Abstract
Experimental or traumatic nerve injury causes the degeneration of associated taste buds. Unlike most sensory systems, the sectioned nerve and associated taste buds can then regenerate, restoring neural responses to tastants. It was previously unknown whether injury-induced immune factors mediate this process. The proinflammatory cytokines, interleukin (IL)-1α and IL-1β, and their requisite receptor are strongly expressed by anterior taste buds innervated by the chorda tympani nerve. We tested taste bud regeneration and functional recovery in mice lacking the IL-1 receptor. After axotomy, the chorda tympani nerve regenerated but was initially unresponsive to tastants in both WT and Il1r KO mice. In the absence of Il1r signaling, however, neural taste responses remained minimal even >8 weeks after injury in both male and female mice, whereas normal taste function recovered by 3 weeks in WT mice. Failed recovery was because of a 57.8% decrease in regenerated taste buds in Il1r KO compared with WT axotomized mice. Il1a gene expression was chronically dysregulated, and the subset of regenerated taste buds were reinnervated more slowly and never reached full volume as progenitor cell proliferation lagged in KO mice. Il1r signaling is thus required for complete taste bud regeneration and the recovery of normal taste transmission, likely by impairing taste progenitor cell proliferation. This is the first identification of a cytokine response that promotes taste recovery. The remarkable plasticity of the taste system makes it ideal for identifying injury-induced mechanisms mediating successful regeneration and recovery.
SIGNIFICANCE STATEMENT Taste plays a critical role in nutrition and quality of life. The adult taste system is highly plastic and able to regenerate following the disappearance of most taste buds after experimental nerve injury. Several growth factors needed for taste bud regeneration have been identified, but we demonstrate the first cytokine pathway required for the recovery of taste function. In the absence of IL-1 cytokine signaling, taste bud regeneration is incomplete, preventing the transmission of taste activity to the brain. These results open a new direction in revealing injury-specific mechanisms that could be harnessed to promote the recovery of taste perception after trauma or disease.
Introduction
Taste buds are replenished by new taste receptor cells throughout life and maintained by trophic support from innervating sensory nerve fibers (Beidler and Smallman, 1965; Guth, 1971; Cheal and Oakley, 1977; Guagliardo and Hill, 2007; Golden et al., 2020). The chorda tympani (CT) nerve, which transmits taste information from taste buds on the anterior tongue to the CNS, is vulnerable to injury sustained during dental and ear surgery and chronic inflammation of the middle ear (Berling et al., 2015; Snyder and Bartoshuk, 2016). Experimental CT sectioning causes the degeneration and subsequent regeneration of distal nerve fibers and associated taste buds. Recovery from nerve injury culminates in the transmission of normal taste-elicited responses and restored behavioral sensitivity to taste stimuli (Guth, 1971; Cheal and Oakley, 1977; Cheal et al., 1977; Hill and Phillips, 1994; St John et al., 1995; Cain et al., 1996; Shimatani et al., 2002).
Taste function typically returns after experimental nerve axotomy, but taste loss because of nerve injury, cancer treatments, or viral infection can persist, highlighting the need to identify mechanisms governing regeneration and recovery (Wang et al., 2009; Snyder and Bartoshuk, 2016; Cooper et al., 2020; Mainland et al., 2020; Risso et al., 2020). Some factors required for taste bud development and/or renewal also promote nerve and taste bud regeneration, including BDNF (Nosrat et al., 1997; Sun and Oakley, 2002; Lopez and Krimm, 2006; Ma et al., 2009), the Wnt/β-catenin signaling regulator and Lgr4/5/6 taste stem cell ligand, R-spondin (Lin et al., 2021), and sonic hedgehog (Shh) (Kumari et al., 2015; Ermilov et al., 2016; Castillo-Azofeifa et al., 2017; Kumari et al., 2017; Lu et al., 2018). Gaps in our understanding of taste regeneration remain as additional signaling pathways triggered by injury may be needed to restore neural sensitivity to tastants. Cytokines promote peripheral nerve regeneration and recovery in other injured adult tissues, but whether the same is true in the peripheral taste system is unknown (Dubovy et al., 2013; Bastien and Lacroix, 2014).
The proinflammatory master cytokine, interleukin (IL)-1, generally promotes regeneration in the PNS (Korompilias et al., 1999; Temporin et al., 2008; Nadeau et al., 2011; Dubovy et al., 2013) in contrast to the brain (Denes et al., 2011; Thome et al., 2019) depending on the injury model (Mason et al., 2001). The IL-1 receptor (IL-1R) binds the cytokines IL-1α and IL-1β to stimulate innate immune responses important for host defense (Dinarello, 2009; Garlanda et al., 2013a) and wound healing (Graves et al., 2001). The IL-1R accessory protein (IL-1RAcP) dimerizes with IL-1R and is needed for efficient signal transduction (Dinarello, 2009). The IL-1 family also includes negative regulators of IL-1, such as IL-1R2, a decoy receptor that binds ligand without initiating signaling, and the IL-1 receptor antagonist (IL-1RA), which in a recombinant form is used to treat autoimmune disease (Dinarello, 2009; Garlanda et al., 2013b). Together, these molecules tightly regulate IL-1R signaling to keep inflammatory responses in check. Although IL-1 is best known for stimulating inflammation, the pleiotropic cytokine also regulates the differentiation and/or proliferation of nonimmune cells, including keratinocytes (Werner and Smola, 2001) and glial cells (Mason et al., 2001; Vela et al., 2002).
IL-1 family members, including IL-1β and IL-1R, are expressed and functionally relevant in the normal and degenerating peripheral taste system as demonstrated in vivo and in polarized taste buds ex vivo (Shi et al., 2012; Kumarhia et al., 2016). Here we record CT nerve responses to tastants in Il1r KO and WT mice to test whether IL-1 signaling impacts taste bud regeneration and the recovery of neurophysiological taste responses. Although the CT nerve regenerated in the absence of Il1r signaling, taste buds did not fully regenerate and taste-elicited activity remained minimal. Dysregulated Il1 family gene expression and taste progenitor cell proliferation indicate that injury unmasks a requirement for cytokine signaling in the injured peripheral taste system.
