Effects of a Cyclooxygenase-2 Preferential Inhibitor in Young Healthy Dogs Exposed to Air Pollution: A Pilot Study

Abstract

Residency in cities with high air pollution is associated with neuroinflammation and neurodegeneration in healthy children, young adults, and dogs. Nonsteroidal anti-inflammatory drugs may offer neuroprotection. The authors measured the plasma concentrations of 3-nitrotyrosine and the cerebro-spinal-fluid concentrations of prostaglandin E2 metabolite and the oligomeric form of amyloid derived diffusible ligand; measured the mRNA expression of cyclooxygenase-2, interleukin 1β, CD14, and Aquaporin-4 in target brain areas; and evaluated brain MRI, cognition, and neuropathology in 8 dogs treated with a preferential cyclooxygenase-2 inhibitor (Nimesulide^®) versus 7 untreated litter-matched Mexico City dogs. Nimesulide^® significantly decreased nitrotyrosine in plasma (p < .0001), frontal gray IL1β (p = .03), and heart IL1β (p = .02). No effect was seen in mRNA COX2, amyloid, and PGE2 in CSF or the MRI white matter lesions. All exposed dogs exhibited olfactory bulb and frontal accumulation of Aβ₄₂ in neurons and blood vessels and frontal vascular subcortical pathology. White matter hyperintense MRI frontal lesions were seen in 4/6 non-treated and 6/8 treated dogs. Nonsteroidal anti-inflammatory drugs may offer limited neuroprotection in the setting of severe air pollution exposures. The search for potentially beneficial drugs useful to ameliorate the brain effects of pollution represents an enormous clinical challenge.

Introduction

There is mounting evidence that exposure to air pollution can cause systemic inflammation (MacNee and Donaldson 2000; van Eeden & Hogg 2002; Brook et al. 2004; Frampton 2006; Rückerl et al. 2007; Swiston et al. 2008), neuroinflammation, neurodegeneration, neurotoxicity (Calderón-Garcidueñas et al. 2003, 2004; Calderón-Garcidueñas, Franco-Lira, et al. 2007; Calderón-Garcidueñas, Solt, et al. 2008; Peters et al. 2006; Mohankumar et al. 2008), and structural brain changes and cognitive deficits (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Particulate matter in the fine (<2.5 μm) and ultra-fine (<100 nm) size translocate to the brain and has been recorded in red blood cells (RBC), endothelial cells, and perivascular macrophages in olfactory bulb and frontal samples of humans and dogs resident in Mexico City (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Neuroinflammation is a critical component of the brain’s responses to insults and plays a crucial role in neurodegenerative disorders such as Alzheimer’s and Parkinson’s diseases (Benveniste 1998; Akiyama et al. 2000; Block et al. 2004; Cunningham et al. 2005; Allan, Tyrrell, and Rothwell 2005; Minghetti 2005; Ferrari et al. 2006; Forero et al. 2006; P. McGeer, Schulzer, and McGeer 1996; P. McGeer, Rogers, and E. McGeer 2006; E. McGeer and McGeer 2007). We have shown in highly exposed children and young adults, average age 25.1 ± 1.5 years, a significant upregulation of cyclooxygenase-2 (COX2), interleukin 1β (IL1β), and CD14 in olfactory bulb, frontal cortex, substantia nigrae, and vagus nerves; disruption of the blood-brain-barrier (BBB) and endothelial activation; as well as inflammatory cell trafficking when compared to matched subjects resident in low-pollution areas (Calderón-Garcidueñas, Solt, et al. 2008). We have previously demonstrated that health effects from air pollution in dogs mimic similar effects in humans (Calderón-Garcidueñas et al. 2001, 2002, 2003; Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008), underscoring the utility of comparative studies. Healthy young dogs < 1 year of age exhibit neuroinflammation along with disruption of the BBB and accumulation of beta amyloid 42 (Calderón-Garcidueñas et al. 2002). Neuroinflammation represents a potential target to slow or to prevent major neurodegenerative diseases with considerable public health burden: Alzheimer’s and Parkinson’s diseases. Nonsteroidal anti-inflammatory drugs (NSAIDs) may offer neuroprotection through several pathways, and epidemiological studies suggest that these drugs may offer some protection against Alzheimer’s disease, although the appropriate dose, duration, and risk/benefit of NSAIDs’ use are unclear (Andersen et al. 1995; McGeer, Schulzer, and McGeer 1996; Sastre et al. 2006; Szekely, Breitner, and Zandi 2007; Etminan, Gill, and Samii 2008; Kotilinek et al. 2008). Nimesulide n-(4-nitro-2-phenox-yphenyl) methane sulfonamide is a preferential COX2 inhibitor used in clinical practice as an analgesic, antipyretic, and anti-inflammatory drug and experimentally as a neuroprotectant in brain ischemia and to improve endothelial dysfunction (Suleyman et al. 2008; Candelario-Jalil 2008; Abdelrahman and Suleimani 2008; Kerola et al. 2009). Major concerns in the setting of prolonged exposures to air pollution are the association of neuroinflammation with vascular pathology and neurodegeneration and the need to prevent or halt such effects—thus our interest in the potential use of antiinflammatory agents. Dogs are sentinels of exposures for humans and a good model for examining behavior, cellular, and molecular processes involved in early phases of human brain aging and AD (Cummings et al. 1996; Pugliese et al. 2005).

