Amelioration of Subglottic Stenosis by Antimicrobial Peptide Eluting Endotracheal Tubes

Abstract

Introduction

Pediatric subglottic stenosis (SGS) results from prolonged intubation where scar tissue leads to airway narrowing that requires invasive surgery. We have recently discovered that modulating the laryngotracheal microbiome can prevent SGS. Herein, we show how our patent-pending antimicrobial peptide-eluting endotracheal tube (AMP-ET) effectively modulates the local airway microbiota resulting in reduced inflammation and stenosis resolution.

Materials and Methods

We fabricated mouse-sized ETs coated with a polymeric AMP-eluting layer, quantified AMP release over 10 days, and validated bactericidal activity for both planktonic and biofilm-resident bacteria against Staphylococcus aureus and Pseudomonas aeruginosa. Ex vivo testing: we inserted AMP-ETs and ET controls into excised laryngotracheal complexes (LTCs) of C57BL/6 mice and assessed biofilm formation after 24 h. In vivo testing: AMP-ETs and ET controls were inserted in sham or SGS-induced LTCs, which were then implanted subcutaneously in receptor mice, and assessed for immune response and SGS severity after 7 days.

Results

We achieved reproducible, linear AMP release at 1.16 µg/day resulting in strong bacterial inhibition in vitro and ex vivo. In vivo, SGS-induced LTCs exhibited a thickened scar tissue typical of stenosis, while the use of AMP-ETs abrogated stenosis. Notably, SGS airways exhibited high infiltration of T cells and macrophages, which was reversed with AMP-ET treatment. This suggests that by modulating the microbiome, AMP-ETs reduce macrophage activation and antigen specific T cell responses resolving stenosis progression.

Conclusion

We developed an AMP-ET platform that reduces T cell and macrophage responses and reduces SGS in vivo via airway microbiome modulation.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12195-023-00769-9.

Introduction

Pediatric subglottic stenosis (SGS) is the severe narrowing of the airway below the vocal folds that most commonly occurs after prolonged intubation in intensive care units [1, 13]. Although the etiology of the disease can be either iatrogenic, idiopathic, trauma induced, or autoimmune related in adults [1, 3, 37], children predominately develop iatrogenic SGS following intubation. Pediatric patients incur a 50% greater chance of developing stenosis for every 5 days intubated owing to the semi-permanent presence of the endotracheal tube (ET) that continuously contacts and irritates the epithelial lining of the trachea [32]. Following this initial injury of the airway epithelium, an inflammatory cascade proceeds resulting in infiltration of macrophages [24, 35], CD8⁺ cytotoxic T cells [20], CD4⁺ helper T cells, and γδ T cells that secrete proinflammatory cytokines such as interleukin 1-beta (IL-1β), IL-6, GM-CSF, and IL-17A [10, 18]. Subsequently, fibroblasts in the lamina propria (LP)—the layer between the airway epithelium and cricoid cartilage—are activated and differentiate into myofibroblasts leading to the aberrant secretion of extracellular matrix. In normal wound healing, myofibroblasts would subsequently apoptose allowing the remodeling of the matrix by other fibroblasts reducing the LP to its original thickness; however, in SGS, myofibroblasts remain activated leading to continuous fibrous scar deposition that over time occludes the airway [13].

Current treatments for pediatric SGS span from short term balloon dilation and serial intralesional steroid injections (SILSIs) requiring repeat interventions every 2–12 weeks to invasive surgeries that physically expand the airway with a graft of autologous cartilage [1, 27, 43]. Overall, these procedures are limited in success with a 6–13% failure rate and often result in complications necessitating repeated surgeries [5, 33, 44]. This poor success rate is associated with patients’ comorbidities that are often the primary reason for the initial intubation and make SGS a challenging disorder to understand and treat. In an attempt to find an alternative treatment, attention has been drawn to the microbiome of the upper airway. In recent work, distinct microbial communities have been reported in the upper airway microbiome of adult patients with SGS compared to controls. Specifically, SGS patients exhibited increased Streptococcus and Mycobacterium and decreased Prevotella species abundances, along with changes in the alpha (intra-community heterogeneity) and beta (inter-community dissimilarity) diversity between disease groups [19, 25]. These observations suggest a significant shift in the airway microbial ecology could be an exploitable difference for treatment. Moreover, these findings were observed in both the airway mucosa and LP suggesting that the injury from ETs rupturing the airway epithelial barrier allows bacterial access to deeper tissue layers. Therefore, we hypothesized that modulating the laryngotracheal microbiome could alter SGS progression and function as a druggable target for an innovative, preventative treatment.

In this work, we utilized our drug-eluting ET platform to administer antimicrobial peptides (AMP) to the laryngotracheal space during intubation for localized airway microbiome modulation [3]. This platform is an advantageous method for delivery as it will release our therapeutic AMPs directly to the site of injury since the inducer is the ET itself. AMPs were selected as the antimicrobial agent for their customizability, potency, and potential for selectivity; AMPs also have decreased propensity to develop multidrug resistance compared to conventional antibiotics, therefore improving their clinical relevance [41]. To ensure a successful treatment, we need to reduce SGS associated bacteria while protecting the healthy commensal airway microbes. With this treatment, we aim to modulate the upper airway microbiome by AMPs at the time of intubation-driven injury, which will diminish the inflammatory response resulting from intubation and prevent the progression of SGS.

