Inhibition of the sodium-dependent HCO3- transporter SLC4A4, produces a cystic fibrosis-like airway disease phenotype

Bicarbonate secretion is a fundamental process involved in maintaining acid-base homeostasis. Disruption of bicarbonate entry into airway lumen, as has been observed in cystic fibrosis, produces several defects in lung function due to thick mucus accumulation. Bicarbonate is critical for correct mucin deployment and there is increasing interest in understanding its role in airway physiology, particularly in the initiation of lung disease in children affected by cystic fibrosis, in the absence of detectable bacterial infection. The current model of anion secretion in mammalian airways consists of CFTR and TMEM16A as apical anion exit channels, with limited capacity for bicarbonate transport compared to chloride. However, both channels can couple to SLC26A4 anion exchanger to maximise bicarbonate secretion. Nevertheless, current models lack any details about the identity of the basolateral protein(s) responsible for bicarbonate uptake into airway epithelial cells. We report herein that the electrogenic, sodium-dependent, bicarbonate cotransporter, SLC4A4, is expressed in the basolateral membrane of human and mouse airways, and that it’s pharmacological inhibition or genetic silencing reduces bicarbonate secretion. In fully differentiated primary human airway cells cultures, SLC4A4 inhibition induced an acidification of the airways surface liquid and markedly reduced the capacity of cells to recover from an acid load. Studies in the Slc4a4-null mice revealed a previously unreported lung phenotype, characterized by mucus accumulation and reduced mucociliary clearance. Collectively, our results demonstrate that the reduction of SLC4A4 function induced a CF-like phenotype, even when chloride secretion remained intact, highlighting the important role SLC4A4 plays in bicarbonate secretion and mammalian airway function.

absence of detectable bacterial infection. The current model of anion secretion in mammalian 23 airways consists of CFTR and TMEM16A as apical anion exit channels, with limited capacity for 24 bicarbonate transport compared to chloride. However, both channels can couple to SLC26A4 25 anion exchanger to maximise bicarbonate secretion. Nevertheless, current models lack any details 26 about the identity of the basolateral protein(s) responsible for bicarbonate uptake into airway 27 epithelial cells. We report herein that the electrogenic, sodium-dependent, bicarbonate 28 cotransporter, SLC4A4, is expressed in the basolateral membrane of human and mouse airways, 29 and that it's pharmacological inhibition or genetic silencing reduces bicarbonate secretion. In fully 30 differentiated primary human airway cells, SLC4A4 inhibition induced an acidification of the 31 airways surface liquid and markedly reduced the capacity of cells to recover from an acid load. 32 Studies in the Slc4a4-null mice revealed a previously unreported lung phenotype, characterized by 33 mucus accumulation and reduced mucociliary clearance. Collectively, our results demonstrate that 34 the reduction of SLC4A4 function induced a CF-like phenotype, even when chloride secretion 35 remained intact, highlighting the important role SLC4A4 plays in bicarbonate secretion and 36 mammalian airway function.

