PM2.5-induced cardiovascular dysregulation in rats is associated with elemental carbon and temperature-resolved carbon subfractions

Background We tested the hypothesis that cardiovascular responses to PM2.5 exposure will be enhanced in hypertensive rats and linked to specific carbonaceous pollutants in an urban industrial setting. Methods Spontaneously hypertensive rats were exposed by inhalation to concentrated PM2.5 in an industrial area of Dearborn, Michigan, for four consecutive summer days. Blood pressure (BP), heart rate (HR) and HR variability (HRV) metrics (SDNN, RMSSD) were assessed by radiotelemetry and compared to 1 h- and 8 h-averaged fluctuations in PM2.5 composition, with a focus on elemental and organic carbon (EC and OC, respectively), and temperature-resolved subfractions (EC1-EC5, PC (pyrolized carbon), and OC1-OC4), as well as other major and minor PM components. Results Mean HR and BP were increased, while HRV was decreased over 4 days of exposure. Using 1 h averages, EC (1 μg/m3 increase) was associated with increased HR of 11-32 bpm (4-11% increase), 1.2-1.5 ms (22-27%) decreases in SDNN, 3-14 mmHg (1.5-8%) increases in systolic BP, and 5-12 mmHg (4-9%) increases in diastolic BP. By comparison, associations with OC were negligible. Using 8 h averages, EC subfractions were linked with increased heart rate (EC1: 13 bpm; EC2, EC3, PC: <5 bpm) and SDNN (EC1> > EC2 > EC3, EC4, PC), but with decreased RMSSD (EC2, EC5 > EC3, EC4). Minimal effects were associated with OC and OC1. Associations between carbon subfractions and BP were negligible. Associations with non-carbonaceous components and trace elements were generally non-significant or of negligible effect size. Conclusions These findings are the first to describe associations between acute cardiovascular responses and thermally resolved carbon subfractions. We report that cardiovascular responses to PM2.5 carbonaceous materials appear to be driven by EC and its EC1 fraction.

In addition to associations with PM 2.5 mass, two recent meta-analyses using Medicare records and particle composition data from the Environmental Protection Agency Chemical Speciation Network found significant associations between elemental carbon (EC), a combustion product most often linked to diesel engine exhaust, and hospital admissions for cardiovascular causes [11,12]. Ambient EC concentrations have been associated with increased systolic pressure [13][14][15] and HRV [13,16]. Another major carbonaceous fraction of PM 2.5 is organic carbon (OC), which is derived from mobile sources, biomass burning and industrial processes and has also been linked to decreased HRV and elevated BP [6,16]. Extensive analyses of PM 2.5 EC and OC using thermal/optical analytical approaches yields up to five OC subfractions (OC1-OC4 and pyrolized carbon [OP]) and four EC subfractions (EC1-EC4). This method applies step-wise increases in temperature in a controlled atmosphere to oxidize first organic and then elemental carbon samples to yield eight fractions with decreasing volatility and increasing structural complexity. Thus OC1 and EC1 are more volatile and simpler in structure than OC4 and EC4, respectively. These carbon subfractions have been used to refine source apportionment analysis in the U.S. [17] and more recently in China [18,19]; however, the relative health effects of these temperature-resolved subfractions of EC and OC have not been evaluated.
We have previously reported altered HRV in hypertensive rats exposed to concentrated PM 2.5 in Detroit, MI and Steubenville, OH that was linked to specific PM components and related industrial and mobile sources [20,21]. In these studies we found EC to consistently have the strongest association of any PM 2.5 component with changes in heart rate and HRV. Our goal in the present study was to extend these observations to include blood pressure responses in hypertensive rats, as well as provide a detailed characterization of both EC and OC and their subfractions in an industrial urban center in Dearborn, MI. Our results describe clear differences between EC and OC in PM2.5induced responses, as well as provide initial insights into the relative potencies of carbon subfractions in relation to acute cardiovascular responses.

