Large pH oscillations promote host defense against human airways infection

Abstract

The airway mucosal microenvironment is crucial for host defense against inhaled pathogens but remains poorly understood. We report here that the airway surface normally undergoes surprisingly large excursions in pH during breathing that can reach pH 9.0 during inhalation, making it the most alkaline fluid in the body. Transient alkalinization requires luminal bicarbonate and membrane-bound carbonic anhydrase 12 (CA12) and is antimicrobial. Luminal bicarbonate concentration and CA12 expression are both reduced in cystic fibrosis (CF), and mucus accumulation both buffers the pH and obstructs airflow, further suppressing the oscillations and bacterial-killing efficacy. Defective pH oscillations may compromise airway host defense in other respiratory diseases and explain CF-like airway infections in people with CA12 mutations.

Graphical abstract

Figure 1.

Nasal pH oscillations in vivo. (A) Double-barreled H⁺-selective capillary electrode. (B) Upper and middle confocal images show x-z views of artificial ASL before and after positioning electrode in artificial ASL, respectively. Bottom image shows the x-y view of same electrode tip. ASL was visualized using SNARF (SemiNaphthoRhodaFluor)-dextran (red) and cell nuclei using DAPI (blue). Vertical white scale bar = 100 µm in middle image. Red area shows ASL meniscus. Dashed line in x-y section shows location of x-z section. These images are representative of five experiments. (C) Cartoon showing electrode orientation. (D) Representative voltage recording from the nasal mucosa of a non-CF subject during breathing followed immediately by calibration in buffers with pH values (from top to bottom) 6.96, 7.62, 8.01, 8.60, and 9.10. (E) Calibration curve for the electrode shown in D. (F) Typical pH recording during tidal breathing and breath holding. (G) Summary of pH oscillation amplitude (peak to peak) on nasal mucosae of healthy subjects (s1–s6, four male and two female subjects). The error bars represent mean ± SE.

Figure 2.

Large pH oscillations on cell cultures are catalyzed by membrane-bound carbonic anhydrase. (A) Confocal image of cells stained with FM-red and covered with 30 µl artificial ASL containing pH indicator BCECF-dextran. Scale bar = 10 µm. (B) Level of CO₂ at the inlet (red) and outlet (black) of the cell chamber showing washout kinetics. (C) pH oscillations induced by variable PCO₂ (dashed line) when artificial ASL contains 3.2 mM (red) or 10 mM (blue) HCO₃⁻, measured using BCECF-dextran. (D) pH during PCO₂ oscillations measured using a conventional microelectrode (Orion) in 100 ml artificial ASL that was bubbled vigorously using an airstone. (E) pH oscillations when mucus is removed (blue) or not removed (brown). (F) Rate of alkalinization after a step decrease in PCO₂ (5% → 0.035%) when mucous secretions are removed by washing (no mucus) and when they remain (with mucus). Mean ± SE, t test; n/Pvalue: a (6/0.000007). (G) Exogenous mucus blunts oscillations when ASL containing 10 mM HCO₃⁻ is exposed to varying PCO₂. (H) Rate of alkalinization after step decrease in PCO₂ (5% → 0.035%) with ASL containing 2.5% or 8% mucin solids. Mean ± SE, t test; n/Pvalue: a (6/0.0021) and b (6/0.00000041). (I) Buffer capacity (mM/pH) of mucins in the range of physiological pH oscillations (gray region). (J) Simulations show that pH oscillations do not reach full amplitude in the absence of carbonic anhydrase (CA) despite slow PCO₂ oscillations (dashed line). (K) Same as J but with carbonic anhydrase. (L) Alkalinization rates after step decrease in PCO₂ on control bronchial culture (Con.), with carbonic anhydrase inhibitor acetazolamide (Aceta., red), with exogenous soluble carbonic anhydrase (blue), or without washing off accumulated mucus (purple). (M) Summary of alkalization rates on bronchial epithelial cells obtained as in L. Mean ± SE, one-way ANOVA with post-hoc Tukey’s test. a, b, and c differences are significant; P = 1.8 × 10⁻¹² (n = 6–8). Topi., topiramate; WO wash, without washing; CA, carbonic anhydrase.

Figure S1.

Programmable fast gas-mixing pump. (A) Schematic of computer-driven valve and CO₂ controllers, humidification system, and cell chamber. (B) Recording of the PCO₂ supplied by the gas-mixing pump, measured using an infrared CO₂ analyzer (CD-3A; Applied Biosystem) at oscillation frequencies of 0.05 Hz and 0.1 Hz.

Figure 3.

