Cyclophosphamide has Long-Term Effects on Proliferation in Olfactory Epithelia

Abstract

Chemotherapy patients often experience chemosensory changes during and after drug therapy. The chemotherapy drug, cyclophosphamide (CYP), has known cytotoxic effects on sensory and proliferating cells of the taste system. Like the taste system, cells in the olfactory epithelia undergo continuous renewal. Therefore, we asked if a single injection of 75 mg/kg CYP would affect cell proliferation in the anterior dorsomedial region of the main olfactory epithelium (MOE) and the vomeronasal organ (VNO) from 0 to 125 days after injection. Both epithelia showed a decrease in Ki67-labeled cells compared to controls at day 1 and no Ki67+ cells at day 2 postinjection. In the sensory layer of the MOE, cell proliferation began to recover 4 days after CYP injection and by 6 days, the rate of proliferation was significantly greater than controls. Ki67+ cells peaked 30 days postinjection, then declined to control levels at day 45. Similar temporal sequences of initial CYP-induced suppression of cell proliferation followed by elevated rates peaking 30–45 days postinjection were seen in the sustentacular layer of the MOE and all 3 areas (sensory, sustentacular, marginal) of the VNO. CYP affected proliferation in the sensory layer of the MOE more than the sustentacular layer and all 3 areas of the VNO. These findings suggest that chemotherapy involving CYP is capable of affecting cell renewal of the olfactory system and likely contributes to clinical loss of function during and after chemotherapy.

Introduction

Cancer is the second most common cause of death in the United States, closely following cardiovascular disease (Siegel et al. 2013). Cancer treatment often involves radiation and chemotherapy, each with side effects that can negatively affect prognosis and quality of life of the patient (Spotten et al. 2017). Chemotherapy can have damaging side effects such as alopecia, nausea, immune suppression, and chemosensory alterations (Epstein et al. 2002) such as increases in thresholds of taste and smell during and after therapy (Berteretche et al. 2004; Steinbach et al. 2009; Walliczek-Dworschak et al. 2017). In addition, chemotherapy patients have self-reported changes in nutritional intake, with as many as 82% of cancer patients reporting food avoidances during chemotherapy (Shils 1977). Decreased appetite, weight loss, anorexia, and depression all lead to a lower quality of life and poorer overall prognosis. Over 40% of cancer patients have reported a loss of ~10% of their body weight (Capra et al. 2001). Although this weight loss may be related to treatment-induced conditioned taste aversions (Bernstein 1978; Mattes 1987; Mattes et al. 1987), recent work showed drug-induced morphological changes in the taste system that coupled a loss of taste sensory cells in lingual taste buds with a loss of taste function (Mukherjee and Delay 2011; Mukherjee et al. 2013a,b; Mukherjee et al. 2017; Jewkes et al. 2018). Chemotherapy patients also experience qualitative and quantitative changes in olfactory functions, with decreases in threshold sensitivity more apparent than deficits in their ability to identify odors or discriminate between odors (Steinbach et al. 2009). However, the effects of chemotherapy on olfactory basal cell proliferation has received little attention. Kai et al. found that a subset of chemotherapeutic drugs such as vincristine sulfate, which prevents polymerization of microtubules during cell replication, are capable of inducing cell death in the main olfactory epithelium (MOE) and the vomeronasal organ (VNO) of mice (Kai et al. 2002; Kai et al. 2004). However, a comparable low dose of cisplatin, another chemotherapy agent with alkylating-like properties which inhibits DNA replication, had little observable effect on MOE or the VNO in the same study. The intent of this study was to determine if cyclophosphamide (CYP), an antineoplastic DNA-alkylating agent (Juma et al. 1979), also adversely affects olfactory epithelia.

CYP was one of the first chemotherapy drugs used and is still widely used for a plethora of cancerous and non-cancerous ailments (Weiner et al. 1984; Kushner et al. 1990). Like most chemotherapy drugs, CYP has known cytotoxic effects. CYP is a pro-drug activated by the hepatic enzyme P-450 cytochrome oxidase and is converted to acrolein and phosphoramide mustard, both cytotoxic (Ludeman 1999; de Jonge et al. 2005). CYP has a dose-dependent half-life of about 4–9 h. CYP and its metabolites are cleared within 24 h (Ludeman 1999). Although these metabolites can be toxic to all cells, they primarily target cells undergoing proliferation, causing inter- and intra-strand cross-linkages in DNA, which interfere with replication and potentially result in necrosis and apoptosis. Unfortunately, CYP does not discriminate between normal and cancerous cells, but targets all cells, including those necessary for cell replacement in normal tissues (Ludeman 1999; Mukherjee et al. 2017).

The chemosensory systems may be particularly sensitive to the effects of chemotherapy drugs such as CYP because their end organs are populated with cells that have relatively short life spans and are constantly being renewed. Previous research (Mukherjee and Delay 2011; Mukherjee et al. 2013a,b; Mukherjee et al. 2017; Delay et al. 2019) demonstrated that a single dose (75 mg/kg) of CYP is capable of disrupting taste sensory cell renewal and taste functions, for example, reducing sensitivity. In view of these effects, we postulated that CYP might also disrupt cell proliferation within the olfactory epithelium. A wealth of research has shown that there is a population of stem cells lining the epithelia of the peripheral olfactory system that divides to replenish the epithelia, including new cells in each of 3 broad categories: olfactory neurons, sustentacular cells, and secretory cells (Barber and Raisman 1978a,b; Wilson and Raisman 1980; Farbman 1997; Farbman et al. 1999; Weiler and Farbman 1999; Weiler et al. 1999a; Brann and Firestein 2010; Weng et al. 2016; Schwob et al. 2017). Since the olfactory epithelia are continually being replaced, the proliferative cell populations are likely to be susceptible to the cytotoxic properties of CYP. To evaluate this possibility, we studied the impact of CYP on cell proliferation in the MOE and the VNO, the 2 main peripheral olfactory epithelia in the mouse.