The primary purpose of this pilot work was to measure the effects of the chronic oral administration of Nimesulide^® by real-time polymerase chain reaction on two key inflammatory genes COX2, and IL1β, and the lipopolysaccharide (LPS) receptor CD14 in target brain regions of young dogs resident in southwest Mexico City (SWMC) treated with Nimesulide^® versus untreated littermates. Given that the neuronal and blood vessel accumulation of βA_1-42 is seen in highly exposed young dogs, we also quantified the amyloid-derived diffusible ligand (ADDL) in CSF samples and did immunohistochemistry (IHC) for Aβ₄₂ in two anatomical targets of air pollutants: olfactory bulb and frontal cortex. Brain nuclear magnetic resonances were done on each dog to evaluate structural changes that may be related to exposure to air pollution (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). There is a considerable interest in developing strategies for preventing Alzheimer’s disease, and dogs represent an excellent sentinel to monitor anti-inflammatory drugs with potential use in humans.

Materials and Methods

Drugs and Reagents

Nimesulide^® was purchased from commercially available tablets Eskaflam^® GSK Mexico, and indomethacin (used as internal standard for the determination of Nimesulide^® by HPLC) was purchased from Sigma-Aldrich (Seelze, Germany). Acetonitrile chromatographic grade was obtained from Merck (Darmstadt, Germany). De-ionized water was obtained from a Milli-Q system (Continental Water Systems, El Paso, TX, USA). All other reagents were of analytical grade purchased from Sigma-Aldrich.

Study Areas

For this study we selected a large, polluted megacity: Mexico City (MC). MC represents an extreme of urban growth and environmental pollution (Molina et al. 2007). The metropolitan area of ~2,000 km² lies in an elevated basin at an altitude of 2,250 m above sea level and is surrounded by volcanic and discontinuous mountain ranges that restrict the natural ventilation of the basin. The combination of topography and meteorology in MC contributes substantially to its air pollution. In this geographical setting, the metropolitan’s 20 million residents, almost 4 million vehicles, and over 40,000 industries consume more than 40 million liters of petroleum fuels per day, producing an estimated annual emission of 2.6 million tons of particulate and gaseous air pollutants (Bravo-Alvarez and Torres-Jardón 2002; Comisión Ambiental Metropolitana 2006; Molina et al. 2007). Air pollution is generally worse during the winter season, when rain is sparse and thermal inversions are more frequent. Residents in MC including dogs are chronically exposed to significant concentrations of ozone, particulate matter (PM), and lipopolysaccharides associated with PM (LPS-PM) (Calderón-Garcidueñas, Vincent, et al. 2007). Southwest MC dogs are exposed to a yearly average concentration of PM_2.5 of 25 μg/m³, a value above the annual standard of 15 μg/m³. LPS detected in PM₁₀ samples show a range of 15.3 to 20.6 nanograms per milligram of PM₁₀, and south MC PM samples show the highest endotoxin concentrations at 59 EU/mg PM₁₀ (Bonner et al. 1998; Osornio-Vargas et al. 2003).

Guidelines for the Use and Care of Experimental Animals

Procedures used were in accordance with the guidelines of the Instituto Nacional de Pediatría (INP) on the Use and Care of the Animals and conform to the standard in the Guide for the Care and Use of Laboratory Animals (NIH Pub. no. 86-23). The INP provided full veterinary daily care of the dogs included in this study.

Dog Population

The study protocol was approved by the Institutional Animal Care and Use Committee at the INP. We selected 15 healthy 12- to 19-month-old mongrel dogs, bred and raised at the INP animal facility located in southwest MC. Dogs were whelped and housed in an outdoor-indoor kennel; husbandry was in compliance with the American Association of Laboratory Animal Certification Standards. Dogs were under daily veterinary observation during their entire life: at no time was there any evidence of overt respiratory, cardiovascular, or neurological diseases; they had all applicable vaccines; and they were treated with antihelmintics regularly. All dogs except one had a brain MRI 2 to 4 days prior to their euthanasia, which was conducted in accordance with established American Veterinary Medical Association guidelines (Panel on Euthanasia 2001). Blood samples were obtained for complete blood count and N-tyrosine shortly before anesthetizing the animals. N-tyrosine was determined according to previously described protocols (Kumarathasan and Vincent 2003). Plasma samples were pooled for this test: 7 and 8 samples, respectively, for nontreated and Nimesulide^®-treated dogs.

Experimental Design

We had two litters of mongrel dogs (total n: 15) available for this study. Four dogs from each litter were started with the Nimesulide^® treatment at ages 78 or 202 days. Seven dogs belonging to the same litters (4 and 3) served as controls. Dogs were treated with oral Nimesulide^® 5 mg/kg/day for 394 (age 202 days at start of Rx) or 451 (age 78 days at start of Rx) days. This dosage regimen was selected since it has been reported that plasma concentrations reached at 5 mg/kg oral dose totally inhibit COX2, and partially inhibit COX1 (Toutain, Cester, Haak, and Metge 2001; Toutain, Cester, Haak, and Laroute 2001). Prior to the beginning of the treatment, a catheter was placed in the cephalic vein and a control blood sample was obtained. Then dogs received an oral Nimesulide^® dose according to weight, and blood samples were obtained at 10, 20, and 30 minutes and 1, 1.5, 2, 4, 8, 12, and 24 hours after drug administration. At 6 and 12 months of treatment, blood samples were obtained immediately before and 2 hours after drug administration. Samples were centrifuged and stored frozen at − 80°C until analyzed by a previously reported high-performance liquid chromatographic method (Carrasco-Portugal et al. 2000).

Pharmacokinetic Analysis

Plasma level-time curves were constructed, and pharmaco-kinetic parameters and maximal concentration (C max) and time to reach this maximum (T max) were directly obtained from these curves. Area under the plasma concentration against time curve (AUC) was obtained by the trapezoidal rule (Rowland and Tozer 1995). Half-life was obtained by log-linear regression of the terminal decay phase. Extrapolation to infinity was calculated by dividing the last measurable concentration by the elimination rate constant. All pharmacokinetic analysis was carried out using the WinNonLin Professional version 2.1 software (Pharsight, Palo Alto, CA, USA).