Mucus buffers and carbonic anhydrase inhibitors diminish pH oscillations in vivo. (A) Nasal pH of healthy nonsmoker volunteers was measured with capillary ion-selective electrodes. The inferior turbinate on each side was lightly sprayed with saline solution, and pH was measured after ~1 min. (B) pH differences during expiration and inspiration (each point is mean of three measurements, 10 subjects). (C and D) Effect of exogenous sterile crude mucins (8 mg/ml) on pH oscillations. (E and F) pH excursions blunted after application of carbonic anhydrase inhibitor acetazolamide (500 µM). (G) Summary of the oscillation amplitudes showing effect of mucus and carbonic anhydrase inhibition on pH excursions. Mean ± SE t test, compared with control (n = 10 subjects); n/Pvalue: differences a (6/0.00016) and b (10/0.000036). Aceta., acetazolamide.

Figure 4.

Expression of membrane-associated carbonic anhydrases CA4, CA9, CA12, and CA14 in human and pig airway epithelia. (A) mRNA levels in primary human bronchial cells (n = 9–12 measurements from ≥3 cultures). Error bars represent SE. (B) CA12 mRNA levels normalized to GAPDH. Human: freshly isolated HBE cells, cultured non-CF and CF HBEs, and HEK cells (with endogenous CA12 expression as positive control). Pig: freshly isolated trachea, cultured PTE cells, skeletal muscle (negative control), and kidney cortex (positive control). Mean ± SE t test, compared with pig trachea; n/Pvalue: a (4/0.0048), b (4/0.016), and c (4/0.0778). (C) Immunoblot showing CA12 protein expression with tubulin as loading control. Fresh tissues were flash frozen in liquid N₂, pulverized, and extracted into lysis buffer. HEK cells were transfected with CA12 as positive control (HEK-CA12). Immunostaining was lost when antibody was preincubated with CA12 as a blocking protein. (D) Summary of relative levels of human and pig CA12 protein. Band densities were normalized to tubulin, and the normalized values were plotted relative to that for pig. Mean ± SE t test, compared with pig trachea; n/Pvalue: differences a (5/0.0036), b (5/0.0022), and c (4/0.0159). (E) Upper image: x-y view of HBE cells showing CA12 (red) at the apical pole (non-CF donor). Lower image: x-z view showing the apical localization of CA12 and acetylated tubulin in ciliated cells. White scale bar = 10 µm. AP, apical; BL, basolateral. (F) Localization of CA12 (red) and tubulin or MUC5AC (green) in isolated, well-differentiated HBE cells from non-CF donor. White scale bar = 10 µm. (G) Cartoon showing reversible hydration of CO₂ in the ASL catalyzed by membrane-bound CA12.

Figure S2.

CA12 is expressed in ciliated cells not goblet cells. (A) Well-differentiated HBEs from non-CF (upper images) and CF (lower images) donors show the colocalization of CA12 with tubulin. Left images show cilia as tubulin immunofluorescence. Middle images show strong CA12 expression after refocusing on the apical cell surface. Merging these images on the right reveals that CA12 expression is exclusively in ciliated cells. Top right image (x-z view) shows CA12 immunofluorescence immediately beneath cilia. AP, apical; BL, basolateral. Scale bars = 20 µm. (B) CA12 (red) and MUC5AC (green) immunostaining of well-differentiated non-CF (upper images) and CF (lower images) HBEs. CA12 and MUC5AC immunofluorescence was not observed in the same cells. CF cultures had more goblet cells and fewer CA12-expressing cells, which may explain the reduced CA12 expression shown in Fig.4. Scale bars = 20 µm.

Figure S3.

CA12 antibody specificity and immunostaining in human lung tissue. (A) Most CA12 immunostaining at the apical pole of ciliated cells was abolished by excess recombinant CA12 as a blocking protein. Scale bars = 10 µm. (B) CA12 signals (red) were detected only in ciliated cells in human lung tissue. Left image: ciliated cell marker tubulin (green) localized with CA12 immunofluorescence (red). Right image: goblet cell marker MUC5AC (green) did not colocalize with CA12 (red). Scale bars = 20 µm.

Figure S4.

CA12 immunostaining is reduced in CF bronchial and nasal cell cultures. (A) Apical localization of CA12 in bronchial and nasal brushings (left images) and well-differentiated bronchial and nasal cells isolated from air–liquid interface cultures. White scale bars = 10 µm. (B) Summary of CA12 immunofluorescence intensities measured as in A with identical microscope settings. Mean ± SE, one-way ANOVA with Tukey’s post-hoc test. a and b are significantly different; n/Pvalue: 6–8/7.8 × 10⁻⁶. (C) CA12 immunofluorescence in well-differentiated non-CF and CF HBEs (four images each condition). Fluorescence intensities were determined using the same settings and analyzed using IMARIS software. White scale bars = 20 µm. (D) Summary of all single-cell CA12 immunofluorescence intensities from 8–10 different cultures (n = 751 non-CF cells and 568 CF cells). Error bars represent mean ± SE. (E) Mean CA12 immunofluorescence intensities from individual cultures (each average of 50–100 cells was considered n = 1). Mean ± SE, Student’s t test; P = 0.0084. (F) Number of cells with CA12 immunofluorescence in cultures of cells from two non-CF and two CF patients. Mean ± SE, Student’s t test; P = 0.0033. A.U., arbitrary units.

