The Mucosal Immune Response to Laryngopharyngeal Reflux

Abstract

Rationale: Laryngopharyngeal reflux (LPR) affects up to 20% of Western populations. Although individual morbidity is usually moderate, treatment costs are high and there are associations with other diseases, including laryngeal cancer. To date, there have been no studies of the mucosal immune response to this common inflammatory disease.

Objectives: To determine the mucosal immune response to LPR.

Methods: We performed a prospective immunologic study of laryngeal biopsies from patients with LPR and control subjects (n = 12 and 11, respectively), and of primary laryngeal epithelial cells in vitro.

Measurements and Main Results: Quantitative multiple-color immunofluorescence, using antibodies for lymphocytes (CD4, CD8, CD3, CD79, CD161), granulocytes (CD68, EMBP), monocytic cells (CD68, major histocompatibility complex [MHC] class II), and classical and nonclassical MHC (I, II, β₂-microglobulin, CD1d). Univariate and multivariate analysis and colocalization measurements were applied. There was an increase in percentage area of mucosal CD8⁺ cells in the epithelium (P < 0.005), whereas other leukocyte and granulocyte antigens were unchanged. Although epithelial MHC class I and II expression was unchanged by reflux, expression of the nonclassical MHC molecule CD1d increased (P < 0.05, luminal layers). In vitro, laryngeal epithelial cells constitutively expressed CD1d. CD1d and MHC I expression were inversely related in all subjects, in a pattern which appears to be unique to the upper airway. Colocalization of natural killer T (NKT) cells with CD1d increased in patients (P < 0.01).

Conclusions: These data indicate a role for the CD1d–NKT cell axis in response to LPR in humans. This represents a useful target for novel diagnostics and treatments in this common condition.

AT A GLANCE COMMENTARY

Scientific Knowledge on the Subject

Laryngopharyngeal reflux affects up to 20% of the United States population. Although morbidity is low, treatment costs are high, and it is associated with other diseases, including mucosal cancer.

What This Study Adds to the Field

The nonclassical major histocompatibility complex molecule CD1d appears to be involved in the response to laryngopharyngeal reflux. This finding suggests novel approaches to the diagnosis and treatment of this common condition.

The larynx represents the crossroads between the gastrointestinal and respiratory systems, and is the junction of the IgA-dominated upper and IgG-dominated lower airways. It is the narrowest point of the airway and a point of great airway turbulence, leading to the deposition of high densities of inhaled, and sometimes ingested, challenges. It also has a bacterial flora distinct from that of other parts of the respiratory tract (1), and there is evidence supporting a distinct immunologic role for the larynx.

Work by our group and others has demonstrated a high density of immunologically active cells in the laryngeal mucosa of animals (2–6) and humans (7–10). There are occasional highly organized lymphoid collections, similar to Peyer's patches (larynx-associated lymphoid tissue) (2, 7), although, as with the bowel (11), more than 90% of immune cells are diffusely distributed throughout the epithelium and lamina propria (8, 9). Environmental challenges alter laryngeal immunology: exposure of rat larynxes to viruses increases numbers of mucosal dendritic cells (4), whereas we have shown previously that the laryngeal mucosa of smokers contains more CD4⁺ T cells (9). Also, patients who have had a laryngectomy suffer increased chest and other infections, which could be partly due to the loss of laryngeal immune responses (12).

Laryngopharyngeal reflux (LPR; synonyms are posterior laryngitis, extraesophageal reflux) is common, affecting an estimated 20% of the adult population in the United States (13). On the basis of U.K. audit figures (unpublished data), 4% of the £500 million ($1 billion) spent on proton pump inhibitors annually by the National Health Service (i.e., £20 million, $40 million), is used to treat LPR. The main clinical features are mucosal inflammation and edema, but symptoms as diverse as globus pharyngeus and chronic cough have been linked to this disease. Furthermore, a recent case-control study found an odds ratio of 2.1 for the development of laryngeal cancer in persons with reflux (14). We found significantly less carbonic anhydrase type III in the laryngeal mucosa of patients with reflux, contributing to the lower pH environment necessary for pepsin-mediated damage (15). In a pig laryngeal ex vivo model, such damage could be seen as epithelial “holes” (16). However, to date, there have been no reports of the changes in immunologically active cells in the laryngeal mucosa of this common chronic inflammatory condition.

