Nanoparticle delivery of RNA-based therapeutics to alter the vocal fold tissue response to injury

Abstract

Objective(s)

Our laboratory and others hypothesize Smad3 is a principle mediator of the fibrotic phenotype in the vocal folds (VF), and we further posit that alteration of Smad3 expression through short interfering (si)RNA holds therapeutic promise, yet delivery remains challenging. To address this issue, we employed a novel synthetic oligomer, lipitoid, complexed with siRNA to improve stability and cellular uptake with the goal of increased efficiency of RNA-based therapeutics.

Methods

In vitro, lipitoid cytotoxicity was quantified via colorimetric and LIVE/DEAD assay in immortalized human vocal fold fibroblasts and primary rabbit vocal fold fibroblasts. In addition, optimal incubation interval and solution for binding siRNA to lipitoid for intracellular delivery were determined. In vivo, a rabbit model of vocal fold injury was employed to evaluate Smad3 knockdown following locally injected lipitoid-complexed siRNA.

Results

In vitro, lipitoid did not confer additional toxicity compared to commercially available reagents. In addition, 20 minute incubation in 1xPBS resulted in maximal Smad3 knockdown. In vivo, Smad3 expression increased following VF injury. This response was significantly reduced in injured vocal folds at 4 and 24 hours following injection (p=0.035 and 0.034, respectively).

Conclusion

The current study is the first to demonstrate targeted gene manipulation in the VFs as well as the potential utility of lipitoid for localized delivery of genetic material in vivo. Ideally, these data will serve as a platform for future investigation regarding the functional implications of therapeutic gene manipulation in the vocal folds.

INTRODUCTION

Vocal fold (VF) fibrosis is among the most challenging laryngeal abnormalities resulting in significant handicap. In healthy vocal folds, oscillatory symmetry of the pliable epithelium and superficial lamina propria allows for complementary cycle-to-cycle contact and favorable voice quality.¹ Injury and the subsequent fibroblastic response disrupt this delicate architecture via activation and proliferation of pro-fibrotic cells, resulting in disruption of mucosal wave and altered glottic closure.^1–4 Due to the difficulties modulating biochemical processes underlying fibrosis, current therapies for VF scarring primarily target glottic closure.^4,5 Our laboratory and others implicated transforming growth factor (TGF)-β1 as a master regulator of fibroplasia via its interactions with fibroblasts to induce proliferation, migration, adhesion, apoptosis, and extracellular matrix (ECM) metabolism.^4,6–10 These fibroplastic activities are largely mediated via Smad3, a receptor-activated protein. Ligand binding to the receptor phosphorylates Smad3 leading to heterodimerization and nuclear translocation to regulate multiple transcriptional events. Our laboratory demonstrated Smad3-dependent regulation of TGF-β1-mediated cellular activities in vocal fold fibroblasts and downstream transcriptional events following Smad3 knockdown in vitro. We also described the temporal pattern of Smad3 expression following vocal fold injury in vivo. Specifically, Smad3 expression increased 3 days following injury with maximal expression 7 days following vocal fold injury.^4–6,11 These data provide some insight into temporal opportunities for intervention targeting Smad3 and serve as the foundation for the timing of treatment in the current study. Globally, we hypothesize that Smad3 is an ideal therapeutic target in vivo, as modulating this pathway can likely redirect wound healing toward a more regenerative, less fibrotic phenotype.

Our laboratory recently employed RNA interference (RNAi) via short interfering RNA (siRNA) to alter Smad3 expression in vitro.^4,6 siRNA can temporarily attenuate expression of target genes by triggering sequence-specific mRNA degradation.^12,13 In vivo, systemic administration of Smad3 siRNA, however, is likely associated with morbidity. Animals deficient in Smad3 develop osteoarthritis and humans with Smad3 mutations are at risk for aortic aneurysms and dissections, as well as skeletal abnormalities.¹⁴ In contrast, localized delivery of siRNA oliognucleotides could circumvent off-target effects associated with systemic delivery. Disease states involving defined tissues such as the eye, skin, and certain types of tumors are amenable to localized siRNA treatment. Several siRNA-based late-stage clinical trials currently employ localized delivery strategies. For example, an industry-funded, dose escalation trial of localized delivery of QPI-1007 for non-arteritic anterior ischemic optic neuropathy and glaucoma (NCT01064505) recently closed. Similarly, intralesional injection of TD101 siRNA for pachonychia congenital had favorable outcomes in a Phase Ib trial (NCT00716014). Given their anatomic location and accessibility, the VFs are an ideal model for local administration of siRNA to ensure precise targeting of Smad3 with minimal off-target toxicity. To date, this approach has not been described in the vocal folds.

