Abstract

   A reduced sense of smell is a common condition in people with cystic
   fibrosis (CF) that negatively affects their quality of life. While
   often attributed to nasal mucosa inflammation, the underlying causes of
   the olfactory loss remain unknown. Here, we characterized gene
   expression in olfactory epithelium cells from patients with CF using
   single-nuclei RNA sequencing and found altered expression of olfactory
   receptors (ORs) and genes related to progenitor cell proliferation. We
   confirmed these findings in newborn, inflammation-free samples of a CF
   animal model and further identified ultrastructural alterations in the
   olfactory epithelium and bulbs of these animals. We established that
   CFTR, the anion channel whose dysfunction causes CF, is dispensable for
   odor-evoked signaling in sensory neurons, yet CF animals displayed
   defective odor-guided behaviors consistent with the morphological and
   molecular alterations. Our study highlights CF’s major role in
   modulating epithelial structure and OR expression, shedding light on
   the mechanisms contributing to olfactory loss in CF.
     __________________________________________________________________

   Cystic fibrosis alters olfactory structures and olfactory receptor
   expression, causing smell loss beyond chronic inflammation.

INTRODUCTION

   Smell deficits, such as hyposmia or anosmia, are common symptoms in
   patients with cystic fibrosis (CF) ([52]1–[53]7), although the
   underlying causes of these deficits are not yet understood.
   CF-associated olfactory deficits mainly affect odor thresholds but not
   odor identification ([54]1, [55]7), suggesting a dysfunction of the
   olfactory periphery. Smell loss in CF is usually interpreted as a
   consequence of chronic rhinosinusitis (CRS), a frequent condition in
   CF. However, several pieces of evidence suggest a more direct role of
   the pathology in olfactory loss, independent of chronic inflammation:
   (i) Patients with CF exhibit more persistent olfactory deficits
   compared to non-CF patients with CRS ([56]1, [57]8), even after
   undergoing sinonasal surgery ([58]3, [59]7) or in the absence of nasal
   obstruction ([60]1). (ii) Olfactory impairment has also been reported
   in children with CF when the long-term effects of sinus disease and
   chronic inflammation are not yet pronounced ([61]2, [62]5). (iii)
   Treatment with elexacaftor-tezacaftor-ivacaftor (ETI; also known as
   Trikafta or Kaftrio), which improves a number of respiratory CF
   symptoms including CRS, does not ameliorate olfactory performance in
   patients with CF ([63]9–[64]11). (iv) Cystic fibrosis transmembrane
   conductance regulator (CFTR), the anion channel whose dysfunction
   causes CF, is present in the olfactory epithelium as both mRNA and
   protein ([65]12–[66]15). Considering the strong link between olfactory
   loss and reduced quality of life ([67]16), gaining a deeper
   understanding of the causes of olfactory impairment in CF and
   developing therapeutic strategies can significantly enhance the quality
   of life for individuals with CF.

   Volatile odorant molecules are detected by olfactory receptors (ORs)
   expressed by olfactory sensory neurons (OSNs) located in the olfactory
   mucosa, which then send odor information to the olfactory bulbs (OBs)
   ([68]17). Each OSN expresses one of hundreds of distinct ORs, and OSNs
   expressing the same OR converge on different sets of OB glomeruli that
   encode odor information ([69]18). Canonical OSNs use a cyclic adenosine
   3′,5′-monophosphate (cAMP)–mediated cascade to transduce odor stimuli
   and a Ca^2+-activated chloride conductance for response amplification
   ([70]19, [71]20). Whether CFTR chloride conductance may play a role in
   olfactory signal transduction has not been fully determined. The
   olfactory epithelium also retains lifetime regenerative capacity
   mediated by stem and progenitor cells—horizontal and globose basal
   cells (HBCs and GBCs)—which reside in the basal layer of the epithelium
   and are able to replenish lost OSNs ([72]12, [73]21, [74]22).
   Conditions encountered in CF such as persistent inflammation,
   mechanical stress, or dehydration of airway surface mucus may induce
   deleterious effects on epithelial cell proliferation and regeneration.
   However, whether CFTR plays a more direct role in epithelial cell
   proliferation to alter tissue regeneration and olfactory function has
   not been yet investigated.

   Here, we characterized gene expression in olfactory epithelial cells
   collected from healthy and CF human subjects and found altered
   expression of genes encoding ORs, genes involved in cell cycle and
   tissue development in progenitor cells, as well as extensive CFTR
   expression in actively proliferating cells. We confirmed these results
   in newborn, inflammation-free samples of a CF animal model (CFTR^−/−
   pig) ([75]23) and further identified structural alterations in the
   olfactory epithelium and OBs in these animals. We identified a
   previously unrecognized role of CFTR in modulating epithelial
   proliferation and OR expression, thereby shedding light on the
   mechanisms through which CF causes olfactory loss independent of
   infection and inflammation.

RESULTS

Patients with CF show smell loss and altered gene expression in the olfactory
neuroepithelium

   We performed an initial olfactory assessment on 10 CF individuals (age
   range 18 to 52 years; five female), eight of which had at least one
   copy of F508del mutation, and seven were under ETI treatment, which
   improves a number of CF symptoms. We assessed symptoms of CRS using the
   SNOT-22 sino-nasal test and their sense of smell by the Sniffin’ Sticks
   test battery ([76]Fig. 1, A and B), a standardized smell test to
   categorize odor threshold and identification ([77]24, [78]25). When
   compared to healthy control individuals, patients with CF showed poorer
   outcomes in SNOT-22, subjective smelling ability, threshold, and
   threshold/identification index, but not in odor identification
   ([79]Fig. 1B), confirming hyposmia in these individuals. Statistical
   comparison between ETI-treated versus nontreated groups revealed no
   significant differences (P values: 0.09 to 0.98), which is consistent
   with previous reports showing that ETI treatment did not seem to
   improve olfactory dysfunction ([80]9, [81]11).

Fig. 1. CF individuals show smell loss and altered gene expression in
olfactory cells.

   [82]Fig. 1.
   [83]Open in a new tab

   (A) Schematic representation illustrating olfactory testing with the
   Sniffin’ Sticks test and collection of olfactory epithelium cells by
   nasal brushing from CF and healthy donors. (B) Scores of the SNOT-22
   test (left) and different parameters of the Sniffin’ Sticks test
   battery showing subjective smelling ability, threshold, identification
   scores, and the threshold/identification (TI) index (right).
   Blue-filled circles indicate ETI-treated individuals. *P < 0.05; ns,
   not significant (Student’s t test); n = 10 CF and 3 controls; n = 10 CF
   and 17 controls for SNOT-22 test. (C) UMAP dimensionality reduction
   plot of gene expression in 26,493 integrated CF and control cell nuclei
   (n = 7 CF and 9 control individuals). (D) UMAP depicting expression of
   CFTR (top) and violin plots of CFTR normalized expression in the
   different cell types (bottom). (E) Percentage of each cell type that
   express CFTR. (F) Volcano plot and heatmap of the 25 most significantly
   DE genes (up- and down-regulated) in the GBC cell cluster. CF samples
   correspond to patients not treated with ETI. Two of the control samples
   contained no GBCs and are not displayed in the analysis. (G) Average
   expression of the 16 most expressed ORs in the OSN cluster of the CF
   sample. pDCs, plasmacytoid dendritic cells.

   We next collected olfactory epithelium samples from the same patients
   via nasal brushing on the olfactory cleft under constant endoscopic
   control. Samples were snap-frozen, and later nuclei were extracted and
   analyzed by single-nuclei RNA sequencing (snRNA-seq) ([84]Fig. 1A). We
   generated an integrated single-cell sequencing dataset from a total of
   16 individuals (9 healthy controls and 7 CF individuals), permitting
   robust cluster annotation from >26,000 cells ([85]Fig. 1C). Cell types
   annotation was performed using known markers from ([86]12) and the
   Human Lung Cell Atlas (fig. S1) ([87]26). Uniform manifold
   approximation projection (UMAP) plots captured the distribution of the
   different olfactory, respiratory and immune cells ([88]Fig. 1C). We
   then identified all cells expressing CFTR ([89]Fig. 1D), a previously
   reported marker of olfactory microvillar cells ([90]12), which share
   many features with ionocytes from the airways ([91]27–[92]29). As
   expected, CFTR was expressed in 95.5% of olfactory microvillar/ionocyte
   cells, whereas OSNs showed little or no expression of CFTR ([93]Fig.
   1E). Unexpectedly, a substantial fraction of other olfactory and
   respiratory cell types also expressed CFTR, including sustentacular
   cells (SUS) (27%), olfactory HBCs (11.1%), GBCs (21.3%), and Bowman’s
   cells (23.5%) ([94]Fig. 1E). To identify CF-specific differences in
   gene expression, we performed a differential expression (DE) analysis
   between the CF and control datasets. We analyzed cells from CF subjects
   not treated with ETI to exclude any potential pharmacological
   interference with CFTR function. DE analysis identified marked
   transcriptional alterations between CF and control GBCs: 813
   up-regulated and 197 down-regulated genes ([95]Fig. 1F). Many of the
   up-regulated genes were related to the immune response, such as
   cytokines and interferon-induced proteins, which is consistent with
   chronic inflammation and recurring viral infections common in patients
   with CF ([96]Fig. 1F). Among the most down-regulated genes, we
   identified genes implicated in differentiation, replication, and
   proliferation of undifferentiated stem cells—such as ZNF365, ADRA1B,
   RADX, NMBR, NSG1, and GPER1— suggesting a potential role of CF in
   modulating olfactory cell renewal.

