The Acute Effects of Age and Particulate Matter Exposure on Heart Rate and Heart Rate Variability in Mice

Abstract

Exposure to ambient particulate matter (PM) is associated with increased cardiac morbidity and mortality with the elderly considered to be the most susceptible. The purpose of this study was to determine if exposure to PM would cause a greater impact on heart regulation in older DBA/2 (D2) male mice as determined by changes in heart rate (HR) and heart rate variability (HRV). D2 mice at the ages of 4, 12, and 19 months were instilled with 100 μg of PM or saline by aspiration. Before and after the aspiration, 3-min echocardiogram (ECG) samples for HR and HRV were recorded at 15-min intervals for 3 h along with corresponding measurements of homeostasis, such as temperature, metabolism, and ventilation. PM exposure resulted in an increase in HRV, declines in HR, and altered measures of homeostasis for a subset of the 12-mo mice. The PM aspiration did not affect cardiac or homeostasis parameters in the 4- or 19-mo mice. Our results suggest that a select group of middle-age mice are more susceptible to alterations in their heart rhythm after PM exposure and highlight that there are acute age-related differences in heart rhythm following PM exposure.

Introduction

Exposure to ambient particulate matter (PM) is associated with increased cardiovascular morbidity and mortality, especially in older individuals [1, 2]. Altered heart rhythm is one manifestation of cardiovascular disease that has been associated with an increase in exposure to ambient PM [2, 3]. This association occurs within hours to days following PM exposure, especially in susceptible individuals such as the elderly [4].

A change in autonomic input to the heart is one of several proposed mechanisms regarding how PM deposition in the lung can impact changes in heart rhythm leading to cardiovascular mortality [4]. It is well-established that measurements of heart rate (HR) and heart rate variability (HRV) are two methods for assessing changes in heart rhythm [5]. Two commonly used measures of HRV are the standard deviation of normal-to-normal intervals (SDNN) and the square root of mean squared differences between successive normal-to-normal intervals (rMSSD). Also, a decline in HRV, specifically SDNN, is considered a predictor for mortality in older individuals [6, 7].

The elderly are considered one of the highest at-risk groups for increased morbidity and mortality to acute PM exposure [8]. Specifically for heart rhythm, studies have shown declines in HRV for the elderly population after exposure to PM [9, 10]. However, not all studies in the elderly have shown a decreased HRV after acute PM exposure. For example, exposure to PM in an elderly population resulted in no significant change in HRV in either the cardiovascular or non-cardiovascular groups [11]. In summary, while the overall results are inconsistent, a majority of the findings indicate that HRV likely decreases with age in adults following acute PM exposure. Similar inconsistent age-related effects are also seen in animal studies [12–14].

The purpose of the current study was to examine the age-dependent differences in HR and HRV responses in healthy mice immediately following an acute exposure to urban ambient PM collected in Baltimore, MD. Since susceptibility to PM exposure generally increases with age in adults, we hypothesized that an age-dependent exposure interaction exists resulting in a greater impact of acute PM exposure on HR and HRV in older mice. The three age groups used in this study closely approximate early adulthood, mid-life, and old age [15, 16], and were selected because they represent a broad spectrum of age groups spaced nearly equidistance apart. Overall, this type of study design assessing the age-dependent PM effects on HR regulation has not been conducted previously in mice.

Methods

Animals

Male DBA/2 (D2) mice at 3 months of age (n = 13) were purchased from Jackson Laboratory (Bar Harbor, ME). 18 month-old (mo) male D2 mice (n = 11) were purchased from the National Institute of Aging. For the 12 mo male D2 mice (n = 18), 8 were born at Johns Hopkins School of Public Health (JHSPH) and 10 were purchased from Jackson Laboratory (Bar Harbor, ME) at 9 months and housed at JHSPH until 12 months of age. All mice were housed on a 12-h light/dark cycle in the JHSPH animal facility, and were provided rodent chow and tap water ad libitum. The animal facilities at JHSPH are accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) International. All experiments performed involving animals were conducted with approval from the Johns Hopkins University Medical Institutions Animal Care and Use Committee and in accordance with the ethical standards of the institution.

Surgical Procedure for Transmitter Implant

The surgical procedures used for transmitter implant (model TA10ETA-F20; Data Sciences International, St. Paul, MN) to record electrocardiograms (ECGs) and core body temperature (Tco) have been described previously [17]. Briefly, the mice were anesthetized by a continuous flow of 2% isoflurane (Forane; Baxter Healthcare Corporation, Deerfield, IL) with 98% oxygen (Airgas Puritan Medical, Linthicum Heights, MD) using a VetEquip isoflurane vaporizer (VetEquip, Inc., Pleasanton, CA). Chest fur was removed from the abdomen and Betadine was applied to the skin of the exposed region. After making a midline incision in the abdominal wall, the transmitter was inserted and sutured to the abdominal muscle. The negative ECG lead was then placed at the right shoulder and the positive ECG lead was situated above the left thigh. Both leads were sutured to the muscle tissue, thus creating a traditional lead II signal. The mice were then allowed to recover from the surgery for at least 3 weeks prior to PM exposure.