INTRODUCTION. 38
Bicarbonate (HCO 3 -) and chloride (Cl -) are actively secreted in the lumen of airways by the lining 39 epithelial cells. Even though, the pathophysiological consequences of decreased secretion of these 40 anions has been extensively documented in cystic fibrosis (CF), the most common, autosomal 41 recessive, disease in humans, there is still much discussion whether HCO 3 secretion per se affects 42 airway homeostasis (1, 2). Impaired Clsecretion reduces the volume of the fluid that covers the 43 airway epithelium, the airway surface liquid (ASL), leading to ciliary dysfunction and promoting 44 mucus stasis and airway obstruction (3, 4). Deficient HCO 3 secretion reduces ASL pH which 45 compromises post-secretory mucin maturation and clearance (5), impairs the antimicrobial 46 function of epithelial cells (6, 7) and increases fluid absorption, further decreasing ASL hydration 47 (8). Whilst, direct measurement of pH in the distal airways of CF children showed no acidification 48 of the ASL (9), it has been demonstrated that the addition of HCO 3 induced an increase in ASL 49 height in human airway epithelial cells (hAECs) cultures and restored the normal properties of 50 mucus from CF patients (8, 10). Moreover, the use of aerosolized HCO 3 into CF-pig airways 51 increased bacterial killing, clearly indicating that HCO 3 supplementation can correct inherent 52 defects of CF in the airways (6, 11). 53 Even though, impaired HCO 3 secretion is recognised as a detrimental component of CF airway 54 disease, a full mechanistic understanding of the process and players involved in transcellular HCO 3 -55 transport in the airways is lacking. In normal airways, HCO 3 is secreted by CFTR and TMEM16A 56 channels, but after inflammatory signalling HCO 3 secretion is augmented through increased 57 expression of the Cl -/HCO 3 exchanger Pendrin (SLC26A4) (12-16). Importantly, the mechanism of 58 basolateral HCO 3 transport/uptake in native pulmonary tissues is still lacking. We reasoned that 59 the identification and functional inhibition of such basolateral membrane proteins could be used as a proof-of-concept to better understand the role of HCO 3 secretion in airway homeostasis 61 without altering Clsecretion. Here using in vitro fully differentiated human bronchial epithelial 62 cells, we identified the Na + -dependent, HCO 3 electrogenic cotransporter (NBCe1), SLC4A4, as the 63 basolateral protein responsible for HCO 3 influx, which couples to apical CFTR for HCO 3 secretion 64 into the ASL. Importantly, SLC4A4 inhibition induced acidification of ASL, revealing the pivotal role 65 of this cotransporter in airway pH homeostasis. These observations were further tested in murine 66 models, and revealed that SLC4A4 is involved in both basal and Ca 2+ -stimulated HCO 3 secretion in 67 mice airways. Strikingly, an Slc4a4 -/mouse model showed significant pathological signs of muco-68 obstructive disease and reduced mucociliary clearance, confirming that inhibition of HCO

Airway cells isolation and intracellular pH determinations: Tracheae were incubated with 185
Pronase 25 µg/ml at 37°C for 30 minutes. Then the trachea was placed in a petri dish with DMEM-186 F12, and the airway epithelium was dissociated by scrapping with tweezers, the cells were 187 collected and spun at 3000 r.p.m. for 5 minutes at room temperature and the supernatant was 188 removed. The cell pellet was incubated with 500 µl of trypsin 1X at 37°C for 5 minutes and 189 centrifuged at 3000 r.p.m for 5 minutes at room temperature and the supernatant was removed. computed and transformed to pH units by performing a pH-clamp. Briefly, cells were exposed to 5 206 µM nigericin and 20 µg/ml gramicidin in a buffer composed of (in mM) 10 HEPES, 129 KCl , 10 207 NaCl, 1.25 MgCl2, 1 EGTA, 10 glucose, with pH values ranging between 6.8-7.8 and the observed 208 changes in fluorescence were quantified and used to construct a calibration curve.    (Table S1) for each NCBT revealed that SLC4A4, A5 and A8 are expressed at 286 mRNA levels in primary hAECs from 3 different individuals (P1, P2, P3, Fig S1A). SLC4A7 and 287 SLC4A10 showed very low to none level of mRNA expression (Fig S1A). Interestingly, isoforms B/C 288 of NBCe1 (known as the pancreatic isoform) were more highly expressed than isoform A (known 289 as the kidney isoform). 290