Additionally, transport into target cells remains challenging. siRNA oligonucleotides administered without delivery media are vulnerable to rapid hydrolytic cleavage and cellular uptake may be attenuated due to the molecular weight and polyanionic character of siRNA oligonucleotides.¹⁵ Transfection efficiency could be enhanced by carrier molecules to facilitate passage of siRNA across cell membranes, reducing the concentration required for optimal therapeutic benefit.^6,16–19 Due to their amphiphilic character, cationic lipids are a class of carrier molecules well-suited for intracellular delivery of nucleic acid cargo. Electrostatic association of carrier molecules with polyanionic molecules such as siRNA, and the ability of the resulting complexes to cross hydrophobic cell membranes, enhance cellular uptake.^20–22 Despite these delivery capabilities, most commercial cationic lipid transfection reagents are unattractive for clinical application, as their design has not been optimized for in vivo use.^6,23

Peptoids are a class of peptidomimetic, N-substituted glycine oligomers with a range of biomedical applications.^24–27 A family of sequence-specific peptoids bearing cationic side-chains and conjugated to phospholipid tails have been investigated as transfection agents and are referred to as “lipitoids”.²⁸ The structure of a representative lipitoid is shown in Figure 1. Lipitoids share many characteristics with cationic lipids, however, they have particularly desirable properties including reduced cytotoxicity, high transfection efficiency, and resistance to proteolytic degradation.^28–32 Lipitoids have exhibited nucleic acid transfection capabilities in a variety of cell types, including primary cells, but have not previously been validated in vivo.

Figure 1.

Structure of the lipitoid oligomer. The term lipitoid can refer generally to a family of N-substituted glycine peptoid oligomeric transfection reagents bearing multiple cationic side chains and a phospholipid moiety. Lipitoids are capable of forming nanoparticle complexes with nucleic acids to enhance cellular uptake. For the current study, lipitoid refers specifically to the oligomer depicted.

Our laboratory previously described the in vitro utility of lipitoid; specifically, lipitoid outperformed traditional transfection reagents for gene knockdown in vocal fold fibroblasts.⁶ In the current study, we sought to establish the utility of localized, in vivo siRNA therapy to alter wound healing in the vocal folds via lipitoid delivery. Cytotoxicity of lipitoid, as well as the optimal conditions for siRNA-lipitoid binding, were established in vitro in order to optimize translation to clinical application. The in vivo effects of lipitoid-bound siRNA on Smad3 mRNA expression were then quantified in a rabbit model of vocal fold injury to serve as a foundation for future investigation regarding the functional implicaitons for localized gene manipulation. Ultimately, we seek to identify optimized lipitoid sequences that lead to physiologically-targeted therapeutics for the millions of patients with voice disorders.

MATERIALS AND METHODS

In Vitro

Lipitoid Synthesis

Synthesis of the lipitoid (2140.7g/mol molecular weight, determined by mass spectrometry) was described previously and purified via high-performance liquid chromatography. Lipitoid has been previously shown to condense plasmid DNA into uniform particles 50–100nm in diameter.²⁸

Cell Lines

An immortalized human vocal fold fibroblast cell line (HVOX) developed in our laboratory was employed.¹⁰ In addition, primary rabbit vocal fold fibroblasts (RVFF) were provided from the Rousseau Laboratory at the Vanderbilt University School of Medicine.

