levels of 6.23 ± 1.90% and 6.49 ± 3.20% in cells treated with both unpurified and purified HDT-CRISPR-AuNP, respectively (Figure 4C). Cell viability was lower across all sample sets than observed in Jurkats (compare Figure 4D with Figure 3D), with purified, but not unpurified HDT-CRISPR-AuNP showing statistically significant differences in viability between nanoparticle-treated samples (49% for unpurified and 39.8% for purified) and mock-treated samples (59%) (Figure 4D).

We next sought to distinguish transgene expression in Jurkat and naïve T cells resulting from translation of exogenous double-stranded template DNA delivered into the cytoplasm from endogenously integrated HDT (the intended gene editing outcome). To this end, we formulated HDT-CRISPR-AuNP with a gRNA targeting the T cell receptor alpha constant region (TRAC) and an HDT encoding an antigen-specific TCR for the New York esophageal squamous cell carcinoma 1 (NY-ESO-1) neoantigen [14]. This gRNA and transgene combination was pioneered by Roth et al. and shown to be result in effective HDR-mediated template insertion when electroporated into primary T cells [14]. The transgene encoded in this HDT sequence consists of the full-length beta (β) chain and variable alpha (α) chain of the TCR, with no constitutive promoter included in the construct. Expression of the transgenic TCR will occur only when it is inserted correctly downstream of the endogenous TCR-α promoter [14]. HDT-CRISPR-AuNP synthesized with these novel elements were stable and monodisperse (Figure S2A).

Having established that HDT-CRISPR-AuNP can be synthesized with modular HDT and gRNA cargo, we sought to determine where the nanoparticles trafficked in primary cells. Tracing cell uptake and cytosolic distribution lends us a better understanding of the mechanisms by which nanoparticles are endocytosed by the cells and therefore lets us know whether we are achieving cytosolic delivery or if any observed editing is the result of low-level leakage of gene editing cargo from intact endosomes. We synthesized HDT-CRISPR-AuNP with a fluorescent ATTO550 tag conjugated to the tracrRNA portion of the duplexed gRNA which colocalizes with nuclease in the RNP complex. Fluorescently tagged HDT-CRISPR-AuNP maintained stability as measured by DLS (Figure S2B). Confocal microscopy demonstrated diffuse ATTO550 fluorescence throughout the cell cytoplasm and nucleus, indicating that at least tracrRNA cargo are capable of cellular entry, broad cytosolic distribution, and nuclear entry (Figure 5), though treatment with purified nanoparticles results in lower fluorescence in the cells imaged in line with results from protein-loading characterization assays (Figure 2C).

Non-fluorescently labeled HDT-CRISPR-AuNP was used to treat primary human T cells in culture. Following treatment with 20 μg AuNP equivalent of unpurified or purified HDT-CRISPR-AuNP, pan CD3+ T cells showed some knockout of endogenous TCR by both flow cytometry and next-generation sequencing (Figure 6A,C). However, dextramer binding to NY-ESO-1 TCR on the cell surface was not observed above background for any treated sample (Figure 6B). Notably, low dextramer binding was also observed in electroporated positive controls.

In the absence of confirmation of HDR by protein expression, we instead designed a PCR assay wherein the forward primer bound in the TRAC locus upstream of the 5′ homology arm of the HDT and the reverse primer bound within the variable beta chain sequence of the TCR transgene. PCR on a genomic DNA (gDNA) sample that contains the integrated insert should therefore yield 645bp product, while gDNA with no integrated HDT should yield no product (Figure 6E). Additionally, this assay cannot amplify exogenous HDT. PCR products from cells electroporated with RNP and HDT, as well as cells treated with unpurified HDT-CRISPR-AuNP, showed faint bands of the expected length (Figure 6F). Sanger sequencing of bands demonstrated upward of 95% homology to the target 645bp band (Figure S3), indicating successful, low level, integration of the transgene into the target locus. No other sample set, including RNP + HDT polyplex-treated cells and purified HDT-CRISPR-AuNP-treated cells, demonstrated visible PCR products in this assay (Figure S3).