Materials and Methods
Animals and CT nerve sectioning
Adult male and female C57BL/6J (WT #000664, The Jackson Laboratory) and B6.129S7-Il1r1tm1Imx/J (Il1r KO #003245, The Jackson Laboratory) mice were 8-10 weeks old at the time of nerve section. Sex did not affect the recovery of neural responses to taste stimuli in initial experiments consistent with previous studies in rodents after CT nerve injury (Cheal et al., 1977). Mice were housed in an SPF barrier facility on a 12:12 h light:dark cycle with lights on at 6:00 A.M. and provided free access to rodent chow (Envigo Teklad) and filtered tap water. Animal numbers are reported in the figure legends. Animal procedures followed National Institutes of Health guidelines and were approved by the Augusta University Institutional Animal Care and Use Committee. We confirmed genotype using The Jackson Laboratory protocols. Aseptic unilateral CT nerve sectioning was performed under intraperitoneal ketamine (50 mg/kg) and xylazine (10 mg/kg) anesthesia. Body temperature was maintained with a water-circulating heating pad. The CT nerve was approached ventrally in the neck and transected after it bifurcates from the lingual nerve with severed ends left in place as previously described in rats (McCluskey, 2004; He et al., 2012; Shi et al., 2012). Ketoprofen (5 mg/kg bw, s.c.) was administered prior, and the surgical site was sutured. Sham-sectioned mice received the same treatment, except that the CT nerve was visualized but not severed.
Neurophysiology
Mice were anesthetized with ketamine (80 mg/kg) plus xylazine (10 mg/kg) intraperitoneally and body temperature maintained at 36°C-39°C. Supplemental anesthetic was injected as needed. Mice were tracheotomized, and the hypoglossal nerves sectioned bilaterally to stop tongue movement. The CT nerve was approached laterally, dissected from surrounding tissue, and placed on a platinum electrode (Zhu et al., 2014). Neural activity was amplified (Grass Instruments), integrated with a time constant of 1.5 s, and the summated signal recorded with PowerLab hardware and software (AD Instruments); ∼3 ml of room temperature taste solution was applied to the anterior tongue with a syringe. Taste stimuli included series of NaCl (0.05, 0.10, 0.25 and 0.50 m), sodium acetate (NaAc; 0.05, 0.10, 0.25 and 0.50 m), monosodium glutamate (MSG; 0.1 and 0.3 m), 1.0 m sucrose, 0.01 m quinine (QHCl), and 0.01N HCl mixed in distilled water. CT nerve response magnitudes were calculated by subtracting the baseline from the height of the summated, integrated traces at 20 s after stimulus application. Relative responses were normalized to mean responses to 0.5 m NH4Cl applied at the beginning and end of the stimulus series and series analyzed if 0.5 m NH4Cl responses deviated by ≤10%. Relative response magnitudes were used to compare sham-sectioned Il1r KO and WT groups with CT-sectioned WT mice, which recovered neural responsivity. Since only low-magnitude responses were recorded from Il1r KO mice after nerve section, we compared absolute responses between groups. At the end of recordings, anesthetized mice were overdosed with isoflurane without awakening, followed by bilateral thoracotomy, which was also the method of death for all experiments below unless otherwise noted.
qRT-PCR
Tongues were harvested anterior to the intermolar eminence, bisected longitudinally, and stored in RNAlater (Fisher Scientific) at −80°C. In a subset of tongues, we peeled the lingual epithelium from the underlying mesenchyme using enzymatic and manual dissection and analyzed Il1 family gene expression separately in the two tissue compartments (Zhu et al., 2014; Meisel et al., 2020). RNA was isolated using the RNeasy Fibrous Tissue Mini kit (QIAGEN) and reverse-transcribed with the QuantiTect Reverse Transcription kit (QIAGEN). Primers with published sequences were purchased from Integrated DNA Technologies (Table 1) (Martin et al., 2013; Robson et al., 2016). qPCR was performed using the QuantiTect SYBR Green PCR kit (Applied Biosystems) and signal detected with the QuantStudio three Real-Time PCR Systems (Applied Biosystems). Expression levels of Il1 family transcripts were normalized to gapdh1 and the 2-ΔΔCT method (Livak and Schmittgen, 2001) used to analyze the change in expression following CT sectioning relative to WT sham-sectioned tongue.
Immunofluorescence
Mice were killed and tongues removed on the mandible and fixed for 5-7 h, then dissected free of the mandible, bisected longitudinally, and cryoprotected overnight in 30% sucrose in PBS (Kumari et al., 2018). Mice were killed between 14:00 and 17:00 h for Ki67 assays. Tongues were frozen in OCT (Sakura) and serially cryosectioned at 8 µm for immunostaining, except when noted below. Sagittal sections of anterior tongue or positive control spleen sections were incubated in the primary antibodies (Table 2) for 2 h at room temperature or overnight at 4°C. Keratin 8 (K8)+, DAPI+ Type I, II, and III taste cells were identified with the markers NTPdase2, PLCβ2, and carbonic anhydrase IV (Car4;), respectively (Kim et al., 2006; Chandrashekar et al., 2009; Vandenbeuch et al., 2013). Tissue sections for Ki67 proliferation assays were additionally antigen retrieved in heated 0.1 m Tris EDTA buffer (pH 6.0) for 2 min before incubating in primary antibodies. Nonspecific immunofluorescence was minimal in WT and Il1r KO sections incubated in primary antibody-omitted diluent or pre-immune serum. However, several additional commercially available IL-1R antibodies yielded similar staining in Il1r KO and WT tissues and were not used further. Tissue sections were rinsed, incubated in species-appropriate AlexaFluor secondary antibodies for 1 h at room temperature (Invitrogen, Fisher Scientific), nuclei stained with DAPI (#D1306; Invitrogen/Fisher Scientific), and slides coverslipped with Fluoromount-G (#0100-01; Southern Biotechnology).