To test this, we first created a mouse-sized ET, and then coated them with Lasioglossin-III (Lasio)—a previously used broad-spectrum AMP—terming the platform Lasio-ET. The Lasio-ETs yielded a prolonged linear release over 1 week and were stably formulated with other AMPs indicating the robust applicability of this platform. Following tube coating characterization by fluorescent and scanning electron microscopy, we showed strong in vitro gram-positive and gram-negative antimicrobial activity toward three clinically relevant airway bacteria (Streptococcus epidermidis, methicillin-sensitive Staphylococcus aureus, Pseudomonas aeruginosa) and biocompatibility with airway fibroblasts and epithelium. We tested the activity of the Lasio-ETs first in an ex vivo model of excised murine airways, then in a robust in vivo SGS mouse model (Fig. 1). Overall, we demonstrate the ability to locally deliver AMPs via our ET platform and successfully prevent SGS through the microbiome-driven modulation of the T cell and macrophage responses.

Fig. 1.

Schematic of antimicrobial peptide coated endotracheal tubes (AMP-ETs) implanted in vivo to locally modulate the upper airway microbiome as a treatment for subglottic stenosis

Materials and Methods

Materials

Agar, BD Difco Nutrient Broth, BD Bacto Tryptic Soy Broth, BD Bacto Brain Heart Infusion, normal goat serum, bovine serum albumin (BSA), xylenes, and histoclear were purchased from Fisher Scientific. Sodium citrate was purchased from VWR. Dichloromethane (DCM), PLGA (50:50, 7–17 kDa), poly(vinyl alcohol) (PVA), albumin–fluorescein isothiocyanate conjugate (FITC–BSA), NaOH, NaCl, HEPES, hydrocortisone, l-(−)-glucose, Tween-20, Triton-X100, and hydrogen peroxide were purchased from Sigma Aldrich. Fetal bovine serum (FBS), Antibiotic–Antimycotic (Anti–Anti), Dulbecco's Modified Eagle Medium (DMEM), minimum essential amino acids (MEM NEAA), Ham's F12 medium, GlutaMAX 100 ×, and Insulin–transferrin–selenium 100 × were purchased from ThermoFisher Scientific. A4K14-citroin 1.1 and Lasio were purchased from Genscript.

ET Tube Fabrication

1 mm diameter cylindrical rods were punched from plasticized poly(vinyl chloride) (McMaster–Carr) with a 1 mm biopsy punch at a length of 8 mm (Supplementary Fig. S1). This diameter was selected based on murine trachea diameter; however, these conditions made it impractical to punch a lumen. Nonetheless, these cylindrical rods are henceforth referred to as ETs. For coating, an emulsion was prepared as previously described [3]. In brief, 1 mL of 1 mg/mL 50:50 7000–17,000 kDa poly(lactic-co-glycolic) acid (PLGA) in DCM was combined with 50 or 200 µL of 1 mg/mL FITC–BSA dissolved in 1% polyvinyl acetate (PVA). The solutions were then subjected to 20 s of a 40J sonic dismembrator (Fisher Scientific) at 25% amplitude. ETs were subsequently dipped thrice for 10 s each, allowing > 20 s between dips to allow coating drying, and lyophilized overnight to remove all residual solvent. For AMP-ETs, Lasio and A4K14-citropin 1.1 (Citropin) were prepared at 8 mg/mL in 1% aqueous PVA and used at 200 µL per emulsion. Tubes were stored in desiccant conditions until use.

Surface Characterizing Microscopy

Surface characterization was conducted on FITC–BSA–ETs via fluorescent microscopy on a BZ-X810 All-in-One Fluorescent Microscope with a GFP filter (Keyence) with identical exposure times. Fluorescence intensity was quantified from fluorescent micrographs by ImageJ averaging 20 random regions on the surface of the tube. Uncoated, PLGA only, and AMP coated ETs were then imaged on a FEI Quanta 600 FEG Mark II ESEM with a 5 kV accelerating voltage to visualize tube surface topography.

Release Profile and Total Encapsulation

Lasio-ETs were placed in a 96 well round-bottom plate and immersed in 150 µL of PBS to monitor release. The study was conducted at 37 °C shaking at 100 rpm in humidified conditions for 7 days, completely replacing the solution at 1, 4, 24, and every 24 h following. Quantification of peptide was assessed by Micro BCA Protein Assay Kit (ThermoFisher Scientific) and absorbance measured at 562 nm on a Synergy H1 Microplate Reader (BioTek, Vermont). Total encapsulation was achieved by combining Lasio-ETs with 50 mM NaOH for 3 days at 37 °C and measuring protein content as previously described. PLGA only ETs were employed as controls for release and total encapsulation. All release and total encapsulation experiments were performed with four fabrication replicates.

Antibacterial Activity

Staphylococcus epidermidis (Winslow and Winslow) Evans (ATCC 13990) was cultured in nutrient broth, Escherichia coli (Migula) Castellani and Chalmers (ATCC 25922) and P. aeruginosa (Schroeter) Migula (ATCC 27853) were cultured in trypticase soy broth, and methicillin-sensitive S. aureus subsp. aureus Rosenbach (ATCC 25904) was cultured in brain heart infusion broth aerobically at 37 °C. Minimum inhibitory concentrations (MICs) and in vitro inhibition were determined as previously described [3]. In brief, MICs were determined by adding 75 µL of serial diluted 2 × peptide resulting in a final concentration range of 0.3125–80 µM and mixing with equal volume of diluted bacteria (OD₆₀₀ = 0.001) into each well. MIC was then determined by visualizing the lowest concentration that completely inhibited all bacteria growth. Experiments were conducted for five separate colonies with three technical replicates. For AMP-ET bacterial inhibition assessment, AMP, PLGA only, and uncoated ETs were placed in 96 well round-bottom plates and mixed with 150 µL of diluted bacteria (OD₆₀₀ = 0.001) for 24 h at 37 °C shaking at 100 rpm. To quantify planktonic bacteria, 100 µL of the supernatant was removed from each well and measured for OD₆₀₀. Adherent bacteria were quantified by sonicating ETs in 250 µL of HEPES/Saline (70 mM NaCl, 0.75 mM Na₂PO₄, 25 mM HEPES) for 15 min and serial diluting the supernatant onto appropriate agar plates. Colonies were counted by an experimental blinded researcher using ImageJ software.