The VNO is a crescent-shaped organ adjacent to the nasal septum and encased in a bony capsule with a duct that links it to the nasal cavity (Doving and Trotier 1998; Weiler et al. 1999a; Weiler et al. 1999b; Francia et al. 2014; Finger et al. 2017). It is present in most mammals and vertebrates but is absent in humans. In the mouse, the MOE is located superior and posterior to the VNO and, like the VNO, is a pseudostratified epithelium of various cell types including bipolar sensory neurons, sustentacular cells (also called support cells), microvillar cells, globose basal cells, and horizontal basal cells (Francia et al. 2014; Finger et al. 2017; Schwob et al. 2017). Neurogenesis in olfactory tissues is highly regulated through apoptosis and continuous replacement by developing neurons derived from the basal layers (Wilson and Raisman 1980; Farbman 1990; Weiler et al. 1999b; Carter et al. 2004; Gokoffski et al. 2010; Francia et al. 2014). Cells within the different regions of the olfactory system have various turnover rates that are thought to decrease as the animal ages (Wilson and Raisman 1980; Weiler and Farbman 1997, 1998a,b; Weiler et al. 1999b). For example, while estimates vary, adult olfactory sensory neurons are generally reported to have a life span of 30–90 days, although some have been reported to live as long as a year (Wilson and Raisman 1980; Brann and Firestein 2010). Because the population of sensory neurons within the MOE are relatively short-lived and are constantly being renewed, patients undergoing chemotherapy could be experiencing a detrimental change in turnover of cells within the olfactory system (Epstein et al. 2002).

The primary goal of this study was to examine the effects of a single injection of CYP (75 mg/kg) on the MOE and VNO over time after administration. We hypothesized that: 1) CYP-injected animals will show a decrease in proliferating cells in both olfactory epithelia, and 2) the marginal zones of the VNO will be more affected than other regions in the VNO due to their higher rate of proliferation (Barber and Raisman 1978a; Weiler et al. 1999b). Overall, we observed an initial decrease in Ki67+ cells at early time points after CYP injection followed by an increase in proliferating cells peaking about 30 days after CYP injection. The MOE was affected more by CYP than the VNO and the marginal zones were affected for a longer time than the rest of the VNO.

Materials and methods

Animals

All mice were handled in accordance with University of Vermont’s Office of Research Protection (IACUC protocol 14-003). Adult C57BL/6J male mice were obtained from Jackson Laboratory (Stock No: 000664; https://www.jax.org/strain/000664, Bar Harbor, ME, USA). All olfactory tissues were obtained through tissue sharing with other experiments studying the effects of CYP on the taste system. Although female mice may respond differently to the effects of CYP on the olfactory system, there were insufficient female subjects to populate all of the experimental conditions. To avoid confounding the results, only male mice were used for this study. All mice were housed in open wire-topped cages in groups of 3–4 per cage to optimize odor exposure. The mice were kept on a 12 h light/12 h dark cycle and given ad libitum access to water and Purina Mouse Chow RMH 3000. Initially, mice were randomly selected for the CYP drug condition, then randomly assigned to one of 13 postinjection time points. Once a CYP mouse was assigned, an age-matched mouse was assigned to the same time point as a saline control. Mice ranged in age from 61 to 121 days old at the time of injection (saline mean age = 85.7 days, CYP mean age = 87.1 days).

Fifty-two mice received a single intraperitoneal injection (75 mg/kg body weight) of CYP (cyclophosphamide monohydrate, 97%; Acros Organics) prepared fresh each day and the 52 mice chosen as age-match controls received a single intraperitoneal injection of physiological saline (0.9%) at the same volume/kg body weight (n = 3–5 mice/group). Four additional mice were treated as no-injection controls (identified as day 0). The postinjection time points were as follows: 1, 2, 4, 6, 18, 30, 45, 60, 75, 90, 105, and 125 days. Selection of these time points was based on preliminary data suggesting that the effects of CYP might extend well beyond 60 days postinjection. At the assigned time point, mice were euthanized with Beuthanasia D (NDC 0061-0473-05, Intervet Inc.) and a cardiac perfusion of 0.1 M phosphate buffer (PB) with 0.1% Heparin, followed by 4% paraformaldehyde in PB was performed. The upper jaw was removed and soaked in paraformaldehyde fixative overnight. The next day the nasal cavities were injected with 7.5% gelatin fill and then put back in paraformaldehyde fixative for an hour before washing with PB. Jaws were then decalcified in a 14% EDTA in PB solution for 4 days, followed by cryoprotection using graded sucrose solutions from 0.5 M to 2.0 M.

All tissues were sectioned at 20 μm using a cryostat and prepared for free-floating immunohistochemical processing to identify Ki67 positive cells. Once the VNO was detected during sectioning, 10 serial sections were collected in a single well. The next 15 sections were eliminated before the next 10 sections were collected in another well. This process was repeated until the VNO was completely sectioned. One section from each well was selected for immunohistochemical processing. Thus, the minimum distance between sections was 300 μm. Free-floating methodology enabled antibodies and solutions better access to both sides of tissue sections, yielding robust labeling of Ki67 positive cells. However, this also resulted in occasional damage to sections that made it difficult to reliably identify and quantify comparable sections of the MOE, especially in the more posterior regions of the MOE. Therefore, to increase confidence that we sampled tissues from comparable areas of the MOE from animal to animal and group to group, we chose to process tissue sections that included both VNO and anterior MOE (Figure 1). Also, since toxic injury can cause a general thinning of the MOE within the first few days post-injury (Bergman et al. 2002; Nathan et al. 2010), we measured the thickness of the MOE. The thickness of each section was measured at a minimum of 5 locations equidistant from each other, then all values were averaged to obtain a single score for the section. While this approach ignored possible medial-lateral differences, it improved the reliability of thickness measurements in free-floating sections. In a preliminary experiment, tissues throughout the extent of the MOE were examined to determine if there were detectable anterior-posterior differences in the effects of CYP on Ki67 labeling or the thickness of the epithelium. T-tests detected a significantly lower label cell count in CYP mice relative to control mice (t (8) = 4.216, P < 0.005), but neither anterior–posterior differences in labeling or epithelial thicknesses reached significance.

Figure 1.

