Cyclophosphamide-Induced Inflammation of Taste Buds and Cytoprotection by Amifostine

Abstract

Taste buds in the oral cavity have a complex immune system regulating normal functions and inflammatory reactions. Cyclophosphamide (CYP), a chemotherapy drug, has wide-ranging disruptive effects on the taste system including loss of taste function, taste sensory cells, and capacity for taste cell renewal. In bladder epithelium, CYP also induces inflammation. To determine if CYP induces inflammation in taste buds, we used immunohistochemistry to examine tumor necrosis factor alpha (TNF-α) (a proinflammatory cytokine) expression over a 72-hour period. Expression of TNF-α increased in a subset of PLCβ2 labeled (Type II) cells, but not SNAP-25 labeled (Type III) cells, between 8 and 24 h postinjection and declined slowly thereafter. This inflammatory response may play an important role in the disruptive effects of CYP on the taste system. Further, pretreatment with amifostine, a sulfhydryl drug known to protect normal tissues during chemo- or radiation therapy, reduced the amount of CYP-induced TNF-α expression in taste buds, suggesting this drug is capable of protecting normal cells of the taste system from adverse effects of CYP. Amifostine, used as a pretreatment to CYP and possibly other chemotherapy drugs, may offer clinical support for preventing negative side effects of chemotherapy on the taste system.

Introduction

The gustatory system is important for distinguishing nutritious from poisonous substances and preparing the gastrointestinal system for digesting food. Disturbances in the taste system by infection, drug-induced injuries, radiation, or other forms of insult can be deleterious for the overall health and well-being of humans. Taste dysfunctions are frequent side-effects reported by patients undergoing chemotherapy (Comeau et al. 2001; Hong et al. 2009) because these drugs do not differentiate between cancer cells and healthy cells such as taste sensory cells (TSCs) (Crook et al. 1986; Anderson et al. 1995; de Jonge et al. 2005; Aaldriks et al. 2013). Numerous studies have shown that chemotherapy drugs can distort taste sensations and alter detection thresholds for basic tastes (Berteretche et al. 2004; Bernhardson et al. 2007, 2008; Steinbach et al. 2009; Nolden et al. 2019; Drareni et al. 2021).

TSCs are located within taste buds (TBs) in the oral cavity where they are exposed to a wide variety of external and internal substances, some of which can harm TSCs. There are 50–100 TSCs within a TB that can be categorized into 4 general cell types: Types I, II, III, and IV (Farbman 1965; Finger 2005). Type I cells have protruding structures that intertwine around other cell types and appear to act as glial-like cells. Type II cells are elongated cells with G-protein-coupled receptors and downstream phospholipase C (PLC) signaling system for sweet, bitter, or umami taste transduction. Type III cells, also elongated, are characterized by synaptic contacts and the expression of synaptic membrane proteins such as SNARE complexes. Type IV cells are immature cells located adjacent to TBs and enter the TB before differentiation and maturation (Roper and Chaudhari 2017). TSCs have short life spans ranging between 6 and 24 days, depending upon the cell type (Perea-Martinez et al. 2013). To replace TSCs lost due to aging or injury, progenitor cells ventral to TBs are continuously engaged in cell renewal processes. The high turnover rates of TSCs and their progenitor cells make them susceptible to chemotherapy drugs.

There are several effects of chemotherapy drugs that can cause disturbances in taste functions. For example, chemotherapy or radiotherapy can reduce salivary output by damaging von-Ebner glands (xerostomia), thereby reducing nutrients available to activate taste receptors in apical ends of TSCs (Mese and Matsuo 2007). Cyclophosphamide (CYP), an alkylating agent, is a well-established chemotherapy drug with cytotoxic properties that damage DNA and induce oxidative stress in cancer and normal cells (Crook et al. 1986; Pantel et al. 1990; Anderson et al. 1995; de Jonge et al. 2005; Pinto et al. 2009). In the taste system, CYP can raise taste thresholds through direct cytotoxic effects on peripheral taste structures and by disrupting cell proliferation involved in replacing TSCs when they die (Mukherjee and Delay 2011; Mukherjee et al. 2013a, 2013b, 2017). Cellular assays confirmed a decline in fungiform papillae within 2–4 days after CYP injection that did not recover for 12–16 days. In contrast, circumvallate TBs did not decline until 7–8 days postinjection (Mukherjee and Delay 2011). Research using behavioral assays showed biphasic declines in taste sensitivity after a single dose of CYP (Mukherjee and Delay 2011; Mukherjee et al. 2013b; Jewkes et al. 2018). These findings indicate TSCs are susceptible to the toxic effects of CYP.

TUNEL and caspase-3 assays, both cell death assays, have verified that CYP causes TSCs to die within hours after injection. TUNEL, which detects fragmented DNA, showed a peak response at 6–8 h postinjection whereas caspase-3, which detects apoptosis, revealed a peak signal 18–24 h postinjection. Results from these assays suggest early drug-related necrosis, followed by apoptosis in TSCs (Mukherjee et al. 2017). Similar TUNEL and caspase-3 effects were seen in rat bladder after CYP injection (Jezernik et al. 2003). These effects were accompanied by a significant expression of cytokine mRNA indicating an inflammatory response within the bladder (Malley and Vizzard 2002; Jezernik et al. 2003).

