Immunocytochemistry assessment of vocal fold regeneration after cell-based implant in rabbits

2.1 In vitro culture and COVR fabrication

The experimental setup mirrored the established protocols that have been previously described.¹⁶ To summarize, hASCs were sourced from the American Type Culture Collection and underwent expansion under conventional culture conditions. COVR implants were crafted through a combination of hASC, rabbit fibrinogen (obtained from innovative research), and bovine thrombin (sourced from Sigma-Aldrich) in a 4:1:1 ratio. Each 600 μl COVR gel encapsulated a total of 600,000 cells to replicate the cellular density found in VF LP. The solutions were blended within 12-mm Transwell inserts from Corning and left to incubate at 37C, with gelation occurring swiftly within 30 min. Differentiation medium enriched with 10 ng/mL human recombinant EGF from Promega Corp was administered at the basal surface. After a 2-week cultivation period, the developed COVR was harvested for transplantation by incising and peeling the Transwell membrane away from the insert and gel, although preserving the lumenal-basal orientation. The resultant COVR constructs measured 12 mm in diameter with thicknesses ranging between 2–3 mm.

2.2 Implantation in rabbit model

All animal procedures were conducted in accordance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals, the NIH Guide for the Care and Use of Laboratory Animals, and the Animal Welfare Act, and were approved by the local Institutional Animal Care and Use Committee. New Zealand white rabbits from two separate implantation series were examined here. In the first set, bilateral COVR implant was performed in 3 male and 3 female rabbits at the time of VF injury, then allowed to heal for two months. One additional male had preoperative poor feeding but proceeded with a bilateral COVR implant. Hypoxia developed during wound closure, and the animal was humanely euthanized although still in the operative suite. Necropsy and larynx harvest were completed immediately, which revealed severe pulmonary edema manifested as pink frothy fluid filling the lungs. That animal is reported as “Day 0”.

A separate set of slides were obtained from animals reported in a prior study.¹⁴ Briefly, these were randomly allocated to one of three treatment groups: (1) open transcervical unilateral cordectomy followed by immediate COVR implantation (acute COVR group); (2) endoscopic unilateral cordectomy followed by transcervical COVR implantation 2 months later (chronic COVR group); or (3) endoscopic unilateral cordectomy with transcervical sham surgery 2 months later (chronic scar group).

COVR implants were performed as previously described.^{13, 14} Vertical neck incision exposed the larynx and trachea, and a temporary tracheotomy provided airway control during the surgery only. Laryngofissure revealed the endolarynx. Vibratory mucosa from one or both VF was excised through sharp dissection, down to the thyroarytenoid muscle. The COVR gel was applied to the defect and secured in place with anterior and posterior sutures of 5–0 fast gut. The larynx was closed with a Prolene suture, an endotracheal tube was removed, and layered wound closure was performed. Perioperative antibiotics and corticosteroids were administered for 3 days. Following the 2-month maturation period, the animals were euthanized for larynx retrieval.

2.3 Tissue harvest and processing

The excised larynges from the first group (with bilateral COVR) were bisected into hemilarynges. One hemilarynx was fixed in formalin, embedded within paraffin, and then sectioned for analysis. The other hemilarynx was used for molecular studies, which are ongoing. The excised larynges from the second group (with unilateral injury or implant) were initially studied with ex vivo phonation and indentation, as previously described.¹⁴ They were then bisected into hemilarynges, immersed in OCT freezing compound, and flash frozen in liquid nitrogen for frozen section cutting. Some microscopic findings were previously reported, and additional imaging was performed here.

2.4 Histology, immunohistochemistry, and microscopy

The formalin-fixed, paraffin-embedded (FFPE) slides were first rehydrated by immersing the samples for 5 min each under the following solutions: 100% xylene, 100% ethanol, 70% ethanol, and dH2O. One section of each animal was used for Pentachrome histological staining to show elastic fibers in black, collagen in yellow-brown, muscle in fuschia, and glycosaminoglycans in blue. Pentachrome slides were digitally scanned to show the entire section at 4X magnification.

