Urban air pollution adversely impacts brain functions in human populations and animal models. Emerging findings show associations of airborne pollutant levels with mild cognitive impairments (Calderon-Garciduenas et al. 2008; Chen and Schwartz 2009; Power et al. 2011, 2013; Weuve et al. 2012). Brains from a highly polluted city had premature inflammation and neurodegeneration (Block and Calderon-Garciduenas 2009). Rodents chronically exposed to particulate matter (PM) from diesel engines or urban traffic emissions also developed glial inflammatory responses (Kleinman et al. 2008; Levesque et al. 2011a,b, 2013; Morgan et al. 2011; Win-Shwe and Fujimaki 2011) and oxidative stress with protein nitrosylation (Levesque et al. 2011b) and lipid peroxidation (Zanchi et al. 2010). Exposure of rats to diesel exhaust also impaired memory functions associated with the hippocampus (Fonken et al. 2011; Win-Shwe et al. 2012).
The basis for hippocampal memory impairments from inhalation of urban air pollutants could include glutamate receptors, which are altered in rodent models by exposure to nPM from diesel exhaust (Win-Shwe et al. 2009) or by nPM of < 0.2 μm fractioned from Los Angeles urban freeway air in our prior study (Morgan et al. 2011). The nano-sized PM (ultrafine PM) from combustion engines has consistently shown higher toxicity than larger PM in vivo, e.g. (Li et al. 2013) and in vitro (Li et al. 2003; Gillespie et al. 2013). Inhalation of nPM for 150 h during 10 weeks decreased hippocampal levels of the GluA1 subunit of AMPA receptors (Morgan et al. 2011). The selectivity of responses to nPM is indicated by the absence of changes in GluA2 levels or in the associated synaptic proteins post synaptic density 95 (PSD95) or synaptophysin.
In vitro primary hippocampal neuronal cultures also showed inhibition of neurite outgrowth during exposure to nPM at 2 μg/mL for 48 h, an effect rescued by the NMDA receptor antagonist AP5 (Morgan et al. 2011). nPM induced lactate dehydrogenase release by hippocampal slice cultures, a measure of cell damage, which was also rescued by AP5. Inhibitory effects of nPM on neurite outgrowth may share mechanisms with the regression of hippocampal cornu ammonis area 1 (CA1) and CA3 dendrites after in vivo exposure to vehicular-derived PM of 2.5 μm size (Fonken et al. 2011).
To further analyze the mechanisms of nPM on glutamatergic functions, we examined the effects of acute nPM on synaptic proteins in hippocampal slices and dissociated neurons. nPM cross cell membranes by non-phagocytic mechanisms (Geiser et al. 2005) consistent with their relatively high hydrophobicity (Xia et al. 2006). Because nPM rapidly induced free radicals in macrophages (Li et al. 2003; Xia et al. 2006), we investigated free radical production in slices and neuronal cultures, with emphasis on nitric oxide (NȮ). Glutamatergic subunit nitrosylation was examined because NMDA receptors are vulnerable to oxidative damage (Aizenman et al. 1989, 1990; Manzoni et al. 1992; Shi et al. 2013). Neuronal selectivity was assessed by comparing effects of nPM on CA1 pyramidal neurons, which are more vulnerable than DG neurons to nPM toxicity (Fonken et al. 2011), ischemia (Kawasaki et al. 1990), and Alzheimer disease (Morrison and Hof 1997). To evaluate functional outcomes of nPM, synaptic transmission was examined by patch clamp recording.