cells. 292
We then tested whether there was an active NCBT under resting conditions in primary hAECs. The 293 cell cultures were mounted in Ussing chambers in buffers containing either HCO 3 -(but no Cl -) or 294 HEPES, and treated basolaterally with the inhibitor S0859 (30 μM). Results, shown in Figure 1  295 confirmed that, under basal conditions, HCO 3 secretion was inhibited by S0859 (Fig 1A,B) and that 296 this pharmacological inhibitor did not have any effect on short-circuit current (Isc) in the absence 297 of HCO 3 - (Fig 1C,D). These results show that there is an electrogenic HCO 3 transporter at the 298 basolateral membrane of hAECs, which is consistent with SLC4A4, since SLC4A7 and A8 are 299 electroneutral, and SLC4A5 has been shown to be localized to the apical membrane of renal 300 epithelial cells (25). Next, we used intracellular pH (pH i ) measurements to functionally investigate 301 NBCe1 activity using a CO 2 -induced acidification protocol (26). As shown in Figure 1E, exposing 302 cells bilaterally to a HCO 3 -/CO 2 -gassed KRB solution induced a transient acidification. On the other 303 hand, an apical-only CO 2 exposure, in the absence of basolateral HCO 3 -, induced a sustained 304 acidification (of the same amplitude as with bilateral HCO 3 -/CO 2 , Figure 1F,H), that recovered 305 when HCO 3 was re-introduced basolaterally ( Figure 1F,I). This pH i recovery depended on the 306 presence of Na + in the basolateral solution ( Figure 1G) consistent with a Na + -coupled HCO 3 -307 transporter, which was confirmed using S0859 which blocked the pH i recovery from the CO 2 -308 induced acidification. In order to isolate NBCe1 only changes in pH i , the contribution of Na + /H + 309 exchanger NHE was inhibited using 100 μM Dimethyl amiloride (DMA). In this condition, S0859 still 310 significantly decreased the rate of pH i recovery from the CO 2 -induced acidification ( Figure 1J,K). 311

Basolateral HCO 3 uptake is essential for ASL pH homeostasis. 312
In order to test whether HCO 3 transport by basolateral SLC4A4 impacted apical HCO 3 secretion, 313 we measured the effect of S0859 on ASL pH. First, S0859 was added basolaterally to primary 314 hAECs and ASL pH continuously measured. NBCe1 inhibition significantly decreased ASL pH under 315 resting conditions (Fig 2A,B), and partially prevented the forskolin-induced increase in ASL pH 316 which we have previously shown was via CFTR (19, 20) (Fig 2C,D). Moreover, when S0859 was 317 added after forskolin, it significantly reduced the forskolin-induced, CFTR-dependent, increase in 318 ASL pH (Fig 2E,F) confirming the central role of SLC4A4 cotransporter in ASL pH homeostasis under 319 both resting and stimulated conditions. 320 Finally, immunolocalization of SLC4A4 protein in human airways showed intracellular and 321 basolateral membrane staining in epithelial cells that were also positively stained for acetylated-322 tubulin indicating that SLC4A4 is preferentially expressed in ciliated cells in human airways (Fig  323   2G,H). 324

