The effects of age and sex on the detection of pure tones by adult CBA/CaJ mice (Mus musculus)

Abstract

Age-related hearing loss (ARHL) is a neurodegenerative disorder characterized by a gradual decrease in hearing sensitivity. Previous electrophysiological and behavioral studies have demonstrated that the CBA/CaJ mouse strain is an appropriate model for the late-onset hearing loss found in humans. However, few studies have characterized hearing in these mice behaviorally using longitudinal methodologies. The goal of this research was to utilize a longitudinal design and operant conditioning procedures with positive reinforcement to construct audiograms and temporal integration functions in aging CBA/CaJ mice. In the first experiment, thresholds were collected for 8, 16, 24, 42, and 64 kHz pure tones in 30 male and 35 female CBA/CaJ mice. Similar to humans, mice had higher thresholds for high frequency tones than for low frequency pure tones across the lifespan. Female mice had better hearing ability than males after 645 days of age. In the second experiment, temporal integration functions were constructed for 18 male and 18 female mice for 16 and 64 kHz tones varying in duration. Mice showed an increase in thresholds for tones shorter than 200 ms, reaching peak performance at shorter durations than other rodent species. Overall, CBA/CaJ mice experience ARHL for pure tones of different frequencies and durations, making them a good model for studies on hearing loss. These findings highlight the importance of using a wide range of stimuli and a longitudinal design when comparing presbycusis across different species.

Graphical Abstract

Operant conditioning and a longitudinal design can be used to trace hearing loss for pure tones in CBA/CaJ mice. Male and female mice exhibit similar hearing sensitivity up to 645 days of age. Later in life, male mice lose hearing at a faster rate than females, paralleling hearing loss in humans.

4. Introduction

Age-related hearing loss (ARHL), or presbycusis, is a neurodegenerative disorder characterized by reduced hearing sensitivity, with a progressive loss of abilities to detect high frequencies followed by a decrease in sensitivity to low frequencies (Homans et al., 2016). Some other symptoms of ARHL include deteriorated sensitivity to signals with longer durations and other temporal cues, a decrease in speech comprehension (especially in noisy environments), slowed processing of acoustic signals, and impaired sound localization (Florentine, Fastl, & Buus, 1988; Gates & Mills, 2005; Huang & Tang, 2010; Ozmeral, Eddins, Frisina, & Eddins, 2016; Schuknecht & Gacek, 1993). The symptoms of ARHL are also expressed differently across the sexes, with males becoming progressively less sensitive earlier and at a faster rate than females (Gates & Cooper, 1991; Pearson et al., 1995; Tambs, Hoffman, Borchgrevink, Holmen, & Samuelsen, 2003). According to the National Institute on Deafness and Other Communication Disorders (2016), ARHL is a serious impairment affecting one in three adults between the ages of 65 and 74. The prevalence and severity of ARHL makes it an important focus of research studies. Cross-sectional and longitudinal animal studies play a critical role in understanding the bases of ARHL and for developing diagnostic and treatment strategies through genetic, physiological, and behavioral studies of hearing and hearing loss (Gates & Mills, 2005; Ohlemiller, 2006; Ohlemiller, Jones, & Johnson, 2016).

One commonly used strain of animals for auditory research is the CBA/CaJ mouse, which retains normal auditory sensitivity into middle adulthood (Henry, 2004; Kobrina & Dent, 2016; Prosen, Dore, & May, 2003; Zheng, Ding, Yu, Salvi, & Johnson, 2009), making it an excellent model for hearing loss in humans. Hearing in this mouse strain has been extensively studied using electrophysiological methodologies and cross-sectional experimental designs (Henry, 2004; Ohlemiller, Dahl, & Gagnon, 2010; Zheng, Johnson, & Erway, 1999). Zheng et al. (1999) used the auditory brainstem response (ABR) to trace correlates of hearing loss in CBA/CaJ mice and showed that 11-month-old mice had higher thresholds for clicks and 8, 16, and 32 kHz tones than 9-month-old mice. In a more detailed study, Henry (2004) discovered that male and female mice exhibited differences in hearing by 12 months of age. Male CBA/CaJ mice had higher ABR thresholds at high frequencies than female mice, paralleling sex differences for hearing loss in middle-aged humans. Ohlemiller and colleagues (2010) used compound action potential (CAP) recordings to measure hearing abilities and rates of hearing loss in the CBA/CaJ mouse strain. The authors found that female and male mice had similar patterns of hearing loss up to 24 months of age for 5 and 28.3 kHz tones. Interestingly, female mice over 24 months of age showed accelerated hearing loss for high frequency tones while hearing in male mice deteriorated progressively in a linear trend. This hearing loss pattern is similar to the hearing deterioration in elder humans, with women showing an accelerated rate of hearing loss for high frequencies after the onset of menopause (Hederstierna, Hultcrantz, Collins, & Resenhall, 2010). Although, it is unlikely that menopause alone causes the acceleration of hearing loss in female mice, because the onset of menopause in this mouse strain is between 12 and 14 months of age (Silver, 1995). Ohlemiller and colleagues (2010) proposed that changes in the stria vascularis and neuronal loss in female mice could account for the observed differences in presbycusis between the sexes in CBA/CaJ mice. These differences in findings across methodologies must be reconciled with a studying using a longitudinal design in order to evaluate the utility of the CBA/CaJ mouse as an appropriate model for hearing loss in humans.