We counted taste buds on thin sections but also stained K8 taste buds on lingual whole mounts to visually compare taste bud density following sham or CT sectioning (Venkatesan et al., 2016). To separate the lingual epithelium from the underlying lamina propria/mesenchyme, tongues were dissected from the mandible and injected with collagenase A (1 mg/ml; Roche Diagnostics) and dispase II (2.5 mg/ml; Roche Diagnostics) in PBS for 30 min before manual dissection and fixation in PFA for 1 h. Whole mounts were blocked in 10% NGS and incubated K8 and secondary antibodies (above) overnight at 4°C. The dorsal and ventral sides of the whole mounts were viewed and photographed on a Leica M165 FC stereoscope while suspended in dishes of distilled water, then flattened and coverslipped on glass slides. Once coverslipped, regions of whole mounts were also imaged on a confocal microscope at 10×.
We determined taste bud and innervation volumes using immunofluorescence on thick sections. Mice were anesthetized with ketamine and xylazine, then intracardially perfused with ice-cold PBS followed by 4% PFA and tissue postfixed, cryoprotected, and frozen as described above; 50 µm coronal cryosections were collected in PBS, fixed again in 4% PFA for 10 min at room temperature, blocked with 1.0% Triton X-100, 2.0% BSA, and 0.1% sodium azide overnight at 4°C, then incubated in K8 and P2X3 (Table 2) in PBS with 1.0% Triton X-100, 1.0% BSA, and 0.1% sodium azide for 5 d. P2X3 is present on CT axons innervating fungiform taste buds (Finger et al., 2005; Ishida et al., 2009; Meng et al., 2017). After rinsing in PBS, floating sections were incubated in AlexaFluor secondary antibodies (1:000; Invitrogen/Fisher Scientific) for 2 d, rinsed, mounted on slides, and coverslipped with FluomountG.
Image analyses
We analyzed IL-1 family cytokine expression on serial, 1 µm optical scans of fungiform taste buds captured at 60× with the zoom set at 3.5 on a Nikon A1R confocal/multiphoton microscope. Images were then converted from NIS Elements (version 4.30.01) to merged 3D projection format using Fiji open-source software (Schindelin et al., 2012; Schneider et al., 2012). Background fluorescence was subtracted using a small ROI outside of the tissue section. The integrated density of IL-1R, IL-1α, and IL-1β fluorescence was measured in a circular ROI with a 75 pixel radius placed within the central section of a K8+ taste bud or adjacent to taste buds in the nontaste lingual epithelium. One value for IL-1R fluorescence in a WT mouse was removed after identification by Grubb's test as an outlier (G = 4.682, a = 0.05).
K8+ taste buds on the right or CT-sectioned side of the tongue were counted in serial sagittal sections using a BX51 microscope equipped with epifluorescence (Olympus), a digital monochrome camera (Cool Snap, Roper Scientific), and MetaMorph software (MDS Analytical Technologies). Each immunopositive taste bud was followed through sections so that we only counted complete taste buds. Taste progenitor proliferation and differentiation were analyzed with 40× images of the center section of each taste bud and associated fungiform papilla on the regenerated side of the tongue. We counted Ki67+/DAPI+ cells in the perigemmal region adjacent to K8+ taste buds and in the apical or basal wall of fungiform papillae (Kumari et al., 2017). The number of DAPI+ taste receptor cell nuclei within the K8+ taste bud was also determined in these sections. Type II and III taste cells with a DAPI+ nucleus within the K8+ taste bud limits were counted in 40× images of the center section of each taste bud. We identified Type I taste cells with the marker, NTPdase2, which is difficult to attribute to individual K8+ cells for counting. Therefore, we thresholded NTPdase2+ pixels using MetaMorph software. We then calculated % NTPdase2+ pixels/total pixels in a taste bud to assess the proportion of Type I cells (Vandenbeuch et al., 2013; Miura et al., 2014; Gaillard et al., 2015).
We quantified K8+ and P2X3+ label volumes to analyze taste bud and innervation on the sham or CT-sectioned side of the tongue. Confocal images of fungiform taste buds on the anterior 2 mm of the tongue (Meng et al., 2015, 2017) were captured at 60× with the zoom set at 3.5. Serial optical sections of 1 µm were separately scanned for each fluorophore. We analyzed image stacks of background-subtracted complete taste buds, either on one tissue section or across two or three sections in NIS Elements software. The perimeter of K8+ taste buds was outlined and P2X3+ fibers thresholded as confirmed across optical sections and in the 3D volume view (NIS Elements version 4.30.01, Nikon). P2X3 has been demonstrated to label CT fibers in regenerated and uninjured taste buds in rodents (Finger et al., 2005; Ishida et al., 2009; Meng et al., 2017). Taste bud area and thresholded P2X3+ pixels within the K8+ limits of the taste bud were summed across taste bud sections and multiplied by the optical thickness of 1 µm to calculate taste bud and innervation volumes (Shuler et al., 2004; Meng et al., 2015).
We analyzed the innervation of taste bud-containing and empty papillae using 8 µm sections from additional sham or CT-sectioned Il1r KO and WT mice. Confocal images of the center section of papillae were captured at 60×, 1.5× zoom, converted to merged 3D projection images, and the integrated density of P2X3 staining was measured in a standard ROI using Fiji software as described above. Papillae sectioned at a tilted plane in which innervation could not be fully visualized were not analyzed. We measured innervation across spatial location by dividing the fungiform field anterior to the intermolar eminence into equal regions corresponding to anterior, mid, and posterior regions.
Experimental design and statistical analyses
To account for repeated measurements on the same mouse, linear mixed models were used to identify significant factors, including time post-section, strain, and surgical treatment (i.e., sham or CT section) on Il1 family gene expression, taste bud and innervation volumes, and Ki67+ proliferating cell numbers. Mixed models were fitted using R package lme4 (version 1.1-29). p values were calculated using t tests with the degrees of freedom approximated by the Satterthwaite's method.
