Unprecedently High Targeting Specificity Toward Lung ICAM-1 Using 3DNA Nanocarriers

Abstract

DNA nanostructures hold great potential for drug delivery. However, their specific targeting is often compromised by recognition by scavenger receptors involved in clearance. In our previous study in cell culture, we showed targeting specificity of a 180 nm, 4-layer DNA-built nanocarrier called 3DNA coupled with antibodies against intercellular adhesion molecule-1 (ICAM-1), a glycoprotein overexpressed in the lungs in many diseases. Here, we examined the biodistribution of various 3DNA in mice. A formulation consisted of 3DNA whose outer-layer arms were hybridized to secondary antibody-oligonucleotide conjugates. Anchoring IgG on this formulation reduced circulation and kidney accumulation vs. non-anchored IgG, while increasing liver and spleen clearance, as expected for a nanocarrier. Anchoring anti-ICAM changed the biodistribution of this antibody similarly, yet this formulation specifically accumulated in the lungs, the main ICAM-1 target. Since lung targeting was modest (2-fold specificity index over IgG formulation), we pursued a second preparation involving direct hybridization of primary antibody-oligonucleotide conjugates to 3DNA. This formulation had prolonged stability in serum and showed a dramatic increase in lung distribution: the specificity index was 424-fold above a matching IgG formulation, 144-fold more specific than observed for PLGA nanoparticles of similar size, polydispersity, ζ-potential and antibody valency, and its lung accumulation increased with the number of anti-ICAM molecules per particle. Immunocytochemistry showed that anti-ICAM and 3DNA components colocalized in the lungs, specifically associating with endothelial markers, without apparent histological effect. The degree of in vivo targeting for anti-ICAM/3DNA-nanocarriers is unprecedented, for which this platform technology holds great potential to develop future therapeutic applications.

GRAPHICAL ABSTRACT

3DNA, a nanocarrier built of DNA, provides highly specific accumulation in the lungs in vivo when targeted to ICAM-1 vs. controls.

INTRODUCTION

The use of deoxyribonucleic acid (DNA) as a carrier in drug delivery applications is a relatively new and rapidly growing field [1, 2]. This natural polymer has great potential in this context because of its unique physicochemical and biological properties [1, 2]. For example, DNA is soluble in physiological fluids, fully biodegradable, and has groups amenable for chemical modification and conjugation of therapeutic, targeting, or imaging agents, while agents can also intercalate within its structure or be intrinsically built in (e.g. therapeutic oligonucleotides, aptamers, etc.) [1, 2]. Sequence-controlled, base complementarity provides the opportunity to manufacture highly reproducible DNA structures with exquisite precision in terms of size and architecture [1, 2]. In addition, nucleic acid constructs can be designed with tunable levels of flexibility, biocompatibility, bioactivity and biodegradability, and fabricated to respond to various environmental cues such as temperature, pH, ionic strength, cell signaling, and degradation processes [1, 2]. This high versatility, reproducibility, and precision makes DNA a desirable material for drug delivery, which complements well the available toolbox of synthetic polymers and other materials more classically employed in this field. Not surprisingly, some designs are exploring the use of DNA hybrid materials for therapeutic and other biomedical applications [2, 3].

Many and varied DNA-based configurations for drug delivery have been reported [1, 4], including “spherical DNA” consisting of oligonucleotide-decorated gold nanoparticles (NPs) [5–7], dendrimeric conformations [8–10], various origami and geometrical cages [11–15], and liposomal- and tubular- like structures [16–18]. The size of these assemblies is diverse and ranges from a few nanometer particles to mesoscale configurations [1–3]. While natural DNA is typically highly labile due to fast enzymatic cleavage in the extra- and intra-cellular environment, engineered DNA conformations possess enhanced resistance to nucleases due to the tight packing of nucleic acids in their structure [19–21].

Most of these systems are just beginning to be studied at the cellular and in vivo levels, rendering important insight into their behavior in the biological environment and potential for further development toward translation [1, 4]. The majority of mechanistic studies have been conducted at the cellular level and show DNA-built carriers mainly interact with scavenger receptors [5, 6]. This is postulated to be mediated by incorporation of serum proteins, which appears to drive cellular endocytosis via macropinocytosis- and caveolae-like routes [5, 13, 14, 16]. In addition, specific targeting of DNA-built drug carriers using affinity moieties such as antibodies (Abs), aptamers, vitamins, etc. has been shown to enable preferential interaction with cells expressing the selected receptors or disease markers. For instance, HER2, nucleolin, biotin receptor, folic acid receptor, transferrin receptor, mucin 1, and several endothelial cell adhesion molecules are examples of markers to which DNA-built drug carriers have been targeted [6, 8, 10, 11, 17, 21, 22]. In some cases, low number of affinity moieties (targeting valency) per particle rendered modest site-specific targeting with remnant contribution by scavenger receptors, while high targeting valency enabled great specificity toward the selected receptor and endocytic pathway associated, which illustrates a wide specificity range amenable for design [6, 23].

At the cellular level, many DNA-built drug carriers have been shown to deliver cargo (e.g. plasmid DNA, siRNA and other biologically active nucleic acids, as well as proteins and enzymes) to intracellular locations [6, 7, 16, 23–26]. Most reports, however, have used DNA-built carriers for intracellular delivery of agents such as doxorubicin and similar chemicals, which are loaded via intercalation [9, 13–16, 21, 22, 27, 28]. Although the majority of applications being investigated focus on cancer therapeutics, inflammation, immunomodulation and vaccination are also being explored [7, 23, 29, 30].