In humans, presbycusis is most often studied using behavioral methods, yet behavioral research is scarce in studies of hearing loss in mice. Generally, behavioral audiograms are more sensitive across frequencies than electrophysiological measures of hearing (Heffner & Heffner, 2001; Heffner, Koay, & Heffner, 2008; Radziwon et al., 2009). The Go/No-go operant conditioning procedure is a functional equivalent of the Yes-No paradigm of hearing assessments in humans, and can be used to draw parallels between mouse and human hearing. Using this method, Radziwon et al. (2009) discovered that the adult CBA/CaJ mouse showed peak sensitivity in the 8 to 24 kHz range, with higher thresholds (i.e., lower sensitivity) for tones outside of this range. Kobrina and Dent (2016) used these methods, a cross-sectional, and a longitudinal experimental design to trace hearing loss for ultrasonic vocalizations (USVs) and 42 kHz pure tones in CBA/CaJ mice. This mouse strain can behaviorally detect USVs and pure tones up to 1000 days of age, indicating that the electrophysiological studies mentioned above do not accurately depict the mouse’s auditory acuity across the entire lifespan. Mice progressively lost hearing for USVs and 42 kHz pure tones, with males losing hearing at a significantly faster rate than females for USVs, but not 42 kHz tones. The lack of sex differences between older male and female mice detecting 42 kHz tones contradicts previous electrophysiological findings (Henry, 2004; Ohlemiller et al., 2010). Thus, a longitudinal design and operant conditioning methodologies in mice can and should be used to reconcile discrepancies between electrophysiological and behavioral research.

The basic measure of auditory acuity is a pure tone audiogram, which has been collected for many strains of mice (reviewed in Dent, Screven, & Kobrina, 2018). However, there are no full behavioral audiograms comparing hearing across the lifespan of the CBA/CaJ mouse. In the first experiment, we used a longitudinal design and operant conditioning procedures to measure audiograms in CBA/CaJ mice across their lifespan. We hypothesized that, similarly to humans, mice would lose hearing gradually, starting with high frequencies and progressing to lower frequencies. We also predicted that male mice would show earlier signs of ARHL (e.g., increasing thresholds at a younger age) than female mice.

In addition, we measured temporal integration functions for mice at various points in their lives. The previously reported audiograms in mice were measured using tones of differing durations across studies, making comparisons problematic (Ehret, 1974; 1976a; 1976b; Prosen et al., 2003; Radziwon et al., 2009). Longer stimuli are easier to detect than shorter stimuli, and longer tones often lead to lower absolute thresholds in humans (Watson & Gengel, 1969) and rodents (Ehret, 1976b; Gleich, Kittel, Klump, & Strutz, 2007; Henderson, 1969). Humans and other animals benefit from longer tone durations as pure tones increase in length up to 500 ms (Ehret, 1976b; Fay, 1988). This effect is known as temporal summation or temporal integration. Humans and other animals become less sensitive to duration benefits with increases in age and hearing loss (Carlyon & Sloan, 1987; Florentine et al., 1988; Gleich et al., 2007; Meddis & Lecluyse, 2011; Solecki & Gerken, 1990). Temporal integration has not yet been measured in the CBA/CaJ mouse, and the effects of AHRL on temporal integration functions in CBA/CaJ mice are unknown. Thus, in the second experiment, we collected temporal integration functions for aging CBA/CaJ mice. We hypothesized that CBA/CaJ mice would benefit from an increase in pure tone duration, and that this benefit would decrease with age.

5. Methods

5.1. Subjects

Mice of the CBA/CaJ strain were used in these experiments. Adults ranging from 2-3 months through 3 years of age were studied. The original breeding pairs were acquired from The Jackson Laboratory. Our subjects were bred at the University at Buffalo, SUNY. Sixty-five adult (30 M, 35 F) CBA/CaJ mice were used for experiment 1 and 36 (18 M, 18 F) CBA/CaJ mice were used for experiment 2. Sixteen (8 M, 8 F) out of 36 mice in experiment 2 also participated in experiment 1. No training effects were observed in these mice. Fifty-seven of the experimental animals were trained for these experiments exclusively, while twenty eight mice (15 M, 13 F) were obtained after participating in other behavioral studies in our laboratory. The number of animal subjects was based on previous studies that used operant conditioning methods to measure hearing in aging mice (Kobrina & Dent, 2016).

The mice were individually housed in a colony room at the University at Buffalo and kept on a reversed day/night cycle. Fifty-five subjects (25 M, 30 F) began training and testing when they were two to three months old. Thirty mice (15 M, 15 F) were trained between 6 and 12 month of age. During testing, the mice were water restricted to approximately 90% of their free-watering weights. The animals had unrestricted access to food, except while participating in the experiment. All procedures were approved by the University at Buffalo, SUNY’s Institutional Animal Care and Use Committee, and complied with the ARRIVE guidelines and the associated NIH guide for the care and use of laboratory animals.