Absolute neural response magnitudes from CT-sectioned groups were compared with sham-sectioned groups within and between mouse strains at each time post-surgery for the highest (or only) concentration tested for each stimulus. p values were adjusted using the False Discovery Rate (FDR). Additional one-way or two-way ANOVAs were performed with Prism 9.0 software using mouse strain (WT or Il1r KO) and surgical treatment (sham or CT sectioning) as main factors. Baseline IL-1 family fluorescence intensities in uninjured mice were analyzed using one-way ANOVA with Bonferroni post-tests comparing expression in taste buds or nontaste epithelium between strains. We analyzed group differences between relative neural response magnitudes with repeated-measures two-way ANOVAs, followed by Bonferroni post-tests when multiple concentrations of stimuli were tested or unmatched two-way ANOVAs and Bonferroni post-tests when a single stimulus concentration was applied. We analyzed taste cell numbers, the number of Type I, II, III taste receptor cells, and papillae innervation with separate two-way ANOVAs followed by Bonferroni post-tests. p values ≤ 0.05 were considered significant for all analyses.
Results
IL-1 is widely expressed in the uninjured peripheral taste system
We analyzed IL-1 family protein expression by immunofluorescence in taste buds and nontaste epithelium of uninjured WT and Il1r KO mice (Fig. 1). We confirmed that taste receptor cells identified by K8 expressed IL-1R in WT but not Il1r KO mice (Fig. 1A), while ligands IL-1α and IL-1β were expressed in K8+ taste buds in both strains (Fig. 1B,C). Semiquantitative analyses indicated significant differences in IL-1R expression between mouse strains [Fig. 1D; F(3, 18) = 22.70, p < 0.0001] in taste buds (p < 0.0001) and in nontaste epithelium (p = 005). IL-1α staining (Fig. 1E) also differed among groups [F(3, 16) = 5.70; p = 0.008] with higher expression in WT compared with Il1r KO taste buds (p = 0.010). IL-1β levels were similar across strain and region [Fig. 1F; F(3,18) = 2.032; p = 0.146]. IL-1R was also expressed by PGP9.5+ nerve fibers within fungiform papillae (Fig. 2A), lingual blood vessels (Fig. 2B), perigemmal taste progenitor cells (Fig. 2C), and a subset of macrophages (Fig. 2D). Together with previous results in rat (Shi et al., 2012), this indicates that taste buds, CT nerve fibers, and nearby cells are sources of IL-1 ligands as well as targets of IL-1 signaling (Kumarhia et al., 2016). Moreover, taste bud IL-1α protein expression is reduced in Il1r KO taste buds before CT sectioning.
Il1r KO prevents the recovery of taste function after axotomy
Sham-sectioned Il1r KO mice and WT exhibited similar neurophysiological taste responses, as did WT mice in which the CT regenerated (Fig. 3). There was a significant main effect of the stimulus concentration of NaCl [F(3,92) = 76.56, p = 0.0001] and NaAc [F(3,92) = 45.98, p < 0.0001] on relative CT responses as expected. However, neither strain nor treatment significantly affected CT responsivity. We recorded similar mean relative CT responses to NaCl [Fig. 3A; F(2,92) = 1.746; p = 0.180], NaAc [Fig. 3B; F(2,92) = 1.133; p = 0.327], monosodium glutamate [Fig. 3C; F(2,46) = 1.530; p = 0.227], sucrose [F(2,21) = 0.198; p = 0.822], quinine [QHCl; F(2,22) = 1.201; p = 0.320], and HCl [F(2,22) = 0.573; p = 0.572] from sham-sectioned WT and Il1r KO mice and from the regenerated CT nerve of WT mice. Since amiloride-sensitive sodium responses were affected by IL-1RA treatment in a prior study (Shi et al., 2012), we recorded responses to NaCl mixed in the epithelial sodium channel antagonist, amiloride. There was an predictable effect of NaCl concentration on relative CT responses in the presence of amiloride [F(3,88) = 35.11; p < 0.0001], but groups did not significantly differ [Fig. 3A; F(2,88) = 0.771; p = 0.467]. CT responses, an indirect measure of anterior taste bud function, were thus unaffected by Il1r KO. We conclude that IL-1 signaling does not affect taste bud development or peripheral taste function in the absence of injury.
The recovery of taste function after axotomy was strikingly different in Il1r KO compared with WT mice as shown by response traces in Figure 4A-C. Sham-like neural responses to all tastants were recorded in most WT mice at day 18-21 after axotomy. This timing is similar to the electrophysiological recovery period in the regenerated CT nerve following transection in gerbil (Cheal et al., 1977) and WT mice (Yasumatsu et al., 2007). At the same time, the CT was present but unresponsive in Il1r KO mice (Fig. 4A,B). Only nominal steady-state responses to the strongest stimuli were observed in some KO mice at day 60 post-sectioning (Fig. 4C). The results from male and female mice are collapsed and summarized in Figure 3D, since recovery time did not vary by sex. We note that some absolute response magnitudes are high in some sham-sectioned mice of either strain reflecting exceptionally strong neural activity in these preparations. We show median and mean values because of this technical variability. NH4Cl typically elicits robust CT responses, as demonstrated in sham-sectioned mice of both strains (Fig. 4D). The CT was unresponsive in 6 of 6 WT mice at day 5-8 after sectioning (p = 0.004 vs WT Sham). From day 11-17 post-injury, responses were variable in WT mice. We recorded normal response magnitudes to 0.5 m NH4Cl in 3 of 15 WT mice, small neural responses in 3 of 15 WT mice, while the CT of the remaining 9 of 15 mice was unresponsive (p = 0.007 vs WT sham). By day 18-22, responses from sham and nerve-sectioned WT mice were similar (Fig. 4D; p = 0.099).
