The transcriptional changes observed in this study indicate that just 8 hour exposure to a low humidity environment can adversely affect vocal fold biology. To the best of our knowledge, this is the first study to demonstrate the effects of surface dehydration on vocal fold tissue in vivo. Important to our methodology, evaluation of the change in PCV following experimental challenge ruled out systemic dehydration as an unintended confounding factor in our analysis. There is considerable evidence that systemic dehydration negatively impacts phonation (20-23). Surface dehydration represents a loss of water from the mucosal surface of the larynx, and while some level of local tissue water loss may be experienced through compensatory rehydration of the epithelial surface, we would not expect systemic dehydration to result. We hypothesize that the homeostatic responses to surface and systemic dehydration are governed by different cellular mechanisms, thus we used % PCV change to control for unintended systemic consequences of low humidity exposure with the concomitant withholding of food and water.
We developed a method to efficiently challenge rabbits to low humidity. We achieved average low relative humidity of approximately 20%, representing physiologically-realistic and substandard occupational conditions per Occupational Safety and Health Administration (OSHA) recommendations (56). Moderate humidity control exposures were conducted in the same chamber with all compartments open to room air of variable temperature within housing guidelines for rabbits. Low humidity challenge and moderate humidity exposure could not be conducted at the same time because preliminary tests demonstrated that a fully closed air circuit that is needed to lower humidity in the chamber measurably increased the interior temperature of the compartments. By separating them, we successfully maintained appropriate ambient temperatures for the low humidity exposures (57) and maintained a 2-fold increase in moderate humidity exposures.
It is noteworthy that exposure to low relative humidity below the Occupational Health and Safety Administration (OHSA) recommended limit of 20% induced transcriptional changes within functional gene categories including inflammation, ion transport, and keratinocyte development. The most robust functional enrichments identified by STRING were stress, defense, and inflammatory responses. Additionally, outside of the STRING analysis, various genes for immunoglobulin chains were identified, three of which were downregulated and one that was upregulated. Interestingly, this cluster presents two opposing interpretations of innate immune dampening and possible macrophage activation.
While none of these genes or corresponding proteins are described within the larynx, the downregulated cluster can be interpreted as a dampening of acute inflammatory response. ORM1 and SAA1 are both acute phase proteins. ORM1 is an acute phase protein that has been shown to polarize M2 macrophage differentiation (58) and to enhance epithelial integrity in a culture model of the blood-brain barrier (59). While ORM1 exhibits anti-inflammatory activity and its downregulation may allow for the development of a more robust inflammatory process, it may also be interpreted as indicative of surface dehydration not contributing to an activating inflammatory event. SAA1 is also an acute phase protein and is associated with a variety of pathological conditions, but it has also been shown to positively influence keratinocyte activity (60). The S100 proteins are diverse with involvement in several cellular processes, but both S100A9 (61) and S100A12 (62) have been described as damage associated molecular patterns in the literature. Taken together, these results suggest that either surface dehydration is not inducing inflammatory pathways or that there is active repression of pro-inflammatory mediators. The latter is substantiated by the increase of IL1RN which encodes the IL-1 receptor antagonist (IL1RA). IL1RN was upregulated in the posterior cricoarytenoid muscle one week following transection of the recurrent laryngeal nerve in a rat model (63), and IL1RA was significantly increased following 8 hours of industrial exposure to respirable and inhalable dust in humans (64). Together this substantiates a role for the increased IL1RN we observed and of a possible active innate immunity repression in response to the low humidity challenge.