Particulate Matter Collection, Characterization, and Exposure

The PM used for exposure was collected by a 3-stage cyclone system [18]. Only the PM from stage 3 was used for the exposure. Stage 3 of the cyclone system collects PM in the size range of 0.3–3.5 μm [18]. The PM was collected on the JHSPH campus in Baltimore, MD (i.e., an ambient urban environment) for 23 days during April and May in 2008. Each week, the cyclone PM sample was removed and stored in the same amber vial at 4 °C under argon, which minimizes oxidation, degradation, and contamination. In total, the bulk amount of PM was 1.5 g.

The analytical technique to assess the metal composition has been described previously [19] and the metal analysis was completed within 6 months after final collection in May of 2008. Briefly, samples were initially digested in a microwave oven (CEM Mars5Xpress; CEM Matthews, NC) with optima grade nitric acid (HNO₃; Fisher Scientific, Columbia, MD) following a two stage ramp-to-temperature method with a maximum temperature of 175 °C. The hold time between stages was 30 min. Following the first digestion, additional H NO₃ and optima grade hydrofluoric acid (Fisher Scientific, Columbia MD) were added and samples were heated according to the microwave method above. Samples were diluted to 1% H NO₃ with ultra pure water (Milli-Q Synthesis; EMD Millipore Corporation, Billerica, MA) prior to analysis. Standard reference material for urban particulate matter (i.e., stock #1648a; National Institutes of Standards and Technologies, Rockville, MD) and reagent blanks were used for quality control. Total metals analysis was performed using an Agilent 7500ce inductively coupled plasma mass spectrometer (ICP-MS Agilent Technologies, Santa Clara, CA).

Before each aspiration, PM from the bulk collection was weighed on a Mettler-Toledo M5 microbalance (Mettler-Toledo, LLC, Columbus, OH), placed into individual amber vials, and stored at 4 °C under argon (Airgas Puritan Medical, Linthicum Heights, MD) until resuspension. On the day of exposure, 0.9% sodium chloride (i.e., saline; Hospira, Inc., Lake Forest, IL) was added to each amber vial to create a working concentration of 2 μg/μl. The vial was then shaken vigorously, vortexed for 10 s, and sonicated for 30 min in a water-bath sonicator (Bransonic Ultrasonic Cleaner 1510; Branson Ultrasonic Corporation, Danbury, CT). After 15 min, sonication was paused and the vial was again shaken vigorously and vortexed to resuspend the larger particles; sonication was then recommenced for the remaining 15 min. Once resuspension was complete, the animals were exposed to PM by aspiration. The aspirations were performed in late 2010 and early 2011.

The aspiration procedure began with each mouse being sedated under a continuous flow of 4% isoflurane (Forane; Baxter Healthcare Corporation, Deerfield, IL) and 96% oxygen (Airgas Puritan Medical, Linthicum Heights, MD) for 2 min using a VetEquip isoflurane vaporizer (VetEquip, Inc., Pleasanton, CA). After being sedated, each mouse was suspended on a board at a 60° incline. The tongue was then gently extended and a 50 μl aliquot of the 2 μg/μl resuspended PM or saline vehicle was placed at the back of the oral cavity for aspiration by the mouse. Immediately prior to each aspiration, the amber vial was vortexed and vigorously shaken. The mice were monitored for their recovery from the anesthesia, which took < 2 min.

Measurement Acquisition for Ventilation, Metabolism, Body Temperature, and Heart Rhythm

All measurements before and after the aspiration were obtained from mice housed in a whole-body Plexiglass chamber under unrestrained conditions. The chamber conditions, for the measurements of ventilatory function, oxygen consumption, and heart beat acquisition, have been described previously [12]. The mice were routinely placed in the chamber at 8:00 am and removed for aspiration at 11:00 am so as to reduce the measurement variability associated with circadian rhythms. The day before each experiment, mice underwent an acclimation event to the chambers. The mice were placed in the chamber at 8:00 am and were briefly handled at 11:00 am to simulate the aspiration without being placed under anesthesia or exposed to saline or PM. While in the chamber for both the pre-aspiration day and day of aspiration, the mice received un-humidified medical grade compressed air (Airgas Puritan Medical, Linthicum Heights, MD).

The magnitude and patterns of breathing were analyzed from 60-s tracings captured every 75 min by a differential pressure transducer (model 8510B-2; Endevco, San Juan Capistrano, CA). Breathing frequency $\left(f\right)$ and tidal volume ($V_{\mathrm{T}}$) were recorded and minute ventilation ($V_{\mathrm{E}}$) was computed as the product of $f$ and $V_{\mathrm{T}}$. Data were transmitted from the pressure transducer to a PowerLab 4 data acquisition system (ADInstruments, Colorado Springs, CO) and analyzed using LabChart 5 version 5.5.5 (ADInstruments, Colorado Springs, CO) and Igor Pro version 5.0.2 (WaveMetrics, Inc., Lake Oswego, OR). Oxygen (O₂) consumption (${\mathrm{V}\mathrm{O}}_2$), carbon dioxide (CO₂) production (${\mathrm{V}\mathrm{C}\mathrm{O}}_2$) and the respiratory exchange ratio (RER) were sampled at 15-min intervals using an indirect open-circuit calorimetric system (Oxymax Deluxe; Columbus Instruments, Columbus, OH).

