Structure-Dependent Biodistribution of Liposomal Spherical Nucleic Acids

Abstract

Spherical nucleic acids (SNAs) are a class of nanomaterials with a structure defined by a radial distribution of densely packed, short DNA or RNA sequences around a nanoparticle core. This structure allows SNAs to rapidly enter mammalian cells, protects the displayed oligonucleotides from nuclease degradation, and enables co-delivery of other drug cargoes. Here, we investigate the biodistribution of liposomal spherical nucleic acid (LSNA) conjugates, SNA architectures formed from liposome templates and DNA modified with hydrophobic end groups (tails). We compared linear DNA with two types of LSNAs that differ only by the affinity of the modified DNA sequence for the liposome template. We use single-stranded DNA (ssDNA) terminated with either a low-affinity cholesterol tail (CHOL-LSNA) or a high-affinity diacylglycerol lipid tail (DPPE-LSNA). Both LSNA formulations, independent of DNA conjugation, reduce the inflammatory cytokine response to intravenously administered DNA. The difference in the affinity for the liposome template significantly affects DNA biodistribution. DNA from CHOL-LSNAs accumulates in greater amounts in the lungs than DNA from DPPE-LSNAs. In contrast, DNA from DPPE-LSNAs exhibits greater accumulation in the kidneys. Flow cytometry and fluorescence microscopy of tissue sections indicate that different cell populations—immune and nonimmune—sequester the DNA depending upon the chemical makeup of the LSNA. Taken together, these data suggest that the chemical structure of the LSNAs represents an opportunity to direct the of nucleic acids to major tissues outside of the liver.

Graphical Abstract

Nucleic acid therapeutics have tremendous potential, but their widespread use has been limited largely due to challenges with effective delivery. Delivery of unmodified, linear oligonucleotides results in rapid clearance, nuclease-mediated degradation, and poor internalization by cells.^1–3 Spherical nucleic acids (SNAs) are a class of nucleic acids composed of a dense shell of radially oriented oligonucleotides surrounding a nanoparticle core. This architecture allows SNAs to overcome many of the limitations associated with delivery of linear oligonucleotides. SNA architectures rapidly enter over 50 different cell types,^4,5 resist nuclease degradation,^6,7 and transcytose across different biological barriers, including the skin,⁸ blood–brain barrier, and blood–tumor barrier.^9,10

There are a number of factors, such as nanoparticle size, shape, and surface charge, that affect the bioavailability of systemically administered nanomedicines.^11–13 This includes their interaction with serum proteins, mechanism of cellular entry, and clearance from the body.¹⁴ Previously, we determined that gold-based SNAs (Au SNAs) primarily distribute to the liver and spleen with minor changes due to varying the presented DNA sequence or backfilling the SNA surface with PEG.^9,15 This follows the pattern of many nanoparticles,^16–23 which are cleared by the cells of the mononuclear phagocyte system (MPS) located primarily in organs such as the liver, spleen, and bone marrow.^24,25 Because the Au SNAs are not extensively used clinically,²⁶ we sought to explore the more modular and clinically relevant liposomal SNA (LSNA)^27–29 in order to exploit structural changes to direct DNA biodistribution.

The highly modular LSNA architecture enables modification of both the nanoparticle core and surface chemistry. With LSNA architectures, the affinity of the DNA shell to the liposome template can be modified to control overall nanostructure stability and the release rate of oligonucleotides from the liposome core. For example, increasing the hydrophobicity of the 3′ tail of the DNA sequence by changing it from a cholesterol group to a C16 diacyl lipid anchor (DPPE)^30–32 increases the affinity of the DNA shell for the liposome template. In serum-containing media, this modification increases the half-life of DNA attachment to the LSNA’s lipid bilayer by greater than 20-fold.³³ This increased stability leads to greater cellular uptake and potency with respect to innate immune receptor stimulation.^33–35 These observations highlight how LSNA stability may dictate interactions with immune cell populations In Vivo as well as the tissues and cell populations to which the LSNAs distribute. To determine these In Vivo structure–function relationships, we synthesized LSNAs with either cholesterol-or lipid-anchored DNA and measured the immune response, tissue distribution, and cellular level distribution of each LSNA construct in immune-competent mice. The results highlight the advantages of LSNA architectures over linear DNA and the importance of LSNA stability in tuning the delivery of systemically administered oligonucleotides to target difficult-to-reach tissues.

