Exploring the anti-inflammatory potential of topical hyaluronic acid for vocal fold injury in a rat model

Forty-two rats were initially enrolled in the study and divided into four groups: an uninjured control group, an injured control group and two injured treatment groups in which the rats were treated with either topical HA alone or in combination with diclofenac. One rat was later excluded from the study as described below.

After induction of anesthesia, all rats had their necks shaved and dissected to expose the trachea. All rats, regardless of which group they were assigned to, were kept alive under a heat lamp for 1 h after the intervention. In the uninjured control group, surgery was stopped at this point. In the remaining animals, which were assigned to the injured groups, the trachea was opened and a cannula was inserted into the trachea. Under microscopic guidance, the larynx was split medially and after identification of the VF, these were injured bilaterally with a needle. The rats assigned to the injured control group were also placed under the heat lamp for 1 h. The remaining rats received topical therapy. Each of these steps is explained in more detail in the following sections and visualized in Fig. 1.

2% (w/w) HA acid sodium salt (Mw 30.000–50.000 Da, Prod.No. 53747; Sigma Aldrich, Munich, Germany) was stirred in MilliQ water at 55 °C for 2–3 h in a closed vial until a clear gel was formed. Subsequently, 1.5% (w/w) diclofenac sodium was slowly added and the mixture was stirred for 15 min.

The viscoelastic behavior in terms of storage modulus G′, loss modulus G″, and complex viscosity η* was investigated using a Physica MCR 301 rheometer (Anton Paar, Graz, Austria) and a CP-50-1 measurement system with cone-plate geometry at 37 °C. Shear rates between 0.1 and 10 rad/s were applied. The loss factor tanδ was calculated as the ratio of G″/G′. The thixotropy behavior of the gels was investigated using the same device in rotational measurement mode. The reduction and recovery of the structural strength of the gels was investigated by applying a constant shear stress of 250 rad/s for 60 s, followed by a rest period of 300 s.

The slip angles were evaluated using a glass microscope slide, and the drop weight of the gels was evaluated using the standard pharmacopeia method for liquid and semisolid formulations (Ph. Eur. 2.1.1).

Forty-one male Sprague–Dawley rats (older than 12 weeks) with an average weight of 569.31 g ± 32.05 g were used. Rats were housed in pairs in the pathogen-free facility of the Medical University of Graz for 4–6 weeks for acclimatization with ad libitum water and standard food. All procedures were approved by the Austrian Federal Ministry of Science, Research and Economy (approval number: BMBWF-66.010/0083-V/3b/2018) and complied with the institution´s animal care guidelines.

For the experiments, rats were randomly assigned to four groups: an uninjured control group, an injured control group, and two treatment groups in which HA alone or HA combined with diclofenac was applied to the injured area (Fig. 1). One rat from the HA group was excluded due to lack of genetic material in the harvested tissue.

Anesthesia was induced by exposing the rats to isoflurane (5% delivered at 2 L/min) in an induction chamber for 10 min, followed by maintenance with intramuscular application of medetomidine (75–101.25 µg/kg), midazolam (1–1.35 mg/kg), and fentanyl (2.5–3.38 µg/kg). The neck was shaved anteriorly and disinfected with ethanol. Rats were then placed supine on an operating platform and a vertical midline incision was made with a No. 21 surgical blade. The pre-laryngeal tissue was bluntly retracted to avoid major injury and bleeding. In rats assigned to the uninjured control group, only the thyroid cartilage was visualized, but the larynx and trachea were not dissected. Rats assigned to VF injury were tracheostomized between the first and second tracheal rings. A plastic tube from a 17-gauge venous cannula was then inserted into the trachea for intubation (Fig. 2a). The larynx was dissected using a vertical median incision. VF injury was induced under the microscope (Vision SX45) by multiple vertical scratches with a 23-gauge needle. Minor bleeding due to mucosal injury was controlled by local compression with cotton swabs. Depending on the assigned group, either diclofenac or HA gel was applied to the injured VF. Rats in the injured control group were not exposed to any substance. The surgeon was blinded to the gel treatment applied.

After injury, animals were placed under a warming lamp to prevent hypothermia until tissue harvest.

One hour after surgery, rats were euthanized by intracardiac injection of 1 mL pentobarbital sodium (400 mg/mL). The larynx was excised, mounted on a mounting board, and placed under a microscope (Fig. 2b). Mucus, blood, and gel were removed with cotton swabs, and the VF mucosa was harvested, snap frozen in liquid nitrogen in MagNa Lyser Green Beads tubes (Roche Diagnostics, Mannheim, Germany) and stored at -80 °C until further processing.