Inflammation is an immune response to infection, tissue damage, or chemical toxicity and involves the expression of cytokines as immune-modulating factors (Vakkila and Lotze 2004). Research has shown that enhanced expression of cytokines in TBs can cause taste-related disorders (Wang et al. 2009) and increase programmed cell death of taste cells when IFN-α and IFN-γ are activated (Wang et al. 2007). Tumor necrosis factor alpha (TNF-α), IFN-γ (both proinflammatory cytokines), and interleukin-10 (IL-10) (anti-inflammatory cytokine) were activated in subsets of TSCs within 6 h after injecting lipopolysaccharide (LPS), a bacterial cell wall component that mimics infection model. TNF-α was found co-localized and highly expressed in sweet and umami (Type II) TSCs (Feng et al. 2012). TNF-α is a trimeric protein molecule released by immune cells such as monocytes and macrophages, and is activated by T lymphocytes. TNF-α induces its effect by binding to 2 types of receptors, TNFR1 or TNFR2. Besides inducing inflammation and immune responses, TNF-α has other functions within cells. In TBs, it is released constitutively during normal physiological conditions, regulating physiological processes such as differentiation, proliferation, and cell survival (Feng et al. 2012). It has some protective effects, but over-expression can be deleterious or even fatal to a cell (Idriss and Naismith 2000). Since increases in TNF-α is a reliable effect of TB inflammation, we used it as a marker to determine if CYP induces an inflammatory response within the taste system.

Given the potential disruptive effects of chemotherapy on normal cells, it is important to try to protect the taste system from the effects of drugs such as CYP. Studies have shown the sulfhydryl drug, amifostine (AMF), is able to protect normal cells during chemo- and radiation therapy by scavenging free radicals in competition with oxygen (Denekamp et al. 1982; Travis 1984; Andreassen et al. 2003; Mell and Movsas 2008; Gu et al. 2014). It also protects the taste system against some of the toxic effects of CYP. Pretreatment with AMF reduces the effects of CYP on the taste system, for example, prevents loss of papillae, protects proliferating cells involved in taste cell renewal, and reduces CYP-induced loss of taste sensitivity (Mukherjee et al. 2013a, 2013b; Delay et al. 2019). Therefore, the second goal of this study was to determine if pretreatment with AMF prevents CYP-induced inflammation in TBs.

In this study, we examined TNF-α expression to determine if it was influenced by CYP over a 72-hour period using immunohistochemistry. Since TNF-α is thought to be expressed in a subset of Type II but not Type III cells, we co-labeled TSCs in fungiform and circumvallate papillae with antibodies to label PLCβ2 (Type II cells) or SNAP-25 (Type III cells) to determine if CYP differently affected TNF-α in these TSCs. We also evaluated the impact of AMF on TNF-α expression in TSCs when administered alone or as a pretreatment to CYP injection.

Materials and methodology

Ethical consideration

The animal care and experimental protocol were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Vermont under protocol 14-003. All efforts were made to minimize suffering and the number of animals used for these experiments. Because this study was intended as an initial investigation into the inflammatory effects of these drugs on the taste system and the window of the investigation was 72 h, only male mice were used to avoid confounding the 4-day female cycle with the effects of the drugs.

Animals

Eight-week-old male C57BL/6J mice, obtained from Jackson Laboratories (Stock No: 000664; https://www.jax.org/strain/000664; Bar Harbor, ME, USA), were acclimatized to the colony in groups of 2–4/cage for at least 7 days before experimental procedures were performed. The animals were between 24 and 30 g at the start of the experiment. The mouse colony was maintained on 12/12 h light and dark cycles. Food (Purina Chow, Prolab RMH 3000) and water were provided to the animals’ ad libitum.

Drugs and reagents

CYP (cyclophosphamide monohydrate, 97%) was obtained from Arcos Organics, (Cat. No. AC203960010, Fisher Scientific, Hampton, NH, USA). Injectable sodium chloride solution (0.9%) was obtained from Hospira Inc. (Cat. No. 0409-7101-68, Lake Forest, Illinois). AMF (2-(3-Aminopropyl) aminoethyl phosphorothioate, 97%) was purchased from Sigma-Aldrich (USP 1019406, St. Louis, MO). Both drugs were freshly prepared in saline solution before injection.

Dosage

The dose for CYP was chosen using a dose-response curve where 3 different doses (37.5, 75, and 150 mg/kg) were evaluated. A dose of 37.5 mg/kg caused some difficult to detect damage to TBs in mice. On the other hand, a dose of 150 mg/kg causes nephrotoxicity in mice (Vizzard 2000; Girard et al. 2011). Ultimately, a dose of 75 mg/kg was chosen, since it has detectable changes within TBs and has been used in previous research (Mukherjee and Delay 2011; Mukherjee et al. 2013b, 2017; Delay et al. 2019; Awadallah et al. 2020; Joseph et al. 2020).