   We further performed an analysis of the expression of ORs in CF
   samples. OR expression was not restricted to OSN, and it was observed
   in 18 different olfactory, respiratory, and immune cell types (fig.
   S1). Within the OSN cluster, we analyzed the expression of the 16 most
   expressed OR genes in the CF sample ([97]Fig. 1G). Only 5 of these 16
   ORs (OR5A1, OR5A2, OR6C4, OR5AN1, and OR9K2) have been reported
   previously as among the most frequently expressed in healthy controls
   ([98]12), suggesting that CF individuals may express a distinctive set
   of ORs. Together, these findings show that CF individuals display
   olfactory loss and marked transcriptional alterations related to the
   immune response, stem cell homeostasis and OR expression, suggesting a
   potential more direct CF role in olfactory loss in addition to
   CRS-related hyposmia.

CFTR is expressed in proliferating cells of the human olfactory
neuroepithelium

   To study a potential role of CFTR in olfactory cell proliferation, we
   assessed the cell cycle pathway based on the expression of G[2]M and S
   cell cycle marker expression ([99]Fig. 2A). We found that the large
   majority of proliferative activity concentrates on GBCs ([100]Fig. 2, A
   and B). The analysis of CFTR expression on the GBC cluster revealed
   that CFTR is expressed in cells from G[1]-S and G[2]M phases ([101]Fig.
   2C), being significantly more expressed in G[1]-S phase (P < 0.005,
   Mann-Whitney test; [102]Fig. 2D). Differential analysis of CF versus
   control groups revealed a significantly higher G[2]M score in the CF
   GBC cell cluster (P < 0.001, Mann-Whitney test; [103]Fig. 2D),
   suggesting an active role of CFTR in GBC proliferation.

Fig. 2. CFTR is expressed in proliferating human olfactory cells.

   [104]Fig. 2.
   [105]Open in a new tab

   (A) UMAP projection of the computationally assigned cell cycle scores S
   (left) and G[2]M (right). (B) Phase scoring shows an elevated G[2]M
   score in the GBC cluster. (C) UMAP projection of the G[2]M, G[1], and S
   scores (left) and CFTR expression (right) in the GBC cell cluster. (D)
   CFTR is significantly more expressed in the G[1]-S phase in the GBC
   cluster, and G[2]M score is significantly higher in the CF group; **P <
   0.005 and ***P < 0.001 (Mann-Whitney test); n = 41 G[1]-S and 26 G[2]M
   that express CFTR; n = 214 CF and 101 control. (E) Olfactory cell
   culture after nasal brushing from healthy donors showing CFTR (green)
   RNA coexpression with DCX, PCNA, and Ki67 antibodies (magenta)
   visualized by dual in situ hybridization and immunostaining. Scale
   bars, 20 μM. (F) Effect of the CFTR inhibitor CFTR[inh]172 on PCNA and
   Ki67 immunolabeling showing an increase on the percentage of positive
   cells. Gaussian probability density functions show a shift toward an
   increased PCNA and Ki67 fluorescence intensity in CFTR[inh]172-treated
   cells. Vertical dashed line indicates the background threshold. n =
   1298 (mock), 890 (10 μM), 748 (20 μM), and 885 cells (30 μM); P < 0.001
   (Kolmogorov-Smirnov test) in all comparisons. a.u., arbitrary unit; NK,
   natural killer; DAPI, 4′,6-diamidino-2-phenylindole.

   To further explore CFTR role and expression, we cultured human
   olfactory cells collected by nasal brushing from 10 healthy control
   donors ([106]Fig. 2E). We characterized cell composition by
   immunoreactivity for 10 markers specific for the different olfactory
   cell types. Labeling of olfactory marker protein (OMP), G-protein
   subunit Gαolf, microtubule-associated protein doublecortin (DCX),
   proliferating cell nuclear antigen (PCNA), Nestin, and SRY-box 2 (SOX2)
   ([107]Fig. 2E and fig. S2) is consistent with the presence of OSNs,
   GBCs, and HBCs at different stages of differentiation ([108]12,
   [109]30). Quantitative reverse transcription polymerase chain reaction
   (RT-PCR) revealed consistent CFTR and neuroblast marker expression
   (PAX6, DCX, and NEUROD1) among donors (fig. S2). In situ hybridization
   revealed CFTR expression in 47% of all cells (fig. S2), which combined
   with immunolabeling, showed CFTR colocalization with the proliferation
   markers DCX and PCNA and the proliferation marker protein Ki-67 (Ki67)
   ([110]Fig. 2E and fig. S2). To better understand a potential functional
   CFTR role on proliferation, we treated the human olfactory cell
   cultures with CFTR[inh]172, a selective CFTR blocker. We observed an
   increase on both the percentage of PCNA^+ and Ki67^+ cells and the
   labeling intensity ([111]Fig. 2F), suggesting that cells were arrested
   at a premitotic stage after treatment. Together, these results indicate
   that CFTR is expressed in undifferentiated, actively proliferating
   cells and that CFTR activity may have an impact on olfactory cell
   proliferation.

A CFTR-null animal model shows deficits on olfactory-driven behavior

   We hypothesized that CF-associated olfactory loss may be consequence of
   a combination of indirect factors, such as CRS, and direct factors
   intrinsic to CFTR function. However, assessing olfactory function in CF
   out of the context of chronic infection and inflammation is
   challenging: patients with CF start to develop CRS during childhood
   becoming persistent during their whole lifetime, likely affecting the
   integrity of the olfactory tissue. In this context, CF animal models
   appear as a suitable alternative to study olfactory function. However,
   common model species such as rodents do not develop many of the human
   alterations in the case of CF, particularly in the respiratory tract
   ([112]31–[113]33). For these reasons, we decided to use a CFTR knockout
   pig model that better recapitulates the different traits of the human
   anatomy and CF pathology, including intestinal and lung pathologies
   similar to the alterations described in CF patients ([114]23). We
   performed all experiments with newborn piglets ([115]Fig. 3A), which
   are devoid of any inflammatory condition ([116]34–[117]36). To assess
   olfactory function in the newborns, we developed a robust behavioral
   assay by quantifying the ability of suckling in nonexperienced
   (newborn) piglets (movie S1). In most mammal species, including humans
   and pigs, the first suckling episode involves a strong maternally
   guided olfactory component ([118]37–[119]41). We thus quantified the
   latency to the initial suckling episode in 17 newborn CFTR homozygous
   mutants (CFTR^−/−) and 15 control (CFTR^+/+) littermates, which were
   macroscopically indistinguishable ([120]Fig. 3, A and E). No noticeable
   motor activity issues were observed, as initial mobility after birth
   was normal, with both genotypes showing similar times to stand (fig.
   S3). However, we observed that mutant piglets exhibited significant
   delays in initiating suckling, taking more time to locate the nipple
   and begin feeding ([121]Fig. 3, B to D), which suggests a deficiency
   consistent with reduced olfactory function.

Fig. 3. Newborn CFTR^−/− piglets display deficits in an olfactory-driven
behavior.

   [122]Fig. 3.
   [123]Open in a new tab

   (A) Litter containing newborn piglets of all three genotypes: CFTR^+/+,
   CFTR^+/−, and CFTR^−/−. (B) Latency-to-suckle in CFTR^−/− piglets
   (179.7 min ± 33.9 SEM; n = 17) and CFTR^+/+ (73.9 min ± 18.6 SEM; n =
   15) (**P < 0.01, Mann-Whitney U test). (C) Percentage of pigs suckling
   and their latency to locate the nipple over the 400-min assay period
   (***P < 0.001, Kolmogorov-Smirnov test). (D) Data from (C) binned in
   25-min windows [did not suckle (DNS)]. A total of 23.6% (4 of 17) of
   CFTR mutants fail to locate the nipple after 400 min. (E) Birth weights
   (P = 0.26, Mann-Whitney U test). (F) Immunolabeling for OMP in freshly
   dissociated pig olfactory epithelium cells (28.9% CFTR^+/+ and 26.9%
   CFTR^−/−; n = 1932 CFTR^+/+ cells and 1917 CFTR^−/− cells from three
   animals per genotype; P = 1, Mann-Whitney U test). Scale bar, 20 μM.
   (G) Example of Ca^2+ imaging showing an activated cell (left, arrow)
   and the corresponding time course (right) evoked by stimulation with
   1-octanol (10 μM). Scale bar, 20 μM. (H) Mean Ca^2+ peak amplitudes
   (ΔR/R[0]) in CFTR^+/+ (gray) versus CFTR^−/− (blue) cells after
   stimulation. (I) Proportion of activated OSNs (F[1,14] = 1.64, P =
   0.27, two-way ANOVA). (J) Peak amplitudes (F[1,558] = 0.13, P = 0.72,
   two-way ANOVA). n = 10 to 149 activated cells of a total of 18,781
   cells analyzed (10479 CFTR^+/+ and 8302 CFTR^−/− cells from five
   animals/genotype).

