Adcy3^+/- mice were obtained from the Storm Laboratory, University of Washington, Seattle, United States. Adcy3^+/+ and Adcy3^-/- mice were bred from heterozygotes and genotyped using polymerase chain reaction (PCR), as previously described (Wang and Storm, 2006). All mice were housed under a 12 h light/dark cycle and provided with access to food and water ad libitum in a specific pathogen-free animal room in Hebei University. All experimental procedures were performed under the Guiding Opinions on the Treatment of Experimental Animals issued by the Ministry of Science and Technology, People’s Republic of China and approved by the Animal Ethics and Caring Committee of Hebei University.

Main olfactory epitheliums from Adcy3^+/+ (n = 3) and Adcy3^-/- (n = 3) mice at P7, P30, and P90 were dissected. RNA of MOE was isolated using traditional TRIzol method (Ambion, cat. nos. 208054). cDNA was synthesized using PrimeScript^TM RT Reagent Kit with gDNA Eraser (TaKaRa, cat. nos. RR047A). Quantitative Adcy3 expression was determined through three-step quantitative reverse transcription polymerase chain reaction (qRT-PCR) using QuantiNava^TM SYBR Green PCR Kit (QIAGEN, cat. nos. 208054) (sequences of Adcy3 primers are as follows: Adcy3 F, AGATGTTCGGTGCCACCTG; Adcy3 R, ACTTCACCAGGGCTTCGTAAG). Adcy3 mRNA expression was determined using the 2^-ΔΔCT method (Livak and Schmittgen, 2001) and normalized to that of the housekeeping gene, β-actin [sequences of β-actin primers are as follows: β-actin F, CTAAGGCCAACCGTGAAAAG; β-actin R, ACCAGAGGCATACAGGGACA (Sciacovelli et al., 2016)].

Mice were anesthetized with intraperitoneal injection of pentobarbital sodium (100 mg/kg body weight). Animals were transcardially perfused with 0.9% physiological saline followed by 4% paraformaldehyde (PFA, pH 7.4). The nasal cavity was postfixed overnight in 4% PFA at 4°C and was subsequently transferred into 10, 20, and 30% sucrose in phosphate-buffered saline (PBS) until sinking (older mice were decalcified in 1 M Ethylenediaminetetraacetic acid (EDTA), pH 7.4, at 4°C). Postfixed specimens were then embedded under optimal cutting temperature. The olfactory tissue was coronally sectioned using a freezing microtome (Leica 1950). The sections were used for either hematoxylin–eosin (HE) staining or immunofluorescence staining procedures.

For SEM observations, Adcy3^-/- and Adcy3^+/+ male mice (n = 3 each) aged P7, P30, and P90 were transcardially perfused with 2.5% glutaraldehyde (in 0.1 M phosphate buffer, pH 7.4). After fixation, the heads were hemisected, the nasal cavity was exposed under a dissecting microscope, and olfactory mucosa was removed from the dorsal zone along the medial aspect (where robust location-dependent change in cilium length was observed) and further fixed overnight in 2.5% glutaraldehyde at 4°C. The samples were washed four times with 0.1 M phosphate buffer (pH 7.4), dehydrated in increasing ethanol concentrations (50, 70, 80, 90, and 100%), and dried using a vacuum pump. The dried samples were then coated with gold particles (EIKO IB-3) and examined using the scanning microscope (Hitachi S-3500N) at 20 KV.

For TEM observations, similar regions of the olfactory mucosa observed via SEM were examined using TEM in both Adcy3^-/- and Adcy3^+/+ male mice (n = 3 each). The olfactory mucosa from mice aged P30 and P90 was fixed for 4 h in 4% glutaraldehyde (in 0.1 M phosphate buffer, pH 7.4) at 4°C. The samples were washed and postfixed for 2 h in 1% osmium tetroxide (pH 7.4) at 4°C. Washed specimens were dehydrated in increasing ethanol concentrations (50, 70, 80, 90, and 100%). The specimens were then dried in propylene oxide and infiltrated overnight with an Epon resin mix/propylene oxide (1:1) mixture. This step was followed by infiltration with 100% Epon resin for 48 h. The specimens were embedded and cured in plastic following incubation in an oven at 60°C for 48 h. The sections (thickness = 1 μm) were cut using an ultramicrotome (Lecia UC-7) until the desired region was reached. Ultrathin sections (50 nm thick) were then prepared, stained with uranyl acetate and lead citrate, and examined using TEM (Hitachi H-7000).