Conversely, the upregulation of MMP12 and MCP1 genes may suggest the activation of inflammatory macrophages. MMP12 was the most significantly upregulated gene in this study by RNA-Seq and RT-qPCR. MMP12 exhibits proteolytic activity on multiple ECM components including elastin, fibronectin, entactin, and type IV collagen (65), all of which are expressed within the vocal folds. Although called “macrophage elastase”, it is also expressed in human vocal fold fibroblasts (66) and bronchial epithelial cells in vitro (67), and in both superficial and deep epidermal layers of the skin in response to ultraviolet radiation (68). MMP12 has a potential role in the development of dysphonia following low humidity exposure since type IV collagen and elastin play an important role in the viscoelasticity and phonatory function of the vocal folds (69, 70). MMP12 may contribute directly to inflammation though epidermal growth factor receptor (EGFR) dependent induction of IL-8 from the respiratory epithelium (71). Interestingly, MMP12 has been shown to positively influence wound healing following epithelial injury to the cornea (72), so it is unclear if the upregulated response to low humidity would be deleterious or influence a reparative response in the vocal folds. MCP1 is an α-defensin expressed in the lungs of fetal and adult rabbits (73); it is secreted from neutrophils and rabbit lung macrophages and exhibits broad antimicrobial activity In our study, the expression of MCP1 was novelly detected in the rabbit larynx, and its upregulation in repsose to low humidity warrants further investiation including targeted anaylsis of differential expression between inflammatory cells and the larygeal tissue.
It is not surprising to find evidence of a pro-inflammatory response with surface dehydration as other environmental stressors such as simulated acidic reflux (74), hypertonic challenge (43), and phonotrauma (52, 75) can perturb the epithelial tight junctions of the vocal folds—indicative of the activation of proinflammatory pathways. As we did not investigate for cell-specific gene expression in this study, we are limited to conclude if the upregulation of these genes reflects activation of macrophages or activity of the epithelium or lamina propria fibroblasts, and further study is warranted. An intriguing hypothesis for a case of macrophage activation would be altered response to local microbiome or pathogens resulting from changes to the laryngeal microenvironment following dehydration.
The perturbation of ion transport or other lubrication mechanisms is anticipated as a response to the altered hydration state of the laryngeal surface (76). Although no gene or protein interaction enrichment cluster was identified within the 103 DEGs analyzed, presumably due to the diversity of substrate and transporter type, a considerable set of ion and solute transporter related genes were identified by RNA-Seq, including ECCP, SLC5A1, SLC13A5, SLC23A1, SLC27A2, and ZACN. All SLC family members were downregulated. This set represents predominantly ion transport, with SLC13A5 and SLC27A2 being involved in glucose transport and fatty acid ligation. In vitro studies of human nasal epithelial cells (41) and human bronchial cell culture (77) demonstrated that apical osmotic pressure can result in altered epithelial electrolyte transport; however, studies with canine tracheal and bronchial cell culture (78) and an in vivo canine model (79) concluded that not all epithelial fluid flux is coupled to electrolyte transport. This evidence suggests that the epithelium may respond to either aberrant electrolyte concentrations or non-ionic osmotic pressure. It is not surprising to find evidence of altered chloride secretion specifically, as balanced sodium and chloride ion secretion is attributed to volume regulation of the airway surface fluid, but the contribution of transport of other ionic and non-ionic species is not well described for airway surface fluid regulation. Our results suggest the pertinence of future targeted study of noncanonical secretion products in the respiratory tract.
Although the ECCP is annotated as an epithelial chloride channel protein, the translation product for ECCP is neither well characterized nor has a direct ortholog in humans. It may belong to the calcium-activated chloride channel proteins (CLCA) family as identified by conserved functional domains, although it exhibits limited homology to the rabbit CLCA proteins. The genes for CLCA1, CLCA2, and CLCA4 lie within the same genetic neighborhood as ECCP but were identified by RNA-Seq with FDR > 0.99, indicating they are not differentially expressed in our model of surface dehydration (Additional file 1: Table S1). This suggests a distinct role for ECCP and its downregulation that warrant further investigation as an ion channel protein newly described in the context laryngeal surface dehydration. In contrast to ECCP, ZACN was upregulated in low humidity compared to moderate humidity but failed to reach statistical significance by RT-qPCR. ZACN is a cation channel expressed in the human trachea and other tissues and demonstrates permeability to potassium ions but not to chloride ions (80); there is no discussion of its expression in the vocal folds in the literature, and it is unclear if it may also be sodium ion permeable. Taken together with the SLC family members identified, these results support a potential role for solute flux as a homeostatic response to surface dehydration. Interestingly, however, the downregulation of chloride transportation would be a counterintuitive response to surface dehydration at the apical membrane as chloride is generally directed out of the cell and aberrant chloride transport can be detrimental in the airways as seen in cystic fibrosis. There is a distinction between the respiratory epithelium of the airways and the nonkeratinized stratified squamous epithelium of the vocal folds, so care must be taken with direct translations of actions between the two.