Information regarding the in vivo biodistribution characteristics of DNA-built drug carriers is still scarce, due to the relative novelty of this research area. Previous studies have focused on a few ex vivo applications, local administration directly into the target tissue, such as topical application, transplantation, ophthalmic or intracranial administration [7, 9, 25, 29, 31]. The enhanced permeability and retention (EPR) effect, by which drug carriers accumulate in areas with leaky vasculature, has been capitalized in a handful of works involving intravenous (i.v.) administration [8, 10, 12, 14, 26]. Even fewer DNA-built carriers have been tested in vivo in the realm of affinity targeting: their tracing has mostly focused on localization within tumor areas after clearance from other regions [8, 10, 11, 21]. Therefore, more studies are necessary to establish the full potential and guide the design of this new type of drug carriers.

In the present work, we aimed to investigate in vivo targeting and biodistribution aspects, particularly focusing on the DNA-built scaffold we previously examined in cell culture [23], called 3DNA. 3DNA® nanocarriers are three-dimensional, highly branched, multilayered structures built from interconnected monomeric subunits of DNA, designed by Genisphere LLC [8, 9, 24, 32]. The structure is photo-crosslinked using psoralen for enhanced stability and can be further modified with other nucleic acids, dyes, small molecules, Abs, etc. [32, 33]. Although 3DNA was originally used as a technology for signal amplification in life science and diagnostics [33], early work in our laboratory showed great potential for drug delivery in various cell culture models [23]. We used commercial 3DNA structures hybridized to secondary Ab-oligonucleotide conjugates and independently anchored specific Abs against intercellular adhesion molecule 1 (ICAM-1), platelet-endothelial cell adhesion molecule 1 (PECAM-1), transferrin receptor, or mannose-6-phosphate receptor. In this model, the resulting formulations showed about the highest specificity reported from a DNA-built system (from 24- to 190-fold increased specificity depending on the target, over 3DNA coupled to non-specific Abs), with full avoidance of scavenger-mediated uptake pathways [23]. This specific targeting was demonstrated in endothelial, epithelial, mesothelioma, and fibroblastic cells, showing potential for a broad number of applications [23]. Further, these formulations demonstrated intracellular delivery of small toxins (phalloidin), polysaccharides (dextran), proteins (albumin), plasmidic DNA (encoding EGFP-RhoA) [23].

Subsequently, the 3DNA platform has been tested ex vivo and in vivo by other groups. Examples include targeting the transferrin receptor to deliver a plasmid encoding diphtheria toxin in the context of pancreatic cancer [24], folate targeting to deliver siRNA in an ovarian cancer model [8], and using a unique monoclonal Ab to specifically deliver doxorubicin and deplete problematic cells in an ophthalmic application [9]. Despite the encouraging functional results encountered in all of these directions the biodistribution and, hence, targeting potential of 3DNA in vivo was not established.

Here we aimed to study these biodistribution and targeting properties, with particular focus on targeting ICAM-1. ICAM-1 is a glycoprotein expressed on the surface of endothelial and other cells in the body, and is involved in inflammation [34]. Its expression is markedly increased upon a plethora of pathological conditions, for which it is an attractive target for drug delivery interventions, as shown by many groups [35, 36]. In particular, due to high ICAM-1 expression, extensive endothelial surface, and receiving full cardiac output, the pulmonary vasculature is a preferential target site for drug nanocarriers addressed to ICAM-1 [37, 38]. Given our previous experience and available data on the biodistribution and specificity of ICAM-1 targeting in vivo using polymeric NPs [38–40], the relevance of this target for therapeutic applications [34, 35, 37], and the need to characterize the targeting behavior of DNA-built nanostructures in vivo, here we examined the biodistribution of various formulations of ICAM-1-targeted 3DNA after i.v. injection in mice, and also compared it to that of non-targeted counterparts and polymeric (poly-lactic-co-glycolic acid; PLGA) NPs.

MATERIALS AND METHODS

Reagents

Rat monoclonal immunoglobulin G (IgG) Ab against mouse ICAM-1 (anti-ICAM) was clone YN1, produced in a hybridoma from American Type Culture Collection (Manassas, VA). Non-specific IgG Ab (called IgG hereafter) was from Jackson Immunoresearch (Pike West Grove, PA). Polyclonal anti-PECAM-1, FITC-labeled secondary Abs and Alexa-fluor 488 labeled secondary antibodies were from Novus Biologicals (Centennial, CO), Jackson Immunoresearch (West Grove, PA) and Thermo fisher Scientific (Waltham, MA), respectively. DNA oligonucleotide (72-mer) modified with thiol at 5’ was from Oligo Factory (Holliston, MA). Pierce bond-breaker TCEP solution, LC-SMCC crosslinker, 7k MWCO Zeba spin columns, 10k MWCO Amicon spin filters, Sartorius™ Vivaspin™ 500 centrifugal concentrator filters (1,000 KDa), thiophilic adsorption resin, heterobifunctional Pierce crosslinking kit, and bovine serum albumin (BSA) were from Fisher Scientific (Kerrville, TX). Carboxylic acid-terminated PLGA (50:50 copolymer ratio; 32 kDa average molecular weight) was from Sigma-Aldrich (Saint Louis, Missouri). Iodogen iodination tubes were from Pierce (Rockford, Illinois) and BioSpin Tris Columns from BioRad (Hercules, California). All other reagents were from Sigma Chemical (St. Louis, MO).

