Long-term Air Pollution Exposure Is Associated with Neuroinflammation,an Altered Innate Immune Response,Disruption of the Blood-Brain Barrier,Ultrafine Particulate Deposition,and Accumulation of Amyloid β-42 and α-Synuclein in Children and Young Adults

Abstract

Air pollution is a serious environmental problem. We investigated whether residency in cities with high air pollution is associated with neuroinflammation/neurodegeneration in healthy children and young adults who died suddenly. We measured mRNA cyclooxygenase-2, interleukin-1β, and CD14 in target brain regions from low (n = 12) or highly exposed residents (n = 35) aged 25.1 ± 1.5 years. Upregulation of cyclooxygenase-2, interleukin-1β, and CD14 in olfactory bulb, frontal cortex, substantia nigrae and vagus nerves; disruption of the blood-brain barrier; endothelial activation, oxidative stress, and inflammatory cell trafficking were seen in highly exposed subjects. Amyloid β42 (Aβ42) immunoreactivity was observed in 58.8% of apolipoprotein E (APOE) 3/3 < 25 y, and 100% of the APOE 4 subjects, whereas α-synuclein was seen in 23.5% of < 25 y subjects. Particulate material (PM) was seen in olfactory bulb neurons, and PM < 100 nm were observed in intraluminal erythrocytes from lung, frontal, and trigeminal ganglia capillaries.

Exposure to air pollution causes neuroinflammation, an altered brain innate immune response, and accumulation of Aβ42 and α-synuclein starting in childhood. Exposure to air pollution should be considered a risk factor for Alzheimer’s and Parkinson’s diseases, and carriers of the APOE 4 allele could have a higher risk of developing Alzheimer’s disease if they reside in a polluted environment.

Introduction

Air pollution is a complex and dynamic mixture of gases, particulate matter (PM), and organic compounds present in outdoor and indoor air. Exposure to air pollution is associated with respiratory, cardiovascular, and stroke-related sickness and death (Banauch et al. 2006; Brunekreef and Holgate 2002). Children living in Mexico City (MC) exhibit evidence of chronic inflammation of the upper and lower respiratory tracts, alterations in circulating inflammatory mediators, and breakdown of the nasal epithelial barrier (Calderón-Garcidueñas et al. 2001; Calderón-Garcidueñas, Franco-Lira et al. 2007; Calderón-Garcidueñas, Mora-Tiscareño et al. 2003). These children also have heart rhythm alterations and decreased vagal responses associated with sustained high plasma endothelin-1, a potent vasoconstrictor peptide involved in the homeostatic regulation of vascular smooth muscle tone, and upregulated after exposure to air pollutants including PM (Thomson et al. 2004, 2005; Calderón-Garcidueñas et al. 2006; Calderón-Garcidueñas, Vincent et al. 2007). Dogs exposed to the polluted environment in MC exhibit chronic respiratory tract inflammation; early expression of neuronal nuclear NFκB and endothelial/glial inducible nitric oxide synthase; disruption of the nasal and olfactory barriers and the blood-brain barrier (BBB); accumulation of amyloid β42 (Aβ42) in neurons; and increased olfactory bulb (OB) and hippocampal apurinic/apyrimidinic sites, indicators of oxidative DNA damage (Calderón-Garcidueñas et al. 2001, 2002; Calderón-Garcidueñas, Maronpot et al. 2003). Breakdown of the nasal respiratory and olfactory epithelium and the BBB facilitates the access of systemic inflammatory mediators and components of air pollution to the central nervous system (CNS) (Calderón-Garcidueñas et al. 2004). Chronic inflammatory processes in the CNS play an important role in the progressive neuronal death seen in neurodegenerative diseases such as Alzheimer’s (Akiyama et al. 2000; McGeer et al., 2006; Selkoe 2001, 2002). A coherent pathway linking exposure to air pollution and brain damage includes a chronic inflammatory process involving the respiratory tract, which results in a systemic inflammatory response with the production of inflammatory mediators capable of reaching the brain; continuous expression of crucial inflammatory mediators in the CNS at low levels; and the formation of reactive oxygen species (ROS) (Calderón-Garcidueñas et al. 2002, Calderón-Garcidueñas et al. 2004; Calderón-Garcidueñas, Maronpot et al. 2003; Calderón-Garcidueñas, Mora-Tiscareño et al. 2003). Ultrafine PM (UFPM), particulate-matter–associated lipopolysaccharides (PM-LPS), and metal uptake could take place through olfactory neurons, cranial nerves such as the trigeminal and vagus, the systemic circulation, and macrophage-like cells loaded with PM from the lungs (Calderón-Garcidueñas et al. 2001, 2002; 2004; Calderón-Garcidueñas, Maronpot et al. 2003; Calderón-Garcidueñas, Mora-Tiscareño et al. 2003). Activation of the brain innate immune responses could follow the interaction between circulating cytokines and constitutively expressed cytokine receptors located in endothelial brain capillary cells, followed by activation of cells involved in adaptive immunity (Nguyen et al. 2002; Simard and Rivest 2006). Monocytes are the main innate immune response mediator cells, producing and secreting TNFα, IL-6, and interleukin-1β (IL-1β), which in turn recruit and increase the activity of other immune cells (Simard and Rivest 2006). In the sustained upper and lower respiratory tract chronic inflammatory process elicited on exposure to significant concentrations of air pollutants in megacities such as MC, particularly fine and ultrafine PM could serve as the crucial trigger for a chain of events leading to endothelial activation, disruption of the BBB, altered response of the innate immune system, neuroinflammation, and neurodegeneration. We previously reported that adult residents of highly polluted urban areas, average age 54.7 ± 4.8 years, exhibit significantly higher expression of cyclooxygenase-2 (COX2)—a powerful inflammatory gene—in brain target areas when compared with matched age/gender/educational level subjects from cities with low pollution levels (Calderón-Garcidueñas et al. 2004). Highly exposed subjects also exhibited a significant neuronal and astrocytic accumulation of the 42 amino acid-isoform (Aβ42) of β amyloid, which is more hydrophobic and prone to aggregation than other Aβ isoforms (Selkoe 2001, 2002). Given that pollutant levels in MC vary within a relatively narrow range throughout the year, its residents are exposed all year long to a significant burden of air pollutants. The pollution levels have been sustained or have worsened in the past twenty years (Bravo-Alvarez and Torres-Jardón 2002), so the exposure of today’s children and teenagers is truly life long, as it began in utero. Moreover, there is a relatively low mobility of MC residents, so individuals tend to be exposed to the same environment for long periods, thus allowing for the opportunity to study chronic health effects associated with prolonged sustained exposure to severe air pollution.

The primary purpose of the present work was to measure by real-time polymerase chain reaction two key inflammatory genes, COX2 and IL-1β, and the LPS receptor CD14; this selection was based on the increasing evidence that neuroinflammatory processes contribute to the cascade of events that lead to neurodegeneration. These markers of neuroinflammation were measured in target brain areas including the OB, frontal cortex, hippocampus, substantia nigrae, periaqueductal gray, and vagus nerves in a cohort of cognitively intact Mexican children, adolescents, and young adults who died suddenly and were residents from low- or high-polluted urban areas in Mexico. Given that inflammatory responses involve the microvasculature and the trafficking of inflammatory cells, we also explore the integrity of the tight junctions in the brain capillaries, the nature of the inflammatory responsive cells, and the expression of endothelial inflammatory markers. We assessed zonula occludens-1 (ZO-1), a scaffolding protein marking tight and adherens junctions. Immune cells were identified immunohistochemically using antibodies to CD68, surface HLA-DR antigens, and CD163 (a macrophage scavenger receptor that identifies brain perivascular macrophages). Leukocyte adhesion molecules investigated included vascular adhesion molecule-1 (VCAM-1), and intercellular adhesion molecule-1 (ICAM-1). Since we have seen the transfer of UFPM from alveolar type I cells to the alveolar epithelial basement membrane, to endothelial cells, and finally to macrophage-like cells in the lumen of exposed MC dog lung capillaries (Calderón-Garcidueñas et al. 2001; Calderón-Garcidueñas, Franco-Lira et al. 2007), we did extensive electron microscopy in samples from the lungs and brains of both control and exposed MC subjects to look for PM. The accumulation of Aβ42 and α-synuclein was also investigated. The trigeminal ganglia were examined given the evidence by Lewis et al. of trigeminal uptake and clearance of inhaled manganese in rodents. In addition, the cohorts were genotyped for the APOE alleles and allelic frequencies of the Asp299Gly TLR4 polymorphism to determine if subjects had a known risk factor for Alzheimer’s disease (i.e., APOE ɛ4 allele carriers) and if they were capable of responding to lipopolysaccharides (one of the major organic components in MC PM).