RESULTS AND DISCUSSION

Synthesis and Characterization of Dual Fluorophore Labeled LSNAs.

The biodistribution profiles of intravenously injected LSNAs were studied in healthy, wild-type C57Bl/6 mice using a cyanine 5 (Cy5) fluorophore-labeled, phosphorothioate backbone single-stranded DNA (ssDNA) sequence. Liposomes onto which the DNA was functionalized were labeled with 10% tetramethylrhodamine (TAMRA)-PC (Tables S1, S2). The DNA used to synthesize cholesterol-tail LSNAs (CHOL-LSNAs) was terminated with cholesterol, while the DNA used to synthesize DPPE-tail LSNAs (DPPE-LSNAs) was terminated with a dibenzocyclooctyne-modified thymidine nucleobase (DBCO-dT) (Figure 1A). The sequence used was ODN 2138, a sequence designed as a GpC nonimmunogenic control to the CpG-containing TLR9 agonist ODN 1826.^36,37 LSNAs were synthesized by first forming a 50 nm diameter small unilamellar vesicle (SUV) template composed of 100% 1,2-dioleyl-sn-glycero-3-phosphocholine (DOPC). For DPPE LSNAs, DOPC SUVs were modified with 5% (mol/mol) azide-capped 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (DPPE-Azide) (Figure 1B). Following SUV formation, DNA sequences and liposome templates were mixed and shaken overnight at room temperature in 20 mM HEPES-buffered saline. This facilitated cholesterol-terminated DNA insertion into the SUV bilayer, forming CHOL-LSNAs, as well as the copper-free click reaction of DBCO-terminated DNA with DPPE-Azide lipids on the SUV surface, forming DPPE-LSNAs. The liposome size and spherical architecture were confirmed using dynamic light scattering (DLS) and transmission electron microscopy (TEM) (Figures S1, S2). Using UV–vis spectroscopy to measure DNA concentration and inductively coupled plasma optical emission spectrometry (ICP-OES) to quantify phosphorus content, we determined the number of DNA strands per LSNA. The mixing ratio of cholesterol-tailed DNA to lipids that results in the maximum number of DNA strands per liposome was determined to be approximately 15.4 μM DNA to 1 mM lipids (Figure S3). The DNA loading into each respective formulation is comparable at this reaction stoichiometry, as the average number of strands per liposome was 123 ± 28 for the CHOL-LSNA and 96 ± 18 for the DPPE-LSNA. As a control, we used a mixture (Mix) of linear DNA with no hydrophobic tail (Tables S1, S2) with the same number of liposomes, such that the Mix contains the same ssDNA sequence, but cannot form LSNAs.

Figure 1.

Oligonucleotide components used to synthesize each LSNA. (A) Structures of 3′ DNA tails that anchor DNA to each liposome template. (B) Each DNA sequence is reacted with either 50 nm liposomes composed of (top) 100% DOPC or (bottom) 95% DOPC/5% DPPE-Azide (mol/mol) to form each respective LSNA.

LSNAs Elicit a Reduced Cytokine Response Compared to Equivalent Linear DNA Sequences.