MOE and VNO regions divided into layers. (A) A drawing of a bisected head with the nasal septum removed and the approximate locations of 4 areas of olfactory tissue are shown. These are the MOE, septal organ (SO), Grüneberg Ganglia (GG), and VNO, all of which project to the olfactory bulb (OB). The drawing is adapted from Mohrhardt et al. to illustrate the approximate plane of a coronal tissue section (vertical black line) through the anterior MOE and VNO (Mohrhardt et al. 2018). (B) A drawing of a coronal section taken at the black line in (A) showing the nasal cavity (NC), with the MOE and VNO epithelia outlined (blue and pink, respectively). (C) A coronal section of the MOE superior to the VNO is shown. The sensory layer is outlined with the pink/purple line, and the sustentacular layer bordering the lumen is outlined with the green line. (D) In the VNO, the sensory layer is outlined with the pink/purple line, the sustentacular layer bordering the lumen is outlined with a green line and the 2 marginal layers are outlined with yellow lines. The area outlined in blue is non-sensory and was not counted. The fine print in 1C and 1D are areal measurements of the region of interest generated by the NIS-Elements Basic Research program.

Immunocytochemistry

Free-floating coronal sections were treated with 1:200 rabbit monoclonal primary Ki67 antibody (RM-9106; Thermo Fisher Scientific; RRID: AB_2341197) in tissue culture plates to identify proliferating cells. Sections were washed with PB, incubated in 90% methanol mixed with a 3% H₂O₂ solution, washed with PB, and then placed in blocking solution (5% normal goat serum, 0.1 M PB, 0.3% Triton X-100) for 1 h. Sections treated with the primary antibody were incubated overnight (16 h) at room temperature.

After treatment with the primary antibody, the sections were washed in PB and then incubated with Alexa 546 goat anti-rabbit secondary antibody (1:1000 dilution, Cat# A11010, Invitrogen, ThermoFisher Scientific; RRID: AB-2534093) for 1 h before washing in PB. Control sections were treated the same, except they were not incubated with the primary antibody. Cover slips were mounted using Fluoromount-G (Cat. No.: 0100-01, SouthernBiotech). Tissues were imaged with a fluorescent microscope (Zeiss Axioskop 2 upright microscope) with a TRITC filter and captured through a Photometric Cool SNAP EZ camera. Images of the VNO and MOE were taken, regions of interest were identified, and Ki67+ cells were quantified using NIS-Elements Basic Research (Nikon Instruments, Inc.). During analysis and quantification of Ki67+ cells, images were contrast enhanced by adjusting Levels or Brightness/Contrast settings to visualize the unlabeled tissues.

Quantification

Two to five sections for each mouse were selected for quantification from the processed sections mounted on slides. Because areas of tissue samples could vary depending upon mouse size, and location and angle of the tissue section, each of the regions of interest were outlined and measured. The MOE was divided into sensory and sustentacular layers (Figure 1A–C). Ki67+ cells within each layer were manually counted by blind observers and then converted to the number of cells/10 000 µm² of tissue using NIS-Elements. To increase accuracy, overlapping Ki67+ cells were reexamined with the microscope. If individual cells within a cluster could not be identified, a count of one was assigned. A single score for each mouse was computed for each layer by averaging across all sections. The thickness of the MOE was measured across the epithelial surface of a section at a minimum of 5 locations equidistant from each other, then all measurements were averaged for the section.

The VNO was divided into 3 regions (Figure 1D): sensory and sustentacular layers, and 2 marginal zones, one on either side of the organ. The marginal zones were conservatively defined by drawing a straight line between the tips of the crescent ends of the lumen and then bisecting the line to find the midpoint between the tips. To identify the marginal zone of one tip of the crescent, a second line was drawn at a 30° angle from the midpoint to the outer boundary of the VNO (Figure 1D). This was repeated to identify the marginal zone of the opposite tip. The sustentacular layer was defined as the area adjacent to lumen where sustentacular cell bodies were located and the sensory zone was defined as the remaining area of the VNO (Figure 1D). Areal measurements and counts of Ki67+ cells for each area were converted to the number of cells/10,000 µm². The data for the 2 marginal zones were combined for all analyses. The thickness of the VNO was measured at the same planes used to define the medial edges of the marginal zones and at a perpendicular line drawn from the midpoint between the marginal zones to the outer border of the sensory layer. A single score for each mouse was computed by averaging the 3 measurements across all sections.

Cell counts and measurements were evaluated using a linear analysis of variance (ANOVA) procedure treating drug treatment and days as between-subject factors. This was followed with simple effects tests and Sidak-corrected post hoc t-tests to make individual comparisons between groups (Howell 2016). Data for saline-injected mice were evaluated first to establish baseline levels of Ki67 activity within each region over the 125 days of the experiment. Then the counts per micron area for the saline mice were compared to the data obtained for CYP-treated mice. Since Ki67+ counts for day 0 mice and saline-injected mice at day 1 did not differ for any condition, the day 0 data are shown in each graph but were not included in any analysis. Because the group sizes were small, covariance was treated as first order auto-regressive or AR(1), assuming no sphericity of the data. All statistical tests were performed with SPSS version 24.0 (IBM Software) and graphs were made with GraphPad Prism 8 (GraphPad Software Inc.). All graphs depict the average and standard error of the mean for each time point.

Results

Although olfactory neurons can be found in 4 regions in the nasal cavity, we chose to examine only the MOE and the VNO. These 2 regions have the largest concentrations of sensory neurons, are easily identified, and have been shown to be responsive to many odorants and pheromones (Doving and Trotier 1998; Weiler et al. 1999a; Wirsig-Wiechmann and Wiechmann 2001; Delay and Restrepo 2004; Zhang and Delay 2006; Zhang et al. 2008, 2010; Ogura et al. 2010; Yang and Delay 2010; Vick and Delay 2012; Cherian et al. 2014).

For this study, the region above the basal lamina of the MOE to just below the cell bodies of the sustentacular cells was identified as the sensory layer. This layer contains horizontal basal cells, globose basal cells, and the cell bodies of the immature and mature olfactory sensory neurons. The sustentacular layer extended from the bottom of the cell bodies of the sustentacular cells to the surface of the lumen and contained the cell bodies of the sustentacular and microvillar cells. Both layers have cells surrounding the ductal region of Bowman’s glands. The same delineation was used for the VNO, except the marginal zone was defined as the area between the tip of the crescent and the line from the midpoint defining a 30° angle (see above and Figure 1D).

MOE

Examination of Ki67-labeled cells in the sensory and sustentacular layers of the MOE (Figures 1C and 2) revealed noticeably different rates of proliferation between saline and CYP-treated animals over time. CYP had a greater impact on the magnitude and duration of posttreatment proliferation in the sensory layer than in the sustentacular layer.

Figure 2.