5.2. Apparatus

The mice were tested in a wire cage (23 × 39 × 15.5 cm, see Figure 1) placed in a sound-attenuated chamber (53.5 × 54.5 × 57 cm) lined with 4-cm thick Sonex sound attenuating foam (Illbruck Inc., Minneapolis, MN). The chamber was illuminated at all times by a small lamp with an 8-W white light bulb and the behavior of the animals during test sessions was monitored by an overhead web camera (Logitech QuickCam Pro, Model 4000). The test cage contained an electrostatic speaker (Tucker-Davis Technologies (TDT), Gainesville, FL, Model ES1), a dipper (Med Associated Model ENV-302M-UP), and two nose poke holes surrounded by infrared sensors (Med Associates Model ENV-254).

Figure 1.

Schematic of the experimental setup used in both experiments. Mice began a trial by nose poking to the observation hole. When they detected a stimulus presented from the speaker, they poked to the report hole. If correct, the mice received Ensure® from the dipper.

The experiments were controlled by Dell Optiplex 580 computers operating TDT modules and software. Stimuli were sent through an RP2 signal processor, a PA5 programmable attenuator, an ED1 electrostatic speaker driver, and finally to the speaker. Inputs to and outputs from the testing cage were controlled via RP2 and RX6 processors. Power supplies were used to drive the dipper (Elenco Precision, Wheeling, IL, Model XP-603) and infrared sensors (Elenco Precision, Model XP-650). Custom Matlab and TDT RPvds software programs were used to control the hardware.

5.3. Stimuli

The stimuli for experiment 1 were five pure tones: 8, 16, 24, 42 and 64 kHz (500 ms in duration with 5 ms rise/fall cosine ramps). These stimuli and durations were chosen in order to span the hearing range of the mouse, and to provide a reliable comparison with previous studies in this strain (Kobrina & Dent, 2016; Kobrina, Toal, & Dent, 2018; Radziwon & Dent, 2014). A random order of five testing conditions was generated for each mouse. If the mouse completed testing on all stimuli, a new random order was generated and they were tested again. The stimuli for experiment 2 were pure tones (16 and 64 kHz) of five durations (20, 50, 200, 400, and 800 ms, with 10% rise/fall cosine ramps). The stimuli were chosen to cover peak behavioral sensitivity (Radziwon et al., 2009) and the mouse’s vocalization range (Burke, Screven, & Dent, 2018; Portfors, 2007). A random order of testing on these 10 conditions was generated for each subject. If the mouse completed testing on all stimuli, a new random order was generated and they were tested again.

The stimuli for these experiments were created and edited in Adobe Audition (v.5). Sound pressure levels for stimuli were calibrated using an ultrasound recording system (Avisoft, Model USG 116-200) and a custom Matlab script, with the microphone (Avisoft Bioacoustics Ultra Sound Gate CM116) placed at the approximate location where the animal’s head would be during testing. Calibrations were conducted weekly.

5.4. Procedure

Mice were trained using a Go/No-go operant conditioning procedure on a detection task. They were trained by shaping to poke to the left observation nose poke hole twice to initiate the trial. Once initiated, mice had to wait for the stimulus, and then poke to the right report poke hole once when the stimulus was detected. The first stage in the training process was to shape the mice to nose poke to the observation hole and then approach the dipper for the chocolate Ensure® reinforcement. Ensure® is a type of nutritional supplement with a consistency like chocolate milk, and it was used in an attempt to maximize the number of trials the mice would complete. The mice were then trained to repeatedly poke to the observation hole until they heard a pure tone, after which they would nose poke to the report hole for the reinforcement. The training stimulus varied in frequency and duration across mice. Next, catch trials were phased into the training and the variable waiting interval was systematically increased.

During testing, the mouse began a trial by nose poking through the observation nose-poke hole, which initiated a variable waiting interval ranging from 1 to 4 s. After the waiting interval, a single test stimulus was presented. This constitutes the “Go” portion of the “Go/No-go” procedure. If the mouse detected the stimulus, it was required to nose poke through the report nose-poke hole within 2 s of the onset of the target. In this trial type, a “hit” was recorded if the mouse correctly responded within the response window and the animal received 0.01 ml of Ensure® as reinforcement. A “miss” was recorded if the mouse failed to nose poke through the report nose-poke hole during the waiting interval. When a “miss” occurred, the trial was aborted, and the mouse could initiate the next trial right away. Sessions with hit rates of at least 80% were included in data analysis. Animals obtained a hit rate of less than 80% approximately 1% of the time, thus very few sessions were excluded for this reason.

Thirty percent of all trials were “No-go” or sham trials. No stimulus was presented during the sham trials. These trials were required to measure the false alarm rate and calculate the animal’s response bias. If the subject nose poked to the report hole during a sham trial, a “false alarm’ was recorded and the mouse was punished with a 3 to 5 s timeout interval. However, if the mouse continued to nose poke to the observation hole, a “correct rejection” was recorded and the next trial would begin immediately. In either case, no reinforcement was given. Chance performance was represented by the animal’s false alarm rate. Testing sessions with false alarm rates greater than 20% were excluded from final data analysis, and approximately 3% of all sessions were discarded for this reason. False alarm rates ranged from 0 to 19%. Mice typically completed between 50 and 400 trials per testing session. A pure tone of only one frequency (experiment 1) or one frequency and duration (experiment 2) was tested per session, with only stimulus amplitude varying from trial to trial. Within a session, all of the stimuli were presented to the subjects in randomized blocks of 10 (7 targets (“Go”) and 3 shams (“No-go”) per block). The pure tone stimuli were presented to mice according to the Method of Constant Stimuli (MOCS). Once performance stabilized for very loud sounds, 2 out of the 7 “Go” targets were attenuated in steps of 5 dB for animals under 800 days old (d.o. hereafter), or 10 dB for 800 d.o. animals and older, until performance consistently dropped below 50% for one of those quieter stimuli. Four hundred trials were then collected and thresholds were calculated from the final 200 trials. Once a threshold was calculated for that stimulus, subjects were moved on to the next stimulus type. Data collection continued throughout the subjects’ lifespan and was terminated when mice responded less than 50% of time to the loudest target of any frequency, or upon deterioration of subjects’ health conditions.