A radiotelemetry system (Data Sciences International, St. Paul, MN) was used to acquire measurements for HR, HRV, and core-body temperature (Tco) at 15-min intervals. HR and HRV measurements were computed from 3-min ECG samples recorded every 15 min. Each 3-min ECG sample was analyzed using a QRS peak detection algorithm (LabChart 5 version 5.5.5 with HRV module version 1.0; ADInstruments, Colorado Springs, CO), and examined individually to mark missed heart beats and remove errors in intervals that did not display a normal pattern (i.e., P wave preceding a QRS peak), which included arrhythmias and signal noise. Normal-to-normal (NN) intervals were exported and time-domain parameters for HR, rMSSD, and SDNN were calculated. NN intervals represent the intervals between adjacent QRS complexes that originate after depolarization of the sinus node causing a normal contraction of the heart.

Statistical Analysis

A total of 43 mice were used in the experiment. 13 mice were 4 months with 7 exposed to PM. 18 mice were 12 months with 10 exposed to PM. 11 mice were 19 months with 6 exposed to PM. This design yielded a total of 172 pre- and 430 post-exposure measurements for HR and HRV (4 pre- and 10 post-exposure repeated measurements per mouse). The data analysis was performed using Stata Statistical software release 11 (StataCorp, College Station, TX). A two-way analysis of variance (ANOVA) was performed to test for age and exposure differences in body weights. When results were significant, a post hoc pairwise comparison utilizing a Bonferroni correction was used to determine if there were statistical differences between specific groups for body weights.

Using R Statistical Computing Environment version 2.13.0, the mean time-series analysis of variance was conducted. It used a two-step process to determine the variation that occurs around the average of the parameter being assessed for a group. In this study, the parameter assessed was variation, which was done independently for HR, rMSSD, and SDNN. In other words, within each age and exposure group, each mouse’s individual variation value was determined, and then the mean and standard deviation for those variation values were assessed prior to exposure using linear regression. Confidence intervals were estimated robustly with a nonparametric bootstrap by animal. Bootstrap is a technique used for obtaining confidence intervals by carrying out a random resampling procedure by sampling with replacement using the original data as the population for resampling [20].

Functional principal component analysis (PCA) was used to analyze HR and HRV time-series data jointly. PCA is a dimension-reduction (i.e., factor analysis) tool used for quantifying major variation (i.e., different patterns) in data by determining individual principal component (PC) scores (i.e., PCA for this study created multiple patterns in the data and reduced each pattern to a single value, PC score, for each mouse) [21]. After determining the individual PC scores for each mouse, the PC scores were analyzed using linear regression to compare for differences between age and exposure groups. PC 1 was log transformed (ln PC 1) and PC 2 was Box–Cox transformed to generate normalized distributions. The PC 1 scores for the mice were then used to classify the PM mice into two categories, responders and non-responders. To classify a mouse, its individual PC 1 score was assessed by a student’s t test to determine if the PC 1 score was significantly different from the average PC 1 score of the saline-exposed mice for its respective age group. Each mouse in both the PM-exposed and saline-exposed groups were tested to see if the individual PC 1 score was different from the average PC 1 score of the saline-exposed mice for its respective age group. The mice were classified as responders if their statistical significance was at or below p < 0.05 after a Bonferroni correction.

Before and after classification of responders, time-series analysis was performed using a generalized estimating equation (GEE) model with an autoregressive correlation for repeated measures over time; this approach accounts for the longitudinal design of the study with measurements being repeated at regular intervals in the same mouse. The pre- and post-exposure data were analyzed separately. Model parameters included age, exposure, time, responder classification, a two-way interaction between responder and time, and a three-way interaction between age, exposure, and time to test for changes in HR, rMSSD, SDNN, Tco, ${\mathrm{V}\mathrm{O}}_2,{\mathrm{V}\mathrm{C}\mathrm{O}}_2$, and RER. Parameters of rMSSD and SDNN were log-transformed (ln rMSSD and ln SDNN) and Tco was Box–Cox transformed to generate normalized distributions. A knot at 0.75 h was used in the post-exposure models since responses to PM exposure appeared to be biphasic. In linear regression, knots define break points in the data, which allow for the fitting of straight lines in nonparametric data. Tco, ${\mathrm{V}\mathrm{O}}_2,{\mathrm{V}\mathrm{C}\mathrm{O}}_2$, and RER models did not have a knot in the post-exposure analysis because the data were not collected until 1 h after exposure.