Immunofluorescence staining of tissue sections was performed according to a previously published protocol (Wang et al., 2011). The sections were rinsed with PBS (pH 7.4) and incubated in a blocking solution (containing 10% serum, 5% BSA, and 0.2% Tritonx-100) for 1 h at 37°C. The sections were then incubated with the following primary antibodies at 4°C overnight: goat anti-OMP (1:500; Wako, cat. nos. 019-22291), rabbit anti-growth associated protein-43 (GAP43) (1:500; Millipore, cat. nos. AB5220), mouse anti-acetylated (Ac)-α-tubulin (1:1000; Sigma–Aldrich, cat. nos. T6199), rabbit anti-cleaved caspase3 (1:200; cell signaling, cat. nos. 9661S), mouse anti-bromodeoxyuridine (BrdU, 1:200; Sigma–Aldrich, cat. nos. B5002), mouse anti-Ki67 (1:300; BD, cat. nos. 556003), and mouse anti-Mash1 (1:150; BD, cat. nos. AF796). After washing thrice with Phosphate Buffer Solution +Tween-20 (PBST), sections were incubated with reciprocal Alexa Fluor dye-conjugated secondary antibodies (1:500, Invitrogen) for 1 h at room temperature. The signal in Ki67 and Mash1-staining was amplified by the deposition of fluorescein using the TSA Fluorescein System (1:50; PerkinElmer Life, cat. nos. NEL701A001KT). The sections were counterstained with 4′6-diamidino-2-phenylindole (DAPI) (2 μg/mL; Sigma, cat. nos. D9542) and coverslipped with PBS, which instead of being used as primary antibody, was used in negative controls.

To examine cell proliferations in MOE, male mice (n = 3 each group) aged P7 and P30 were intraperitoneally injected with BrdU (Sigma–Aldrich, cat. nos. B8434) (100 mg/kg body weight, dissolved in sterile saline) 2 h prior to perfusion. Male mice (n = 3 each group) aged P7, P30, and P90 were also injected thrice intraperitoneally (100 mg/kg body weight) (with intervals of 2 h) for three consecutive days. Mice were perfused 4 weeks later, and lifespan of neurons in the MOE was examined. PFA-fixed sections were incubated with 2 N HCl for 30 min at 37°C before immunostaining with anti-BrdU antibody. The subsequent immunostaining procedure was identical to that used for other antibodies.

HE-stained sections were examined and photographed using × 40 lens (NA, 0.95) of an Olympus BX53 under bright-field illumination. Immunofluorescence-stained sections were imaged using Olympus IX81 FLUOVIEW confocal microscope. Z-stack images (1024 × 1024 pixels) were captured using ×20 (NA, 0.75) and ×40 (NA, 0.95) objective lenses. In these image stacks, all labeled cells were counted, and we simultaneously determined whether these cells were double-labeled for OMP and caspase3. All images were only processed to facilitate contrast, brightness, and color-balance optimization using Adobe Photoshop CS software.

The thickness of MOE in HE sections, defined as the length from the basement membrane to the top of the knobs, was measured as documented in literature (Weiler and Farbman, 1997; Barrios et al., 2014). The same threshold was applied for all images obtained from the experiment. Laser intensity and detector gain were individually adjusted for each filter for maximum dynamic range, and settings were left unchanged for all the images collected in a section. In addition, no other attempts were made to normalize measurements background immunostaining. The background is a random factor, which is unrelated to the genotypes or treatments. Therefore, this measurement should lead to more conservative inferences. Positive cells under higher magnification images (40× objective) were manually counted using the “Cell Counter” tool of ImageJ plugins. For all quantifications, the experimenter was blinded to genotypes of the studied animals. The epithelial length along olfactory mucosa was measured using AxioVision software. Cell density was calculated by dividing the number of immunopositive cells by epithelial length (cells number/mm) for each age group. The number of immunopositive cells was finally converted based on the method published by Abercrombie (1946): corrected number = count × [section thickness/(section thickness + mean nucleus size)] (Abercrombie, 1946).

