Nominal carbonic anhydrase activity minimizes airway‐surface liquid pH changes during breathing

Abstract The airway‐surface liquid pH (pHASL) is slightly acidic relative to the plasma and becomes more acidic in airway diseases, leading to impaired host defense. CO 2 in the large airways decreases during inspiration (0.04% CO 2) and increases during expiration (5% CO 2). Thus, we hypothesized that pHASL would fluctuate during the respiratory cycle. We measured pHASL on cultures of airway epithelia while changing apical CO 2 concentrations. Changing apical CO 2 produced only very slow pHASL changes, occurring in minutes, inconsistent with respiratory phases that occur in a few seconds. We hypothesized that pH changes were slow because airway‐surface liquid has little carbonic anhydrase activity. To test this hypothesis, we applied the carbonic anhydrase inhibitor acetazolamide and found minimal effects on CO 2‐induced pHASL changes. In contrast, adding carbonic anhydrase significantly increased the rate of change in pHASL. Using pH‐dependent rates obtained from these experiments, we modeled the pHASL during respiration to further understand how pH changes with physiologic and pathophysiologic respiratory cycles. Modeled pHASL oscillations were small and affected by the respiration rate, but not the inspiratory:expiratory ratio. Modeled equilibrium pHASL was affected by the inspiratory:expiratory ratio, but not the respiration rate. The airway epithelium is the only tissue that is exposed to large and rapid CO 2 fluctuations. We speculate that the airways may have evolved minimal carbonic anhydrase activity to mitigate large changes in the pHASL during breathing that could potentially affect pH‐sensitive components of ASL.

The pH ASL is influenced by a number of H + -and HCO 3 --dependent mechanisms (Fischer and Widdicombe 2006). HCO 3 is a powerful buffer because it is in an open-buffering system with CO 2 , following the elementary equilibrium reaction: Within the cytoplasm and some extracellular fluid compartments, the CO 2 /HCO 3 buffering system is catalyzed to a nearly diffusion-limited rate by carbonic anhydrase, 1.6 9 10 6 sec À1 for carbonic anhydrase isozyme II (Supuran 2008). For most cell types and compartments, the CO 2 concentration is a nearly constant 5%. However, the CO 2 concentration in large proximal airways changes during the respiratory cycle (Cochrane et al. 1982). This change is observed in a clinical setting by variations in end-tidal CO 2 . During inspiration (1-2 sec), the CO 2 concentration in the large airway lumen approaches that of air (0.04% CO 2 ), while during expiration (2-4 sec) the CO 2 concentration approaches 5%. If carbonic anhydrase is present in the ASL, then fast CO 2 equilibration will change pH ASL in the large airways during breathing. Here, we tested this prediction using differentiated cultures of porcine airway epithelia. We were surprised to discover that changing the CO 2 concentration between 0.04% and 5% produced little change in pH ASL over a time course associated with respiration (i.e., a few seconds). We found that nominal carbonic anhydrase activity prevented short-term changes in pH.

Ethical approval
The protocol for isolation of porcine airway epithelia was reviewed and approved by the University of Iowa Animal Care and Use Committee. Newborn piglets were obtained from Exemplar Genetics (Exemplar Genetics, Sioux Center, IA) and were euthanized upon arrival by an overdose of Euthasol via cardiac puncture after sedation with 20 mg/kg Ketamine and 2 mg/kg xylazine.
Isolation, expansion, and culture of porcine airway epithelia Piglet lungs were excised and proximal large airways, including trachea and main stem bronchi, were dissected out from the airway tree as described ). Next, primary porcine airway epithelia were isolated according to an adapted procedure originally developed for tracheal airway cells. Isolated large airway cells were expanded as recently described ) using a method developed to conditionally reprogram airway epithelial cells Suprynowicz et al. 2012). Briefly, large airway cells were cultured in F media in the presence of 10 lmol/L Y-27632, a ROCK inhibitor, and low passages of irradiated fibroblast feeder cells NIH-3T3-J2 obtained from Dr. H. Green's laboratory at Harvard University (Rheinwald and Green 1975) and maintained at 37°C with 5% CO 2 . After two passages of amplification, expanded epithelial cells were separated from feeder cells and seeded onto collagen-coated, semipermeable membranes (Corning #3470) at density of 10 6 cells/cm 2 and cultured at the air-liquid interface 37°C in a 5% CO 2 atmosphere as previously described (Zabner et al. 1996) for 2-3 weeks in the absence of feeder cells and ROCK inhibitor. In the first week of seeding at the air-liquid interface, cells were maintained in Small Airway Growth Media (Lonza, Basel, Switzerland) supplemented with 10 ng/mL keratinocyte growth factor for 1 week, after which cells were maintained in USG media. We used the expanded epithelial cells between passage 2 and passage 5 to obtain consistent results.