CFTR is not required for sensory function in OSNs

   Next, we investigated the basic physiological properties of OSNs, which
   detect small volatile odor molecules, in CF animals. We collected cells
   from the olfactory mucosa of CFTR^−/− animals by freshly dissociating
   the pig olfactory epithelium using an adapted protocol originally
   developed for mouse olfactory cells (see Materials and Methods). We
   confirmed the presence of mature OSNs (mOSNs) via OMP immunolabeling
   ([124]Fig. 3F). We then performed live-cell Ca^2+ imaging in 18,781
   olfactory cells ([125]Fig. 3, G to J) testing three odorant molecules:
   1-octanol, 2-heptanone, and octanal. Subsets of OSNs from CFTR^−/− and
   control animals responded reliably to each stimulus with similar
   response rates (0.59 to 0.85% versus 0.41 to 0.9%) and amplitudes
   ([126]Fig. 3, H to J), indicating intact sensory activity. Similarly,
   Ca^2+ responses to olfactory signaling pathways activators forskolin,
   3-isobutyl-1-methylxanthine (IBMX), and 8-bromoguanosine 3′,5′-cyclic
   monophosphate (8-Br-cGMP) also remained unaffected ([127]Fig. 3, H to
   J, and fig. S3). Together, these data establish that CFTR^−/− OSNs can
   still detect small organic molecules and that CFTR is not required for
   olfactory transduction in OSNs.

CFTR deletion induces transcriptional changes related to cell cycle and
development

   To explore whether the core transcriptional signature of the olfactory
   mucosa is affected by CFTR deletion, we performed bulk RNA sequencing
   (RNA-seq) and compared gene expression in the whole olfactory mucosa
   between CFTR^−/− and CFTR^+/+ pig samples. Genes (211) displayed a
   significant DE at a false discovery rate of 5%; 81 of these were
   up-regulated in CFTR^−/− samples, and 130 were down-regulated
   ([128]Fig. 4A and fig. S4). On the basis of published RNA-seq data in
   mice ([129]30), we determined that the largest fraction (49%) of
   down-regulated genes were not neuron specific ([130]Fig. 4, B and C).
   Gene Ontology analysis on the DE genes revealed a significant
   enrichment of terms related to organic compound processes, nucleic acid
   processes, DNA replication, cellular aromatic processes, and
   nucleobase-containing processes ([131]Fig. 4D). Many of these DE genes
   have been reported to interact and form a regulatory network likely to
   underlie alterations in cell cycle and development (fig. S4). Gene
   annotation enrichment analysis highlighted cell cycle/DNA signaling
   pathways, nucleotide metabolism, and development among the top hits of
   broad transcriptional changes ([132]Fig. 4E and fig. S4). We did not
   detect DE molecules involved in inflammation or immune response
   signaling pathways. Most cell cycle/development–related genes (41 of
   56) were down-regulated ([133]Fig. 4, F and G), suggesting altered cell
   proliferation. By contrast, the largest fraction (56%) of up-regulated
   genes was neuron specific ([134]Fig. 4C). The analysis of cell-specific
   marker genes revealed significant down-regulation of genes expressed by
   undifferentiated GBCs (PCNA, NES, KRT7, and PAX6), but not in genes
   specific for other cell types ([135]Fig. 4H). Marker genes for more
   differentiated GBCs such as ASCL1, NEUROG1, and NEUROD1 were not
   significantly DE ([136]Fig. 4H), suggesting that CFTR effect is
   specific to progenitor cells in an early stage of differentiation. We
   also observed altered expression of 16 OR genes, 13 being up-regulated
   and 3 down-regulated ([137]Fig. 4I). These results showing altered
   transcription of proliferation marker genes and ORs align with the
   expression changes observed in human CF samples.

Fig. 4. Transcriptome analysis of the CFTR^−/− piglet olfactory epithelium
shows alterations related to cell cycle and development.

   [138]Fig. 4.
   [139]Open in a new tab

   Differential gene expression of the whole olfactory mucosa
   transcriptome of CFTR^−/−compared to control CFTR^+/+ piglets. (A)
   Statistically significant DE genes are highlighted in blue. (B) DE
   genes classified as OSN- and non-OSN–specific according to previously
   published mouse RNA-seq datasets in Saraiva et al. ([140]30). A
   majority (48.5%, 63 genes) of the 130 downregulated (DR) genes are
   non-OSN. (C) Of the 81 up-regulated genes, 55.6% (45 genes) were
   specific to OSNs. (D) Pathway enrichment analysis of all genes that
   were significantly DE (P < 0.05) using Panther Classification System.
   (E) Gene Ontology analysis of all DE genes (P < 0.05) using UniProt
   database for classification. (F) DE genes related to cell cycle/DNA or
   development. A total of 41 of 58 (70.7%) were down-regulated. n = 4 per
   genotype. (G) Heatmap of cell cycle and development-related DE genes.
   (H) Heatmap of cell type–specific markers: GBCs, HBCs, mOSNs, iOSNs,
   microvillar cells (microvilli), sustentacular cells (sus) and Bowman’s
   gland cells (Bow). Significantly DE genes (in green) correspond to GBCs
   markers. (I) Heatmap and RNA-seq normalized counts from the 16 DE OR
   genes. DR, down-regulated; UR, up-regulated.

Olfactory structures exhibit morphological defects in CF pigs

   To better understand the impact of CFTR deletion and subsequent
   transcriptional changes, we performed a detailed histological
   characterization of the newborn nasal neuroepithelium in the
   CFTR-knockout pig. The macroscopic nasal anatomy appeared normal (fig.
   S5), yet examination by microscopy revealed a marked reduction in the
   olfactory epithelium thickness (79.7 versus 131.9 μm, P < 0.01)
   ([141]Fig. 5, A and B). CF preparations presented significant global
   cell loss (135.3 versus 236.4 cells/100 μm, P < 0.01) and reduced cell
   density (15.9 versus 19 cells/1000 μm^2, P < 0.05) ([142]Fig. 5B).
   Remaining OSNs displayed otherwise intact cell morphology and presence
   of Gαolf in the cilia (fig. S5). However, layer-specific
   quantifications determined major cell loss in mature OMP^+ neurons
   (41.3 versus 85.3 cells/100 μm, P < 0.01) ([143]Fig. 5C). We also
   observed significant cell reduction in nerve growth factor receptor
   positive (NGFR^+) cells (8.4 versus 16.8 cells/100 μm, P < 0.05),
   immature OMP^−/NGFR^− neurons (46.3 versus 76.6 cells/100 μm, P <
   0.05), and SUS cell layer (39.2 versus 57.7 cells/100 μm, P < 0.05)
   ([144]Fig. 5C).

Fig. 5. Morphological alterations in the CFTR^−/− piglet olfactory
epithelium.

   [145]Fig. 5.
   [146]Open in a new tab

   (A) Immunostaining for NGFR and OMP in olfactory epithelium sections.
   Epithelium thickness is delimited by the arrows. (B) Quantification of
   the olfactory epithelium (OE) thickness, cell number, and density. (C)
   Quantification of the cell number of each layer: OMP^+, NGFR^+,
   sustentacular cells (OMP^− cells above OMP^+ layer) and iOSNs (OMP^−
   NGFR^− cells between layers). (D) Representative images of OB glomeruli
   immunolabeled for OMP in CFTR^+/+ and CFTR^−/− piglets. (E)
   Quantification of the number and size of the OB glomeruli. *P < 0.05
   and **P < 0.01 (Mann-Whitney U test); n = 6 animals per genotype. Scale
   bars, 50 μm.

   We then assessed the morphology of 819 OB glomeruli (469 CFTR^+/+ and
   350 CFTR^−/−) in the brain, which receive axons from specific OSN
   subsets. CFTR mutants displayed significantly fewer (59 glomeruli/mm^2
   versus 108.9 glomeruli/mm^2, P < 0.05) and larger (7892 μm^2 versus
   4686.8 μm^2, P < 0.01) glomeruli ([147]Fig. 5, D and E), characteristic
   of a less developed olfactory system ([148]42). Thus, we also assessed
   the olfactory epithelium and bulb morphology in 75-day pig fetuses,
   over a total gestation period of 114 days. CFTR^−/− fetuses still
   displayed lower number of OMP^+ cells in the epithelium and larger,
   less abundant glomeruli (fig. S5), indicating that alterations observed
   at birth emerge earlier in development.