The dose for AMF was based on clinical reports and previous work with mice. When used as a pretreatment to CYP, 100 mg/kg AMF administered subcutaneously 30 min prior to CYP has some protective effects over taste cells of mice without apparent side effects (Mukherjee et al. 2013b; Delay et al. 2019). Clinical data suggest higher doses can cause numerous side effects such as nausea and hypotension (Glick et al. 1984; Glover et al. 1984; Adamson et al. 1995). To avoid these potential adverse effects while enabling us to compare our results with previous data, a dose of 100 mg/kg was chosen for this study.

Tissue collection

Mice were randomly assigned to a group (n = 4 mice/group) prior to CYP or AMF administration and perfused at 1 of the following postinjection time points: 8, 16, 24, 48, or 72 h. At the assigned time point they were euthanized using sodium pentobarbital (Beuthanasia D (NDC 0061-0473-05), Intervet Inc., Madison, NJ, USA). The 0 h or control group (n = 4 mice) was injected with saline (1 mL/kg) followed by immediate perfusion. A group of mice (AMF-CYP group, n = 4) was injected with AMF 30 min prior to CYP injection and a second group (AMF group, n = 4) was injected with only AMF. Both groups were perfused 24 h later. After euthanasia, animals were perfused with 0.1M phosphate-buffered saline (PBS) mixed with 0.1% heparin, followed by 4% paraformaldehyde (Cat. No. 15710, EM Grade, Electron Microscopy Sciences, Hatfield, PA, USA) mixed with PBS to immediately fix tongue tissues. After dissection, tissues were sequentially soaked in 4% paraformaldehyde solution for 3 h, incubated in 30% sucrose solution in PBS for at least 2 days, blocked, and stored in a cryo-mold with O.C.T. Compound (Tissue Tek, Sakura Finetec USA, Torrence, CA, USA) at −80 °C. Frozen tissue was sectioned at 12 µm and mounted onto Super-frost Plus glass slides (Cat. No. 12-550-15, Fisher Scientific).

TNF-α antibody

We used a goat anti-TNF-α antibody (Cat no. AF510NA, R&D Systems, Minneapolis, MN, USA) to immunoreact with mice taste cells. Mouse TNF-α shares 94% sequence identity with this antibody within the extracellular domain. To ensure that this antibody would react with TNF-α expressed in mouse TBs, we employed the LPS-inflammation induction method Feng et al. (2012) used to detect and analyze the function of TNF-α in mouse TBs. This method required intraperitoneal injection of 5 mg/kg (body weight) LPS 6 h before tissue harvest. Tissue processing and evaluation followed the procedures described below. Like Feng et al., we found a strong reaction of the antibody after LPS treatment (Figure 1A, right panels) and no reaction when the primary antibody was absent (left panels).

Figure 1.

Examples of TNF-α, PLCβ2, and SNAP-25 images used to quantify cell counts and intensity measurements. DAPI signals are grayscale (DAPI only images) or are blue in merged images. PLCβ2 or SNAP-25 immunostain is shown as green, and TNF-α immunostain is shown as red. Asterisks indicate the location of taste pores. All scale bars = 20 µm. (A) LPS-induced inflammation 6 h postinjection showing the effectiveness of TNF-α antibody in TBs. Left 2 panels show control tissues (no primary antibody) with no TNF-α reaction detectable. The third panel shows 2 circumvallate TBs and the right panel shows a fungiform TB after reaction with the TNF-α antibody (labeled TNF-α AB). Both show strong detectable TNF-α signals. (B) Circumvallate TB of a mouse 8 h after CYP injection. Dashes in the DAPI image outline the ROI area of the TB where cells were counted and TNF-α signal intensity was measured. The green image shows the PLCβ2 signal within the TB. (C) Fungiform TB of a mouse 8 h after CYP injection. Dashes in the DAPI image outline the ROI area of the TB where cells were counted and TNF-α signal intensity was measured. The green image shows the PLCβ2 signal within the TB. (D) This is an expanded image of the fungiform TB shown in (C) to illustrate additional areas in which the intensity of the TNF-α signal was measured. The intensity of the apical portion of the TB was measured between the 2 slanted bars on either side of the TB. An area directly above the TB (rectangular box) was measured to determine ambient signal intensity. The signal intensity of non-taste epithelium was also measured (rectangular box, right side). (E) Circumvallate TB of a mouse 8 h after CYP injection. The green image shows the SNAP-25 signal of Type III cells within the TB. The TB is from the same mouse shown in (B).