Methods

Study Cities and Air Quality Data

We selected a large, polluted megacity and two control cities. Mexico City (MC) was the selected megacity, and Tlaxcala and Veracruz were the low-polluted cities. Mexico City represents an extreme of urban growth and environmental pollution (Bravo-Alvarez and Torres-Jardón 2002). Mexico City is a megacity that covers an area of 2000 km² surrounded by a series of volcanic and discontinuous mountain ranges that limit the natural ventilation of the basin. The basin has more than 30,000 industrial facilities and 4 million vehicles, with an estimated annual emission of 2.6 million tons of particulate and gaseous air pollutants. The critical air pollutants are ozone (O₃), and PM. The climatic conditions in MC are relatively stable through the seasons, thus air pollutant concentrations are relatively consistent. Residents in MC have been chronically exposed to significant concentrations of O₃, PM, and LPS for the past 2 decades. The marked increase in O₃ concentrations initially started in the fall of 1986, coinciding with the introduction of a new gasoline with lower tetraethyl lead concentration and higher levels of short-chain aliphatic hydrocarbons and aromatic compounds (Bravo-Alvarez and Torres-Jardón 2002). The change in gasoline composition led to an increase in reactive hydrocarbon emissions and O₃ ambient concentrations. By the end of 1989, in an apparent move to further reduce atmospheric carbon monoxide and hydrocarbon emissions, methyl-ter-butyl ether (MTBE) was introduced as an additive in gasoline. However, the use of MTBE in the absence of catalytic converters on motor vehicles led to a further increase in reactive hydrocarbons (i.e., isobutene and formaldehyde). As a result, O₃ production chemistry changed, leading to an additional rise in ambient O₃ concentrations. Around the same time, MC authorities imposed a regulation banning residents from driving cars on specific days of the week. This measure significantly increased total driving in MC, because many people bought an additional car because of the inefficient public transportation. The resulting effect of the greater use of old cars not equipped with catalytic converters, overcrowded streets, and increased weekend driving was a serious boost in vehicular emissions, which, combined with the use of MTBE in gasoline, led to very high O₃ levels that peaked in 1991 (Bravo-Alvarez and Torres-Jardón 2002). Citizen concerns about air pollution and car market pressure to introduce new cars equipped with catalytic converters forced authorities to consider the distribution of reformulated gasoline free of tetraethyl lead. As a consequence, beginning with 1991 car models, catalytic converters were required in Mexico, although the turnover rate to new converter-equipped cars was slow because of the weak economic situation. A slight reduction in O₃ ambient levels started in 1992. Additional measures, such as the vehicle emission inspection program and more strict control of hydrocarbon emissions from gas stations, helped to reduce O₃ levels. However, the growth of the population and the number of cars in MC, the continuing usage of MTBE in reformulated gasoline, and the very high levels of volatile organic compounds have slowed and delayed the reduction of O₃ to acceptable levels. A serious problem in MC is the contribution of aromatic compounds to secondary organic aerosol formation through atmospheric transformation and the formation of oxidation products that are partially absorbed into organic films on pre-existing PM_2.5. Concentrations of PM_2.5 and PM₁₀ in MC are above the current annual standards. Lipopolysaccharides (LPS) detected in PM₁₀ samples show a range of 15.3 to 20.6 nanograms per milligram of PM₁₀, and PM samples from South Mexico City show the highest endotoxin concentrations at 59 EU/mg PM₁₀ (Bonner et al. 1998). Mexico City has significant sources of environmental endotoxins, including open-field waste areas, waste disposal dust, waste water treatment plants, open sewer channels, and daily outdoor deposits of 500 metric tons of animal and human fecal material.

Control Cities

The control cities included Tlaxcala and Veracruz. Because of the combination of the relatively few contributing emission sources from industry and cars and the good ventilation conditions by the regional wind, criteria pollutants (O₃, PM₁₀, SO₂, NO₂, CO, and Pb) levels in control cities are below the current US standards. Three additional factors for the selection of the control cities included: (1) altitude above sea level similar to Mexico City (i.e., Tlaxcala); (2) dog necropsies from these cities have shown minimal pathology in lungs and hearts (Calderón-Garcidueñas et al. 2001); and (3) clinical studies in children in these cities have shown healthy children with no evidence of air-pollution–associated pathology (Calderón-Garcidueñas et al. 2003).

Autopsy Selection

The study protocol was approved by the Institutional Review Boards for Human Studies at the institutions involved. We studied 47 subjects from 2 cohorts of clinically healthy, cognitively and neurologically intact children and adults, ages two to forty-five years, with an average age of 25.1 ± 1.5 y. The control cohort included subjects from low-polluted cities (n = 12) and the exposed cohort (n = 35) from MC. The forty-seven subjects had complete autopsies and neuropathological examinations and were included in the immunohistochemistry (IHC) and the real-time polymerase chain reaction (RT-PCR) studies. Data available for all subjects included age, gender, place of birth, place of residency, occupation, smoking habits, clinical histories, cause of death, and time between death and autopsy. Cause of death was considered for all subjects to rule out the possibility that infection, inflammatory events, drug exposure, brain ischemia, and hypoxia might impact the mRNA levels of the inflammatory markers measured in the study. Therefore, the selected cohorts had no clinical history or pathological evidence of short- or long-term inflammatory processes, administration of drugs, anti-inflammatory medications, hormones, or events such as cerebral ischemia or epilepsy.

Necropsy and Tissue Preparation

Autopsies were performed 3.9 ± 1.1 hours after death. The postmortem period was similar for controls and pollution-exposed subjects. The skull was opened, and the OBs, trigeminal ganglia, and brain were removed. The right and left vagus nerves were exposed and dissected at the neck level, and a 10-cm section was cut along the OBs and selected areas from alternating right and left cerebral hemispheres, then quickly frozen and kept at −80°C. Frozen tissues for the RT-PCR were taken from the cortex and the white matter, taking care to make a perpendicular cut to the brain surface and keeping similar amounts of cortex and white matter for each method. In the midbrain section taken at the level of the superior colliculi, we dissected the substantia nigrae and the central grey stratum around the cerebral aqueduct. The right side was selected for the RT-PCR studies, and the left side was fixed in formaldehyde. Brain sections adjacent to the frozen material were immersed in 10% neutral formaldehyde, fixed for 48 hours, and transferred to 70% alcohol. Sections were taken from the OB, superior frontal gyrus, anteriomedial temporal lobe, hippocampus, basal ganglia, midbrain at the level of the superior colliculi, pons, medulla, neocerebellum, and trigeminal ganglia. Sections from lungs (upper-right lobe), peribronchial lymph nodes, heart (left and right ventricles), kidney, and liver were also taken. Paraffin sections 8 μm thick were cut and routinely stained with hematoxylin and eosin (H & E).

Immunohistochemistry (IHC) was performed on sections from the OB, frontal lobe, hippocampus, midbrain, pons, trigeminal ganglia, heart, and lungs. The sections were deparaffinized and immunostained as described previously (Calderón-Garcidueñas et al. 2004). Negative controls included omission or substitution of primary antibodies by nonspecific, isotype-matched antibodies. Positive and negative controls were included for each antibody. In double IHC, detection of β amyloid_1–42 was followed by GFAP staining, and CD163 was combined with glucose transporter type 1 (Glut-1). The brain histopathologic parameters evaluated included: vascular changes; the presence of histological elements characteristic of neuronal, glial necrosis, or apoptosis; and the distribution and characteristics of astrocytes. Sections were read blindly by one neuropathologist and one general pathologist with no access to the codes regarding the subjects’ data. Electron microscopy was performed in frontal, trigeminal ganglia, and lungs of control and MC samples. Samples were fixed in 2% paraformaldehyde, 2% glutaraldehyde in sodium phosphate buffer (0.1M, pH 7.4), post-fixed in 1% osmium tetraoxide, and embedded in Epon. Semithin sections (0.5–1 μm) were cut and then stained with toluidine blue for light microscopy examination. Ultrathin sections (60–90 nm) were cut and collected on slot grids previously covered with formvar membrane. Sections were stained with uranyl acetate and lead citrate and examined with a Carl Zeiss EM109T (Germany) or a JEM-1011 (Japan). For immunofluorescence staining, prior to staining, 10- to 20-μm paraffin-embedded tissue sections were dewaxed, rehydrated, and pretreated by incubation with warm (37°C) trypsin 0.1% in phosphate-buffered saline (PBS) with CaCl₂ (PBS-CaCl₂) for ten minutes. Sections were then washed in PBS and incubated with the primary antibodies overnight at 4°C. After washing, incubation with secondary antibodies was done for four hours at room temperature. Primary antibodies were diluted as follows in PBS with 0.5% BSA: rabbit anti-Glut-1, mouse CD163, VCAM-1, and ZO-1. Secondary antibody included goat antirabbit cyanine 5 and goat anti mouse Alexa fluor 488 and 568 at 1:100 (InVitrogen). Sections were mounted in PBS/glycerol (2:1) containing 170 mg/mL Mowiol 4–88 (Calbiochem, VWR International). For the confocal microscopy using the ZO-1 antibodies, we prepared smears of frontal fresh brain of seventeen cases, six controls, and eleven MC (APOE 3/3 and 4) subjects, fixed them in cold acetone for ten minutes, and air-dried the slides. Vessel diameters and tight juntion (TJ) abnormalities were assessed by two independent observers, and vessels were scored as normal or abnormal on the basis of the ZO-1 staining of their TJs. Selected areas with blood vessels were examined, and an average of one hundred vessels were visualized for the integrity of the ZO-1 staining. Fluorescence was examined using a BioRad Radiance 2000 laser scanning confocal on an inverted Nikon TE 300 microscope. Images were processed and visualized with LaserSharp software (version 2000, BioRad Microscience, Hertfordshire, UK).