At day 5-8 post-sectioning, the CT nerve was unresponsive to stimuli in 4/4 KO mice tested (Fig. 4D; p = 0.0003 vs KO Sham), similar to WT mice during this period. However, recovery in the two strains diverged in the next periods tested. The CT was unresponsive or barely responsive to 0.5 m NH4Cl in all 5/5 Il1r KO mice from day 11-17 post-sectioning, although magnitudes were not statistically different from those in WT mice (p = 0.085) (Fig. 4D; p = 0.0003 vs KO Sham). From day 18-22 post-sectioning, responses remained significantly lower in the Il1r KO group compared with axotomized WT mice (p = 0.02) and to sham-sectioned KO mice (p = 0.0003). Taste responses remained minimal at day 25-45 (p = 0.0006 vs KO Sham) and day 56-60 (p = 0.0006 vs KO Sham; p = 0.017 vs WT) after sectioning in the absence of Il1r.
We saw a similar pattern in responses to additional tastants. The regenerated CT in Il1r-deficient mice responded poorly to NaCl, NaAc, MSG, and QHCl compared with WT mice at day 22-60 (Fig. 4E-J; p = 0.0009-0.0032) and to the sham-sectioned KO group (p = 0.0003-0.011). KO responses to sucrose were reduced in sham versus sectioned KO mice (Fig. 4I; p = 0.01), while the decrease in HCl responses did not reach significance (Fig. 4J; Sham KO vs Cut KO; p = 0.057). CT nerve responses to the lower stimulus concentrations were at or near zero so are not depicted. The CT also responds to tactile and temperature stimuli through mechanosensitive endings that survive taste bud degeneration by Hedgehog pathway inhibition but we did not test those responses here (Finger et al., 2005; Donnelly et al., 2022).
We recorded from an additional 2 Il1r KO mice at day 100-101 post-sectioning. The mean absolute response to 0.5 m NH4Cl was 0.22 units in 1 KO mouse, well below the dashed line in Figure 3D. In the other KO mouse, the CT was absent, suggesting atrophy from the distal axotomy site. This was rare, observed in 1 other Il1r KO mouse at day 56-60 post-injury. In sum, the sectioned CT nerve regenerated in all but 2 Il1r KO mice but remained unresponsive to tastants >14 weeks after axotomy while taste function returned in most WT mice after 3 weeks. Since the CT nerve remained minimally responsive at each post-injury period, we conclude that Il1r KO prevented rather than delayed recovery.
Il1r KO alters IL-1 family gene expression dynamics during regeneration
We first confirmed that Il1r is expressed in both the lingual epithelium and underlying connective tissue/mesenchyme in WT but not KO tongues since recovery could be affected by receptor signaling in both tissue compartments (Fig. 5A). We also analyzed Il1 family gene expression in the right/denervated side of the tongue following sham or CT sectioning (Fig. 5B-F). As shown in Figure 5B, Il1r is significantly upregulated in WT by day 18-20 post-injury (p = 0.002) and remains chronically elevated at day 56-60 (p = 0.023) compared with sham-sectioned WT mice. In contrast, Il1rap expression (Fig. 5C) acutely decreased at day 7 after injury in both WT (p = 0.011) and KO mice (p = 0.047). Il1r2 expression was significantly reduced in KO versus WT shams (Fig. 5D; p = 0.005) but then upregulated at day 56 in both strains (p = 0.046 KO; p = 0.023 WT). Il1a expression was distinctly regulated by strain (Fig. 5E). Specifically, Il1a expression was elevated at baseline (p = 0.001) and day 7 post-sectioning (p = 0.028) in Il1r KO compared with WT mice but not significantly elevated by axotomy. In contrast, Il1a significantly increased at day 18 (p = 0.0004) and 56 (p = 0.039) in injured WT mice compared with shams. The increased Il1α mRNA in sham KO mice contrasts with reduced immunofluorescence intensity (Fig. 1E). This may reflect post-transcriptional or post-translational modifications or analyses limited to taste buds (Fig. 1E) versus the sectioned side of the tongue (Fig. 5E) (Mehra et al., 2003; Shebl et al., 2010; Israelsson et al., 2020; Young et al., 2020). Il1b was upregulated at day 56 compared with sham-sectioning in both WT (Fig. 5F; p = 0.039) and KO mice (p 0.0017), although expression of this ligand reached significantly higher levels in KO mice (p = 0.043). Il1r and Il1a gene expression thus increased in parallel with the successful recovery of taste function.
Fewer taste buds regenerate in Il1r KO mice
We asked whether peripheral taste function fails to recover because few taste buds regenerate in Il1r KO mice. As shown qualitatively in whole-mounted lingual epithelia (Fig. 6A), fewer K8+ taste buds are present on the anterior tongue in Il1r KO compared with WT mice at day 56-60 post-axotomy. We verified this by counting K8+ taste buds in sagittal cryosections (Fig. 6B). The number of taste buds was significantly affected by surgical treatment [F(2, 13) = 13.41; p = 0.007] and mouse strain [F(1,13) = 11.62; p = 0.0047]. The number of taste buds was similar between strains at day 5 post-injury, indicating that degeneration was not faster or more severe in the absence of Il1r (Fig. 6C). However, the number of regenerated taste buds was significantly reduced by 57.8% in Il1r KO compared with WT mice at day 56-60 post-injury (Fig. 6C; p = 0.0091).
