|Home | About | Journals | Submit | Contact Us | Français|
Accumulating evidence suggests that outdoor air pollution may have a significant impact on central nervous system (CNS) health and disease. To address this issue, the National Institute of Environmental Health Sciences/National Institute of Health convened a panel of research scientists that was assigned the task of identifying research gaps and priority goals essential for advancing this growing field and addressing an emerging human health concern. Here, we review recent findings that have established the effects of inhaled air pollutants in the brain, explore the potential mechanisms driving these phenomena, and discuss the recommended research priorities/approaches that were identified by the panel.
The association of air pollution with a number of adverse respiratory and cardiovascular health effects has been well documented. More recent epidemiological and animal toxicology studies have raised concerns about the potential impact of air pollution on central nervous system (CNS) outcomes including chronic brain inflammation, microglia activation, and white matter abnormalities leading to increased risk for autism spectrum disorders, lower IQ in children, neurodegenerative diseases (Parkinson’s disease, PD; Alzheimer’s disease, AD), multiple sclerosis, and stroke, as discussed below. Research needs to be pursued to gain comprehensive understanding of the contributions of perturbations in the brain due to the diverse particulate- and gas-phase components of air pollution mixtures. Alterations in the CNS may have direct health consequences or play a secondary role (s) by modulating the responses of the cardiovascular, pulmonary and immune systems. Conversely, alterations in the cardiovascular, pulmonary and immune systems due to air pollution may have effects on the brain due to the production of circulating pro-inflammatory mediators. Given the importance of understanding the effects of toxic exposure over the life span and evidence for the special vulnerability of the fetus, young children and the elderly to air pollution, the need for more research is obvious.
To evaluate how to support this emerging environmental health issue, the National Institute of Environmental Health Sciences (NIEHS) consulted with leaders in air pollution health effects research, behavioral neurotoxicologists, and basic neuroscientists studying the molecular and cellular basis of learning, cognition, and neurodegeneration. They were charged with describing what is currently known so that NIEHS could then identify research priorities and gaps for future studies and recommend the most promising approaches for moving this nascent field forward.
While it is acknowledged that indoor air pollutants are likely to contribute to changes in brain health, the workshop focused on outdoor air pollutants. As such, this mini-review will discuss the following areas:
Air pollution is a complex mixture that includes particulate matter (PM); ozone, carbon monoxide, sulfur oxides, nitrogen oxides, methane, and other gases, volatile organic compounds (e.g., benzene, toluene, and xylene), and metals (e.g., lead, manganese, vanadium, iron). It is derived from numerous natural and anthropogenic sources. The chemical composition of air pollution shows both spatial and temporal variations that reflect local sources (e.g., traffic character, industry, and natural biologic processes) and meteorological (including sunlight intensity) conditions. Although outdoor pollutants can penetrate the indoor environment, there are also unique sources of indoor air pollution from combustion (e.g., oil, gas, kerosene, coal, wood, and tobacco products), biologics (e.g., mite and cockroach fragments, molds), and vapors from building materials, furnishings, household cleaning and maintenance products. Importantly, several constituents of air pollution have the potential to reach the brain and adversely impact the CNS.
The PM component of air pollution is defined by its aerodynamic diameter (PM10, <10 μm; PM2.5 < 2.5 μm; ultrafine (UF), PM0.1, < 0.1 μm). Inhaled particles are deposited differentially within the upper airways (nasopharynx, larynx), conducting airways, and alveolar region depending upon their aerodynamic size, airway anatomy, and ventilation pattern (Phalen et al. , 2010). The location where particles deposit can influence their rates of uptake by lung macrophages, their movement into the interstitium, and their potential translocation from the respiratory tract to other sites within the body, including the brain (Geiser et al. , 2005, Oberdorster et al. , 2004, Phalen, Mendez, 2010). Soluble components (e.g., metals, some organics) and very small particulate fragments or UF particles may also be absorbed through the respiratory tract surface into the systemic circulation. PM has received significant attention as a human health concern, where effects associated with PM include cardiopulmonary, cardiovascular, and more recently, CNS damage and disease.
There are substantial data showing that PM and other sources of air pollution are causally related to cardiovascular disease (Brook et al. , 2010), can worsen asthma (Auerbach and Hernandez, 2012) especially in children (Breysse et al. , 2010, Patel et al. , 2011), and exacerbate chronic obstructive pulmonary disease (Eisner et al. , 2010). Epidemiological studies also demonstrate that exposure to ambient PM is associated with significantly higher risks of lung cancer mortality (mortality relative risk of 1.14 for a 10 μg/m3 increase of PM2.5) (Pope et al. , 2002). Accumulating evidence also indicates that the health effects of air pollution extend past the respiratory tract, resulting in cardiovascular damage and consequent elevation of morbidity and mortality (Brook, Rajagopalan, 2010).
Numerous studies in humans have demonstrated deleterious effects of air pollutant exposure on various measures of cardiovascular health (Gill et al. , 2011) including systemic inflammation (Peters et al. , 2001, Seaton et al. , 1999), blood pressure (Auchincloss et al. , 2008), heart rate variability (Liao et al. , 2004, Park et al. , 2010, Park et al. , 2005, Whitsel et al. , 2009), vascular function (Adar et al. , 2010, Diez Roux et al. , 2008), altered cardiac structure and function (Van Hee et al. , 2009, Van Hee et al. , 2010) and atherosclerosis (Diez Roux, Auchincloss, 2008, Kaufman, 2010). Investigations of these cardiovascular disease endpoints in population-based studies and in studies of sensitive subgroups have enhanced our understanding of the mechanistic pathways underlying associations between air pollutants and cardiovascular morbidity and mortality. The three broad pathways that have been identified include: (1) the release of inflammatory and oxidative stress mediators by the lungs into the systemic circulation; (2) interactions between pollutants and pulmonary neuronal afferents resulting in autonomic imbalance; and (3) direct translocation of particle constituents or particles from the respiratory tract into the systemic circulation. It is hypothesized that these same mechanisms underlying the effects of air pollution on pulmonary and cardiovascular health may also mediate responses in the brain.