   We next investigated the subpopulations of actively proliferating basal
   cells in the newborn olfactory epithelium using KRT5, NGFR, or SOX2
   antibodies to discriminate between HBCs and GBCs ([149]Fig. 6, A to C).
   We observed fewer KRT5^−/NGFR^+ and KRT5^−/SOX2^+ cells (1.8 versus 5.5
   cells/100 μm, P < 0.01) ([150]Fig. 6C), consistent with a reduced
   number of GBCs. By contrast, KRT5^+/SOX2^+ cells remained unaltered
   (14.5 versus 14.2 cells/100 μm, P = 1) ([151]Fig. 6C), indicating that
   quiescent HBCs are less affected by CFTR deletion.

Fig. 6. CFTR deficiency alters the GBC proliferative phenotype.

   [152]Fig. 6.
   [153]Open in a new tab

   Double immunostainings for KRT5/NGFR (A) and KRT5/SOX2 (B). Arrowheads
   indicate NGFR^+/KRT5^− and basal SOX2^+/KRT5^− labeling specific for
   GBCs. (C) Quantification of SOX2^+/KRT5^− (GBCs) and SOX2^+/KRT5^+
   (HBCs) cells. **P < 0.01 (Mann-Whitney U test); ns, not significant. n
   = 7 CFTR^+/+ and 5 CFTR^−/− animals. (D) Dual fluorescent in situ
   hybridization with probes for CFTR and NGFR in control CFTR^+/+. CFTR
   labeling is not only robust in microvillar-like cells in the apical
   epithelium (white arrowheads, top) but is also detected in other cell
   types in the OSN cell layer (gray arrowhead, middle), basal cells
   (black arrowheads, bottom, colocalized with NGFR), and Bowman glands
   (right panels). (E) Fluorescent in situ hybridization for CFTR and
   immunostaining for SOX2 in the basal portion of the epithelium in
   CFTR^+/+ showing colabeling. (F) Dual fluorescent in situ hybridization
   with probes for CFTR and NGFR in CFTR^−/− showing no evident CFTR
   labeling. Bottom panels show magnified views of boxed area. Scale bars,
   20 μm.

CFTR is expressed by different cell types in the olfactory epithelium

   Our RNA-seq results in human samples revealed CFTR expression in
   different olfactory cell types ([154]Fig. 1E). We analyzed CFTR
   localization in epithelial tissue from CFTR^+/+ newborn piglets by in
   situ hybridization. CFTR probes showed robust labeling in scattered
   cells in the apical part of the epithelium, consistent with microvillar
   cell location and morphology ([155]Fig. 6D). Notably, we also observed
   abundant basal CFTR labeling in NGFR^+ ([156]Fig. 6D) and SOX2^+ cells
   ([157]Fig. 6E), consistent with expression in GBCs and HBCs. CFTR
   labeling was also present in the SUS and OSN cell layers, as well as in
   Bowman’s glands ([158]Fig. 6D). This broad expression pattern of CFTR
   in the olfactory mucosa of pigs is in line with our findings on human
   cells ([159]Fig. 1). CFTR labeling was absent in the epithelium of
   CFTR^−/− piglets, confirming the loss of CFTR mRNA expression
   ([160]Fig. 6F). Previous studies in mice have suggested a role of
   CFTR-expressing microvillar cells, which share many features with
   ionocytes from the airways ([161]27–[162]29), in modulating stem cell
   activity by the release of neuropeptide Y (NPY) ([163]43). However, in
   our samples, NPY was absent in microvillar cells and did not overlap
   with CFTR labeling (fig. S6), indicating distinct properties between
   species.

Olfactory mucosa snRNA-seq reveals GBC and OR alterations

   To further investigate the presence of CFTR transcripts in neurogenic
   progenitors, we analyzed the pig olfactory mucosa by snRNA-seq on a
   total of 8923 nuclei, comprising 4085 CFTR^+/+ and 4838 CFTR^−/−. Data
   were displayed on UMAP plots to analyze cellular heterogeneity after
   cluster identification based on known murine and human marker genes
   ([164]12) ([165]Fig. 7A and fig. S7). CFTR^−/− and control piglets
   contained a similar cell type repertoire, yet CFTR^−/− samples
   displayed higher proportions of non-olfactory cells ([166]Fig. 7B).
   When focusing exclusively on olfactory cell populations (a total of
   2863 CFTR^+/+ and 2697 CFTR^−/− nuclei), we observed a reduced
   proportion of iOSNs (−32.4%), GBCs (−27.5%), and HBCs (−11.8%)
   ([167]Fig. 7C). UMAP projection in control CFTR^+/+ cells confirmed
   CFTR expression across many cell types ([168]Fig. 7, D and E),
   consistent with our in situ hybridizations ([169]Fig. 6C) and human
   snRNA-seq data ([170]Fig. 1E). Similar to humans, the percentage of
   CFTR-expressing cells (95.6%) was highest in microvillar/ionocyte cells
   (fig. S7), but we also detected expression in Bowman’s cells (56.5%),
   SUS cells (17.6%), HBCs (15.3%), GBCs (7%), and some mature (4.8%) and
   immature (3.8%) OSNs (fig. S7).

Fig. 7. snRNA-seq analysis of CFTR^−/− piglet olfactory mucosa.

   [171]Fig. 7.
   [172]Open in a new tab

   (A) UMAP dimensionality reduction plot of gene expression in 8923
   integrated CFTR^+/+ and CFTR^−/− olfactory and respiratory mucosal cell
   nuclei (n = 73 to 1835 cells). (B) Cell proportions for every cell
   cluster. (C) Olfactory cell types, as a percentage of the total
   olfactory cells. (D) UMAP depicting expression of CFTR in the CFTR^+/+
   cell sample. (E) UMAP projection of the computationally assigned cell
   cycle (G[2]M, G[1], S; left) and CFTR expression (right) in the GBC
   cell cluster from the CFTR^+/+. (F) Ratio of G[2]M versus G[1] + S
   cells in the CFTR^+/+ GBC cell sample. G[2]M cells are 3.8 times more
   abundant among the CFTR-expressing GBCs . (G) DotPlot visualization of
   the expression of 11 GBC marker genes. (H) Percentage of GBCs that
   express the genes in (G) that are in G[2]M phase.

   We further assessed the G[2]M cell cycle pathway score as a proxy for
   proliferation based on the expression of G[2]-M and G[1]-S phase cell
   cycle markers, confirming that the bulk of proliferative activity
   occurs in GBCs (fig. S7). In control CFTR^+/+ samples, GBCs in G[2]M
   phase were 3.4 times more abundant among the CFTR-expressing versus
   CFTR-nonexpressing population ([173]Fig. 7F and fig. S7), indicating
   that most (77%) CFTR-expressing GBCs are actively proliferating. We
   observed that CFTR expression was most prominent in the less
   differentiated GBC cluster comprising cells in G[2]-M phase, whereas
   more differentiated GBCs in G[1]-S phase in transition to immature
   neurons, expressed little or no CFTR ([174]Fig. 7E). We also observed a
   consistent reduction in gene expression among specific GBC marker genes
   ([175]Fig. 7G), but not in other cell type–specific markers (fig. S7).
   Furthermore, GBC cell subsets expressing specific markers were
   consistently less frequent in G[2]-M phase (29.6% average reduction)
   ([176]Fig. 7H), indicating less proliferation in the CFTR^−/−
   epithelium.

   We assessed the expression of 200 ORs across mature and immature OSNs
   ([177]Fig. 8). Two genes annotated in the pig database as ORs,
   ENSSSCG00000033601 and ENSSSCG00000013696, were the most frequently
   expressed ORs (detected in 1190 distinct OSNs). However, the expression
   pattern of these two ORs did not match typical OR expression as they
   were coexpressed with other ORs and were not specific to neurons. We
   therefore consider unlikely that these two genes encode for real ORs.
   After exclusion of these two genes from the analysis, our data
   indicated a lower global OR expression in the CFTR^−/− sample
   ([178]Fig. 8B). Of the 198 identified ORs, 151 were detected in
   CFTR^+/+ cells, while 93 were detected in CFTR^−/− cells. We compared
   the composition of the two OR populations ([179]Fig. 8C and fig. S8).
   Some OR genes such as OR4E1 were largely underrepresented in the
   CFTR^−/− samples (nine CFTR^+/+ versus zero CFTR^−/− cells), and others
   such as OR51E2 were overrepresented (zero CFTR^+/+ versus four CFTR^−/−
   cells) ([180]Fig. 8A). Subsequent analysis identified three OR
   populations in mOSNs: 53% of the detected ORs were specific to
   CFTR^+/+, 23.7% to CFTR^−/−, and 23.2% common to both ([181]Fig. 8C).
   We then analyzed the expression of OR51E2 alongside another related OR,
   OR51E1, which showed no expression differences, using in situ
   hybridization in the pig olfactory epithelium. Consistent with our
   RNA-seq data, the OR51E2 probe labeled more cells in CFTR^−/−, while
   OR51E1 labeling remained similar across both genotypes ([182]Fig. 8D).
   We also detected differences in OR expression in immature OSNs
   ([183]Fig. 8, A and B). At this stage, it may be difficult to conclude
   that CFTR deficiency directly affected OR gene choice on early
   differentiation stages. Overall, our findings provide strong support
   for a model in which CFTR modulates the proliferative activity, and its
   deficiency induces structural changes in the olfactory epithelium and
   OB, as well as OR repertoire expression influencing odor responses.