Figure 5.

Excursions of pH are reduced in CF. (A and B) Recordings from non-CF (A) and CF (B) nasal mucosae showing more acidic basal pH and smaller alkalinizations during inhalation in CF subject with genotype 3876delA/1811+1643G>T. (C and D) Summary of nasal pH during inhalation and exhalation in non-CF and CF subjects. Mean ± SE, t test, n = 10; n/Pvalue: difference a (P = 7/1.6 × 10⁻⁷) and difference b (P = 7/0.0032). (E and F) Recordings of nasal pH during tidal breathing in CF subjects with genotypes G542X/711+1G>T (#1) and 2789+G>A/2347delG (#2), respectively. (G and H) Nasal pH during breathing in the same subjects as in E and F after nasal mucosa was lightly sprayed with 50 mM bicarbonate solution. (I and J) Surface pH on CF nasal epithelia during exhalation and inhalation when sprayed with saline lacking (I) or containing (J) HCO₃⁻, respectively. Subject genotypes: F508/W1282X, 3876delA/1811+1643G>T, G542X/711+1G>T, 1677delTA/1677delTA, 2789+5G>A/2347delG, F508/3272-26A>G, F508del/F508del (tested without HCO₃⁻); 1677delTA/1677delTA, G542X/711+1G>T, 3876delA/1811+1643G>T, and F508del/F508del (tested with HCO₃⁻). (K) Oscillation amplitudes on CF nasal surface sprayed with saline or 50 mM HCO₃⁻ solution. Mean ± SE, t test, comparison to control (n = 7); n/Pvalue: difference a (4/0.0017).

Figure 6.

Oscillations enhance bacterial killing at high pH. (A) pH dependence of Pseudomonas killing after 24-h exposure to air-equilibrated artificial ASL (mM: 115 NaCl, 1.2 CaCl₂, 1.2 MgCl₂, and 25 NaHCO₃) plus 100 mM NaHPO₄/Na₂PO₄ with ratio adjusted to give indicated pH. Mean ± SD, n = 3. (B) Time course of killing in air-equilibrated artificial ASL (pH ~9.2, mean ± SD, n = 3). (C) Antimicrobial effects of airway epithelial secretions. Bacteria were incubated for 6 h in fluid secreted by Calu-3 cells treated with DMSO (vehicle control) or FSK (10 µM) to stimulate bicarbonate secretion. (i) No proliferation when equilibrated with air (0.035% CO₂). Viable bacteria (CFUs) were unchanged in control secretions (blue) and reduced 10-fold in stimulated secretions (red). (ii) Proliferation in constant 5% CO₂. (iii) No proliferation or killing in constant 2.5% CO₂. (iv) Oscillations with median PCO₂ ≈ 2.5% prevented growth in control secretions but caused killing in stimulated secretions. (D) Comparison of antimicrobial effect of secretions when equilibrated with constant 2.5% CO₂ or with CO₂ oscillations with mean CO₂ = 2.5%. Dashed red line shows initial CFUs. Oscillations increase sensitivity to alkaline pH. Mean ± SE, t test, compared with 2.5% PCO₂ (n = 6); n/Pvalue: difference a (6/0.0207) and difference b (6/0.0079). (E) Antimicrobial effect of PCO₂ oscillations in artificial ASL (mM: 115 NaCl, 1.2 CaCl₂, 1.2 MgCl₂, and varying concentrations of NaHCO₃ from 3.2 to 50 mM). Mean ± SE, t test comparing constant PCO₂ versus oscillations (n = 4); n/Pvalue: differences a (4/0.0121), b (4/0.0035), c (4/0.0367), and d (4/0.0026). (F) Antimicrobial effect of 2-h exposure to 100 mM NaHPO₄/Na₂PO₄ solution without HCO₃⁻ adjusted to pH 7.4 (red) or 8.6 (blue; i); ii same as i but with 10 NaHCO₃ added and pH 7.6 and 9.3. Error bars in C and F represent SE. RLU, relative luminescence units.

Figure S5.

HCO₃⁻-dependent killing is enhanced by PCO₂ oscillations in the presence of nutrients. (A) Equilibration with room air (0.035% CO₂) is bacteriostatic when artificial ASL (115 mM NaCl, 1.2 mM CaCl₂, and 1.2 mM MgCl₂) + 1% TBS is supplemented with 13.2 mM HCO₃⁻ and bactericidal when it contains 50 mM HCO₃⁻. (B and C) Bacteria proliferate in nutrient-containing artificial ASL at both concentrations when exposed to constant 5% and 2.5% CO₂, respectively. (D) Oscillations that yield the same mean PCO₂ (2.5%) as in C prevent bacterial growth and may be bacteriostatic (13.2 mM HCO₃⁻) or bactericidal (50 mM HCO₃⁻). The error bars represent mean ± SE.