A Student’s $t$ test was used to determine differences between groups for ventilation parameters. The parameters include $f,V_{\mathrm{T}}$, and $V_{\mathrm{E}}$. The analysis was performed at pre-, immediately post-, and 3 h after aspiration. Statistical significance at each time interval was accepted at a p < 0.05 after a Bonferroni correction. $V_{\mathrm{T}}$ and $V_{\mathrm{E}}$ were adjusted for differences in body weight.

Results

Forty-three mice in three different age groups (4-, 12-, and 19-mo) were aspirated with either PM or saline and assessed for pre- and post-exposure differences in cardiac (HR and HRV) and homeostasis (temperature, ventilation, and metabolism) parameters. The metal composition of the PM used for the aspiration is displayed in Table 1. When comparing the mice, there were no significant differences in weight within each age group between PM and saline controls as shown in Table 2 prior to exposure. However, the 12- and 19-mo mice were significantly (p < 0.05) heavier than 4-mo mice, as expected.

Table 1.

Metal content of the Baltimore particulate matter

Metal	Metal content (μg/mg)
	Mean	Standard error
Be	0.020	0.028
Ag	0.055	0.009
Al	3.262	0.572
As	0.040	0.058
Ca	3.118	0.839
Cd	0.013	0.019
Co	0.032	0.051
Cr	0.057	0.065
Cs	0.020	0.032
Cu	0.061	0.054
Fe	2.721	0.175
K	0.946	0.167
Mg	2.360	0.721
Mn	0.113	0.047
Mo	0.030	0.027
Na	5.795	1.164
Ni	0.042	0.054
Pb	0.035	0.029
Sb	0.034	0.034
Se	0.015	0.009
Sn	0.026	0.030
Ti	0.215	0.053
Tl	0.166	0.291
V	0.039	0.050
Zn	0.207	0.077

Table 2.

Body weights by age and exposure

	4 months	12 months	19 months
Weight (g) (mean ± SE)
Saline	29.4 ± 0.7	33.6 ± 1.2*	34.1 ± 1.2*
PM	28.4 ± 0.4	33.0 ± 1.0*	32.9 ± 1.2*

^*p < 0.05 for each age compared to 4 months within exposure

To assess if there were any differences in HR and HRV between the age groups of mice prior to exposure, mean variance analysis was also conducted. For the mean variance (Table 3), the 19-mo mice had a significantly (p < 0.05) lower mean variance compared to the 4- and 12-mo mice for HR. This means that the 19-mo mice have a smaller variation in HR among individuals within that age group compared to the other two age groups. However, for rMSSD, the 19-mo mice had a mean variance that was greater than the 4-mo mice. Thus, there appears to be an age-dependent trend towards increasing variance in rMSSD as well as decreasing variance in HR (Table 3) prior to any aspiration.

Table 3.

Pre-exposure mean time-series variance for HR, rMSSD, and SDNN by age

	Variance mean variance (95% CI)
	Heart rate	rMSSD	SDNN
4 month	8.3 (7.7, 8.6)*	3.6 (2.5, 4.0)	3.9 (3.1, 4.1)*
12 month	8.6 (7.8, 8.8)*	4.0 (1.9, 4.6)	3.7 (2.7, 4.1)
19 month	7.4 (6.5, 7.7)	5.8 (4.1, 6.3)	4.5 (3.3, 4.9)

^*p < 0.05 for each age compared to 19 months

HR and HRV

PCA and time-series analyses were used to determine PM effects by age and exposure on HR and HRV. Prior to the exposure, based on time-series analysis, there were no significant differences in HR and HRV between any of the groups of mice. After the exposure, there also was no significant difference in HR and HRV between PM and saline exposed mice within each age group. However, analysis by PCA did reveal a significant difference (p < 0.05) between PM-exposed mice and saline for the 12-mo age group (Table 4).

Table 4.

Principal component scores by age and exposure

	Principal component scores (mean ± SE)
	PC 1	PC 2	PC 3
Saline
4 month	1.42 ± 0.26	2.84 ± 0.47	0.05 ± 0.81
12 month	1.29 ± 0.22	2.72 ± 0.40	0.59 ± 0.70
19 month	1.15 ± 0.28*	3.48 ± 0.51	0.05 ± 0.89
PM
4 month	1.85 ± 0.24	3.07 ± 0.43	−1.55 ± 0.75
12 month	2.06 ± 0.20#	2.11 ± 0.36	0.14 ± 0.63
19 month	1.24 ± 0.26	3.32 ± 0.47	0.69 ± 0.81

^*p < 0.05 for each age compared to 12 months within exposure

^#p < 0.05 for saline vs. PM exposure within each age

PCA is a dimension-reduction analysis used to quantify major directions (i.e., unique patterns) of variation in functional data. For the ECG data, three principal components (i.e., patterns) explained 76% of the variation that occurred in the HR and HRV data. The first PC pattern (i.e., PC 1 score) accounts for 48% of the variation seen in the data and captured the drop in HR and the increases in SDNN and rMSSD that was visible during exploratory data analysis in some, but not all mice following PM exposure. The PC 1 scores were then analyzed using linear regression. For the saline-exposed mice, only the 19-mo mice were found to have a significantly (p < 0.05) lower PC 1 score compared to the 12-mo mice. When comparing within each age group between the PM-exposed mice and the saline-exposed mice, a significant difference (p < 0.05) was found in the 12-mo mice (Table 4). The other two principal components did not display any significant differences based on age or exposure (Table 4). Therefore, based on PCA analysis, the only apparent response to PM exposure occurred in the 12-mo age group.