Cell Proliferation and Viability

HVOX and RVFF were grown in 96-well plates using Dulbecco’s Modified Eagle Serum (DMEM; Life Technologies, Carlsbad, CA) with 10% Fetal Bovine Serum (FBS) (Life Technologies) and 1% antibiotic/antimycotic (Life Technologies). Both cell lines were then treated with the experimental concentration of lipitoid or Lipofectamine® (Invitrogen Thermo Fisher Scientific, Waltham. MA) for 24 hours in antibiotic free media. The media was aspirated and the cells were treated with 100μL of complete media and 20μl of CellTiter 96^® Aqueous One (Promega, Fitchburg, WI) solution for 2 hours. Absorption was quantified at 490nm.

LIVE/DEAD Assay

HVOX grown to confluence in a six well plate were treated with lipitoid or Lipofectamine® (Invitrogen) in DMEM with 10% FBS for 24 hours. Cells were then harvested and labeled employing the LIVE/DEAD^® Cell-Mediated Cytotoxicity Kit (Molecular Probes^®, Thermo Fisher Scientific, Waltham, MA) following standard protocols. Cell images were analyzed and counted using a Nikon TIRF/Epi-Fluorescent Microscope (Nikon, Tokyo, Japan).

Transfection/Incubation times

HVOX were grown to 80% confluency and 5μM of siRNA was combined with 1.07mg/mL of lipitoid in 500μL of Opti-MEM (Life Technologies, Carlsbad CA). Opti-MEM was included in the current study as it is the recommended reagent for use with Lipofectamine®. These solutions were incubated at room temperature for 20 minutes, 2 hours, and 24 hours, respectively, for each sample. The solution was then added to cells with 1.5mL DMEM and 10% FBS. After 24 hours, mRNA was extracted and subjected to quantitative polymerase chain reaction.

Transfection/Incubation solutions

HVOX were grown in a 6-well plate until 80% confluency and 5μM of siRNA was combined with 1.07mg/mL of lipitoid in 500mL H₂O, deionized H₂O, 1x Phosphate Buffer Saline (PBS), 10x PBS, or Opti-MEM. This solution was incubated for 20 minutes and added to the cells with 1.5mL DMEM and 10% FBS. After 24 hours, mRNA was extracted and subjected to quantitative polymerase chain reaction.

Quantitative Polymerase Chain Reaction

mRNA was harvested employing the Qiagen RNeasy Kit (Qiagen, Valencia, CA) according to manufacturer’s protocols and quantified with the NanoDrop 2000 UV-Vis Spectrometer (Thermo Scientific, Wilmington, DE). The Taqman High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Grand Island, NY) was used to perform quantitative reverse transcriptase-polymerase chain reaction (RT-PCR). Gene sequences for Smad3 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were obtained in the form of Taqman gene expression assays (Applied Biosystems). Quantitative RT-PCR was performed on the Applied Biosystems StepOne Plus Real-Time PCR System following manufacturer’s protocols. The ΔΔCt method was used with normalization via GAPDH.

Statistical analysis

All in vitro experiments were performed in triplicate at minimum. One-way analysis of variance (ANOVA) was employed. If the main effect was significant at p<0.05, post hoc comparisons were performed via the Scheffé method using StatView 5.0 (SAS Institute, Berkeley, CA). Data are expressed as means ± standard error.

In Vivo

Vocal Fold Injury

All procedures were approved by the Institutional Animal Care and Use Committee at the New York University School of Medicine. Thirty adult, 2–3kg New Zealand white rabbits were randomly assigned to 1) no injection (control), 2) lipitoid + nonsense siRNA, and 3) lipitoid + Smad3 siRNA (n=10 in each group). The control group did not undergo any injection to represent an untreated clinical cohort; the nonsense siRNA group served injection control. Each group was further categorized according to time of sacrifice: 4 hours and 24 hours after injection (each n=5). All animals received intravenous injection of ketamine (70mg/body) and xylazine (2mg/body) for induction of anesthesia. Inspired isoflurane was added for paralytic maintenance. The animals were placed supine on an operating table and the larynx was visualized via a slotted laryngoscope (Size 1; Karl Storz, Flanders, NJ). Under the guidance of 2.7mm, 0° or 30° telescope (Karl Storz), the vocal folds were injured unilaterally by separating the lamina propria from the thyroarythenoid muscle via insertion of a 22-gauge needle at the lateral portion of right vocal fold, followed by removal of the lamina propria with microscissors and microforceps. This model and the surgical technique have been described previously by our group.^11,33