Fig. 8. snRNA-seq analysis of CFTR^−/− piglet olfactory mucosa.

   [184]Fig. 8.
   [185]Open in a new tab

   (A) Specific expression of 33 ORs in individual iOSNs and mOSNs. (B)
   Relative OR expression in mature and immature OSNs. (C) Expression map
   identifying 198 ORs, specific to either CFTR^+/+ (red) or CFTR^−/−
   (blue) or coexpressed in both (gray). Each row represents an individual
   cell and each column is a single OR. Gene identities and cell numbers
   can be found in data S1. (D) Dual fluorescent in situ hybridization
   with probes for the OR genes OR51E2 [indicated in red in (A)] and
   OR51E1 in the olfactory epithelium of CFTR^+/+ and CFTR^−/− piglets.
   Right: Ratio of OR51E2/OR51E1 cells per section. *P < 0.05
   (Mann-Whitney U test); n = 6 per genotype. Scale bars, 50 μm.

DISCUSSION

   The molecular pathology of olfactory loss in CF is poorly understood.
   While airflow obstruction and tissue inflammation contribute to
   olfactory deficits in many cases, a reduced number of OSNs and OB
   glomeruli as well as differential OR expression may also play major
   roles. In the current study, our findings elucidate the capacity of
   CFTR to influence progenitor cell proliferation and OR gene expression.
   CFTR deficiency reduces neurogenesis in the olfactory epithelium and
   alters the expression of subsets of OR genes, leading to olfactory
   deficits.

   The contribution of CFTR to local proliferation and tissue development
   has been established in other peripheral tissues ([186]44–[187]48).
   Emerging evidence suggests that defects in CFTR function alter airways
   development and epithelia differentiation ([188]49). Congenital defects
   in the trachea and mainstem bronchi have been recognized in both
   neonatal CFTR null pigs and young CF children ([189]47). The olfactory
   epithelium maintains a stable population of mOSNs via continual
   neurogenesis ([190]21), normally sustained through differentiation of
   actively proliferating neurogenic progenitor cells ([191]22, [192]50).
   We found CFTR expression in actively proliferating cells, in both human
   and pig samples ([193]Figs. 1 and [194]2). We also observe in human and
   pig CF olfactory tissue a significant down-regulation of progenitor
   cell genes together with other markers involved in cell cycle and
   tissue development ([195]Figs. 1 and [196]4) as well as a reduction of
   epithelium thickness and abnormal OB glomeruli organization in CFTR^−/−
   pigs. Together, these observations are consistent with a model in which
   CFTR^−/− olfactory epithelium exhibits fewer proliferation, resulting
   in a thinner epithelium and fewer OB glomeruli. Our results highlight
   the importance of CFTR in epithelial development and may indicate
   potential strategies for therapeutic intervention to improve olfactory
   function in CF individuals.

   Our results may provide an explanation for some abnormalities affecting
   the olfactory system function in patients with CF ([197]1–[198]7).
   Altered olfactory epithelium structure, and hence olfactory function,
   can occur as a primary defect, independently of secondary consequences
   of the disease. By studying newborn animals, we excluded the
   possibility of secondary CF-related conditions such as chronic
   inflammation, bacterial infection, or persistent desiccation of the
   sensory surface layer. Our analysis of RNA-seq data showed no evidence
   of inflammation in newborn animals, indicating that the epithelium
   largely remains under uninjured, steady-state conditions. The reasons
   for poor olfactory function in both CF humans and pigs are likely
   related to the lower number of sensory neurons in the olfactory
   epithelium, which may result in fewer electrical signals in response to
   odorants. In addition, we also observed that human CF individuals
   express a different OR repertoire compared to controls [[199]Fig. 1 and
   ([200]12)], which may further alter olfactory performance. Similar
   results were observed in pig CFTR^−/− samples, indicating that altered
   OR expression is an intrinsic consequence of the pathology. Changes in
   OR expression are bidirectional, as a different set of ORs is
   up-regulated in the CFTR^−/−, providing support for improved olfactory
   perception of some odors. This would imply that patients with CF may
   show not only a reduced smell sensitivity but also differences in
   perceiving certain odors. Excessive representation of certain ORs in CF
   individuals may lead to a larger number of fibers from a specific OSN
   type reaching the OB, resulting in enlarged glomeruli, as seen in CF
   pigs. Whether these enlarged glomeruli arise from differences in
   specific OSN types or a more general delayed OB development remains
   unclear, but in both cases, they likely contribute to altered olfactory
   function.

   CF pigs better recapitulate human CF than the mouse, particularly in
   the upper airways ([201]31–[202]33, [203]36, [204]51), allowing the
   collection of newborn samples void of infection and inflammation. Using
   this model, we found that CFTR channel function is not necessary for
   odor transduction signaling in OSNs as they exhibit normal odor-induced
   calcium transients ([205]Fig. 3). Yet, CFTR^−/− piglets display a
   defective suckling behavior, which is compatible with reduced olfactory
   function. In most mammals, the initiation of the milk suckling behavior
   is critically dependent on olfactory signals ([206]37–[207]39). Pups
   from mice mutant for key olfactory transduction genes (Gnal, Adcy3,
   Cnga2, Scn9a) die at birth because of an inability to locate the
   mother’s nipple ([208]52–[209]55). In pigs, anesthesia delivery into
   the nose of newly born piglets or washing the mother’s nipples strongly
   impairs the ability to initiate sucking ([210]40, [211]41). Although
   other sensory modalities may also play a role, this evidence supports
   the involvement of olfaction in the control of initial sucking
   performance in pigs. We also validated previous observations showing
   high CFTR expression in olfactory microvillar cells ([212]12, [213]14,
   [214]43), a cell population that shares many traits with ionocytes
   ([215]27–[216]29). The function of CFTR in olfactory microvillar cells
   is not well understood, but, in the mouse, it has been suggested to
   play a role in maintaining tissue homeostasis by stimulating stem cell
   activity via release of NPY ([217]43). However, it is unlikely that
   they share the same function in pigs and humans because NPY is not
   expressed in microvillar cells nor coexpressed with CFTR in these
   species [fig. S6 and ([218]12)]. Moreover, NPY expression remains
   unchanged in the CFTR^−/−.

   Limitations of this study include the restriction of our biopsy samples
   to a limited number of human subjects resulting in low cell numbers,
   particularly OSNs, and the lack of access to intact (post-mortem)
   tissue for immunohistochemistry or RNA in situ hybridization. Despite
   the substantial similarities in RNA-seq data between human and pig CF
   samples, we cannot exclude the relevance of species-specific
   differences, especially given the much larger OR gene repertoire in
   pigs. In addition, our data from newborn animals do not address the
   role of CFTR in the adult olfactory system, and therefore, we are not
   able to exclude the possibility that CFTR contribution to the adult
   olfactory neurogenesis and OR expression differs to our observations in
   newborn individuals.

   In summary, our findings elucidate the capacity of CFTR to regulate
   progenitor cell proliferation and OR expression in the olfactory
   epithelium. The structural abnormalities in the sensory epithelium and
   OBs altered OR expression, and olfactory deficits reveal CF-related
   smell deficits as an intrinsic consequence of CFTR loss, independent
   from chronic inflammation of the upper airways.

MATERIALS AND METHODS

Experimental design and participants

   The objective of this study was to establish the causes of smell
   alterations observed in patients with CF. In a prospective study
   design, patients who presented themselves with CF were recruited from
   CF outpatient clinic consultations in the sense of regular
   disease-specific follow-up. Healthy participants were recruited for
   comparison. For the Sniffin’ Sticks test, a total of three healthy
   participants (aged 23 to 52 years, one woman) and 10 patients with CF
   (aged 18 to 52 years, five women) participated in the study from 4 to
   25 July 2023. For the SNOT-22 test, 17 healthy controls (23 to 52 years
   old, 10 women) and the previous 10 patients with CF were used. For
   snRNA-seq, samples from seven CF and nine controls were sequenced. The
   information on the genetic form of the disease comes from the
   documentation of earlier DNA tests. For healthy participants, exclusion
   criteria were neurological diseases, systemic diseases associated with
   smell disorders such as chronic renal failure, subjective smell
   impairment, CRS, allergic rhinitis, alcohol or drug abuse, and
   pregnancy. The study was conducted according to the Declaration of
   Helsinki and had been approved by the Ethics Committee at the TU
   Dresden (EK 552122022). All participants gave written informed consent.