Increasing evidence links air pollution to CNS diseases, behavior deficits, neuroinflammation, and neuropathology in human and animal studies (Block and Calderon-Garciduenas, 2009, Lucchini et al. , 2011). However, the underlying mechanisms and sequence of events that culminate in neurotoxicity remain poorly understood. Consistent with prior air pollution toxicology research, pre-existing disease conditions, age, and critical periods of development may modulate CNS effects, but at the present time, few studies have systematically explored these issues. Here, we summarize the current findings of what is known about air pollution effects in the brain and highlight potential mechanisms.
While the majority of prior studies have focused on the effects of air pollution in cardiovascular and pulmonary disease, evidence now points to a potential role for air pollution in CNS disease (Block and Calderon-Garciduenas, 2009). In fact, postmortem sampling has identified PM in the human brain, albeit of unknown composition and origin (Calderon-Garciduenas et al. , 2008b). In a small controlled-exposure study utilizing a blinded randomized crossover design, a significant relationship was observed between exposure to high levels of diesel exhaust and increased brain (frontal cortex) activity as measured by quantitative electroencephalography (Cruts et al. , 2008). Importantly, human studies have also shown that living in conditions with elevated air pollution is linked to decreased cognitive function (Calderon-Garciduenas et al. , 2008a, Chen and Schwartz, 2009, Power et al. , 2011, Ranft et al. , 2009, Suglia et al. , 2008, Weuve et al. , 2012), lower neurobehavioral testing scores in children (Wang et al. , 2009), a decline in neuropsychological development in the first 4 years of life (Morales et al. , 2009), AD- and PD- like neuropathology (Calderon-Garciduenas et al. , 2010, Calderon-Garciduenas et al. , 2012, Calderon-Garciduenas et al. , 2004, Morales, Julvez, 2009), increased stroke incidence (Mateen and Brook, 2011) (Donnan et al. , 1989, Henrotin et al. , 2007, Villeneuve et al. , 2006), and elevated autism risk (Volk et al. , 2011). Some air pollution components such as manganese have been associated with CNS pathology, as elevated levels of manganese in the air have been shown to correlate with enhanced PD risk (Finkelstein and Jerrett, 2007)
Recent reports reveal that controlled inhalation exposure to ambient PM (Campbell et al. , 2009, Campbell et al. , 2005), ozone (Santiago-Lopez et al. , 2010), diesel exhaust (Gerlofs-Nijland et al. , 2010, Levesque et al. , 2011a, Levesque et al. , 2011b), and manganese (Antonini et al. , 2009, Elder et al. , 2006) results in elevated cytokine expression and oxidative stress in the brain. At present, the mechanisms responsible for the initiation of neuroinflammation in response to air pollution are poorly understood and may be exposure-specific. One theory implicates activated brain microglia in neuroinflammation.
Microglia are the resident innate immune cells in the brain and are predominant regulators of neuroinflammation, evidenced by their production of pro-inflammatory (e.g., Interleukin-1! , tumor necrosis factor (TNF)-α, prostaglandin E2, and interferon-γ) and reactive oxygen species (e.g., •NO, H2O2, O•2−, and ONOO−/ONOOH) (Block et al. , 2007). Although microglia are necessary for normal brain function, excessive and chronic activation can result in neurotoxicity, such as the initiation and/or amplification of neuronal damage (Block and Hong, 2005). Importantly, microglia are activated in neurodegenerative diseases, as indicated by analysis of postmortem brains from AD and PD patients (McGeer et al. , 1988), and have been implicated in the progressive nature of each disease (Block and Hong, 2005, 2007, Lull and Block, 2010, McGeer and McGeer, 2010). Microglia are also activated in the pre-frontal cortex of autistic children (Morgan et al. , 2010) and in response to ischemic stroke (Yenari et al. , 2010). In vitro studies have shown that PM activates microglia, amplifies the microglial response to pro-inflammatory stimuli, and causes cellular damage (Block et al. , 2004, Morgan et al. , 2011, Sama et al. , 2007). This is in concert with additional in vitro studies showing that PM activates macrophages through a toll-like receptor (TLR) 2 and 4-dependent mechanism (Shoenfelt et al. , 2009). Microglia are known to express many pattern recognition receptors, including TLRs, where TLR’s initiate a cascade of inflammatory responses that are dependent on the activation of TLR’s pathways (Lehnardt, 2010). Furthermore, dopaminergic neurons underwent cell death only in the presence of microglia following nanometer-sized diesel exhaust particle exposure (Block, Wu, 2004), suggesting that this cell type may be important for neurotoxicity.
Consistent with human studies (Calderon-Garciduenas, Solt, 2008b), recent reports have demonstrated in rodent models that air pollution activates microglia (Levesque, Taetzsch, 2011b, Morgan, Davis, 2011). In rodents exposed to diesel exhaust via inhalation, measures of neuroinflammation were reported to be the highest in brain regions with the most microglia (Levesque, Taetzsch, 2011b) . Pro-inflammatory markers have also been shown to correlate with manganese translocation along the olfactory neuronal pathway (Elder, Gelein, 2006). These studies suggest that specific brain region vulnerability is likely a combination of the cellular architecture of the region (e.g., microglia density), the characteristics of the exposure in question, and the route of administration. These studies support the hypothesis that air pollution may be a common and continuous environmental source of microglial activation.