Ki67+ cells in both layers of the MOE of saline- and CYP-treated mice. The number of Ki67-labeled cells (seen as white-labeled cells in these fluorescent images) in the MOE varied over the 125 days after the CYP injection. Images in the left column are taken from saline-injected animals, whereas those in the right column are from CYP-injected animals. Images in the first row are 2 days postinjection, second-row images are 30 days postinjection, third-row images are 60 days and the last row images are 125 days postinjection. The scale bar applies to all the images. While saline controls showed minor variation in the number of KI67+ cells over time, CYP-induced large shifts in Ki67+ cells postinjection with no detectable labeled cells on day 2 and high levels of labeled cells on day 30.

Sensory layer

Saline mice

In the saline control tissues, most of the Ki67+ cells were in the sensory layer, directly above the basal lamina, although a few Ki67+ cells were scattered above this first layer of labeled cells (Figure 2, left column). There were a few Ki67+ cells scattered in the sustentacular layer but these were not as numerous as those in the sensory layer. The number of Ki67+ cells throughout the epithelium appeared to decrease as the postinjection time increased.

An ANOVA of Ki67+ cell counts/10 000 µm² of the MOE sensory layer of saline-injected mice indicated significant variation over the 125 days posttreatment (F(11, 35) = 3.825, P < 0.001; Figure 3A,B). Post hoc analysis revealed that the number of Ki67+ cells 2 and 90 days after injection were significantly higher than the number of Ki67+ cells 30, 75, and 105 days postinjection (Ps < 0.05). Ki67 labeling in saline mice on day 45 approached but did not reach a significantly higher level than labeling in saline mice on day 30 (P = 0.059). The ANOVA of Ki67+ cells in the sustentacular layer of saline control mice did not find any significant differences over the course of the experiment (Figure 3C,D).

Figure 3.

Ki67+ cells in the MOE of saline- and CYP-injected mice varied over time posttreatment. (A) Sensory layer days 0–125. Saline-injected mice exhibited more Ki67 labeling 2 and 90 days postinjection than observed at 30, 75, and 105 days postinjection. In contrast, the number of Ki67+ cells in CYP-injected mice in the sensory layer dropped to zero by 2 days postinjection. Once proliferation began again on day 4, it increased to levels exceeding saline mice between days 6–30 postinjection. At 45 and 60 days postinjection, there was a large decrease in Ki67+ cells in CYP mice compared to saline controls. From day 75 on, there were no significant differences between the sensory layers of saline and CYP mice. (B) Sensory layer days 0–6. Expanded view of Ki67 expression during the first 6 days after injection. (C) Sustentacular layer days 0–125. Ki67+ cells of saline mice were relatively constant over the course of the experiment in this layer. Immediately following CYP injection, cell cycling in CYP mice decreased to zero (day 2), but restarted again by day 4 with the highest rate in this layer seen at day 30. Thereafter Ki67 expression in CYP mice was similar to saline mice. (D) Sustentacular layer days 0–6. Expanded view of Ki67 expression during the first 6 days after injection. Saline v CYP mice: ^aP < 0.05, ^bP < 0.01, ^cP < 0.001.

CYP mice versus saline mice

In contrast to saline mice, Ki67 labeling in the sensory layer of the MOE of CYP-injected mice had a significantly different pattern of expression over days (days factor: F(11, 66) = 5.543, P < 0.001; days by drug interaction: F(11, 66) = 9.867, P < 0.001; Figure 3A,B). Simple effects tests comparing expression of Ki67 of the saline mice to CYP mice identified several time points after drug injection when the 2 groups differ. These tests showed the number of Ki67+ cells in CYP mice was decreasing compared to saline mice the first day postinjection (P < 0.05) and was completely absent on the second day (P < 0.001; Figures 2 and 3A). Cell renewal in CYP mice appeared to begin again about day 4, and by day 6, Ki67+ cells exceeded levels observed in saline-injected mice (P < 0.01). Proliferating cells in CYP mice remained significantly higher than saline mice on days 18 (P < 0.01) and 30 (P < 0.001), reaching the highest levels of expression 30 days postinjection. The majority of the Ki67+ cells were located adjacent to the basal lamina (Figure 2, CYP day 30). In the sensory layer, Ki67 labeling in CYP mice declined to a level significantly below saline mice at 60 days (P < 0.05; Figure 2, CYP 60 days). After day 60, Ki67 expression tended to remained similar to saline control levels, including 90 days after injection when an increase in Ki67+ cells was also seen in saline mice. The initial postinjection suppression of cell proliferation followed by a prolonged higher level of cell cycling suggests that the mouse MOE is particularly sensitive to the effects of CYP.

Sustentacular layer

Saline mice

In general, most of the Ki67+ cells in the sustentacular layer of the MOE were scattered across the layer, although small clusters were also observed. By the later time points most of the labeled cells of saline mice were in the sensory layer with only a few in the sustentacular layer (Figure 2, left column). The only difference between saline groups was seen when Ki67+ cells on day 2 were significantly higher than those seen on postinjection day 75 (P < 0.05).

CYP mice versus saline mice

Within the MOE sustentacular layer of CYP mice, Ki67+ cells decreased significantly within 24 h and dropped to zero for all mice by the second day postinjection (both Ps < 0.05; Figure 2, CYP day 2; Figure 3C,D). Proliferation in CYP mice restarted by day 4. By day 6, however, Ki67 expression was increasing and reached its highest level of expression at day 30, a level significantly greater than saline control mice (day 30: P < 0.001; Figure 3C,D). Thereafter Ki67+ cells of CYP mice dropped to near zero at day 60 and remained low for the rest of the experiment.

In general, proliferative cells within both layers of the MOE were adversely affected by a single dose of CYP. Cell proliferation in both layers was completely suppressed after drug administration, then gradually increased until it peaked at 30 days postinjection. Moreover, the sensory layer was much more sensitive to the effects of CYP than the sustentacular layer of the MOE.