For FFPE slides undergoing immunocytochemistry, antigen retrieval was performed by boiling in citrate acid buffer (ph 6) and then maintained at a sub-boiling temperature for 20 min. After cooling for an additional 20 min the sections were immersed in distilled water for five minutes and then washed with PBS. Slides were blocked with 5% goat serum diluted in PBS for 30 min. Primary antibody staining was done with either anti-HLA (Abcam cat#: ab70328) at a 1:500 dilution or anti-CD31 (Novus cat#: NB0122181) at a 1:400 dilution. Primary antibody was added in 1% BSA (1:500 dilution) to slides and stored overnight for 20 h. A negative control slide receiving only 1% BSA without primary antibody was performed for each animal. Secondary antibody staining was done with 2.5% goat serum blocking solution for 15 min. Fluorescently labeled secondary antibody, goat-anti mouse (Alexa Flour 633, Thermal Fisher), in 1% BSA (1:400 dilution) was later added.

Frozen sections were prepared for immunocytochemistry in a similar fashion except without antigen retrieval. Some of these slides were also labeled with alpha-smooth muscle actin (SMA) antibodies at a dilution of 1:500. The secondary antibody was performed the same as in the paraffin slides. All sections received one drop of antifade mounting medium with DAPI (Vectashield) prior to the addition of a cover slide.

2.5 Western blot analysis

Frozen sections from operated vocal folds were scraped off slides and collected for protein. Cells and tissue were lysed in an SDS sample solution containing 2% SDS and 10% glycerol in 0.5 M Tris-buffer at pH 7.5. The lysates were snap-frozen in liquid nitrogen to prevent degradation. Approximately 300 μL of ice-cold lysis buffer was rapidly added to 5 mg of tissue, which was then homogenized using an electric homogenizer. The lysate was centrifuged at 12,000 rpm for 20 min at 4°C, and the supernatant was collected. The lysate was then heated in a sample buffer at 100°C for 5 min. Samples were loaded onto a 10% Mini-PROTEAN TGX precast gel and resolved by SDS-PAGE at 120 V for 85 min. The resolved proteins were transferred from the gel to a PVDF membrane at 100 mA for 1.5 h. The membrane was blocked for 30 min and then incubated with the primary antibody, anti-SMA (1:2000), or B-actin as a loading control, for 1.5 h at room temperature. After washing, the membrane was incubated with the secondary antibody, Goat anti-Mouse IgG-HRP (1:3000), for 75 min. Proteins were detected using an ECL chemiluminescent system, with optimal exposure time between 30–300 s.

Quantification of Western Blot lanes was done through the utilization of ImageJ Software (https://imagej.net/ij/) and analysis was performed through the Gel Analysis methods outlined in ImageJ¹⁷ with reference to an identical protocol as discussed by Stael.¹⁸ Individual bands of the gels were marked with rectangular selections within the ImageJ Software. Intensity profile plots were then created for each band, and total intensity was calculated using the “Plot profile” function. The intensity of SMA was corrected by our loading protein, B-actin, through utilization of the formula (SMA / (SMA + B-actin)).

3.1 Human leukocyte antigen expression

HLA labeling was identified in scattered cells within the implant region in all animals receiving COVR (Figure 1). The frozen section findings (group 2, unilateral implant) were previously reported.¹⁴ Scar group animals had no labeling for HLA, and BSA substitution for the HLA primary antibody eliminated labeling. Day 0 sections, from the animal suffering perioperative death, showed more HLA-positive cells than in the other animals. Still, cells in the implant region did not uniformly exhibit HLA, consistent with incomplete labeling due to antigen degradation or variable HLA expression by the cells.

3.2 Vascular endothelial structure

CD31 was selected as a marker of vascular endothelium. It is also found in platelets and some types of leukocytes. Figure 2 demonstrates sections from the FFPE larynges, in which all animals underwent bilateral COVR implantation. All animals in this group demonstrated both HLA and CD31+ in the same area, anatomically corresponding to the implant site. The labeling quality was inadequate to identify if both markers were present in the same cells.