Olfactory testing

   A detailed medical history was taken including age, gender, CF
   symptoms, medication, surgery, and questions regarding olfactory
   function. Further, the questionnaire sino-nasal outcome test (SNOT-22)
   was filled out to focus on lists of symptoms and social/emotional
   consequences of rhinosinusitis. Patients underwent olfactory tests
   using the Sniffin’ Sticks test battery (Burghart Messtechnik, Holm,
   Germany) to categorize the olfactory function regarding odor threshold
   and identification ([219]24, [220]25). A nasal endoscopy by an ear,
   nose, and throat specialist was performed to categorize endonasal
   diseases such as polyps or any form of CRS according to the Lildholdt
   score ([221]56). Nasal brushing was performed on the olfactory cleft on
   both sides under constant endoscopic control using a sterile cotton
   swab (CLASSIQSwabs, Brescia, Italy). The tips of the brushes were
   stored in CryoStor cell cryopreservation media (~1 to 2 ml) for 10 min
   at 4°C, then transferred to a cooling device (Thermo Fisher Scientific,
   Dreieich, Germany), and stored at −80°C until further processing.
   Patient records were assigned to codes and anonymized.

Human snRNA-seq

Sample preparation

   Cellular heterogeneity of human CF olfactory epithelium was analyzed by
   snRNA-seq. Frozen individual olfactory epithelium swabs were thawed and
   pooled for nuclei extraction in four pools containing three distinct
   samples from three individuals each (16 total individuals, 9 controls
   and 7 CF). This pooling ensured a sufficient amount of material for
   nuclei isolation. Genetic polymorphisms between samples were then used
   to discriminate individual samples (see below). Nuclei extraction was
   performed according to ([222]57), with some modifications. Briefly,
   each cryotube was thawed at 37°C supplemented with 400 μl of preheated
   saline medium. Brushes were cleared on ice by pipetting, transferred
   into 15-ml tubes, and centrifuged for 5 min at 290g at 4°C. After the
   removal of the supernatant, the cell pellet was resuspended into 1 ml
   of a buffer containing 25 mM citric acid and 0.25 M sucrose. The
   suspension was homogenized by 15 strokes with loose and then tight
   pestles, interrupted by 2× 3-min incubation on ice. After a filtration
   on a 20-μm filter (Miltenyi, Paris, France), nuclei were centrifuged
   for 5 min at 500g, resuspended in 500 μl of previous citric buffer, and
   centrifuged for 5 min at 500g. Nuclei were resuspended in a buffer
   containing 25 mM KCl, 3 mM MgCl[2], 5 mM tris-buffer (pH 8), RNAsin
   Plus ribonuclease inhibitor (0.4 U/μl), SUPERaseIn RNase inhibitor (0.4
   U/μl), and 1 mM dithiothreitol (DTT). After a control of morphology on
   a Floid cell imaging station (PicoQuant, Berlin, Germany), nuclei were
   counted with a Countess II FL Automated Cell Counter (Thermo Fisher
   Scientific, Waltham, USA). Nuclei were then diluted at a concentration
   of 1200 nuclei/μl to target a capture of ~10,000 nuclei, and the
   suspension was rapidly loaded on a 10X Genomics Next GEM Single Cell 3′
   v3.1 system (10X Genomics, Pleasanton, USA). Libraries were prepared
   according to the constructor specifications, with sequencing on a
   NextSeq 2000 (Illumina, San Diego, USA).

snRNA-seq processing and analysis

   Human raw data were analyzed using Cell Ranger Single Cell software
   v6.0.0, with inclusion of the intron sequences. The raw gene-barcode
   matrices were aligned to the hg38 genome. Demultiplexing was performed
   using Demuxafy within a Singularity container to accurately assign
   reads to individual samples ([223]58). After demultiplexing, all sample
   data were processed using Seurat package v.4.0 ([224]59) in R v.4.0.2.
   Cells with less than 200 genes or more than 5% mitochondrial content,
   and genes expressed in fewer than three cells were excluded from the
   analysis. Each individual Seurat object was normalized, and highly
   variable features were identified using the variance-stabilizing
   transformation (vst) method, selecting the top 2000 variable features.
   Integration features were selected from the variable features
   identified in each object. Each Seurat object was then scaled and
   subjected to principal components analysis (PCA) using the selected
   integration features. Integration anchors were identified using
   reciprocal PCA (rPCA), and the data were integrated on the basis of
   these anchors. The integrated data were further processed by setting
   the default assay to “integrated,” scaling the data, running PCA, and
   performing UMAP dimensionality reduction on the first 30 principal
   components (PCs). A neighbor graph was then constructed, and clustering
   was performed at a resolution of 1.0. Differentially expressed genes
   were identified for each cluster using the FindMarkers function.
   Pseudobulk analysis was conducted using the edgeR-LRT from Libra R
   package ([225]60) to aggregate single-cell data into bulk-like samples
   for DE analysis between CF and control. Codes used for all analyses can
   be found at [226]https://github.com/ymbouamboua/Human_Pig_Olfactory_CF.
   Numbers for all cell types are: B cells (n = 106), Bowman’s gland (n =
   84), club (n = 2600); deuterosomal cells (n = 405), GBCs (n = 315),
   Goblet cells (n = 1785), ionocytes/microvillar cells (n = 430),
   macrophages (n = 166), monocytes (n = 345), neutrophils (n = 32),
   natural killer cells (n = 308), olfactory HBCs (n = 4104), OSNs (n =
   31), plasmacytoid dendritic cells (n = 192), respiratory HBCs (n =
   183), respiratory multiciliated cells (n = 10298), sustentacular cells
   (n = 383), T cells (n = 1147), and Tuft cells (n = 18).

Pig snRNA-seq

Sample preparation

   Pig main olfactory epithelia were snap frozen in liquid nitrogen and
   included in OCT. Sections (300 μm) were made using a cryostat and kept
   at −80 °C. In a glass dounce tissue grinder previously cooled on ice,
   seven tissue slices and 1 ml of citric acid–based buffer (0.25 M
   sucrose and 25 mM citric acid) were added. Tissue samples were
   homogenized with three to five strokes of loose pestle and incubated on
   ice for 5 min. After five more strokes using a loose pestle, samples
   were incubated on ice for 5 min, homogenized again with three strokes
   using a loose pestle and five strokes using a tight pestle ([227]57).
   The suspension was filtered through a 40-μm cell strainer, and 1 ml of
   citric acid–based buffer was used to wash the containers. After a
   centrifugation for 5 min at 500g at 4°C, the supernatant was carefully
   removed and the sample was resuspended in 1 ml of wash buffer [10 mM
   tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl[2], 1% bovine serum albumin,
   0.1% Tween 20, 1 mM DTT, RNaseIn (0.6 U/μl), and SuperaseIn (0.2
   U/μl)]. Sample was filtered through a 5-μm cell strainer and
   centrifuged for 5 min at 500g at 4°C. Nuclei were resuspended in 50 to
   100 μl of diluted nuclei buffer [Nuclei Buffer 1X; Multiome kit, 10X
   Genomics), 1 mM DTT, RNaseIn (0.6 U/μl), and SuperaseIn (0.2 U/μl)].
   The nuclei morphology was verified using a Floid cell imaging station.
   Nuclei were counted on a Countess II FL Automated Cell Counter, diluted
   to the desired concentration (for a target capture of 10,000 nuclei).
   An RNA quality test was performed for each preparation before the 10x
   sequencing library was made. To do so, an aliquot of purified nuclei
   was placed in QIAzol Lysis Reagent, and RNA was isolated using miRNeasy
   Micro kit. The RNA integrity number was checked using the Agilent
   Bioanalyzer System. snRNA-seq was performed according to the protocol
   provided by 10X Genomics.