In situ bacterial inhibition was conducted on excised laryngotracheal complexes (LTCs) from euthanized mice by implanting uncoated, PLGA only, or Lasio-ETs in them and culturing in nutrient broth for 24 h at 37 °C shaking at 100 rpm. Viable bacteria on the tube were quantified as described above.

Biocompatibility

Human vocal fold fibroblasts were a kind gift from the laboratory of Dr. Susan Thibeault and cultured in supplemented DMEM (10% FBS, 2% Anti–Anti, 1% MEM NEAA). NL20 bronchus epithelial cells (CRL-2503) were cultured in supplemented Ham's F12 medium (1.5 g/L sodium bicarbonate, 2.7 g/L glucose, 2.0 mM l-glutamine, 0.1 mM nonessential amino acids, 0.005 mg/mL insulin, 0.001 mg/mL transferrin, 500 ng/mL hydrocortisone, and 4% FBS). Biocompatibility was assessed as previously described [3]. In short, fibroblasts were seeded at 2500 cells/well and epithelial cells at 5000 cells/well in a 96 well plate and allowed to adhere overnight. At the same time, AMP, PLGA only, and uncoated ETs were placed in 150 µL of appropriate cell culture medium and allowed to release at 37 °C shaking at 100 rpm for 24 h. Then, 100 µL of releasate was added to a well of cells in addition to pure AMP at 10 µM for comparison against blank media and 20% DMSO serving as controls, respectively. After 24 h of incubation, biocompatibility was quantified using Invitrogen AlamarBlue HS Cell Viability Reagent (ThermoFisher Scientific) and quantified by fluorescence at excitation/emission 560/590 nm. To further test the biocompatibility of the tube, cells were seeded as described above, allowed to adhere overnight, and cultured directly with each tube condition for 24 h prior to determining viability by AlamarBlue HS Cell Viability Reagent.

In Vivo SGS Model

30 C57BL/6 mice 8-week of age were purchased from Jackson Laboratories following IACUC 01184 at the Corporal Michael J. Crescenz Philadelphia Veterans Affairs Medical Center Institutional Animal Care and Use Committee. All C57BL/6 mice were housed in a temperature-controlled specific-pathogen-free facility under 12-h light/dark cycles. This number of animals was selected to provide six replicates for each condition. Mice were randomly separated into the following groups that would receive either sham surgery or SGS induction and ET condition. The conditions to be tested were as follows: (1) no ET/sham, (2) uncoated ET/sham, (3) no ET/SGS, (4) uncoated ET/SGS, and (5) Lasio-ET/SGS. SGS induction was conducted following our previously established subcutaneous SGS mouse model where tracheas are excised from a donor mouse, induced with stenosis, and then implanted in the subcutaneous flap on the dorsum of an acceptor mouse to allow stenosis progression over 7 days [21, 22]. In brief, three donor tracheas were excised from euthanized mice with their tracheas cut at the vocal folds. To induce SGS, a 0.02-in. twisted wire brush (McMaster–Carr) was inserted into the trachea and brushed 15 times. Next, an incision was made at the eighth tracheal ring completely removing the trachea. If the trachea was in a group with an ET, the tube was then placed inside the lumen of the excised trachea. Simultaneously, one acceptor mouse was anesthetized with ketamine/xylazine and implanted with the three tracheas in a deep subcutaneous flap on the dorsal side of the mouse. The incision was glued closed with Ethicon-Dermabond (Medline) and monitored for 7 days. At 7 days, mice were euthanized and tracheas excised from the dorsal pockets.

Histology and Immunohistochemistry

Excised tracheas were fixed in 10% buffered formalin for 1 h at room temperature and serially dehydrated in ethanol diluted in PBS (25, 50, 75, 95, 100%). Following clearing in xylenes, tracheas were paraffin embedded and sectioned at 5 µm between the cricoid cartilage and fourth tracheal ring. Hematoxylin and eosin staining and Masson’s trichrome were conducted following slide rehydration. LP thickness was measured at four different points around the trachea and averaged for each trachea.

Immunohistochemistry for T cells and macrophages was conducted on sections as follows. Slides were prewarmed at 60 °C for 1 h, cleared in Histoclear twice, and serially rehydrated. Slides for T cell staining were exposed to 30 min of proteinase K (Qiagen 19131) diluted 1:1000 in PBS and those for macrophages were placed in citrate buffer (10 mM sodium citrate, 0.05% Tween-20, pH 6) for 2 h at 60 °C for antigen retrieval. Endogenous peroxidase was blocked by incubating the slides for 30 min in 0.3% hydrogen peroxide at room temperature. Slides were then blocked with blocking buffer (5% goat serum, 3% BSA in 0.01% TritonX-100 in PBS) for 2 h at room temperature. Rabbit anti-mouse CD3ε (Cell Signaling 78588) was used for T cells and rabbit anti-mouse F4/80 (Cell Signaling 70076) was used for macrophages, both diluted 1:400 in blocking buffer. Slides were then placed at 4 °C overnight. Slides were then washed thrice for 5 min each in PBS and incubated with HRP conjugated goat anti-rabbit IgG (Abcam 6721) diluted 1:1000 at room temperature for 1.5 h. Following washing in PBS three times for 5 min each, slides were developed with a DAB substrate kit (Abcam 64238) until minimal background was observed on a slide that was not incubated with primary antibody. Slides were then mounted in limonene and imaged.