Tissue RNA isolation and PCR
Tissue samples from large airway epithelial cells, kidney, intestine, and salivary gland were fixed and stored frozen in RNAlater solution (Invitrogen) followed by thawing and three washes of PBS without divalent cations. Next, the tissues were homogenized in TRIzol reagent solution (Invitrogen) and RNA was extracted using Invitrogen's PureLink RNA Mini Kit according to the manufacturer's protocol. The samples were then processed to cDNA by loading 1000 ng of RNA from each sample using Super-Script IV VILO Master Mix (Invitrogen) according to the manufacturers' protocol. cDNA concentrations of each sample were quantified using NanoDrop 2000 (Thermo-Fisher), then diluted to 500 ng/lL. Using Kapa HiFi HotStart ReadyMix PCR Kit a reaction mixture for PCR was made according to the manufacturer protocol using 100 ng of cDNA from each sample with each of the following primer sets for membrane-bound CA isoforms; CA-IX F: 5 0 -TCCCACCACAGGAGGAGATT-3 0 IX R: 5 0 -ATCCCTGGGAGCCTCAGTAG-3 0 CA-XII F: 5 0 -CCCACTCAACGGATCCAAGT-3 0 XII R: 5 0 -CAAACTGCTGGTTGGCAGTC-3 0 Pig Beta Actin F: 5 0 -CTGCGGCATCCACGAAAC-3 0 R: 5 0 -GTGATCTCCTCCTGCATCCTGTC-3 0 The thermocycler was programmed with the following settings: Initial denaturation of 3 min at 95°C, followed by 25 cycles of denaturation for 20 sec at 98°C, annealing for 15 sec at 63°C, and extension for 15 sec at 72°C; finishing with a final extension of 1 min at 72°C. PCR product was run using electrophoresis on a 5% agarose gel.
pH ASL measurements pH ASL was measured using the ratiometric fluorescent pH indicator SNARF-1 conjugated to dextran (Molecular Probes, Eugene, Oregon), as previously reported (Pezzulo et al. 2012). SNARF has an estimated response time of 60 msec (Chen et al. 2010b), which is faster than observed pH changes in this study (10 sec to 5 min). SNARF-1 dextran was applied to the airway-surface liquid as a finely strained powder and 2 h later epithelia were placed in a humidified chamber with 5% CO 2 /21% O 2 /balanced N 2 heated to 37°C and examined by confocal microscopy (Zeiss 510 Meta NLO). pH was calculated as previously described (Pezzulo et al. 2012), but using a nonlinear fit, as pH ASL was not completely within the linear range of the dye in this study (e.g., when exposed to apical air). Calibration curves were fitted in GraphPad Prism with a modified acid-base titration equation (Boyarsky et al. 1988): Where variables a and b, as well as pKa, are obtained from the fit. The ratio observed is the recorded 580/640 emission. pH and pKa have their classic definitions. In this study, SNARF-1 dextran had a pKa of 7.44 consistent with 37°C values reported by various groups (Buckler and Vaughan-Jones 1990;Blank et al. 1992;Westerblad et al. 1997;Ch'en et al. 2003). In some experiments, the CO 2 was changed from 5% CO 2 to air (compressed air containing 0.04% CO 2 ) every 4 min and the change in pH ASL was monitored in real time as described above. To test carbonic anhydrase activity, cells were pretreated with 20 lmol/L acetazolamide (Sigma-Aldrich, St. Louis, MO), a carbonic anhydrase inhibitor, or 1.5 U carbonic anhydrase isozyme II (Sigma-Aldrich, St. Louis, MO), in a final volume of 0.5 lL on the apical surface, then pH ASL was monitored as described above.