Transfection Solutions

10μL of 5μM Smad3 siRNA was complexed with 10μl of 210μM lipitoid diluted in deionized water/1x PBS buffer by adding the siRNA to the lipitoid solution. The final injected lipitoid concentration was determined via 1.07mg/mL (initial lipitoid concentration in H₂0) x 5μL=5.35μg in 20μL of 1x PBS. This solution was allowed to stand at room temperature for 20 minutes to optimize nanoparticle dimension, as previously described,³⁴ but also based on our in vitro findings described in this manuscript. In addition, siRNA-lipitoid complexes were formed at an approximate lipitoid to siRNA molar ratio of approximately 42:1, establishing a positive/negative charge ratio of 3:1. Nonsense siRNA solution was prepared similarly, according to the manufacturer’s protocol.

Vocal Fold Injection

Seven days after injury, animals underwent anesthesia and laryngeal visualization as described above. The cricothyroid approach was then employed to inject 20μl of 0.1mg/kg administered dose of nonsense siRNA or Smad3 siRNA complexed with lipitoid into the subepithelial layer of the scarred right lamina propria using a microinjection system (Hamilton Company, Reno, NV) with a 27-gauge needle. The location of the injection site was confirmed visually via marked distention of the vocal fold mucosa. At 4 or 24 hours following injection, animals were sacrificed and the larynges were harvested and immersed in RNAlater RNA Stabilization Regent (Qiagen Inc., Valencia, CA) and stored at −80°C until analysis.

Quantitative Real-Time Polymerase Chain Reaction

The bilateral vocal folds were dissected under magnification. mRNA extraction, quantification of mRNA, and RT-PCR were performed as previously described. Expression was presented as fold change compared to the left (uninjured) vocal fold in the control cohort. The uninjured, left vocal fold served as an internal control.

Statistical Analyses

Two-way ANOVA followed by the Scheffé post hoc test was employed to investigate differences in gene expression at each time point. When interactions were present between treatment and time point, a one-way factorial ANOVA followed by the Scheffé post hoc test were performed. Statistical significance was defined as p<0.05. All data are expressed as means ± standard error.

RESULTS

In Vitro

Cell Proliferation and Viability

Both Lipofectamine and lipitoid reduced viability of rabbit and human fibroblasts compared to control; however, no significant differences were observed between transfection reagents for either cell type. In human immortalized fibroblasts (Figure 2A), 5mg/μL and 10mg/μL of Lipofectamine (Invitrogen) resulted in cell viability of 80% and 69% relative to the control group, respectively. This dose effect was not present in response to lipitoid; cell viabilities of 69% and 67% were observed at 5mg/μL and 10mg/μL, respectively. In addition, this dose response was not observed in primary RVFFs treated with Lipofectamine (Invitrogen); 78% and 76% at 5mg/μL and 10mg/μL, respectively (Figure 2B). RVFFs treated with lipitoid also resulted in minimal difference between doses; 66% and 64% at 5mg/μL and 10mg/μL, respectively. To further quantify cytotoxicity in human vocal fold fibroblasts, Dead/Live data were standardized as a percentage of control cell survival (Figure 2C–E). A significant dose response (p=0.0398) was noted with Lipofectamine (Invitrogen); percent alive was 86% and 69% at 5mg/μL and 10mg/μL, respectively. Lipitoid treatment resulted in survival rates of 82% and 74% for 5mg/μL and 10mg/μL, respectively, lacking a significant dose response (p = 0.4430). No significant differences in cell survival were noted between lipitoid and Lipofectamine (Invitrogen) across concentrations.

Figure 2.