All data were expressed as mean ± standard error, unless otherwise indicated. Statistical analyses were performed with SPSS 21.0. Normality of distribution of variables and homogeneity of variances were checked through Shapiro–Wilk’s and Levene’s tests, respectively. We analyzed the data using one- or two-way ANOVA with post hoc pairwise comparisons. p < 0.05 was considered statistically significant.

By using qRT-PCR analysis, we confirmed that the Adcy3 expression levels in MOEs of P7, P30, and P90 Adcy3^-/- mice are significantly lower compared with those of Adcy3^+/+ mice with the same ages (Figure 1I). To evaluate whether the overall structure of the MOE is affected by Adcy3 deletion, we performed HE staining on specimens from P7, P30, and P90 Adcy3^+/+ and Adcy3^-/- mice. Our analyses were restricted to the mid-dorsal mucosa in each specimen to avoid morphological variations in the analyzed regions. The restriction was based on different regions of the MOE with different thicknesses. In general, the dorsal regions are thicker than the corresponding ventral regions, the medial regions are thicker than the lateral regions, and the middle and posterior regions are thicker than the anterior areas (Barrios et al., 2014).

In wild-type mice, results showed that the MOE at P7 was relatively thicker than those at P30 and P90, and the cells were loosely arranged (Figure 1A). Upon adulthood, the MOE became thinner, and neurons were more densely arranged (Figures 1B,C). These observations were consistent with those of previous histological studies, which showed the age-dependent decrease in epithelial thickness of murine MOE (Kondo et al., 2010). In Adcy3^-/- mice, the MOE cell density varied little from that of wild-type mice at P7 (Adcy3^-/-, 949.36 ± 24.19/mm; Adcy3^+/+, 951.07 ± 21.21/mm; p = 0.858; Figures 1A,D,G) and the MOE cell density at P30 was slightly reduced when compared with that of Adcy3^+/+ mice (Figures 1E,G). However, in P90 Adcy3^-/- mice, the cell density of MOE sharply decreased (Adcy3^-/-, 479.92 ± 31.67/mm; Adcy3^+/+, 780.11 ± 17.05 /mm; p = 0.000; Figures 1C,F,G), and the MOE was much thinner (Adcy3^-/-, 37.79 ± 1.34 μm; Adcy3^+/+, 65.16 ± 1.95 μm; p = 0.000; Figures 1C,F,H) when compared with wild-type specimens. Compared with the cell density of Adcy3^+/+ mice at P90, the cell density of MOE in Adcy3^-/- mice was 38.5% less at the same period of observation (Figure 1G). Together, these data suggest that Adcy3 deletion results in the gradual decline in cell number after birth and profound reduction in adult mice. The complete structure of the MOE is apparently affected by Adcy3 loss as animals age reach adulthood after the postnatal stages.

To further characterize OSN maturation in Adcy3^-/- mice, we examined the expressions of the immature OSN marker, GAP43, and the mature OSN marker, OMP, in the MOE. In Adcy3^+/+ mice, the number of GAP43⁺ cells decreased from age P7 (589.25 ± 13.60/mm) and P30 (222.55 ± 20.10/mm) until P90 (157.49 ± 26.91/mm) (Figures 2A,C,E,H). Concurrently, an increase in OMP⁺ cells was observed following MOE development from age P7 and P30 until P90 (Figures 2A,C,E,G). By contrast, the Adcy3^-/- MOE exhibited a higher number of GAP43⁺ cells and significantly reduced OMP⁺ cells at all postnatal ages examined (Figures 2B,D,F,G,H). GAP43 expression was apically increased (Adcy3^-/-, 319.23 ± 21.69/mm; Adcy3^+/+, 157.49 ± 26.91/mm; p = 0.000) (Figures 2E,F,H), mOSN layer was reduced and mainly restricted to the most apical layer of the MOE, below the supporting cells in the MOEs of P90 Adcy3^-/- mice (Adcy3^-/-, 134.38 ± 17.79/mm; Adcy3^+/+, 501.59 ± 28.43/mm; p = 0.000; Figures 2E,F,G). Remarkably, the number of OMP⁺ cells in P90 Adcy3^-/- mice was much less than that observed in P30 Adcy3^-/- mice (P30, 213.46 ± 35.52/mm; P90, 134.38 ± 17.79/mm; p = 0.000; Figures 2D,F,G), which is obviously not consistent with normal OSN development patterns exhibited in Adcy3^+/+ mice. Although the total OSN number (OMP⁺ and GAP43⁺) between Adcy3^+/+ and Adcy3^-/- mice varied little at P7 and P30, it significantly decreased in Adcy3^-/- mice at P90 (Figure 2I). Together, these data demonstrate that Adcy3 deletion reduces mOSN number and inhibits OSN development at the immature neuronal stage of mice during adulthood.