Kinetic pH ASL analysis
The experimental data (n = 5-8 independent experiments performed on cells isolated from N = 3-4 pigs) were fitted using the least-squares method in GraphPad Prism with the following equations: Association phase (removal of CO 2 ): Decay phase (addition of CO 2 ): Where pH ASL is the instantaneous ASL pH, t is the instantaneous time, t 0 is the time of the removal (association equation) or addition (decay equation) of CO 2 , pH ASL0 is the baseline pH ASL , pH ASLmax is the fitted maximal pH ASL (association equation only), pH ASL min is the fitted minimum pH ASL (decay equation only), and k is the obtained rate constant for the fit. R 2 values were >0.98 for all fitted data.
To calculate the mean pH, the following equation was used: Where pH mean is the mean pH of the proton activities of pH 1 (steady state under 5% CO 2 ) and pH 2 (steady state under air).
To determine the rate constants for large airway epithelia exposed to acetazolamide, carbonic anhydrase, or control treatment, the data were fitted using the least-squares method. Consistent with carbonic anhydrase being an enzyme, we observed that equilibrium pH ASL values for each experiment were not different. Therefore, we could compare rate constants among these three treatments.

Simulation of pH ASL changes in the respiratory cycle
A program to calculate real-time pH ASL changes during the respiratory cycle was coded in MATLAB version R2015a (Mathworks) and is freely available upon request. User-defined input values are the respiration rate, inspiratory:expiratory (I:E) ratio, initial pH ASL , and fitted curves obtained using the association and decay equations described above. The program uses pH-dependent rates, which accommodates for pH-dependent unknown variables (e.g., intrinsic buffering capacity) to predict pH ASL values in the time domain. Briefly, the program computes the rate of pH change (dpH/dt) along the fitted inspiration curve for a user-defined initial pH value and outputs a computed pH with a user-defined time binning (10 points for the smallest respiratory phase in this study). This process is looped for each bin until the end of inspiration is achieved. After completing the inspiration phase, ª 2018 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society the program begins the same process on the expiratory curve initiating at the final pH of the inspiration phase. The program then loops between these two phases until a user-defined simulation time is reached. Data were output to a Microsoft Excel file. From these simulations, the mean oscillatory pH (pH OSC ) and the root mean square (RMS) of the oscillatory pH (pH RMS ) were computed. The pH RMS was calculated using the RMS equation for a triangular wave: Where a is the amplitude of the triangular wave. In some simulations, the steady-state simulation was acidified or alkalinized and the time to recovery to baseline was monitored and fitted with the association (acid addition) or decay equations (alkaline addition) defined above in prism to obtain time constants (s).

Statistics
Data are expressed as mean AE SEM or regressions fitted by the least-squares method. A P value less than 0.05 was accepted as significant. Means were compared using ANOVA with Bonferroni or Dunnet correction (for comparing acetazolamide and carbonic anhydrase application to control), and unpaired, or paired t-tests where appropriate. Statistical significance among regressions was computed by comparing the residual sum-of-squares to obtain an F ratio. P values were obtained in Excel for a given F ratio with corresponding degrees of freedom.

Results
Rate of pH ASL change in response to CO 2 To assess how changes in CO 2 affect pH ASL , we studied differentiated cultures of porcine airway epithelium. We varied apical CO 2 between 5% and 0.04% (air) while maintaining basolateral CO 2 at 5%. Switching from 5% CO 2 to air led to a slow pH ASL change that plateaued after~4 min (Fig. 1A). This change was reversible upon reexposure to 5% CO 2 . These changes occurred with a time constant of approximately 50 sec.
The reaction CO 2 + H 2 O ↔ H + + HCO 3 is catalyzed by carbonic anhydrase (Kim et al. 2014). Because the observed pH change is slow, we predicted that carbonic anhydrase activity is low within intact ASL. We first assessed mRNA transcripts of carbonic anhydrase isozymes (Ca) predicted to be enzymatically active outside of cells, which include the membrane-bound and secreted carbonic anhydrase isozymes. Microarray analysis from Li et al. 2016 (Fig. 1D) revealed that most mRNA transcripts for these isozymes were present at levels lower than Cftr mRNA transcripts, which is present in low abundance in airway epithelia (Trapnell et al. 1991;Chu et al. 1992). CaXII mRNA transcripts were present at similar levels to Cftr, however, carbonic anhydrase XII protein has low enzymatic activity (Tureci et al. 1998). We also investigated the expression of CaIX and CaXII in pig trachea by RT-PCR. As shown in Figure 1E CaIX and CaXII are expressed in intestine and renal cortex tissues, respectively, but not in pig trachea.