Since a significant difference was seen in the 12-mo mice, the individual PC 1 scores were used to determine if it was all or a subset of the 12-mo PM exposed mice that responded to the PM exposure. This comparison was performed by testing to determine if the PC 1 score for each 12-mo PM-exposed mouse was statistically different (p < 0.05) in comparison to the overall distribution of the PC 1 scores for all the 12-mo saline-exposed mice (i.e., average PC 1 score for the 12-mo saline-exposed mice). Based on this analysis, three of the ten 12-mo PM exposed mice were found to have a significantly different (p < 0.05) PC 1 score. These three mice were then labeled as responders. Of the three responder mice, two were born at JHSPH and the other one was from Jackson Laboratory. The other seven 12-mo PM-exposed mice that did not have significantly different PC 1 scores from the saline-exposed were classified as non-responders. However, since PCA is a pattern-reduction method, it does not provide any means of analyzing the magnitude or change over time in HR or HRV following PM aspiration. Therefore, the classification of responders, based on statistically significant differences in PC 1 scores for individual PM-exposed mice compared to the saline average PC 1 score of the same age group, was utilized in the time-series analyses to describe the magnitude and change over time in HR and HRV both pre- and post-exposure.

Based on time-series analysis utilizing the responder/non-responder classifications, there still remained no differences between age or exposure groups prior to aspiration (i.e., baseline) for the average HR, rMSSD, or SDNN, except within the 12-mo mice (Fig. 1a). For the 12-mo mice, the 12-mo non-responders had a significantly (p < 0.05) lower HR of 410 beats/min (bpm; 95% CI 372–448) compared to the 12-mo saline controls with an average HR of 498 bpm (95% CI 442–554) prior to exposure.

Fig. 1.

Each time interval displays the mean ± SE for heart rate (a1–3), ln rMSSD (b1–3), and ln SDNN (c1–3) pre- and post-aspiration for each age and exposure group

After exposure to PM, a biphasic response in HR and HRV occurred in the 12-mo responder mice. No response occurred in the other age groups, 4- or 19-mo, of PM-exposed mice. Aspiration of saline as a control also had no impact on HR and HRV. The biphasic response had an immediate response phase followed by a recovery phase that started 1 h after exposure. Immediately following the exposure, the average HR (Fig. 1a) in the 12-mo responder mice of 309 bpm (95% CI 250–368) was significantly (p < 0.001) lower compared to the 12-mo saline-exposed mice and the 12-mo non-responder mice showing average HRs of 561 bpm (95% CI 490–631) and 531 bpm (95% CI 477–585), respectively. The HR of the 12-mo responder mice was also significantly (p < 0.001) lower than 4- and 19-mo PM-exposed mice. This initial drop in HR was not seen in any other age group exposed to PM or saline. Concurrently, while HR was lower, both ln rMSSD (Fig. 1b) and ln SDNN (Fig. 1c) were significantly (p < 0.001) higher in the 12-mo responders compared to 12-mo saline and 12-mo non-responder groups as well as the 4- and 19-mo PM-exposed mice (p < 0.004). No differences in HR or HRV occurred when comparing PM-exposed mice to saline in either the 4- or 19-mo age groups.

During the recovery phase, the 12-mo responder mice showed a gradual increase in HR after 1 h (Fig. 1a). The increase in HR during recovery returned to similar levels as the 12-mo saline and 12-mo non-responder groups after 3 h. Similarly, both ln rMSSD (Fig. 1b) and ln SDNN (Fig. 1c) decreased in 12-mo responder mice from peak levels at 1-h post-exposure to similar levels as the 12-mo saline and 12-mo non-responders within 3 h post-exposure. In 19-mo saline-exposed mice, there was a significant (p < 0.05) decline in HR over the 3-h recovery period at an average rate of − 29 bpm/h (95% CI − 41 to − 16), which was not evident in the 19-mo PM mice showing a rate of − 2 bpm/h (95% CI − 20 to 17). Despite the difference in HR for the 19-mo mice, there were no detectable differences for ln rMSSD or ln SDNN between the exposure groups at 19-mo. For the 4-mo mice, there were no detectable differences during the recovery phase in HR, ln rMSSD, and ln SDNN between the PM- and saline-exposed groups.