The mucins are equally important to maintain satisfactory hydration of the laryngeal surface as ion and fluid flux. MUC12, MUC21, and TFF1 were identified as downregulated by RNA-Seq. Both mucins are members of the cell-surface associated mucin family, and as such, should originate directly from the epithelial cells. The first exon of MUC12 exhibited increased expression in laryngeal epithelium from laryngeal reflux patients compared to reflux negative patients (81). Exogenous surface expression of MUC21 in in vitro cell culture reduced intercellular adhesion and adhesion to extracellular components (82). It is interesting then to observe all three to be downregulated. However, in addition to roles as epithelial protectants, mucins and related proteins also serve roles in cell signaling with physiological consequences. This is recently shown for MUC21 overexpression as influencing the development of lung adenocarcinoma (83) and TFF1 influencing epithelial-mesenchymal transition. Together, this may be a contributing factor to the STRING cluster of keratinocyte differentiation factors discussed below, but further study is warranted to determine which cell types are expressing these genes and which cell signaling may be impacted.
Although there was no gross inflammation observed, some level of epithelial cellular response to surface dehydration is expected. The vocal folds are covered by a non-keratinized stratified squamous epithelium for which some aspects of development are well understood, such as embryological developmental factors and differentially expressed structural components (84, 85), but a comprehensive molecular description is not available as for other epithelia like the epidermis. It is interesting that several keratinocyte developmental and epithelial structural factors were identified with RNA-Seq, enriched in the low humidity group by GSEA, and as a protein interaction cluster in the STRING analysis: CDSN, CNFN, CRNN, KRT80, KRTDAP, and TGM3. Also identified by RNA-Seq were SPBN, another keratinocyte factor, and CDHR4, a cell interaction mediator. All of these were downregulated. SPBN, CDHR4, and CDSN were selected for RT-qPCR validation. All three gene products may be involved in maintaining the integrity of the stratified squamous epithelium, though none have been described specifically within the vocal folds until this study. SPBN is expressed in the suprabasal layers of tongue, stomach, and epidermis (86). It is required for keratinocyte differentiation in an in vitro skin model (87) and skin development in murine embryos (88). The specific activity of CDHR4 is not described in the literature, but family member CDHR2 is expressed in gastrointestinal epithelial cells and is associated with microvillus development (89), while family member CHDR3 is expressed in ciliated respiratory epithelial cells and is associated with ciliary development and intercellular interactions (90). CDSN is expressed in the stratum granulosum of human skin and appears to participate in cellular cohesion at this level, with its loss associated with desquamation (91, 92). That the entire cluster was downregulated substantiates surface dehydration as capable to influence vocal fold epithelial maintenance. Further study is required to elucidate the specific roles of these proteins within the vocal folds, as this epithelium is distinct from the epidermis.
A limitation of designing an environmental chamber as described here was that it precluded the provision of relative humidity lower than 15%. While environmental rooms and chambers are commercially available, they are cost prohibitive and their small size precludes the use of certain animal models, such as rabbits. Another limitation of the study is that we only observed a single time point after low humidity exposure. It has been shown that local response to vocal fold injury is transient and time-dependent (48, 54, 93). Further studies specifically observing for inflammatory response at multiple times points within a single challenge or within repeated or chronic challenges would be helpful in further characterization of vocal fold biology. Finally, the dissected vocal fold tissue included striated muscle and small amounts of respiratory epithelium immediately above and below the region of the vocal folds. Therefore, genes associated with muscle or respiratory epithelium were not selected for the discussion.