MOE epithelial thickness

The average thickness of the MOE epithelium of all saline mice was 80.66 (±1.36 SEM) µm. Analyses comparing the thickness of the MOE epithelium of CYP mice with saline-injected mice across days postinjection showed a significant reduction in thickness of the MOE of CYP mice (mean = 49.34 ± 1.96 µm) the first 2 days after administration of CYP (both Ps < 0.001). By day 4, the MOE was beginning to thicken (mean = 67.79 ± 3.43 µm) in the CYP mice, but was still significantly less than the day 4 saline mice (mean = 80.18 ± 3.83 µm; P < 0.05). By day 6, the thickness of the MOE of CYP mice was nearly equivalent to saline mice. The mean thickness of the MOE epithelium of CYP mice for the remainder of the experiment was 76.18 ± 1.65 µm, except day 60 when some thinning was observed (mean = 60.22 ± 3.83 µm).

These data, combined with the Ki67 data, suggest that CYP had a cytotoxic effect on the MOE which caused thinning of the MOE and suppressed cell renewal of the epithelium for up to 4 days after injection. Once cell proliferation was reinitiated, cell renewal continued at higher than normal rates for 30 days or more postinjection.

VNO

The 2 layers and the marginal zone of the VNO were evaluated to determine if this system was also susceptible to the effects of CYP. Ki67+ cells were visible in all 3 layers of the VNO. Compared to the MOE, only a few labeled cells were observed in either the sensory or the sustentacular layer. Most of the Ki67+ cells in the VNO of saline-injected mice were located in the marginal layer (Figure 4, left column). In general, the effects of CYP became apparent almost immediately (Figures 4 and 5). After an initial drop in the rate of cell proliferation in both layers and the marginal zones after injection, proliferation gradually increased before appearing to oscillate over time.

Figure 4.

Variation in Ki67+ cells in the VNO between saline- and CYP-injected mice. The fluorescent images shown in the left column are from the saline controls, whereas those in the right column are from CYP-injected animals. Shown are images of the VNO at 1, 30, 60, and 125 days postinjection (top to bottom). All of the images were taken at the same magnification. At day 1 postinjection, there were more Ki67+ cells (white-labeled cells) in saline mice than in CYP-treated mice. By day 30 postinjection, little Ki67 label was observed in the saline-treated mice compared to that seen in the CYP mice. In CYP mice, Ki67+ cells were found in the sustentacular layer as well as the sensory layer. At this time point, fewer Ki67+ cells were observed in the marginal zones with either treatment. While few labeled cells were found in either group at day 60, later time points (e.g., day 125) showed higher levels of expression.

Figure 5.

Mean (±SEM) of Ki67+ cells the VNO of saline- and CYP-injected mice over time posttreatment. (A)Sensory layer days 0–125. Ki67 labeling of saline mice was higher 2 days postinjection than observed in saline mice 30, 60, 75, and 125 days postinjection (Ps < 0.05, no symbols). In addition, Ki67 labeling was lower in saline mice at 30 days postinjection compared to mice at 18 days (P < 0.05, no symbol). When compared to CYP mice, however, Ki67 expression was higher in saline mice than CYP mice at days 1 and 2, and lower than CYP mice on days 30 and 75 posttreatment. (B)Sensory layer days 0–6. Expanded view of Ki67 expression during the first 6 days after injection. (C)Sustentacular layer days 0–125. Ki67+ cells in saline mice did not vary significantly over the course of the experiment. In contrast, labeled cells in the CYP mice were lower than saline mice at days 1 and 2, and higher than saline mice at days 30 and 45. (D)Sustentacular layer days 0–6. Expanded view of Ki67 expression during the first 6 days after injection. (E)Marginal layer days 0–125. In saline mice, Ki67+ cells were generally constant throughout the experiment except for days 30 and 75. Ki67+ cells on day 75 was significantly lower than all days (P < 0.05, no symbol) except day 30. Labeling on day 30 was significantly lower than day 18 (P < 0.05, no symbol) and approached significance compared to day 45 (P = 0.065). Compared to CYP mice, Ki67 expression in saline mice was significantly higher at days 1 and 2, and lower at 75 days posttreatment. In general, CYP mice had significantly higher levels of Ki67+ cells than saline mice between days 18 and 105 postinjection (P < 0.05). (F)Sensory layer days 0–6. Expanded view of Ki67 expression during the first 6 days after injection. Saline v CYP mice: ^aP < 0.05, ^bP < 0.01, ^cP < 0.001.

Sensory layer

Saline mice

Ki67+ cells observed in the sensory layer of saline mice appeared in small clusters, often with one cell superior to the one adjacent to the basal lamina (Figure 4, left panel). Proliferation in the sensory layer of the VNO followed a cycle-like pattern with alternating highs (mean ~ 1 cell/10 000 µm²) and lows (near zero) occurring every 30–45 days (F(11, 33) = 4.128, P < 0.001; Figure 5A). Saline-injected mice had higher levels of Ki67+ cells at day 2 than saline mice at 30, 60, 75 and 125 days postinjection (Ps < 0.05). Also, saline mice at day 30 had lower Ki67 labeling than saline mice at day 18 (P < 0.05) and approached but did not reach significantly lower levels (P = 0.065) compared to saline mice at day 45.

CYP mice versus saline mice

When compared to saline-injected mice, the effects of CYP on Ki67 expression in the sensory layer of the VNO are immediately apparent as a significant days by drug interaction (F(11, 68) = 3.682, P < 0.001; Figures 4 and 5A,B). Ki67+ cells of CYP-injected mice were diminished within 24 h (P < 0.01) and completely absent 2 days (P < 0.001) after injection. Cell proliferation in the sensory layer restarted by day 4, when the number of Ki67+ cells in CYP mice were once again comparable to saline mice, and continued to increase to day 45. At day 30, Ki67 expression was significantly higher in CYP mice than saline mice (P < 0.05). However, even though proliferation continued to increase between days 30 and 45 in CYP mice, the differences between CYP and saline mice were not significant because saline mice also showed an increase in Ki67+ cells. By day 60, the number of Ki67+ cells in CYP and saline mice had decreased to a similar level. At postinjection day 75, Ki67+ cells of CYP mice again increased significantly while remaining low for saline mice (P < 0.05). Differences in Ki67 labeling between the 2 drug conditions did not reach significance in the sensory layer after that time point.

Sustentacular layer

Saline mice

Like the sensory layer of the VNO, Ki67+ cells in the sustentacular layer of saline controls were sparse over all of the time points examined (Figure 4). An ANOVA of these data showed only minor fluctuations in Ki67+ cells over the course of the experiment, which did not quite reach significance for the days factor (F(11,33) = 2.084, P = 0.051; Figure 5C,D). However, a Sidak-corrected t-test indicated that there were significantly fewer Ki67+ cells in saline mice at day 30 compared to saline mice at days 18 and 90 (Ps < 0.05).