Exposure characterization
The average (±SD) chamber PM 2.5 mass concentrations during the four 4-day exposure studies were as follows: Study 1, 415 ± 99; Study 2, 642 ± 294; Study 3, 767 ± 256; Study 4, 364 ± 58 μg/m 3 . Figure 1 depicts the distribution of major chemical components of PM 2.5 collected during each 4-day inhalation exposure period. Table 1 shows the combined numeric average for all exposures. As previously observed in this Dearborn study area [22,23], sulfate and OC dominated during the summer months. Figure 2 shows the average carbon fraction distribution of concentrated Comparisons of responses in rats exposed to AIR vs. PM 2.5 using 1 h averaged data resulted in several-exposure-related differences ( Table 2). Changes in blood pressure responses were the most sensitive indicator of PM 2.5 exposure, with significant associations with MAP, systolic and diastolic pressures during 3 of the 4 exposure studies. Other significant associations were found for heart rate during studies 1 & 3, and SDNN during Study 3.

Associations with major components and trace elements
Assessment of major non-carbonaceous PM components (nitrates, sulfates and ammonia) and 29 elements revealed a number of minor, but statistically significant associations with HR, HRV and BP (Table 3, Additional file 1: Tables S1  and S2). For example, significant responses of less than 0.1 bpm for HR and less than 0.01 ms for lnSDNN were associated with the change in IQR for a few specific components or elements. The only PM 2.5 constituent with a larger effect estimate was uranium, which was associated with a 15 mm Hg increase in diastolic pressure (Table 3), although the associations between this element and both systolic BP and MAP were not significant (Additional file 1: Table S2). The size of the uranium effect estimate compared to other components may be due to its low ambient concentration (0.19 ng/m 3 ), and relative IQR (0.2 ng/m 3 ).

Associations with carbon fractions
During all four exposure studies, PM 2.5 EC concentration was consistently associated with increased HR with estimated changes of 11-32 bpm (4-11% increase) for a 1 μg/m 3 increase in EC (Table 4). EC was also associated with 22-27% decreases in HRV during Studies 1 & 4. By comparison, OC and PM 2.5 mass had little to no associations with cardiac automaticity.
Changes in blood pressure also appeared to be more influenced by variations in EC than by OC or PM mass (Table 5). Increases in systolic BP ranged from 3-14 mmHg (1.5-8% increases) for a 1 μg/m 3 increase in EC for Studies 1-4. MAP was elevated by 4-9% in Studies 1 & 2 while diastolic BP increased significantly with EC by 4-10% in three of the four exposure studies.

Associations with carbon subfractions
Overall the daily concentrations of EC subfractions had greater associations with heart rate and HRV changes than OC subfractions ( Figure 3A-C). Subfractions EC1, EC2, and EC3 were all significantly associated with increased heart rate and SDNN, with EC1 having the largest effect estimate (HR increase of 13 bpm; SDNN increase of approximately 7.5%). Interestingly, no effects were found for EC itself using the 8 h averages. EC2, EC3, EC4, and EC5 were all significantly negatively associated with reductions in RMSSD. EC1 showed extremely wide confidence intervals. Small effect estimates were seen for OC and OC1, the only organic carbon fractions with significant associations. Carbon subfractions had weak associations with vascular responses ( Figure 3D-F). Except for total EC which had small effect estimates (>1 mmHg) for both MAP and DBP, no EC or OC subfractions were significantly associated with any change in blood pressure.