Preparation of PLGA and 3DNA nanocarriers

PLGA NPs were prepared using a nanoprecipitation/solvent evaporation method, as described [38, 41]. Briefly, PLGA was dissolved at 20 mg/mL in acetone and poured under stirring into nanopure water at final concentration of 2.22 mg/mL. NPs formed instantaneously and the dispersion was kept under mild stirring for acetone evaporation, followed by concentration in a rotary evaporator. The NP concentration was estimated by dynamic light scattering (DLS; Malvern Zetasizer, Worcestershire, UK) in comparison to similar-size standards and from the weight of PLGA polymer of a given NP volume measured after freeze-drying. PLGA NPs were then coated by surface adsorption with 0.9 μM anti-ICAM or IgG control, followed by centrifugation at 11,000 rpm for 3 minutes to remove non-coated Abs, as previously reported [38, 42].

The 3DNA manufacturing process has also been published [32, 33]. Briefly, pairs of seven unique single-stranded DNA strands hybridize into double-stranded DNA “monomers” with single-stranded ends. Five unique monomers are then hybridized to one another, layer-by-layer, with psoralen crosslinking at each step to make covalent bonds in specified nucleic acid regions. The resulting tridimensional structure is mostly double-stranded; only the peripheral regions are single-stranded. The final size and number of peripheral arms available for subsequent hybridization depends upon the number of layers assembled [32, 33]. These studies utilized 4-layer 3DNA assemblies with average MW of 11,000 kD. As per the Ab coat, secondary Ab, anti-ICAM, or IgG control were conjugated to short oligonucleotides (oligo) using NHS-maleimide chemistry. For this, 72-mer DNA oligo with 5’thiol modification was reduced in 50 mM TCEP, followed by ethanol precipitation and resuspension in 5 mM EDTA in phosphate buffer saline (PBS). In parallel, Ab was reacted with LC-SMCC, followed by crosslinker removal in Zeba spin columns. Reduced oligo was added to Ab-LC-SMCC for conjugation, chromatography was used to remove excess thiol-oligo, and the resulting Ab-oligo conjugate was concentrated using Amicon 10 kDa MWCO spin filters. Conjugates were then hybridized to 3DNA by incubation at 37 °C for 30 min with a Tm of 72°C (16.9 μM conjugate, as oligo, to 89 μM 4-layer 3DNA). As a model, 4-layer 3DNA hybridized with secondary Ab-oligo conjugate was further incubated with primary anti-ICAM (or non-specific IgG control). Additional formulations tested in this work include 4-layer 3DNA directly hybridized with primary anti-ICAM (or IgG control)-oligo conjugates, and formulation where 3DNA had been similarly conjugated to fluorescent Cy3 for visualization.

The average diameter, polydispersity index (PDI), and ζ-potential of Ab-coated NPs and 3DNA were measured via dynamic light scattering (DLS) and electrophoretic mobility using the Malvern Zetasizer (Worcestershire, UK). Their visualization was additionally conducted by transmission electron microscopy (TEM), applying negative staining in the case of 3DNA. The number of Abs coated was measured using ¹²⁵I-labeled Abs and quantification in a gamma counter (PerkinElmer, Boston, Massachusetts) as described [42]. The findings from this characterization are presented in the Results and Discussion section.

Stability of 3DNA hybridized with antibody-oligo conjugate

Filtration through a 1,000 KDa centrifugation filter was first used to study the stability of ¹²⁵I-labeled 3DNA (11,228 KDa) hybridized or not with ¹³¹I-Ab-oligo conjugate (174 KDa), before and after incubation with 50% bovine serum for 1 h at 37°C. Radioisotope tracing in a gamma counter was utilized to quantify the retained and filtered fractions for both counterparts, and trichloroacetic acid (TCA) precipitation was implemented to detect free iodine and correct data to avoid artifacts. Free-iodine corrected cpms were then used to calculate the percentage of retained and filtered species.

To monitor degradation by gel electrophoresis, naked 3DNA or Ab/3DNA were incubated in 75% human serum or control phosphate buffer saline at 37 °C for 16 hours. Densitometry was used to calculate the percentage of intact 3DNA over time.

Biodistribution and visualization of targeted formulations in mice

Anesthetized C57BL/6 mice were injected with either “naked” ¹²⁵I-labeled Abs, ¹²⁵I-Abs coupled to PLGA NPs or 3DNA, or 3DNA without Abs. For comparisons between targeted versus non-specific 3DNA, injections encompassed 2.15 × 10¹³ particles per Kg body weight. For comparisons between 3DNA and PLGA counterparts, injections included 1.29 × 10¹³ particles per Kg body weight, with matching number of ¹²⁵I-Abs per particle (90 Ab molecules per particle; 249 μg Ab per Kg body weight). Blood samples were collected at the indicated time points and at sacrifice either 30 min or 60 min after injection, as indicated, followed by collection and weighing of main organs (brain, heart, kidneys, liver, lungs, and spleen). Blood and organs were subjected to homogenization and TCA precipitation to correct data for free ¹²⁵Iodine which may be present in the formulation or arise from degradation after injection. Radioactivity measurements obtained using a gamma counter were then utilized to calculate the percentage of injected dose in blood and each organ (%ID), where the injected dose is the dose measured prior to injection minus the dose remnant in the syringe after the injection. We also calculated the %ID per gram of organ (%ID/g) which compares relative accumulation in organs with different weight, the localization ratio (LR = %ID/g in an organ : %ID/g in blood) to express the tissue-to-blood distribution, and the specificity index (SI = LR of a targeted formulation : LR of the non-targeted counterpart) to estimate the targeting advantage [39, 40]. All biodistribution data is compiled in Supplementary Table S1, yet to facilitate result description and discussion in an organized manner, specific comparisons are presented in separate Figures. Please, note that small discrepancies in the values reported may be possible due to the use in calculations of more decimals than shown.