Statistical Analysis

All data were graphed and reported as means ± standard deviation. For antibacterial activity analysis, ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons was used on N = 6 per condition. Biocompatibility was also analyzed with ANOVA and Tukey’s multiple comparisons for N = 5. LP thickness was measured using ImageJ at four locations on the trachea, averaged, and compared using ANOVA with Tukey’s multiple comparisons for N = 5. T cells and macrophages were counted by an individual blinded to slide condition or treatment and quantified on two slides per trachea, one form the first tracheal ring and one from the third ring. Comparisons were conducted using ANOVA with Tukey’s multiple comparisons for N = 5. Plots of immune cells versus LP thickness were prepared on R 4.3.0 using the Tidyverse 2.0.0 package and fit for ellipses using stat_ellipse(). Data analyses were performed using GraphPad Prims Software Version 9.3.1.

Results and Discussion

Murine AMP-ET Fabrication and Coating Characterization

In our previous work, we created a peptide-eluting platform by coating ETs with an emulsion of AMP and PLGA and demonstrated the ability to inhibit planktonic and biofilm bacterial growth in vitro. With this technology in hand, we aimed to test our hypothesis that modulating the upper airway microbiome would alter SGS progression. Since previous studies have shown differences in the microbiome in adults, we aim to test our hypothesis by modulating the bacterial populations in the upper airway as an innovative therapeutic approach [25]. To test this in vivo, we aimed to implant our ETs in murine tracheas for up to 7 days in a previously established SGS mouse model and therefore needed to downsize the platform from an ET of 4 mm diameter sized for a child’s airway to a 1 mm diameter to fit a murine airway. Mouse sized ETs were fabricated by punching plasticized polyvinyl chloride with a 1 mm diameter biopsy punch to create a cylinder. This ET was too small in diameter to punch a lumen; therefore, we would need to utilize our subcutaneous SGS model that we have previously validated [21, 35]. To assess the coating procedure on this smaller tube, we first used fluorescent bovine serum albumin (FITC–BSA) as a model protein by dipping tubes thrice in an emulsion of 1 mg/mL PLGA in DCM and 1 mg/mL FITC–BSA in 1% aqueous PVA. Importantly, when increasing the volume of protein containing aqueous phase from 50 to 200 μL in the coating emulsion, we observed a corresponding 4-fold increase in fluorescence intensity (Fig. 2A, B; Supplementary Fig. S2). Based on our previous studies comparing FITC–BSA to AMP loading, we decided to fabricate all AMP-ETs with 200 µL peptide aqueous phase to achieve a clinically relevant concentration considering the reduction in surface area.

Fig. 2.

Mouse size ETs were successfully coated in FITC–BSA when incorporating A 50 µL and B 200 µL of aqueous phase in the coating emulsion (scale bar = 1 mm). C Release of FITC–BSA reveals a burst dominated release while the AMP Lasioglossin demonstrated a 1 h burst release (inset) followed by a continuous linear release of 1.16 µg/day for 7 days (N = 4, r² = 0.9696). Scanning electron micrographs of D uncoated, E PLGA coated, and F Lasio-ETs show increased stability of the coating with AMP inclusion as observed with a smoother topography (×100 magnification inset scale bar = 500 µm and × 5000 magnification scale bar = 10 µm)

It is well established that the different properties of amino acids in peptides define their activity and selectivity, as normally determined by structure activity relationships [2]. For this study, the AMPs Lasio [42] and Citropin [14] were selected for incorporation into the AMP-ET platform as their antibacterial activity has been examined to exhibit broad spectrum activity and general biocompatibility with the host’s cells [4]. Additionally, they cover a range of physiochemical properties such as sequence length, formal charge, and logD (hydrophobicity) (Table 1). Given that ETs are typically replaced weekly in the ICU [3], the ideal release profile for this technology would be a two-phase profile: (1) an initial burst release to quickly alter the host microbiota followed by (2) a linear, sustained release to maintain the community modulation. Therefore, Lasio-ETs were placed in PBS at 37 °C and the concentration of Lasio present in the supernatant was quantified over the course of 1 week with periodic sampling. Our results showed that while FITC–BSA released via burst-dominated behavior, Lasio eluted from the ET at a linear rate of 1.161 ± 0.04 μg/day following an initial burst release lasting 4 h, which equates to 3.16 ± 0.11 μM/day with complete solution replacement (Fig. 2C, r² = 0.9696). Additionally, 94.61 ± 5.33% of the encapsulated AMP released over the span of 1 week, which matches clinically relevant timeframes.

Table 1.

Antimicrobial peptide properties and antimicrobial activity

Peptide	Sequence	Length	Formal charge	LogD	MIC (µM)
					S. epidermidis	S. aureus (MSSA)	P. aeruginosa	E. coli
Citropin	GLFAVIKKVASVIKGL-NH2	16	+ 4	− 6.39	0.3125	10	40	2.5
Lasioglossin	VNWKKILGKIIKVVK-NH2	15	+ 6	− 10.51	0.625	20	20	40

Based on previous studies, we would expect that a platform of this geometry would release a molecule in a burst-phase dominated pattern followed by minimal sustained release similar to what has been shown with drug eluting stents [15, 16]; however, to our surprise this was not observed with our platform. We hypothesize the primarily linear release is a result of interaction between the amphipathic peptide and PLGA resulting in a more controlled release from the surface. Specifically, we expect the charged, basic residues in the form of lysine in positions 4, 5, 9, 12, and 15 of Lasio to interact with the PLGA and slow its release from the coating (Table 1). This has previously been observed in PLGA microparticles where the charged residues within peptide sequences exhibited hindered release rates as the peptides attempt to diffuse outward from the microparticle [7]. Additional studies showed increasing molecular weight of PLGA from 7–17 kDa as we used in this study to 43 kDa significantly prolonged the release. Therefore, we would expect that if a modified profile were to be required to release longer for our Lasio-ET, we would just need to utilize a higher molecular weight PLGA.