5.5. Data analysis

Signal detection analysis was performed to factor out the animals’ motivational biases as bias is independent of sensitivity (Steckler, 2001). Mean hit and false alarm rates were used to calculate thresholds using signal detection theory with a threshold criterion of d’= 1.5. We chose this conservative d’ value in order to compare these results with other findings for this and other mouse strains (e.g., CBA/CaJ mice, Kobrina & Dent, 2016; Radziwon et al., 2009; NMRI mice, Klink, Bendig, & Klump, 2006) and due to its correspondence to a low false alarm rate. The relationship between rates of false alarm and age was examined using regression analysis.

5.5.1. Data analysis experiment 1

Given the longitudinal design of this study, this data set contained an unequal number of observations from every mouse due to differences in lifespan and productivity. In other words, every stimulus category contained a different number of observations from male and female mice, with some mice contributing multiple points per stimulus, while others were only able to complete one set of conditions during their lifespan.

Exploratory regression analyses of the data were performed in order to determine which function for the data fit best explained changes in hearing in male and female mice across the lifespan. Ohlemiller et al. (2010) showed that male mice lost hearing progressively across the lifespan in a linear pattern, while females had a polynomial pattern of hearing loss with greatest degree of deterioration later. For this experiment, second order polynomial functions explained the greatest amount of variability in the threshold data in both sexes of CBA/CaJ mice (threshold (dB) = y0 + a*x + b*x², see Supplementary table 1 for derived equations). We used a linear mixed-effects model to examine whether age (x) and age² (x²) predict hearing loss in male and female mice across stimuli (LMM, lmer in the lme4 R package) (Bates, Bolker, & Walker, 2015; R Core Team, 2016). In this model, we examined if thresholds (dB SPL) were affected by fixed factors of age and age² (continuous), sex (male or female), frequency (8, 16, 24, 42, and 64 kHz), and by interactions between age² and sex, as well as between frequency and age. To control for dependencies within our data from sampling each mouse repeatedly we included a random intercept for mouse identity across age. Post-hoc comparisons using Tukey’s method were performed to assess significance in the relationship between age, frequency, and sex (emmeans R package).

We also examined the consistency of an individual’s detection performance across the lifespan. For this purpose, repeatability associated with the 95% confidence intervals (CI) was calculated for the random effect of individual with the mixed effect models. This was done separately for male and female mice. Uncertainty estimates were generated by bootstrapping the data 1000 times with the R package rptR (Stoffel, Nakagawa, & Scheilzeth, 2017).

5.5.2. Data analysis experiment 2

Similar to experiment 1, data for this experiment also contained an unequal number of observations from every animal across the lifespan. In addition, the number of data points varied across age and durations. Data were sorted by frequency (16 and 64 kHz), duration (20, 50, 200, 400, and 800 ms), and sex (male and female). Exploratory regression analyses of the data were performed in order to determine which function for the data fit best explained changes in hearing in male and female mice across the lifespan. Second order polynomial functions explained greatest amount of variability in the threshold data across the lifespan for male mice (thresholds (dB) = y0 + a*x + b*x²), while both linear (threshold (dB) = y0 + a*x) and polynomial functions explained variability in threshold data across the lifespan for females (see Supplementary table 2 for derived equations). Data from male and female mice were analyzed separately to examine whether differences in trends of hearing loss were significant (LMM, lmer in the lme4 R package) (Bates, Bolker, & Walker, 2015; R Core Team, 2016). A mixed-effect model was used to assess whether thresholds (dB SPL) were affected by fixed factors of age (continuous), age² (continuous), frequency (16 and 64 kHz), and duration (20, 50, 200, 400, and 800 ms) for male and female mice separately. To control for dependencies within our data from sampling each mouse repeatedly we included mouse identity across age as a random effect. Post-hoc comparisons using Tukey’s method were performed to assess significance in the relationship between age, frequency, and duration for male and female mice (emmeans R package). Similar to experiment 1, repeatability was calculated.

In order to construct average temporal integration functions for adult CBA/CaJ mice, we calculated average thresholds and standard deviations for all stimuli in male and female mice. In order to examine the effect of signal duration on thresholds we calculated threshold shifts across conditions relative to the longest duration signal (800 ms). Slopes $(m=\frac{\Delta y(dB)}{\Delta x(ms)})$ were computed for each function in order to examine the degree of benefit that mice experience from the increase in duration. The greatest degree of change in thresholds (Δy (dB)) for 16 and 64 kHz occurred between 20 and 200 ms (Δx (ms)).