Estimation of mRNA abundance was by real-time RT-PCR. Total RNA was extracted from frozen tissues including lungs, OB, frontal cortex, hippocampus, substantia nigrae, periaque-ductal grey, and vagus nerves, using Trizol Reagent (Invitrogen Corp, Carlsbad, CA) according to the manufacturer’s instructions. Random-primed first-strand cDNAs were generated as described (Calderón-Garcidueñas et al. 2004). Relative abundances of mRNAs encoding COX2, IL-1β, and CD14 were estimated by quantitative fluorogenic 5’ nuclease (TaqMan) assay of the first-strand cDNAs as described (Calderón-Garcidueñas et al. 2004). Primers and fluorophore-labeled TaqMan probes targeting human COX2, IL-1β, and CD14 were designed using Primer Designer software (Scientific and Educational Software, Durham, NC) based on sequence information in GenBank.

For Asp299Gly and APOE genotyping, Asp299Gly genotype was determined using an allelic discrimination assay protocol according to Applied Biosystems (ABI). The aspartic-acid-to-glycine change at residue 299 results from the substitution of an adenosine to glycine at nucleotide 896 from the start codon of the TLR4 cDNA.The portion of the TLR4 gene containing the polymorphism was amplified using the PCR on the ABI Prism 7700 instrument. For the APOE genotyping, DNA was isolated from the frontal cortex as described and genotyped for the HhaI restriction site polymorphism in the APOE gene.

Statistics

Statistics were performed using Stata statistical software (College Station, TX). We applied the parametric procedure that considers the differences among variances of the variables of interest—COX2, IL-1β, and CD14 mRNA abundance in controls and exposed subjects. Significance was assumed at p < .05. Data are expressed as mean values ± SD.

Results

Air Quality Data

Residents in Mexico City have been chronically exposed to significant concentrations of O₃ and PM for the past two decades (Figure 1). The climatic conditions in Mexico City are relatively stable, thus pollutants concentrations are consistent year after year. Figure 1A illustrates the long-term trend (1986–2006) of the number of exceedances per year of the eight-hour O₃ air quality standard (0.085 ppm over any eight-hour period, not to be exceeded in three years) average concentrations as well as their 90th and 50th percentiles for eight-hour averages determined for the whole Mexico City Metropolitan Area (MCMA). The higher eight-hour average O₃ concentrations coincide with the times children and teens are outdoors during the school recess and physical education periods as well as when they play outdoors at home (Villarreal-Calderón et al. 2002). Figure 1B shows the trends of the number of days above the PM₁₀ (10 μm or less in aerodynamic diameter) twenty-four-hour average air quality standard (150 μg/m³, not be exceeded more than once per year), the maximum of the daily PM₁₀ average concentrations, and the 50th percentile for 24-hour PM₁₀ concentration data registered in the whole MCMA from 1990 to 2006. Because of the existing high correlation between secondary organic aerosols and photochemical processes, PM₁₀ concentrations in Mexico City also tend to peak during the midafternoon hours, coinciding with children’s activities (Villarreal-Calderón et al. 2002). Figure 1C illustrates PM_2.5 (2.5 μm or less in aerodynamic diameter) twenty-four-hour and annual concentrations for five different regions in MC for the years 2003–2006. Residents of MC are exposed to concentrations of PM_2.5 above the standards year after year. The air pollutant data from MC were obtained from the MC Ambient Air Monitoring Network.

Study Population

The primary cause of death was accidents resulting in immediate death. The average age for the study cohorts of twelve controls and thirty-five highly exposed subjects was 26.4 ± 3.7 and 24.6 ± 1.6 years, respectively (p = .66) (Tables 1 and 2). The cohorts included thirteen children aged two to seventeen years (n = 4 in the control and n = 9 in the MC group), average age 13.2 ± 3.7 and 12.6 ± 1.7 respectively (p = .33), and within both cohorts there were twenty-three subjects younger than twenty-five years (Table 1). The occupations in both cohorts included elementary, middle, and high school as well as college students and blue- and white-collar workers. Based on the careful evaluation of the medical information available and the results of the autopsy, each subject was considered to be clinically healthy and cognitively and neurologically intact prior to his or her demise.

Real-time PCR mRNA Analysis of COX2, IL-1β , and CD14

Real-time, rapid-cycle PCR analysis of COX2, IL-1β, and CD14 in lungs, OB, frontal cortex, hippocampus, substantia nigrae, periaqueductal gray, and vagus nerves from 47 subjects indicated that the corresponding mRNA was present in each of the samples analyzed (Table 3). When samples were stratified according to the subject’s residency (MC vs. low-polluted cities), there was a significant difference in mRNA for COX2 in lung (p = 0.01), OB (p = .0002), frontal cortex (p = .008), substantia nigrae (p = .03), left vagus (p = .03), and right vagus (p = .0002), whereas it was not significant for hippocampus (p = .1) and periaqueductal gray (p = .1) (Table 3). When MC subjects were graphed by age for OB mRNA COX2 values, there were five subjects (APOE ɛ 3/3) identified with the highest mRNA COX2 values; these subjects ranged in age between seven and thirty-four years, and four of them were teenagers. Three of these teens already exhibited Aβ42 in their OBs, and one of them also exhibited α-synuclein. The youngest child with the high OB COX2 value did not have Aβ42 or α-synuclein in his OBs. Age graphs for the substantia nigrae pars compacta SNC/COX2 dataset showed a cluster of four subjects with the highest mRNA COX2 values ranging in age from two to forty-five years; three of these subjects, including an eleven-year-old boy, had α-synuclein in OB and/or neurons in brain stem nuclei. When frontal mRNA COX2 samples were graphed by age, the higher values were seen in subjects in the third decade and older. The higher values of lung COX2 were seen in eight MC subjects aged two to forty-five years; these subjects, with the exception of a two-year-old boy, exhibited significant deposition of PM in interstitial spaces, alveolar macrophages, and subpleural regions. For the vagus nerves, the subjects with the higher COX2 values for the left were different—except for one twenty-year-old male APOE ɛ 3/4 subject—from the subjects with the higher COX2 right vagus values, and two of the higher COX2 vagus subjects also had the higher IL-1β values. The subject with the higher COX2 values for the left vagus, a thirty-four-year-old male APOE ɛ 3/3, exhibited α-synuclein in the dorsal nucleus of the vagus and in neuronal groups in the pons and medulla. For IL-1β, the frontal cortex (p = .0002) and the OB (p = .003) were significantly higher in MC subjects vs. controls, whereas it was not significant for lung, hippocampus, substantia nigrae, periaqueductal gray, and vagus nerves. The higher IL-1β mRNA values for both OB and frontal cortex corresponded to teens and young adults. Significant upregulation of CD14 was present in the OB (p = .04), and the right (p = .02) and left (p = .01) vagus in MC subjects. The left vagus had the highest CD14 values across all subjects.