Cell viability as a function of Lipofectamine and lipitoid concentration in human vocal fold fibroblast (A) and rabbit vocal fold fibroblast (B). Results of Dead-Live assay displaying the percentage of live human vocal fold fibroblasts when treated with Lipofectamine and lipitoid relative to control (C); Dead-Live assay of lipitoid treated cells; green=alive, red=dead (D). All data shown as percent control; *p<0.05.

Transfection Conditions

Smad3 knockdown as a function of media was quantified. Twenty minute incubation in H₂O, deionized H₂O (dH₂O), Opti-MEM, 10x PBS, and 1x PBS yielded RQ values of 0.40, 0.24, 0.16, 0.16, and 0.14, respectively (Figure 3A). Transfection with 1xPBS resulted in the greatest Smad3 knockdown; this difference (as for all of the experimental solutions tested) was significant relative to H₂O (p<0.0001 for each). However, differences between 1xPBS and dH₂O, Opti-MEM, and 10x PBS did not reach statistical significance. The influence of siRNA + lipitoid incubation time was then explored. Incubation times of 20 minutes, 2 hours, and 24 hours resulted in RQ values of 0.36, 0.45, and 0.52, respectively, compared to the control sample (p<0.0001 for all incubation times compared to control; Figure 3B). However, no significant differences between incubation times were noted (p=0.1253 for 20 minutes compared to 24 hours).

Figure 3.

Percentage of Smad3 expression after siRNA transfection with lipitoid as a function of siRNA/lipitoid reagent (A) and incubation time (B).*p<0.05 relative to non-transfected control, #p<0.05 relative H₂O.

In Vivo

Lipitoid-complexed siRNA suppressed Smad3 gene expression after vocal fold injury

An established rabbit model of vocal fold injury was employed to investigate the Smad3 knockdown efficacy of injection of lipitoid-complexed siRNA. As shown in Figure 4, expression of Smad3 mRNA increased significantly in response to injury. Smad3 expression was significantly reduced in injured vocal folds following injection of lipitoid+Smad3 siRNA at both 4 and 24 hours post-injection (p=0.035 and 0.034, respectively) relative to the control and injection without Smad3 siRNA. Qualitative suppression of Smad3 expression was also observed in the uninjured vocal folds at both time points following lipitoid+Smad3 siRNA injection; this difference was not significant.

Figure 4.

Expression of Smad3 mRNA at 4 and 24 hours after injection of lipitoid-siRNA complex (n=5; p=0.035 at 4 hours and p=0.034 at 24 hours). Smad3 expression was standardized to GAPDH and presented as fold change relative to control (mean +/−SEM). *p<0.05; N/S-non-sense siRNA.

DISCUSSION

Vocal fold fibrosis remains a substantial problem in clinical laryngology. A promising step towards meeting this challenge is in vivo modulation of biological pathways underlying the fibrotic phenotype during wound healing. Suppression of Smad3 has been previously shown to reduce fibrotic activities in vitro,⁴ and, in vivo, Smad3 deficient mice are resistant to bleomycin-induced pulmonary fibrosis, carbon tetrachloride-induced hepatic fibrosis, and glomerular fibrosis.^35–39 Downregulation of Smad3 expression is thought to blunt the TGF-β1-mediated fibroplastic response in fibroblasts and interrupt TGF-β1 signaling in inflammatory cells within the wound milieu.⁴⁰ Smad3 was recently silenced in murine skin using a topical gel, resulting in resistance to radiation-induced cutaneous fibrosis.⁴¹ Cumulatively, therapeutic manipulation of Smad3 likely holds potential for redirecting wound healing towards a regenerative outcome. In spite of the prevalence of voice disorders and relatively direct access to the vocal folds, the current study is the first to describe manipulation of local gene expression in the vocal folds. Although encouraging, these data must be considered preliminary with tempered enthusiasm as the functional implications of localized knockdown are unclear.