One hypothesis regarding the reduced cell numbers in the OSNs is that this phenomenon arises from reduction in cell lifespan, increase in apoptotic cell number, and/or decreased proliferation of progenitors. To test this possibility, we utilized long-term BrdU incorporation analysis to measure the OSN lifespan. Adcy3^-/- mice and wild-type littermates were injected with BrdU at ages P7, P30, and P90. Mice were subsequently sacrificed 4 weeks after injection. The number of BrdU⁺ cells in Adcy3^-/- mice significantly decreased relative to Adcy3^+/+ controls at each of the examined stage (Figures 3A–G). These data indicate that Adcy3 deletion shortens the lifespan of OSNs.

To assess whether reduction of OSNs in Adcy3^-/- mice resulted from alterations to cell proliferation mechanisms, we analyzed the expression of the neuronal progenitor marker, Mash1, a marker for globose basal cells (GBCs). Mash1 expression did not differ between Adcy3^-/- and Adcy3^+/+ mice at P7 and P30 (Figures 4A–E), indicating that OSN progenitors are possibly not affected by the loss of Adcy3 and develop normally into neurons at the early stages of neurogenesis. We identified actively proliferating cells using two markers, BrdU and Ki67, which are both used as labels for multiplying cells. Animals were injected with a single dose of BrdU and perfused after 2 h. The MOE sections were immunolabeled with antibodies against BrdU and Ki67. At P7, Adcy3^+/+ and Adcy3^-/- MOE showed a substantial number of BrdU-labeled proliferating cells (Figures 4F,G). BrdU⁺ cells were predominantly scattered near the basal neuroepithelium at both genotypes of P30 mice (Figures 4H,I). Furthermore, BrdU⁺ cell numbers also decreased with aging in Adcy3^+/+ and Adcy3^-/- mice (Figure 4J). However, significant difference was not observed between Adcy3^+/+ and Adcy3^-/- mice at P7 and P30 in terms of BrdU⁺ cell numbers (Figure 4J). Ki67⁺ cells were expressed in both the basal and apical layers of MOEs in the Adcy3^+/+ and Adcy3^-/- mice at P7 (Figures 4 K,L), and were mainly restricted to the basal layer at P30 (Figures 4M,N). Compared with P7 mice, the Ki67⁺ cells decreased at P30 in both Adcy3^+/+ and Adcy3^-/- MOEs (Figure 4O). Similarly, the number of Ki67⁺ proliferating cells varied minimally between Adcy3^-/- and Adcy3^+/+ mice at P7 and P30 (Figure 4O). Together, these data indicate that Adcy3 deletion does not suppress cell proliferation in MOE.

To determine whether the decrease in OSN in Adcy3^-/- MOE resulted from an overall increase in postnatal apoptosis, we tested for the expression of cleaved-caspase3, an enzyme critically involved in mammalian apoptotic pathway. Compared with P7 Adcy3^+/+ MOE, the number of caspase3⁺ cells decreased at P30 and P90 (Figure 5G). Only relatively few apoptotic cells were observed in the basal layer of MOE at P30 and P90 Adcy3^+/+ mice (Figures 5B,C,G). The number of caspase3⁺ cells differed minutely between Adcy3^-/- and Adcy3^+/+ mice at P7 (Figures 5A,D,G). However, we observed a significant increase in the number of caspase3⁺ cells in P30 and P90 Adcy3^-/- mice compared with their Adcy3^+/+ controls (Figures 5B,C,E,F,G). The number of apoptotic cells in P90 Adcy3^-/- mice was approximately 10-fold higher than that observed for Adcy3^+/+ littermates (Figure 5G). These results demonstrate that Adcy3 deletion leads to pronounced elevation in apoptotic OSNs during MOE development.