Clinical and Gross Pathological Observations

Non-CNS Findings

Tracheal epithelium showed patchy areas of squamous metaplasia and submucosal chronic inflammatory infiltrates in MC subjects over the age of twenty-five years. Nonperfused lungs from MC subjects displayed patchy clusters of alveolar macrophages filled with PM, bronchiolar smooth muscle cell hyperplasia, chronic mononuclear cell infiltrates, and macrophages filled with PM surrounding the bronchiolar walls and extending into adjacent vascular structures. In nine MC subjects there was extensive deposition of PM-laden macrophages in the subpleural regions along with mononuclear inflammatory infiltrates, smooth muscle cell hyperplasia of the pulmonary veins, and clusters of macrophages in the submucosa of the medium-sized bronchi. Peribronchial lymph nodes were grossly black in subjects over the age of twenty-five years and were loaded with PM (Figure 2A). Subjects from control cities exhibited small numbers of alveolar macrophages and rare foci of inflammatory cells in association with either terminal bronchioles or pulmonary blood vessels. Peribronchial lymph nodes showed small clusters of PM-containing macrophages. The higher lung values of mRNA COX2 were seen in eight of nine subjects, with the higher loads of PM in subpleural regions. Ganglion cells present in the bronchial walls, as well as Schwann cells in bronchial nerves, exhibit punctuate α-synuclein (Figures 2B and 2C). Nerve fibers in large bronchi also exhibit foci of mononuclear cells. Heart sections in MC residents showed clusters of perivascular partially degranulated mast cells, whereas nerve fibers on the epicardial surface exhibit positive α-synuclein punctuate pattern not seen in the control subjects. Liver sections from six MC residents showed PM in Kupffer cells (Figure 2D), and in macrophage-like cells in the portal spaces. These six subjects also had the most PM in their lungs. No liver abnormalities were seen in the control cohort.

Lung Electron Microscopy

One-micrometer–thick toluidine blue sections from MC teenagers and young adults showed neutrophils attached to alveolar capillary endothelial cells (Figure 2E). The alveolar walls exhibited collagen interstitial fibers, and the endothelial cells exhibited numerous fronds surrounding red blood cells (RBC) (Figure 2F). The RBC exhibited aggregation of particles along the cytoplasmic membrane and established discrete contacts with endothelial cell cytoplasmic vacuoles lined by particulate material (Figure 2G). Higher magnifications of RBC in lung capillaries revealed ultrafine PM (Figure 2H).

CNS Gross Findings

Gross brain examination was unremarkable in all subjects.

Brain Histopathology

For the olfactory nerve and bulb, four of the thirty-five MC subjects, including a fourteen-year-old boy, exhibited a significant amount of black PM in the cytoplasm of neuron-specific enolase (NSE)-positive cells at the glomerular region (Figure 3A). COX2 stained the cytoplasm of mitral and tufted neurons and olfactory ensheathing cells. Aβ42 was seen in ensheathing cells, astrocytes in the olfactory nerve, and in OB neurons in six of eighteen MC APOE ɛ 3/3 subjects younger than twenty-five years (Table 2), the youngest an eleven-year-old boy (Figure 3B). Corporae amylacea were numerous along the length of the olfactory nerves starting in the late teens. Reactive gliosis (GFAP-positive astrocytes) was present in all layers of the OB in all exposed subjects (including external and internal plexiform, mitral cell layers, and the olfactory glomeruli). Alpha-synuclein was present in the form of Lewy neurites, as well as granular punctuate cytoplasmic deposits in NSE-positive cells in the glomerular, mitral, and granular cell layers in four of eighteen MC subjects younger than twenty-five years old (Table 1, Figure 3C); the youngest was an eleven-year-old boy (Figure 3D). In the trigeminal ganglia and nerves, partially degranulated mast cells were seen in close proximity to the ganglion cells (Figure 3E). Perineurial blood vessels exhibited vacuolated endothelial cells and marginal WBCs.

Cortical Sections

As to vascular changes, teens exhibited significant amounts of lipofuscin in endothelial cortical capillaries cells. Perivascular hemosiderin-laden macrophages were seen around small venules and arterioles in both gray and white matter, the latter being foremost; these changes were already prominent in the eleven-year-old MC boy in this series. Intact RBCs inside macrophages were identified alongside the hemosiderin-laden macrophages (Figure 3F). In five MC subjects, cortical blood vessels exhibit platelet thrombi (Figure 3G). Exposed subjects exhibited positive prothrombin (PT) staining outside blood vessel walls, predominantly in the white matter (Figure 3 H). Perivascular macrophages as well as reactive astrocytes were positive for prothrombin in the proximity of blood vessels with PT outside the walls. No PT outside of blood vessels was seen in the controls. Clusters of mononuclear cells around blood vessels in the frontal and temporal cortex, subicular area, and the brain stem were a frequent finding in MC subjects regardless of age (Figure 4A). These mononuclear cells were positive for CD68, CD163, and HLA-DR. CD163-positive cells were present predominantly in perivascular locations in the cortex and to a lesser degree in the neuropil as activated positive microglia (Figure 4B). CD68 stained numerous white matter microglia-like cells in highly exposed individuals. Positive CD68 and HLA-DR cells were seen predominantly in the white matter (Figures 4C and 4D), and positive perivascular cells were seen in the cortex in MC subjects as young as two years of age. In MC subjects, numerous partially degranulated mast cells exhibit positive tryptase granules (Figure 4E), particulary in the white matter, whereas frontal neurons exhibited positive staining in their cytoplasm. In controls, however, only occasional tryptase positive perivascular cells were seen, and the neurons were negative. VCAM-1 strongly stained cortical endothelial cells in MC subjects (Figure 4F), whereas ICAM-1 was positive for astrocytes and microglia cells in both the cortex and the white matter. Nitrotyrosine (NT) positive cells were present in all exposed individuals. NT immunorectivity was present as diffuse cytoplasm neuronal staining in frontal neurons as well as inclusions in glial cells, including astrocytes and microglia. Abundant NT-positive, macrophage-like cells were seen in perivascular white matter locations (Figure 4G), as well in endothelial cells. Control subjects exhibited an occasional perivascular positive cell. NFκB was positive in the nuclei of endothelial cells in cortical capillaries (Figure 4H) and perivascular macrophages in MC residents. NFκB nuclear positivity was not seen in control subjects. iNOS-positive cells included astrocytes and neurons in cortical regions and the OB of MC residents. COX2 immunoreactivity was seen in neuronal cell bodies and dendrites, as well as endothelial cells of small capillaries and arterioles in the frontal cortex. Exposed subjects exhibited strong endothelial COX2 staining in both cortex and white matter. In control subjects, the staining was confined to neurons. 8-0HdG positivity was present predominantly in pyramidal frontal cells and, to a lesser degree, in astrocytes in the white matter of MC subjects. Astrocytes with a small amount of cytoplasm were seen around blood vessels and neurons in the frontal cortex; a few of these astrocytes were positive for GFAP. Patchy cortical GFAP-positive astrocytes were prominent in MC children (present in the youngest, two years old) and teens. GFAP-positive astrocytes increased in the cortex with age. Reactive astrocytes were focally prominent in subpial areas and perivascular deep white matter of all exposed individuals. Immunoreactivity for Aβ42 was seen in the cytoplasm of neurons in the frontal and temporal cortices, in the smooth muscle cells of cortical vessels, and in both diffuse and mature senile plaques. In MC residents, Aβ42 selectively accumulated in the perikaryon of pyramidal frontal neurons as discrete granules and was present in cortical and white matter astrocytes and in sub-arachnoid and cortical blood vessels. Neuronal Aβ42 was identified in APOE ɛ 3/3 children as young as eleven years old, whereas diffuse plaques were first seen in seventeen-year-olds (Figure 5A). Mature Aβ42 plaques were abundant in subjects in the fourth decade (Figure 5B). In the cohort of APOE ɛ 3/3 MC subjects under twenty-five years of age, nine of seventeen exhibited Aβ42 positivity in the frontal cortex (Table 1). In sharp contrast, in the cohort of APOE ɛ 4 MC subjects (Table 4), the four youngest subjects (ages fifteen, twenty, twenty-two, and twenty-five) all exhibited Aβ42 in the OB, blood vessels, cortical neurons, and/or in diffuse plaques. There were three controls who were heterozygous for APOE 4 (ages thirty-six, forty-four, and forty-five), and two of these subjects had Aβ42 immunoreactivity. None of the forty-seven subjects fulfilled morphological Alzheimer’s criteria as described in the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), Braak stages, and NIA-Reagan Institute criteria (Lilian Calderón-Garcidueñas, unpublished data).

Confocal Microscopy for Tight Junctional Abnormalities Zonula Occludens Ab (ZO-1)

The majority of examined vessels were < 100 μm in diameter. In the control group APOE β 3/3 (n = 4, age = 16 ± 4.88 years), there were 3.8 ± 1.08% of vessels with abnormal ZO-1 TJs. Mexico City subjects APOE β 3/3 (n = 8, age 13.25 ± 2.36 years) exhibited 30.8 ± 5.9%, whereas APOE β 4/4 and 3/4 (n = 5, including 2 controls) had 62.2 ± 7.36% of the vessels with discontinuous or punctuate staining in the frontal cortex (Figure 5C). There was a significant difference in the number of abnormal tight junctions between APOE β 3/3 controls and MC subjects (p = .01) and vs APOE β 3/4 (p = .0002), whereas there was also a significant difference between MC APOE β 3/3 vs. 3/4 (p = .007).