Taste cell proliferation dynamics are altered by nerve sectioning in Il1r KO mice
Reduced progenitor cell proliferation could result in incompletely regenerated taste buds and functional deficits, particularly since this population expresses the IL-1R (Fig. 2C). We counted Ki67+ cells adjacent to taste buds (i.e., perigemmal) and in the apical and basal papillae wall since these regions repopulate taste buds (Fig. 7A) (Okubo et al., 2009; Liu et al., 2013; Kumari et al., 2017). We focused on three periods: day 14 post-sectioning before normal taste responses can be recorded, day 56-60 when taste function recovers in WT but not Il1r KO mice, and at day 34-35 midway through this period to test whether progenitor cell proliferation lags in KO mice. Results from linear mixed-model analyses revealed significantly fewer Ki67+ proliferating cells in the perigemmal region of Il1r KO compared with WT mice 34-35 d after axotomy (Fig. 7B; p = 0.0012). The same pattern occurred in the apical (Fig. 7C; p = 0.0010) and basal (Fig. 7D; p = 0.040) papillae wall at this period as the number of Ki67+ cells decreased in KO mice after nerve sectioning. This resulted in significantly fewer regenerated taste receptor cells in Il1r KO compared with WT mice at day 34-35 post-injury (Fig. 7E; p = 0.0011). However, by day 56-60 post-sectioning, the number of proliferating cells and taste receptor cells were similar among strains and surgical groups in all regions (p = 0.238-0.749). Ki67+ proliferating cells were located in the papillae walls but absent in the core of papillae without taste buds during the last post-sectioning period (Fig. 7A). Although empty papillae were uncommon in WT mice, these also lacked Ki67+ perigemmal cells. In sum, taste progenitor cell proliferation was delayed in Ilr KO mice but stabilized by 8 weeks post-injury in the subset of papillae with regenerated taste buds.
Regenerated Il1r KO taste buds are reinnervated and repopulated by all cell types
The loss of taste function in Il1r KO mice could be worsened by changes in the cellular composition or innervation of the subset of taste buds that did regenerate. We determined the proportion of Type I, II, and III cells in K8+ taste buds at day 56-60 post-sectioning (Fig. 8A-C). The density of NTPdase2 staining for Type I taste cells was significantly affected by strain [Fig. 8D; F(1,187) = 17.37; p < 0.0001] because of a decrease in KO compared with WT taste buds after sham sectioning (p < 0.0001). The percentage of Type II receptor cells, derived from the number of PLCβ2+/K8+ taste cells, was significantly affected by surgical treatment [Fig. 8E; F(1,159) = 4.30; p < 0.039], but group differences were not significant in post-tests (p = 0.656). The percentage of Car4+ Type III cells was significantly affected by surgical treatment [Fig. 8F; F(1,94) = 11.17; p = 0.001] and the interaction with mouse strain [F(1,94) = 4.69; p = 0.032] resulting in a significant increase in Type III cells in sham-sectioned Il1r KO taste buds (p = 0.030). Thus, in the absence of injury, Type III cells were more prominent at the expense of Type I cells in KO taste buds, although neural responses were unaffected (Fig. 3). Taste buds that regenerated in the absence of Il1r were fewer in number but comprised of normal proportions of Type I, II, and III cells.
We also determined the volume and innervation of regenerated taste buds by analyzing P2X3+ CT fibers within taste buds on 3D reconstructions from confocal stacks (Fig. 9) (Meng et al., 2017). Most taste buds were reinnervated regardless of strain (Fig. 9A-C). Specifically, 100% of WT and 96.3% of KO taste buds contained P2X3+ nerve fibers from day 18-60 post-injury, although as shown in Figure 9, innervation volume was slower to recover and taste buds remained smaller in KO mice. In WT mice, taste bud volume decreased at day 12-16 (p = 0.0006) and trended toward, but did not reach significance compared with sham-sectioned groups at day 17-25 after axotomy (Fig. 9D; p = 0.055). In contrast, Il1r KO taste buds were significantly smaller compared with KO shams at day 17-25 (p 0.000024), 26-40 (p = 0.040), and 56-60 (p = 0.024). Compared with WT taste buds, Il1r KO taste buds that regenerated were smaller at day 17-25 (p = 0.0006) and 56-60 (p = 0.004) post-sectioning. CT nerve volume shown in Figure 9E was significantly decreased at day 12-16 post-injury (p = 0.0003) in WT mice compared with shams but recovered by day 17-25. The nerve volume in KO taste buds was significantly reduced compared with shams at day 17-25 (p = 0.04) but reached sham KO levels by 8 weeks post-injury. In sum, the subset of taste buds that did regenerate in Il1r KO mice regained innervation more slowly and never reached full volume.
Papillae lacking taste buds are sparsely innervated and more frequent in Ilr KO mice
While regenerated taste buds were eventually reinnervated in Il1r KO mice that might not be true of papillae without taste buds. We determined the density of P2X3+ CT nerve fibers in the center section of fungiform papillae with and without taste buds at day 56-60 post-injury (Fig. 10). Papillae structure appeared grossly similar regardless of strain or surgical treatment (Fig. 10A,B), although occasional papillae were crowned with an atopic filiform spike (Oakley et al., 1990; St John et al., 1995). In sham-sectioned mice, papillae were more likely to contain a taste bud regardless of strain (Fig. 10C). However, at 56-60 d after axotomy, there were significant main effects of mouse strain [F(1,8) = 9.02; p =0.017], surgical treatment [F(1,8) = 13.67; p = 0.006], and their interaction [F(1,8) = 8.46; p = 0.020] on the percentage of papillae with taste buds which decreased in Il1r KO compared with WT mice (p = 0.006; Fig. 10C). Conversely, empty papillae were more numerous in Il1r KO mice (p = 0.006). Together, these results indicate that papillae structure was maintained after axotomy though fewer taste buds regenerated in the absence of IL-1R signaling.
In papillae with a taste bud (Fig. 10D), the total density of P2X3+ fibers was significantly affected by surgical treatment [F(1,8) = 52.27; p < 0.0001] and the interaction with mouse strain [F(1,8) = 10.57; p = 0.012] but not strain as a main factor [F(1,8) = 0.068; p = 0.801]. However, differences in papillae innervation did not reach significance in post-tests comparing sham (p = 0.076) or CT-sectioned (p = 0.135) KO and WT mice, consistent with similar innervation volumes in KO and WT taste buds at day 56-60 in reconstructed taste buds (Fig. 9E). Innervation was sparse in papillae without a taste bud across strain and surgical treatment (Fig. 10E). There was a significant main effect of surgical treatment [F(1,8) = 64.97; p < 0.0001] on the innervation density in empty papillae but no difference in a post hoc test (p > 0.9999). Thus, papillae without taste buds were more prevalent in Il1r KO mice after axotomy but poorly innervated regardless of strain.