Other CNS cells are also predicted to be affected by air pollution exposure. For example, astrocytes are the most prevalent cell type in the brain and are reported to be activated in humans (Calderon-Garciduenas, Mora-Tiscareno, 2008a) and animals (Elder, Gelein, 2006) following exposure to certain air pollutants. Oligodendrocytes exist in the brain as both progenitor cells that can produce cytokines and growth factors (Hahn et al. , 2010) and mature cells that insulate neurons (Mekhail et al. , 2012). While oligodendrocytes have been largely overlooked in air pollution research, it is likely that these cells are affected in children (Calderon-Garciduenas, Mora-Tiscareno, 2008a) and dogs (Calderon-Garciduenas et al. , 2002) with air pollution-induced white matter lesions. Finally, neurons themselves can be affected by air pollution directly, a possibility suggested by their ability to internalize PM (Calderon-Garciduenas, Solt, 2008b), and are also likely damaged in white matter lesions in humans (Calderon-Garciduenas, Mora-Tiscareno, 2008a). Furthermore, neuronal damage may also contribute to the pro-inflammatory milieu in the brain (Block, Zecca, 2007).
Changes in α-synuclein (Chahine and Stern, 2011, Shi et al. , 2011), Aβ42 (Perrin et al. , 2011), and tau (Perrin, Craig-Schapiro, 2011) have been proposed as potential indicators of preclinical neurodegenerative diseases, such as PD, AD, and frontotemporal dementia. Neuropathological analyses of post-mortem tissues from humans living in highly polluted areas have shown elevations in Aβ42 (Calderon-Garciduenas, Kavanaugh, 2012, Calderon-Garciduenas, Solt, 2008b), hyper-phosphorylated tau (Calderon-Garciduenas, Kavanaugh, 2012), and α synuclein (Calderon-Garciduenas, Solt, 2008b). Importantly, these findings have been confirmed in animal studies using subchronic (6 months) inhalation exposures to DE, demonstrating that Aβ42, phosphorylated tau, and α synuclein are elevated in rat brain (Levesque, Surace, 2011a). Together, these findings suggest that exposure to air pollution has the potential to impact neurodegenerative disease pathways. However, at this time epidemiology data are unavailable and the ability of the complex mixture of air pollution to damage neurons, either directly or through inflammation is poorly understood.
There is increasing evidence that air pollution has neuropsychological effects in humans and in animal models (Guxens and Sunyer, 2012). Several experimental studies have linked air pollution to neurochemical changes (Suzuki et al. , 2010, Yokota et al. , 2009) and markers of synaptic plasticity (Win-Shwe et al. , 2009, Win-Shwe et al. , 2011) that may precede or be independent of overt neuronal death. Animal experiments provided evidence that diesel exhaust inhalation decreases measures of spatial learning (Win-Shwe, Yamamoto, 2011), motor behavior (Suzuki, Oshio, 2010, Yokota, Mizuo, 2009), and even impacts behavioral measures of emotion (Yokota et al. , 2011). Exposure to near roadway air pollution in mice for 10 months has also been linked to decreases in spatial learning and depressive behavior (Fonken et al. , 2011). While the mechanisms driving these phenomena are unclear, these studies emphasize that there are clear behavioral effects associated with air pollution exposure, supporting a potential link to human disease.
While air pollution is a complex mixture that likely affects human health through multiple pathways, some of the individual constituents or characteristics of specific components have already been implicated in neurotoxicity and nervous system disease independently. Oberdörster and Utell (Oberdorster and Utell, 2002) first proposed that the brain may be vulnerable to ultrafine ambient PM. It was further shown that model nanosized particles could cross the BBB and physically enter the CNS of animals (Lockman et al. , 2004). Biological components of PM such as endotoxins, mold, and pollen have been associated with neurodevelopmental disorders (e.g., schizophrenia, autism, mental retardation) and neurodegenerative diseases (e.g., AD, PD) (Meyer et al. , 2009). A major biological inflammatory component of PM is endotoxin. In fact, it has been shown that the “activation” properties and cytokine production of PM is directly related to the content of endotoxin (Shoenfelt, Mitkus, 2009). Short term exposure to endotoxins such as bacterial lipopolysaccharides administered intranasally in rats induces cytokine expression in several brain regions, activates the stress axis and alters behavioral coping responses to stress mimicking all the effects of long term exposure to PM (Tonelli et al. , 2008). In addition, metals, solvents, and PAHs found in air pollution and carried on PM have been associated with neurodegenerative diseases (e.g., PD, AD) (Budimir, 2011, Hozumi et al. , 2011, Meloni and Vasak, 2011, Wirdefeldt et al. , 2011) and cognitive deficit disorders (Edwards et al. , 2010, Freire et al. , 2010, Pearce and Braverman, 2009).
Lead and manganese are strongly implicated in neuronal damage and CNS disease. The clinical manifestation of lead neurotoxicity includes impaired intellectual function and attention, encephalopathy, and convulsions (Mendola et al. , 2002). Neurotoxicity associated with the organic solvents such as toluene has also been reported occupationally and in experimental settings (Win-Shwe and Fujimaki, 2010). More specifically, toluene exposure is clinically associated with frequent headaches, memory impairments, and CNS depression, among other effects (Win-Shwe and Fujimaki, 2010). Manganese exposure produces a form of parkinsonism with bradykinesia, rigidity, and occasional resting tremor (Lucchini et al. , 2009). Willis and coworkers (2010) have recently reported that the incidence of PD is greater in urban U.S. counties with high industrial release levels of copper or manganese (Willis et al. , 2010). Similarly, some investigators have suggested that exposure to manganese and other welding fumes could predispose individuals to earlier onset PD (Racette et al. , 2001). In addition, various air pollutants, such as PAHs, can cause damage to the sensory neurons present in the olfactory epithelium and associated functional changes (Feron et al. , 2001, Tonelli and Postolache, 2010, Upadhyay and Holbrook, 2004). Interestingly, olfactory damage and loss of smell are commonly seen in both PD and AD (Tonelli and Postolache, 2010).