To test the hypothesis that ASL upon differentiated airway cultures has minimal carbonic anhydrase activity, we performed two experiments. We added the carbonic anhydrase inhibitor acetazolamide and found it did not affect pH ASL kinetics induced by apical CO 2 changes (Fig. 1B). In contrast, adding recombinant carbonic anhydrase to the ASL accelerated the rate of pH ASL change in response to changes in apical CO 2 (Fig. 1C). Rate constant analysis (Fig. 1F and G) suggests that airway epithelia lack substantial carbonic anhydrase activity resulting in slow changes in pH ASL when CO 2 concentration changes. These slow changes would minimize the change in pH ASL during breathing.

Computational modeling of dynamic pH ASL changes in airway epithelia
To further explore how respiratory cycle properties (e.g., respiratory rate and inspiratory:expiratory (I:E) ratio) affect pH ASL , we performed computational modeling using the pH-dependent rates obtained from Figure 1. This approach models the real-time pH changes and allows for integration of the pH-dependent processes (e.g., ASL buffering capacity).
The initial model parameters included a 15 breath per minute (bpm) respiratory rate and an I:E ratio of 1:2, which reflects a typical human respiratory cycle (Cheng et al. 2010). To validate the model, we performed simulations using data beginning at the minimum pH ASL value (i.e., the pH ASL for 5% CO 2 ) or the maximum pH ASL value (i.e., the pH ASL for air) for each culture. Example simulations for one culture are shown in Figure 2A. For these simulations, we quantified the magnitude of the pH oscillations as root-mean-squared values (pH RMS ) and the mean pH during steady-state oscillations (pH OSC ) (Fig. 2B). For the experiment shown, a pH OSC of 7.36 was reached after~4 min with little oscillation (pH RMS = 0.002 pH units). These simulation data suggest that the pH ASL is more alkaline during the respiratory cycle than in the constant presence of 5% CO 2 , under which most airway culture studies are performed, and the pH does not fluctuate appreciably from breath to breath.
Next, we modeled changes in the respiratory rate (15, 20, and 60 breaths per minute) and the I:E ratios to mimic physiologic or pathologic scenarios (1:2physiologic ratio, 1:4obstructive airway disease, and 4:1 -  2C) and was directly correlated to the I:E ratio (Fig. 2D). pH RMS was inversely correlated to respiratory rate (Fig. 2E), but unaffected by the I:E ratio (Fig. 2F). In summary, the I:E ratio affects the mean steady-state pH (pH OSC ) and the respiration rate affects the magnitude of pH oscillations (pH RMS ). Based on our rate analysis, exogenous carbonic anhydrase would increase pH oscillations. To demonstrate this, the simulation was repeated using rates obtained when we added exogenous carbonic anhydrase, vehicle, or acetazolamide (Fig. 3A). The presence of carbonic anhydrase increased ASL pH RMS (Fig. 3B), but not steady-state pH OSC (Fig. 3C) relative to controls. These results were expected because adding carbonic anhydrase enzymatic activity should alter the pH kinetics, but should not alter the pH equilibrium achieved.