snRNA-seq processing and analysis

   Pig raw data were analyzed using Cell Ranger Single Cell software
   v6.0.0. The raw gene-barcode matrices were aligned to the Sscrofa11-1
   genome. Cell Ranger output was analyzed with the Seurat package v.4.0
   ([228]59) using R v.4.0.2. All individual data were filtered to keep
   cells with 200 to 7000 genes, less than 99% of dropouts, and less than
   5% of mitochondrial sequences. Doublet cells were removed with
   DoubletFinder ([229]61). After normalization and variance stabilization
   with SCTransfor and dimensionality reduction by PCA, data were
   visualized using UMAP embedding based on the first 20 PCs. Global
   integration was performed after normalization and identification of the
   4000 most variable features of each dataset. Features that varied
   across datasets were selected, and experiments were integrated with the
   matching mutual nearest neighbors method as described ([230]62). Cell
   clusters were annotated according to canonical gene markers: Bowman’s
   gland (SOX9, SLC6A11, and TMEM163), endothelial (MMRN1, CLVS1, and
   CCL21), fibroblast/stromal (DCN, LUM, and PTPRD), GBCs (EZH2, CXCR4,
   and HES6), immature neurons (GNG8, GAP43, and TPD52), immune (CD163,
   ARHGAP15, and MRC1), mature neurons (GNG13, STXBP5L, and PEX5L),
   olfactory ensheathing glia (SORCS1 and ZNF536), olfactory HBCs
   (CCDC129, CAPN13, and MGAM2), olfactory microvillar (STAP1, CLNK, and
   SLC35F3), pericytes (NOSTRIN, LRFN5, and CXCL2), respiratory ciliated
   (CFAP126 and FOXJ1), respiratory epithelial (NDAL, GPS, and EIF1),
   respiratory HBCs (KRT5, MET, and TP63), respiratory secretory (TTC6 and
   EIF2AK2), sustentacular (MOCOS, GLDN, and ERMN), and vascular smooth
   muscle (TAGLN, TPM2, and MYL9). Bowman’s gland cells (n = 460),
   endothelial cells (n = 167), fibroblasts/stromal cells (n = 1475), GBCs
   (n = 313), immature neurons (n = 1241), immune cells (n = 144), mature
   neurons (n = 1835), olfactory ensheathing glia (n = 360), olfactory
   HBCs (n = 108), olfactory microvillar/ionocyte cells (n = 128),
   pericytes (n = 129), respiratory multiciliated cells (n = 314),
   respiratory epithelial cells (n = 73), respiratory HBCs (n = 222),
   respiratory secretory cells (n = 546), sustentacular cells (n = 1115),
   and vascular smooth muscle cells (n = 293).

Bulk RNA-seq processing and analysis

   Olfactory epithelia from CFTR-null and control piglets were dissected
   and immediately frozen on liquid N[2]. RNA was extracted using the
   RNeasy Mini Kit (QIAGEN) with on-column deoxyribonuclease (DNase)
   digestion. mRNA was prepared for sequencing using the TruSeq RNA sample
   preparation kit (Illumina). All samples were multiplexed together and
   sequenced on four lanes on the Illumina HiSeq 2500 platform to generate
   100–base pair paired-end reads. Sequencing reads were mapped using STAR
   2.3 to the 11.1 Sus scrofa reference genome, annotation version 11.1 in
   the Ensembl pig genome database. The number of fragments aligned to
   each gene was counted using the HTSeq package, with the script
   htseq-count (mode intersection-nonempty). Any read that maps to
   multiple locations in the genome (also called multireads) was not
   counted toward the expression estimates as it cannot be assigned to any
   gene unambiguously. To compare the expression values across samples,
   raw count data were normalized to account for the depth of sequencing.
   Size factors were calculated using DESeq2’s function
   estimateSizeFactorsForMatrix, and raw counts were divided by the
   corresponding size factor for each sample. To test for DE, DESeq2 was
   used with standard parameters. Genes were considered to be
   differentially expressed if they had an adjusted P value of ≤0.05
   (equivalent to a false discovery rate of 5%). A total of 11 DE genes
   located on the Y chromosome were excluded from the analysis to prevent
   sex-specific bias. To find terms that are enriched in our list of DE
   genes the over/under-representation algorithm from GeneTrail
   ([231]http://genetrail.bioinf.uni-sb.de/) was used. The background
   provided were all those genes tested for DE. To assess whether the DE
   genes form putative regulatory networks, STRING
   ([232]http://string-db.org/) was used with default settings for the 212
   DE genes only. All normalized data and detailed results of the DE and
   enrichment analyses can be found in the data S1. Heatmaps of genes with
   significantly changed expression were generated using OriginPro 2021b
   program. Functional classification was analyzed by Gene Ontology
   Consortium (Panther Classification System) and UniProt database
   ([233]www.uniprot.org) using the biological process terms.
   Classification into OSN- or non-OSN–specific genes in [234]Fig. 4 (B
   and C) was performed by direct comparison with the dataset published by
   Saraiva et al. ([235]30) obtained by RNA-seq of mouse OMP^+ mOSNs.
   Cell-type assignments for specific marker genes were based on
   previously published mouse and human expression studies ([236]12,
   [237]30).

Human olfactory neuroepithelium cell culture

   Human olfactory cell samples were obtained independently from snRNA-seq
   samples using a slightly different technique. Cells from the olfactory
   mucosa were collected by nasal brushings from 10 healthy control,
   nonsmoker, severe acute respiratory syndrome coronavirus 2–negative
   subjects between 18 and 45 years old (both males and females) as
   previously described ([238]63). Every subject gave written informed
   consent for the study and the procedures involved. The study was
   approved by the local institutional ethics committee CEIm Ethical
   Committee Parc de Salut MAR, IMIM-Hospital del Mar Research Institute,
   Barcelona (study no. 2018/7942/I). Samples from the middle and upper
   turbinates were maintained in 250 μl of cold Dulbecco’s modified
   Eagle’s medium/Ham F-12 (DMEM/F12) enriched with 10% fetal bovine serum
   (FBS), 2% glutamine, and 1% streptomycin-penicillin (GibcoBRL), as
   previously described ([239]63). At 80% confluence, cells were expanded
   using 0.25% trypsin (GibcoBRL) and replated in 75-cm^2 flasks. Cells
   were then expanded until a maximum of five passages. Cells from two
   healthy individuals were treated with 0, 10, 20, or 30 μM CFTR[inh]172
   selective CFTR blocker (Tocris) on the culture media and incubated for
   24 hours. Cells were washed with phosphate-buffered saline (PBS) fixed
   with 4% paraformaldehyde (PFA) for 15 min, and in situ hybridization
   and/or immunolabeling was performed as described below.

Quantitative RT-PCR

   RNA from cultured human olfactory cells from 10 controls was purified
   with the PureLink RNA Mini Kit (Invitrogen) and stored at −20°C.
   Extracted total RNA was converted to cDNA by reverse transcription of
   20 ng of RNA using the Nucleospin RNA XS kit (Macherey-Nagel, Düren,
   Germany) according to the manufacturer’s protocol. The cDNA was applied
   in TakyonTM No ROX SYBR 2X MasterMix blue dTTP (Eurogentec), and the
   primers used are listed in data S1. All RT-qPCR reactions were
   performed in 96-microwell plates using The LightCycler 480 Real-Time
   PCR System (Roche Applied Science). Three technical replicates were
   averaged, and the quantitative RT-PCR data were analyzed using the
   2ΔΔCt method. GAPDH and RPL19 genes were used as a control for
   normalization. The results are expressed as relative fold change.

CFTR-deficient pigs

   Male and female CFTR^+/− heterozygous transgenic pigs ([240]23) were
   provided by the LMU Munich, Germany, transferred to France. Pigs (large
   white breed) were housed together and had access to a standard
   grain-based diet and water ad libitum. Heterozygous pigs were mated to
   generate CFTR^+/+, CFTR^+/−, and CFTR^−/− piglets. Newborn piglets (0.8
   to 1.5 kg) were genotyped within 6 hours after birth and euthanized for
   olfactory tissue collection. E75 pig fetuses were collected from
   amniotic sacs by cesarean section in pregnant dams. Pig littermates
   homozygous for CFTR (CFTR^+/+) served as controls. Littermates were
   randomly assigned to experimental groups keeping the same sex ratio
   between groups. All experiments were conducted in accordance with the
   guidelines of the Institutional Animal Care and Use Committee at the
   French National Research Institute for Agriculture, Food and
   Environment (INRAE). All experimental procedures were evaluated by the
   Ethics Committee of the Val de Loire (CEEA VdL, committee no.19) and
   approved by the Ministry of Higher Education and Research
   (APAFIS#1166-2015071615392426 Notification and
   APAFIS#10125-20170602162555 Notification).

Neonate suckling assay

   Newborn piglets were tested for suckling latency right after birth.
   Delivery was induced after 112 days of gestation by injection of 175 μg
   of cloprostenol and 20 units (IU) of oxytocin, and each piglet was
   marked right after birth and monitored, and latencies to stand up after
   birth and to suckle were scored (see movie S1). Testing lasted until
   the piglet reached the mother’s nipple and started to suckle or for a
   maximum of 400 min after the birth of each piglet. Each test group
   contained piglets from one single litter. Each litter was tested only
   once to ensure no learning occurred. Time scoring was performed by an
   experimenter blind to the genotype of the piglets, as biopsies for
   genotyping were collected after behavior testing.