CYP mice versus saline mice

When CYP mice were compared to saline mice, ANOVA results found a significant day main effect (F(11, 69) = 3.091, P < 0.01) and day by drug treatment interaction (F(11, 69) = 2.292, P < 0.05). When the number of Ki67+ cells in CYP mice increased, the pattern of expression was not evenly distributed across the sustentacular layer. Instead, there were areas with numerous labeled cells and other areas with a more diffuse pattern of labeled cells (Figure 4, right column). Simple effects tests indicated that Ki67 expression in CYP mice was significantly lower than saline mice day 1 postinjection (P < 0.01) and no Ki67+ cells were observed in CYP mice by day 2 (P < 0.001; Figure 5D). Cell cycle activity restarted 4 days after CYP treatment and gradually increased over the next several weeks. The Ki67+ cells detected in CYP mice approached but did not reach significantly higher levels than saline mice at 30 days postinjection (P = 0.056), but were significantly higher than saline mice at 45 days (P < 0.01). However, Ki67+ cells in CYP mice dropped to levels similar to saline mice at day 60 and remained comparable to saline mice throughout the rest of the experiment.

Marginal zone

Saline mice

The number of Ki67+ cells observed in saline-injected mice showed some oscillation over the course of the 125 days of the experiment (F(11, 33) = 2.534, P < 0.05; Figures 4 and 5E,F). Ki67+ cells in saline mice at days 2, 45, and 90 were significantly higher than Ki67 labeling seen on days 30 and 75 (Ps < 0.05).

CYP mice versus saline mice

Cell cycle activity in the marginal zone of the VNO of CYP mice also showed significant shifts over the course of the experiment, but these shifts were different from those seen for saline mice (F(11, 69) = 3.254, P < 0.001; Figures 4 and 5E,F). For CYP treated mice, Ki67+ cells were significantly lower than saline mice 1 day postinjection (P < 0.05), and completely absent 2 days postinjection (P < 0.001), but reappeared at levels comparable to saline mice on day 4. While the number of Ki67+ cells was higher than saline mice at day 30, this difference was not significant (P = 0.064). Like saline mice, the number of Ki67+ cells in CYP mice increased again between days 30 and 45, then dropped again on day 60. Another increase in Ki67+ cells of CYP mice was detected at day 75, the same time point at which Ki67+ cells in saline mice continued to decrease (P < 0.05). After the initial CYP-induced depression in Ki67 expression, cell proliferation in the marginal zone appeared to be generally higher in CYP mice than in saline mice over the experiment. To test this possibility, an ANOVA comparing the number of Ki67+ cells between days 18 and 105 revealed significantly higher levels of Ki67 in CYP mice than saline mice, F(1, 38) = 4.846, P < 0.05. This indicates that cell proliferation was generally greater in CYP mice than in saline mice throughout this period (the only exception was day 60).

Thus, saline mice showed a mild increase in proliferation in all 3 regions of the VNO shortly after saline injection, after which proliferative activity seemed to settle into small cyclic shifts, probably as a result of oscillations in cell populations within these areas. On the other hand, cell proliferation in all 3 regions of the VNO of CYP-injected mice was initially suppressed, reinitialized by the fourth day postinjection, and then appeared to cycle at levels well above saline animals, especially between 30 and 45 days after CYP treatment.

VNO epithelial thickness

The average thickness of the VNO epithelium of saline mice was 165.28 ± 4.07 µm throughout the experiment. Two days after CYP injection, the average thickness of the VMO was 139.91 ± 13.93 µm (P < 0.05). During the remainder of the study, the VNO of CYP mice averaged 157.72 ± 4.58 µm.

Comparison of sustentacular layer in the MOE and VNO

Cell proliferation in the sustentacular layers of the MOE and the VNO showed similar trends after CYP treatment. Proliferation decreased to zero by the second day postinjection before restarting. Once restarted, proliferation in CYP mice gradually increased to levels higher than saline mice before slowing. Cell cycling in the CYP mice exceeded saline levels again 30 days post CYP injection in the MOE and 45 days in the VNO. Cell cycling returned to saline levels in both epithelia for the remainder of the experiment. Examination of sections with proliferating cells did not show morphological evidence of glandular tissue or blood vessels associated with labeled cells (Figure 6).

Figure 6.

Ki67+ cells in the sustentacular layers of the MOE and VNO. Superimposing DIC image of the same section with the fluorescent image clarified that the Ki67-labeled cells were in the sustentacular layer of the MOE and the VNO. The left image is of the MOE at day 30 postinjection and the VNO at day 125 postinjection. Both illustrate Ki67+ cells in the sustentacular layer of their respective epithelium.

Effects of age

The mice used in this study ranged from 61 to 121 days of age at the time of injection. Previous studies have suggested that the age of the mouse at the time of injection can influence cell cycling in the olfactory epithelium, especially if they are less than 2 months of age (Weiler and Farbman 1998b; Brann and Firestein 2010; Brann et al. 2015). To determine if this might be a factor in this experiment, or if the effects of CYP might be influenced by the age of the mice in this study, analysis of covariance treating age as the covariant, was conducted. This analysis did not detect a significant effect related to the age of the mice at injection.

Qualitative observations

Processing and mounting free-floating sections generally worked well except for tissues collected from CYP mice within 4 days of their injection. These tissues were treated in the same manner as other olfactory tissue processed with them, but they were often more difficult to mount. Sections were not as firm and tended to fold or tear more easily, as if there was less cohesiveness between cells. A similar loss in tissue integrity has been observed for tongue sections obtained within the first 4 days following administration of CYP (unpublished data, E.R. Delay). Previous research has reported that CYP can cause a loss of cell adhesion (Yoon et al. 1997; Lee et al. 2008b), which may account for these observations.