6. Results

6.1. Results experiment 1

Male and female mice were able to detect pure tones of all frequencies well into old age (Figure 2 a-e, Table 1). Hearing sensitivity significantly decreased with age across frequencies in mice, but rates of false alarming did not change across the lifespan. Mice also lost hearing for high frequency pure tones differently than for low frequency tones. The model revealed main effects of age, age², sex, and frequency (Table 2). Mice were more sensitive to 16 kHz pure tones than to tones of other frequencies (p < 0.000) and less sensitive to 64 kHz tones than to 8, 16, and 24 kHz tones (p < 0.000). There were no significant differences between detection thresholds for 8 kHz and 24 kHz (p = 0.809), for 8 kHz and 42 kHz (p = 0.241), and for 24 kHz and 42 kHz tones (p = 0.783). In line with our hypothesis, hearing loss progressed differently across the lifespan for 42 and 64 kHz pure tones than for other frequencies as revealed by the significant interaction between frequency and age (Table 2). In general, hearing loss onset for 8, 16, and 42 kHz was between 500 and 600 d.o. (p = 0.001), while for 24 and 64 kHz tones the onset was approximately between 400 and 500 d.o. (p < 0.000). The model also revealed a significant interaction between sex and age (Table 2). Male mice had significantly higher thresholds across frequencies than females after 645 d.o. (t = −1.99, p = 0.049) and up to 1000 d.o. (t = −2.18, p = 0.030) (Figure 3, a-b).

Figure 2.

Second order polynomial regression plots for 8, 16, 24, 42, and 64 kHz tones (a-e). Each plot depicts thresholds from multiple mice across their lifespans (males = filled blue symbols, females = open symbols). Lines represent the best data fits of hearing across the lifespan for 8–64 kHz (a-e). The amount of variability in the data explained by aging is expressed in the form of r² for males and females separately.

Table 1.

Descriptive statistics for male and female mice in Experiment 1¹.

	Males				Females
Stimulus	Age (days)	Range (dB)	N	N (obs.)	Age (days)	Range (dB)	N	N (obs.)
8 kHz	187–958	2 - 65	19	27	210–957	3 – 55	26	37
16 kHz	186–992	1 – 59	20	28	193–1004	−3 – 63	26	39
24 kHz	206–1007	9 – 76	19	30	185–1002	3 – 66	24	33
42 kHz	171–969	6 – 84	17	28	214–952	5 – 72	24	35
64 kHz	166–898	24 – 102	21	34	200–871	31 – 89	28	44

¹N = number of subjects, N (obs.) = number of observations.

Table 2.

Mixed-effects model analysis and significance testing comparing hearing loss in male (N = 30, N_observations=188) and female (N = 35, N_observations =147) mice across the lifespan for 500 ms pure tones of different frequencies¹.

Fixed effects	B	SE	t-value	p
Intercept (Id)	10.45	2.36	4.42	<0.000
Age	0.03	0.00	7.90	<0.000
Poly (age)	69.07	13.95	4.95	<0.000
Sex	3.24	2.71	2.07	0.044
Frequency (8 kHz)	−3.78	2.65	−1.43	0.152
Frequency (24 kHz)	−3.99	2.57	−1.56	0.121
Frequency (42 kHz)	2.77	2.71	1.02	0.307
Frequency (64 kHz)	18.47	2.52	7.33	<0.000
Sex*poly(age) 1st order	52.30	21.73	2.41	0.017
Sex*poly(age) 2nd order	8.52	19.39	0.44	0.661
Frequency (8 kHz) * age	0.00	0.00	−0.61	0.546
Frequency (24 kHz) * age	0.00	0.00	0.03	0.973
Frequency (42 kHz) * age	0.00	0.00	−2.00	0.046
Frequency (64 kHz) * age	0.02	0.00	−3.51	0.00
Random effects	σ2
Mouse Id	4.59
Residual	8.24

¹Significant values are bolded. The LMM formula in R was lmer (threshold ~ sex * poly(age, 2) + age * frequency + (1∣ Id)). B = model estimate, SE = standard error, σ² = standard deviation. Fixed effects for sex are compared to females, frequencies are compared to 16 kHz, for sex*poly (age) for 1^st order and 2^nd order interactions are compared to sex (female)* poly(age) for 1^st order and 2^nd, and for frequency*age to frequency (16 kHz) * age. Poly(age) and age² represent the same value.

Figure 3.

Mean audiograms with standard error bars as predicted by the model for male (a) and female (b) subjects under 300 d.o. (300 = grey circles), 301–450 d.o. (450 = red squares), and 451–645 d.o. (645 = pink stars), 645–750 d.o. (750 = green hexagons), 751–900 d.o. (900 = blue diamonds), and 901–1000 d.o. (1000 = black circles). Hearing between male and female mice significantly diverged around 645 d.o..

Individual differences between male and female mice accounted for a significant, but small amount of variability in hearing loss. Repeatability measures suggested that 15% of variability in our hearing loss data set was associated with individual differences between male mice over their lifespan (p = 0.006, CI: 0.000 to 0.361). For females, 36% of variability was associated with individual differences between female mice (p < 0.000, CI: 0.198 to 0.528), suggesting that female subjects in this experiment were more variable than males.