Cognitive Questionnaire

To evaluate the cognitive status of the 15 dogs, we used the cognitive 16-item test used by Pugliese et al. (2005). The test was filled out by the veterinarian in charge of the study. For each item, score 1 indicated the normality of the specific behavior, and scores 2, 3, and 4 the degrees of abnormality. Similar to the Pugliese et al. approach, we then selected the 9 items that in the authors’ hands evaluated more accurately the cognitive status of each dog and subdivided the dogs into 3 categories: normal, <12; light cognitive deficit, 12 to 26; and severe cognitive deficit, >26. The final second score reflected the cognitive status of the animal.

Dogs’ Magnetic Resonance Imaging (MRI)

Fourteen of 15 dogs had a brain MRI 2 to 4 days prior to the termination of the study. The 3D MRI was acquired on a 1.5 Tesla 5T Signa Excite HD MR (General Electric, Milwaukee, WI, USA) with an 8 Channel Brain Array. The total scanning time was approximately 35 minutes, and dogs were anesthetized with Zoletil-Virbac 10 mg/kg IM. White matter lesions (WML) were defined as hyperintense focal areas observed in two different sequences: T2 and T2 weighted with fluid-attenuated inversion recovery (FLAIR) axial and coronal images. White matter lesions were scored by lobe location and number, and we obtained a quantitative measure of load by multiplying each lesion by a size-dependent constant: 0.0042 ml for 1(<3 mm), 0.114 ml for 2 (4–10 mm), and 0.90 ml for 3(>10 mm) according to the method used by Kruit et al. (2004). The studies were coded and two neuroradiologists reviewed the studies independently, having access only to age information in each case. The final index of number and extent of white matter lesions for each dog (Kruit score) was the result of the evaluation of the two readings.

Dogs’ Necropsy and Tissue Preparation

Pathology procedures were as in our earlier works (Calderón-Garcidueñas et al., 2002, 2003, 2008b). Full necropsies were performed within 2 to 4 days after the MRI study to allow for the review of the studies conjointly by the veterinarians, neuroradiologists, and neuropathologist. Dogs received the last oral dose of Nimesulide^® between 24 and 28 hours prior to their sacrifice. Prior to opening the skull, a sample of CSF on average 8 cc was taken from the cisterna magna and frozen at − 80°C. Full necropsies were performed immediately after euthanizing the dogs. Samples from the brain and olfactory bulbs were taken within 3 minutes after euthanizing the animals. Samples from frontal and parietal cortex, hippocampus, nasal respiratory and olfactory mucosa, lungs, heart, and liver were taken. Samples were fixed in 10% neutral formaldehyde for 48h and then transferred to 70% alcohol before paraffin embedding within 1 week of the necropsy. Paraffin sections 8 μm thick were cut and routinely stained with hematoxylin and eosin (H&E). Fresh frontal samples for the RT-PCR studies were taken from the cortex and the white matter separately, and we used either direct homogenization of the tissue or a procedure for the isolation of microvascular endothelial cells (Ge and Pachter 2006). The pathology of MC dogs was compared with the materials from 5 age-matched control dogs from Tlaxcala, a city located 114 km east of Mexico City at 2,254 m above sea level, with low levels of air pollutants (Calderón-Garcidueñas et al. 2003). A veterinarian pathologist and a human neuropathologist read sections with no access to the codes regarding the dogs’ identification data.

Immunohistochemistry (IHC) was performed as previously described (Calderón-Garcidueñas, Solt, et al. 2008; Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Briefly, the sections were deparaffinized and immunostained, and for this work, we used 88% formic acid as an epitope retrieval method for the Aβ. Negative controls included omission or substitution of primary antibodies by nonspecific, isotype-matched antibodies. Positive and negative controls were included for each antibody. Confirmation of the IR was done with 2 different antibodies in serial sections with a minimum of 10 slides for each case. Selected antibodies included COX2 (Santa Cruz Biotechnology, Santa Cruz, CA, 1:200; and Cayman, Ann Arbor, MI, USA, 1:500), β amyloid 1-42 (Abcam, Cambridge, MA, USA, 1:200), β amyloid, 17-24 4G8 (Signet) (Covance, Emeryville, CA, USA 1:1000), and glial fibrillary acidic protein GFAP (Abcam, 1:500). Transmission electron microscopy was performed in selected frontal gray and white matter samples. Samples were fixed in 2% paraformaldehyde and 2% glutaraldehyde in sodium phosphate buffer (0.1M, pH 7.4), postfixed in 1% osmium tetraoxide and embedded in Epon. Semithin sections (0.5–1 μm) were cut and 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 (Oberkochen, Germany) or a JEM-1011 (Japan Electron Optics Laboratory [JEOL], Tokyo, Japan).

Determination of Amyloid β_1-42 and Prostaglandin E Metabolite in CSF and 3-Nitrotyrosine in Plasma

Aggregated beta amyloid and Prostaglandin E metabolite were measured in CSF samples, while 3NT was measured in plasma. Aggregated beta amyloid was measured by an immunoassay kit from Biosource International, Inc. (Catalog #KHB3491, Camarillo, CA, USA). The oligomeric form of Aβ (amyloid-derived diffusible ligand ADDL) was determined by centrifuging the CSF at 100,000 × g for 1 hour, followed by the Biosource protocol. ADDL results were expressed in ng/ml. Prostaglandin E metabolite was measured from an EIA kit from Cayman Chemical Company (Ann Arbor, MI, USA). Results were expressed as % B/B₀ (ratio of the absorbance of the sample to that of the maximum binding well). Plasma samples (250 μL) treated with DETPA and butylated hydroxytoluene to prevent any autoxidation were analyzed for 3-nitrotyrosine using a previously reported HPLC-CoulArray method (Kumarathasan and Vincent 2003). Results are expressed in pmoles/ml.