For tissue homogenization, 700 µl of QIAzol reagent (Qiagen, Hilden, Germany) was added to the frozen samples. Four homogenization steps (20 s at 6500 rpm) were performed on the MagNa Lyser instrument with intermittent cooling of the samples on ice for 1 min. Total RNA was then isolated using the miRNeasy Mini kit (Qiagen) according to the manufacturer´s instructions and RNA concentration was determined using the NanoDrop 2000c UV–Vis spectrophotometer (Thermo Scientific, Waltham, MA, USA). Reverse transcription (RT) was performed using the QuantiTect Reverse Transcription kit (Qiagen) according to the manufacturer´s protocol. RT-qPCR was performed as previously described [16]. All primers used are listed in Table 1. Samples were measured in technical triplicates and data were corrected using a Universal Rat Reference RNA (cat.no.QS0641; Invitrogen) for interplate calibration. Relative quantification of mRNAs of interest was calculated based on the 2-ΔΔCq method [17] with minor modifications: the geometric mean of the internal reference genes Werner helicase interacting protein 1 (Wrnip1) and ubiquitously expressed, prefoldin-like chaperone (Uxt) was used for internal normalization, and the mean of all analyzed samples was used as the reference to calculate ΔΔCq values.

The pure and drug loaded HA gels show a shear thinning and viscoelastic behavior. The viscosity of both gels is in a similar range. However, the viscoelastic properties differ slightly. For HA gels, the elastic modulus G′ dominates the viscous modulus G′′ over the applied shear range, resulting in tanδ values between 0.78 and 0.63. Through the addition of diclofenac–sodium, the viscous modulus is high at the starting point at 0.1 rad/s. After a crossover point at low shear rates, i.e., 0.6–0.8 rad/s, the elastic modulus dominates the viscous modulus indicating more stable network interactions. The loss factor tanδ increased from a starting value of 1.24 at 0.1 rad/s to 0.86 after 0.8 rad/s. Both gels exhibit a thixotropic behavior with a complete recovery of the structure as soon as the stress is stopped. This is particularly important for gel application with syringes, as the formulation liquefies due to shear stress, but rebuilds its structure within a very short time and thus remains at the site of action. The slip angle of the gels was 48° ± 2° for HA gels and 52° ± 1° for the drug-loaded gels, and the average drop weight (n = 20) was 8.044 ± 0.683 mg, which equals an average calculated drug loading of 0.12 mg/drop (Fig. 3).

The gene expression of cyclooxygenase (Ptgs)1 and 2 were upregulated by injury, though only Ptgs2 reached statistical significance. Both, HA gel and diclofenac gel treatment tended to reduce this upregulation similarly (Fig. 4a, and b, respectively). The gene expression profiles of the interleukins interleukin 1 beta (Il1b) and interleukin 10 (Il10) were significantly increased after injury, with no statistical effect for either gel treatment (Fig. 4c, and d, respectively). Transforming growth factor beta 1 (Tgfb1) gene expression was increased after injury, which was significantly decreased after HA gel treatment and, although not significant, reduced after diclofenac gel treatment (Fig. 4e).

The gene expression profiles of tumor necrosis factor alpha (Tnfa), nuclear factor kappa B subunit 1 (Nfkb1), and interferon gamma (Ifng) remained unchanged after injury and treatment (Fig. 4f–h, respectively).

The gene expression profiles of collagen type I alpha 1 chain (Col1a1), collagen type I alpha 2 (Col1a2), collagen type III alpha 1 chain (Col3a1) remained unchanged after injury and treatment (Fig. 5a–c, respectively).

Hyaluronan synthase (Has) 1 and 2 gene expression levels were significantly upregulated after injury, with no statistical effect for either gel treatment (Fig. 5d, e, respectively). Has3 gene expression was upregulated after injury without reaching significance, whereas HA gel treatment significantly reduced this upregulation (Fig. 5f).

Elastin (Eln) gene expression did not show any change by injury, but was reduced by both gel treatments, whereas the reduction by HA gel treatment reached significance (Fig. 5g).

Wound healing can be divided into three partially overlapping phases: inflammatory, proliferative, and tissue remodeling. During the inflammatory phase, immune cells migrate to the site of injury to remove cellular debris and pathogens [3]. In the proliferative phase, fibroblasts produce new ECM to provide structural support for the developing tissue. Finally, during the remodeling phase, excess ECM components are degraded, refining the tissue and forming a mature scar [20, 21].

The foundation of wound healing is laid during the inflammatory phase, as dysregulation can lead to chronic inflammation and impaired wound healing [22]. It is, therefore, particularly attractive to target this phase to improve the functional properties of the remodeled tissue. Indeed, many approaches have been investigated to improve the oscillatory properties of the VF after injury. However, none of these approaches has achieved the desired effect [4, 23]. As a result, there is currently no agent applied in clinical routine in laryngology.

The inflammatory phase involves hemostasis and an inflammatory response. Studies have shown that the destruction of keratinocytes leads to the release of IL-1, the body's first signal after tissue injury. This leads to the activation of platelets and the release of other factors, such as growth factors, which are designed to attract more inflammatory cells and thus disarm the infectious agents [19, 24]

In the present work, a significant upregulation of Has1, Ptgs2, Il1b and Il10 was seen in all three injury groups compared to the uninjured control group. Similarly, Welham et al. reported a significant upregulation of Has1, Ptgs2, Il1b and Tnfa after 1 h of injury [2]. The fact that Il1b, Il10 and Ptgs2 were significantly upregulated, suggests that the VF injury model in the present study was functional. Although the other inflammatory markers measured (Ptgs1, Tnfa, Tgfb1, Nfkb, Ifng) are also described as key regulators in the inflammatory phase and would, therefore, be expected to be upregulated, they were not significantly in these experiments [25]. We attribute this to the short observation period.