The first aim of this study was to examine changes in the expression of immunologically relevant cell subsets and antigen-presenting molecules within the laryngeal mucosa as a result of LPR. The second aim was to use these changes to develop hypotheses about the immunologic function of the laryngeal mucosa. Some of the results of these studies have been previously reported in the form of an abstract (17).

METHODS

Patients and Control Subjects

Patients and control subjects were obtained from nonconsecutive consenting persons presenting to the Center for Voice Disorders, Wake Forest University Baptist Medical Center, North Carolina, between August 2004 and April 2005. Demographics of patients and control subjects are shown in Table 1.

TABLE 1.

DEMOGRAPHIC CHARACTERISTICS OF PATIENTS AND CONTROL SUBJECTS

	Control	LPR
Total recruited	11	12
Female	5	10
Non-white	1	2
Median BMI (range)	28 (22–40)	28 (23–35)
Median age, yr (range)	40 (30–57)	49 (45–60)
Median RSI (range)	3 (0–7)	24 (21–30)

Definition of abbreviations: BMI = body mass index; LPR = laryngopharyngeal reflux; RSI = reflux symptom index.

Both groups had relatively high median BMIs. See online supplement Methods for further details of diagnostic criteria.

Biopsies of posterior (nonvibrating edge) vocal cord epithelium were taken using transnasal esophagoscopy under topical anesthesia (17) and immediately snap frozen in isopentane which had been precooled over liquid nitrogen. Specimens were mounted on cork discs in OCT (Sakura, Torrance, CA) and stored at −80°C.

Quantitative Multiple-Color Immunofluorescence Histology

Five-micrometer frozen tissue sections were subjected to two- or three-step multiple-color immunofluorescence as previously described (8–10, 18–20). See Table E1 of the online supplement for details of antibodies used. Sections were mounted, multiple fields at ×20 magnification were digitized, and grayscale images captured on a Leica DMRA microscope using a Hamamatsu Orca-ER camera and Q-Fluoro software (Leica, Milton Keynes, UK). Percentage areas of positive pixel staining were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov), and mean fluorescence intensity of staining across the basal to apical layers of the epithelium was measured using Image ProPlus software (Media Cybernetics, Silver Spring, MD). See online supplement for further details of quantitative image analysis.

In Vitro Staining of Laryngeal Epithelial Cells

Laryngeal epithelial cells were isolated from biopsies of normal patients undergoing routine surgery for nonlaryngeal disease, and cultured in serum-free medium using methods previously described (19). Cytospins of epithelial cells at first passage were subjected to dual-color immunofluorescence as described above and in the online supplement, using a mouse anti-human monoclonal antibody to CD1d (see Table E1).

Colocalization Method

The levels of colocalization of CD1d, CD3, and CD161 were calculated as described previously (20) using ImageJ. Briefly, preferential colocalization of molecules (and therefore fluorochromes) was determined by comparisons of the observed distributions of colors with those predicted from a null hypothesis that each element of positive staining was randomly and independently distributed (the expected levels of colocalization). The data were log₁₀ transformed to obtain a normal distribution, and the mean observed and expected values were calculated for the reflux and control groups.

Statistical Analysis

Average pixel area for each fluorochrome was calculated for each biopsy using macros developed by our group (20). Analysis of log₁₀(n + x) transformed data used analysis of variance (ANOVA) and t tests (SPSS version 12.0 for Windows; SPSS, Inc., Chicago, IL). For line analysis, paired t tests were used to compare intensities of antigen expression between epithelium and lamina propria. Where standard deviations were unequal, central tendency was expressed using a geometric mean and a paired t test was performed on log₁₀-transformed data. The observed levels of colocalization of fluorochromes were compared with the expected levels using a two-tailed, paired Student's t test.