Immunohistochemical labeling

Co-labeling TNF-α with PLCβ2

To co-label with these 2 molecular markers, tissues collected after each treatment condition were thawed and washed in PBS. Tissue processing was completed in batches of slides. Each batch of tissue included sections representing at least 1 mouse from each group injected with CYP and perfused at 0, 8, 16, 24, 48, and 72 h postinjection. Each batch also included sections of an AMF and an AMF-CYP animal. For antigen retrieval procedure, slides were soaked in 10 mM sodium citrate solution in PBS for 20 min at 90 °C. The slides were allowed to cool for 40 min before washing them in 0.1M PBS with 0.3% Triton-X 100 (Cat no. T9284, Sigma Aldrich, St. Louis, MO, USA). The slides were then incubated in a permeabilization buffer comprised of 0.1% Saponin (Cat no. SAE0073, Sigma Aldrich, St. Louis, MO, USA), PBS, and 0.3% Triton-X 100 for an hour. Subsequently, the slides were blocked using 5% normal donkey serum, 0.1% Saponin, and PBS buffer for 1.5 h at room temperature. Primary antibodies against TNF-α and PLCβ2 were applied using goat anti-TNF-α antibody and rabbit anti-PLCβ2 antibody (Cat no. sc-206, Santa Cruz Biotechnology, Dallas, TX, USA) mixed in 5% normal donkey serum solution at 1:75 and 1:500 concentrations, respectively, for 24 h at 4 °C. The slides were rinsed in PBS 0.3% Triton-X 100 before applying Alexa Fluor 488 donkey anti-rabbit IgG (Cat #A21206) and Alexa Fluor 546 donkey anti-goat (A11056) secondary antibody (Life Technologies ltd., Rockville, MD, USA) for 1.5 h at 1:1000 concentration. The slides were subsequently washed in 0.1 M PBS, and then incubated in diamidino-2-phenylindole (DAPI) (1:10,000, Cat. No. D9542, Sigma Aldrich), a nuclear stain to facilitate cell counting. Coverslips (Gold Seal 3323, Cat. No. 6377001, Electron Microscopy Sciences) were mounted with Fluoromount-G (Cat. No. 0100-01, Southern Biotech, Birmingham, AL, USA).

Co-labeling TNF-α with SNAP-25

Additional slides were labeled to detect TNF-α in Type III cells. Type III cells are identifiable with the biomarker SNAP-25 antibody produced in rabbits (1:3000, Cat # S9684, Sigma Aldrich). A protocol similar to the co-labeling of TNF-α with PLCβ2 was used to co-label SNAP-25 with TNF-α.

Image analysis and cell counts

Data collection was performed using fluorescence imaging with a Nikon Eclipse E600 fluorescent microscope with a photometric Cool SNAP EZ camera and SPOT Advanced (Version 5.1) software for image capturing. During preliminary experiments, the optimum gain and exposure time were determined for each antibody and then held constant across all images. TNF-α was imaged with a red fluorescent filter, PLCβ2 and SNAP-25 were imaged in the green channel, and DAPI was imaged in the blue channel. Subsequently, the images were sharpened and merged using Adobe Photoshop CS6 (https://www.Adobe.com). Cell counts were obtained with NIH ImageJ (https://imagej.nih.gov/ij) using the cell counter plugin. Because the TNF-α signal intensity within TBs and adjacent tissues appeared to vary over time after CYP injection, we also measured the mean intensity of TNF-α signal emission within specific regions of interest (ROIs) of original images using Analyze and Measure functions of ImageJ.

TBs on either side of the circumvallate papilla close to the trench were selected for analysis. A TB was counted if the TSCs were oriented along their long axis from the pore region to the basal membrane and the section was from the central third of the bud. Fungiform TBs were selected if they had a taste pore and their cells appeared oriented along their long axis. For each mouse, 5–10 of each type of TB were examined to quantify labeled cells.

Once a TB was identified as meeting criteria for analysis, the outer boundary of keratinocytes of the TB was determined by visual examination of the DAPI signal. The cells within the TB were then evaluated (Figure 1B,C). Quantification of each TB included counts of cells within a TB (DAPI), cells labeled with either PLCβ2 (Type II) or SNAP-25 (Type III), cells labeled with TNF-α, and cells double-labeled with TNF-α and PLCβ2 or SNAP-25. PLCβ2, SNAP-25, and TNF-α are all expressed in the cytoplasm, therefore, to be conservative a cell was not counted unless the cytoplasmic and the nuclear regions were distinguishable. This was particularly important when evaluating TNF-α since this molecule is released into extracellular space, which could potentially lead to an inflated cell count. Cell counts were made by simultaneously viewing each bud with ImageJ and Adobe Photoshop. If labeling of a cell was in question, images were examined by a second observer and then, if needed, by microscopic examination of the section. In addition, random audits of cell counts were made to ensure adherence to criteria was maintained throughout the process.

Intensity/area measures of TNF-α were taken of the same TBs labeled with PLCβ2 antibodies using ImageJ. Sampling of the circumvallate included ROIs of: 1) the region inside each TB (Figure 1B), 2) non-taste epithelium (NTE) at the top of the trench or outside of the trench, and 3) an area of the image without any tissue immediately above the papilla. This last measure represented ambient signal from processing and was subtracted from all other measures before any analyses were performed. Preliminary examination suggested TNF-α signal intensity changed over time in tissues around the pores of fungiform TBs (but not circumvallate papillae). Thus, intensity measures of fungiform TBs were taken at 4 ROIs: 1) interior TB (Figure 1C), 2) tissue surrounding the pore of the TB, 3) NTE, and 4) non-tissue area of the slide immediately above the taste pore (Figure 1D). The average of each ROI was computed for each mouse and then used for all data analyses. A software error occurred during the acquisition of fungiform TB images taken 16 h after injection. While cell counts were still possible, the error prevented accurate measurements of the intensity of these images.