6.2. Results experiment 2

Male and female mice were able to detect 16 and 64 kHz pure tones of all durations well into old age (Table 3, Figure 4). Male and female mice lost hearing differently, indicating that there may be a sex-based trend to hearing loss (Ohlemiller et al., 2010). Hearing loss in male mice was best explained by a polynomial data fit (χ² = 5.68, p = 0.017), but no significant differences were detected between linear and polynomial fits in female data (χ² = 0.07, p = 0.789) (Figure 4 a-e). The models revealed main effects of age, age², frequency, and duration for both male and female mice (Table 4). Male and female mice lost hearing across frequencies and durations with age. Detection thresholds were lower for 16 kHz than for 64 kHz for both male and female mice (p < 0.000). Interestingly, the onset of hearing loss across frequencies and durations for male mice was between 400 and 500 d.o. (t = −3.18, p = 0.015), and between 300 and 400 d.o. for female mice (t = −3.32, p = 0.001).

Table 3.

Descriptive statistics for male and female mice in Experiment 2¹.

		Males				Females
Stimulus (kHz)	Duration (ms)	Age (days)	Range (dB)	N	N (obs.)	Age (days)	Range (dB)	N	N (obs.)
16	20	270–917	11 – 48	6	17	240–890	10 – 51	8	20
16	50	242–930	−3 – 23	8	15	231–910	−6 – 22	9	20
16	200	187–955	−4 – 52	10	15	229–940	−2 – 27	9	16
16	400	284–958	1 – 48	13	19	193–1002	−3 – 64	13	20
16	800	205–909	3 – 53	9	18	247–911	−4 – 25	8	17
64	20	215–831	20 – 71	7	20	260–850	33 – 67	7	20
64	50	269–870	24 – 61	8	20	228–888	27 – 64	8	20
64	200	141–900	18 – 96	6	20	177–851	28 – 69	8	19
64	400	221–898	24 – 101	8	19	225–860	36 – 89	14	20
64	800	299–887	29 – 71	7	16	194–873	42 – 68	8	20

¹N = number of subjects, N (obs.) = number of observations.

Figure 4.

Second order polynomial regression plots for 20, 50, 200, 400, and 800 ms, 16 and 64 kHz tones (a-e). Each plot depicts thresholds from multiple mice across their lifespans (males = filled blue symbols, females = open symbols). Lines represent the best data fit of hearing across the lifespan for 16 kHz (solid lines, blue = males, black = females) and 64 kHz (dashed lines, blue = male, black = females). The amount of variability in the data explained by aging is expressed in the form of r² for each frequency and for males and females separately.

Table 4.

Mixed-effects model analysis and significance testing across the lifespan, comparing the fit of hearing loss trends in male (N = 18, N_observations =179) and female (N = 18, N_observations = 192) mice for 16 and 64 kHz pure tones of different durations.¹

Sex	Fixed effects	B	SE	t-value	p
Male	Intercept (Id)	17.73	2.06	8.62	<0.000
	Poly(age) linear	133.64	11.17	11.96	<0.000
	Poly(age) 2nd order	76.32	10.71	7.13	<0.000
	Frequency (64 kHz)	30.71	1.55	19.84	<0.000
	Duration (50 ms)	−5.50	2.42	−2.27	0.024
	Duration (200 ms)	−3.80	2.39	−1.59	0.114
	Duration (400 ms)	−2.33	2.37	−0.99	0.326
	Duration (800 ms)	−1.58	2.39	−0.66	0.510
	Random effects	σ2
	Mouse Id	2.84
	Residual	9.98
Female	Intercept (Id)	23.73	1.64	14.46	<0.000
	Poly(age) linear	111.88	9.47	11.82	<0.000
	Poly(age) 2nd order	23.13	8.96	2.58	0.011
	Frequency (64 kHz)	36.80	1.26	29.31	<0.000
	Duration (50 ms)	−13.30	1.93	−6.88	<0.000
	Duration (200 ms)	−11.00	2.02	−5.46	<0.000
	Duration (400 ms)	−7.61	1.95	−3.90	<0.000
	Duration (800 ms)	−10.25	1.95	−5.27	<0.000
	Random effects	σ2
	Mouse Id	1.99
	Residual	8.43

¹Significant values are bolded. The LMM formula in R was lmer (threshold ~ poly(age linear, polynomial) + frequency + duration + (1∣Id∣)). B = model estimate, SE = standard error, σ² = standard deviation. Fixed effects for frequency are compared to 16 kHz, for duration to 20 ms.

Individual difference between male mice accounted for a significant, but small amount of variability in hearing loss. In females, individual variability was not associated with hearing loss across stimuli. Repeatability measures suggested that 8% of variability in our hearing loss data set was associated with individual differences between male mice over their lifespan (p = 0.014, CI: 0.000 to 0.224). For females, individual differences were not associated with changes in hearing loss across the lifespan (p = 0.182, CI: 0.000 to 0.181).