RESULTS

To investigate the immunologic consequences of LPR, we first studied the composition and location of the infiltrating cells in the laryngeal mucosa of the patient and control samples. No differences in numbers of B cells between patients with LPR and control subjects were found, confirming our earlier observation that B cells are almost entirely confined to the lamina propria (9). Similarly, there was no change in the expression of neutrophil, eosinophil, or monocytic lineage markers with reflux.

Next, the distribution of CD8⁺ lymphocytes was compared. There was a highly significant increase in overall CD8⁺ staining in the laryngeal epithelium of patients with reflux (P < 0.005; Figure 1A). When the epithelium was then stratified into luminal and basal layers, the ratios of areas of expression of CD8 in the luminal and basal epithelial layers was significantly higher in patients compared with control subjects (P < 0.0001; Figure 1B), suggesting that the response to LPR may involve accumulation of CD8⁺ T cells in the luminal epithelial layer.

Figure 1.

(A) Log₁₀(n + 2) average percentage pixel area staining of CD8 in patients with laryngopharyngeal reflux (LPR) and control subjects. Bars indicate geometric means. Epithelial CD8⁺ cells increased in reflux (P < 0.005). (B) The ratio of expression of average luminal to basal epithelial CD8 staining in patients with LPR and control subjects. Bars indicate geometric means. Patients with LPR had significantly greater expression of CD8 in the luminal epithelial layer (P < 0.0001). (C) Two-color immunofluorescence image of laryngeal mucosa of a control showing CD8+ T-cells (red) concentrated in basal layer of the epithelium (EP), the area with the highest epithelial cell expression of MHC Class I (green). LP = lamina propria; white line = basement membrane.

Next, we investigated whether changes in the expression of the antigen-presenting major histocompatibility complex (MHC) class I and class II molecules were associated with LPR. In these experiments, as described previously (10), the thickness of the epithelium was divided into 10 equal distances (deciles), and changes across these layers from basal to luminal levels were used to assess the intensity of expression of a given molecule. In concordance with our previous results, MHC class II molecules were expressed consistently throughout all layers of the epithelium in the healthy control subjects. This pattern was preserved in patients with LPR and there was no statistically significant difference in MHC class II expression between the control and the patient group (Figure 2A).

Figure 2.

Average log₁₀ fluorescence intensity of expression of (A) major histocompatibility complex (MHC) class II, (B) MHC class I, and (C) β₂-microglobulin (β₂m) across the laryngeal epithelium of patients with laryngopharyngeal reflux (LPR) and control subjects. Intensity of expression has been measured at each of 10 levels for each group, with basal layers to the left and luminal layers to the right. Patients with LPR had increased β₂m expression near the basement membrane (P < 0.05).

In keeping with previous reports, the expression of MHC class I molecules was polarized, with more antigen present in the deeper layers and gradually decreasing in abundance toward the luminal side in control samples. The staining pattern in patients with LPR was identical, and the level of expression did not show any differences between control and LPR larynxes (Figure 2B). The expression of β₂-microglobulin (β₂m), the molecule associated with classical MHC class I molecules, showed the same polarized behavior, with levels decreasing toward the lumen. However, as shown in Figure 2C, comparison of individual points between the two groups showed a trend to increased β₂m levels in the LPR samples, and this difference reached statistical significance in the deepest layers of the epithelium (P < 0.05; Figure 2C). This suggests that not all β₂m protein may be complexed with classical class I molecules in patients with reflux, raising the possibility that one of the nonclassical class I molecules could be associated with β₂m in laryngeal epithelial cells. We explored whether the difference between patterns of MHC class I and β₂m expression might be due to the expression of CD1d in the larynx.