TRAC-targeted HDT-CRISPR-AuNP carrying the NY-ESO-1 TCR construct were also used to treat primary human CD34 + HSPC in vitro. HSPCs do not express TCR, but several studies show that HSPCs edited with transgenic chimeric antigen receptors (CAR) express these transgenes after differentiation into a T cell lineage [25]. Endogenous TCRα promoter is inactive in HSPC, meaning no transgene expression would be expected on the surface of CD34+ cells even if HDR occurred. As such, gDNA was extracted from each sample and subjected to gDNA sequencing as well as the integration PCR assay we developed described above. Sequencing showed up to 5.08% indels at the TRAC locus for HDT-CRISPR-AuNP but fell short of statistical significance (Figure 7A).

Cell viability remained high regardless of treatment, with no statistical difference between mock-treated cells and any treated sample (Figure 7B). Integration PCR showed bands at the expected length for all four of the electroporated RNP + HDT samples and for two of the HDT-CRISPR-AuNP-treated samples (Figure 7C), with no bands for any of the other sample sets (Figure S3). Altogether these data indicate that HDT-CRISPR-AuNP is active in both primary human T cells and HSPC.

All PCR primers and gRNAs were purchased from IDT (Coralville, Iowa, the United States). CMV-GFP reporter plasmids were synthesized by and purchased from GenScript USA (Piscataway, New Jersey, the United States). NY-ESO-1 TCR plasmid was a gift from Alexander Marson (Addgene plasmid # 207 482; http://n2t.net/addgene:207482↗; RRID:Addgene_20748). SpCas9 nuclease was purchased from Aldevron (Fargo, North Dakota, the United States). Chloroauric acid, sodium citrate dibasic trihydrate, hydrochloric acid, and β-mercaptoethanol (BME) were purchased from Sigma-Aldrich (St. Louis, Missouri, the United States). Neon Electroporation System 100 μL Kit (Invitrogen, Waltham, MA, the United States) was used on the Neon Transfection System (Invitrogen) for all electroporations. Flow cytometry antibodies were purchased from Biolegend (San Diego, California, the United States). MHC dextramer was purchased from Immudex (Copenhagen, Denmark). PEI–PEG copolymers were purchased from Nanosoft Polymers (Winston-Salem, NC, the United States). Other materials not listed were purchased through ThermoFisher Scientific (Waltham, Massachusetts, the United States).

Nanoparticle synthesis involves less than 2 h of hands-on time. AuNP of 15–20 nm in diameter is first synthesized using the Turkevich method [43] and then characterized via TEM (TALOS120C, ThermoFisher Scientific) and visible–ultraviolet (UV–Vis) absorbance at 520 nm wavelength (Nanodrop One, ThermoFisher Scientific) [16] to accurately estimate nanoparticle molarity for binding cargo. AuNP molarity was calculated using 1 cm Beer–Lambert law absorbance at 520 nm as previously described by Lane et, al [16]. AuNP is then incubated with thiol-oligoethylene glycol-modified HDT encoding a transgene construct with 300-bp homology arms (Table S1). HDT is added at a 4:1 molar ratio and a similarly modified 25 nt ssDNA at a 200:1 molar ratio in the presence of 25 mм citrate buffer. The short ssDNA coats the remainder of the gold surface and prevents aggregation of gold particles in the presence of citrate buffer. For the sake of replicability, different batches of AuNP were diluted to have equivalent molarities such that the amount of HDT loaded into a 20-μg dose of AuNP was consistently 4 μg dsDNA. After incubation with thiol-modified DNA, HDT-AuNP is washed twice by centrifugation at 15,000RCF at 4°C for 45 min at to remove unbound DNA. Meanwhile, 100 pmol Cas9 RNP is formed by combining Cas9 protein and thiol-modified gRNA at a 1:1 ratio. This thiolated RNP is combined with 10 μg 2000 W-graft-20 PEI–PEG thiolated polymer to form an RNP polyplex, by incubating at room temperature for 1 h prior to HDT-AuNP addition and vigorous trituration for 10 s. Final formulations were stored at 4°C for up to 3 days.