Overall, CBA/CaJ mice differed in how they detected shorter 16 and 64 kHz tones. This can be seen in the steeper temporal summation functions for 16 kHz than for 64 kHz tones (Figure 5 a). Female and male mice were less sensitive to 16 kHz 20 ms tones than to 800 ms tones. At 64 kHz, female mice were less sensitive to 20 ms than to 800 ms tones. Male mice, however, showed similar sensitivity for 20 and 800 ms tones at 64 kHz. On average, the greatest degree of change for 16 and 64 kHz tones occurred between 20 and 200 ms (Figure 5 b). Male mice benefited slightly less from the increase of 16 kHz stimuli from 20 ms to 200 ms (m = −10.85 dB/ms; improvement of 16.59 dB/dec) than females (m = −8.43 dB/ms; improvement of 21.36 dB/dec). Male mice also benefited less from the increase of 64 kHz tones from 20 to 200 ms (m = −108.64 dB/ms; improvement of 1.66 dB/dec) than female mice (m = −46.06 dB/ms; improvement of 3.91 dB/dec).

Figure 5.

Mean temporal summation functions with standard deviation error bars (a) and average threshold shifts relative to the longest duration signal (dashed line) as a function of signal duration (b) for male (blue symbols, blue line) and female (open symbols, black lines) mice. These data are potted in comparison to mean temporal summation functions and threshold shifts for 15 kHz (green triangles, green line) and 60 kHz (purple triangles, purple line) for young male NMRI mice retrieved from Ehret (1976b). In CBA/CaJ mice, the greatest change in threshold shifts occurs between 20 and 200 ms (dotted line).

7. Discussion

The goal of this set of experiments was to construct behavioral audiograms and temporal integration functions in aging CBA/CaJ mice using operant conditioning procedures. Our secondary goal was to evaluate the CBA/CaJ mouse as a behavioral animal model for presbycusis in humans. This is the first study, to our knowledge, to employ a longitudinal design and behavioral methodology to measure age-related hearing loss for pure tones of different frequencies and durations. CBA/CaJ mice of all ages had the most sensitive hearing for 16 kHz pure tones, with higher thresholds in the lower and higher frequency ranges. The mice in this experiment experienced an onset of hearing loss for 500 ms tones by the middle of their lifespan. Similar to high-frequency hearing loss in humans (Ahmed et al., 2001; Shim et al., 2009), hearing loss for 42 and 64 kHz progressively declined in mice across their lifespan. The sexual dimorphism seen in presbycusis in humans emerged in mice in the last third of their lifespan, with male mice exhibiting higher thresholds than females across all tested frequencies. The hearing loss from our mice found using behavioral methods was less than that found by Li and Borg (1991) and Ohlemiller et al. (2010) using physiological methods, highlighting the importance of testing hearing in awake and behaving animals in order to make the best possible comparisons to humans. These behavioral results show that operant conditioning procedures can be used to draw parallels between mouse and human presbycusis.

Young CBA/CaJ mice had similar hearing sensitivity compared to other mouse strains (reviewed by Dent et al., 2018). In general, CBA/CaJ mice retained their hearing later into their lifespan and showed overall better hearing sensitivity than NMRI mice (Ehret, 1974), CBA/J mice (Ohlemiller et al., 2010), and C57B/6J mice (Prosen et al., 2003), making this mouse strain a suitable model for ARHL in humans. It is important to note, however, that in these experiments age-related hearing loss was measured without the confounds of noise exposures and other environmental factors that affect humans. Consequently, the hearing loss that male and female mice experienced in this study should be examined within the boundaries of a more “pure” form of presbycusis.

Temporal processing has been shown to degrade with hearing loss both in humans (Florentine et al., 1988) and in rodents (Gleich et al., 2007). CBA/CaJ mice showed a decrease in duration benefits, exhibiting an earlier onset of deafness for shorter durations (20 and 50 ms). Adult CBA/CaJ mice had higher detection thresholds for short 16 kHz pure tones than for longer stimuli, although this pattern did not hold up for 64 kHz tones. CBA/CaJ mice experience benefits from increasing duration up to 200 ms for 16 kHz tones, plateauing out at shorter durations than other rodents (reviewed by Dent et al., 2018; Fay, 1988) and humans (Florentine et al., 1988; Meddis & Lecluyse, 2011). At a higher frequency, however, temporal integration functions from females were generally flatter, showing only a slight threshold decrease with increasing tone durations, though this was not true for male mice. NMRI mice also showed a greater benefit (lower thresholds) for duration increases at lower frequencies than for high frequencies (Ehret, 1976b). Together, this experiment suggests that sex, signal duration, and stimulus frequency likely all have an effect on ARHL expression in CBA/CaJ mice.

In contrast to experiment 1, female mice in experiment 2 progressively lost hearing across durations and frequencies while males showed a polynomial hearing loss trend that began after 400 d.o.. While we do not know why these asymmetrical differences emerged, we hypothesize that duration may interact with other acoustic stimulus parameters in the processing of simple and behaviorally relevant signals. Previous electrophysiological studies showed that chinchillas and mice have duration-tuned neurons that may be used for the detection and discrimination of natural vocalizations (Brand et al., 2000; Chen, 1998; Holmstrom et al., 2000). These neurons can alter their activity and sometimes abolish duration tuning characteristics based on other acoustic qualities (e.g., intensity and frequency) of signals (Brand et al., 2000). As a result, it is possible that the role of stimulus duration changes with age, sex, and context. The asymmetrical changes across age and sex that we observed are not unique to the stimuli tested here, as it has been demonstrated that female mice benefit from an increase in USV duration (Kobrina & Dent, 2016). In the future, the interaction of duration and other acoustic parameters should be examined in both male and female mice.