Bicarbonate secretion is calcium-activated in mouse trachea.
To investigate the expression of SLC4 exchangers in mouse airway epithelium, we performed RT-326 PCR of epithelial cells from mouse tracheas and observed that several members of the SLC4 family 327 including Slc4a4, Slc4a5, Slc4a7 and Slc4a10 were expressed (Fig S2). Studies of Slc4a4 isoforms 328 demonstrated that isoform B/C but not isoform A was expressed in mouse airways. Next we 329 characterized HCO 3 secretion in the mouse. Ussing chamber experiments performed in freshly 330 excised mouse trachea using HCO 3 - (Fig 3A) or HEPES (Fig 3B) buffered solutions, showed that UTP-331 induced an anion current that was significantly reduced in the absence of HCO 3 -(-138 ± 28 vs -82 ± 332 15 µA cm -2 ; p<0.01 Mann-Whitney test), with no significant effect on the cAMP-induced anion 333 secretion, or the amiloride-sensitive sodium absorption (Fig 3C). Complementary studies in HCO 3 -334 buffer showed that de novo synthesis of HCO 3 was not participating in the UTP-induced 335 electrogenic anion secretion, as incubation of tracheas with acetazolamide didn't affect the 336 magnitude of the UTP-induced current (-124 ± 14 µA cm -2 ; p>0.05 One-way ANOVA). 337 Using the SLC4A4 blocker S0859, we observed the inhibition of the cAMP-induced anion current ( 338 ΔIsc -12.2 ± 2.4) suggesting that SLC4A4 might participate in the cAMP-response (Fig S3A-C). 339 Nevertheless, when the TMEM16A/CFTR inhibitor, CaCCinhA01, was used to block the cAMP-340 induced current, further addition of S0859 was still able to induce a reduction in the current and of 341 similar magnitude as shown in Fig S3A (Fig S3D-F; -10.1 ± 2.8 µA cm -2 ), confirming that 342 electrogenic-bicarbonate secretion was not significantly stimulated by cAMP, indicating that basal 343 HCO 3 secretion occurs in mouse trachea. To confirm this last hypothesis we added S0589 to 344 tissues pre-incubated with amiloride and observed a reduction in the basal current only in HCO 3 -345 buffer (-13.9 ± 3.3 vs -3.0 ± 1.6 Δ µA cm -2 for HCO 3 vs HEPES buffer; p<0.02; Mann-Whitney; Fig  346   3D-F). The magnitude of basal HCO 3 secretion inhibited by S0859 was similar to the experiments 347 summarized in Fig S4C and 4F. Of note, the addition of S0859 to the tracheas induced a fast and transient negative change in I sc as observed in Fig 3D and E and Fig S3A and D, that has been also 349 observed in human cells (27), but for which the cause remains to be identified. 350 To further characterize the Ca 2+ -activated anion secretion, the UTP response was tested with no 351 involvement of cAMP-induced secretion. As can be observed (Fig 3G,H) the UTP-induced anion 352 secretion in tracheas maintained in HCO 3 buffer was significantly reduced by previous addition of 353 S0859 (-368 ± 25 vs -200 ± 17 µA cm -2 ; p<0.001 One-way ANOVA). The UTP response was also 354 reduced when HCO 3 was replaced with HEPES buffer (Fig 3I) (-199 ± 25 µA cm -2 ; p<0.001 One-way 355 ANOVA), but the addition of S0859 induced no significant reduction of the UTP-induced anion 356 secretion in tissues maintained in HEPES buffer (Fig 3J) (-142 ± 12 µA cm -2 ; p>0.05 One-way 357 ANOVA). 358

SLC4A4 participates in intracellular pH homeostasis in mouse airway epithelial cells. 359
We reasoned that the UTP-induced HCO 3 exit would lead to cytoplasmic acidification and 360 therefore we monitored intracellular pH of BCECF-loaded murine airway cells. As shown in Fig 3L,  361 UTP induced an intracellular acidification (ΔpH -0.25 ± 0.02) that was significantly reduced when 362 cells were placed in low Clbuffer (ΔpH -0.11 ± 0.02), indicating the existence of Cl -/HCO 3 -363 exchange. Fig 3M summarizes

changes in intracellular pH and includes experiments in HEPES 364
buffer, which shows that UTP was almost unable to induce intracellular acidification (ΔpHi -0.01 ± 365 0.01) in absence of HCO 3 -. To test if the UTP-induced intracellular acidification was dependent on 366 SLC4A4 activity, we tested the S0859 inhibitor and observed acidification of the intracellular 367 compartment and prevention of UTP-induced acidification (Fig S3G-H). Washout of S0859 partially 368 restored pHi and UTP-induced acidification. (Fig S3G-H; -0.05 ± 0.01 vs -0.13 ± 0.03 ΔpHi, for UTP 369 with S0559 and UTP post washout respectively; p>0.002; Mann-Whitney). These data suggest that both basal and UTP-induced HCO 3 secretion are dependent on SLC4A4 activity in mouse airway 371 epithelial cells. 372 The genetic inactivation of Slc4a4 induces a cystic fibrosis-like phenotype in mouse airways. 373 As explained in the methods section, we decided to work with wild type and Slc4a4 -/on the hybrid 374 background, at 16-20 days of age. First, we observed that Slc4a4 -/animals were affected by 375 defects in tracheal cartilage formation with presence of ventral gaps and abnormal patterns on the 376 rostrocaudal side (Fig 4A). In wild type animals, immunolocalization of SLC4A4 showed strong 377 localization in the airway epithelium but the signal was nearly absent in the airways from the 378 Slc4a4 -/mice ( Fig 4B). Using the same antibody, we showed that SLC4A4 was preferentially 379 expressed in CCSP-positive cells that correspond to Club cells and was excluded from cells positive 380 to acetylated-Tubulin, that identify ciliated cells (Fig 4C). This pattern of expression was 381 maintained in distal airway bronchi and bronchioles (Fig S4B). Further histological examination of 382 the Slc4a4 -/mouse airways demonstrated the presence of adherent mucus at the surface of the 383 tracheal epithelium (Fig 4D and S4C,D) as well as in the bronchi (Fig S4E-G) and bronchioli (Fig  384   S5H). Signs of damaged epithelium was also observed as interruptions in the epithelial layer facing 385 the lumen in the Slc4a4 -/airways (Fig 4D and S4F,H), that might explain the decreased R te of the 386 tracheas placed in Ussing chambers and that prevented electrophysiological measurements in the 387 Slc4a4 -/tracheas (Fig 4E). 388