Estimation of mRNA Abundance by Real-Time RT-PCR

Protocols were performed as described previously (Calderón-Garcidueñas et al. 2004; Calderón-Garcidueñas, Solt, et al. 2008; Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Relative abundances of mRNAs encoding COX2, IL1β, CD14, GFAP, and Aquaporin-4 were estimated by quantitative fluorogenic 5’ nuclease (TaqMan) assay of the first-strand cDNAs as described in Calderón-Garcidueñas et al. (2004). The amount of COX2, IL1β, CD14, GFAP, and Aquaporin-4 cDNA in each sample was normalized to the amount of GAPDH cDNA yielding an index: molecules per fentomol proportional to the relative abundance of each mRNA in each sample. To compare the results of the mRNA in the genes of interest, we used fresh frozen frontal white matter samples from 5 age-matched control Tlaxcala dogs from our previous studies (Calderón-Garcidueñas et al. 2003). For six different anatomical regions and with direct homogenization of the tissue or a procedure for the isolation of microvascular endothelial cells (Ge and Pachter 2006), we quantified the genes of interest in untreated versus Nimesulide^®-treated MC dogs.

Statistics

Statistics were performed using Stata (StataCorp 2005). We applied a parametric Student t-test or a nonparametric Wilcoxon or Mann-Whitney test procedure to compare two independent samples for the mRNA dog results in the target genes, and univariate descriptive measurements were summarized as mean values ± SEM (Table 5). Significance was assumed at p < .05.

Results

Air Quality Data

Dogs in southwest MC are chronically exposed to significant concentrations of ozone, and particulate matter. Figures 1A–C illustrate the monthly distributions of the data corresponding to the 8-hour average ozone and the 24-hour average PM₁₀ and PM_2.5 measured concentrations in the southwest MC area in the period between May 2005 and November 2006 that includes the dog study period. Boxplot time series graphs show the 10th, 25th, 50th (median), 75th, and 90th percentiles as well as the maximum of the set of concentration values as indicators of the air quality for O₃, PM₁₀, and PM_2.5, respectively. A comparison of the data with associated air quality standards is included in the graphs.

The concentration values for the construction of these figures were obtained from data available to the public from the Mexico City Automatic Atmospheric Monitoring Network (http://www.sma.df.gob.mx/simat/consultas.htm).

Figure 1A shows the boxplot time series for ozone concentrations averaged in mobile 8-hour periods and its comparison with the new primary U.S. EPA 8-hour ozone standard of 0.075 ppmv. Ozone levels above this standard are considered to be unhealthy. During the study period, a great number of exceedences (i.e., values above the standard) to the U.S. EPA 8-hour ozone standard were observed in every month, coinciding with the months with higher UV solar radiation levels. Figure 1B illustrates the monthly variation of PM₁₀ 24-hour average levels and its comparison with U.S. EPA PM₁₀ air quality standard of 150.0 μg/m³. The highest PM₁₀ levels were observed during the typical dry season in central Mexico, when coarse particles can be easily lifted by the wind. The PM₁₀ air quality standard was not exceeded any day along the whole study period. Figure 1C shows the monthly variation of PM_2.5 24-hour average levels and its comparison with the U.S. EPA PM_2.5 air quality standard of 35.0 μg/m³. As shown in Figure 1C, the occurrence of exceedences to the PM_2.5 EPA 24-hour air quality standard is more frequent during months in which the meteorological conditions in central Mexico conduct to restricted ventilation conditions and higher UV radiation. PM_2.5 air pollution is the result of both the primary emissions of fine particulates mainly from motor vehicles combustion processes and the secondary formation of aerosols from primary emitted gases that are converted in fine particles through complicated chemical atmospheric processes. The annual PM_2.5 average concentration for the southwest MC area can also be compared with the U.S. EPA PM_2.5 annual air quality standard of 15.0 μg/m³. The annual PM_2.5 average concentration for any 12-month period combination between May 2005 and November 2006 in the study area was in the range 22.8 to 24.5 μg/m³.

A comparison between the information shown in Figures 1B and 1C indicates that most of the PM₁₀ particles are within the fine fraction. In general, residents in Mexico City have been exposed to concentrations of PM_2.5 above applicable standards year after year (Sistema de Monitoreo Atmosférico 2006).

Pharmacokinetic Analysis

The results of the pharmacokinetic analysis are shown in Figure 2 and Table 1. There was good tolerance of the medication at the 5 mg/kg/day selected oral dose. Plasma level-time curves after administration of Nimesulide^® are shown in Figure 2. Nimesulide^® is absorbed reaching a maximum of about 4 μg/ml in 3 hours. Then levels decayed with a half-life of about 4 hours. At 6 and 12 months of treatment, there was no accumulation of the drug (Figure 2), as expected with the dosage regimen employed and the half-life of the compound. Table 1 shows the pharmacokinetic parameters obtained after administration of the first dose of 5 mg/kg of Nimesulide^® to eight dogs.

Cognitive Questionnaire

No cognitive deficits were observed in any of the Nimesulide^® Rx dogs and only 1/7 control dogs had a score of 18 consistent with a light cognitive deficit (Pugliese et al. 2005). This female dog exhibited alterations in self-control, aggressive behavior, and decreased interaction with other animals and caretakers. This aggressive behavior was severe to the point that no attempt was made to perform a brain MRI due to the difficulty handling the animal by the caretakers and veterinarians.