It is important to note that short inhalation exposures (1 month) to some forms of complex air pollution mixtures such as diesel exhaust (Levesque, Taetzsch, 2011b) and concentrated urban air pollution (Veronesi et al. , 2005) fail to cause neurotoxicity in healthy, young adult animals. On the other hand, short term ozone exposure in rodents has been reported to result in cerebral edema (Cretu et al. , 2010), neurodegeneration in the hippocampus, striatum, and substantia nigra (Pereyra-Munoz et al. , 2006, Rivas-Arancibia et al. , 2010, Santiago-Lopez, Bautista-Martinez, 2010), and altered behavior (Rivas-Arancibia et al. , 2003, Santucci et al. , 2006, Sorace et al. , 2001). Further, exposure to carbon monoxide in animals has been associated with increased oxidative stress in cochlear blood vessels, damage to spiral ganglia neurons, inner hair cells(Lopez et al. , 2008) and ultimately with hearing loss (Schwela et al. , 2005). Thus, the potential neurotoxic effects of air pollution are likely to be exposure specific, time-dependent and dose-dependent, emphasizing the importance of employing accurate occupational and ambient exposure models in future studies.
Subclinical and clinical cardiovascular disease are established risk factors for cognitive decline and dementia (O’Brien, 2006, Stampfer, 2006), and it is likely that the observed impact of air pollutants on the brain occurs, at least in part, downstream of air pollution-induced cardiovascular effects. It is also plausible that the generalized mechanistic pathways described by Brook et al. (Brook, Rajagopalan, 2010) act independently of cardiovascular effects. Circulating cytokines, markers of systemic inflammation, have been shown to result in neuroinflammation (Rivest et al. , 2000) and neurotoxicity (Qin et al. , 2007) and to exacerbate neurodegenerative progression (Perry et al. , 2007). Moreover, it has been implicated that systemically circulating ultrafine particulate matter can reach the brain causing inflammation and oxidative stress (Block and Calderon-Garciduenas, 2009, Peters et al. , 2006). Potential mediators of these effects are astroglia, cells important in maintaining the BBB and the regulation of neurotransmitter uptake and response to injury, and microglia, the immune cells of the brain (Block and Calderon-Garciduenas, 2009). In addition, the pathways that confer cardiovascular risk by way of the autonomic nervous system(Liao, Duan, 2004, Park, O’Neill, 2005, Whitsel, Quibrera, 2009) imply that air pollutants exert direct quantifiable effects on preganglionic neuronal activity. Thus, cardiovascular disease and air pollution may also interact to influence CNS health.
The nasal cavity provides an alternative transport pathway for direct brain delivery of some inhaled materials via olfactory neurons into the olfactory bulbs. Further CNS distribution can then occur to the piriform cortex, olfactory tubercule, amygdala and entorhinal cortex (Lucchini, Dorman, 2011). Further, anterograde pathways are proposed to connect from these regions to the orbito-frontal cortex, thalamus, hypothalamus and hippocampal formation (Sunderman, 2001b). Reports indicate that ultrafine PM deposited on the olfactory epithelium can indeed distribute throughout the brain as a result of olfactory transport (Elder, Gelein, 2006, Oberdorster et al. , 2009, Oberdorster, Sharp, 2004). Various metal compounds (manganese, iron, cadmium, thallium, mercury, cobalt, zinc) as well as carbon particles appear to move into brain following inhalation or intranasal/tracheal instillation exposures (Sunderman, 2001b). However, not all studies confirm these transport routes (Rao et al. , 2003) and movement is influenced by the chemical species. Mechanisms by which such transport occurs remain unknown although roles for metal-binding carrier molecules or transporter proteins in facilitating the olfactory nerve transport of solutes have been proposed (Genter et al. , 2009, Sunderman, 2001a). It should also be noted that animal species differences are important since the relative surface area of the nasal olfactory mucosa in humans (5%) is much smaller than that of rodents (50%)(Dorman et al. , 2002).
A direct “nose-to-brain” transport of manganese has been demonstrated by several investigators. The presence of manganese (as 54Mn) in the olfactory bulb of rats, mice, and freshwater pike on the side ipsilateral to the instilled nostril (Tjalve and Henriksson, 1999) were demonstrated after intranasal instillation of radiolabeled manganese chloride. Olfactory transport of inhaled Mn has also been shown (Dorman, Brenneman, 2002) in an animal model in which one nostril was occluded, thus restricting olfactory transport of Mn to the side of the rat brain ipsilateral to the patent nostril. In these studies, direct delivery along the olfactory route accounted for nearly all of the 54Mn found in the olfactory bulb and tract of the rat brain following acute Mn inhalation, indicating that Mn can cross synapses within the olfactory pathway and travel via secondary and tertiary neurons to more distal sites within the brain. Lewis and coworkers (2005) reported that the rat trigeminal nerve may also translocate Mn from the nasal cavity (Lewis et al. , 2005). Although olfactory transport rapidly delivers Mn to brain structures in the olfactory pathway, the trigeminal route appears to be relatively slow (and perhaps inefficient) in delivering inhaled Mn to the rat striatum and other more distant brain structures (Leavens et al. , 2007).
