Three-Dimensional Mapping of Ozone-Induced Injury in the Nasal Airways of Monkeys Using Magnetic Resonance Imaging and Morphometric Techniques

Abstract

Age-related changes in gross and microscopic structure of the nasal cavity may alter local tissue susceptibility as well as the dose of inhaled toxicant delivered to susceptible sites. This article describes a novel method for the use of magnetic resonance imaging, 3-dimensional airway modeling, and morphometric techniques to characterize the distribution and magnitude of ozone-induced nasal injury in infant monkeys. Using this method, we generated age-specific, 3-dimensional, epithelial maps of the nasal airways of infant Rhesus macaques. The principal nasal lesions observed in this primate model of ozone-induced nasal toxicology were neutrophilic rhinitis, along with necrosis and exfoliation of the epithelium lining the anterior maxilloturbinate. These lesions, induced by acute or cyclic (episodic) exposures, were examined by light microscopy, quantified by morphometric techniques, and mapped on 3-dimensional models of the nasal airways. Here, we describe the histopathologic, imaging, and computational biology methods developed to precisely characterize, localize, quantify, and map these nasal lesions. By combining these techniques, the location and severity of the nasal epithelial injury were correlated with epithelial type, nasal airway geometry, and local biochemical and molecular changes on an individual animal basis. These correlations are critical for accurate predictive modeling of exposure-dose-response relationships in the nasal airways, and subsequent extrapolation of nasal findings in animals to humans for determining risk.

Introduction

The nasal cavity is often the first part of the respiratory tract exposed to inhaled airborne pollutants. The mucosal membranes lining the nasal airways play an important role in protecting the more distal regions of the respiratory tract by absorbing water-soluble gases and vapors (Brain, 1970; Morgan, 1977), filtering airborne particulate matter (Lippmann, 1970; Proctor and Anderson, 1982), and metabolizing xenobiotic agents (Dahl et al., 1982; Dahl and Hadley, 1991). Their proximal location in the respiratory tract also makes the nasal mucous membranes susceptible to injury from certain highly reactive inhaled toxicants, including the common oxidant air pollutant, ozone.

Understanding the pathogenesis of ozone-induced airway injury and remodeling in the upper respiratory tract requires knowledge of the dose delivered to the target airway tissues and the airway tissue’s response to exposure. Nasal injury caused by ozone exposure has been shown to be site-specific in both laboratory rodents and primates (Harkema et al., 1987a). The intranasal location of these ozone-induced lesions may be attributable to the local dose of ozone reaching the affected area, the site-specific tissue susceptibility, or a combination of these factors (Morgan and Monticello, 1990). To determine the relative importance of each factor, it is essential to first microscopically identify the ozone-associated lesions in the nasal passages and precisely map the distribution of these specific lesions relative to airway geometry and epithelial susceptibility.

In previous reports, this has been accomplished experimentally using a sequential series of cross-sectional airway profiles covering the entire length of the nasal airways of each experimental animal. This systematic approach has been used to identify patterns of nasal toxicity to inhaled chemicals in adult monkeys (Young, 1981; Harkema et al., 1987d; Mery et al., 1994; Kepler, 1995), and relies on precise characterization and identification of the distribution of surface epithelial populations (Gross et al., 1982, 1987; Harkema et al., 1987a; Harkema, 1991). This type of detailed characterization, however, has not been previously reported for the developing nasal airways of infant monkeys. Specific age-related differences between adult and infant monkeys in nasal airway geometry, epithelial distribution, and mucosal biochemistry make the upper airways of adult monkeys unsuitable as predictive models for nasal toxicology in children.

The infant rhesus macaque serves as an important animal model of upper airway toxicology because their nasal airways resemble, at a gross and microscopic level, the nasal airways of humans (Harkema, 1991; Yeh et al., 1997). Knowledge of the patterns of ozone-induced airway lesions in the developing nasal passages of immature monkeys provides insight into the relative roles played by regional dose and local tissue susceptibility in the toxic responses to ozone in the nose, and ultimately, assists in the assessment of the potential risk of ozone exposure to the health of children. Using a combination of magnetic resonance (MR) imaging, 3-dimensional (3D) modeling, and standard histopathologic and morphometric techniques, we developed a detailed and flexible approach of nasal analysis to identify, localize, and precisely map normal nasal airway epithelium and ozone-induced nasal airway lesions. This method also provides a means to correlate morphologic changes to site-specific biochemical and molecular changes in the airways (e.g., regional antioxidant regulation, local expression of pro-inflammatory genes). Identifying these correlations is essential for developing predictive models of dosimetry, and defining exposure-dose-response relationships in nasal airways of immature non-human primates that are crucial for determining the mode of action in the nasal toxicity of ozone and other inhaled toxicants.

While the use of these techniques provides important and relevant human risk assessment information, the use of a non-human primate model presents both ethical and financial challenges that cannot be overlooked. Technical methods must be considered carefully to ensure that the experimental design is both efficient and thorough. During the past several years, we have established a collaborative effort to investigate the mechanisms of age-dependent airway injury in response to ambient ozone exposure using this model. In this article, we review the methods that we have employed for ozone exposure, nasal airway imaging, nasal tissue processing, and 3D nasal airway mapping in the infant rhesus monkey. We focus attention here on the details of these methods that provide an efficient and comprehensive characterization of both airway geometry and histopathology to maximize the information obtained from each individual animal.

Materials and Methods

Animals and Exposure Regimens

The principal laboratory animal used in our ozone inhalation studies was the infant, male Rhesus macaque (Macaca mulatta). All monkeys selected for these studies were born and raised at the California National Primate Research Center. Care and housing of animals before, during, and after treatment complied with the provisions of the Institute of Laboratory Animal Resources and conformed to practices established by the American Association for Accreditation of Laboratory Animal Care (AAALAC).