Systemically administered linear oligonucleotides often lead to off-target effects, such as nonspecific cytokine production and stimulation of inflammatory pathways.^38–40 To quantify the difference between linear DNA and LSNA structures in this context, we measured the production of various cytokines in the serum after intravenous administration of LSNAs into C57/Bl6 mice. At 30 min post-injection, linear DNA increased production of the pro-inflammatory cytokine MCP-1 by 2.46-fold over CHOL-LSNAs and 1.80-fold over DPPE-LSNAs. Linear DNA also induced IFNγ production when it was not detected in LSNA-treated mice (Figure 2A). This suggested a more severe acute inflammatory response to linear DNA than for the equivalent dose of LSNAs. IL-6 and TNF production were also increased in linear DNA treated mice compared to LSNA formulations, but changes were not statistically significant. Enhanced production of anti-inflammatory cytokine IL-10 was also observed in response to linear DNA compared to untreated mice (Figure 2A).

Figure 2.

Effect of linear oligonucleotides and LSNAs on cytokine production. Cytokines were measured following intravenous administration of linear DNA, CHOL-LSNAs, or DPPE-LSNAs at (A) 30 min and (B) 24 h post-injection in C57/Bl6 mice (ND = not detectable; statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test; *P < 0.05, **P < 0.01, ***P < 0.005, error bars represent standard error, N = 3).

By 24 h, most cytokines produced in response to linear DNA and LSNAs returned to basal levels, with IL-6 and TNF not detected. The only exception was MCP-1, which showed elevated levels due to linear DNA, 2.39-fold enhancement over CHOL-LSNAs and 2.71-fold over DPPE-LSNAs (Figure 2B). These changes in cytokine levels suggest that the LSNA architectures studied herein are inherently less inflammatory than linear DNA In Vivo, an observation we previously described in vitro.⁴¹ Because the inflammatory cytokines MCP-1 and IFNγ are known to recruit and activate cells of the MPS,^42,43 the difference in production of these cytokines in response to each form of DNA (linear or LSNA) is likely linked to differences in trafficking and sequestration of each of these architectures after injection.

Organ Level Distribution of LSNAs Is Structure Dependent.

To evaluate the tissue level distribution of administered DNA, mice received a peripheral intravenous (iv) injection of either LSNA structure or a control mixture (Mix) containing the same amount of Cy5-DNA- and TAMRA-PC-labeled liposomes (Table S2). After 30 min or 24 h of circulation, organs were recovered and analyzed. Using the spectral unmixing function on the IVIS instrument, fluorescence of the 10% TAMRA-PC liposomes was separated from that of the Cy5-tagged DNA. When comparing all organs at 30 min, the greatest fluorescence from both DNA and the liposome cores of LSNAs came from the liver, kidneys, and spleen (Figures 3A, S4–S6). However, when each organ was imaged separately, more significant differences were apparent. For most organs examined, except the small intestine and pancreas, DNA derived from either the CHOL-LSNA or DPPE-LSNA had greater tissue accumulation relative to linear DNA. In the liver and serum, twice as much DNA fluorescence was observed with LSNA-treated mice than linear DNA. The LSNA core fluorescence was at a comparable ratio, with both LSNAs exhibiting 1.5- to 2-fold enhancement in the liver and greater than 50-fold enhancement in the serum. There was a skewing toward CHOL-LSNA trafficking to the lungs and lymph nodes, where we observed 5-fold and 3-fold increased Cy5-DNA fluorescence compared to linear DNA (Figures 3B, S5). Compared to DPPE-LSNAs, DNA from CHOL-LSNAs accumulated in the lungs by greater than 2-fold. The liposome core fluorescence was also 2-fold greater with CHOL-LSNAs, which suggested that the cholesterol-DNA may not be released from the liposome core before lung accumulation (Figure 3B). In contrast, DNA from DPPE-LSNAs trafficked in greater amounts to the heart, brain, and kidneys (Figures 3B, S5, S7). Most notably, DPPE-LSNAs exhibited nearly 2-fold enhanced Cy5-DNA fluorescence in the kidneys compared to the Mix and CHOL-LSNAs, but showed no TAMRA fluorescence enhancement (Figures 3B, S6). This suggests that the enhanced kidney delivery may be due to DNA dissociation from DPPE-LSNAs prior to kidney accumulation.