To further characterize the surface and robustness of our platform, coatings were prepared with Citropin in addition to Lasio and assessed by scanning electron microscopy. Micrographs revealed a smoother topography in both AMP-ETs compared to uncoated or PLGA only coated controls suggesting peptides increase the coating’s stability (Fig. 2D–F). Additionally, there was no appreciable difference in coating between Lasio and Citropin indicating the robust nature of the AMP-ET platform across different peptides that do not alter coating stability. This finding suggests most peptides could be incorporated into our platform thus significantly increasing its clinical relevance as different bacteria could be associated with SGS in pediatric patients. Furthermore, different patients can have acute or chronic infections when they are intubated either during surgery, in the intensive care units, or with a permanent tracheostomy. Therefore, having a robust platform that can contain multiple different peptides offers increased applicability to treating pediatric patients in a targeted and potentially more personalized manner.

AMP-ETs Inhibit Airway Bacteria In Vitro and Ex Vivo

In humans, the upper airway is normally inhabited by a plethora of microbes, some of which include Staphylococcus, Streptococcus, Moraxella, Neisseria, Prevotella, Dolosigranulum, and Haemophilus species [12]. During infection, however, the microbiome can become imbalanced resulting in abnormally elevated levels of pathogens. A previous study showed that pediatric patients exhibited greater populations of S. aureus and P. aeruginosa in their nares and tracheal aspirate, which are pathogens that can lead to cartilage infection and increased stenosis intensity [27]. Additionally, patients with SGS often experience gastroesophageal reflux that increases tracheal aspiration thus increasing the probability of E. coli colonizing the upper airway [27]. Therefore, the gram-positive bacteria S. epidermidis and the methicillin-sensitive S. aureus (MSSA) and the gram-negative pathogens P. aeruginosa and E. coli were selected as model airway bacteria for assessment of antibacterial activity of the AMP-ETs.

Importantly, both planktonic inhibition and biofilm inhibition must be assessed to simulate the natural environment of the trachea where bacteria exist in the laryngeal mucosa and form robust biofilms on ETs. Planktonic inhibition relies on diffusion of the AMP from the tube. Lasio-ETs reduced planktonic bacterial growth to 0.04 ± 0.39% for S. epidermidis, 14.86 ± 2.26% for MSSA, 40.04 ± 3.66% for P. aeruginosa, and 88.69 ± 7.94% for E. coli (Fig. 3A, C, E, G). Similarly, Citropin-ETs reduced planktonic bacterial growth to 0.13 ± 0.33% for S. epidermidis, 18.35 ± 8.35 for MSSA, 51.81 ± 5.72% for P. aeruginosa, and unlike Lasio-ETs, − 0.058 ± 0.48% for E. coli (Fig. 3A, C, E, G). This shows the ability for AMP-ETs to differentially inhibit bacterial growth depending on the species, which is important as our platform serves to modulate the bacterial communities present in the airway and not eliminate all populations, such as those that are commensal organisms that positively influence airway health [12].

Fig. 3.

Bacterial inhibition of uncoated, PLGA only, AMP-ETs, and free peptide at the respective MIC were tested against each bacterium. Inhibition against S. epidermidis, a gram-positive microbe, reveals A significant reductions in planktonic viability for both Lasio and Citropin-ETs as well as a B 4-log reduction in adherent viability per tube. Inhibition against methicillin-sensitive S. aureus, a common gram-positive airway bacterium, showed C significant inhibition for both peptides and a D significant 1-log reduction in adherent viability per tube. Pseudomonas aeruginosa, a gram-negative airway pathogen, was E partially inhibited planktonically by each peptide and exhibited F 10-fold reduction in adherent bacterial populations per tube. E. coli, a gram-negative pathogen likely in the airway following gastrointestinal reflux or aspiration, was G significantly inhibited by Lasio and completely inhibited by Citropin-ETs with H partial inhibition of adherent bacterial communities on each tube. N = 5, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 for all panels

When assessing biofilm inhibition on the ETs, similar trends were observed as seen with planktonic inhibition by Lasio-ETs by eliciting a significant 4-log reduction in S. epidermidis, and 1-log reduction in MSSA, P. aeruginosa and E. coli (Fig. 3B, D, F, H). Citropin-ETs elicited significant reductions identical to that for S. epidermidis, MSSA, P. aeruginosa, and a 5-log reduction in E. coli (Fig. 3B, D, F, H). Reducing biofilms is imperative in this context as ETs are established to harbor biofilms after a mere 12 h of intubation [47]. When biofilms start to grow on ETs, patients are significantly more likely to experience upper airway infections that can intensify the severity of diseases such as SGS. Notably, our AMP-ET would serve the function of reducing biofilms to decrease the likelihood of airway infection that exacerbates SGS while also modulating the surrounding mucosal bacterial populations to reduce stenosis.