- Top of page
- Material and methods
- Authors Contributions
- Supporting Information
We report three new findings on rapid neuronal damage by nPM derived from vehicular traffic using in vitro models of acute hippocampal slices and dissociated neurons. (i) Hippocampal slices responded rapidly to nPM with dose-dependent increases of nitric oxide (NȮ) within 15 min. (ii) Nitrosylation of several glutamate receptor subunits was increased by 2 h, while phosphorylation of other sites was decreased. In both slice and neuronal cultures, levels of post-synaptic proteins PSD95 and spinophilin were increased. Several changes induced by nPM were blocked by the NMDA receptor antagonist AP5. (iii) The amplitude of excitatory post-synaptic currents in CA1 neurons was decreased, while paired-pulse facilitation was unchanged. These findings document that acute exposure to nPM can alter properties of glutamate receptors that are critical to neuronal plasticity and memory processes. These findings suggest mechanisms that contribute to cognitive impairments associated with vehicular-derived pollutants and to the vulnerability of CA1 neurons to excitotoxicity in Alzheimer disease and ischemia (see introduction).
Dose responses to nPM corresponded to prior studies with these locally derived nPM, in the range of 1–10 μg nPM/mL for in vitro brain cell models (Morgan et al. 2011) and macrophages (Xia et al. 2006). Other PM sources collected from diesel exhaust (Levesque et al. 2011b) or from urban Baltimore (Zhao et al. 2009) were not as active in cytokine induction at levels < 10 μg/mL on macrophages or bronchial epithelial cells, respectively.
The components of nPM that induced NȮ and altered glutamate receptors could include redox active metals, water soluble organic carbon (WSOC), and long-lived free radicals that persist for 30 days after initial collection (Morgan et al. 2011). Relative to ambient nPM, the present filter-extracted nPM had similar levels of WSOC and redox active metals, but relatively less black carbon, polyaromatic hydrocarbons, steranes, and organic acids because of differential extraction from filters by sonication (Morgan et al. 2011 and its Table S2).
NȮ production in hippocampal slice cultures was increased within 15 min of exposure to nPM, followed by nitrosylation of proteins at 2 h. The rapid increase of NȮ within 15 min may be the most rapid free radical response to combustion engine-derived PM in refereed reports. For comparison of the time course, macrophages (RAW 264.7 murine line) exposed to 10 μg nPM/mL had increased H2O2 production at 30 min (Li et al. 2002), followed at 60 min by decreased mitochondrial membrane depolarization (ΔΨm), and then increased MitoSox fluorescence at 2 h of exposure; mitochondrial swelling arose by 4 h, followed at 16 h by loss of cristae and increased mitochondrial [Ca2+] (Xia et al. 2006). Note that these prior reports used alternate terminology (ultrafine particles, UFP), which were collected at the same site by co-author Sioutas (Morgan et al. 2011). Although the size range of UFP was similar to the present nPM, they may have differed in chemical composition because they were not extracted from filters (see above). Since this report was submitted, Gillespie et al. (2013) observed increased NȮ after 24 h incubation of dopaminergic neurons with 8 μg/mL. Responses to nPM may prove to be even faster than 15 min because nPM appear to cross cell membranes directly without endocytosis (Xia et al. 2006). Thus, the mitochondrial [Ca2+] influx may begin earlier than the 16-h observation time of Xia et al. (2006).
Other reactive oxygen species and oxidants may be induced by nPM besides NȮ. As shown for a different urban PM sample by Zhao et al. (2009), we observed that nPM increased the oxidation of mitochondrially targeted hydroethidine (also called MitoSOX™) by about 50%. Although the increased fluorescence of Mito-SOX™ may indicate increased superoxide, it can also be caused by other mitochondrial derived radicals and oxidants (Zielonka and Kalyanaraman 2010). The subsequent induction of the Nrf2-dependent detoxifying enzymes observed after 10 weeks of in vivo exposure (Zhang et al. 2012) may thus involve multiple oxidants and free radicals.
In the present acute responses, the blockade of NȮ production by the NMDA receptor antagonist AP5 has implications for the mechanisms underlying toxic effects of nPM. Activation of either AMPA or NMDA receptors in hippocampal slices rapidly increased NȮ production in CA1 neurons within 5 min (Frade et al. 2009). Moreover, NȮ production in neurons depended on extra-cellular Ca2+ and recruitment of the PSD95/NOS complex to post-synaptic NMDA receptors (Sattler et al. 1999). In vivo, glutamate and NMDA infusion also rapidly increased NȮ levels more in CA1 neurons than DG neurons (Lourenço et al. 2011). Further studies are needed to resolve the role of Ca2+, glutamate, and nNOS in the rapid induction of NȮ in relation to possible neuronal depolarization by nPM.