Figure 3.

Whole organ analysis of DNA trafficking. Following iv injection of LSNAs, organs were harvested at (A, B) 30 min and (C, D) 24 h and imaged ex vivo. TAMRA-PC (liposome) and Cy5 (DNA) fluorescence were separated using the spectral unmixing function of the IVIS instrument. Relative tissue level distribution normalized to untreated mice was assessed by imaging all organs simultaneously (A, C). Individual organs were imaged, and the relative fluorescence was calculated at (B) 30 min and (D) 24 h (N = 3–5; statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test; *P < 0.05, **P < 0.01, error bars represent standard error).

Similarly, at 24 h, the liver and kidneys exhibited the overall highest DNA accumulation (Figures 3C, S9). Both LSNAs exhibited higher liposome core fluorescence than the Mix in the serum but had little Cy5-DNA fluorescence at this time point. The highest serum DNA signal was from CHOL-LSNAs (Figure 3D). In the liver, both LSNAs continued to show higher Cy5 fluorescence compared to linear DNA (Figure 3D). Cy5-DNA fluorescence from CHOL-LSNAs remained higher than linear and DPPE-LSNAs in the lungs and lymph nodes (Figures 3D, S9, S11) and also had the highest accumulation in the brain and bone marrow. CHOL-LSNA-treated mice exhibited the highest TAMRA liposome fluorescence in the liver and lungs at this time point (Figures 3D, S8, S10). DPPE-LSNAs continued to deliver the most DNA to the kidneys, with nearly 3-fold Cy5 fluorescence relative to CHOL-LSNAs (Figures 3D, S9). Akin to the 30 min time point, there was little TAMRA-PC fluorescence in the kidneys from DPPE-LSNAs, suggesting Cy5-DNA release before accumulation.

Linear DNA accumulation, which was higher in the pancreas and small intestine at 30 min (Figure S7), was not significantly different from either LSNA in these organs at 24 h (Figure S11).

To further probe tissue-level distribution and the colocalization of both labeled LSNA components, we imaged cryosectioned tissues from mice injected with both dual fluorophore-labeled LSNAs and the Mix control. In agreement with the IVIS data, livers from animals treated with either LSNA had greater levels of Cy5 fluorescence at both time points (Figure 4A,E). Although LSNAs accumulated in the liver in greater total amounts, the location of linear DNA and both LSNAs appeared similar within our organ sections (Figures 4A,E and S12), suggesting that DNA from LSNAs and linear DNA may be sequestered by similar cell types. This was confirmed by flow cytometry (Figure 5A). For the lungs, the highest fluorescence was observed in the case of the CHOL-LSNA, both at 30 min (Figure 4B) and 24 h (Figure 4F), also consistent with IVIS imaging (Figure 3B,D). While IVIS imaging of whole organs cannot distinguish the location of both fluorophore-labeled components within tissues, cryosections indicated that the liposome and DNA of CHOL-LSNAs were colocalized within lung tissue (Figure S13). This confirmed that the cholesterol-tail DNA is not released from CHOL-LSNAs before lung accumulation. The spleen showed a high level of linear DNA accumulation in what appears to be a blood vessel at 30 min (Figure 4C). In contrast, the spleens from LSNA-treated animals exhibited more evenly distributed fluorescence throughout the organ (Figures 4C, S14). At 24 h, this evenly distributed fluorescence signal remained in the spleens from animals treated with DPPE-LSNAs and was not observed in spleens from animals treated with linear DNA (Figure 4G). The tubules of the kidney showed very high Cy5 signals for both linear DNA- and DPPE-LSNA-treated animals compared to animals treated with CHOL-LSNAs, 30 min (Figure 4D) and 24 h (Figure 4H) post-injection, consistent with the IVIS data. While it is not surprising that linear DNA is observed in the tubules, accumulation of DNA from DPPE-LSNAs in these structures was surprising. The 50 nm diameter of the DPPE-LSNAs is large compared to glomerular capillary pores, the largest of which have radii of approximately 80 Å.^44,45 Thus, DPPE-LSNAs cannot be filtered if the LSNA is intact. We hypothesize that DNA derived from DPPE-LSNAs dissociates from the liposome template in the glomerulus, leaving lipid-tail DNA alone to pass through the glomerular fenestrations. To confirm that the DNA was released, we checked for colocalization of the liposomes and DNA of DPPE-LSNAs in the cryosections. We observed high Cy5-DNA signal in the kidneys, but little TAMRA fluorescence above the untreated background, suggesting that the DNA must have been released from DPPE-LSNAs prior to accumulation within the tubules (Figure S15). This is a finding specific to LSNAs, as significant accumulation of other SNAs in the kidneys has not been previously observed. We reported previously that Au SNAs exhibit the greatest accumulation in the liver and spleen⁹ with very little accumulation in the kidneys. In contrast, LSNAs exhibited the greatest DNA accumulation in four different organs: the liver, lungs, spleen, and kidneys. This suggests that the more dynamic nature of LSNAs compared to Au SNAs may play a role in their In Vivo bioavailability. We also observed DNA trafficking from LSNAs in the small intestines, lymph nodes, and pancreas, suggesting that there may be other possible tissue targets for future LSNA therapeutic development (Figures S7, S11).