Brain MRI

Four of six control dogs (66.6%) had white matter lesions (WML) in the frontal lobes, while 6/8 Nimesulide^®-treated dogs had lesions (75%) (Table 2, Figures 3A and B). There were no statistical differences in the number of WML between nontreated and treated dogs (χ² = .117, p = .73). The white matter lesions were very similar in their characteristics, size, and location to the lesions found in southwest MC 10.7 ± 2.4-year-old children (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008).

Real-Time PCR mRNA Analysis of COX2, IL1β, CD14, and Aquaporin-4

Real-time, rapid-cycle PCR analysis of COX2, IL1β, and GFAP were done in a cohort of matched Mexico City untreated dogs from this study (n: 7) and 5 age-matched controls from our previous dog studies (Table 3) to show that there is significant upregulation of COX2 and IL1β in exposed healthy dogs versus low-pollution-exposed animals. We then proceeded to do RT-PCR analysis of COX2, IL1β, CD14, GFAP, and Aquaporin-4 in olfactory bulb, frontal cortex, cerebellum, hearts, and lungs from the treated and untreated MC dogs (n: 15) (Table 4). When samples were stratified according with the dog’s treatment (no Rx versus Nimesulide^®), there was a significant decrease in mRNA for IL1β in frontal gray endothelial fraction in Nimesulide^®-treated dogs, p = .03, and in the heart, p = .02, while for GFAP, the decrease was significant in the frontal white matter (p = .04). No differences were detected with COX2, CD14, or Aquaporin-4 (Tables 4 and 5) among the groups. The AQP 4 gray/white matter frontal ratio between nontreated and Nimesulide^® dogs was not statistically significant (2.5 and 3.90, respectively).

Aggregated β Amyloid and Prostaglandin E Metabolite in CSF Samples

There was no statistical difference between non-treated and treated dogs in either CSF oligomeric form of Aβ (ADDL) (no Rx, 0.54 ± 0.45 vs. Rx, 0.04 ± 0.016 ng/ml; p = .18) and the prostaglandin E metabolite PGE2 (no Rx, 17.92 ± 5.9 versus Rx, 15.85 ± 3.9 % B/B₀; p = .76).

N-Tyrosine in Pooled Plasma

Nontreated dogs had 701.37 pmoles/ml of N-tyrosine in plasma, while Nimesulide Rx dogs had 75.07 pmoles/ml (p < .0001).

Dogs’ Pathological Data

The nasal respiratory epithelium displayed patchy replacement of the mucociliary epithelium by squamous metaplasia in both MC cohorts. Changes in the olfactory epithelium ranged from patchy absent olfactory neurons and mild accumulation of lipofuscin to moderate decrease in sensory and basal layer cells. Lungs displayed bronchiolar epithelial and mild smooth muscle hyperplasia. Chronic mono-nuclear cell infiltrates and scattered mast cells along with macrophages filled with particulate matter were surrounding vascular structures and occupying alveolar spaces. Peribronchial and periaortic lymph nodes contained PM in macrophage-like cells. There were no differences in extra cerebral pathology between nontreated and Nimesulide^® -treated MC animals. Gross examination of the brain revealed no abnormalities in either group. All nontreated and treated dogs exhibited Aβ amyloid in neurons and blood vessels (capillaries and arterioles) in the olfactory bulb and frontal samples (Figures 4A–D). One Nimesulide^®-treated dog age 19.8 months exhibited few isolated neurons and blood vessels with IR to Aβ. Olfactory bulb and frontal sections showed COX2 IR in 7/7 untreated dogs and 4/8 Nimesulide^® Rx dogs. COX2 reactivity was especially prominent in hyperplastic endothelial cells of arteriolar blood vessels and in large neurons. Intense search by light microscopy for particulate matter in the olfactory bulbs yielded negative results. Vascular pathology was prominent in olfactory bulbs and frontal white matter in all dogs (Figures 4E–H). Small vessels exhibited fibrin thrombi (Figure 4E). Frontal white matter sections showed widespread perivascular gliosis and enlarged Virchow-Robin spaces (Figure 4F). Reactive GFAP + astrocytes were focally prominent in subpial, subcortical, perivascular, and deep white matter of all Mexico City dogs (Figure 4F). This observation was also prominent in areas identified by MRI as hyperintense lesions (Figure 4G). Significant loosening of the neuropil with leaky blood vessels, extravasation of red blood cells, along with strongly positive GFAP reactivity and enlarged Virchow-Robin spaces (Figure 4G) were the main histological features of the hyperintense lesions identified by MRI. Frontal white matter arterioles exhibited hyperplastic endothelial cells with elongated fronds that protruded into the lumen and significantly reduced the vessel lumen (Figure 4H). Olfactory bulb architecture was preserved; beta amyloid neuronal and vascular accumulation were present in all dogs (Figures 5A–B). Electron micrographs of frontal white matter vessels showed perivascular macrophage-like cells, with abundant lipid vacuoles in intimate contact with vessel walls exhibiting abundant fibrils (Figures 5C–D). Fibrils with a repeated core β-sheet structure were numerous in the walls of the frontal arterioles (Figure 5E).

Discussion

Oral administration of Nimesulide^®, a nonsteroidal anti-inflammatory COX2 preferential inhibitor to Mexico City healthy young dogs for 394 or 450 days versus littermatched dogs receiving no treatment resulted in (1) a reduction in peripheral oxidative stress as evidenced by a significant decrease in plasma nitrotyrosine (p < .0001); (2) a significant decrease in mRNA expression of frontal gray IL1β (p = .03), heart IL1β (p = .02), and frontal white matter GFAP (p = .04); and (3) no significant effect on mRNA COX2, CD14, and Aquaporin-4, the concentrations of the oligomeric form of Aβ, PGE2 in CSF, or the MRI hyperintense white matter lesions. Pharmacokinetic parameters obtained after the first Nimesulide administration were similar to those previously reported by other authors (Toutain, Cester, Haak, and Metge 2001); the selected oral dose was adequate to produce inhibition of COX2 for the whole period of the treatment without clinical or pathological evidence of secondary effects.