Biochemical modifications of glutamate receptor subunits were also observed. The S-nitrosylation of cysteine residues in GluA1 and GluN2A was increased by 50% after 2-h exposure to nPM in CA1; again, DG neurons were unresponsive. The blockade of S-nitrosylation by AP5 is consistent with its inhibition of NȮ induction. Concurrently, nPM decreased phosphorylation of GluA1 (pS831 and pS845) and of GluN2B (pS1303) in slices and neuronal cultures. Because phosphorylation regulates glutamate receptor trafficking to synapses (Santos et al. 2009), the decreased phosphorylation in response to nPM could be a factor in the increased levels of GluA1 and GluN2B. However, the opposite directions of nitrosylation and phosphorylation responses to nPM were not predicted by the induction of both nitrosylation and phosphorylation in GluA1 by glutamate agonists, in which phosphorylation of S831-GluA1 required C875 nitrosylation (Selvakumar et al. 2013). Further studies may address possible oxidative damage by nPM to cysteines or other residues, which may directly modulate phosphorylation; alternatively, indirect actions through phosphatases or kinases may occur.
By western blots of hippocampal slices, post-synaptic proteins were selectively altered by acute nPM, with increased GluA1 (but not GluA2 or mGluR5), and of GluN2A and GluN2B (but not of GluN1). PSD95 and spinophilin (but not synaptophysin) were also increased. The direction of these changes is consistent with rapid increases of GluN2A and of spine puncta in cultured hippocampal neurons at 70 min after depolarizing pulses of KCl (Baez et al. 2013). Notably, these KCl induced increases of GluN2A were blocked by inhibitors of RNA and protein synthesis. Thus, early responses of synaptic protein levels to nPM may be mediated by gene expression as well as by post-translational mechanisms of nitrosylation and phosphorylation.
Cytotoxicity from nPM is also indicated by the nitrosylation of GAPDH measured in whole slice proteins. Excitotoxic glutamate levels also increased GAPDH nitrosylation (Hara et al. 2005). By EthD-1 uptake, CA1 neurons showed the greatest membrane damage, which may contribute to their decreased EPSC. This ranking follows the in vivo induction of NȮ by depolarizing KCl to a greater extent in CA1 than DG neurons (Baez et al. 2013) and the well-known relative vulnerability of CA1 neurons to ischemia and Alzheimer disease.
Hippocampal neuron cultures responded to nPM in parallel with the hippocampal slices for decreased phosphorylation of GluA1 at S831 and S845. In both slice and neuron cultures, nPM increased PSD95 and spinophilin. In neuron cultures, these responses to nPM were blocked by AP5; because the E18 neurons had limited synapse formation after 7 days of culture, we suggest that AP5 was acting on extrasynaptic NMDA receptors. However, levels of GluA1 protein decreased in cultured neurons, but increased in slices. This divergence may be attributed to the earlier developmental stage of dissociated neuron cultures, whereas the trisynaptic hippocampal circuit was fully formed in slices from 1-month-old mice. The neuronal cultures also do not model glial-neuron interactions, as observed in conditioned media from mixed glia exposed to nPM, which altered neurite outgrowth (Morgan et al. 2011). The TNF-alpha secretion induced in mixed glia by nPM is a candidate for modulation of neurite outgrowth (Cheng et al. 2012), and for modulation of AMPA receptor subunits (Santello and Volterra 2012)
These electrophysiological studies indicate that nPM can directly alter post-synaptic functions in CA1 pyramidal neurons, which showed the greatest membrane vulnerability by EthD-1 assay. Thus, acute exposure to nPM caused a large reduction (50%) of evoked EPSC amplitudes across a range of stimulus intensity with minor immediate impact on pre-synaptic neurotransmitter release. We suggest a role for the impaired phosphorylation of GluN2B receptor, which in turn phosphorylates GluA1 for trafficking and insertion into the post-synaptic membrane. A post-synaptic location of these changes is indicated by the normal paired-pulse facilitation at Schaeffer collateral synapses, implying maintenance of pre-synaptic transmitter release. The paired-pulse plasticity expressed with synapses activated at short interstimulus intervals is a measure of neurotransmitter release (Fioravante and Regehr 2011). An alternative mechanism, which was not investigated, could involve effects of post-synaptically generated NȮ on the excitability of pre-synaptic terminals. Pilot data suggest greater alteration of NMDA receptor currents (Davis et al. 2012). Because of the importance of NȮ-dependent GluA1 phosphorylation to memory (Traynelis et al. 2010), we predict that long-term potentiation LTP will be impaired by nPM exposure. For example, LTP is impaired in the mouse GluA1 S845A mutant that cannot be phosphorylated (Lee et al. 2010).