Calcium imaging

   Ca^2+ imaging was performed in freshly dissociated pig olfactory
   epithelium cells adapting protocols previously developed for mouse
   olfactory cells ([241]64, [242]65). The olfactory neuroepithelium was
   detached from the cartilage and minced in PBS at 4°C. The tissue was
   incubated (20 min at 37°C) in PBS supplemented with papain (0.22 U/ml;
   Worthington), DNase I (10 U/ml; Fermentas), and urea (40 mM;
   Sigma-Aldrich) gently extruded in DMEM (Invitrogen) supplemented with
   10% FBS and centrifuged at 100g (5 min). Dissociated cells were plated
   on coverslips previously coated with concanavalin-A type IV (0.5 mg/ml;
   overnight at 4°C; Sigma-Aldrich). Cells were used immediately for
   imaging after loading with Fura-2/AM (5 μM; Invitrogen) for 60 min.
   Coverslips containing cells were placed in a laminar-flow chamber
   (Warner Instruments) and constantly perfused at 22°C with extracellular
   solution Hank’s balanced salt solution (Invitrogen) supplemented with
   10 mM Hepes. Cells were alternately illuminated at 340 and 380 nm, and
   the light emitted above 510 nm was recorded using a C10600-10B
   Hamamatsu camera installed on an Olympus IX71 microscope. Images were
   acquired at 0.25 Hz and analyzed using ImageJ [National Institutes of
   Health (NIH)], including background subtraction, region of interest
   (ROI) detection, and signal analyses. ROIs were selected manually and
   always included the whole-cell body. Peak signals were calculated from
   the temporal profiles of image ratio/fluorescent values. Results are
   based on recordings from five piglets for each condition and genotype
   (n = 10 to 149 activated cells of a total of 18,781 cells analyzed;
   10,479 CFTR^+/+ and 8302 CFTR^−/− cells). Cells were stimulated
   successively and in random order using bath application. The following
   criteria for stimulus-induced Ca^2+ responses were applied: (i) A
   response was defined as a stimulus-dependent deviation of fluorescence
   ratio that exceeded twice the SD of the mean of the baseline
   fluorescence noise. (ii) Cells showing a response to control buffer
   were excluded from analysis. (iii) A response had to occur within 1 min
   after stimulus application. In time-series experiments, ligand
   application was repeated to confirm the repeatability of a given Ca^2+
   response. Chemostimuli were freshly prepared each day and diluted in
   extracellular solution to give the following final concentrations: 10
   μM 1-octanol (Sigma-Aldrich); 10 μM 2-heptanone (Sigma-Aldrich), 10 μM
   octanal (Sigma-Aldrich), 50 μM forskolin (Sigma-Aldrich), 100 μM IBMX
   (Sigma-Aldrich), 500 μM 8-Br-cGMP (Sigma-Aldrich), and 90 mM KCl.
   Volatile odorants were initially prepared in dimethyl sulfoxide
   (Sigma-Aldrich) and further diluted in extracellular solution.

Immunostaining

   Human olfactory cells were fixed in PBS containing 4% PFA for 15 min at
   room temperature, washed 3X in PBS, and incubated in blocking solution
   (PBS solution containing 0.1% Triton X-100 and 3% horse serum) for 30
   min at room temperature. For newborn pigs, olfactory tissues were
   removed, postfixed overnight in 0.1 M phosphate buffer (PB) containing
   4% PFA, and later cryoprotected in 0.1 M PB buffer containing 30%
   sucrose. Samples were embedded in Tissue-Tek OCT compound, snap-frozen
   in cold isopentane, and processed on a Leica CM 3050S cryostat.
   Olfactory epithelia were cut in 16-μm-thick coronal sections and were
   directly mounted on SuperFrost Plus slides glasses (Thermo Fisher
   Scientific). Olfactory bulbs were cut in 30-μm serial free-floating
   sagittal sections in a PBS solution. Sections were treated with 10 mM
   sodium citrate for 5 min at 95° to 100°C for antigen retrieval, washed
   (3× 5 min) with PBS, incubated in blocking solution (PBS solution
   containing 0.1% Triton X-100 and 5% normal horse serum) for 2 hours at
   room temperature (RT), incubated overnight at 4°C in blocking solution
   supplemented with the primary antibody, washed in PBS solution, and
   incubated in blocking solution supplemented with secondary antibody for
   1 hour at RT. Nuclei were counterstained with
   4′,6-diamidino-2-phenylindole (DAPI; 0.5 μg/ml; Sigma-Aldrich) for 5
   min. We used the following primary antibodies: goat anti-SOX2 (1:300;
   R&D Systems, AF2018), rabbit anti-KRT5 (1:800; BioLegend, 905501), goat
   anti-OMP (1:2000; Wako, 019-22291), mouse anti-NGFR (1:1000;
   Sigma-Aldrich, N5408), mouse anti-PCNA (1:1000; Sigma-Aldrich, P8825),
   rabbit anti-DCX (1:2000; Abcam, AB18723), rabbit anti-Ki67 (1:100;
   Abcam, ab9021), mouse anti-Gαolf (1:500; Santa Cruz Biotechnology,
   sc-55545), rabbit anti-ITPR3 (1:500; Millipore, AB9076), rabbit
   anti-TRPM5 (1:400; Alomone Labs, ACC-045), and rat anti-NES (1:600;
   Santa Cruz Biotechnology, sc-33677). Secondary antibodies used for the
   corresponding target species were conjugated with Alexa Fluor 488,
   Alexa Fluor 546, and Alexa Fluor 647 (1:500; Thermo Fisher Scientific).
   Epithelium thickness, cell density, areas, and OB glomeruli
   quantifications were preformed from images acquired on a Zeiss LSM-780
   confocal laser-scanning microscope. Image regions were analyzed in the
   entire z axis with 3-μm step intervals, and images were reconstituted
   using the maximum intensity projection tool of Zen software. Images
   were analyzed with Fiji/ImageJ (NIH). Demarcations of epithelium limits
   and cell-specific layers were based on OMP and NGFR immunoreactivities,
   and DAPI^+ cells were counted using the particle analyzer plug-in of
   Fiji. Counts of the number of cells were evaluated blindly for each
   animal. Five slices per animal and genotype were used. Cells were
   counted from five to seven independent animals. For each sample, images
   were acquired at least from five different anatomical levels. Olfactory
   bulb glomeruli demarcation was based on OMP immunoreactivity in 5 to 17
   slices per animal from six independent animals per genotype. A total of
   819 glomeruli (469 CFTR^+/+ and 350 CFTR^−/−) covering all main OB
   topography (dorsal, ventral, lateral, and medial) were analyzed.

In situ hybridization

   Staining for CFTR, NGFR, OR51E2, OR51E1, and NPY mRNAs was performed
   using multiplex fluorescent in situ hybridization (ISH). Human
   olfactory cells and olfactory tissue sections were prepared as
   described above. RNAscope Fluorescent MultiplexV2 labeling kit (ACDBio
   323110) was used to perform the ISH assays according to the
   manufacturer’s recommendations. Probes used for staining are hs-CFTR
   (ACDBio 603291), ss-CFTR (ACDBio 541401-C2), ss-NGFR (ACDBio 828841),
   ss-LOC100625684 (ACDBio 1216061-C2), ss-LOC100737531 (ACDBio
   1216070-C3), and ss-NPY (ACDBio 318751). Negative control slides were
   performed in parallel. After incubation with fluorescent-labeled
   probes, slides were counterstained with DAPI and mounted with antifade
   fluorescent mounting medium (Dako). Fluorescent images were captured
   using sequential laser scanning confocal microscopy (Zeiss LSM-780).
   CFTR^+ cells were counted from four independent human subjects.

Combined ISH and immunohistochemistry

   After a standard ISH protocol, cells and tissue sections were directly
   incubated in blocking solution followed by primary and secondary
   antibodies for DCX, PCNA, Ki67, and SOX2 as described above.

Quantification and statistical analysis

   Statistical analyses were performed using the statistical package R
   version 3.6.0 and the packages ggplot2, drc and ggpubr (R-studio
   Software, 4.1.1), and OriginPro 2021b (OriginLab Corporation,
   Northampton, MA, USA). Statistical details of experiments can be found
   in the figure legends, including statistical test used, the exact value
   of n, as well as dispersion and precision measures. Assumptions of
   normality and homogeneity of variance were tested before conducting the
   following statistical approaches. Mann-Whitney U test was used to
   measure the significance of the differences between two distributions.
   Multiple groups were compared using a two-way analysis of variance
   (ANOVA). Kolmogorov-Smirnov test was used to compare cumulative
   distributions. The probability of error level (alpha) was chosen to be
   0.05. P values and the specific statistical test performed for each
   experiment are included in the appropriate figure legend or main text.
   Unless otherwise stated, results are presented as individual data
   points combined with boxplot indicating median lines and 25 to 75%
   ranges.

Acknowledgments