Figure 4.

Distribution of Cy5-labeled DNA within tissues. The distribution of Cy5-labeled linear DNA, CHOL-LSNAs, and DPPE-LSNAs in the liver, lungs, spleen, and kidneys at (A–D) 30 min and (E–H) 24 h post-injection (blue = DAPI (nuclear stain), green = phalloidin (actin filament stain), red = Cy5-DNA, taken at 40× magnification, scale bar = 20 μm).

Figure 5.

Analysis of cellular distribution using flow cytometry. Flow cytometry was used to assess the total accumulation of Cy5-DNA in immune and nonimmune cells from linear DNA and LSNAs in the liver, lungs, spleen, and kidneys at (A–D) 30 min and (E–H) 24 h (N = 3 post-injection; statistical significance comparing percent Cy5+ cells was calculated by one-way ANOVA with Tukey’s post hoc test; *P < 0.05, error bars represent standard deviation).

LSNA Architecture and DNA Attachment Chemistry Alter Cellular Trafficking of DNA within Major Organs.

After examining macroscopic differences in DNA distribution between each architecture using IVIS and fluorescence microscopy, we determined structure–distribution relationships on the cell population level. We developed two flow cytometry panels, one for staining immune cells and another for nonimmune cells, which would capture the majority of cell types present in each organ. Within the immune cell panel, single cell suspensions derived from each organ were stained for a general immune cell marker (CD45), T cells (CD3), B cells (CD19), neutrophils (CD11b), dendritic cells (CD11c), and macrophages (CD68). The nonimmune cell panel included the general immune cell marker (CD45), to exclude those cells that stained positively, as well as markers for epithelial cells (EPCAM), endothelial cells (CD31), fibroblasts (CD140a), and blood-derived stem cells (CD34). The general gating strategy used is depicted in Figure S16.