Discussion
Results of this study clearly demonstrate that the EC and EC subfractions drive the acute changes in HR, HRV and BP in hypertensive rats exposed to an urban-industrial aerosol. These findings are consistent with our previous report of EC's relation with HR in the same rat model [21], but we now extend these observations to identify EC-associated effects on BP as well, suggesting a more profound overall impact on cardiovascular health. Furthermore, we detected associations between EC subfractions and cardiac function, especially between EC1 and EC2 with HR and HRV. By comparison to EC, OC was associated with only modest and sometimes opposing cardiovascular effects in PM 2.5 -exposed rats. To our knowledge, our current findings are the first to describe associations of cardiovascular health effects with inhaled PM2.5 carbon subfractions using thermal/optical analytical methods.
Although not all source apportionment studies include EC and OC in their factor resolution, those that do typically find that these PM components load onto mobile source emissions factors [24,25]. Furthermore, the separation of carbonaceous fractions derived from progressive oxidation temperatures suggest that the subfraction EC1 is generally linked to diesel engine sources, whereas OC fractions (OC1-4) are normally associated with gasoline emissions [17,26]. It is notable, therefore, that the adverse cardiovascular responses we describe for EC and its  subfractions are consistent with HR, HRV and BP responses reported during controlled exposures to whole diesel engine exhaust in the same SH strain of rats we used in the current study [27,28]. While major components of laboratory-generated diesel exhaust are gaseous inorganic compounds (nitrogen oxides, sulfur dioxide, and carbon monoxide), the minor components of particulate EC and of volatile and semivolatile organic compounds (OC) have been linked to stimulation of both sympathetic and parasympathetic cardiovascular effects in SH rats where exposures compared whole versus filtered exhaust [29,30]. A major difference in our field studies compared to laboratory-generated diesel engine exhaust is the atmospheric transformation of EC core particulates that might result in surface adherence of volatile organic hydrocarbons such as carbonyls, or in oxidative modifications that alter the particle's toxicity [31,32]. OC was a major component of PM 2.5 mass at our urban site, and though not strongly associated with health effects in our study, it may contribute to cardiovascular effects as a surface component of diesel soot particles. Our study location in Dearborn, MI is located near automotive production industries with heavy vehicle and diesel traffic in the surrounding community. Of note are several active trucking facilities within a mile of the site, with several hundred trucks loading and unloading cargo daily, as well as a railyard within 250 m of the site. We recently compared cardiometabolic responses in twentyfive volunteers before, during and after ambient exposures at this same industrial site, relative to their residences in rural upwind areas of Dexter, MI [33]. After five daily exposures, decreases in HRV and insulin resistance were associated with increased PM2.5. Further analyses of source: health effect relationships found that changes in HR, BP and trends for impaired endothelial function were associated with the diesel source factor that impacted this site [34]. Fewer health effects were associated with other PM source factors, with motor vehicle sources being linked to changes in BP, and iron/steel and secondary aerosol source factors being associated with changes in HR. We recently reported dramatic drops in BP and HR in fructose-fed rats with cardiometabolic syndrome exposed to PM2.5 at this same site in Dearborn; however, analyses to attribute specific sources with CV responses have not been completed   [35]. Our findings with EC and subfractions in the current study adds to our previous work with PM2.5-exposed SH rats where EC and traffic sources had robust associations with changes in HR and HRV in urban Detroit, MI and Steubenville, OH [20,21].
To date, carbon subfraction analysis has been employed in air pollution studies primarily to incorporate fractions into source apportionment analyses and, ideally, to improve source factor resolution [36,37]. For the most part, such studies have also been able to differentiate gasoline from diesel emissions, as discussed above. In addition, some work has investigated indoor and non-indoor sources of carbon fractions in residential homes [38]. We were able to identify one in vitro study including carbon subfractionation in which human lung epithelial cells were exposed to dust from soil and road surfaces in the western United States [39]. Release of inflammatory mediators was most highly correlated with the EC1 fraction, while lesser correlation coefficients were observed for OC fractions and pyrolized carbon. Compared to other EC fractions, compounds that comprise EC1 and EC2 would be more volatile, of smaller molecular weight, and a less complex structure. Less clear are the molecular targets, receptors or proteins with which different subfractions may preferentially interact to elicit biological responses. In addition to the greater responses induced in airway cells, smaller sized EC1 compounds would theoretically be better able to translocate and influence extarpulmonary responses. However, epidemiological or clinical evidence for EC subfractionassociated health effects is lacking, and we speculate that EC and EC1may be markers for other pollutants or pollutant mixtures that underlie the health effects we describe in exposed rats.