CD1d was present on epithelial cells with a membranous expression pattern both in the healthy control individuals and the patients. For both groups, there was significantly more CD1d expression in the superficial (luminal) layers (P < 0.001 multiple range test; Figure 3A) decreasing toward more basal deciles. These results indicate a topographic switching of expression of MHC class I^hiCD1d^lo epithelial cells in the basal layers, gradually transitioning to the MHC class I^loCD1d^hi phenotype seen in the luminal layers (Figures 3 A and 3B). In LPR samples, there was an overall trend (ANOVA F = 0.4, P = 0.2) toward an increased CD1d expression, which reached significance for the most superficial levels (P = 0.023; Figure 3A)

Figure 3.

CD1d expression by human laryngeal epithelium. (A) Average log₁₀ fluorescence intensity of expression of CD1d. Expression was measured in the same way as for Figure 2, with basal layers to the left and luminal to the right. Patients with laryngopharyngeal reflux (LPR) exhibited greater expression of CD1d toward the luminal edge of the epithelium (P = 0.023). (B) Three-color immunofluorescence image of laryngeal mucosa of a patient with LPR showing topographic switching between MHC class I/β₂m (basal layers, green) and CD1d (luminal layers, blue). CD161⁺ cells (red) may be seen colocalizing with CD1d (white arrow). White line denotes basement membrane (BM), EPI = epithelium; LP = lamina propria.

To explore whether laryngeal epithelial cells express CD1d constitutively, or if the presence of this antigen-presenting molecule is regulated by environmental factors in situ, we studied the presence of CD1d on cultured epithelial cells. Cytospins of epithelial cells, derived from healthy donors at first passage, were stained for the presence of CD1d using the same antibody as above. As shown in Figure 4, all cells expressed CD1d in a predominantly membranous pattern, indicating that bacterial products are not necessary to induce the transcription/translation of this protein (Figure 4).

Figure 4.

Cytospins of cultured normal human primary laryngeal epithelial cells stained for CD1d. Green = CD1d, blue = nuclei. There is constitutive expression of CD1d (green) in a membranous pattern.

CD1d internalizes, processes, and presents glycolipid antigens to adjacent T cells (21), but also presents antigens to a small subset of T lymphocytes called natural killer T (NKT) cells (22–24). Therefore, we proceeded to measure the expression of NKT cell markers in our subjects. Our choice of antigens for identification of NKT cells was limited by the need to have different immunoglobulin isotype–specific primary and secondary antibodies for each molecule in any given combination, and by the importance of colocalizing these cells with CD1d. Thus, we defined NKT cells as CD3⁺CD161⁺ lymphocytes. Because CD161 is predominantly present on natural killer (NK) cells, we compared the prevalence and location of NK cells (CD161⁺CD3⁻ lymphocytes) with NKT cells, which also express the T-cell marker CD3. In these experiments, NK cells exhibited a significantly reduced tendency to associate with CD1d in the laryngeal mucosa (P < 0.01). The majority of CD161⁺ cells colocalizing with CD1d were CD3⁺ T cells (P < 0.01; Figure 5), indicating the possibility that NKT cells were likely to interact with CD1d in the mucosa. In turn, this suggests antigen recognition by NKT cells at this location. Finally, whereas the number of CD3⁺ CD161⁺ NKT cells did not change in patients with reflux, there was a marked increase in the number of CD161⁻CD3⁺ T cells accumulating in the luminal, CD1d^hi layers of the epithelium, which corroborates our finding that patients with reflux had significantly increased CD8⁺ T-lymphocyte staining in the luminal layer of the laryngeal epithelium (Figure 1B).

Figure 5.

Colocalization studies for CD161, CD3, and CD1d in laryngeal epithelium in control subjects (open and hatched bars for observed and expected results, respectively) and patients with laryngopharyngeal reflux (LPR) (solid and cross-hatched bars for observed and expected results, respectively). The graph shows the observed and expected areas of colocalization. Error bars represent standard errors. *P < 0.01.