Statistical analyses

All statistical tests were performed with SPSS version 25.0 (IBM Software). Cell counts and intensity measurements were analyzed using a linear analysis of variance (ANOVA) procedure treating postinjection time as a between-subject factor. This was followed by Sidak-corrected post hoc t-tests to compare groups. Since group sizes were small, covariance was treated as a first-order auto-regressive or AR(1) factor, which does not assume any sphericity of the data. All graphs were created in GraphPad Prism 9 (GraphPad Software Inc.). An alpha of P < 0.05 was used throughout.

Results

In total, 616 circumvallate TBs (PLCβ2 = 298, SNAP-25 = 318) and 551 fungiform TBs (PLCβ2 = 286, SNAP-25 = 265) were examined. There were 23.65 ± 0.29 (mean ± SEM) DAPI-labeled cells per circumvallate TB and 19.89 ± 0.23 DAPI-labeled cells in fungiform TBs. In circumvallate tissue sections, there were 5.62 ± 0.08 PLCβ2 positive cells and 3.30 ± 0.06 SNAP-25 positive cells (Figure 1E) per TB. In fungiform sections, there was an average of 4.08 ± 0.09 PLCβ2 positive cells and 1.99 ± 0.03 SNAP-25 positive cells per TB. The total number of cells per TB, and the number of PLCβ2 and SNAP-25 positive cells per TB were relatively constant across groups. The effects of CYP on TNF-α in Types II and III cells over the 72-hour span after injection are presented first and the group comparisons to evaluate the effects of AMF are presented second.

Time postinjection cell counts

PLCβ2 and TNF-α

Images of circumvallate TBs with cells double-labeled with PLCβ2 and TNF-α are shown in Figure 2A. At 0 h (saline controls), there is little TNF-α detectable but within 8 h after CYP administration, the cytokine is clearly evident and remains elevated throughout the 72 h after injection, although it begins to decline by 48 h. By hour 16 TNF-α is dispersing throughout the TB, making it more difficult to identify individual cells in which the cytokine is activated. A similar temporal pattern of TNF-α activation was observed in fungiform TBs (Figure 3A). Diffusion of TNF-α into interstitial space was also pronounced and appears to spread into the pore region and the apical area or crown of the TB.

Figure 2.

This illustrates the effects of CYP on TNF-α expression in circumvallate TBs 8–72 h postinjection. Data for 0 h are from saline control mice. (A) Micrograph examples of TSCs double-labeled with PLCβ2 and TNF-α at 0–72 h after CYP injection. Left column shows the PLCβ2 labeling of cells within each TB. Center column shows the TNF-α labeling within the TB. Right column shows the merged images for the 2 labels. Scale bar = 20 µm. (B) Graph shows the mean cell count (filled circle) for each mouse and the horizontal bars are means for each group. Double-label cells were significantly higher at 8, 16, and 24 h postinjection (**P < 0.01) compared to saline mice. (C) Graph shows the average + SEM intensity difference scores for each group. The score for each mouse was calculated by subtracting the mean intensity of the TNF-α signal measured for NTE from the mean signal intensity measured inside the TB. Differences at 16 and 24 h were significantly greater (*P < 0.05) compared to saline mice.

Figure 3.

This illustrates the effects of CYP on TNF-α expression in fungiform TBs 8–72 h postinjection. Data for 0 h are from saline control mice. (A) Micrograph examples of TSCs double-labeled with PLCβ2 and TNF-α at 0–72 h postinjection. Left column shows the PLCβ2 labeling of cells within each TB. Center column shows the TNF-α labeling within the same TB. Right column shows the merged images for the 2 labels. Scale bar = 20 µm. (B) Graph shows the mean cell count (filled circle) for each mouse and the horizontal bars are means for each group. Double-label cells were significantly higher at 8, 16, 24, and 48 h postinjection (***P < 0.001, *P < 0.5) compared to saline mice. (C) Graph shows the average + SEM intensity difference score for each group. The score for each mouse was calculated by subtracting the mean intensity of the TNF-α signal measured for NTE from the mean signal intensity measured inside the TB and at the apex of the TB. TB-NTE (dash line-filled circles) differences were significantly greater at 24 and 48 h (*P < 0.05) compared to saline mice. Apex-NTE differences (solid line-open squares) were significantly greater at 48 and 72 h (#P < 0.05) than saline mice.