Expected in the Nimesulide^® -treated dogs was the significant reduction in plasma nitrotyrosine (nitrated tyrosine residues), an indirect measurement of the peroxynitrite concentrations formed by the fast reaction of nitric oxide and superoxide radicals, and ultimately reflecting the extent of damage caused by oxidative stress (Orhan et al. 1999; Boulos, Jiang, and Balazy 2000; Kopff, Kopff, and Kowaslczk 2007; Ferrer-Sueta and Radi 2009). In keeping with its antioxidant capacity, Nimesulide^® has been reported to prevent oxidative stress following transient forebrain ischemia in the rat hippocampus at clinically relevant doses (Al-Majed et al. 2004). In brain ischemia models, reduction in measures of oxidative damage seems to correlate with the neuroprotective efficacy of Nimesulide^® against CA1 hippocampal neuronal death after ischemia (Candelario-Jalil 2008). Significant attenuation of lipid peroxidation, that is, determinations of malondialdehyde, 4-hydroxy-alkenals, and lipid hydroperoxides are also reported (Candelario-Jalil et al. 2003). We were expecting lower concentrations of PGE2 metabolite in CSF given that Nimesulide^® inhibits the prostaglandin PG synthesis (Zweifel et al. 2002; Ko et al. 2008; Candelario-Jalil 2008) and that CSF PGE-2 concentrations likely reflect COX2 activity in hippocampal and cortical neurons as well as inflammatory activation in neurons and glial and endothelial cells (Kaufmann et al. 1997; Combrinck et al. 2006). However, the CSF concentrations of PGE2 were not different among the treated and untreated animals. One plausible explanation for our PGE2—CSF results could be the time between the last Nimesulide^® dose and the dogs’ sacrifice 24 to 28 h later. CSF penetration of COX2 inhibitors reaches optimal medication concentrations within 10 h of the last dose and no effects at 22 h after the last dose (Dembo, Park, and Kharasch 2005; Kokki et al. 2007, 2008). The PGE2 endpoint is relevant in the clinical setting, since increased CSF concentrations of prostaglandin E2, a major product of COX2 activity, have been described in patients with mild memory impairment and with probable AD, consistent with the hypothesis that neuroinflammation is an important component of the disease (Montine et al. 1999; Combrinck et al. 2006). The high concentrations of CSF PGE2 in patients with mild cognitive impairment also relate to high TNFα concentrations and tend to decrease as the patient’s cognitive status deteriorates (Tarkowski et al. 2003; Combrinck et al. 2006). Kotilinek et al. (2008), investigating the effects of NSAIDs on memory function and Aβ-mediated inhibition of hippocampal long-term plasticity (LTP) in rats, reported that the beneficial effects on memory were inversely related to forebrain (minus hippocampus) prostaglandin E2 levels and proposed that NSAIDs block COX2-mediated production of PGE2 in dendritic spines, preventing the effects of Aβ on synaptic plasticity and memory function. Furthermore, Melnikova et al. (2006) suggested that pathological activation of COX2 may potentiate the toxicity of Aβ peptides without significantly affecting Aβ accumulation. Thus, our expectations of ameliorating both the accumulation of Aβ and the mRNA expression of COX2 were not fullfilled.

The decrease in COX2 IR in 4/8 Nimesulide^® -treated dogs is important given that in endothelial cells COX2 plays a central role in producing prostanoids that regulate the functions of vascular smooth muscle cells and circulating cells (Capone et al. 2007). Inflammatory stimuli including lipopolysaccharides (LPS), reactive oxygen species, and some cytokines upregulate COX2 expression in endothelial cells (Brian et al. 1998). The reduction of brain endothelial COX2 IR could exert a beneficial effect on the vascular responses to inflammation, since prostanoids are potent vasoactive and inflammatory substances (Brian et al. 1998). Abdelrahman and Suleimani (2008) used Nimesulide^® in a rat model of streptozotocin-induced diabetes and described significant improvement in endothelial dysfunction in treated animals, suggesting that COX2 products may be involved in the pathogenesis of endothelial dysfunction in this model. The downregulation of IL1β in the endothelial fraction of frontal gray matter samples, along with similar downregulation in the heart of Nimesulide^® -treated animals, are not unexpected findings given that COX2 pathway mediates IL1β regulation in different cell types (Bartfai et al. 2007). The results are interesting since IL1β markedly induces angiogenesis in vitro and in vivo, and this angiogenic effect is inhibited by COX2 selective inhibitors (Kuwano et al. 2004). Moreover, the association between the use of selective COX2 inhibitors and adverse cardiovascular events likely relates to an alteration of vascular homeostasis (Hong et al. 2008). Hong et al. (2008) investigated the effect of Nimesulide^® on thrombus formation in a dog model of carotid artery thrombosis and noted that when animals were given LPS to induce a systemic inflammatory response, COX2 became the major local isoform responsible for the production of antithrombotic prostaglandins during systemic inflammation. This antithrombotic-associated PG effect was markedly reduced by Nimesulide^® (Hong et al. 2008). Although we saw no marked differences in the number of cerebral blood vessels exhibiting thrombi in the two groups, in theory the Nimesulide^® animals could have more thrombotic vessels and thus more hypoperfusion and white matter damage. Moreover, if angiogenesis is also being inhibited in the gray matter and the heart (the downregulation of IL1β), this effect will have detrimental brain and heart consequences added to the antithrombotic prostaglandin-inhibited effect in the Nimesulide^® -treated animals. These observations are key given that the inducible vascular COX2 serves important functions in maintaining vascular homeostasis and plays a role in the reported adverse cardiovascular events observed in clinical trials with selective COX2 inhibitors (Konstantinopoulos and Lehmann 2005; Bresalier et al. 2005).