Taken together, these results indicate that AMP-ETs remain active against bacteria after coating and elute from the ETs to kill planktonic bacteria and adherent bacterial communities. In fact, not all bacteria were entirely inhibited demonstrating the ability to modulate present communities. As seen with Lasio and Citropin, different peptides exhibit varying degrees of potency toward bacterial species [40]. This beneficial characteristic implies different peptide sequences could be incorporated into the AMP-ET platform to target specific bacterial taxa while minimizing the impact to the healthy commensal microbiota. For example, the peptide MAD1 could be added to the coating to selectively inhibit M. tuberculosis [40], which have also been implicated to have a role in the development of SGS in adults [19, 30]. Furthermore, it has been reported that 90% of patients with tracheobronchial tuberculosis have some degree of stenosis [38], an alarming incidence rate for patients that could receive a preventative treatment with MAD1-ETs. Additionally, peptides that selectively kill gram-positive [28] or gram-negative [6, 29] bacteria could be useful in treating patients with known bacterial infections during intubation.

AMPs Are Strongly Biocompatible

Prior to placing AMP-ETs in vivo, we assessed biocompatibility toward airway fibroblasts and epithelium cells. AMP-, PLGA only, and uncoated ETs were subjected to 24 h of release in the respective cell culture medium followed by transfer of the releasate onto the cells. After 24 h of treatment, there were no significant decreases in viability of fibroblasts between AMP conditions and controls; unexpectedly, free Citropin significantly increased fibroblast viability compared to all other conditions (Fig. 4A, p < 0.01). When tested against airway epithelial cells, Lasio-ETs and free Lasio exhibited strong biocompatibility with 89.21 ± 5.30% and 102.16 ± 10.33% viability, respectively (Fig. 4B). Citropin-ETs and free Citropin, on the other hand, significantly reduced viability to 61.79 ± 9.25% (p < 0.0001) and 71.92 ± 3.78% (p < 0.05), respectively (Fig. 4B). These viability results were recapitulated when cells were directly incubated with both AMP-ETs indicating the tube itself does not alter biocompatibility (Supplementary Fig. S3). Due to the cytotoxicity of Citropin toward healthy cells, all experiments moving forward were conducted with Lasio. This observation was not completely unexpected as previous research on Citropin revealed moderate toxicity toward human umbilical vein endothelial cells and human erythrocytes [2, 4, 14]. Unfortunately, some AMPs are known to exhibit properties of cytotoxicity in addition to their antibacterial activity; therefore, the propensity to interact with human cells can be high with certain sequences and should always be verified. Lasio, on the other hand, exhibited strong biocompatibility toward the cell types tested in this study and demonstrated limited predicted toxicity toward other eukaryotic cell types [8, 42].

Fig. 4.

Biocompatibility of all tube conditions was assessed against common mammalian cells of the upper airway. A Uncoated, PLGA, and AMP-ETs exhibited strong biocompatibility toward vocal fold fibroblasts, as did free AMP, with viabilities all above 85% (N = 5). B Bronchus epithelial cells were biocompatible what treated with uncoated, PLGA, and Lasio-ETs, but exhibited significant viability loss with Citropin-ETs below 75% (N = 5). C To assess antimicrobial activity in situ, Lasio-ETs and ET controls were inserted into excised mouse laryngotracheal complexes from the vocal folds to the eighth tracheal ring and cultured in nutrient for 24 h revealing Lasio-ETs significantly reduce adherent bacteria within the trachea (N = 5).*p < 0.05, **p < 0.01, ****p < 0.0001 for all panels

Knowing that the Lasio-ET is highly biocompatible, we next assessed the platform’s ability to reduce bacterial adherence in situ. To achieve this, we compared Lasio, PLGA only, and uncoated ETs and inserted them in an excised LTC for 24 h of ex vivo culture. Quantification of viable adherent bacteria revealed a significant 2-log decreases in the Lasio-ET condition compared to PLGA only and uncoated ETs (Fig. 4C, p < 0.0001). Thus, we show that our platform technology directly reduces biofilm formation when placed in the airway, which can reduce hospital-acquired infections that lead to prolonged hospital stays, chronic deficits in lung health, and overall worsened patient outcomes.

Locally Administered AMPs Ameliorate Stenosis In Vivo

To test the effect of locally delivered AMPs on SGS progression in vivo, we implemented a previously described and validated SGS mouse model [21, 22]. Briefly, LTCs are first removed from donor C57BL/6 mice and brushed with a twisted-wire brush to induce stenosis before implantation into a deep subcutaneous dorsal flap of an acceptor C57BL/6 mouse. We have previously shown that this model successfully recapitulates SGS pathogenesis following a mechanical injury, which leads to immune cell infiltration, fibroblast differentiation, matrix deposition, and scar tissue formation. In an additional study by Gosh et al., severe combined immunodeficiency (SCID) mice were used as the LTC acceptor mouse to investigate the role of circulating immune cells and distinguish their role from that of resident immune cells [21]. The results of that study confirmed that circulating immune cells are required for stenosis progression indicated by a significant thickening of the LP, indicating also the formation of neo-vasculature connecting the implanted trachea to the host making this model more similar to the conditions of native trachea. This suggests that the pathology of stenosis is maintained after transplantation. It is however a limitation of this study that the composition and abundances of bacterial communities might be affected by transplantation ex situ and this should be the object of further studies.