Brainstem: vascular changes in the brainstem, including the midbrain, were similar to the ones described for the neocor-tex. Exposed subjects exhibit significant VCAM-1 staining of endothelial cells in capillaries and small venous and arteriolar vessels in the midbrain. CD163 and HLA-DR strongly stained mononuclear perivascular cells, whereas CD68 stained microglia-like cells in the substantia nigrae pars compacta (SNC) region, superior colliculus, red nucleus, tegmental tract, and medial lemniscus. A few tryptase-stained perivascular cells were seen in some of the exposed subjects. There was a significant degranulation of SNC in subjects in their twenties and early thirties. The degranulation was accompanied by numerous macrophages loaded with melanin pigment around the degranulated neurons and in perivascular locations. These changes were also seen in the Asp299Gly TLR4 polymorphism subjects. The MC woman APOE ɛ 4/4 exhibited α-synuclein positivity in substantia nigrae neurons and mesencephalic V neurons and displayed significant degranulation of SNC pigmented cells with numerous melanin-laden macrophages. In MC residents, NT and 8-0HdG positivity were present in raphe neurons, mesencephalic V neurons, and glial cells in the medial raphe (Figure 5E). 8-0HdG-positive neurons were also seen in the trigeminal thalamic ventral tract. Alpha-synuclein granular cytoplasmic neuronal staining involved neurons in the trigeminal thalamic tract, mesencephalic V, reticular and raphe nuclei, the glossopharyngeal-vagus complexes, and the SNC (Figure 5F) in exposed subjects as young as seventeen years of age (Table 1). Perivascular macrophages with hemosiderin pigment and intact RBC were seen in capillaries throughout the brain stem, including the ones in the regions of cranial nerve nuclei (i.e., mesencephalic trigeminal neurons). Fibrin thrombi were seen at all levels of the brain stem in small blood vessels in exposed subjects.

Evaluation of APOE ɛ 3/4 and 4/4 Subjects

All MC APOE ɛ 4 either heterozygous or homozygous subjects had Aβ42 in neurons and blood vessels in the frontal cortex and the hippocampus, including the twenty-year-old male with a mutant TLR4 genotype (Table 4). Vascular changes, cortical reactive GFAP-positive astrocytes and white matter gliosis were more prominent than in the APOE ɛ 3/3 age-matched MC cohort. The only APOE ɛ 4/4 subject, a thirty-two-year-old MC woman, had scattered foci of perivascular monuclear cells in the hippocampi, as well as platelet thrombi in small blood vessels. This woman had Aβ42 in her OBs, cortical neurons, and cortical and subarach-noid blood vessels, in addition to α-synuclein immunoreactivity in the substantia nigrae and neurons from the mesencephalic trigeminal nerve. Subjects with the APOE ɛ 4 allele displayed ZO-1 discontinuous or punctuate staining in 62.2% of their vessels throughout the frontal cortex, in sharp contrast to the 30.8% in APOE ɛ 3/3 subjects (p = .007). Assessment of accumulation of Aβ42 and α-synuclein as a function of age and residency (Tables 5 and 6) showed that 58.8% of APOE ɛ 2/3, 3/3 subjects under the age of twenty-five who were residents of MC exhibit Aβ42 accumulation (average age 17.4 years), whereas in the same group 23.5% already had α-synuclein detectable by IHC. Accumulation of both Aβ42 and α-synuclein starts in the teen years in MC residents (Tables 5 and 6).

Electron Microscopy

Frontal capillaries exhibited RBC with ultrafine particles in their cytoplasm, along with aggregation of intramembrane UFPM and establishment of discrete contacts with endothelial cell cytoplasmic membranes (Figure 5G). Mononuclear cells established a close contact with the capillary outside the BBB, and UFPM was seen frequently at the interphase. Trigeminal ganglia capillary sections also revealed the presence of discrete contacts between the RBC and the endothelial cells and increased caveoli (Figure 5H with insert).

Discussion

Clinically healthy, cognitively and neurologically intact children, teenagers, and young adults with a lifetime exposure to significant concentrations of air pollutants including O₃, PM, and PM-LPS exhibit an upregulation of mRNA COX2, IL-1β, and a key innate immunity receptor CD14 in the OB, frontal cortex, substantia nigrae, and/or vagus nerves, as well as early disruption of the tight junctions in frontal blood vessels. These subjects exhibit nuclear NFκB in brain endothelial cells as well as evidence of an activated inflamed cerebral endothelium, with an altered BBB and trafficking of inflammatory cells in perivascular areas and in the neuropil. The OB mRNA COX2 upregulation, the frontal disruption of the BBB, and the endothelial nuclear NFκB are early key findings in the highly exposed cohort. Inflammatory cell trafficking and Aβ42 accumulation in the OB and frontal cortex are seen in prepuberal children with no known risk factors for Alzheimer’s disease. Alpha-synuclein Lewy neurites and punctuate α-synuclein neuronal accumulation are seen in the OB in children as young as eleven years of age, and in teens and young adults the α-synuclein immunoreactivity was also identified in the dorsal nucleus of the vagus, mesencephalic V, trigeminal thalamic tract, substantia nigrae, and in lung and heart autonomic ganglia and nerves.

The presence of PM in olfactory bulb neurons, in luminal erythrocytes from capillaries in lung, frontal and trigeminal ganglia, and in Kupffer cells, along with the translocation of UFPM from RBC to endothelial cells in capillary lungs, and from RBC to endothelial cells and to perivascular macrophage-like cells in frontal capillaries are very important observations in these highly exposed individuals. The fact that PM is directly reaching the brain parenchyma, along with the early disruption of the BBB and the vagal upregulation of CD14 capable of activating inflammatory processes in the brain stem, are key findings that need to be analyzed in terms of their potential impact on neuroinflammation and neurodegeneration.

There has been a growing interest in the identification of fine and ultrafine PM in urban air and their health effects (Donaldson 2003; Oberdorster et al. 2002), as well as how these particles reach the brain (Dorman et al. 2002; Henriksson et al. 1997). Moreover, neurodegenerative effects have been reported in experimental animals using UFPM (Block et al. 2004; Peters et al. 2006), and in dogs and human beings exposed to urban environments (Calderón-Garcidueñas et al. 2002; Calderón-Garcidueñas, Maronpot et al. 2003; Peters et al. 2006). Fine and ultrafine PM exhibit biological activities that are detrimental to cells, including induction of oxidative stress with the consequent depletion of cell antioxidants, direct cytoxicity including mitochondrial dysfunction and altered phagocytic function, alteration of cell signaling pathways, and DNA and lipid damage (Donaldson 2003). Portals of entry of PM are of utmost importance in highly exposed subjects in MC, particularly children, since we have documented breakdown of the nasal barrier with significant accumulation of PM in and around nasal epithelial cells (Calderón-Garcidueñas et al. 2001; Calderón-Garcidueñas, Franco-Lira et al. 2007) and the transport of metals associated with PM to OB neurons (Calderón-Garcidueñas, Maronpot et al. 2003). Factors such as age, gender, weight, race, nostril shape, exercise level, minute ventilation, and outdoor time all contribute to the particle deposition and to lesser or higher risk from inhalation of pollutant PM in ambient air (Bennett et al. 2005; Villarreal-Calderón et al. 2002).

Early disruption of the BBB and translocation of UFPM likely contribute to damage of the BBB. An intact BBB is necessary for the proper functioning of the CNS by actively controlling cellular and molecular trafficking between the systemic circulation and the brain parenchyma (Abbott 2005). Brain capillaries represent the largest surface area blood–CNS interface where tight intercellular junctions constitute the morphological basis of the BBB (Lossinsky et al. 2004). The issue of a damaged BBB is important, since this barrier has the ability to respond to LPS, IL-1β, TNFα, and IL-6 (Nadeau and Rivest 1999; Rivest 2001). LPS and IL-1β upregulate adhesion molecules, increase leukocyte migration across the CNS endothelial cells, and regulate BBB permeability (Hickey 2001; Rothwell and Luheshi 2000), whereas TNFα and IL-6 disrupt the BBB through the release of endothelial nitric oxide or, in the case of a transgenic animal overexpressing IL-6, the lack of BBB development (Brett et al. 1995; Farkas et al. 2006). Clinically healthy children in MC have evidence of systemic inflammation with increased sustained levels of prostaglandin E metabolite, IL-6, IL-1β, and a systemic response to their LPS-PM exposure through upregulation of mCD14 and two transporting LPS proteins: lactoferrin and heat shock protein 60 (Calderón-Garcidueñas Mora Tiscareño et al. 2003; Calderón-Garcidueñas, Franco-Lira et al. 2007; Calderón-Garcidueñas, Vincent et al. 2007). Since brain blood vessels express receptors for TNFα, IL-1β, and IL-6 (Nadeau and Rivest 1999), and TNFα and IL-1β can evoke expression of inflammatory mediator genes, such as COX2 (Rivest 2001) within the brain capillary endothelium, our findings of an early BBB disruption and CD14 upregulation suggest that systemic cytokines could be key early CNS vascular aggressors. Moreover, circulating cytokines can gain access to the brain by being transported across the BBB (Nguyen et al. 2002; Rivest 2001; Pan and Kastin 2001) and are able to evoke additional inflammatory mediator expression by vascular-associated microglia (Griffin et al. 2002), further increasing the permeability of the BBB (Blamire et al. 2000). Systemic and local brain production of cytokines are implicated in contributing to the initiation, propagation, and regulation of immune and inflammatory circuits (Benveniste 1998; Cunningham et al. 2005). IL-1β is the most important molecule capable of modulating cerebral functions during systemic and localized inflammation (Ferrari et al. 2006; Griffin et al. 2002; Rothwell and Luheshi 2000). Zhang and Rivest proposed that circulating LPS and cytokines could bind to their cognate receptors onto endothelial and/or monocytic cells lining the BBB, which in turn will lead to proinflammatory signaling and transcription of the receptors for different proinflammatory ligands that can stimulate NFκB kinases and mitogen-activated protein (MAP) and the enzymes responsible for PGE2 formation in the cerebral tissue (Zhang et al. 2003). Zhang and Rivest proposed responses to systemic immune stimuli likely apply to our chronically air-pollution–exposed subjects.