Neuroepithelial damage and biotransformation of ambient air pollutants may also impact bioavailability in the brain and consequent CNS effects. The epithelium of olfactory mucosa has high activity and expression of cytochrome P450 (CYP) isoforms including CYP1A2, 2 A, 2B, 2E, and 2G (Livertoux et al. , 1996, Longo et al. , 2000, Voigt et al. , 1993). These phase 1 enzymes catalyze the biotransformation of ambient air pollutants such as polycyclic aromatic hydrocarbons, semi-quinones and quinones into toxic metabolites, such as mutagenic epoxide or reactive oxygen species. The presence and activity of flavin-contamining monooxygenase in olfactory epithelium also produces deleterious reactive metabolites from air pollutants (Genter et al. , 1995). Importantly, phase II enzymes in the nasal cavity epithelium, such as UFP-glucuronosyltransferase, epoxide hydrolase, sulfotransferases, and glutathione S-transferases (Bond et al. , 1988),function to detoxify reactive metabolites formed by CYP-dependent activities. Thus, the appropriate balance of phase I and II enzymes in the epithelium of olfactory mucosa may provide a first line of defense for the brain, limiting axonal transport of ambient air pollutants and toxic metabolites from the olfactory nerve cells to the olfactory bulb. However, while the level and activity of CYP enzymes in the brain are low when compared to those in the nasal mucosa (Ghersi-Egea et al. , 1988), it has been proposed that air pollutants escaping the metabolic barrier of the olfactory epithelium may reach the brain and undergo local cerebral metabolism which could have deleterious CNS effects (Minn et al. , 2002). In fact, a recent study demonstrated that aged mice failed to upregulate phase II enzymes in the brain in response to nanoparticle urban air pollution exposure (Zhang et al. , 2012), offering some insight into the role of aging on CNS effects. Clearly, there is a need for more research to sort out these biological and chemical factors that may contribute to the potential for nasal or other transport of air pollution to the brain.
Systemically delivered ultrafine PM or its bioavailable components may enter the brain through the BBB. The BBB itself comprises multiple cell types (e.g., endothelia, pericytes, astrocytes) and transporter systems which together provide an “impermeable” barrier covering most regions of the brain and spinal cord. When intact and not impaired by disease, age, infection, or toxicants, the BBB effectively protects the CNS interstitium from macromolecules, toxins, and small organic drugs (Miller, 2010). Recent evidence indicates that air pollutants have been shown to alter BBB function. For example, aluminum nanoparticles have been reported to injure endothelial cells and damage the BBB (Chen et al. , 2008). Human exposure to air pollution shows endothelial cell damage in the cerebral vasculature, with increases in ICAM and VCAM present (Calderon-Garciduenas, Solt, 2008b). In addition, in vitro studies using whole brain rat capillaries reveal that treatment with particulate matter causes production of cytokines and reactive oxygen species, which signal changes in transporter expression and function (e.g. P-glycoprotein and Multidrug Resistance Associated Protein-2) and a decrease in expression of various tight junction proteins (Hartz et al. , 2008). Thus, brain capillaries respond to air pollution components by regulating the physical and chemical barrier function and producing pro-inflammatory signals. In addition, this response may serve as a pro-inflammatory sensor (Hartz, Bauer, 2008) and ultimately distribute reactive oxygen species, cytokines, and particulate matter to the brain parenchyma, further contributing to CNS pathology.
There are distinct brain regions (e.g., median eminence, area postrema, subfornical organ and periventricular regions of the hypothalamus) in which the BBB is comprised of a more fenestrated endothelial layer (Abbott et al. , 2010). Perivascular microglia use their proximity to this “leaky” barrier structure to survey the systemic environment for adverse stimuli such as inflammatory cytokines, and possibly soluble or particulate pollutants that have exited the respiratory tract and entered the systemic circulation. The perivascular microglia are activated through innate immune pathways in an inflammatory cascade that includes NF-κB activation and up-regulation of numerous surface molecules such as major histocompatibility complex (MHC) and various receptors (e.g., Toll-like MAC-1, CD14, chemokines) (Carrillo-Esper, 2003). These microglia subsequently release free radicals and pro-inflammatory cytokines (interleukin-1β, TNF-α) that can damage surrounding neurons. These events represent a common pathway for microglia-mediated neurodegeneration (Block and Hong, 2005).
The latter scenario suggests that physical entry of soluble or particulate pollutants into the brain may be unnecessary to stimulate adverse changes. It further indicates that the inflammatory mediators released by pollutant inflamed organ systems such as the respiratory tract or gut can trigger the microglia located along the fenestrated BBB (Finnerty et al. , 2007, Hetland et al. , 2005, Nel, 2005, Soukup and Becker, 2001). This peripheral inflammation results in the systemic release of inflammatory mediators (e.g., cytokines) over time that can then activate the brain’s immune surveillance cells (i.e., microglia) (Carrillo-Esper, 2003) and alter CNS immune response, function, and/or behavior (Perry, Cunningham, 2007, Tonelli and Postolache, 2010). At this time, the role of the peripheral immune response in the CNS effects of air pollution has not been tested.
An additional potential route of entry for air pollutants is via the sensory afferents found in the gastrointestinal tract as part of the brain-gut axis (Perry, 2004). It is known that PM particulates that are too large to enter the lungs are quickly cleared by swallowing (Mutlu et al. , 2011). Likewise, clearance of particles that deposit in the deep lung eventually occurs via the mucociliary escalator and, so, gastrointestinal tract exposure cannot be discounted. The sensory afferents of the dorsal vagus nerve located in the gut communicate directly with brain stem neurons. It is interesting to note that abnormalities in olfaction-vagal-brainstem routes (gastric systems secondary to swallowing of nasal secretions in saliva) have been studied as the early pathologies of Parkinson’s neurodegeneration (Hawkes et al. , 2007, Jang et al. , 2009). Pulmonary afferents from the airways (Hunter and Dey, 1998, Hunter and Undem, 1999) and/or deep lung may also represent an additional target site.