At the 30 min time point (after injection of labeled linear DNA or LSNA), there was very little difference between groups in the total number of cells from the liver (Figure 5A), spleen (Figure 5C), and kidneys (Figure 5D) that tested positive for the Cy5-DNA. However, an investigation of specific cell types within the spleen and kidneys revealed differences in the total amount of DNA being delivered to cells by each respective construct. In the spleen, there was a trend for higher accumulation of DNA from the DPPE-LSNA in the nonimmune cells, particularly in epithelial cells (Figure 5C), and higher accumulation of DNA from CHOL-LSNA in CD11b+ immune cells. The kidney showed higher linear DNA accumulation in nonimmune cells, but no difference between groups in immune cells. This was an expected outcome because the count of total immune cells (CD45+ cells) in the kidney was very low (<1% of cell counts). The most significant difference at 30 min for total Cy5-positive cells was observed in the lungs (Figure 5B), where CHOL-LSNAs showed the highest DNA accumulation, which is consistent with IVIS and fluorescence imaging results. In particular, DNA from CHOL-LSNAs was preferentially trafficked to the nonimmune cells of the lungs, and there was no significant difference between CHOL- and DPPE-LSNAs in immune cells, although the total DNA fluorescence from both LSNAs in immune cells was higher than linear DNA.

At the later time point, the trend for total Cy5 accumulation in each organ remained the same, with the liver, spleen, and kidney showing no significant difference between groups and the lungs showing higher CHOL-LSNA delivery. However, there were differences in the total DNA in specific cell types. In the liver at 24 h (Figure 5E), both CHOL- and DPPE-LSNAs exhibited higher fluorescence in nonimmune cells, which suggested that LSNA architecture is responsible for enhanced DNA delivery to the liver. Noticeably, in the immune cells within the liver, CHOL-LSNAs showed higher delivery to T cells, but all three treatments were less distinguishable in B cells, neutrophils, dendritic cells, and macrophages. In the spleen (Figure 5G), the same trend was observed in nonimmune cells as at 30 min. In contrast, immune cells sequestered DNA from DPPE-LSNAs in greater amounts in CD11b+ cells than DNA from CHOL-LSNAs. A slight preference of CHOL-LSNA DNA trafficking to CD11c+ and CD68+ cells at 24 h was also observed. In the kidney (Figure 5H), there was a dramatic reduction in the trafficking or accumulation of DNA from CHOL-LSNAs in all nonimmune and CD19+ immune cells. Whereas linear DNA showed the highest accumulation at 30 min, DPPE-LSNAs exhibited the highest fluorescence intensities from most cell types at 24 h. Specifically, fibroblasts, epithelial cells, and T cells showed very high DPPE-LSNA accumulation. Finally, in the lungs (Figure 5F), CHOL-LSNAs exhibited higher Cy5 fluorescence intensity than linear DNA in all cell types, with a preference for the trafficking of DNA from CHOL-LSNAs to immune cells.

CONCLUSION

In summary, the results from this study suggest that LSNAs are not immediately cleared from circulation and that they may be used to direct nucleic acids to cells and organs outside of those rich in cells of the mononuclear phagocyte system. This is an important finding, as the MPS is a major hurdle in the delivery of nanoparticle-based therapeutics,^46–48 but understanding how structure dictates where LSNAs (or the nucleic acids that comprise SNAs) accumulate upon intravenous administration allows for the rational design of targeted LSNA therapeutics. This insight broadens the scope of the clinical indications that could benefit from LSNA therapies.

We show that the architecture of LSNAs offers a delivery advantage over linear DNA, as the DNA derived from LSNAs is observed in greater quantity in most tissues and circulation after 30 min and 24 h post-injection. Distribution differences between linear DNA and LSNAs are likely due to a decreased inflammatory cytokine response and a different clearance mechanism. The SNA architecture’s ability to enhance nucleic acid transport across barriers within the body and uptake into many cell types may also drive these distribution differences.

In addition, we have shown that the affinity of DNA to its liposome template affects the distribution of LSNAs In Vivo. This is a particularly important design consideration for therapeutic LSNA development. CHOL-LSNAs show high DNA trafficking to the lungs, which could lead to therapeutic development for indications such as chronic obstructive pulmonary disease, pulmonary fibrosis, or lung cancer. DPPE-LSNAs show high DNA accumulation in the kidneys at the time points examined, which could be beneficial for treating glomerular diseases. These LSNAs also exhibit high accumulation in the spleen, indicating potential as cancer vaccines. The dense DNA shell on LSNAs also changes the tissue-level distribution of the liposome core. The CHOL-LSNA architecture enhanced the delivery of the liposome core to the liver and lungs, while the DPPE-LSNA architecture significantly increased the delivery of liposomes to the brain. As the liposome components of LSNAs can be loaded with other drug cargoes, LSNAs have the potential to co-deliver drugs and nucleic acids to several major organs. We envision that the readily tailorable distribution we describe in this article will inform further applications of this technology, especially in targets where structure dictates significant delivery enhancement.