Figure 3 depicts the expression of CD31 and HLA, in frozen sections taken 2 months after COVR implantation or sham surgery.¹⁴ CD31 expression was observed in all animals, whereas HLA-positive cells were only present in the Chronic COVR group.

In the Chronic COVR group, both CD31 and HLA-positive cells were present around lumenal structures that appear to be vascular endothelial in nature. In each of those animals, examples were found where HLA and CD31 appeared to be labeling the same cells in adjacent slides.

3.3 Fibroblast activation state

Slides and tissues from a previously reported study¹⁴ were probed here for SMA expression by immunocytochemistry and Western blot. Figure 4 shows IHC labeling for SMA and HLA in frozen sections 2 months after COVR implant or sham surgery. These markers were observed in nearby regions but did not appear to co-localize in individual cells in sequentially-cut sections. As anticipated, HLA was absent in all scar sections, consistent with the absence of implanted human cells. Figure 5 presents the quantification of SMA protein by Western blot in the chronic COVR, acute COVR, and scar groups. Expression levels varied in all groups, and there was no statistically significant difference in SMA expression among the three groups.

4.1 Vascular endothelial structure

The structure of human vocal folds is unique, as only small vessels are present near the vocal fold mucosa. This specific vascular structure, where small vessels are located at the vocal fold edge and separated from the vessels in other areas of the mucosa and muscle, helps to minimize hypoxia and ensures adequate vibration.²⁰ Assessing the vascular structure is therefore critical when evaluating the efficacy of vocal fold regeneration.

CD31, a common endothelial cell marker, plays an important role in vasculogenesis.²¹ CD31+ endothelial progenitor cells have been identified within the stromal vascular fraction (SVF) of adipose tissue, from which hASCs are derived. Those cells were associated with the luminal layer of small blood vessels, and have the capacity to proliferate and form vessel-like structures.^{22, 23} SVF has separately been found to have the ability to differentiate into both endothelial and adipogenic cell types.²⁴

To assess whether the COVR implant assisted with revascularization of the vocal folds, IHC labeling was performed for the endothelial marker CD31. In paraffin embedded sections, HLA and CD31 positive cells were visualized in similar areas. However, limited labeling in these FFPE sections made it difficult to determine if the markers co-localized. Frozen sections were then pursued because immunohistochemical labeling is typically more robust with this method.

These frozen sections more clearly exhibited co-localization of CD31 expression with hASC (as evidenced by HLA expression). This finding suggests that cells within the implanted COVR not only survive but also may develop donor-derived vascular structures. Scar group sections also demonstrated CD31 expression and vascular structures, without HLA, confirming that neovascularization occurs in the scar group during the normal wound healing process.

4.2 Fibroblast activation state

Along the fibroblast–myofibroblast spectrum, myofibroblasts express α-SMA, and also produce excessive collagen, both leading to stiffening of the tissue.²⁵ The vocal fold fibroblast expression of α-SMA has been found to be decreased when the tissues were co-cultured with stem cells from either adipose or bone marrow sources.²⁵ Other previous work with mesenchymal stem cells has shown that their injection into scarred vocal folds can decrease the α-SMA expression and concurrently improve function.²⁶

In the current study, we utilized α-SMA as a marker for myofibroblast activation. We hypothesized that COVR implantation would show a reduction in myofibroblast activation when compared to the scar group immunohistochemistry showed variable regions of SMA expression within the implants and no co-localization with HLA. These findings suggest that the human ASC did not differentiate from myofibroblast phenotypes. However, there was no statistically significant difference in the bulk amount of SMA quantified among groups. We found no evidence, then, that hASC reduces myofibroblast phenotype among host cells in this system.

The variability observed could be due to intersubject differences within each treatment group. Potential sources of this variability could include different degrees of vocal fold injury as well as individual animal-to-animal variations. Further investigation is needed to fully elucidate the role of myofibroblasts and their activation state in the context of vocal fold scarring and the effects of the COVR implant.