A critical finding is the endothelial nuclear NFκB activation present in the brain capillaries of young exposed subjects. NFκB activation depends on varied stimuli such as cytokines, LPS, and DNA damage (Pahl 1999); activation is tightly regulated and quickly shortened through feedback inhibition following the initial activating stimulus (Xiao et al. 2006). However, persistant activation (i.e., continuous exposure to significant levels of cytokines, UFPM, and/or PMLPS) results in deleterious effects.

Once the BBB is disrupted, significant leaking of RBC and proteins such as prothrombin may follow. There is an increment in the number of perivascular macrophages and microglia that expresses CD163, a scavenger receptor mediating the removal of hemoglobin-heptaglobin complexes, that is increased in inflammatory disorders (Kim et al. 2006). CD163 perivascular macrophages were common in the deep frontal and temporal white matter of MC subjects. Concomitantly with the increment in CD163, immunoreactivity for CD68 and HLA-DR in microglia, perivascular macrophages and endothelial cells were observed, in keeping with the inflammatory response. Intact and degranulated mast cells identified by means of tryptase monoclonal antibodies were seen in perivascular locations in frontal and temporal cortices, as well in trigeminal ganglia, and in peripheral autonomic nerves innervating the lungs and hearts in MC subjects, whereas in the control subjects mast cells were very rare and intact. Mast cells in the brain are normally observed in small numbers around the third ventricle, thalamus, hypothalamus, and meninges, and in the peripheral nervous system in association with inflammatory processes (Dropp 1979; Theoharides 1990). Mediators released by activated mast cells contribute to local inflammatory responses, regulating BBB permeability and angiogenesis and playing an active role in neuroinflammation (Ibrahim et al. 1996). More importantly, their presence in the context of the disruption of the BBB relates to their arrival in the CNS and PNS via the bloodstream following the trafficking of other inflammatory cells (Ibrahim et al. 1996). The identification of prothrombin in extravascular spaces and perivascular macrophages is a crucial finding in keeping with the BBB disruption (Mhatre et al. 2006), and it could be a contributing factor in the increased apolipoprotein immunoreactivity observed in MC dogs (Calderón-Garcidueñas et al. 2002), and as described by Mhatre et al. in a rat model of intraventricular infusion of prothrombin (Mhatre et al. 2006). The seminal work of Grammas et al. has shown that neurotoxic thrombin and inflammatory proteins are elevated in AD microvessels, a finding that is very relevant to our work. Rupture of the vascular basement membrane and leakage of prothrombin are described in the prefrontal cortex of Alzheimer’s patients (Zipser et al. 2007).

Perivascular mononuclear cells are active and efficient antigen-presenting cells (Lossinsky et al. 2004). In a healthy brain the endothelial cells express very low levels of adhesion molecules required for leukocyte emigration (Lossinsky et al. 2004), a central pathogenic event in CNS inflammation (Hickey 2001). Leukocyte adhesion to endothelial cells is a crucial step to facilitate selective and effective capture of leukocytes (Hickey 2001), and for leukocytes to cross the BBB, they must first roll along the luminal endothelial cell (EC) surfaces to establish the initial cell–cell communication (Abbott 2005; Hickey 2001; Lossinky et al. 2004). In in vitro adhesion assays, binding of lymphocytes to inflamed brain vessels is mainly mediated by leukocyte function-associated antigen-1 and intracellular adhesion molecule-1, the late-activation antigen-4, and the vascular cell adhesion molecule-1 (Hickey 2001). EC activation is seen after the injection of TNFα and LPS (Nadeau and Rivest 1999; Pan and Kastin 2001; Rivest 2001), the latter of which represents an important component of PM in MC. In MC subjects including children, the luminal EC exhibit strong immunoreactivity for adhesion molecules such as VCAM-1 and ICAM-1 both in supra- and infratentorial regions. In keeping with the EC activation, two critical observations were described in this human study: the presence of UFPM in RBCs and the aggregation of intramembrane particles with the formation of patterned discrete contact points between endothelial cells and RBCs in the CNS, trigeminal ganglia, and lung capillaries of highly exposed people. The establishment of contact points between ECs and RBCs could represent a pathway for the exchange of PM between the activated endothelial cell and the UFPM-loaded RBCs, in keeping with the capacity of ultrafine PM to penetrate RBC, as elegantly shown by Geiser and Rothen-Rutishauser (Geiser et al. 2005; Rothen-Rutishauser et al. 2006). Ultrafine particles are not membrane bound, which allows for direct access to intracellular proteins, organelles, and DNA, enhancing their toxic potential (Geiser et al. 2005). Further, the passage of PM to the brain following the RBC-activated EC is likely to be increased in subjects exposed to pollutants owing to the disruption of the BBB as described in previous lines, and it is likely related to the production of NO (Calderón-Garcidueñas et al. 2002; Thiel and Audus 2001). Plasmodium infected RBCs induce endothelial upregulation of ICAM-1 and give rise to endothelial cell microvilli or cytoplasmic fronds that touch the infected RBCs (Tripathi et al. 2006). The formation of endothelial fronds surrounding the RBC in the malaria-infected model is remarkably similar to our findings in the lung and trigeminal ganglia capillaries. Of utmost importance, the endothelial fronds/microvilli in the malaria model interfered with blood flow even after lysis of the infected RBCs (Tripathi et al. 2006), an indication that in our subjects the endothelial frond formation could account for a decreased blood flow in the involved areas.

Breakdown of the nasal barrier in pollution-exposed subjects may also contribute to brain inflammation by increasing the access of PM to the brain through the olfactory and trigeminal pathways. The finding of PM in the glomerular region of the OBs of MC residents indicates that particles are readily transported from the nasal cavity to the brain via the olfactory nerve, a pathway very well known in experimental animals exposed to metals (Dorman et al. 2002; Henriksson et al. 1997). Moreover, there is an early and significant upregulation of COX2, IL-1β, and CD14 in the OB, which is indicative of an ongoing inflammatory process. Further, the accumulation of Aβ42 and α-synuclein in the OB is associated with significant upregulation of mRNA COX2, whereas the presence of α-synuclein in the brain stem is related to COX2 upregulation in the substantia nigrae. The upregulation of COX2 is particularly relevant in these subjects already exhibiting Aβ42 accumulation, since COX2 potentiates β-amyloid peptide generation through alterations in γ secretase activity and apoptotic cell death and is indeed associated with the accumulation of Aβ42 (Qin et al. 2003; Xiang et al. 2002) and α-synuclein (Jellinger 2003). Both Aβ42 and α-synuclein are proteins capable of aggregation and misfolding, leading to progressive neurodegeneration that develops insidiously over the lifetime of the individual (Jellinger 2003; McGeer et al. 2006; Nguyen et al. 2002; Selkoe 2001, 2002). Given that axons from the olfactory sensory neurons project to the OB, and the primary axons of the projection neurons send off collateral branches to the olfactory nuclei, piriform cortex, entorhinal cortex, and amygdaloid nuclei, and then to the hippocampal formation and the parahippocampal gyrus, it is expected that significant inflammatory process in the OB in these highly exposed subjects may translate into olfactory dysfunction, which is indeed the case in young adults (L. Calderón-Garcidueñas and M. Franco-Lira, pers. comm. 2007). Olfactory dysfunction is an early clinical finding in several neurodegenerative disorders, including Alzheimer and Parkinson’s diseases (Hawkes 2003).