Genetic silencing of Slc4a4 impairs intracellular pH homeostasis and mucociliary clearance in 389 mouse airways. 390
In order to validate the role of SLC4A4 in pH homeostasis of murine airways, UTP-induced 391 intracellular acidification was studied in airway cells isolated from the Slc4a4 -/mice. We observed 392 a decrease in the magnitude of intracellular acidification in Slc4a4 -/cells when compared to those 393 from wild type animals during UTP stimulation (-0.24 ± 0.01 vs -0.14 ± 0.01 ΔpH [i] , for wild type vs 394 Slc4a4 -/-) suggesting that an important amount of HCO 3 accumulates in airways cells via SLC4A4 395 (Fig 4F-G). We also noticed that after UTP wash-out, the acidification persisted in the wild type 396 cells but not in the Slc4a4 -/- (Fig S4J; -0.09 ± 0.02 vs -0.02 ± 0.01 ΔpH [i] , for wild type vs Slc4a4 -/-), 397 suggesting that HCO 3 secretion was sustained by SLC4A4 activity. Clearance of plastic beads, as a 398 way to measure MCC, was studied in freshly isolated mouse tracheas whose mucosal side was 399 exposed to air. As shown in the polar plots in Fig 4H the  We noticed that SLC4A4 localization in mouse was different to that in human airway epithelium, as 438 the mouse SLC4A4 was expressed in CCSP+ and not ciliated cells. Previously, CFTR and TMEM16A 439 channels were specifically located in non-ciliated cells of mouse airways (41), a distribution also 440 conserved in the rat (42). Pendrin, was also detected in non-ciliated secretory MUC5AC+ cells, 441 suggesting that a functional coupling for HCO 3 secretion occurs in secretory non-ciliated cells of 442 the mouse airways (43). Such a difference in expression of transporter proteins among human and 443 mouse airways has frequently been described; for example CFTR, is not the principal Cl -444 transporter in the mouse airways and consequently silencing or mutation of Cftr in mice did not 445 produce CF disease of the lungs (44, 45). Furthermore, the expression in human and not mouse 446 airways of another protein that influence ASL pH, the ATP12A H + /K + ATPase, magnifies ASL 447 acidification in human airways in CF disease (7). 448 Slc12a2, which encodes the bumetanide-sensitive NKCC1 co-transporter, which is essential for Cl -457 accumulation and secretion, did not. This suggests that lack of HCO 3 is pathologically more 458 relevant than Clsecretion in mouse airways (47). Indeed, a significant amount of experimental 459 evidence, using different models and species, is consistent with our findings in the Slc4a4 -/mouse. 460

Impaired bicarbonate secretion produces a CF
For example, the acidification of the ASL induced abnormal epithelial immune responses, 461 increased mucus viscosity and reduced MCC that could be reversed after HCO 3 supplementation 462 (6, 7, 48-51). Furthermore, HCO 3 secretion is necessary for proper mucus release from hAECs (27), 463 and maintenance of normal amounts of ENaC-mediated Na + absorption and ASL volume (8), 464 functions that become abnormal due to acidic ASL pH. Even though technically challenging issues 465 prevented us from measuring ASL pH in the mouse trachea, the reduction of HCO 3 transport 466 demonstrated here after SLC4A4 inhibition is consistent with the expected muco-obstructive 467 phenotype, including reduced MCC and mucus accumulation.