To develop a working model, we need to know if nPM exposure induces a cascade starting with Ca2+ influx from nPM that causes glutamate release, leading to induction of NȮ. The role of long-lived free radicals in the nPM suspension (Morgan et al. 2011) to the induced cellular free radicals also remains unknown. Synapse-independent glutamate release can occur at neuritic growth cones (Soeda et al. 1997; Gelsomino et al. 2013). Further downstream may be remodeling of glutamate receptors through nitrosylation and phosphorylation of subunits via pathways that are recognized in LTP. This cascade is applicable to in vivo responses to chronic nPM inhalation initiated by neurons and glial in the olfactory mucosa, which project into the forebrain (Block and Calderon-Garciduenas 2009) and which transport synthetic nPM as far as the hippocampus and cerebellum (Oberdörster et al. 2004).
The present in vitro models do not address the putatively slower adaptive responses to nPM during chronic exposure in vivo, in which GluA1 was decreased (Morgan et al. 2011), in contrast to GluA1 increases from acute in vitro exposure. The present in vitro models of hippocampal slice and neuronal cultures give a basis for dissecting acute effects of nPM on neurons and glia of the nasal mucosa neuroepithelium and the olfactory bulb, which are the initial brain cell contacts of inhaled nPM (Oberdörster et al. 2004; Block and Calderon-Garciduenas 2009).
Lastly, we note potential links of nPM to brain aging and Alzheimer disease (AD) through glutamatergic functions and the amyloid β-peptide (Aβ). Human brains from a highly polluted city showed premature elevations of Aβ (Calderon-Garciduenas et al. 2012). In a rat model, chronic inhalation of diesel exhaust PM2.5 increased brain amyloid β-peptide (endogenous rat Aβ) (Levesque et al. 2011a). Ongoing epidemiological studies address possible associations of air pollution with AD and other dementias. Moreover, glutamate receptors have direct links to Aβ in rodent models. In AD transgenic (ADtg) mice carrying human Aβ, activation of metabotropic glutamate receptors promoted synaptic release of Aβ (Kim et al. 2010), whereas the FDA-approved, NMDA receptor antagonist Memantine decreased the brain Aβ load (Alley et al. 2010). Furthermore, activation of extrasynaptic NMDA receptors (predominantly GluN2B) increased NȮ and caused synapse loss (Talantova et al. 2013). Thus, the hippocampal spine loss from inhalation of PM observed in mice exposed to diesel PM2.5 (Fonken et al. 2011) could involve the endogenous mouse Aβ, NȮ, and glutamate. The newer NMDAR antagonist, NitroMemantine, protected against Aβ-induced synapse loss, as well as induction of NȮ and S-nitrosylation (Talantova et al. 2013). Future studies may consider therapeutic interventions with NMDAR antagonists on the synergies between nPM inhalation and Aβ as factors in cognitive impairments associated with aging, as well as onset and progression of AD.