Nanoparticle solutions were suspended in ultrapure water (UPW; Invitrogen) and analyzed in using a Zetasizer Nano ZS (Malvern Panalytical, Worcestershire, the United Kingdom). At least 5 μg gold solution was diluted into at least 100 μL UPW and loaded into 40 μL disposable cuvettes (BrandTech Scientific, Connecticut, the United States) for size and PDI measurement. Cuvettes were allowed to equilibrate for 1 min prior to reading with three averaged runs of 10 sub-runs each. For zeta potential measurement, nanoparticle solutions were diluted in 800 μL UPW and loaded into folded-capillary cuvettes (Malvern Panalytical). Charge measurements were taken using autosense. Measurements for each sample were taken in triplicate at 20°C.

Fully formulated HDT-CRISPR-AuNP and HDT-AuNP were stained with 1% uranyl acetate as a negative stain to show protein, polymer, and DNA in high contrast. Grid-mounted nanoparticles were imaged using TALOS120C. For diameter calculations, TEM images were analyzed in ImageJ (Version: 1.54f) by measuring diameter on two axes for at least 20 AuNP.

A 2.5-μg CRISPR-AuNP solution based on gold core mass was combined with 4X lithium dodecyl sulfate (LDS) loading buffer (Invitrogen, Waltham, MA, the United States) according to the manufacturer’s instructions and incubated for 15 min at 95°C. Each sample was loaded into a NuPAGE 4%–12% Bis-Tris Mini Protein Gel (Invitrogen) and run at 200 mV for 30 min.

Gels were stained using SimplyBlue SafeStain (Invitrogen) imaged using iBright FL1500 Imaging System (Invitrogen) and analyzed for density using the iBright built-in gel analysis platform. Protein bands were selected and densities used to establish a standard curve for protein mass. Nanoparticle protein content was interpolated from the standard curve and calculated for a 20-μg CRISPR-AuNP dose.

An 8-μg CRISPR-AuNP solution was incubated overnight in 5mм BME to cause release of surface-bound thiolated nanoparticle cargo. Released nanoparticle solution was combined with 10X BlueJuice gel loading buffer (ThermoFisher) and loaded into an agarose gel for electrophoresis. Samples were run for 45 min at 100 mV. Bands were analyzed using iBright and a standard curve established for band density. This was used to interpolate nanoparticle DNA content and calculated for a 20 μg dose.

Jurkat cells (American Type Culture Collection, Manassas, VA, the United States) were thawed and plated in at 0.5 million cells per mL in Roswell Park Memorial Institute 1640 medium (RPMI, ThermoFisher) with 10% fetal bovine serum (FBS, ThermoFisher) and 1% penicillin streptomycin (ThermoFisher). After two passages in culture, cells were replated in serum-free, antibiotic-free RPMI, and serum-starved for 2 h. CRISPR-AuNP or polyplexes were added directly to cell media and incubated for 4 h before serum addition to media. Electroporated samples were electroporated using a Neon Transfection system (Invitrogen) according to manufacturer’s settings for Jurkats (1700 mV, 1 pulse, and 20 ms). Electroporated samples were treated with maximum possible RNP dose (100 pmol RNP) and HDT dose (400 ng HDT) and immediately placed into serum-containing RPMI. Cells were harvested for flow cytometry and gene editing analysis 72 h after nanoparticle treatment.

Primary peripheral blood naïve T cells and primary peripheral blood pan T cells were purchased from STEMCELL Technologies (Vancouver, Canada). Per STEMCELL, each donor’s written informed consent was obtained via IRB-approved consent forms and protocols. For each cell type, four unique donors were sourced, with half female and half male. Cells were thawed and placed into RPMI supplemented with 5 mм BME, 10% FBS (for naïve T cells) or 10% Human AB serum (Sigma-Aldrich) for pan T cells, and 1% penicillin streptomycin, 200 U mL⁻¹ IL-2, 5 ng mL⁻¹ IL-7, and 5 ng mL⁻¹ IL-15 (ThermoFisher). Cells were harvested 72 h after nanoparticle treatment.