The cell counts of PLCβ2⁺, TNF-α ⁺, and those cells positive for both cell markers found in each type of TB were subjected to ANOVA evaluation and post hoc analyses to detect any changes induced by CYP post-administration. The ANOVA of the PLCβ2 data did not detect a change in number of Type II cells over the time course of the experiment in either type of TB. In contrast, TNF-α labeling increased in Type II cells in circumvallate TBs after CYP injection (F(5, 18) = 8.868, P < 0.001). This increase was detected by a shift from a relatively low baseline (mean = 0.95 ± 0.21 cells/TB) at 0 h to a high (mean = 3.30 ± 0.32 cells/TB) at 8 h postinjection. TNF-α remained at this level of expression through hour 24 postinjection (all Ps < 0.01), then showed a decline at 48 and 72 h toward levels seen at 0 h. The analyses of the TNF-α/PLCβ2 double-label cell counts revealed very similar effects over time (F(5, 18) = 9.588, P < 0.001; Figure 2 A,B). Of the 618 TNF-α positive cells identified in circumvallate TBs, 590 cells (95.5%) were co-labeled with the PLCβ2 antibody.

PLCβ2 and TNF-α labeling in fungiform TBs followed similar patterns after CYP injection. The number of PLCβ2⁺ cells showed a small but significant (F(5, 18) = 7.728, P < 0.001) decrease at 48 and 72 h postinjection (Ps < 0.5 or less) compared to the first 24 h. Analysis of the TNF-α cell counts indicated (F(5, 18) = 15.513, P < 0.001) that expression was highest at 8 h postinjection (P < 0.001) and remained elevated almost to 72 h (P = 0.054) compared to control mice. Cells co-labeled with TNF-α and PLCβ2 also varied significantly over the 72 h (F(5, 18) = 11.075, P < 0.001; Figure 3A,B). The number of double-labeled cells peaked at 8 h and remained higher than 0 h through 48 h after injection of CYP (Ps < 0.05 or less). Of the 546 cells labeled with TNF-α, 533 also were labeled with PLCβ2 immunostain.

SNAP-25 and TNF-α

While the number of SNAP-25 positive cells in circumvallate and fungiform TBs did not vary across the 72-hour span, the number of TNF-α positive cells varied significantly between groups after CYP injection in both types of TBs (circumvallate: F(5, 18) = 20.844, P < 0.001; fungiform: F(5, 18) = 7.736, P < 0.001). Post hoc testing of the circumvallate cell counts indicated that TNF-α expression in the 0 h group was significantly lower than in all other groups (Ps < 0.001). Post hoc testing of the TNF-α ⁺ cells in fungiform TBs found significantly fewer labeled cells in saline mice than in mice at 8, 16, and 24-hour postinjection (Ps < 0.01). Only 5 cells appeared to be co-labeled with SNAP-25 immunostain.

Intensity measurements

Relative intensity/area measures of TNF-α for circumvallate TBs and NTE were evaluated using the original unmodified images captured in the red channel (TNF-α). NTE showed only a small but nonsignificant increase in intensity at 8 h postinjection. However, the intensity of the fluorescent signals of TBs was significantly higher than NTE at all postinjection time points, including the 0 h saline controls (all Ps < 0.01 or greater). Since this difference existed in saline mice, we chose to evaluate difference scores between intensities measured for NTE and TBs of each mouse at each time point. This analysis indicated a significant increase in the intensity of the TBs over time (F(5, 18) = 3.799. P < 0.05; Figure 2A,C). Sidak-corrected tests indicated that the intensity of the TNF-α signal within TBs were significantly elevated at hours 16 and 24 postinjection (Ps < 0.05).

Similar analyses were performed on TNF-α intensity measures of fungiform TBs (without the 16 h data above). No differences in NTE signal intensity were found between time point groups. The ANOVA comparing differences in intensity between NTE and area inside fungiform TBs found a significant group effect (F(4, 15) = 5.152, P < 0.01; Figure 3A,C). The post hoc tests found that the intensity of the TNF-α signal was significantly higher inside TBs at 24 (P < 0.05), and 48 h (P < 0.05) post CYP injection compared to saline mice. Because our qualitative observations suggested the TNF-α signal may be migrating to the apex of the fungiform papilla, we computed a second difference score between the intensities of the apex of the fungiform papilla and NTE. Analysis of these data revealed a significant group difference (F(4, 15) = 3.877, P < 0.05) due to elevated intensity of the apical region at 48 and 72 h postinjection relative to controls (Ps < 0.05).

It should be noted that in spite of our attempts to be conservative, the diffusion of the TNF-α throughout the TB may have resulted in over- or underestimates of the number of TSCs labeled with TNF-α. By extension, estimates of the number of Types II and III double-labeled cells may also have been impacted.

AMF experimental results

TB images showing the effects of AMF alone and as a pretreatment to CYP can be seen in Figure 4A. Compared to saline mice, AMF induced a small but detectable increase in TNF-α. These cells appeared the same as those seen at other time points. That is, TNF-α labeled cells were co-labeled with the PLCβ2 label. Gross inspection of these labels did not reveal any unusual cellular features. Nevertheless, AMF as a pretreatment clearly prevented much of the increase seen after CYP injection without increasing its own level of activation.

Figure 4.