Alternatively, similar experiments were conducted using Ab/Cy3–3DNA to visualize these formulations in lung samples isolated from mice 60 min after injection. Fixed lung samples were used for H&E staining, or were immunostained for confocal microscopy using either Alexa fluor 488-labeled secondary Ab to examine the colocalization of Ab and 3DNA counterparts of the formulation, or that of 3DNA with polyclonal anti-PECAM-1 + FITC-labeled secondary Ab, to localize formulations with the vascular endothelium. All animal experiments were performed under protocols approved by the Institutional Animal Care and Use Committee and University regulations.

Statistical Analysis

Data were calculated as mean ± standard error of the mean (S.E.M.), with n ≥ 5 mice being used for PLGA NPs or 3DNA with primary Ab-oligo conjugates, and n ≥ 3 for 3DNA with secondary Ab-oligo conjugate + primary Abs. Significance was determined using the Student’s unpaired t-test, assuming a p-value of 0.05.

RESULTS AND DISCUSSION

In vivo biodistribution of 3DNA with secondary conjugate and non-specific IgG

First, we characterized 4-layer 3DNA pre-hybridized with secondary Ab-oligo conjugate (2°3DNA), which had average diameter of 175±9 nm, PDI of 0.3±0.1, and ζ-potential of −14 mV. We coupled non-specific ¹²⁵I-IgG control to 2°3DNA and the resulting formulation (IgG/2°3DNA) was 211±14 nm in average diameter, with a PDI of 0.4±0.01 and ζ-potential of −20 mV. IgG/2°3DNA was intravenously injected in mice and circulation was followed for the first 30 min, to monitor early events without confounding results of potential degradation. As shown in Figure 1A, IgG/2°3DNA disappeared from the bloodstream faster than IgG: 56% vs. 91% of the injected dose (ID), respectively, 1 min after injection and 24% vs. 75% ID by 30 min. This is in agreement with the fact that Abs naturally circulate in the bloodstream for longer time periods than NPs [43]. Lower circulation of IgG/2°3DNA was expected due to its larger size, multivalent presentation of IgG molecules on 3DNA which may enhance Fc-mediated uptake by phagocytic organs, as well as possible interference with Ab binding to FcRn, known to contribute to natural Ab circulation. Yet, faster clearance did not hinder IgG/2°3DNA from reaching to tissues, although this formulation had different biodistribution compared to IgG. In particular, the majority of IgG accumulated in the kidney and liver (4.5% and 12.7% ID, Supplementary Fig. S1), with the kidney receiving a slightly higher dose per gram of tissue (11.7% and 10.1% ID/g; Fig. 1B). In contrast, IgG/2°3DNA had reduced kidney uptake and enhanced liver accumulation (1.8% and 46.9% ID, Supplementary Fig. S1), with the liver receiving a markedly enhanced fraction (4.6% and 42.6% ID/g; Fig. 1B). Coupling to 3DNA did not affect accumulation of IgG in the lung or spleen (0.9-fold and 0.7-fold change in %ID/g; Fig. 1B), and decreased accumulation in the brain and heart (2.0-fold and 3.6-fold change in %ID/g; Fig. 1B). Nevertheless, biodistribution in these four organs was low for both formulations, i.e. <1.5% ID (Supplementary Fig. S1) and <10% ID/g (Fig. 1B). In addition, the localization ratio (LR), a parameter that provides information about the tissue-to-blood distribution (see Methods section), was increased in most organs for IgG/2°3DNA vs. free IgG. Yet, the LR for IgG/2°3DNA was <1 for all organs except the liver (Fig. 1C), indicating that despite the biodistribution shift observed between the kidney and the liver, no particular targeting was observed. This finding was expected for these non-specific IgG formulations and is in agreement with lack of targeting or uptake of IgG/2°3DNA observed in cell culture in our previous publication [23].

Figure 1. Biodistribution of IgG/2°3DNA.

Non-coupled ¹²⁵I-IgG or ¹²⁵I-IgG coupled to commercial 3DNA bearing a secondary Ab-oligo (2°3DNA), i.v. injected in C5Bl/6 mice. (A) Circulation was calculated at the indicated times between injection and at sacrifice at 30 min, as the percentage of the injected dose (% ID). (B) Organ biodistribution at sacrifice was calculated as % ID per gram (%ID/g) to compare organs of different weight. (C) Organ-to-blood distribution expressed as the localization ratio (LR), was calculated as % ID/g in an organ divided by % ID/g in blood. Data are mean ± S.E.M. * p < 0.05 by Student’s t-test.