Taken together, the results from experiments 1 and 2 suggest that CBA/CaJ mice can be used to model presbycusis in humans. Similar to humans, this mouse strain retains their normal hearing ability late into their lifespan. Humans and CBA/CaJ mice lose hearing at faster rates later in life (Pearson et al., 1995). Mice and humans also have a similar pattern of hearing loss for high frequencies, showing progressive linear ARHL for pure tones in the upper frequencies of their respective audiograms (Ahmed et al., 2001; Shim et al., 2009). We also discovered that CBA/CaJ mice show a sexual dimorphism previously identified in human presbycusis, with male mice exhibiting diminished hearing abilities compared to females later in life (Pearson et al., 1995).

Human presbycusis has been characterized based on the origin of the degeneration (Gates & Mills, 2005; Schuknecht, 1974; Schuknecht & Gacek, 1993). The sensory type of ARHL is characterized by outer hair cell (OHC) loss and neural degeneration of afferent neurons, metabolic ARHL is related to strial atrophy, and mechanical AHRL is caused by cochlear stiffness. These degenerations can occur independently or in combination, thus affecting diagnosis and treatment in humans (Schuknecht & Gacek, 1993). A mixed presbycusis category has been added to describe human presbycusis and animal models with a variety of symptoms. Sergeyenko, Lall, Liberman, & Kujawa (2013) showed that cochlear synaptic loss precedes outer hair-cell loss in aging CBA/CaJ mice. Other electrophysiological studies showed that CBA/CaJ mice exhibit a significant loss of strial cells, sulcus cells in the spiral ligament, and slow OHC loss compared to other mouse strains, suggesting that CBA/CaJ mice may be a good model for metabolic ARHL (Ohlemiller 2009; Ohlemiller et al., 2010). In humans, metabolic ARHL is associated with slowly progressing hearing loss and a flattening of the audiogram (Huang & Tang, 2010). Behaviorally, CBA/CaJ mice showed slow progressive hearing loss only for high frequencies, with rapid hearing loss in the last third of their lifespan. The audiograms did not flatten for these mice, suggesting that this mouse strain likely falls into the mixed presbycusis category. In the future, this mouse strain should be used to examine and model a “mixed” form of presbycusis, which combines age-and noise-induced hearing loss.

There are some notable limitations to the CBA/CaJ mouse model of ARHL. In contrast to humans, mice did not show sex differences in rates of hearing loss for longer stimuli. One possible explanation is that this is a species-specific difference. However, Homans and colleagues (2017) recently reported that the rates of hearing loss in human men and women are more similar than previously established (Pearson et al., 1995). This new finding may be correlated with changes in lifestyle and environment factors in the general population. Another difference between mice and humans is that CBA/CaJ mice developed pronounced sex differences across frequencies in the last third of their lives. On average, human men have lower sensitivity for high frequencies compared to women by 30 years of age, which is approximately the first third of the human lifespan (Pearson et al., 1995). This pattern of hearing loss is reversed in post-menopausal women (Hederstierna et al., 2010), with differences in hearing loss rates disappearing by 80 years of age in the elder population (Wattamwar et al., 2017). Unlike our mice, humans experience a variety of environmental factors that likely lead to a combination of age and noise-related hearing loss. CBA/CaJ mice are highly susceptible to noise induced hearing loss showing progressive loss of cochlear synapses and nerve degeneration after noise exposure (Kujawa & Liberman, 2009). In the future, presbycusis in this mouse strain should be examined in similar environmental conditions.

Lastly, temporal integration functions in the CBA/CaJ mice were shallower than in humans and other mammals (Fay, 1988; Florentine et al., 1988). We suggest utilizing a different behavioral task in order to draw parallels in temporal processing abilities between mice and humans. Radziwon and colleagues (2009) showed that CBA/CaJ mice have gap-detection thresholds for broadband white noise that are comparable to those found in other rodents (reviewed in Dent et al., 2018) and humans (Allen et al., 2002). In humans, gap-detection thresholds increase with age (Pichora-Fuller et al., 2006). Future presbycusis studies should examine whether gap-detection thresholds differ between male and female mice across the lifespan.

8. Conclusion

The data presented here show that operant conditioning procedures can be used to evaluate age-related hearing loss in animals. CBA/CaJ mice can detect pure tones of variable durations late into their lifespan. This mouse strain also developed sex differences in hearing loss, with male mice having worse sensitivity than females, making them an excellent model for mixed presbycusis in humans. Temporal integration functions in this mouse strain differ from humans and other rodents, suggesting that CBA/CaJ mice experience a different magnitude of benefits from increases in stimulus duration. These findings highlight the importance of using a behavioral methodology and a longitudinal design when comparing presbycusis across different species.

3. Significance statement.

Presbycusis is one of the most common health conditions affecting older humans. Mouse studies play critical roles in understanding the genetic, physiological, and behavioral markers of presbycusis. Operant conditioning procedures were used in the present study to track the detection of pure tones by mice across their lifespan. Similar to humans, mice developed sex differences as they aged, with male mice having lower sensitivity and more rapid rates of hearing loss than females starting in the middle adulthood. These results suggest that mice can be used as a model for “pure” form of presbycusis.