The LTC subcutaneous implant model presents a key advantage for our study’s purpose: the ability to incorporate a mouse ET long-term in the LTC since it is placed subcutaneously and does not need to function for respiration. This feature is paramount in our experiment as our ETs are cylinders and are too small to punch a lumen through the middle to permit air flow. Specifically, after airway brushing, ETs were inserted in the airway to test our approach by comparing the following conditions: (1) no tube/sham, (2) uncoated tube/sham, (3) no tube/SGS, (4) uncoated tube/SGS, and (5) AMP-coated tube/SGS. Mice were left for 7 days to allow immune cell infiltration and LP thickening, which have previously been shown to characterize SGS [15, 21]. On day 7, hematoxylin and eosin staining revealed significant thickening of the LP in SGS induced mice compared to the sham controls (p < 0.001), which was prevented when ETs were coated with AMPs (Fig. 5, p < 0.0001). Importantly, the ET did not induce any significant thickening of the LP as seen when comparing no tube/sham and uncoated tube/sham conditions. Interestingly, presence of the ET in an SGS induced airway exhibited significantly reduced thickening of the LP compared to no ET SGS induced airways suggesting that the ET exhibits a stent-like effect (Fig. 5, p < 0.01). Overall, a similar trend was observed after Masson’s trichrome staining which showed increased collagen deposition in the LP in SGS induced mice that received either no tube or uncoated tubes compared to controls and AMP-coated ETs (Supplementary Fig. S4). Due to the previously established antimicrobial activity in vitro and in situ, we believe the reduction in thickness after brushing is a result of microbiome modulation.

Fig. 5.

Histological analysis of mouse tracheas stained by hematoxylin and eosin after 7 days of deep subcutaneous implantation following either sham surgery (unbrushed) or SGS induced (brushed) with or without an ET. No ET [(−)Tube/sham] and uncoated ET [(+)Tube/sham] exhibited thin lamina propria thicknesses, whereas significant thickening of this layer was observed in SGS induced airways with no ET [(−)Tube/SGS] and uncoated ET [(+)Tube/SGS]. This thickening was reduced in Lasio coated ETs [(+)AMP Tube/SGS]. N = 5, × 10 images scale bar = 200 µm, × 40 images scale bar = 50 µm, **p < 0.01, ***p < 0.001, ****p < 0.0001

This is based on evidence provided in other literature indicating Lasio exhibits potent broad-spectrum antimicrobial and anti-fungal activity, but shows minimal effects to healthy cells [8, 42]. The minimal effect on cells is likely a result on their mechanism of action, which is membrane specific that follows pore formation [2]. This mechanism of action depends on two key characteristics of peptides: hydrophobicity which allows them to stably penetrate cell plasma membranes and positive charge which promotes their electrostatic interaction with bacterial cells. Healthy mammalian cells have a net neutral outer-membrane dominated by phosphatidylcholine [4, 26], compared to bacterial cells that have numerous negatively charged components such as lipopolysaccharide, lipoteichoic acid, and peptidoglycan [9, 45]. Therefore, we are confident that AMPs have limited interaction with mammalian cells and preferentially interact with bacteria to elicit their antimicrobial effect, indicating microbiome modulation is the mechanism through which stenosis is reduced.

Further evidence to explain this mechanism involves bacteria and how they interact with epithelial and immune cells in airway mucosa [23, 36]. Bacteria are known to activate various pathways that lead to persistent inflammation when they penetrate the vulnerable airway epithelium following injury through mechanisms such as epithelial-mesenchymal transition [46], mechanotransduction [17], and NLRP3-dependent inflammasome [35]. Although each of these pathways are relevant to SGS through epithelial disruption, inflammasome activation is the most common for recruiting immune cells into the submucosal layer of the airway and causing localized inflammation [39]. The immune cells present during this phase can express pro-healing phenotypes; however, in fibrosis this is not observed, and inflammation continues leading to increased myofibroblast populations and continuous stenosis [13]. Additionally, patients with SGS have increased likelihood for recurrent upper respiratory infections, commonly including S. aureus, P. aeruginosa, and Haemophilus influenzae which often lead to increased inflammation that further intensifies stenosis [27]. Therefore, exploring the immune cell infiltrate into the airway during stenosis in vivo can help explain the immunological role in SGS.

AMPs Modulate T Cell and Macrophage Response After Injury

We then explored how the immune system responds to both ET implantation and SGS induction by brushing. T cells were the first cell population of interest as they respond to antigens from either bacteria, viruses, or other cells [35]. Previous studies have shown T cell infiltration is likely responsible for remodeling the airway epithelium after injury, and, during SGS, leads to thickening of the LP and granulation tissue formation [21, 35]. Specifically, CD8⁺ resident memory T cells have been reported to be significantly enriched in SGS human samples [20], with T-cell receptor sequences exhibiting specificity toward intracellular pathogens. Other studies evidenced that CD4⁺ T cells, whether helper or regulatory, dominate the immune cell population that infiltrates the airway in adult SGS patients [10].

To assess how T cells infiltrate the trachea of mice instilled with AMP-ET and controls, we conducted immunohistochemistry for CD3⁺ T cells on the sectioned mouse tracheas. As expected, after 7 days there were no increases in T cell populations in sham mice with or without an uncoated ET. In SGS induced mice, however, significant increases in T cells were observed in the no ET and uncoated ET condition compared to sham controls (p < 0.01). Notably, these T cell populations were not found in LTCs with AMP-ETs (Fig. 6A; Supplementary Fig. S5, p < 0.0001). When average T cell counts were plotted against LP thickness, distinct clustering was observed co-localizing the Lasio-ETs with sham mice controls (Fig. 6B). Therefore, our combined data on bacterial reduction, lack of T cell infiltration and of SGS in the Lasio condition suggests the reduction of T cell responses occurs through antigen specific mechanisms, which we expect is a result of microbiome modulation. The common antigens that are recognized by T cells are pathogen-associated molecular patterns (PAMPs), which are present either on the surface of bacteria and viruses or get secreted into the extracellular space. Interestingly, a recent study suggests that γδ T cells, a population of T cells that recognize microbial antigens and secrete IL-17A [18], are associated with SGS patients compared to controls [37]. Taken together, Lasio’s ability to reduce bacteria in the upper airway suggests the elimination of PAMPs, which would in turn decrease T cell infiltration resulting in reduced stenosis.