In similar PM 2.5 field exposures in urban areas we have previously identified associations of health effects with a number of trace elements that are linked to industrial activities in the Midwest [21,40]. In the current analyses, several elements had statistically significant associations with HR, HRV and BP, but we interpret the effect sizes as having questionable biological relevance (e.g., < 0.1 bpm HR). Interestingly the greatest and most consistent effects were found with uranium, which had a considerable effect estimate for its association with increases in diastolic BP (13 mmHg). Using x-ray diffraction analysis to assess Detroit PM constituents, our colleagues showed that uranium is colocalized with EC in graphene structures [41]. Its source in the urban industrial airshed in southwest Michigan is unknown although it has been associated with coal combustion [42]. In our study, despite health effects associated with this element, other elements and components typically associated with coalfired power plant emissions such as sulfate, selenium, and arsenic, yielded no adverse health effects findings.
Compared to most of our community-based animal studies, the current investigation is limited by the use of 8 h-integrated PM 2.5 samples to estimate effects on daily changes in cardiovascular responses. Our group is unique in that we have used 30-minute sampling periods for both particle collection and cardiotelemetry recording; however, the necessary instrumentation was not available for all studies in this project. A second limitation is the lack of normotensive control subjects with which to compare our responses in the SH rats. We have previously used Wistar Kyoto rats as our healthy controls, but because of space limitations in the exposure chamber they were omitted to allow for a larger group number of hypertensive rats. As such any translation of our findings to understand potential susceptibility is limited.
In summary, this is the first report of cardiovascular health effects linked to inhalation exposure to ambient PM 2.5 carbon subfractions. Increased BP and HR and decreases in HRV showed robust associations with EC, and our initial findings using thermal optical approaches yielded strong relationships of EC1 and EC2 with changes in HR and HRV. Interestingly, we found relatively fewer and weaker responses with OC fractions or trace elements. Black carbon has been proposed as an important indicator of PM-induced health effects [43], and our results with EC add to this evidence, specifically for adverse cardiovascular responses. Future research efforts that include the analyses of carbon subfractions are needed to confirm our observations, and will help to further characterize the contribution of EC to the health risk of PM 2.5 exposures.  2010. The site is located within 5 km of iron/steel production facilities, a coke oven, oil refinery, sewage sludge waste incinerator, a coal-fired power plant and major highways. Concentrated PM 2.5 was generated from ambient PM 2.5 using a Harvard-type fine particle concentrator and whole body animal exposure chambers as previously described in detail [44]. Exposures were carried out in two stainless steel Hinners-type whole body inhalation chambers; one received PM2.5 while the other received HEPA-filtered clean air at the same flow rate as the experimental group. Eight SH rats were exposed in each chamber from 7:30 am  Lead II configuration to sample cardiac parameters and the pressure catheter placed in the aorta via the femoral artery. Telemetry receivers (RLA3000, DSI) were modified and affixed inside individual cages in exposure chambers that were customized for telemetry studies. Datastreams of 30 second duration were collected and analyzed every 5 minutes during exposures. Automated ECG analysis (DSI, ART3.2) allowed for R-wave detection on a beat-to-beat basis. The R-R intervals for all normal beats (N-N intervals) were used to calculate HR and time-domain measures of HRV: standard deviation of the normal-to-normal intervals (SDNN), an indicator of overall autonomic tone, and the square root of the mean squared differences of successive normal-to-normal intervals (RMSSD), an estimate of parasympathetic tone.

Statistical analyses
To determine exposure -related differences in cardiac and vascular indices we used mixed model analyses using SAS (Version 9.2, Cary, NC); this approach accounts for the longitudinal nature of the measurements on each animal. To reduce the skewness of the HRV measures, we natural logtransformed the SDNN and RMSSD after adding 1. Analyses comparing Air-vs PM 2.5 -exposed rats (Table 2), and the associations between PM 2.5 , EC and OC and cardiovascular responses (Tables 4 and 5) were derived from 1-hour samples of both PM and health effect responses during each of the four field exposure studies. Data for trace elements and carbon subfractions and their associations with health responses (Table 3, Figure 3, Additional file 1: Tables S1 and S2) were derived from 8-hour samples and integrated across the four field exposures (n = 32 samples). The criterion for significance was set at p ≤ 0.05 for all parameters.