Finally, we modeled pH ASL recovery from simulated acid or alkaline addition to the ASL (Fig. 3D). We found no difference in recovery to an effective 20 nmol/L acid or alkaline addition under control conditions or in the presence of acetazolamide, whereas applying carbonic anhydrase significantly accelerated the time to recovery (Fig. 3E). These data indicate that lack of appreciable airway carbonic anhydrase activity in the ASL would minimize pH ASL changes during respiration and exogenous carbonic anhydrase may help acidic airways achieve a more alkaline pH, at least for a fraction of the respiratory cycle (i.e., flanking peak inspiration). Figure 1. Alterations in pH ASL for large airway cultures in response to changes in CO 2 . A-C, Cells were exposed bilaterally to 5% CO 2 , followed by an apical shift to air (noted on traces as "Air"). After monitoring pH ASL for 5 min, apical gas was subsequently shifted back to 5% CO 2 . These maneuvers changed pH ASL from~7.2 to~7.7 in each condition. Each experiment was fitted (black-dotted lines) using the exponential equations described in the Methods. (A) Control conditions. (B) Cells were pretreated with 20 lmol/L acetazolamide, a carbonic anhydrase inhibitor, on the apical surface. (C) Cells were pretreated with 1.5 U carbonic anhydrase on the apical surface. Experiments were performed on matched samples isolated from the same animal. Each trace is a representative example of one experiment. (D) RNA sequencing microarray data for large airway epithelial cells cultured at the air-liquid interface. (E) Pig trachea RT-PCR for CA IX (control: intestine) and CA XII (control: renal cortex). (F and G) Average rate constants for airway epithelia when CO 2 was shifted from (F) 5% CO 2 to air or (G) air to 5% CO 2 . Each dot represents one culture. CTL: control; ATZ: acetazolamide; CA: exogenous carbonic anhydrase.

Discussion
Because the large airway is exposed to oscillations in CO 2 during the respiratory cycle, our goal was to understand how varying CO 2 concentration affects the pH ASL on differentiated airway epithelia maintained at the air-liquid interface. We found that, after altering CO 2 , the time to plateau was on the order of minutes. In contrast, respiratory cycles occur within seconds and therefore pH ASL changes are small during breathing. The slow change in pH ASL was a result of low carbonic anhydrase activity of the ASL. These data are consistent with previous reports that ASL removed from epithelia has minimal carbonic anhydrase activity (Candiano et al. 2007;Kim et al. 2014). Thus, loss of carbonic anhydrase activity in the bulk ASL may represent an evolutionary selection to minimize dynamic changes in pH ASL in response to respiration.
At least 16 different carbonic anhydrase isoforms have been described in mammals Imtaiyaz Hassan et al. 2013). The most relevant isoforms to this study are the membrane-bound and secreted forms, including carbonic anhydrase IV, VI, IX, XII, and XIV (Pinard et al. 2015). Carbonic anhydrase IX has been observed in human airways, but only under hypoxic conditions (Polosukhin et al. 2007(Polosukhin et al. , 2011. Carbonic anhydrase VI, also known as gustin, is the only isoform that is secreted, and it is expressed highly in the salivary glands (Thatcher et al. 1998). Carbonic anhydrase VI has been found by immunocytochemistry in rat lung (Leinonen et al. 2004) and by mRNA detection in human neuroendocrine bodies (Livermore et al. 2015), but not on surface Figure 2. Airway pH ASL is more alkaline during the respiratory cycle than in the constant presence of 5% CO 2 , with minimal breath-to-breath fluctuations. (A) Modeling pH ASL changes during respiratory cycles with a 15 breath per minute (bpm) respiratory rate and 1:2 I:E ratio. Beginning the simulation at varying pH values does not change steady-state pH achieved. (B) Expansion of the steady-state pH changes from panel A. Parameters that correspond to the mean pH during steady-state oscillation (pH OSC ) and the bandwidth of the pH oscillation (pH RMS ) are noted. (C) The I:E ratio was held constant at 1:2, and the respiratory rate was modeled at 15, 20, and 60 bpm. Because pH OSC will be influenced by the steady-state pH values of a given culture, simulated pH OSC was plotted as a function of the midpoint H + activity between 5% CO 2 and air. There was no change in pH OSC . (D) The respiratory rate was held constant 15 bpm (typical human breathing cycle), and the I:E ratios were modeled at 1:2 (physiologic ratio), 1:4 (obstructive airway disease), and 4:1 (airway pressure release ventilation). The pH OSC is directly correlated with relative phase lengths. (E and F) pH RMS varied inversely with the respiration rate (E) and was not affected by the I:E ratio (F). bpm: breaths per minute. (C and D) Filled circles represent a simulation for rates obtained from different cultures. (E and F) Error bars represent standard error of simulations performed using rates obtained from different cultures.  (Livermore et al. 2015). Recent immunohistochemistry data suggest that carbonic anhydrase XII is expressed in the terminal bud of human respiratory epithelial cells (Lee et al. 2016). Mutations in carbonic anhydrase XII have been implicated in lung disease (Lee et al. 2016), but the mechanism remains unclear. Regardless, we speculate that carbonic anhydrase XII is not likely to regulate pH ASL because carbonic anhydrase XII's subcellular localization suggests lack of direct contact with the ASL (Lee et al. 2016). In addition, carbonic anhydrase XII has low enzymatic activity that is 15% of carbonic anhydrase IV (Tureci et al. 1998). This notion is consistent with our data that the ASL of large airways has minimal carbonic anhydrase activity.