EXPERIMENTAL SECTION

Synthesis and Characterization of LSNAs.

DNA oligonucleotides were synthesized using automated solid support phosphoramidite synthesis (model: MM12, BioAutomation, Inc.). The sequence used, ODN 2138, has been previously shown to be nonimmunogenic in linear and SNA form.⁴⁹ The free strand nontargeting (ODN 2138) sequence is 5′-TCCATGAGCTTCCTGAGCTT-Cy5-(spacer18)-(spacer18)-3′. On the nanoparticle, the nontargeting (ODN 2138) sequence is 5′-TCCATGAGCTTCCTGAGCTT-Cy5-(spacer18)-(spacer18)-cholesterol-3′. The DBCO-modified nontargeting (ODN 2138) sequence is 5′TCCATGAGCTTCCTGAGCTT-Cy5-(spacer18)-(spacer18)DBCOdT-3′. All oligonucleotides were synthesized with a phosphorothioate (PS) backbone. Sequences were purified by high-pressure liquid chromatography (HPLC, Agilent Technologies) and characterized using matrix-assisted laser desorption ionization-time-of-flight (MALDI-ToF, Bruker Autoflex III). 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(6-azidohexanoyl) ammonium salt (Azido-Cap PE), and 1palmitoyl-2-(dipyrrometheneboron difluoride)undecanoyl-sn-glycero-3-phosphocholine (Topfluor TAMRA-PC), all purchased from Avanti Polar Lipids, Inc., were dissolved in chloroform and prepared into a lipid film. The solvent was evaporated under nitrogen, and trace chloroform was removed under vacuum for several hours. Following this, the lipid film was rehydrated in a buffer containing 20 mM HEPES and 150 mM NaCl (pH 7.4) and freeze–thaw cycled several times. The solution was then extruded through polycarbonate membranes of increasingly smaller pore size (100 nm, 80 nm, 50 nm) until the resulting small unilamellar vesicles were monodisperse with a hydrodynamic radius of ~50 nm as ascertained by DLS (Malvern Instruments). The concentration of lipids was determined via elemental analysis using ICP-OES (Thermo Fisher Scientific). DNA loading to each nanoparticle was determined by measuring the DNA absorbance at 260 nm of LSNAs dissociated in 0.1% sodium dodecyl sulfate using a UV–vis spectrophotometer and measuring total phosphorus concentration using ICP-OES.

Based on analysis of maximum cholesterol DNA loading on 50 nm liposomes shown in Figure S2, 1.3 mM of total lipids was mixed with 20 μM cholesterol- or DBCO-terminated DNA for 3–4 h at 37 °C under constant agitation. The hydrodynamic radius and polydispersity were measured by DLS.

TEM of LSNAs.

LSNA samples were negative stained with 2% (w/v) uranyl acetate. LSNAs were drop cast on TEM grids containing a carbon film on 300 copper mesh (Ted Pella, Inc.). After 30 s, the liquid was wicked away using filter paper and the sample was rinsed twice with 20 mM HEPES containing 150 mM NaCl to remove particles not adhered to the grid. Subsequently, uranyl acetate stain solution was dropped onto the grid and removed four times using filter paper, and the grid was air-dried. A JEOL 1230 TEM (JEOL, Ltd.) was used for imaging.

Animal Handling.