We have shown that the OB and the substantia nigrae are early targets of air pollution in young people, since the greater upregulation of mRNA COX2 was documented in teens and young adults. Alpha-synuclein accumulated as Lewy neurites and/or punctuate deposits in the OB, trigeminal thalamic tract, mesencephalic V, reticular and raphe nuclei, the glossopharyngeal-vagus complexes, and lung and heart autonomic ganglia in subjects as young as eleven years of age for the OB and the lung ganglion cells, and seventeen years for the brain stem findings. The brain stem findings in teens and young adults bring up three crucial issues: (1) the role the vagus nerves play in the brain stem inflammation development; (2) the accumulation of α-synuclein in target areas as a risk factor for the development of Parkinson’s disease in exposed populations; and (3) the accumulation of α-synuclein as a neuroprotective or neurotoxic effect. We know that systemic cytokines could affect the CNS via sensory nerves such as the vagus. This could be a consequence of exposure to air pollutants, because IL-1β is recognized by chemosensory receptors located in vagal paraganglia in the vagus nerve at several levels, including the cervical, thoracic, and abdominal regions (Elmquist et al. 1997). Activation of the peripheral immune system drives viscerosensory pathways originating in the brain stem nucleus of the solitary tract and ventrolateral medulla in response to cytokine signaling in the vagus nerve (Elmquist et al. 1997). The vagus and glosopharyngeal nerves, with their chemosensitive afferent fibers, are major neural pathways that establish communication between the immune system and the brain, generating responses to proinflammatory mediators (Elmquist et al. 1997). Moreover, indirect activation of the vagus nerve can be accomplished by paraganglia activation (Goehler et al. 1999). Paraganglia are located in the thorax and abdomen and are positioned to sense immune products released in lymph nodes and visceral organs, and they express binding receptors for IL-1β (Goehler et al. 1999). Indeed, the vagus nerves play a role in the lung inflammation caused by diesel-soot rodent exposures (McQueen et al. 2007), further emphasizing their potential to extend the inflammatory response into the brain stem. The upregulation of COX2 in the vagus nerves was an expected finding, given the extensive innervation coverage of the vagus nerves in target organs exposed to PM and endotoxins (i.e., lung, heart, and liver) (Kukanova and Mravec 2006; Uyama et al. 2004). Moreover, the presence of PM in Kupffer cells—the liver-resident macrophages (Uyama et al. 2004; Wake et al. 1989) responsible for the clearance of foreign material arriving from the circulation and the gut (Wake et al. 1989)—is a very interesting finding, given that the liver and the digestive tract are preferred sites for extrapulmonary translocation of ultrafine particles in human beings and rats (Oberdorster et al. 2002). Thus, the mRNA COX2 right vagus nerve’s significant upregulation when compared to the left in MC residents could be an indication of the role the liver plays in the detoxification and clearance of foreign and altered-self substances including PM and LPS-PM in the parenchyma (Nolan 1975). The issue is important from the clinical point of view, since in Parkinson’s disease there is substantial asymmetry of symptoms from the onset, with a marked preference for the right side (Djaldetti et al. 2006). Thus, the asymmetry could be explained for the PM factor, by the type and size of particles people are exposed to, and their fate in the lung and extrarespiratory anatomical areas (Daigle et al. 2003). Thus, in MC residents with a disrupted and ineffective nasal barrier, major concentrations of PM would be swallowed and thus enter the digestive system, the liver pathway, and the right vagus. Interestingly, whereas the substantia nigrae in the highly exposed subjects exhibited an upregulation of COX2 (p = .03), IL-1β did not reach significance (p = .06). In contrast, the periventricular grey adjacent to the SNC but not involved in the same neural pathways did not show upregulation of any of the selected inflammatory genes. Based on these findings we concluded that in subjects exposed to air pollution, the brain stem is taking part in the inflammatory process, either through local pathways or systemic inflammation or both, and the brain stem participation likely also depends on the PM entry pathways (i.e., digestive and lower respiratory systems).

Alpha-synuclein—an abundant brain 140 residue protein—is the culprit in Parkinson’s disease (PD) (Braak et al. 2003; Eriksen et al. 2003; Fink 2006; Jellinger 2003). Synucleins are developmentally expressed, and α-synuclein is present in presynaptic terminals and in both soluble and membrane-associated brain fractions (Fink 2006; Eriksen et al. 2003; Jellinger 2003). Substantial evidence suggests that α-synuclein aggregation is a critical step in PD and other synucleopathies (Fink 2006; Jellinger 2003), and a pathway going from normal soluble to abnormal misfolded filamentous proteins is a key process regardless of the primary disorder (Fink 2006; Jellinger 2003). Factors affecting the kinetics of α-synuclein fibrillation include oxidative stress, pesticides, metals, glycosaminoglycans, lipids, and macromolecular crowding (Fink 2006). Linse et al. demonstrated that nanoparticles enhance the rate of protein fibrillation by decreasing the lag time for nucleation, a novel mechanism that could be applicable to both Aβ42 and α-synuclein in the scenario of air pollution. Oxidative stress is present in the brain stem of these MC subjects, as evidenced by the presence of cells positive for 8-OHdG and NT and the upregulation of COX2 in the substantia nigrae, which could result in the production of ROS (Choi et al. 2006; Minghetti 2005). Early and sustained production of COX2 results in the production of free radicals in the process of converting arachidonic acid to precursors of vasoactive prostaglandins (Choi et al. 2006; Minghetti 2005). The presence of aggregated α-synuclein in target CNS and PNS regions in these high-air-pollution–exposed young cohorts follows the characteristic topographical distribution of early PD lesions, as described by Braak and colleagues (Braak et al. 2003, 2006; Del Tredeci et al. 2002), that is, OB, lower brain stem, and ganglionic autonomic cells. These prepuberal teens and young adults can be identified by their aggregated α-synuclein as being in Braak stages 1 and 2, the presymptomatic PD stage. Recent observations among MC pediatric cardiologists of an increased number of otherwise healthy children with syncope could offer evidence of heart autonomic and lower brain stem involvement as we have shown here (Dr. Alfredo Bobadilla-Aguirre, pers. comm., November 14, 2006). We strongly suggest that the Braak et al. proposal (2003) about a “putative environmental pathogen capable of passing the gastric epithelial lining might induce α-synuclein misfolding and aggregation” could indeed be related to PM gaining access to the brain through the respiratory and gastrointestinal vagus pathway in subjects exposed to significant PM concentrations for long periods of time. A controversial issue has to be addressed in this scenario: is the aggregation of α-synuclein neuroprotective or toxic in these young subjects? (Quilty et al. 2006; Sidhu et al. 2004) It appears that in tissue culture and with relatively low levels of oxidative stress, increased α-synuclein offers neuroprotection (Quilty et al. 2006), and it is also clear that the level of expression is crucial to confer either protection or toxicity (Sidhu et al. 2004). Given that the potential factors (oxidative stress, COX2 upregulation, vascular inflammation, nanoparticles) likely accounting for the aggregation of α-synuclein in these air-pollution–exposed subjects are both intense and prolonged, we favor the idea that α-synuclein acquires neurotoxic properties in these cohorts.

Chronic oxidative stress is a major contributing factor in the pathogenesis of both Alzheimer’s and Parkinson’s diseases (Nunomura et al. 2006; Quilty et al. 2006). We previously described significant oxidative DNA damage (genomic DNA apurinic/apyrimidinic sites) in both MC dogs and human beings in the OB and the frontal cortex (Calderón-Garcidueñas et al. 2002; Calderón-Garcidueñas, Maronpot et al. 2003). Our human data support previous work from Nunomura et al., Forero et al., and Zhu et al. stating that oxidative stress is early and precedes neuropathological manifestations of AD. We could add that brain oxidative stress starts in childhood and the teen years and is accompanied by accumulation of both Aβ42 and α-synuclein in the scenario of air pollution exposure.

Apolipoprotein E is the susceptibility gene with the clearest link to late-onset Alzheimer’s disease, although the ɛ 4 genotype alone is insufficient to predict an individual’s risk for AD (Wishart et al. 2006). In this work we have shown that carriers of an ɛ 4 allele residing in MC accelerate their Aβ42 accumulation by one decade compared to 3/3 carriers (Table 5). On the other hand, there is Aβ42 brain immunoreactivity in 58.8% of young MC residents (17.41 ± 1.5 years) with ɛ 3 alleles, and 80% in the older cohort (>25 years [35.2 ± 1.9 years]). These results suggest that cumulative exposures and age are key factors. It is interesting to point to this observation, because the deposits of Aβ42 start in childhood and the teen years in MC residents, and we know that these subjects do not fulfill current morphological AD criteria, thus these younger years constitute a time frame that is important in terms of pharmacological protection of our exposed populations. Alpha-synuclein accumulation also starts in the teens in highly exposed subjects, moreover, as expected we found no differences in APOE 4 carriers (Table 6).