DISCUSSION

LPR is the most frequently encountered chronic inflammatory condition of the larynx. Despite its high prevalence, and the high associated health care costs, the immunology of this disease state remains to be fully elucidated. The aim of this study was to examine changes in the expression of the main antigen-presenting molecules present on the epithelial cells and the composition and frequency of immunologically relevant cells within the mucosa in LPR. Therefore, we performed a prospective, blinded, controlled, quantitative study on the expression of selected immunologically relevant cell surface molecules in vivo, using biopsies of laryngeal mucosa from patients with this common chronic inflammatory disease and matched control subjects.

The main differences in the immunologic infiltrate between healthy control subjects and patients with LPR consisted of an increase in the numbers of CD8⁺ cells. CD8⁺ TCRαβ⁺ T cells classically respond to antigen associated with MHC class I, which was consistent with our observation that these cells localized to the deeper, MHC class I^hi layers of the epithelium in control subjects. However, in patients with reflux, the expression of CD1d was increased toward the outer layer of the epithelium to a greater degree than in control subjects. Next, we investigated if changes in antigen-presenting molecules could give clues toward the mechanisms of the observed increase in CD8 staining toward the luminal epithelial layer. Consistent with our previous studies (10), MHC class I levels diminished significantly across the epithelium from basal to luminal surfaces. A similar decrease was seen in the expression of β₂m, which forms complexes with the MHC class I heavy chain and a short peptide antigen. However, comparison between healthy control subjects and patients with LPR demonstrated a relative elevation in β₂m production in the patient group. This suggested the existence of β₂m protein in a form not complexed with classical class I molecules and led us to test the hypothesis that nonclassical MHC molecule expression by laryngeal epithelial cells is involved in the mucosal response to refluxate.

A number of nonclassical MHC molecules, including HLA-G, HLA-E, and various CD1 isoforms, associate with β₂m (25). However, HLA-G expression has only been unequivocally demonstrated in extravillous cytotrophoblasts, whereas the formation of HLA-E– β₂m complexes requires the presence of leader peptides derived from classical class I molecules. Thus, the presence of β₂m-associated HLA-E on laryngeal epithelial cells (LEC), especially toward the luminal surface, was unlikely. At the same time, the expression of the group 2, β₂m-independent CD1 molecule, CD1d, has been described on various epithelial structures (26, 27), including the mucosa of the alimentary tract, prompting us to investigate the expression of this molecule. Our results show the expression of CD1d in the laryngeal epithelium with a unique topographic switching of expression of MHC class I^hiCD1d^lo epithelial cells in the basal layers gradually transitioning to the MHC class I^loCD1d^hi phenotype, as they move toward the lumen. This would suggest that, in LPR, there may be a change in the type of antigen presentation occurring in response to refluxate.

Although a similar pattern of CD1d expression has been reported on keratinocytes, with higher abundance of the protein toward the skin surface (28), this is related to the terminal maturation of these cells in the stratum corneum, paralleled with increasing involucrin levels (29). The authors proposed that this switching is a paraphenomenon associated with aging and cell death. However, laryngeal epithelial cells, by contrast, remain viable throughout, with live cells in contact with luminal challenges, including refluxate. Meanwhile, MHC class I was expressed at higher levels toward the stratum basale and declined toward the granulosum in skin (30), which mirrors our findings of classical MHC class I expression in the larynx.

In murine intestinal epithelium, it has been observed that CD1d is most prevalent apically and laterally within crypts (27), and this is also the case in human small intestinal epithelium (31, 32). We performed a limited study of human small intestine that confirmed this (data not shown). The relationship between MHC class I and CD1d expression we have observed in the laryngeal mucosa may be restricted to the upper airway, or to squamous mucosal epithelia generally. This is supported by our observation that MHC class I expression also diminishes progressively toward the luminal surface of human tonsillar epithelium (10). We therefore hypothesize that the observed switch in antigen-presenting molecules, with reduced MHC class I expression and the prevalence of CD1d at the lumen in laryngeal epithelium, has an important functional role, such as the induction and/or maintenance of tolerance to incident antigens. In addition, of all tested antigen-presentation pathways, only this CD1d-mediated antigen-presentation pathway showed significant changes in LPR.