Respiratory rates vary in health and disease. The average respiratory rate in healthy resting humans is approximately 15-20 bpm. During exercise or during respiratory failure, the respiratory rate can increase to 60 bpm. Our modeling data suggest that these changes would have minimal effect on pH ASL . In patients with obstructive lung disease, such as asthma or chronic obstructive pulmonary disease (COPD), the expiratory phase is increased. Our data suggest that the decrease in pH ASL due to the I:E ratio would be modest (~0.1 pH units). Inverse ratio ventilation is routinely used in intensive care units to ventilate patients that do not respond to conventional mechanical ventilation (Cole et al. 1984). Our model showed that an I:E ratio of 4:1 would significantly increase pH ASL .
Why would the airway evolve to lack carbonic anhydrase in the ASL? Based on the Henderson-Hasselbalch equation, fluctuations in CO 2 from 5% to 0.04% in the presence of 20 mM HCO 3 and carbonic anhydrase could result in a pH change from 7.4 to 9.5. Thus, we speculate that the large airway evolved to have minimal carbonic anhydrase activity and nonbicarbonate buffering capacity to mitigate extreme alkaline shifts in pH ASL . Additionally, airway lacking carbonic anhydrase may have been selected for based upon beneficial pH-dependent enzymes, such as b-defensin-3 and LL-37 (Abou Alaiwa et al. 2014b). Simulations with carbonic anhydrase (Fig. 3A) have a pH oscillation magnitude of~0.1 unit, which could affect enzyme function. For example, consider an enzyme whose function is defined by a titration curve and pKa equals that of the midpoint of the ASL pH oscillation. Then an oscillation of 0.1 (midpoint AE 0.05 units) would undergo a~11.5% change in protonation from peak to peak. In contrast, a pH change of 0.01 (midpoint AE 0.005 units), as seen in the absence of exogenous carbonic anhydrase (Fig. 3A), would change the theoretical enzyme protonation state by only~1% from peak to peak.
Our study also has limitations. First, we did not remove mucus from cultured epithelia. However, mucus is present in vivo and it is proposed to have only a minor role in buffering at the physiological pH ASL values (Holma and Hegg 1989;Kim et al. 2014). During inhalation, where CO 2 is low, pH is~7.8, and therefore HCO 3 -is~500 lmol/L, high concentrations of mucus may provide additional buffering capacity through its free cysteines (pKa~8.5). Second, the model does not take into consideration submucosal glands that are present in an intact lung. However, acidic submucosal gland secretions (Widdicombe and Wine 2015) are anticipated to be similar to the perturbations associated with acid loading modeled in Figure 3. Third, prolonged changes in apical CO 2 , from which our rates were obtained, may have a modest effect on the intracellular pH under the apical membrane. Fourth, additional buffering capacity may be excreted in vivo by the epithelium or the presence of other cell types (e.g., leukocytes during disease). None of these limitations would affect our conclusion that pH ASL changes were slow in response to CO 2 due to nominal carbonic anhydrase activity.
In summary, pH ASL oscillations in the respiratory cycle are small, avoiding extreme pH values that may be caustic to airway epithelia. However, given that pH ASL is more acidic in many airway diseases (e.g., in CF due to decreased HCO 3 secretions) and contributes to host defense defects, delivering or increasing ASL carbonic anhydrase may be useful to increase pH ASL . An increase in carbonic anhydrase activity would increase the pH oscillation magnitude, achieving more alkaline pH values for a fraction of the respiratory cycle.