Psychosocial stress and toxicants such as air pollutants are known to influence common physiological pathways including neuroendocrine and autonomic function, oxidative stress and proinflammatory immune pathways (Dantzer et al. , 2008, Tomei et al. , 2003). Therefore, the interaction between these factors is complex and likely bi-directional in which stress may influence the negative outcomes of exposure to air pollution and exposure to air pollution may also lead to maladaptive coping responses to stress. For example, an correlation between stress (community violence) and ambient air pollution (measured as NO2) has been reported for urban childhood asthma risk (Clougherty et al. , 2007) suggesting that stress exacerbates the effects of air pollution on respiratory illness. Furthermore, increased asthma risk has been correlated with poorer socioeconomic status (Drake et al. , 2008). Given that social stress tends to track along socio-economic gradients, poor communities may be at greatest risk for the toxicity of ambient air pollutants due in part to higher levels of psychosocial stress. Conversely, socially enriched environments may protect children from the toxic effects of environmental hazards (Laviola et al. , 2008, Schneider et al. , 2001). On the other side, higher perceived stress on poorer cognitive function have been reported in elderly men and the effect has been shown dependent on low-level lead exposure (Peters et al. , 2010). Animal studies have shown that intranasal instillations of bacterial lipopolysaccharides (LPS) are capable of inducing depressive-like behavior and increased hormonal responses to stress (Tonelli et al, 2009). Both lead and LPS are common components of air pollution with the potential of significantly impact on stress responsiveness. An integrated approach that addresses both cleaner air and social interventions may have significant implications for prevention and intervention in high-risk urban populations (O’Neill et al. , 2003, O’Neill et al. , 2008, Wright, 2006). In sum, understanding potential interactions between ambient air pollution and psychosocial stress promises a more complete understanding of the toxicity of air pollutants at both individual and population level.
Regardless of whether one studies genetic, social or even nutritional susceptibility, besides exposure and disease, there is always a third dimension when considering environmental exposures: timing. Dose-response curves are not likely to be constant across all life stages. Fetal life and early childhood, in particular, are times of vulnerability. These life stages are periods of rapid growth, cell differentiation, organogenesis and, in the case of the brain, network development. Relatively small disruptions in these processes may set individuals on trajectories that have subtle effects in early years and profound effects later in life.
Because many genes are only expressed during specific developmental stages and are subsequently turned off, gene-environment interactions may be partly or even largely dependent on the timing of the environmental exposure. Such interactions likely depend on the life stage at which exposure occurs. For example, growing evidence from animal research indicates that the CNS is highly vulnerable to chemical injury during development (Faustman et al. , 2000, Rodier, 2004). This is due to interference with processes critical to neurodevelopment such as neuronal growth, synaptic network formation, neuronal migration, and development of receptor numbers. These processes are most active during fetal life and in childhood and subside during adult life. If toxic exposures impact these growth and differentiation processes, they could alter the trajectory of brain development, but the same chemical would be less toxic once development is complete. Differential toxicity in children vs. adults has been shown for many chemicals including methyl mercury (Amin-Zaki et al. , 1974, Marsh et al. , 1980), PCBs (Tilson et al. , 1990), organophosphate pesticides (Engel et al. , 2007), and lead (Bellinger et al. , 1987, Schnaas et al. , 2006). With respect to environmental effects that act via gene-environment interactions, effects may be dependent primarily on whether exposure timing corresponds to a critical developmental window during which the gene is highly expressed. A design that cannot address the timing of environmental exposures would likely miss such gene-environment interactions (Tsuang et al. , 2001). Likewise, interactions that also involve other risk factors - social, nutritional, etc,-may depend on the timing of the joint exposure for similar reasons.
Many neurodegenerative diseases are predicted to be the result of cumulative exposures across an entire lifetime (Carvey et al. , 2006) and aging is a well-known risk factor for the two most prevalent neurodegenerative diseases, AD and PD (von Bernhardi et al. , 2010). Further, aging is linked to elevated neuroinflammation, where aged brains show an elevation of cytokines and microglial activation (Sparkman and Johnson, 2008, von Bernhardi, Tichauer, 2010). In addition, aged brains are more vulnerable to environmental insult, particularly pro-inflammatory stimuli (Sparkman and Johnson, 2008, von Bernhardi, Tichauer, 2010), which may be of importance given the extensive reports of a pro-inflammatory response in the brain with diverse forms of air pollution exposure (Block and Calderon-Garciduenas, 2009). Notably, aging may be especially important to inflammation-mediated neurotoxicity in response to peripheral inflammation, e.g. chronic neuroinflammation in response to a single intraperitoneal injection of lipopolysaccharide – a potent inflammatory stimulus – in young adult mice will only culminate in dopaminergic neurotoxicity in aged animals (Qin, Wu, 2007). Further, aged individuals are more vulnerable to the cardiopulmonary health effects of air pollution (Namdeo et al. , 2011). For example, acquired immunity declines with age in the respiratory system (Busse and Mathur, 2010), which has the potential to negatively impact the brain. However, the effect of the interplay of air pollution and aging on CNS health is largely unknown.