NKT cells, CD3⁺ T lymphocytes that recognize glycolipid antigens presented by CD1d, are characterized by the expression of a limited TCRα/β receptor repertoire (22, 33) and the ability to produce cytokines very rapidly after TCR ligation (34). They play a role in protection against bacterial infections, the elimination of malignant cells, and the maintenance of self-tolerance. Given the observation of strong epithelial expression of CD1d, which is known to present certain bacterial antigens, the size of the laryngeal “bacteriome” (1), and the exposure of this epithelium to incident antigens and microbes, NKT cells may be functionally relevant at this site. Limitations of currently available anti-TCR antibodies and CD1d tetramer technologies did not allow us to investigate NKT cell numbers in the larynx in vivo directly using a single cell surface molecule (35, 36). Thus, we used CD161 as an indirect marker in our studies and where this colocalized with CD3 on lymphocytes, cells were classified as NKT cells. This approach has its limitations. Although NKT cells were initially described as TCRα/β⁺ T cells expressing NKR-P1 (the murine equivalent of CD161), this definition is no longer used. It is now clear that a subset of CD1d-reactive NKT cells does not express CD161, whereas a large fraction of CD161⁺ T lymphocytes express a diverse TCR repertoire.

Whether these latter cells respond to CD1d-restricted antigens by secreting the same cytokines remains to be determined. Furthermore, CD161 is almost universally expressed by NK cells. Nonetheless, studying the coexpression of CD161⁺CD3⁺ in the larynx allowed us to investigate the likelihood of NKT cell contact with the CD1d present on epithelial cells. The absence of positive association of CD161⁺CD3⁻ cells, NK cells, with CD1d suggested little functional interaction between the two in vivo, consistent with the lack of CD1d-reactive T-cell receptor on NK cells. By contrast, CD161⁺CD3⁺ cells significantly colocalized with CD1d, strongly suggesting that at least some of these were CD1d-restricted NKT cells. Taken together, the results indicate a physiologic switch from immunologic surveillance mediated by classical MHC class I antigen complexes recognized by CD8⁺ T cells near the basement membrane, to nonclassical MHC class I molecules presenting antigens to NKT cells, and possibly CD1d-restricted CD8⁺ T cells and NKT cells nearer the lumen. Furthermore, elevated CD1d expression by epithelial cells in patients with LPR indicates a regulation of this pathway by acid refluxate. There is some evidence that CD1d-responsive CD8⁺ cells may be involved in the pathogenesis of inflammatory bowel disease in humans (37). NK cells also express CD8 at low levels, and it is possible that because the antibody used to detect CD8 was raised against the α chain of the TCR, some of the positive staining observed was attributed to the presence of NK cells. Further studies of the interaction between antigens presented by CD1d on laryngeal epithelial cells and CD8⁺ T cells in the presence or absence of refluxate seem justified to provide further insights into the immunology of the upper airways.

In summary, we have observed a novel expression pattern whereby MHC class I^hiCD1d^lo epithelial cells in the basal layers gradually transition to MHC class I^loCD1d^hi epithelial cells in the luminal layers. Epithelial cells in culture constitutively expressed CD1d. These data suggest functional separation of luminal and basal layers in a manner not observed in nonsquamous mucosal surfaces, including the small intestine, a known site for the induction of mucosal tolerance. CD161⁺CD3⁺ (NKT) cells significantly colocalize to the CD1d layers, as do CD8⁺ T cells in reflux. These cells, CD8⁺ T cells, and CD1d expression by superficial epithelial cells were all increased in the mucosa of patients with reflux. We therefore hypothesize that the graded change from classical to nonclassical MHC class I expression between the layers of the laryngeal epithelium is crucial in the maintenance of the balance between upper airway tolerance and appropriate inflammatory responses. Selective regulation of the CD1d–NKT cell axis during the chronic inflammatory challenge of LPR appears to play a central role in the immunopathology of this condition. We propose, first, that this interaction is central to the pathogenesis of airway disease and, second, that the larynx is an important and accessible target for novel diagnostic and therapeutic strategies in a range of airway diseases.