Given that all ɛ 4 MC subjects—including the fifteen-year-old boy—had accumulation of Aβ42 obligates us to entertain the possibility that in the context of exposure to severe air pollution, the presence of an ɛ 4 accelerates the AD-like pathology. Since regional brain atrophy in the right medial temporal and bilateral frontotemporal regions (Wishart et al. 2006) and altered fractional anisotrophy (Persson et al. 2006)—a marker of white matter integrity—have been described in cognitively intact adults with homozygous or heterozygous ɛ 4 status, it is possible than MC APOE ɛ 4 teens already display similar alterations, a major issue because maturation of white-matter pathways is crucial in cognitive, behavioral, emotional, and motor development during childhood and the teen years.

There is no doubt that some subjects in this highly exposed MC cohort responded mainly with supratentorial pathology, deposition of Aβ42, and formation of diffuse amyloid plaques, whereas for others the response was mainly infratentorial, with the substantia nigrae and the vagus upregulation of COX2 and the deposition of α-synuclein in brain stem nuclei. A small group of subjects displayed both supra- and infratentorial pathology. This pattern of findings pointing toward Alzheimer’s and/or Parkinson’s-like pathology resembles the distribution of AD/PD patients with overlaps (Kurosinski et al. 2002). Given the common denominators between these two major neurode-generative diseases, our findings are expected.

The age issue of detection of Aβ42 in teens is important, given that higher mRNA IL-1β expression is seen in frontal cortex in teens and young adults under the age of twenty-five, whereas the higher values for mRNA COX2 were seen in adults over the age of thirty. Thus, teens’ frontal cortices exhibit early IL-1β upregulation. Cytokines play a central role in the self-propagation of neuroinflammation, with IL-1β having a prominent function (Minghetti 2005). The early upregulation of IL-1β in the frontal lobe of young subjects is of utmost importance, because this proinflammatory cytokine has been associated with BBB disruption; recruitment of inflammatory cells into the CNS (Ferrari et al. 2006); sustained upregulation of IL-8, VCAM-1, and ICAM-1 in astrocytes (Moynagh 2005); and with neuronal, glial, and endothelial injury through strong activation of the classical IL-1 signaling pathway, activation of MAPKs and NFκB (reviewed in Allan et al. 2005). Equally crucial is the role of IL-1β in the transformation of diffuse Aβ in mature plaques (Akiyama et al. 2000). NFκB is a crucial mediator in the IL-1β signal, and NFκB activation is sustained in astrocytes in response to IL-1 stimulation (Moynagh 2005).

The upregulation of CD14 may significantly contribute to the neuroinflammatory response in air-pollution–exposed subjects, particularly those exposed to significant amounts of PM-associated LPS, hence the upregulation of mRNA CD14 in the OBs and vagal nerves could indicate a brisk response of the innate immune system to LPS. Interestingly, although mCOX2 was significantly upregulated in the right vagus (p =.0002) versus the left (p = .03), CD14 expression was less, presumably reflecting the liver capacity to inactivate LPS. Fassbender et al. showed that CD14 binds Aβ and mediates Aβ-induced microglial and monocytic activation and toxicity for neurons. Further, they demostrated CD14 in Alzheimer’s brains but not in control subjects by immunohistochemistry (Liu et al. 2005). Letiembre et al. showed an altered regulation of innate immune receptors in older nondemented people. Given that TLR4 is necessary to engage the innate immune responses in the brain (Fassbender et al. 2004; Letiembre et al. 2007; Liu et al. 2005), we had hypothesized that TLR4-mutant subjects will have fewer brain inflammatory responses. However, that was not the case; trafficking of inflammatory cells and accumulation of Aβ42 were also observed in the 3 TLR4-mutant subjects, thus suggesting that factors other than TLR4 are also playing a crucial role in cell trafficking and amyloid accumulation observed in these megacity pollution-exposed subjects.

We would like to propose that sustained exposures to significant levels of air pollutants including UFPM, PM_2.5, and PM-LPS produce brain neuroinflammation and neurode-generation through at least four pathways.

1. Induction of upper respiratory, lung epithelial, and endothelial injury leading to persistant chronic inflammation in the respiratory tract and systemic inflammation. The systemic imflammation is accompanied by the production of pro-inflammatory cytokines such as TNFα, IL-6 and IL-1β (all of which are upregulated in MC children), for which brain blood vessels exhibit constitutive and induced expression of receptors. These cytokines can activate endothelial cells in the BBB, disrupt the BBB (early findings in highly exposed subjects and dogs), upregulate COX2 (target brain regions in MC subjects), and trigger cascades leading to activation of MAP kinases/NFκB (nuclear transduction of NFκB in endothelial brain cells in exposed subjects). A high level of activation of NFκB in astrocytes results in increased expression of nitric oxide synthase (seen in three-month-old MC dogs), and nitric oxide production that opens the BBB (seen in dogs, children, and teens). The early disruption of the BBB is followed by leaking of RBC and proteins such as prothrombin (neurotoxic protease, increases APP) (Grammas et al. 2006; Mhatre et al. 2006; Zipser et al. 2007) and trafficking of inflammatory cells expressing CD163, CD68, and HLA-DR, as well as mast cells in keeping with the inflammatory response. A dysregulated inflammatory response involves neural-immune interactions including the upregulation of CD14 that further activate immune cells, glial cells, and neurons (Fassbender et al. 2004; Letiembre et al. 2007; Liu et al. 2005; Minghetti 2005). Chronic oxidative stress is an early component of the brain responses, as evidenced by DNA oxidative damage present in different target brain areas as well as outside the CNS (Calderón-Garcidueñas, Maronpot et al. 2003).

2. We strongly support the importance of the olfactory pathway, especially in children, since olfactory neurons are loaded with PM and a strong early upregulation of COX2, IL-1β, and CD14 is present. Early damage to the OB and its connections and the accumulation of Aβ42 and α-synuclein will potentially translate into an abnormality in the limbic system, including the hippocampus and the parahippocampal gyri, as well as a decrease in the number of stem cells existing in the OB (Bédard et al. 2004).

3. The vagus/trigeminal (Lewis et al. 2005) pathways are also crucial, given that PM enters the respiratory and digestive systems (i.e., the liver). PM-LPS is likely to play an important role in these pathways, as shown by the vagal upregulation of CD14.

4. Direct access of UFPM to the brain, further accentuating an inflammatory response in the brain parenchyma (ROS production in activated microglia and perivascular macrophages), damaging components of the BBB, and potentially enhancing the rate of protein fibrillation affecting Aβ42 and α-synuclein (Linse et al. 2007).

All four pathways are clearly demonstrated in MC children, teens, and young adults, and based on the long-standing view that chronic inflammation, altered innate immune responses, and oxidative stress are detrimental, we propose that inflammatory interactions that take place at the blood-endothelium interface along with early oxidative stress are the bases for the early Alzheimer’s and Parkinson’s-like changes we observed in these populations. Oxidative stress in AD and PD is involved at the earlier stages of the pathological cascades with aggregation of the target proteins: amyloid β, tau and α-synuclein as an initial compensatory response (Eriksen et al. 2003; Fink 2006; Forero et al. 2006; Minghetti 2005; Nunomura et al. 2006; Quilty et al. 2006; Sidhu et al. 2004; Zhu et al. 2006). Persistance of the initial damaging factors translates into a neurodegenerative response.

In summary, exposure to significant concentrations of air pollutants including UFPM and PM_2.5 produces neuroinflammation and altered innate immune responses in crucial brain target anatomical areas in children and young adults. Ultrafine PM could play a role in the enhancement rate of protein fibrillation affecting Aβ42 and α-synuclein (Linse et al. 2007). We strongly propose that neuroinflammation as a result of exposure to air pollution could have a causative role in both Alzheimer’s and Parkinson’s diseases and that sustained brain inflammation confers a higher risk for the development of these two frequent neurodegenerative disorders. In the United States, 158 million people live in areas where O₃ exceeds the eight-hour standard, 29 million are exposed to PM₁₀ , and 88 million are exposed to PM_2.5 (http://www.epa.gov/oar/oaqps/greenbk/03co.html). Neuroinflammation provides a mechanistic link between inhalation/ingestion of air pollutants and neurodegeneration as seen in AD and PD. Neuroinflammation and accumulation of Aβ42 and α-synuclein in key target brain areas start in healthy children with no known risk factors for neurodegenerative diseases. Long-term exposure to air pollution should be considered a risk factor for both Alzheimer’s and Parkinson’s diseases, and APOE ɛ 4 allele carriers could have a higher risk of developing AD if they reside in a polluted environment.