No differences among the dogs treated with Nimesulide^® beginning at ages 78 or 202 days and treated for 394 or 450 days or the untreated matched dogs were seen in terms of the white matter hyperintense lesions detected by brain MRI. Treated and untreated dogs exhibited loosening of the frontal neuropil with leaky blood vessels, extravasation of red blood cells, strongly positive GFAP perivascular reactivity, and enlarged Virchow-Robin spaces in the areas identified as hyperintense WML lesions by MRI (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). In the literature, WML appear to be related to chronic microvascular disease and hypoperfusion and their presence relates to frontal-subcortical circuitry disruption and translates in cognitive impairment (Delano-Wood et al. 2008; Wright et al. 2008). Vascular changes predominantly involving subcortical areas are seen in both Alzheimer’s and vascular dementia patients and in mixed dementia cases (Jellinger and Attems 2007). The common denominator for the hyperintense white matter lesions detected by brain MRI appears to be a vascular lesion with perivascular gliosis and enlarged Virchow-Robin spaces. Similar frontal WML are detected in 56.5% of Mexico City children average age 10.7 + 2.4 y and coincided with cognitive impairment (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Our recent observations of cognitive deficits and white matter alterations prompt us to quantify mRNA Aquaporin-4 in this study for several reasons: (1) AQP4 plays a key role in three distinct processes: brain water balance, astroglial cell migration, and neural signal transduction (Papadopoulos and Verkman 2008); (2) we have alterations in the blood-brain-barrier in both young dogs and humans exposed to air pollutants (Calderón-Garcidueñas, Solt, et al. 2008), and AQP4 abnormal expression patterns relate to BBB disturbances (Warth et al. 2007, Papadopoulos and Verkman 2007, Papadopoulos and Verkman 2008); (3) AQP4 in the most prominent water channel is restricted to the glia limitants and astrocytic endfeet, and its expression in the cerebral cortex is increased in the early stages of Alzheimer’s disease (Pérez et al. 2007); and (4) we have shown a significant increase in mRNA AQP4 in Mexico City subjects older than 25 y versus <25 y (Calderón-Garcidueñas, unpublished results). However, no differences were detected in Aquaporin-4 among the groups of dogs, suggesting that the potential abnormalities in the cerebral microcirculation/glia limitants and/or astrocytic feet alterations are not yet significant enough to alter the AQ4 in these young animals.

Defining the mechanisms by which NSAIDs could have a protective memory/antineuroinflammatory effects is crucial, especially in the setting of protecting against neuroinflammation in children and young adults chronically exposed to significant concentrations of air pollutants (Calderón-Garcidueñas, Solt, et al. 2008; Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). An important point to emphasize is that the protective effect of NSAIDs revealed in epidemiological studies (reviewed in McGeer and McGeer 2007) involved long-term use in keeping with the observations that inflammatory events are present early in the Alzheimer’s disease process (Akiyama et al. 2000; Minghetti 2005). Thus, drugs potentially useful as neuroprotectors should target the key mechanisms responsible for the neuroinflammatory/degenerative process and should have minimal side effects on long-term prophylactic use (Weggen, Rogers, and Eriksen 2007). COX2 is a constitutively expressed enzyme, regulated by synaptic activity with significant contributions to synaptic plasticity under physiological conditions; however, an upregulation of COX2 is associated with neuronal injury, as seen in acute and chronic degenerative neurological diseases (Hewett, Bell, and Hewett 2006).

Mechanisms associated with NSAIDs’ effects that may be relevant to their neuroprotection for Alzheimer’s disease include (1) selective modulation of amyloid β_1-42 production with shifting cleavage of APP, the result of inhibition of γ secretase, an effect independent of the COX2 pathway (Pasinetti 2002; reviewed in Weggen, Rogers, and Eriksen 2007; Hoshino et al. 2007); (2) inhibition of the enzymatic activity of COX1 and inducible COX2, which catalyze the first committed step in the synthesis of prostaglandins with resulting reduction in oxidative damage, decrease activation of NFκB, decrease amyloid burden, and decrease formation of Aβ oligomers (Lukiw and Bazan 1998; Boutaud et al. 2002; Qin et al. 2003; Liang et al. 2005; Jang and Surh 2005; Hewett, Bell, and Hewett 2006); (3) the blockade of a COX2-mediated PGE2 response at synapses (Kotilinek et al. 2008); (4) COX2 inhibition and improved endothelial dysfunction through the inhibition of PGE on endothelial cells (Abdelrahman and Suleimani 2008); (5) immunomodulatory effects including macrophage and T cell–mediated immune responses such as TNFα release and nitric oxide production and lymphocyte proliferation (Cho 2007). The role played by COX isoforms in neurodegenerative diseases is still controversial, and a clear knowledge of the COX2 physiological role in the CNS is missing (Minghetti 2008; Ajmone-Cat, Cacci, and Minghetti 2008; Hoozemans et al. 2008).