Next, we substituted non-specific IgG with anti-ICAM to examine if specific targeting could be achieved. As for the non-specific controls, anti-ICAM/2°3DNA also showed reduced blood circulation compared to anti-ICAM alone (e.g. 31% vs. 47% ID at 30 min; Fig. 2A), reduced kidney accumulation (e.g. 5.5% vs. 14.8% ID/g, Fig. 2B; 2.0% vs. 5.4% ID, Supplementary Fig. S2), and enhanced liver uptake (34.3% vs. 13.8% ID/g, Fig. 2B; 29.8% vs. 16% ID, Supplementary Fig. S2). This was expected since these tissues are not main targets for ICAM-1 and hence, size, Fc multivalency, and other factors would drive this redistribution as for IgG controls (Fig. 1). In contrast, as explained in the Introduction, the lungs are the main target for ICAM-1. As such, anti-ICAM significantly accumulated in this organ (47.6% ID/g, 8.2% ID, 1.8 LR; Fig. 2B,C and Supplementary Fig. S2) compared to control IgG (6.7% ID/g 1.1% ID, 0.2 LR; Fig. 1B,C and Supplementary Fig. S1). Lung accumulation was also high for anti-ICAM/2°3DNA (24.2% ID/g, 3.6% ID, 1.0 LR; Fig. 2B,C and Supplementary Fig. S2), yet decreased compared to anti-ICAM control, which may be due to steric hindrances of Ab presented on 2°3DNA.

Figure 2. Biodistribution of Anti-ICAM/2°3DNA.

Non-coupled ¹²⁵I-anti-ICAM or ¹²⁵I-anti-ICAM coupled to 3DNA bearing a secondary Ab-oligo (2°3DNA), i.v. injected in C5Bl/6 mice. (A) Circulation was calculated at the indicated times between injection and at sacrifice at 30 min, as the percentage of the injected dose (% ID). (B) Organ biodistribution at sacrifice was calculated as % ID per gram (%ID/g) to compare organs of different weight. (C) Organ-to-blood distribution expressed as the localization ratio (LR), was calculated as % ID/g in an organ divided by % ID/g in blood. Data are mean ± S.E.M. * p < 0.05 by Student’s t-test.

Nevertheless, this lung accumulation of anti-ICAM/2°3DNA was also specific compared to that of IgG/2°3DNA (3.4-, 2.8, and 2.0-fold increase in % ID/g, % ID, and LR, respectively; Fig. 3B,C and Supplementary Fig. S3A). In contrast, its liver accumulation was reduced by 1.6- and 1.3-fold for % ID and % ID/g compared to IgG/2°3DNA. Kidney uptake remained similar for both (e.g. 4.6% and 5.5% ID/g), and other organs showed increased accumulation for anti-ICAM/2°3DNA, including the brain, heart, and spleen (Supplementary Fig. S3A and Fig. 3B). This was expected as it had been observed for anti-ICAM polystyrene and PLGA NPs, due to the fact that ICAM-1 is expressed not only by the lung endothelium but also by endothelium (and additional cell types) of other tissues [34, 35, 39, 40, 43]. Primarily, anti-ICAM/2°3DNA moved from the liver to the lung and the spleen (Fig. 3B and Supplementary Fig. S3A), suggesting that specificity was not as profound as reported for other NP formulations [38, 40, 43]. In fact, the tissue-to-blood LR showed the liver remained the organ which received the highest amount of anti-ICAM/2°3DNA (Fig. 3C). Nevertheless, the lung-to-liver LR ratio (not shown) was 0.71 for anti-ICAM/2°3DNA vs. 0.17 for IgG/2°3DNA, showing specificity. In addition, the specificity index (SI), which expresses targeting of anti-ICAM/2°3DNA over IgG/2°3DNA (see Methods), showed that specific lung uptake doubled for anti-ICAM/2°3DNA, while the liveŕs was reduced by half (Supplementary Figure S3B). Therefore, these data were encouraging and provided the rationale to pursue alternative formulations of 3DNA with primary Ab-oligo conjugate directly hybridized to the 3DNA scaffold, to eliminate potential non-specific interactions of the secondary Ab and improve targeting.

Figure 3. Biodistribution of anti-ICAM/2°3DNA.

¹²⁵I-IgG/2°3DNA or ¹²⁵I-anti-ICAM/2°3DNA were i.v. injected in C5Bl/6 mice. (A) Circulation was calculated at the indicated times between injection and at sacrifice at 30 min, as the percentage of the injected dose (% ID). (B) Organ biodistribution at sacrifice was calculated as % ID per gram (%ID/g) to compare organs of different weight. (C) Organ-to-blood distribution expressed as the localization ratio (LR), was calculated as % ID/g in an organ divided by % ID/g in blood. Data are mean ± S.E.M. * p < 0.05 by Student’s t-test.

In vitro characterization of 3DNA directly coupled to primary Abs

Given that a custom-made, instead of a commercial, formulation was then used, we first confirmed that the primary Ab-oligo conjugates do indeed hybridize on 3DNA (see Methods). Table 1 demonstrates this hybridization-driven coupling as measured by size exclusion filtration using radiolabeled 3DNA and Ab counterparts. Complexation was observed regardless whether Ab-oligo (87% retention) or 3DNA (91% retention) were traced, while only minimal retention was found for Ab-oligo alone (7%).

Table 1.

Coupling and stability of Ab/3DNA

Condition	Retained (%)	Filtered (%)
Complexation (prior to serum)
Ab-oligo	6.9	93.1
3DNA	71.8	28.2
Ab/3DNA (track Ab)	87.4	12.6
Ab/3DNA (track 3DNA)	91.4	8.6
Stability (after serum)
Ab-oligo	0.0	100.0
3DNA	33.3	66.7
Ab/3DNA (track Ab)	86.0	14.0
Ab/3DNA (track 3DNA)	100.0	0.0

Therefore, direct Ab incorporation on 3DNA was an efficient process and resulted in an increase in the average hydrodynamic size of 3DNA formulations (Table 2), i.e. from 170 nm prior to Ab coupling to ≈ 180 nm after the coupling of IgG or anti-ICAM. Ab/3DNA also showed more negative ζ-potential vs. 3DNA alone (−43 for IgG/3DNA and −37 mV for anti-ICAM/3DNA vs −19 mV for 3DNA alone), while all formulations had similar PDI ≈ 0.2.