Discussion

Chemotherapy patients often report disruptions of taste and smell functions as strong aversive long-lasting side effects of drug therapies. Chemotherapy drugs such as CYP are designed to attack replicating cells, either by disrupting the cell replication process or killing the cell. Recent studies have shown that CYP can kill taste sensory cells and suppress cell proliferation involved in renewing cells within taste buds (Mukherjee et al. 2013a,b; Mukherjee et al. 2017; Delay et al. 2019). Like the taste system, cell renewal is important for maintaining normal functioning of the olfactory system. Most sensory neurons of murine olfactory epithelia have life spans estimated between 30–90 days before needing to be replaced (Barber and Raisman 1978a; Wilson and Raisman 1980; Schwob et al. 2017). The results of the present study indicate that a single 75 mg/kg dose of CYP can alter proliferation responsible for cell renewal within the MOE and VNO of mice. These findings strongly suggest that olfactory deficits experienced by patients during and after treatment with CYP may be, in part, a result of the drug’s effects on their olfactory epithelia (Epstein et al. 2002; Berteretche et al. 2004; Steinbach et al. 2009).

Established models of MOE injury such as inhalation of methyl bromide or methimazole, an antithyroid drug that can be administered orally or intraperitoneally, cause immediate thinning of the MOE epithelium due to extensive death of all cell types and a diminution of cell proliferation (Genter et al. 1995; Huard and Schwob 1995; Schwob et al. 1995; Huard et al. 1998; Bergman et al. 2002; Jang et al. 2003; Brann et al. 2015; Schwob et al. 2017). In those studies, cell proliferation usually restarts within 24 to 48 h and increases to a level well above baseline within 4–7 days. Thereafter the rate of proliferation usually declines towards baseline for up to 30 days, slowing as the epithelial cell population is restored. The first immature cells can be detected within 3 days and the first new mature cells can be identified 7–14 days post-injury. Functionality returns as new cells begin to differentiate and mature (Huang et al. 2019). In this study, the effects of CYP on proliferation appeared to follow a comparable pattern. The epithelium of the MOE thinned immediately, probably due to the cytotoxic effects of the drug, and cell proliferation in the MOE was suppressed for 2–3 days after administration. By the fourth day after injection, proliferation started to recover and thickening of the MOE was apparent. Cell proliferation, especially in the sensory layer, continued to increase well beyond control levels, peaking 30–45 days postinjection. Thereafter, proliferation decreased until it was near zero 60 days post-CYP injection, perhaps because the epithelium was almost fully repopulated by this time point. After a quiescent period, cell cycling appeared to return to control levels. It is possible that proliferation varied more than we are reporting since the intervals between time points after the first 6 days are broad (12–15 days). However, our preliminary data suggested that proliferation between these time points did not vary radically from the trends observed. Also, we should note that CYP may not have a uniform affect throughout the MOE due to cell diversity. For example, tissues in this study were primarily from anterior dorsomedial areas of the MOE. Olfactory neurons in this area express high levels of the enzyme NADPH:quinone oxidoreductase and little or no olfactory cell adhesion molecule, whereas neurons in ventrolateral areas of the MOE have opposite levels of these markers (Gussing and Bohm 2004; Coleman et al. 2019). Regional differences in cell phenotypes such as these may alter the potency and effects of CYP.

There are several potential mechanisms by which CYP might affect proliferation within the olfactory system. CYP is a prodrug but its metabolites, phosphoramide mustard, and acrolein, either kill cells immediately or disrupt their functioning, particularly their ability to replicate. For example, phosphoramide mustard is an alkylating agent that generates inter- and intra-strand cross-linkages at quanine N-7 positions (Hall and Tilby 1992). Open DNA strands such as when the cell is replicating or encoding messenger RNA are particularly susceptible and, if the DNA sequence cannot be repaired, induces apoptosis (Ludeman 1999; de Jonge et al. 2005). Acrolein is a reactive unsaturated aldehyde that is toxic to a wide range of cells (Moghe et al. 2015; Chen et al. 2017; Chen et al. 2019), including neural and epithelial cells (Chen et al. 2017). It has the capacity to compromise cells by DNA, RNA, and protein adduction, generate oxidative stress, disrupt mitochondrial function, damage cellular membranes, cause endoplasmic reticulum stress (Lopachin and Decaprio 2005; Moghe et al. 2015), and disrupt gene transcription and regulation, for example, those encoding for the antioxidants, superoxide dismutases 1, and glutathione peroxidase (Oraby et al. 2010; Goncalves and Goldstein 2019).

Besides its capacity to interfere with DNA replication, CYP may interfere with molecular signals regulating cell cycling of the basal cells, especially globose basal cells responsible for cell renewal in the MOE, as well as disrupt biological systems that affect cell proliferation or the environment in which proliferation, differentiation, and maturation take place. For example, CYP can induce inflammation in epithelial tissues (Jezernik et al. 2003; Girard et al. 2011), including the taste system (Wang et al. 2007; 2009; Sarkar et al. 2017). This may interfere with cell proliferation, adhesion, and differentiation by creating an environment that is less than optimal for cell renewal (Zhu et al. 1987; Yoon et al. 1997; Lee et al. 2008b). In addition, loss of cell adhesion following CYP treatment has been reported (Yoon et al. 1997; Lee et al. 2008b; Khalili and Ahmad 2015), an effect consistent with our observation of the fragile quality of tissue during the first few days after drug administration. CYP is also capable of altering RNA transcription and interfering with genetic regulation (Oraby et al. 2010; Goncalves and Goldstein 2019). These effects may interfere with pathways such as regulatory mechanisms (e.g., signaling molecules) thought to be responsible for cell differentiation and maturation in olfactory epithelia (Iwai et al. 2008; Nathan et al. 2010; Packard et al. 2011a; Packard et al. 2011b; Jia and Hegg 2012; Holbrook et al. 2014; Herrick et al. 2017; Schwob et al. 2017). Any of these effects or some combination of these effects may induce a longer period of heightened levels of proliferation to reestablish a normal working population of sensory neurons and other cells within the MOE. The slowly developing increase in proliferation of the sustentacular layer also may be the result of a disturbance in cell renewal processes. More research will be needed to address these possibilities and to determine which basal cell populations are affected by CYP.