This illustrates the effects of AMF pretreatment on PLCβ2 and TNF-α expression in TBs 24 hrs after CYP injection. (A) Micrograph examples of TSCs double-labeled in saline, AMF, CYP, and AMF-CYP. Left column shows the TNF-α labeling of cells within each TB. Center column shows the PLCβ2 labeling within the same TB. Right column shows the merged images for the 2 labels. Scale bar = 20 µm. (B) Graph shows the mean + SEM cell counts per TB averaged across circumvallate and fungiform TBs for each group. Double-label cells were significantly higher in CYP mice (***P < 0.001) than all other groups. Saline mice had significantly fewer double-label cells than AMF mice (#P < 0.5) and approached significance for the AMF-CYP mice (P = 0.52).

The effects of AMF on TNF-α were evaluated with 2-way ANOVAs comparing TB types (circumvallate vs. fungiform) and drug group (Saline, CYP, AMF, and AMF-CYP) factors. The analysis of the Type II TSCs labeled with PLCβ2 antibodies revealed significant group differences (F(3, 24) = 26.71, P < 0.001). The Sidak-corrected post hoc tests indicated that saline-injected mice had fewer fungiform and circumvallate TSCs labeled with TNF-α than the other 3 groups (Ps < 0.05 or less). In addition, these tests indicated that the CYP group had significantly more labeled cells than the other 3 groups (all Ps < 0.001). The ANOVA applied to the PLC + TNF-α double-label data also detected a significant group effect (F(3, 24) = 22.995, P < 0.001; Figure 4A,B). Post hoc tests found significantly more double-labeled cells in the 24 h CYP mice than the other 3 groups (all Ps < 0.001) and saline mice had fewer double-labeled cells than the AMF group (P < 0.045) and trended toward the same compared to the AMF-CYP group (P = 0.052). Also, the AMF group had significantly fewer double-labeled cells than the AMF-CYP group (P < 0.001). For the tissues immunolabeled with SNAP-25, TNF-α labeling was significantly altered by drug effects (F(3, 24) = 27.746, P < 0.001). Sidak-corrected post hoc testing found that labeling in saline mice was significantly lower than seen in CYP and AMF-CYP mice (Ps < 0.001 and 0.003, respectively) but was not different from AMF mice. On the other hand, CYP mice had significantly more cells labeled with TNF-α than AMF and AMF-CYP mice (Ps < 0.001). TNF-α labeling in AMF and AMF-CYP groups did not differ. Beyond the differences between saline and CYP mice already described above, the analyses of TNF-α signal intensity did not detect any other differences between saline, CYP, AMF, and AMF-CYP groups.

Discussion

Recent research using LPS has shown that TBs are equipped with an elaborate system to regulate inflammatory responses to injury or disease-inducing agents (Cohn et al. 2010; Feng et al. 2014, 2020). A key component of this system involves the proinflammatory cytokine, TNF-α, which is expressed and released by a subset of Type II cells; more specifically, TSCs that express the T1R3 gene common to sweet and umami receptors (Feng et al. 2012). Moreover, TNF-α receptors are found on the cell membranes of all mature TSC types. Secreting TNF-α into the interstitial space within the TB enables Type II cells to modulate other TSCs. In the present research, we determined that a single, moderate dose of CYP can induce an inflammatory response within TBs and that AMF can reduce this response.

The analyses of TNF-α cell counts spanning 72 h from the time of CYP injection indicated that expression of TNF-α increased within 8 h after injection and continued at roughly the same level at least 24 h postinjection. The initial elevation of TNF-α was expected since previous reports had shown that TUNEL, a measure of cell death, peaked in TBs 6–8 h after CYP administration and continued up to 18 h postinjection. Caspase-3 assays showed that some cell death is detectable up to 36 h postinjection (Mukherjee et al. 2017). The intensity measurements of TNF-α suggest that while the initial expression of TNF-α was localized mostly within the cytoplasm of Type II cells, these cells appeared to release TNF-α into extracellular space where it can migrate to influence other cells within the TB and potentially add to the toxic effects of CYP. The decline in number of labeled cells and intensity measures indicate that TNF-α levels began to drop at 48 h and was gradually approaching saline levels by 72 h postinjection.

In this study, TNF-α expression in circumvallate and fungiform TBs was similar with 1 exception. TNF-α appeared to gradually spread to the apical region of fungiform papillae over time, reaching a peak at 48 h after injection. This was not observed around the pore region of TBs in the circumvallate trench or NTE of either type of TB. Previous research indicated that fungiform TBs may be more susceptible to the effects of CYP (Mukherjee and Delay 2011). The spread of TNF-α into the pore region of fungiform papillae may increase the susceptibility of fungiform TBs to the effects of toxic substances such as CYP. Further research is needed to address this possibility.