Genetic factors may be modifiers of chemical neurotoxicants. The current state of art for gene-environment interaction is to conduct genome wide screening in a discovery stage to generate a candidate gene list for second stage genotyping in an independent sample. The subset of single nucleotide polymorphisms (SNPs) that are significant in the discovery stage are tested in an independent population to validate the findings. A smaller subset in this second stage are expected to be significant, and more likely contain any true positive results. The alternative design is to conduct a biological pathway study, also known as a candidate gene approach. This is less costly and is hypothesis driven, unlike the genomic approach. Typically, investigators select candidate genes which reflect the biological pathways implicated in air pollution toxicity. Examples might include inflammation (IL-6, inducible nitric oxide synthase (iNOS), TNF-α),(Castro-Giner et al. , 2009, Thompson et al. , 2010, Thomson et al. , 2007) the autonomic system (Norepinephrine transport, NET [HUGO name SLC6A2]) (Min et al. , 2009), and HPA axis (glucocorticoid receptor (GCR) [HUGO name NR3C1]and 11- β hydroxysteroid dehydrogenase-2(11-β HSD2) [HUGO-HSD11B2])(Derijk, 2009, Derijk et al. , 2002, Zaehner et al. , 2000). There is no methodology that can comprehensively create a complete list of candidate genes. The potential list of candidate genes is likely impossible to comprehensively select a priori, as genes may play roles in chemical metabolism, cellular toxicity, neurotransmitter metabolism, or could even work indirectly (i.e., disrupt the metabolism of a nutrient involved in brain function). The selection of candidate genes is therefore subjective and will vary from investigator to investigator. This is the major limitation of this approach. In addition, it is unlikely that all the relevant genes and regulatory regions that modify the toxicity of ambient pollution are even known. Thus, a candidate gene approach can never comprehensively interrogate all the potential genetic modifiers of environmental toxicants. The genomic approach might require a much larger sample and may be most feasible in the context of pooled studies or multi-center collaborations. A further complication is that gene-environment interactions could be manifest in a brain region-specific manner, thus requiring careful tissue harvesting to ensure that target areas are not lost of that a signal of interest is not diluted. However, the possibility exists that extremely large gene-environment interaction effects have not yet been discovered, given the paucity of research on such interactions in genomics. If such effects do exists, large sample sizes may not be necessary.
Variation in DNA sequence cannot explain all host susceptibility to ambient pollution. Indeed, phenomena such as fetal programming are difficult to explain in the context of DNA sequence variation. The toxicity of air pollutants is undoubtedly multifactorial, and one potential pathway yet to be fully explored is epigenetics, the study of changes in gene function that are mitotically and/or meiotically heritable and that do not entail a change in DNA sequence (Wu and Morris, 2001). DNA methylation is the best studied of the epigenetic processes that regulate gene silencing and there is a growing body of evidence demonstrating an association between DNA methylation and environmental ambient pollutants (Baccarelli et al. , 2009). DNA methylation changes can occur in somatic tissues during any stage of development and program gene expression changes. Importantly, DNA methylation has been linked to brain development. For example, failure of DNA methylation systems leads to clinical syndromes including mental retardation and autistic-like behaviors (Shahbazian and Zoghbi, 2002), and DNA methylation changes have been linked to other behavioral disorders as well (Levenson et al. , 2006, Petronis, 2003). Furthermore, animal studies increasingly demonstrate that environmental factors can alter DNA methylation patterns and that these changes correlate with animal behavior (Weaver et al. , 2004, Weaver et al. , 2001). Epigenetic modifications of regulatory DNA sequences in response to variations in environmental conditions can be a critical source of variation in gene expression and function. In addition, DNA methylation changes can mediate the relationship between genome and environment throughout each stage of neurodevelopment, including fetal life. Thus, while the role of epigenetics in CNS vulnerability to air pollution has yet to be explored, this area of research may be of significant importance.
There is significant need for well-designed studies investigating the human risk for CNS disease with air pollution exposure. There are several methodological challenges as related to confounding, exposure misclassification, and other potential biases that are unique to conducting epidemiologic research. In the context of studying air pollution and brain health, confounds are proximate causes (or antecedent predictors) of both air pollution exposure and the disease outcomes. For example, maternal mental health status may be a potential confounder in the putative associations of behavioral problems in children and prenatal exposure to air pollution, because the increased risk of various adverse neurobehavioral outcomes in offspring (either through genetic predisposition or not) are associated with mothers’ mental health status, which may affect residential choice and thus determine the estimated residential exposures to ambient air pollution during pregnancy or early childhood. Or if older people with better cognitive function behave differently (e.g., move to neighborhoods with cleaner air) from those with poor brain health, this may lead the appearance of a spurious positive association between air pollution and cognitive decline. Epidemiologic studies of the effects of long-term air pollution exposures on brain health thus have to be designed carefully and be able to address the methodological issues related to exposures and confounders likely to vary over time.
PM may affect those with pre-existing neurological disease as it may affect disease progression rather than or in addition to initiation. Oxidative stress is posited as a possible mechanism that mediates air pollution induced adverse effects on brain health. Therefore, comborbidities with other non-neurological diseases where there may have been underlying increases in oxidative stress (e.g. cardiovascular diseases, heart failure, diabetes, lupus, and obesity) need to be carefully assessed because of their resulting possibly greater susceptibility to the adverse effects on human brains. If particles escape the respiratory tract(Oberdorster, Elder, 2009) and enter the vasculature, those individuals with diseases that promote increased airway particle deposition may be at increased risk; this includes those with mild asthma and COPD(Pope, 2000). In those with more severe asthma, the increased particle deposition is probably counterbalanced by increased mucus production that will limit particle migration to the periphery. This will be an important consideration in future studies.