Table 2.

Characterization of 3DNA nanocarriers and PLGA nanoparticles

Formulation	Size (nm)		PDI		ζ-potential (mV)
	Mean	SEM	Mean	SEM	Mean	SEM
3DNA Nanocarriers
Non-coated	170.4	7.5	0.220	0.033	−19.1	0.6
IgG	181.2	5.1	0.231	0.019	−42.7	0.5
Anti-ICAM	179.5	5.7	0.251	0.017	−37.5	0.6
PLGA Nanoparticles
Non-coated	154.4	1.4	0.071	0.007	−59.4	0.6
IgG	196.0	6.6	0.224	0.012	−36.8	0.5
Anti-ICAM	208.7	1.5	0.174	0.007	−32.3	0.2

Next, the stability of Ab/3DNA was studied under more physiological-like conditions. Incubation for 1 h in the presence of serum did not affect size-dependent retention of Ab/3DNA using the described filtration assay (Table 1), regardless whether Ab-oligo (86% retention) or 3DNA (100% retention) were traced. As control, no retention (0%) was detected for non-coupled Ab-oligo and, interestingly, retention of non-coupled 3DNA decayed from 72% to 33% after serum incubation. This suggests that 3DNA is sensitive to serum, likely due to nuclease degradation. However, when its outer layer displays Abs, the Ab coating seems to protect the 3DNA core, decreasing its degradation. This notion was further validated by an independent assay where the stability of 3DNA, with or without Ab coating, was compared in an electrophoretic setting (see Methods) upon incubation with control buffer or serum. As shown in Fig. 4, high stability was observed by 3DNA in control buffer regardless of Ab coupling (close to 100% intact formulations even after 16 h incubation). As expected, non-coupled 3DNA decayed over time in serum, with only 11% intact 3DNA observed by 16 h, rendering a half-life of ~9.5 h. However, Ab/3DNA appeared 60% intact by 16 h with a corresponding half-life of ~26 h, again suggesting a protective role of the Ab coat on 3DNA (Table 1), rendering a ~2.7-fold longer half-life.

Figure 4. Stability of 3DNA with and without Ab coat.

Non-coupled 3DNA and 3DNA coated with control IgG-oligo conjugate (Ab/3DNA) were incubated in PBS vs. 75% serum up to 16 hours. At the indicated times, 3DNA degradation was quantified using gel electrophoresis and imaging. Data are mean ± S.E.M. * Compares serum vs. PBS for each formulation; # compares Ab/3DNA vs. 3DNA (p < 0.05, Student’s t-test).

These data are in agreement with literature showing that engineered DNA structures tend to be more stable than natural DNA conformations, and additional coatings further enhance stability. For instance, a DNA-based delivery system consisting of DNA strands tightly packed on the surface of metal nanoparticles was shown to have 4-fold increased in vitro resistance to DNAse I vs. naked DNA matching controls (from 24 to 100 min half-life) [19]. This was believed to be mediated by high local salt concentrations due to the negatively charged surface of the spherical DNA and the tight packing of DNA strands, which posed steric hindrances for nucleases to reach this substrate [19]. In addition, fluorescently-labeled DNA origami were also shown to be about 50% stable for 24 h upon uptake by cells in culture while regular DNA degraded much faster [20], and complexation of DNA with other materials, whether intentionally for drug delivery application or naturally, further demonstrated protection from fast degradation [44, 45]. Hence, it is likely that the 4-layered structural conformation of 3DNA makes it relatively stable and Ab coating further enhances its longevity (Fig. 4).

In vivo biodistribution of Ab/3DNA nanocarriers

After this characterization, we examined directly-coupled Ab/3DNA in mice. First, their circulation was monitored for 60 min (Fig. 5A). Comparatively, Ab/3DNA disappeared even faster from the bloodstream in the case that anti-ICAM (not IgG) weas coupled on 3DNA, e.g. 3.3% and 23.5% ID in blood by 30 min, respectively (Fig. 5A), compared to 31.4% and 24.5% ID in blood for corresponding 2°3DNA formulations (Fig. 3A). 3DNA with no Ab coating showed intermediate blood levels, i.e. 7.3% at 60 min vs. 3.0% for anti-ICAM/3DNA and 28.6% for IgG/3DNA at this time. Faster clearance of 3DNA alone vs. IgG/3DNA is in agreement with increased stability of Ab-coated 3DNA observed in vitro. The fact that anti-ICAM/3DNA shows lower circulation pairs well with a phenomenon previously observed for anti-ICAM polymeric NPs, including polystyrene or PLGA formulations, shown to be due to fast targeting of endothelial ICAM-1, which is readily accessible from the bloodstream [34, 40, 43].

Figure 5. Biodistribution of customized 3DNA nanocarriers.

¹²⁵I-IgG/3DNA, ¹²⁵I-anti-ICAM/3DNA, or control ¹²⁵I-3DNA were i.v. injected in C5Bl/6 mice. (A) Circulation was calculated at the indicated times between injection and at sacrifice at 60 min, as the percentage of the injected dose (% ID). (B) Organ biodistribution at sacrifice was calculated as % ID per gram (%ID/g) to compare organs of different weight. (C) Organ-to-blood distribution expressed as the localization ratio (LR), was calculated as % ID/g in an organ divided by % ID/g in blood. Data are mean ± S.E.M. *Compares either Ab/3DNA vs. 3DNA alone; #compares anti-ICAM/3DNA vs. IgG/3DNA (p < 0.05 by Student’s t-test).