Nimesulide^® -treated dogs are showing some potential neuroprotective effects, that is, a decrease of plasma reactive oxidants, while others such as the downregulation of IL1β in the heart could potentially translate in decreased angiogenesis. An issue that needs to be addressed if we would like to use cyclooxygenase inhibitors in individuals exposed to high concentrations of air pollutants is that prevention trials are likely to be the trials of choice, and drugs likely have to be administered early. We know that 100% of Mexico City dogs ages 12 to 19.8 months in this study have an accumulation of Aβ already in target brain regions, 58.8% of APOE 2/3 and 3/3 children and young adults resident in the same city have beta amyloid, while 100% of the APOE 4 carriers do (Calderón-Garcidueñas, Solt, et al. 2008). Furthermore, we also know that 56% of MC children age 10.5 ± 3.7 y have hyperintense white matter lesions by MRI and significant cognitive deficits compared with matched low-pollution residents (Calderón-Garcidueñas, Mora-Tiscareño, et al. 2008). Evidence that chronic exposure to severe air pollution is associated with decreased smell function is also available (Calderón-Garcidueñas et al. 2009). Healthy Mexico City 21.1 ± 2.6-year-olds exhibited deficits in the University of Pennsylvania Smell Identification Test (UPSIT): 35.5% versus 12% of controls residing in a low-polluted city. Moreover, carriers of an APOE ɛ 4 allele with less time of residency in MC failed significantly more UPSIT items known to be sensitive to Alzheimer’s disease than their APOE 2/3 and 3/3 MC and control counterparts, implying that a combination of environmental exposure factors and genetics plays a key role in influencing the brain responses to continuous pollutant exposures over the lifetime of the individual (Calderón-Garcidueñas et al. 2009).

Long-term prevention trials aimed at preventing or delaying onset of disease carry risks on initiation of treatment and the problems of selecting the outcome measurements, approach to recruitment of young people, and issues related to monitoring. Furthermore, the prospect of potential benefit is down the road (Meinert 2008). Thus, work is necessary in animal models to define the timing and pathophysiology of the brain lesions associated with polluted environments; to select the appropriate biomarkers and diagnostic tests (i.e., brain volumetric and segmental MRI); and to follow the selection of drugs that could have an impact on the development of the neuroinflammation, the accumulation of abnormal proteins, the brain MRI structural/volumetric changes, and the cognitive deficits. Dogs are suitable models since they share with humans the brain effects of the aging process (Cummings et al. 1996; Kimotsuki et al. 2005, Head 2009; Opii et al. 2008), including behavioral changes and cognitive decline, as well as progressive β-amyloid deposition, diffuse plaques, amyloid angiopathy, and brain MRI changes.

In the present study, we provide support for the restricted neuroprotection given for Nimesulide^®, n-(4-nitro-2-phenoxyphenyl) methane sulfonamide, a preferential cyclooxygenase-2 inhibitor, under the exposure conditions and the ages at the beginning of the drug administration in the dog model used. Furthermore, our data support previous studies with Tg mice showing that Nimesulide^® is not effective in reducing Aβ_1-40 or Aβ_1-42 levels, reducing Aβ amyloid plaque deposition, or altering the Aβ precursor protein metabolism (Sung et al. 2004 ). In the same study, mice receiving Nimesulide^® showed a partial but not significant reduction in total cerebral cortex and hippocampus PGE2 (Sung et al. 2004). Selection of Nimesulide^® for our dog pilot study was based on the minimal secondary effects observed in dogs and the high pKa 6.5 and its diffusion capacity into the brain (Taniguchi et al. 1997). Since, IL1 β is one of the proinflammatory cytokines with sustained upregulation in both dogs and humans exposed to the severe air pollution in southwest MC, the selection of Nimesulide^® was also based on the reported downregulation of IL1β in treated rats (Taniguchi et al. 1997).

Limitations of this pilot study include (1) the Nimesulide^® treatment was started at ages 78 or 202 days, even though brain pathology associated with air pollution exposures can be detected in the exposed dogs as early as 2 weeks after birth (Calderón-Garcidueñas et al. 2002, 2003); (2) in the evaluation of our endpoints, particularly prostaglandin E metabolite and 3-nitrotyrosine, serial determinations after the last dose of the medication would have been more appropriate and PGE2 determinations in the target brain areas more ideal; (3) a more complete cognitive study with longitudinal evaluations as the dogs aged would have been preferable; and (4) given that in human epidemiological studies NSAIDs are preventive agents associated with a decreased risk for Alzheimer’s disease only when subjects take the medications for long periods >24 m (McGeer and McGeer 2007), an extension of the treatment period in dogs should be considered for future studies.

In summary, dogs exposed to high concentrations of air pollutants developed neuroinflammation and accumulation of Aβ₄₂ in brain target areas at an early age. The administration of a COX2 preferential inhibitor produced some beneficial effects but no significant changes in neuroinflammation or the accumulation of amyloid. We are seeing similar neuropathological findings in highly exposed children and young adults resident in Mexico City, along with structural MRI brain alterations and significant cognitive deficits. Thus, the search for potentially beneficial drugs helping to ameliorate the brain effects of pollution represents an enormous clinical challenge. Dogs may be a useful model for examining behavioral, brain cellular, and molecular processes associated with environmental exposures and their responses to drugs aimed at the prevention and treatment of diseases such as Alzheimer’s. Recognition and mitigation of environmental factors that promote neuroinflammation, accumulation of abnormal proteins, brain structural damage, and cognitive deficits is crucial in limiting the future health impact of air pollutants on our exposed populations. Alzheimer’s disease is increasing at an accelerated rate in both developed and underdeveloped countries. By 2050, the incidence of Alzheimer’s disease is expected to approach nearly 1 million people per year, with a total estimated prevalence of 11 to 16 million persons in the United States alone (Hebert et al. 2003). Exposure to air pollution is likely a risk factor for the development of Alzheimer’s disease, and thus, there is an unprecedented opportunity for prevention.