Ventilation, Metabolism, and Body Temperature

Ventilation, metabolism, and body temperature pre- and post-exposure were analyzed by time-series analysis utilizing the responder/non-responder classification. Prior to exposure, the 12-mo PM non-responders had a significantly (p < 0.05) lower Tco compared to 12-mo saline and 12-mo responders (Fig. 2a). There were no detectable differences in Tco between exposure groups for 4- and 19-mo old mice prior to exposure. Following exposure, the 12-mo PM responders had significantly (p < 0.001) lower Tco compared to the 12-mo saline and non-responders as well as the 4- and 19-mo PM-exposed mice. There was no Tco response in the 4- and 19-mo PM-exposed mice (Fig. 2a). By 3 h post-exposure, there were no detectable differences in Tco between the 12-mo PM responders and all other groups.

Fig. 2.

Each time interval displays the mean ± SE for body temperature—Tco (a1–3), O₂ consumption—${\mathrm{V}\mathrm{O}}_2$ (b1–3), and CO₂ production—${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ (c1–3) pre- and post-aspiration for each age and exposure group

For ${\mathrm{V}\mathrm{O}}_2$ (Fig. 2b) and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ (Fig. 2c), 19-mo saline mice showed significantly (p < 0.05) lower metabolic parameters compared to the 4-mo saline mice prior to exposure. After the exposure, the 19-mo PM mice showed significantly (p < 0.05) lower ${\mathrm{V}\mathrm{O}}_2$ and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ levels compared to both 4-mo PM and 12-mo PM non-responder mice. For the 12-mo PM responders, significantly (p < 0.003) lower ${\mathrm{V}\mathrm{O}}_2$ and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ levels occurred immediately following exposure compared to the other PM-exposed age groups. Significantly (p < 0.001) lower ${\mathrm{V}\mathrm{O}}_2$ and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ levels in the 12-mo PM responders was also evident when compared to 12-mo saline and non-responder mice. Like the Tco responses, the 12-mo responders showed no detectable difference in ${\mathrm{V}\mathrm{O}}_2$ and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$ levels when compared to all other groups at 3 h post-exposure.

With respect to ventilation parameters, the 12-mo PM responders showed a significant (p < 0.05) increase in $f$ at 3 h after PM-exposure compared to 12-mo saline mice (Fig. 3a). There were no other differences between the age or exposure groups in $f$ before or immediately following the aspiration. Prior to the aspiration, there were no differences in $V_{\mathrm{T}}$ or $V_{\mathrm{E}}$ between the age or exposure groups. In contrast, PM exposure resulted in a significant (p < 0.05) drop in $V_{\mathrm{T}}$ and $V_{\mathrm{E}}$ in the 12-mo PM responder mice compared to the other PM-exposed groups. At 3 h following the aspiration, the 12-mo PM responder mice had the same $V_{\mathrm{T}}$ and $V_{\mathrm{E}}$ as the other groups (Fig. 3b, c).

Fig. 3.

Each time interval displays the mean ± SE for frequency—$f\left(\mathbf{a}1-3)\right.$, tidal volume—$V_{\mathrm{T}}(\mathbf{b}1-3)$, and minute ventilation—$V_{\mathrm{E}}(\mathbf{c}1-3)$ pre- and post-aspiration for each age and exposure group

Discussion

The current study focused on age-dependent differences in HR and HRV responses to PM-exposure using urban ambient particulate matter collected in Baltimore City. The effects of PM exposure predominately occurred in a subset of the 12-mo mice, in which a decline in HR and an increase in HRV was dramatically evident. These findings appear to be consistent with other studies in rodents after acute exposure to various components of urban ambient PM [22, 23]. In addition to the effect of PM on HR and HRV in the 12-mo PM responders, the same exposure caused only a modest effect on HR in the 19-mo mice without any effect on HRV. Overall, these findings are unique in terms of assessing the possible responses to PM exposure based on age-dependent differences for a broad spectrum of biological parameters, including cardiac effects.

The use of principal component analysis for classification of time-varying response patterns is also a novel approach. When combining the classification of response patterns from PCA with time-series analysis, it allows for a more detailed description of the magnitude of the time-varying acute effects of PM exposure. When used independently, the same conclusions are not possible to obtain as regression of the PC scores only reveals that there is a significant pattern difference between the 12-mo PM exposed mice and the 12-mo saline mice. Comparatively, when only the time-series analysis is used, no difference is found due to PM exposure in any age group. However, when analysis of the individual PC scores are used to classify responders and the classification is included in the time-series analysis, a difference in the magnitude of HR and HRV for a subset of the 12-mo PM exposed mice can be elucidated (Fig. 1). Overall, this allows for the detection of a subtle difference in HR and HRV as a result of PM exposure, which otherwise might not be possible.

Using the combined PCA and time-series regression, two types of responses were apparent within the 12-mo PM-exposed mice (i.e., responders and non-responders). The result of the PM-exposure in the 12-mo responders was a drop in HR as well as increases in rMSSD and SDNN whereas PM exposure had no effect in the 12-mo non-responders or the 4- and 19-mo PM-exposed mice. Similar decreases in Tco, ventilation, and indicators of metabolism (i.e., ${\mathrm{V}\mathrm{O}}_2$ and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$) were also seen after PM exposure in the 12-mo responder mice but not the 12-mo non-responder mice or the 4- and 19-mo PM exposed mice. One possible mechanism that connects the impact of PM exposure in the lung to other organ systems is the reflex arc.