In accord with the idea that fast clearance does not necessarily impair binding of readily accessible targets such as ICAM-1, anti-ICAM/3DNA remarkably accumulated in the lung, the main ICAM-1 target: 18.0% ID by 60 min (Supplementary Fig. S4A) and 124.3% ID/g (Fig. 5B), well above lung accumulation of IgG/3DNA (0.4% ID and 2.8% ID/g) or anti-ICAM/2°3DNA (3.6% ID and 24.2% ID/g, shown above). Accumulation of anti-ICAM/3DNA in the liver (40.2% ID and 35.2% ID/g) was lower than that of IgG/3DNA (50.2% ID and 46.5% ID/g). In addition, anti-ICAM/3DNA surpassed IgG/3DNA in other organs, including the heart, spleen, etc. (Fig. 5B). 3DNA without Ab coating was least abundant in the liver (20.8% ID and 15.5% ID/g) and most abundant in the kidney (3.2% ID and 8.3% ID/g), which agrees with in vitro observations of its lower stability, and may result in faster liver degradation with kidney clearance of small degradation products. Ultimately, the LR showed good tissue-to-blood retention of anti-ICAM/3DNA over control formulations (Fig. 5C), and also the lung-to-liver LR ratio (not shown) was 3.45 for anti-ICAM/3DNA vs. 0.06 for IgG/3DNA, representing a remarkable targeting specificity: the SI showed 424- and 88-fold enhancement in lung targeting compared to IgG/3DNA and 3DNA, respectively (Supplementary Fig. S4B).

Lung targeting was further verified using fluorescence microscopy to visualize 3DNA labeled with Cy3, which showed profuse presence of anti-ICAM/3DNA vs. IgG/3DNA in the lung tissue (Fig. 6A), as well as colocalization with PECAM-1-positive endothelium (Fig. 6B), as expected. Anti-ICAM/3DNA seemed to locate on the endothelial surface and intracellularly (Fig. 6B), although detailed studies to define mechanistic aspects will need to compare various times points in appropriate transgenic or knockout models. Importantly, dual-color tracing of 3DNA (red) vs. Ab (green) counterparts revealed that both components colocalized in the lung tissue (Fig. 6A), further supporting anti-ICAM-mediated targeting of 3DNA. Histological examination showed no apparent differences between the lungs of mice injected with anti-ICAM/3DNA vs. IgG/3DNA, despite profuse pulmonary accumulation of the former, suggesting relative safety which will be examined in detail in future studies.

Figure 6. Visualization of anti-ICAM/3DNA in the lungs and histology.

Control IgG/Cy3–3DNA or anti-ICAM/Cy3–3DNA were i.v. injected in C57Bl/6 mice and lungs were isolated and processed at sacrifice at 60 min. (A) Confocal microscopy showing Cy3–3DNA (red), while the Ab counterpart was visualized using a FITC-secondary Ab (green). Arrow-heads show colocalization of the Ab with cy3–3DNA. (B) Confocal microscopy showing anti-ICAM/Cy3–3DNA (red) and endothelial cells visualized using polyclonal anti-PECAM1+FITC-secondary Ab (green). Arrows indicate colocalization of cy3–3DNA with PECAM-1. (C) H&E staining. Respective scale bars are shown.

In retrospect, comparing these results to those previously published for anti-ICAM polymeric NPs, it appears the direct Ab coupled, customized 3DNA formulation surpassed previous ones: e.g. lung LR for anti-ICAM polystyrene NPs reported in [40] was 32 and SI against IgG/NPs was 18, while here these values were much greater (as described above). Although they are not built from DNA, the NP formulations reported in [40] had similar size, PDI, and ζ-potential as 3DNA, although they had not been matched side-by-side with 3DNA in terms of targeting valency (the number of antibodies on the coat). Hence, to verify the enhanced targeting of 3DNA designs, we prepared PLGA NPs and 3DNA counterparts with matching parameters. As expected (Table 2), PLGA NPs had increased average diameter and PDI upon coating with Abs (from 155 nm to ≈ 200 nm, and from 0.07 to ≈ 0.2 PDI; Supplementary Fig. S5) and less negative ζ-potential (from −59 mV to ≈−35 mV). In addition, coating PLGA NPs with anti-ICAM or control IgG resulted in similar formulations: 208 vs. 196 nm, 0.17 vs. 0.22 PDI, and −32 vs. −37 mV. Importantly, these values were similar to those observed for 3DNA counterparts; e.g. ≈ 180 nm, ≈ 0.2 PDI, and ≈ −0.4 mV for Ab/3DNA. Importantly, both 3DNA and PLGA displayed similar number of targeting Abs: 78.6 ± 5.6 and 88.6 ± 0.3 Ab molecules per particle, respectively.