The hepatic P-450 system of the liver is known to metabolize CYP into its toxic metabolites, but cells within the olfactory epithelia also have high levels of P-450 system which might intensify the effects of the drug (Dahl et al. 1982; Voigt et al. 1985; Reed et al. 1986; Lazard et al. 1990; Asakawa et al. 2017). However, a single dose of CYP had effects on the MOE that lasted well beyond the pharmacological life of the drug. CYP caused some thinning of the MOE within 24 h, suggesting the drug exerted some cytotoxic effects on these cells. Comparable effects have been described for the taste system in which the same dosing (75 mg/kg) of CYP caused an initial surge in death of taste sensory cells, detectable with a TUNEL assay, that peaked 6–8 h following injection and subsided by 12 h (Mukherjee et al. 2017). This corresponds roughly with the half-life of CYP and suggests the initial loss of chemosensory cells is due to toxic effects of the drug’s metabolites. A second wave of taste cell death was detected 18–36 h postinjection through a caspase-3 assay, suggesting that these taste cells were damaged beyond their ability to recover from the effects of CYP and the canonical apoptotic pathway involving caspase-3 was eventually activated (Mukherjee et al. 2017), findings consistent with the effects of CYP on bladder epithelium (Jezernik et al. 2003). Given the cytotoxic effects of CYP on epithelia, it is possible that the thinning of the MOE was due to drug-induced loss of cells immediately after drug administration. It is likely that CYP can cause more extensive damage to the olfactory epithelia in a therapeutic setting where doses of the chemotherapy drug are higher and/or administered in multiple doses over time (Levin and Hryniuk 1987; Smith et al. 2015; Lamar et al. 2016; Delay et al. 2019). It is important that more research using assays such as TUNEL and caspase antibodies be conducted with more intense dosing and dose regimens to evaluate whether specific cell types throughout the MOE are more susceptible to the effects of CYP and determine if the cells with P-450 compounds contribute to these effects.

There are fewer studies of the effects of injury on neurogenesis in the VNO, but it appears that the neurogenic response to injury in this organ may be similar to the MOE (Kai et al. 2004; Brann and Firestein 2010; Brann and Firestein 2014). The vascular supply of the VNO is extensive and regulation of this system plays an important role in its primary sensory function (Meredith et al. 1980; Doving and Trotier 1998; Meredith 1998; Eisthen et al. 2000; Boehm 2006; Samuelsen and Meredith 2009; Vick and Delay 2012; Cherian et al. 2014). Because CYP and its metabolites were being delivered to the VNO via the vascular system, we anticipated CYP might have a more potent effect on this structure than we detected. The baseline rate of proliferation (number of Ki67+ cells/area) and the magnitude of the effects of CYP on proliferation in the VNO were not as high as observed in the MOE. The reason for this is not clear but might be related to the VNO’s response to noxious stimulation while the drug is present. A wide range of chemical stimuli and bitter substances will induce the duct of the VNO to close (Ogura et al. 2010). Perhaps a component of this system can also protect the VNO from circulating toxic substances. Like earlier reports, proliferation in the VNO was more prevalent in the marginal zones than the sensory or sustentacular layers (Weiler et al. 1999b; Martinez-Marcos et al. 2005; Weiler 2005; de la Rosa-Prieto et al. 2009; Brann and Firestein 2010). Nevertheless, CYP affected cell proliferation in all 3 regions of the VNO much like the sensory layer of the MOE. Cell cycling was initially suppressed for 2–3 days and was reinitiated by the fourth day after drug administration. In general, proliferation in the sensory and sustentacular layers increased to levels above those of saline mice between postinjection days 18–45, then decreased to levels comparable to those of saline mice. Proliferation in the marginal zone of CYP mice tended to follow the same pattern but remained at levels higher than saline mice for a longer period postinjection. This sequala is similar to the pattern of neurogenesis following bulbectomy or nerve damage that led Brann and Firestein (2014) to suggest that the VNO responds to injury much like the MOE. The results of this study support and extend their hypothesis to injury sustained from the systemic effects of CYP.

Saline-injected mice appeared to exhibit modest increases in cell proliferation within the first 2 days after their injections compared to non-injected controls. They also showed some cyclicity in proliferation over the course of the experiment. None of the mice were handled by the experimenters prior to injection, suggesting that the initial increase in cell proliferation in saline-injected mice might have been a reaction to the stress of the injection process (Rogers et al. 1992; Solvason et al. 1992; Lee et al. 2008a). This was more apparent in both layers of the MOE and, to a lesser extent, in the sensory layer of the VNO. Interestingly, the initial effects of CYP appeared to counter the effects of stress from the injection process. The cyclic-like increases and decreases in proliferation in saline mice were also an important finding acquired from multiple samples throughout the 125 days of the experiment. However, the low overall level of cell cycling and the spacing between the time points after day 6 posttreatment, make it difficult to determine if the lows and highs identified in this experiment accurately represent full cycles in the saline mice. The later time points also show increased variability as the animals aged. Even so, similar cycle-like increases and decreases in proliferation, although larger in magnitude, were also identified in the CYP mice. In adult rats and mice, the rate of proliferation varies somewhat over time to meet cellular turnover and changes in the size of the VNO and MOE epithelia throughout adulthood while maintaining neuronal density constant (Farbman 1990; Carr and Farbman 1993; Mahalik 1996; Weiler and Farbman 1997; 1998a; Weiler et al. 1999b). Perhaps, by the age of 2–4 months, such as the mice were at the beginning of this study, cell proliferation underlying renewal of olfactory neurons and sustentacular cells in the MOE and the VNO generates replacement cells in waves rather than in a continuous stream. Injuries to the olfactory epithelium such as that caused by CYP may induce the system to reset the cycle-like renewal process.

In sum, this study found that a single dose of CYP, a chemotherapy drug with alkylating properties, damaged the MOE and VNO after systemic administration. CYP had the greatest effect within the sensory layer of the MOE, interrupting cell proliferation for up to 3 days. Once restarted, cell proliferation remained at relatively high levels for an extended period before eventually returning to control levels. While the magnitude of the effects of CYP were less pronounced, similar patterns of diminished and then heightened cell proliferation were observed in the sustentacular area of the MOE and all 3 regions of the VNO. Most of this cell renewal activity occurred near the basal areas of these epithelia. More research is needed to determine the mechanisms by which CYP alters cell renewal within olfactory epithelia. The findings of this study suggest chemotherapy drugs like CYP can cause significant disruption of the olfactory system that may be the basis for the loss of olfactory function reported during and after chemotherapy. They also offer new questions about how these drugs might affect patient recovery after CYP treatment (Choi and Goldstein 2018).

Funding

This research was supported by a grant from National Institutes of Health, National Institute of Deafness and Other Communicative Disorders 1R01DC012829 awarded to ERD.