Fig. 6.

Immunohistochemistry was conducted on sectioned mouse tracheas for T cells and macrophages. A CD3⁺ T cells were quantified in the lamina propria revealing SGS increased T cell infiltration significantly, which was reduced with Lasio-ET. B T cell counts were plotted against average lamina propria thickness and clustered based on the variance in each axis revealing overlapping clusters of unbrushed shams with AMP-ET treatment after brushing. C F4/80⁺ macrophages were quantified within the lamina propria revealing significant increases in macrophages in sham mice with an uncoated tube. Macrophages were further increased in SGS airways, which was abrogated when treated the SGS airway with AMP-ETs. D This was further visualized by plotting average macrophages against average lamina propria thickness revealing co-localizing clusters between (−)Tube/sham and (+)AMP/SGS. N = 5, **p < 0.01, ****p < 0.0001

In addition to antigen specific responses, foreign body responses primarily driven by macrophages are also impactful in wound healing and SGS [34]. In a recent work, dysregulated pro-healing macrophages, traditionally referred to as M2 macrophages, were shown to infiltrate at early time points and persist in the trachea at longer time points, an observation that was unexpected and hypothesized to lead to fibroproliferation [24]. Additional research has also shown increases in monocyte-derived macrophage in the trachea in both humans with SGS and in SGS induced mice [35]. Therefore, we conducted immunohistochemistry for F4/80⁺ macrophages on SGS induced and sham mice, with all the conditions from uncoated to AMP-ETs.

As expected for a foreign body response, there was a significant increase in macrophage infiltration in the LP of sham mice with an uncoated ET compared to no tube controls (Fig. 6C; Supplementary Fig. S6, p < 0.01). This recapitulates what is observed in humans when chronic intubation results in macrophage infiltration, which precedes stenosis [11]. After SGS induction in mice, macrophage infiltration is significantly increased compared to the no ET sham surgery control (p < 0.01) and further increased when SGS induced airways are implanted with an ET (p < 0.01). Impressively, macrophage infiltration is completely reduced to the levels similar to that of no ET sham surgery when Lasio is incorporated on the ET (p < 0.01). When average macrophage counts are plotted against LP thickness, a clear overlap is evidenced of the AMP-ETs with the no tube sham condition, which was distinct from the other three conditions without Lasio (Fig. 6D). Therefore, our findings strongly suggest AMPs can reduce macrophage driven foreign body responses to ETs in the airway.

We believe this can be a result of two different observations that have previously been discussed. First, the foreign body response caused by the ET in SGS can become exacerbated during epithelial cell injury, which secrete cytokines that recruit macrophages. Furthermore, epithelial cells perform the essential functions of protecting the LP from bacteria, which during the vulnerable phase after injury can become exposed to the airway bacteria. This would then increase macrophage recruitment to the LP. When AMPs are delivered to the airway following SGS injury, the epithelium can heal properly and produce fewer macrophage-recruiting cytokines and overall reduce the macrophage response to the airway. Second, certain AMPs have been shown to neutralize PAMPs, such as lipopolysaccharide, indicating the possibility that Lasio might bind to PAMPs which would subsequently induce a macrophage driven response [31]. This could also play a role in T cell attenuation, as was previously discussed. Overall, out data indicates that Lasio-ETs modulate the immune response of both T cells and macrophages through mechanisms of microbiome modulation.

Conclusion

Although the upper airway microbiome is only moderately characterized in adult populations with SGS and hardly characterized at all in pediatrics, modulating distinct bacterial communities to alter disease levels opens an innovative option for treatment. To that end, we developed a robust platform to deliver AMPs directly to the site of injury that causes SGS by delivering modulatory molecules that manipulate bacterial populations by differential inhibition of various species. The use of AMPs has the advantage of decreasing the risk to develop resistance mechanisms and to potently target specific bacteria, making them advantageous over conventional antibiotics. Additionally, the local administration of AMPs to the site of injury reduces the likelihood of adverse effects or unwanted patient systemic side effects symptoms. In vivo, we show that the placement of an ET does not induce significant stenosis; however, it does induce a strong foreign body response through F4/80⁺ macrophage infiltration. A strong disease phenotype is observed with and without the tube after brushing the airway, which is also associated to rapid LP thickening and to infiltration of antigen specific CD3⁺ T cells and macrophages. When Lasio is incorporated into the AMP-eluting ET platform, however, SGS is abrogated by broad spectrum modulation of the microbiome to reduce bacterial populations, attenuate the immune response by reducing T cells and macrophages, and overall decrease of the thickness of the LP that defines the disease. Future studies are warranted to confirm the efficacy of our AMP delivery platform in situ without the influence of perturbation of bacterial communities that might follow tracheal transplant. Overall, we establish AMP-ETs as a novel treatment for SGS prevention in pediatric patients to reduce stenosis and improve pediatric health outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

MRA, RCB, INJ, NM, and RG conceived the research plan. MRA, AM, RMF, DDG, HCBN, and KSM carried out experiments. MRA, RMF, and RG prepared figures and wrote the manuscript. All authors reviewed, edited, and approved the final manuscript.

Data availability

Data is available upon request.

Declarations

Conflict of interest

MA and RG are inventors on a pending patent related to the technology described in this manuscript. AM, RMF, DDG, RCB, HCBN, KSM, INJ, and NM declare no conflicts of interest.

Research Involving Human and Animal Rights Statement

All institutional and national guidelines for the care and use of laboratory animals were followed and approved by the appropriate institutional committees. No human studies were carried out by the authors for this article.