AMF administered 30 min prior to CYP reduced but did not completely block the level of TNF-α activation by CYP 24 h after injection. These findings are consistent with previous reports that showed the same dose (100 mg/kg, sc) protected the proliferative cell population involved in taste cell renewal along with Types II and III TSCs from the cytotoxic effects of CYP. AMF also reduced the loss of fungiform TBs and losses in taste sensitivity (Mukherjee et al. 2017; Delay et al. 2019). Higher doses would be expected to provide better protection of the taste system but side effects such as emesis, hypotension, bladder toxicity, and mucosal reactions have been reported in humans at higher doses (Adamson et al. 1995; Hensley et al. 1999; Culy and Spencer 2001). AMF also has anti-inflammatory effects. For example, it reduced the inflammatory effects of carrageenan-induced hind paw edema in mice (Bhutia et al. 2010) and cisplatin-induced inflammation, oxidative stress, and disrupted kidney function in rats (Khairnar et al. 2020). In addition, AMF reduced inflammation seen with 5-fluorouracil-induced oral mucositis and hypersalivation in hamsters (Barbosa et al. 2019). In each case, AMF reduced TNF-α and IL-1β as part of its effects. One of the primary pharmacological actions of AMF is to scavenge free radicals within cells in competition with oxygen, thereby countering CYP’s capacity to increase free radicals. Perhaps AMF is able to reduce inflammation by reducing signals such as nuclear factor-ĸb (Nfkb) that activate and/or regulate an inflammatory response in TBs (Feng et al. 2020). The increase in TNF-α in the AMF treatment group was unexpected since this has not been previously reported. It should be noted, however, that most of these studies have examined how much AMF reduces inflammation generated by a known source of inflammation, often without examining the effects of AMF alone. Even so, AMF has been reported to enhance levels of TNF-α in select tissues such as testes (Meistrich et al. 1984). While it is possible that the present finding is the result of relatively small group sizes, perhaps there is an as yet unknown inflammatory component within TBs. It is also possible that normal, non-inflamed TBs are affected somewhat differently by AMF than cells that are inflamed. For example, perhaps the scavenging reactive nitrogen and oxygen species or the hypoxic effects (Koukourakis et al. 2016) of AMF alters the homeostatic level of TNF-α within normal T1R3-positive Type II cells, or AMF alters the signal interaction between T1R3-positive cells generating TNF-α and gustducin-expressing T2R cells generating IL-10 that appears to modulate the T1R3 cells (Feng et al. 2014). Further research can determine if AMF directly alters the taste inflammatory system or acts by preventing excessive injury that triggers inflammation. Regardless, AMF can help improve a patient’s recovery from the effects of chemotherapy.

TBs, with their community of cell types and high rates of cell renewal, are structural units requiring a homeostatic environment to enable normal cell maturation, differentiation, and aging, and normal taste functioning. TBs are equipped with several inflammatory receptors and signaling molecules that appear to be important for maintaining TB structural integrity. Disruptions of this system within TBs can disrupt cell renewal processes, shorten life spans of TSCs, interfere with cell maturation and differentiation, and ultimately alter taste functions (Cohn et al. 2010; Feng et al. 2014, 2015). Cancer patients often report disruptions in taste functions during and after chemotherapy, which can lead to malnutrition and poorer prognosis (Berteretche et al. 2004; Bernhardson et al. 2007; Steinbach et al. 2009; Aaldriks et al. 2013; Nolden et al. 2019). Comparable effects have been reported after treatment with CYP using murine models. CYP-induced disruptions in cell renewal processes in taste and olfactory systems, as well as loss of TSCs, TBs, and taste functions have been observed (Mukherjee and Delay 2011; Mukherjee et al. 2013b, 2017; Jewkes et al. 2018; Delay et al. 2019; Awadallah et al. 2020; Joseph et al. 2020). The results of this study suggest that CYP-induced inflammation may be a contributing factor in these effects.

The findings of this research open new questions that need to be addressed. One of the more pressing is whether CYP and other chemotherapy drugs also impact other elements of the inflammatory response system, such as anti-inflammatory signal molecules such as IL-10 known to modulate inflammatory responses in TBs (Feng et al. 2014). What is the mechanism by which CYP induces the proinflammatory response in TBs? Inflammation is clearly different between male and female subjects (Pace et al. 2017; Rosen et al. 2017; Slavich and Sacher 2019; Dolsen et al. 2020), suggesting an obvious question: are the inflammatory responses of TBs to CYP different in females? Chemotherapy regimens often involve dose fractionation (dividing a dose into smaller components to be administered at different times), but previous research with a murine model found dose fractionation enhanced the negative effects of CYP on the taste system (Jewkes et al. 2018; Delay et al. 2019). Would a fractionated therapeutic approach alter the inflammatory reaction induced by CYP? AMF is able to protect the taste system from the toxic effects of CYP. In view of the potential clinical implications of AMF treatment, what are the mechanisms underlying AMF’s effect on the taste system?

In summary, this study shows that a single moderate dose of CYP induces an increase in the proinflammatory cytokine, TNF-α, in a subset of Type II TSCs. The number of cells expressing TNF-α in circumvallate and fungiform TBs increased significantly by 8 h postinjection and remained high through 24 h postinjection before slowly decreasing toward saline levels. Pretreatment with AMF reduced the inflammatory response, thereby protecting normal cells of the taste system from the toxic effects of the chemotherapy drug. These findings may provide insights into the effects of chemotherapy on the nutritional intake of patients and a method by which some of these effects can be alleviated.

Funding

This research was supported by a grant from the National Institutes of Health, National Institute of Deafness and Other Communicative Disorders (R01DC012829) awarded to E.R.D. and by a John Wheeler Research Grant from the University of Vermont awarded to A.A.S.