We also analyzed regional differences in innervation density and taste bud number, since CT axons may partially regenerate to some papillae (Fig. 10F,G). In the anterior region, P2X3 density was significantly affected by surgical treatment [F(1,158) = 32.50; p < 0.0001] and the interaction with mouse strain [F(1,158) = 17.94; p < 0.0001] though not strain itself [F(1,158) = 1.27; p = 0.261]. Specifically, anterior fungiform papillae received less innervation in Il1r KO compared with WT mice at 8 weeks post-injury (p = 0.0004). There were significant main effects of surgical treatment on P2X3 density in the mid [F(1,40) = 7.817; p = 0.008] and rear [F(1,31) = 5.592; p = 0.025] regions but no differences in post-tests (p = 0.561 and p > 0.999, respectively). Regional differences in taste bud number paralleled changes in innervation (Fig. 10G). More taste buds were located in the anterior fungiform field of sham-sectioned and WT CT-sectioned mice as expected (Zhang et al., 2008). However, there were significant effects of surgical treatment [F(1,8) = 14.40; p = 0.0053] and the interaction [F(1,8) = 14.40; p = 0.0053] with strain, though not strain [F(1,8) = 3.6; p = 0.094], as significantly fewer taste buds regenerated in the anterior lingual field of Il1r KO mice (p = 0.0013). Fewer taste buds are located in the mid and posterior regions of the tongue, and there were no significant effects of strain or surgical treatment (p = 0.320 to >0.999). A subset of papillae across all regions of the fungiform field contained taste buds and were innervated in both mouse strains, however, indicating that the CT has the potential to regenerate to the tip of the tongue when taste buds are present.
In sum, the majority of taste buds failed to regenerate even 8 weeks after CT nerve axotomy in the absence of Il1r signaling. Taste progenitor cell proliferation lagged in Il1r KO compared with WT mice and taste buds were smaller with fewer taste receptor cells at 5 weeks post-injury. Yet, by 8 weeks post-injury, the subset of taste buds that regenerated in Il1r KO mice received normal amounts of innervation and were repopulated by similar numbers of Type I, II, and III cells. Despite the regeneration of 42% of taste buds, Il1r signaling is needed to support the restoration of CT responses to tastants. The failure of this subpopulation of taste buds to sustain neural responsivity is consistent with previous studies. Neurophysiological recovery required the regeneration of >50% of taste buds to regenerate after Shh pathway inhibition (Kumari et al., 2017). The recovery of the ability to behaviorally discriminate between tastants also depended on the number of regenerated taste buds in rat after CT axotomy (St John et al., 1995).
Discussion
Injury-specific factors in taste regeneration have generally been overlooked. We demonstrate for the first time that the recovery of taste function after injury requires endogenous cytokine signaling. Neural taste responsivity to tastants remains negligible even 8 weeks after axotomy in Il1r KO mice because of incomplete taste bud regeneration. Nearly 60% of taste buds fail to regenerate in Il1r KO mice despite the regrowth of CT fibers to the tongue tip. The proliferation of taste progenitor cells and taste bud regrowth and reinnervation were delayed in taste buds that regenerated in the absence of Il1r. Papillae lacking a taste bud were sparsely innervated by CT fibers in both strains. Yet the decrease in regenerated taste buds in Il1r KO mice on the anterior tongue led to chronic deficits in papillae innervation.
Distinct role of Il1r versus growth factors in taste bud regeneration
Some growth factors that mediate taste bud development and maintenance also mediate regeneration. BDNF is critical for the innervation of taste buds during development, adulthood, and after CT nerve sectioning. Inducible deletion of bdnf before nerve injury in adult mice prevents the regeneration of taste axons, in contrast to the loss of taste function we observed in Il1r KO mice because of fewer regenerated taste buds (Meng et al., 2017). R-spondin-2 maintains taste buds during homeostasis and can replace trophic signals supplied by afferent taste nerves after axotomy (Lin et al., 2021; Lu et al., 2022). Sonic hedgehog signaling also maintains taste buds during development and adulthood, and pharmacological inhibition of the pathway leads to taste bud but not neural degeneration (Castillo-Azofeifa et al., 2017; Kumari et al., 2017; Lu et al., 2018). In contrast, the requirement for Il1r signaling in regenerating taste buds is strictly injury-induced since taste buds develop normally and neural taste responses are unaffected by developmental Il1r KO.
Well-orchestrated cytokine responses regulate functional recovery in other injured peripheral nerves and sensory tissues (Shamash et al., 2002; Bastien and Lacroix, 2014; Denans et al., 2019). Exogenous IL-1β promotes myelin clearance and the recovery of sensory function after sciatic nerve injury (Perrin et al., 2005; Temporin et al., 2008), while behavioral recovery is delayed in the absence of Il1β and remains incomplete in double Il1β/tnf KO mice (Nadeau et al., 2011). Cytokines can also initiate detrimental effects, especially when chronically upregulated in the CNS, but are necessary to initiate wound healing in response to tissue injury (Gadani et al., 2015; Denans et al., 2019). Our results demonstrate the need for an Il1r-mediated response to injury-induced taste bud loss in addition to developmental factors.
Potential mechanisms for Il1r-mediated recovery and taste bud regeneration
Most Il1r KO taste buds fail to regenerate, although the CT nerve regenerates to the anterior tongue. We suggest that chronic dysregulation of Il1r and perhaps Il1a disrupts the proliferation of taste progenitor cells following nerve injury. The expression of Il1r and its ligand, Il1α, increases at day 18 and remains elevated in WT but not KO mice. In addition to classic effects on immune cells, IL-1R signaling stimulates the proliferation and/or differentiation of multiple cell types, including neuronal stem cells (García-Ovejero et al., 2013), oligodendrocytes (Vela et al., 2002), and hair follicle stem cells (Lee et al., 2017). IL-1 may play a pro-proliferative role in the injured taste system as indicated by the delayed increase in Ki67+ taste progenitor cells, which express IL-1R in KO mice.