Although the total OSN is almost the same between Adcy3^+/+ and Adcy3^-/- mice at P7 and P30 (Figure 2I), the total OSN substantially decreased in Adcy3^-/- mice at P90 (Figure 2I), and this reduction is mainly caused by the decrease in mOSN (Figure 2G,H). We performed double-staining of OMP and caspase3 in P30 and P90 MOE to determine whether apoptosis caused the mOSN reduction in Adcy3^-/- mice. Some OMP and caspase3 double-labeled cells were observed in Adcy3^+/+ mice at both P30 and P90 (Figures 6A,C,E,F). Compared with Adcy3^+/+ mice, the number of double-labeled mOSN cells significantly increased in Adcy3^-/- mice at both P30 and P90 (Figures 6B,D,E,F). These data suggest that the impairment of OSN survival may account for the decreased mOSN in Adcy3^-/- mice.

Olfactory cilia are essential organelles that permit the conversion of external chemical stimuli into intracellular electrical responses in OSNs (Menco, 1997). Cilia stem from each dendritic knob of OSN and run horizontally in various directions to form a meshwork with cilia from other OSNs. This intertwined mat of cilia substantially increases the sensory surface of OSNs to facilitate odorant detection. In this study, aging culminated in a marked decrease of mOSN in Adcy3^-/- mice. To investigate whether cilia morphology was also altered by loss of mOSN after Adcy3 deletion, we performed an Ac-α-tubulin (a ubiquitous cilium marker) immunofluorescence analysis and SEM.

Ac-α-tubulin immunolabeling revealed a substantial decrease in immunofluorescence intensity of the ciliary layer in Adcy3^-/- mice at P30 and P90 (Figures 7A–F). Following SEM analysis of wild-type specimens at P7, P30, and P90, olfactory cilia were observed to be oriented parallel to the epithelial surface. The structures were also observed to run in various directions, forming a fine and dense meshwork with increasing age (Figures 7G–I). This effect occurred in parallel with an increase in mOSN number of postnatal MOE. Substantial and dense overlaps were observed in the cilia of P90 Adcy3^+/+ mice (Figure 7I). In Adcy3^-/- mice, cilia density was normal during early postnatal stages (P7, Figure 7J) but was significantly reduced at P30 and P90 (Figures 7K,L). Additionally, malformed dendritic knobs were observed in older specimens (especially P90). Numerous Adcy3^-/- cilia were short and stubby with enlarged knobs (Figures 7K,L, solid arrows). These results revealed that the loss of olfactory cilia occurs in conjunction with decreasing mOSNs in Adcy3^-/- mice.

Given that reduced cilium density and abnormal morphology of dendritic knobs were observed at P30 and P90 in Adcy3^-/- mice, we further examined cilium ultrastructure via TEM. Ciliary microtubule configuration (9 + 2 arrangement) and dendritic knobs filled with mitochondria and numerous microtubules were observed in Adcy3^+/+ mice at both P30 and P90 (Figures 8A,B,G; and 9A,B,C,H). These results were consistent with those previously observed (Jenkins et al., 2009). The membrane structures of both cilia and knobs were normal in Adcy3^+/+ mice. By contrast, the membranes displayed abnormal electron-dense outgrowths in P30 Adcy3^-/- mice (Figures 8C,D,F,H; black arrowheads). We frequently observed malformed knobs with dissolved and fractured mitochondria (Figures 8C,D, dashed arrows) and knobs lacking mitochondrial structures (Figures 8E,F). More severe ultrastructural deficiency was observed in P90 Adcy3^-/- mice. Many cilia and knobs had incomplete membranes (Figures 9D,F, black arrowheads, Figure 9I,J). Adcy3^-/- cilia were swollen (Figures 9D,F, red solid arrows), and some ciliary configurations (9 + 2) were disordered (Figure 9J). Medullary mitochondria were detected in the knobs (Figures 9D,E, red arrowheads), indicating severe mitochondrial ultrastructure damage. We also observed “bare knobs” with no cilia (Figure 9G). This phenomenon is rarely encountered in Adcy3^+/+ mice. These findings suggest that Adcy3, which is a key ciliary membrane protein, plays a role in the maintenance of cilium ultrastructure during OSN development.