Tissue repair and remodeling occurs at later stages of wound healing, which explains why there was no significance in the expression of either COL or ELN. The increase of Has expression was, therefore, interesting, as this enzyme is responsible for the production of HA. However, HA is not only a scaffolding protein but plays an important role in immunomodulation and angiogenesis and is, therefore, also relevant at an early stage [26].

From a risk–benefit perspective, topical therapy is the most promising, as no systemic side effects are to be expected. Akdogan et al. investigated the effect of topical retinoic acid on VF wound healing in rabbits. They showed that there was a significant reduction in both collagen and fibroblast deposition in the retinoic acid-treated group [18]. However, local irritation has been reported in dermatologic studies, where it is used to treat acne vulgaris [27]. In VF, this side effect can be serious and may lead to significant deterioration of the voice. Therefore, this agent does not appear to be suitable for the treatment of VF scars in humans.

Another study investigated the effect of HA collagen nanofibers on early wound healing. Twelve New Zealand white rabbits underwent unilateral VF injury. The six rabbits in the control group received superficial saline treatment using cotton swabs for three consecutive days, while the other six rabbits were treated with an HA–collagen mixture for three consecutive days. Half of the rabbits in each group (treatment and control) were then sacrificed on day 7, while the other half were euthanized on day 21. Based on a significant reduction in collagen fiber diameter on day 21 in the treatment group, they concluded that topical application of HA–collagen could lead to reduced scar tissue formation [28]. Although an interesting approach, this study may lack statistical power due to the small number of animals in each group. In addition, the sequential application on three consecutive days as described in this publication may not be feasible in clinical routine as the majority of awake patients would probably not tolerate topical treatment and repeated anesthesia for this purpose would simply not be justifiable. Nevertheless, HA appears to be a promising agent for improving VF wound healing and was used in the present study for the reasons described in more detail in the following paragraphs.

HA is a ubiquitous polysaccharide that plays a key role in wound healing and is widely used in other areas of medicine due to its high biocompatibility and low cost [29]. HA is known to inhibit the profibrotic effect of TGFβ1 via the Smad signaling pathway [30]. This may explain the significant downregulation of Tgfb1 in the present study by the addition of HA to the wound. TGF-β1 plays a critical role in the inflammatory phase of wound healing and excess can lead to hypertrophic scars [31]. Targeting TGF-β1 is, therefore, a popular way to improve wound healing. Studies have shown that inhibition of TGF-β improves wound closure and reduces fibrosis [32, 33]. Therefore, the gene downregulation of Tgfb1 demonstrated in this study may provide a favorable environment for wound healing.

Our results showed that Eln, which is also partially regulated by Smad [34], was significantly downregulated by the addition of HA. ELN is urgently needed for tissue repair, but it appears that the composition of the cross-links is more important than the concentrations themselves. [35] Based on the available results, it is not possible to make a conclusive assessment in this regard and further investigation is required.

In the presence of HA, Has was downregulated, but only Has3 was significant. This may suggest a negative feedback mechanism to maintain HA homeostasis. According to Tammi et al. the relationship between HA concentration and Has expression is poorly understood. This may be due to technical difficulties in reliably detecting HAS at the protein level [36]. Therefore, this feedback mechanism can currently not be demonstrated in a healthy population.

In the group, where a mixture of HA and diclofenac was applied to the wound, no significant changes were observed compared to the injury group. Diclofenac belongs to the group of NSAIDs and exerts its anti-inflammatory effect through COX inhibition. However, the effect of topical diclofenac on wound healing is controversial. While the anti-inflammatory effect may be beneficial, histomorphometric studies have shown a reduction in fibroblasts, which may be associated with impaired wound healing [37, 38]. The lack of significance with the addition of diclofenac in the present study could have several reasons. These include the short observation period, the low dose, or an actual lack of effect.

There are some limitations to this study. One is that the observation period was very short. To study the effect of the drugs over a longer period of time, the animals would have had to be awakened from anesthesia. This was impossible, because the gel would have blocked their airways and they would have suffocated. Therefore, it is not possible to draw definitive conclusions about the effects of the local therapy used here on the fibrotic processes and thus on the vibrational properties of the VF. One way to overcome these problems in the future would be to use larger animals. Rats were used in this work, because they have a very similar structure to the human VF [39]. At the cost of greater histological differences, other animal models (e.g., sheep) could more easily be used to study long-term effects. Another limitation is that only changes at the mRNA level were examined. This was also due to the small size of the rat larynx, as only very little tissue could be obtained. This could have been achieved by increasing the number of animals or switching to another animal model. The latter would be preferable, because it would require fewer animals, taking into account the principle of 3Rs (reduce, replace, refine).