Cytokines released by dying cells elicit inflammation before regeneration in other sensory systems. In the injured zebrafish retina, TNF-α induces Muller glia cell proliferation yielding neural precursor cells that differentiate into new photoreceptors (Nelson et al., 2013). Regenerated olfactory receptor neurons also require the activation of inflammatory pathways after injury. Tnf receptor signaling activates the RelA (p65) subunit of a downstream transcription factor, NF-κB, to stimulate inflammatory cell recruitment and the proliferation of horizontal basal cells which differentiate into olfactory neurons (Chen et al., 2017). Chronic inflammation, however, switches the local cellular response to an immune defense role preventing neuronal regeneration (Chen et al., 2019). TNF-α is also expressed by injured hair cells in the inner ear, where a similar requirement for inflammatory signaling has been suggested in species able to regenerate new hair cells (Denans et al., 2019). IL-1 binding to IL-1R activates NF-κB (Chen et al., 2017), and this pathway may similarly promote the proliferation of taste receptor cell progenitors after axotomy.
Nerve-taste bud induction in Il1r KO mice
Altered reinnervation may also cause incomplete KO taste bud regeneration. Papillae with fully innervated taste buds were interspersed with papillae lacking both a taste bud and significant innervation regardless of strain. This indicates that the CT nerve regenerates from the axotomy site to the tongue tip. Empty papillae, however, were scarce in WT mice and more frequent in Il1r KO mice after nerve sectioning. One possibility is that CT axons never extended from the lamina propria into papillae lacking taste buds. This could occur if local axon guidance signals are maintained in the tongue but not in papillae without taste buds. Alternatively, CT fibers might initially regenerate into the papilla but withdraw after chronic loss of a taste bud.
Molecular cues that guide the innervation of lingual papillae in early development have been identified (Vilbig et al., 2004; Treffy et al., 2016), but less is known about axonal pathfinding during regeneration. Despite the loss of taste buds after CT nerve sectioning the expression of BDNF remains stable because of two sources. Nontaste epithelium continues to express BDNF while taste cells in small, remnant buds increase BDNF expression (Meng et al., 2017). BDNF levels are also likely maintained by the lingual epithelium in Il1r KO mice following axotomy, although empty papillae did not contain remnant taste cells in the current study. Another possibility is that regenerated CT axons in Il1r KO mice are morphologically intact but biochemically defective. For example, disrupted axonal transport could prevent the delivery of R-spondin2 or other taste cell inducing factor from neuronal cell bodies to papillae (Lin et al., 2021). Intriguingly, IL-1R signaling in intestinal mesenchymal cells upregulates the Wnt agonist, R-spondin3, needed for epithelial cell regeneration after infection or injury (Cox et al., 2021). Further studies will determine whether IL-1R signaling modulates Wnt pathways in the injured taste system. Overall, IL-1R signaling could affect taste receptor cells and CT axons at multiple levels revealing a previously unappreciated role of cytokines in taste recovery.
Why do some taste buds regenerate in the absence of Il1r?
Proinflammatory cytokines, particularly TNF-α and IL-1, have pleiotropic and overlapping functions (Dinarello, 2007; Bastien and Lacroix, 2014). Both IL-1β and TNF-α are needed for full behavioral recovery from sciatic nerve sectioning, while single cytokine KO mice exhibit partial recovery (Nadeau et al., 2011). The regeneration of 42% of taste buds suggests that TNF-α or other proinflammatory cytokines may compensate for the loss of Il1r signaling after injury. TNF-α is produced mainly by Type II taste receptor cells within the taste bud (Feng et al., 2012), but infiltrating leukocytes responding to CT injury are also expected to produce this cytokine and could contribute to partial taste bud regeneration (McCluskey, 2004; Steen et al., 2010; Bastien and Lacroix, 2014). Increased TNF-α in response to systemic injection of the immunostimulant LPS decreases the proliferation of taste progenitor cells in foliate and circumvallate papillae. However, inflammatory stimuli and injury likely trigger different inflammatory pathways with distinct effects on progenitor cell proliferation and differentiation.
Regenerated taste buds, though fewer in the absence of Il1r, were morphologically similar in WT and KO mice. Taste cell number, composition of taste cell types, and innervation were all unaffected by strain at day 56-60 post-injury, although taste bud volume was reduced in KO mice. However, neural responses were absent and the number of taste buds was similar in both strains soon after injury, suggesting that degeneration was comparable. While some taste buds regenerate in the absence of IL-1R, they might be incapable of transducing taste stimuli because of faulty synaptic connections with CT fibers. Regardless of the functional status of the subset of regenerated Il1r KO taste buds, taste transmission from taste buds to the CT nerve is impaired.
In conclusion, the sense of taste is fundamental for nutrition and quality of life, as experienced by those with taste loss because of cancer treatment, nerve injury, and COVID-19 (Mainland et al., 2020). We demonstrate that the recovery of taste function after injury requires IL-1R signaling. Cytokine responses at the site of spinal cord and peripheral nerve transection have been dissected (Bastien and Lacroix, 2014), but their role in regenerating distant sensory receptor cells is understudied. The taste system is highly plastic throughout life, making it a tractable model to understand mechanisms underlying successful axonal and sensory organ regeneration. Exogenous treatment with IL-1 ligands is a potential therapy for taste loss. An important next step is to define the window for required Il1r signaling to minimize the potential for enhanced nociceptive responses because of neutrophil recruitment (Nadeau et al., 2011).
Footnotes
This work was supported by National Institute on Deafness and Other Communication Disorders Grants R01005811 and R010016668 to L.P.M.; and the Medical Scholars Program at the Medical College of Georgia to N.V. and E.C. We thank Emma Heisey for editorial comments.
Dedication: Dr. Daniel Linder, Division of Biostatistics and Data Science, Department of Population Health Sciences made significant contributions to statistical analyses used in these studies before his sudden death in May, 2022. We are deeply saddened by the loss of our colleague.
The authors declare no competing financial interests.
- Correspondence should be addressed to Lynnette Phillips McCluskey at lmccluskey{at}augusta.edu