Most ambient (outdoor) air pollution studies often conducted in large human populations base their exposure assessment on routine nationwide monitoring networks for “criteria” air pollutants that measure all or some of the gases NOx, CO, ozone, SO2, as well as particles (lead and PM10, and more recently also PM2.5). These pollutants commonly originate from the same source (e.g., CO and NOx from motor vehicles; sulfate particles and SO2 from power plants and industrial facilities) and even interact with each other chemically. As the ambient levels of these pollutants are correlated in space and time, it is hard to disentangle biologic effects of individual pollutants in air pollution epidemiologic studies with exposure assessment that solely relies on environmental monitoring network data. The relative importance of different ambient air pollutants in producing health effects may vary by endpoint (stroke/neurodegeneration) and pollutants in mixtures may act synergistically. For pollutants that are more spatially homogenous (e.g., PM2.5), monitors placed in urban areas allow estimation of community-wide average exposures and record variation over time very well but these networks are often inadequately dense for spatially heterogeneous pollutants where proximity to sources determines peak exposures; e.g., ultrafine particles (<0.1 μm), black smoke (a marker of exhaust particles), CO and NOx/NO/NO2 as markers of vehicle exhaust have been shown to decrease exponentially with distance (within hundreds of feet) from freeways (Zhou and Levy, 2007, Zhu et al. , 2002a, Zhu et al. , 2002b).
Lack of sufficient exposure characterization to inform the possible sources of pollutant exposures is another limitation of epidemiologic studies that rely only on air monitoring network to assess the exposures. Recently some health effect studies employed simple measures of proximity to sources such as traffic or much more complex models such as land use based (LUR) regression and emissions based models (such as CALINE4). For example, to characterize traffic-related air pollutants in urban environments, LUR models are based on simultaneous measurements of vehicle exhaust markers, such as NOx, taken at many locations throughout an urban area during multiple two-week periods (e.g. see Su et al 2009)(Su et al. , 2009). Various geographic information systems (GIS) parameters (e.g. traffic and roadway density, traffic volumes, truck routes, population density, and land use features and sometimes even satellite-derived data such as soil brightness) are used to predict the measured concentrations (Sahsuvaroglu et al. , 2006). This fine scale spatial model can then be used to estimate exposure concentrations at specific locations based on the GIS parameter values. Also, much more computationally intense air dispersion models have been developed based on vehicle emission factors such as the CALINE4 model that relies on for example the California Air Resources Board’s EMFAC2007 model (http://www.arb.ca.gov/msei/onroad/latest_version.htm), paved road-dust emissions for PM2.5 based on in-roadway measurements (Fitz and Bufalino, 2002) and meteorological factors (hourly wind speed, direction, and temperature from the National Weather Service and average mixing heights) to derive pollutants concentrations for particles or gases (Benson, 1989, Croes and Fujita, 2003, Wu et al. , 2009a, Wu et al. , 2005, Wu et al. , 2009b). The performance of such LUR or emissions based models in health effects studies; however, still needs to be further explored and depends on whether or not the data sources indeed can accurately model exposures in time and space. The accuracy and precision of exposure prediction based on LUR may vary by geographic areas in which these models are employed (Wu et al. , 2011).
In contrast to the population-monitoring networks described above, short term personal monitoring and biomarkers promise in-depth exposure information at an individual level, and even physiologic and target organ level as well. Yet, these approaches are extremely costly and labor intensive and, therefore, feasible only for application to small groups of subjects. These approaches will also need to clearly define the toxin of interest and determine whether personal or biologic monitoring would be possible over the longer periods. Further, it will also be necessary to discern whether non-target organ samples such as blood, urine, or hair would suffice to establish exposures affecting the brain. To date, there is little consensus on appropriate biomarkers for ambient air pollutants except possibly for some PAHs, mostly through measurement of DNA , protein, or albumin-adducts, although measurement of hydroxylated PAH metabolites in urine is another option (Castano-Vinyals et al. , 2004, Smith et al. , 2002). Other issues that need to be considered carefully regarding use of biomarkers is the ability to discern source contributions (e.g., dietary versus inhalation sources of PAHs) and appropriate timing of measurements. For example, urinary PAH metabolites measured at a single point in time will reflect exposures that occurred in the previous few days while DNA-adducts will reflect approximately a one month period (Barr et al. , 2005).
A growing list of studies implicates air pollution as a significant CNS health concern. As described above, these earlier reports have identified the problem and are beginning to define the extent and mechanisms of these CNS effects. However, these initial findings are as of yet incomplete and emphasize the importance of additional scientific inquiry. To address this issue, the NIEHS panel identified four critical research foci to move this developing field forward and address this emerging health concern, which are described in detail below.
The following topics were identified as critical areas of research:
The following research priorities were identified as urgent priorities in the field:
Specific human research priorities are to:
The following recommendations for methodology and approach for future research were identified:
Evidence is accumulating for air pollution related CNS effects at multiple levels, including modulation of molecular/neuochemical/pathobiological pathways, neuroinflammation, neurotoxicity, and neurobehavioral changes that implicate subclinical/clinical manifestation of disease. However, the extent of these effects contributing to ill health, the components of air pollution responsible and the molecular mechanisms underlying the phenomena are poorly understood. As such, there is significant need for thorough epidemiology, mechanistic, and translational studies to identify how air pollution affects CNS health with the goal of eventual identification of early markers of neuropathology, intervention, and prevention of disease. Identification of critical periods, vulnerable populations, and the use of environmentally relevant, life-time exposures will be key to understanding this human health concern. This research program will by necessity involve close collaborations amongst scientists from different fields, including atmospheric science, epidemiology (air pollution, developmental, and neuroepidemiology), clinical medicine (neurodevelopment, neurology, and psychiatry), inhalation and behavioral neurotoxicology, social sciences (including neuropsychology) and basic neuroscience.
This work was supported by NIEHS/NIH-DERT.
CONFLICT OF INTEREST STATEMENT There authors declare that there are no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.