The lung has many sensory neural receptors that are capable of responding to stimulation by air pollutants include pulmonary C-fibers and rapidly adapting receptors (RARs), which are also known as irritant receptors [24, 25]. While, both C-fibers and RARs modulate pulmonary responses, RARs can also affect the cardiac system [26]. Activation of these receptor types can result in a reflex arc after neural integration in the central nervous system (CNS) leading to efferent autonomic (parasympathetic and sympathetic) signals being sent to various organs of the body, including to the heart and back to the lung [24, 25]. Changes in HR, Tco, ventilation, and metabolism are common responses to acute stress, especially in rodents, that could potentially be initiated by the reflex arc [27, 28]. Since the changes in HR, Tco, ventilation, and metabolism did not occur in mice that were only instilled with saline, the response seen in this study is unlikely to be an acute stress response to the method of instillation used for exposure or due the use of anesthesia. Instead, the effects are more likely due to the acute PM exposure.

In mice, the acute stress response to toxicant exposure results in a hypothermic stress response. The hypothermic response is characterized by a decrease in Tco, HR, ventilation, and metabolism [28]. The premise of the hypothermic response is that it may help to diminish the health impact of the PM exposure by potentially reducing uptake and bioactivation of the PM [28, 29] thereby protecting the mice via allostasis. The series of events (i.e., declines in HR, Tco, $V_{\mathrm{T}},V_{\mathrm{E}},{\mathrm{V}\mathrm{O}}_2$, and ${\mathrm{V}\mathrm{C}\mathrm{O}}_2$) that occurred in the 12-mo responders after PM-exposure mimics the classification of the hypothermic response. Yet, not all age groups displayed a hypothermic response.

The absence of a hypothermic response in the 19-mo mice and 12-mo non-responders may be due to alterations in physiologic compensatory function that occur with age. Specifically, with aging comes a decline in physiologic compensatory function, which has been proposed as a reason why the elderly may be susceptible to PM exposure [30]. Aging also has been previously conceptualized as a loss of robustness and progressive decline in the complexity of the control mechanisms (i.e., allostasis) that are necessary to adapt to stress and return to homeostasis [31–33]. Maintaining homeostasis involves having normally functioning physiological systems such as blood pressure, heart rhythm, brain electrical activity, and hormone levels that are able to respond dynamically via short-term fluctuations initiated by a complex network of control mechanisms [34]. It has been suggested that the inability to compensate with aging is due to a reduction in the number of dynamic responses and parallel compensatory pathways that provide potentially corrective actions [31, 32, 34]. Potential changes in cardiac homeostatic compensatory mechanisms, like the loss of HRV with aging [35], could be due to diminished β-adrenergic receptor responsiveness, which has been found in rodents [36]. Modeling the changes in physiological compensatory function that occurs with age could be represented by an inverted U-shape with increasing development of complex responses as the organ systems mature and an eventual gradual loss of complex responses with increasing age as physiological function declines. However, more work should be done to validate if these changes can really be modeled by an inverted U-shape.

In this study, the potential inverted U-shape of changing complexity with age resembles the responses in HR and HRV, or more accurately lack thereof, in the PM-exposed 4- and 19-mo groups and dramatically different individual changes in HR and HRV at 12 months (Fig. 1). In the 19-mo mice, the absence of a response could be due to loss of compensatory mechanisms. Assessing baseline variability is one possible method for predicting an individual’s ability to compensate and respond to a stressful event [34]; for an individual this would be equivalent to a small value for variation. If estimated and assessed as a group, which was done in this study, the mean variation value for the group would be small and there would be little variation in that value between individual animals. The 19-mo mice pre-exposure (i.e., at baseline) had a smaller mean variation in HR compared to either the 4- or 12-mo mice (Table 3) as determined by mean variance analysis. This smaller baseline variability resembles the loss of complex responses that is associated with aging. Conversely, the absence of a response in the 4-mo mice could be due to a full complement of compensatory mechanisms available. Whereas the 12-mo mice, which are mid-life, may represent an age group in which some individuals might begin to be losing compensatory mechanisms, such as the non-responders, while other individuals may not, such as the responders. The difference pre-exposure in HR between the 12-mo responders and non-responders might be an indication of the existence of this difference.

However, some studies, but not all, have seen similar acute and immediate decreases in HR and increases in HRV almost immediately after exposure. Unfortunately, very few have assessed older animals and even less have assessed multiple age groups. In one similarly designed study that also used aspirations, 5-mo chronic ischemic heart failure (CHF) rats were exposed to a 50 μl aliquot of Baltimore PM at a concentration of 20 mg/kg, which resulted in a decrease in HRV [37]. By comparison, most other studies have used inhalation as the method of exposure.