Male mice (C57Bl/6) in the age range of 8–12 weeks were obtained from The Jackson Laboratory and maintained in conventional housing. All animals used were handled according to methods and procedures approved by the Institutional Animal Care and Use Committee at Northwestern University. Briefly, mice were given a single bolus injection of 50 μM linear DNA or LSNAs via peripheral intravenous injection. At predetermined periods of time, mice were anesthetized using a 1:1 mixture of ketamine/xylazine, and blood was collected via cardiac puncture. Organs were cleared of blood by transcardial perfusion with 1× phosphate-buffered saline (PBS).

Evaluation of Cytokine Production.

Once blood was removed via cardiac puncture, it was allowed to clot on ice. Samples were centrifuged at a minimum of 400g for 5 min. The supernatant was isolated, immediately frozen, and stored until analysis. The amounts of IL-6, IL-10, IL-12p70, MCP-1, TNF, and IFNγ were measured using a flow-cytometry-based multiplexing assay (CBA mouse inflammation kit, BD Biosciences) on a FACSymphony flow cytometer (Becton Dickinson), and data were visualized using FlowJo (version 10.5.3, FlowJo LLC).

Organ Harvest for IVIS Imaging.

Organs for imaging were harvested, fixed in 10% neutral buffered formalin overnight, and then stored in 1× PBS until imaging using an In Vivo Imaging System (IVIS, PerkinElmer). An excitation wavelength of 535 nm and an emission wavelength of 580 nm were used to visualize TAMRA-labeled lipids, and an excitation wavelength of 640 nm and an emission wavelength of 680 nm were used to quantify the relative fluorescence of the Cy5-labeled DNA. One-way ANOVA was used to calculate significance between groups.

Fluorescence Imaging.

Following IVIS imaging, the same organs were placed in 15–30% sucrose at 4 °C until the organs sunk to the bottom of the vial. Tissues were then embedded in a glycol/resin mixture (Tissue-Tek O.C.T.) and snap frozen using liquid nitrogen. Tissues were cryosectioned to 5 μm slices and placed on glass slides. The slides were stained with fluorescein-phalloidin and mounted with an antifade mountant containing 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific) and imaged using an inverted microscope (Zeiss Axio 7 inverted microscope with AxioCam 506 mono).

Flow Cytometry.

Organs for flow cytometry were harvested, minced, and incubated in an enzymatic digestion mixture (collagenase with DNaseI, with or without elastase) for 30 min at 37 °C. Once digested, organs were sieved through a 70 μm cell strainer and centrifuged. Red blood cell lysis was performed as necessary (Gibco). Single cells were washed with 1× PBS/2% bovine serum albumin and stored on ice. Cells were stained for immune (CD45, CD3, CD19, CD11b, CD11c, CD68) and nonimmune (EPCAM, CD31, CD140a, CD34) markers (Becton Dickinson, BioLegend) as well as with a fixable live/dead stain (Thermo Fisher Scientific). Cells were fixed in neutral buffered formalin after staining. Analysis of DNA association was done using a FACSymphony flow cytometer (Becton Dickinson), and data were visualized using FlowJo (version 10.5.3, FlowJo LLC). One-way ANOVA was used to calculate significance between treatment groups.

Statistical Analysis.

All results are expressed as the mean ± SE or mean ± SD and number of biological replicates (N) as noted in the figure captions. Outliers were removed using the ROUT method with a false discovery rate (Q) of 1%. IVIS data of whole organ fluorescence of the Cy5-DNA and TAMRA liposomes are normalized to the untreated organs; hence all bar graphs are reported in fold fluorescence enhancement over untreated organs. One-way analysis of variance (ANOVA) was performed and Tukey’s post hoc test was used for multiple comparisons when the result was significant (P < 0.05). In analyzing the flow cytometry data (Figure 5), significance tests were applied to the % Cy5 positive cells in each organ, but not to the MFI values, as removing Cy5-negative cells does not retain a normal distribution in the data. All statistical analyses were performed with GraphPad Prism 8.