Then, we tested PLGA formulations in vivo. Because the lungs, liver, and spleen represented the organs which received the highest %ID of anti-ICAM/3DNA (Supplementary Fig. S4A), we focused on these organs. Direct comparison of 3DNA and PLGA formulations (Fig. 7A) showed anti-ICAM/PLGA disappeared from the circulation similarly as fast as anti-ICAM/3DNA (e.g. 3.9% and 2.6% ID by 60 min). Accumulation of anti-ICAM/PLGA in the intended target, the lungs, was of 3.5% ID (Supplementary Fig. S6A) and 28.3% ID/g (Fig. 7B), which was specific over control IgG/PLGA (e.g. SI was 4.6-fold; Supplementary Fig. S6B). While specific, this lung accumulation was markedly lower than that of anti-ICAM/3DNA (6.0-fold and 5.6-fold lower comparing %ID and %ID/g, respectively). In addition, anti-ICAM/PLGA showed increased liver and spleen accumulation over anti-ICAM/3DNA: i.e. 2.1-fold and 2.5-fold increase in the %ID/g in the liver and spleen, respectively (Fig. 7B). The tissue-to-blood LR was 9.7-fold greater for anti-ICAM/3DNA counterparts (Fig. 7C), and the lung-to-liver LR ratio (not shown) was 11.6-fold higher for anti-ICAM/3DNA, which was 144-fold more specific in targeting the lungs than anti-ICAM/PLGA (compare SI in Supplementary Fig. S6B).

Figure 7. Comparative in vivo biodistribution of 3DNA vs. PLGA formulations.

¹²⁵I-anti-ICAM/3DNA or ¹²⁵I-anti-ICAM/PLGA i.v. injected in C5Bl/6 mice. (A) Circulation was calculated at the indicated times between injection and at sacrifice at 60 min, as the percentage of the injected dose (% ID). (B) Organ biodistribution at sacrifice was calculated as % ID per gram (%ID/g) to compare organs of different weight. (C) Organ-to-blood distribution expressed as the localization ratio (LR), was calculated as % ID/g in an organ divided by % ID/g in blood. Data are mean ± S.E.M. * p < 0.05 by Student’s t-test.

This difference in targeting capacity between 3DNA nanocarriers and polymeric NPs was unexpected a priori. While it is tempting to speculate that the attachment of Abs on the coat of PLGA by surface adsorption may lead to some Ab release upon injection, our previous work has shown surface adsorption on particles, just as in ELISA applications, is relatively stable: no more than 10–15% of the Ab coat detaches upon incubation with serum [46, 47]. Another explanation may be the lower availability for targeting of PLGA-adsorbed Abs vs. Abs hybridized via oligo-conjugates to the outer arms of 3DNA. The DNA configuration may be structurally more flexible and/or may allow for greater target accessibility or more stable binding of the Ab to the targeted marker (ICAM-1) on the endothelial surface, resulting in higher in vivo accumulation in target organs (the lungs). Future studies will examine whether conjugation of anti-ICAM on PLGA NPs with appropriate spacers, for a more accurate comparison, may render these formulations to be more specific and similar to 3DNA counterparts.

An additional observation arises from comparing anti-ICAM/3DNA in Fig. 7 vs. Fig. 5, since these formulations were prepared to carry a different number of anti-ICAM molecules per 3DNA particle (targeting valency): the formulation in Fig. 5 carried 47.2±3.3 anti-ICAM molecules vs. 78.6±5.6 in Fig. 7. As shown in Supplementary Table S1, increasing valency increased lung targeting (e.g. from 124 to 158 %ID/g) while reducing liver accumulation (e.g. from 35 to 26 %ID/g), in agreement with these being, respectively, a specific target vs. a non-specific clearance organ for ICAM-1-addressed formulations This is consistent with previous publications showing that increasing the number of Ab molecules per particle increases avidity and, hence, binding to the target receptor [48, 49]. This effect could be reached when the number of cell-surface receptors are exceeded and/or even cause targeting decay due to steric hindrances, which is to be thoroughly investigated in future studies.

CONCLUSION

DNA-built nanostructures are showing promise in the realm of drug delivery, yet the behavior of these systems in vivo is still poorly characterized due to their relative novelty. In this study, we investigated a DNA-built and highly branched carrier, 3DNA, in terms of physicochemical characteristics, stability in serum, and biodistribution in mice after i.v. injection, using pulmonary ICAM-1 targeting as a translationally relevant model. Our results demonstrate that, for the same size, PDI, ζ-potential, and number of targeting Abs, 3DNA largely outperformed PLGA counterparts, demonstrating superior accumulation in the targeted organ (lungs in the case of ICAM-1) along with reduced accumulation in clearance organs (liver, spleen, kidneys). To the best of our knowledge, no other report has demonstrated such high targeting specificity for a DNA-built nanoscaffold in vivo (per SI, anti-ICAM/3DNA accumulated in the lungs 424-fold over control IgG/3DNA), despite serum proteins and presumed expression of scavenger receptors throughout the body. This level of targeting specificity is remarkable compared to most other reports for any targeted nanocarriers, which highlights the fact that active targeting by affinity ligands such as Abs offer a great opportunity to enhance site-specific drug delivery as long as the selected targets are accessible from the circulation, as is the case for ICAM-1. Whether a similar degree of specificity will be achieved by targeting other markers will have to be investigated. The role of opsonization and protein corona of 3DNA will require careful examination, yet it is expected that this effect would be similar for anti-ICAM/3DNA and IgG/3DNA since these specific and non-specific Abs are similar molecular species, validating that preferential lung biodistribution of the former is due to ICAM-1 targeting, just as observed for control anti-ICAM and other ICAM-1-targeted NPs. In addition, while these efforts are highly supportive of advancing clinical applications, future work must determine the role of the physicochemical characteristics of 3DNA on this outcome and must carefully examine aspects relative to potential side effects of this technology, to rationally guide its development. Nevertheless, given the high versatility with regards to the degree of modification amenable to this type of DNA design, including the use of non-immunological nucleotides, this platform technology is highly attractive and holds great potential to develop into valuable therapeutic applications.