Peripheral blood G-CSF-mobilized CD34+ blood stem cells were purchased from Core Center of Excellence in Hematology at the Fred Hutch following the protocol approved by the Fred Hutch Institutional Review Board (IRB) (protocol no. 985.03) and in accordance with the Declaration of Helsinki and the Belmont Report. Informed written consent was obtained from all human cell donors. Four unique cell donors were used: half female and half male. Cells were thawed and placed into Serum-Free Expansion Medium version II (SFEM II; STEMCELL Technologies) supplemented with 100 ng mL⁻¹ human recombinant stem cell factor, Flt-3L, and thrombopoietin. Cells were rested overnight before being replated into non-supplemented Iscove’s Modified Dulbecco’s Medium (IMDM, ThermoFisher) for 2 h prior to treatment. Electroporated samples were electroporated according to the manufacturer’s instructions (1600 mV, 10 ms, and 1 pulse) and placed immediately into serum-supplemented stem cell media. Nanoparticle and polyplex samples were treated after 2 h of serum starvation and allowed to incubate for an additional 4 h. At 4 h, cells were supplemented with complete media and allowed to incubate a further 72 h before harvest for gene editing analysis.

Cell viability was measured using a Countess Automated Cell Counter (Invitrogen). Cell samples were combined 1:1 with Trypan Blue (Invitrogen) prior to automated counting per manufacturer’s instructions.

HDT-CRISPR-AuNP and HDT-polyplex were synthesized using Cas9 RNP formulated using ATTO550-labeled tracrRNA (IDT). Primary pan T cells were treated with nanoparticles or polyplexes as described previously. Cells were incubated with fluorescently labeled treatments for 4 h and then stained with CellMask Far Red membrane dye (ThermoFisher) and NucBlue Live Cell nuclear stain (Hoechst 33 342, ThermoFisher) according to manufacturer instructions. Cells were resuspended in 50 μL 1% FBS in 1X phosphate buffered saline (PBS, ThermoFisher) and imaged on a Zeiss LSM 780 confocal microscope (Carl Zeiss Microscopy, Oberkochen, Germany). Images were analyzed with Zen Blue (version: 3.1, ZEISS).

Antibody-fluorophore conjugates used in each flow cytometry experiment are detailed in Table S2. For each experiment, cells to be stained and analyzed were washed twice in 3 mL PBS to remove excess polymer and decrease the risk of antibody sequestration [44]. Washed cells were stained with Zombie Near Infrared Live/Dead stain (BioLegend) at a 1:500 dilution for 15 minutes. LiveDead stain was washed by centrifugation. For primary naïve and pan T cells, human FcR block (Miltenyi Biotec, Bergisch Gladbach, Germany) was incubated with cells according to manufacturer’ instructions for 15 min. Cells were washed and resuspended in FACS buffer (2% FBS in PBS) with the antibody cocktail detailed in Table SX for 45 min prior to final wash and acquisition. Cells were acquired on a BD FACSCelesta flow cytometer (BD Biosciences, Franklin Lakes, NJ, the United States).

Genomic DNA (gDNA) was extracted from cell samples using the Invitrogen Purelink Genomic DNA Isolation kit (ThermoFisher) according to manufacturer’s instructions. For MiSeq analysis of target cut sites, gDNA was amplified using MiSeq adapter primers designed in house for either the B2M or TRAC locus (Table S1). Amplicons were indexed and multiplexed using Nextera XT v2 indices (Illumina, San Diego, CA, the United States) and combined into sequencing libraries. Libraries were sequenced using paired end MiSeq. Resulting reads were analyzed for total gene editing and insertion/deletion mutations using an established in house analysis pipeline [16].

PCR primers were designed such that the forward primer bound ~150bp upstream of the TRAC transgene HDT construct and the reverse primer within the variable beta chain of the transgenic TCR[cite] such that any product produced would have a size of 645bp. A 20-ng gDNA per sample was used in each PCR reaction with NEB Q5 HotStart 2x Mastermix (New England BioLabs, Ipswich, MA, the United States) according to manufacturer’s protocol. gDNA samples subjected to this PCR that yielded the expected band were subject to Sanger sequencing using the same forward primer to confirm band identity.

All statistical analyses were performed on raw data as Brown–Forsythe and Welch ANOVA tests. Analyses of nanoparticle characteristics (Figures 1, 2, S1, and S2) used n = 3 independently synthesized samples. Analysis of nanoparticle-treated cell lines (Figure 3) used n = 3 technical replicates per sample. Analysis of nanoparticle-treated primary T cells and HSPC (Figures 4–7) used n = 4 biological replicates from different donors. Graphpad Prism (10.0.4; Boston, MA, the United States) was used to present all results as the mean ± SEM. Statistical significance was defined as p ≤ 0.05.