For example, one study using 4-mo rats fed either a normal diet or a high-fructose diet to induce metabolic syndrome were exposed to P M_2.5 at an average concentration of 356 μg/m³ via inhalation for 8 h/day. The PM exposure decreased HR and increased rMSSD in both groups [38]. Meanwhile, another study found slightly different results. The study exposed 3-mo normal and hypertensive rats for 3 h via nose-only inhalation to 0.45, 1.0, or 3.5 mg/m³ of a synthetic residual oil fly ash (s-ROFA) and found decreased short-term HRV in the normal rats but increased short-term HRV in the hypertensive rats [39]. Another study that also exposed 3-mo hypertensive rats to the same concentrations of s-ROFA for 4 h via inhalation also found an increase in rMSSD as well as a decrease in HR [23]. In contrast, a study looking at 6-mo normal and CHF rats exposed to diesel exhaust PM for 3 h at a concentration of 0.5 mg/m³ via inhalation caused a reduction in rMSSD in both rats but had no impact on HR [40]. Finally, a study performed in about 3-mo hypertensive rats that were exposed for 4 h/day for 4 days via inhalation to PM concentrations of 50, 150, or 500 μg/m³ from various biodiesel sources resulted in varying combinations of responses. The responses to the exposure were decreases in HR and mixed impacts on RMSSD, which included significant effects only caused by exposure to one biodiesel resulting in decreased rMSSD at 50 μg/m³, increased rMSSD at 500 μg/m³, and no effect on rMSSD at 150 μg/m³ [41].

As can be seen from the various studies in rodents, they differed in the method of exposure (inhalation vs. aspiration), concentration of exposure, type of PM, and included various susceptible groups. As a result, all of these differences make it complicated to compare across studies to determine the impact of PM exposure on HR and HRV. Yet, as noted by one author, there appear to be dose-dependent responses in the autonomic system such that sympathetic responses may be obscuring vagal irritant reflexes at lower concentrations, which are even further impacted by varying susceptibility [42]. As this study involves a low dose exposure provided via aspiration, which is similar to a moderate level for a short-term inhalation exposure, it is possible the sympathetic response is minimizing irritant receptor response, except in the 12 month responder mice, which resulted in a decrease in HR. A similar decrease in HR in 12-mo mice was also found to occur shortly after exposure to PM. In that study, the mice were exposed to about 200 μg/m³ carbon black for 3 h [14]. However, given all of these variations in responses to PM exposure and very few studies in older animals, future studies assessing age-dependent inhalation exposures would be more ideal to further elucidate the potential age-dependent effects of exposure to PM on HR and HRV.

There are also several other limitations of the current study. The biggest limitation of this study is the confounding that exists between age and the birth location of the mice. Each age group was acquired from a different source, therefore, it is possible that the age-dependent effects following PM exposure seen in this study might be due to the source of the mice instead of being due to age differences. Therefore, additional studies are necessary to corroborate the findings of this study. In addition, the potential decline in compensatory mechanisms was assessed by evaluating the mean variance of heart rate for only the normal-to-normal beats, while excluding some arrhythmias. Since the presence of arrhythmias could be viewed as providing complexity to the system, further studies would need to be conducted to validate this as a technique for measuring loss of compensatory mechanisms. Another limitation is the number of groups; by adding more age groups between the 4- and 12-mo mice and the 12- and 19-mo mice, one might be able to better assess if it is a gradual change in the increase in responders that occurs with age. Another limitation was the method used for and the frequency of exposure. The mice were aspirated with a single dose of PM instead of breathing in the same concentration over an extended period of time. Also, when instilling the PM, there is a chance that the bolus might have been swallowed instead of aspirating into the lung. While the tongue is held until each mouse takes its first breath to minimize the likelihood of the bolus being swallowed, the chance of the bolus being swallowed does exist and might also lead to some variability in responses seen in the mice. Instead, an acute exposure at a similar concentration performed via inhalation over a longer period would likely be a better method. In addition, exposure only occurred once instead of multiple times, such as subchronic or chronic, as occurs in humans. Therefore, future studies assessing age-dependent chronic inhalation exposures would be more ideal to further understand the potential long-term effects of exposure on age, especially as it relates to humans.

In conclusion, the results indicate the presence of different responses to acute PM exposure across age groups for various indices of homeostasis. These age-dependent responses, especially as it applies to various adult age groups, have not been frequently studied. Also, the changes in HR regulation have not been well studied as a function of aging and PM exposure in animal models, but our work does demonstrate, despite the potential confounding associated with varying sources of the mice for each age group, that for a subset of 12-mo mice, there is an increase in rMSSD and SDNN, a decline in HR, and alterations in other measures of homeostasis, which commonly occur during a hypothermic response to stress in mice. Given the absence of any response to PM exposure in the 19-mo mice, it could possibly be due to a loss of complexity and parallel compensatory mechanisms, such as the hypothermic response, that occurs with age [30]. As other studies have suggested, the hypothermic response in mice does not reflect changes normally seen in humans. However, there may be elements of the pathway involved in the initiation of the vagal response that could be conserved between the species [27]. If a conserved pathway does exist, it might help to explain the association between PM exposure and increases in morbidity and mortality in older individuals that is described in the epidemiology literature.