Over 50 clinical trials utilizing RNA-based therapeutics were identified in a 2012 review,²³ many of which employed “naked” delivery of siRNA. These trials are limited by both the inefficiency of gene silencing and the necessity to administer elevated quantities of the therapeutic agents.^42,43 Motivated by the goal of in vivo siRNA utility to effectuate biologically-targeted therapies, our laboratory previously performed preliminary in vitro investigation regarding a relatively novel lipitoid/siRNA nanoparticle.⁶ Data from the current investigation demonstrated limited cytotoxicity of this nanoparticle in vitro, optimized lipitoid-siRNA binding conditions, and successful knockdown of Smad3 in vivo via lipitoid-complexed siRNA.

Although not a comparative study, it was critical to ensure that lipitoid had no deleterious effect on metabolic activity, in either primary rabbit vocal fold fibroblasts (RVFF) or in immortalized human vocal fold fibroblasts (HVOX) relative to Lipofectamine® (Invitrogen) in vitro. Both lipitoid and Lipofectamine® reduced cell viability compared to control/untreated cells. However, no significant differences were noted between compounds. Based on these favorable outcomes, we sought to acquire clinically meaningful data regarding optimal conditions prior to application/injection to ensure desirable nanoparticle formation and morphology, and, by extension, effective gene silencing in vivo. Relatively brief incubation (e.g., 20 minutes) in PBS resulted in significant in vitro Smad3 knockdown. Determination of these variables is critical for the eventual progression to clinical application. These conditions were then employed in the in vivo component of the current study.

Lipitoid-mediated delivery of Smad3 siRNA in a leporine model of vocal fold injury was effective. Seven days following injury to the vocal folds, animals underwent cricothyroid approach to injection of Smad3 siRNA complexed with lipitoid under endoscopic guidance. This time point was selected because as previous work from our laboratory showed peak Smad3 mRNA expression seven days following vocal fold injury in a similar model.^5,44 Furthermore, this interval allowed for recapitulation of the epithelium to compartmentalize the injectate. Smad3 gene expression significantly decreased at both 4 and 24 hours after injection of Smad3 siRNA complexed to lipitoid. Although functional outcomes following such genetic manipulation remain to be examined, we hypothesize that suppression of Smad3 using siRNA-lipitoid complex, even for a brief interval, likely holds immense therapeutic potential with minimal toxicity. These preliminary experiments support the therapeutic potential of lipitoid-complexed Smad3 siRNA to modify the wound healing response in the vocal folds.

The current study, however, is not without limitations. Most notably, the physiologic and functional sequelae of Smad3 suppression following injury were not examined. Optimal dosing, timing, and frequency of siRNA administration were also not explored; these issues are critical to the eventual clinical utility of this treatment modality. Furthermore, improvements to the oligomer can be readily implemented to tailor the nanoparticle complex for compatibility with tissue type, delivery method, and siRNA sequence.^45,46 Efficient modular synthesis of lipitoids as oligomeric species allows the physicochemical properties of the transfection reagents to be tailored to meet these objectives through the selection of varying side chain chemical groups from the hundreds of commercially available primary amine synthons compatible with peptoid synthesis. In that regard, the current study did not employ a “naked” siRNA condition; in spite of emerging contrary data, one may hypothesize that delivery of siRNA to the vocal folds without a functional delivery reagent may yield similar outcomes. Similarly, only two time points assayed following injection. Clearly, increased temporal insight is warranted to more fully appreciate the metabolic changes associated with Smad3 knockdown. These limitations speak to the preliminary nature of the current study, and these issues must be addressed prior to increased enthusiasm regarding the outcomes. Finally, these experiments examined a model of early wound healing, despite established fibrosis of the vocal folds leading to the most morbidity among patients. Cumulatively, although encouraging, these data represent an initial attempt at localized gene manipulation and enthusiasm regarding clinical utility must be attenuated pending further investigation.

CONCLUSION

The current study is the first successful manipulation of local gene expression in the vocal folds and provides foundational data for the further in vivo investigation, particularly with regard to the functional sequelae of such intervention. These data may represent progress towards the development of effective therapeutics to direct wound healing in the vocal folds toward a less fibrotic, more regenerative outcome.