While several rodent models of environmental airway disease have been previously utilized for human risk assessment, some aspects of the airway anatomy and physiology of rodents make them less desirable for direct extrapolations to humans. For example, rodents are obligate nasal breathers, while humans may alternate between nasal and oronasal breathing. Humans also possess simple nasal turbinates, while rodents possess a very complex nasal turbinate structure (Harkema et al., 2006). Furthermore, the relative surface epithelial populations lining the nasal airways of rodents and humans differ considerably.

Approximately 50% of the surface area of the nasal cavity in F344 rats is lined by olfactory epithelium (Gross et al., 1982), while only 3% of human nasal airways are dedicated to this sensory neuroepithelium (Sorokin, 1988). Each of these factors ultimately changes the dose of inhaled toxicants delivered to susceptible sites between rodents and humans (Kimbell, 2006). A nonhuman primate model species was selected because the structural and cellular aspects of the nasal airways of monkeys most closely represent those in humans (Yeh et al., 1997). Furthermore, infant rhesus monkeys exposed to ozone represent a more relevant model of oxidant airway disease in children (Evans et al., 2003; Larson et al., 2004).

Inhalation studies were designed to compare the epithelial responses to acute ozone exposure to the responses induced by episodic ozone exposure in the nasal airways of 90-day-old and 180-day-old (infant) rhesus monkeys. Acute regimens consisted of 5 consecutive days of exposure (8 hours/day) to 0.5 ppm ozone. Episodic regimens consisted of 5 biweekly cycles of alternating filtered air (9 consecutive days of 0 ppm [FA]) and ozone (5 consecutive days of 0.5 ppm ozone, 8 hours/day) exposure. Episodic exposures were designed to mimic intermittent exposure to clean and polluted air, which likely mimics the nature of exposure experienced by people in high pollution regions. Control animals for each regimen were exposed to FA. Exposures were scheduled so that each animal reached its target age (90 or 180 days) at the end of the exposure regimen. All exposures were performed at the California National Primate Research Center on the campus of the University of California at Davis.

Ozone exposures were conducted in 4.2 cubic meter (m³) stainless steel and glass, whole-body inhalation chambers, similar to those previously described (Hinners et al., 1968). Ozone was produced from vaporized liquid, medical grade oxygen by electric discharge ozonizers. Generated ozone was mixed with filtered dilution air and injected into the inlet air-flow of the exposure chamber. Filtered air was supplied to the chamber at a flow rate of 2.1 m³ per minute, providing 30 air exchanges per hour. Chamber ozone concentration was monitored throughout exposures with an ultraviolet ozone analyzer (Model 1003-AH Dasibi Corporation, Glendale, CA). During exposures, a PC-based data acquisition and control system was used (Dell Computer Corporation, Round Rock, TX and GE Fanuc Automation North America, Inc., Charlottesville, VA). In addition to monitoring ozone concentration, this computer system also monitored chamber airflow, temperature, pressure, and humidity.

Necropsy and Tissue Preservation

At the end of exposure, monkeys were sedated using ketamine hydrochloride (10 mg/kg IM), followed by induction of deep anesthesia using propofol (Diprivan, 0.1–0.2 mg/kg/min IV). The trachea, sternum, and abdominal cavity were exposed by a midline incision. Animals were euthanized by exsanguination via the abdominal aorta.

Immediately after sacrifice, the head was removed from the carcass, and the lower jaw, skin, musculature, and brain were removed from the head. The nasal cavity was exposed by splitting the skull sagittally, 1–2 mm to the right of midline. This sectioning provided a complete left nasal cavity, including an intact nasal septum, and a free right lateral wall, with the ethmoturbinate and maxilloturbinate attached (Figure 1). The left side of the nasal cavity, including the nasal septum, was flushed with a solution of 1% paraformaldehyde and 0.1% glutaraldehyde, and was then immersed in ~100 ml of the same solution for storage at 4°C until preparation for MR imaging.

The right side of the nasal cavity was used to sample nasal mucosal tissues from specific locations throughout the lateral wall of the nasal cavity for RNA isolation or biochemical antioxidant analyses. Mucosal samples were dissected from the anterior lateral wall between the nasal vestibule and the maxilloturbinate (preturbinate region), the anterior and posterior regions of the maxilloturbinate, the anterior and posterior aspects of the ethmoturbinate, and the lateral wall of the nasopharynx (Figure 1). Each mucosal sample was divided into 2 equal parts. One mucosal sample from each region was stored in 0.5 ml of RNALater (Ambion, Austin, TX), allowed to remain at room temperature for 24 hours, and stored at −20°C until further processing for RNA extraction. The other mucosal sample from each region was immediately placed into a solution of 10% meta-phosphoric acid, snap-frozen in liquid nitrogen, and stored at −20°C until further processing for antioxidant analysis via high performance liquid chromatography (HPLC). Results from biochemical and gene expression analyses of these mucosal tissues from the right lateral wall are not presented in this report.

Magnetic Resonance Imaging of Nasal Airways

MR imaging of formalin-fixed nasal cavities has previously been exploited to facilitate measurements of airway geometry in cynomologus monkeys (Harris et al., 2003). In the present study, a similar approach was employed to rapidly compile detailed 3-dimensional descriptions of airway architecture postmortem, but prior to nasal airway sectioning and processing for light microscopic examination.

In preparation for MR imaging, the left nasal cavities were first flushed with a fixative solution of 1% paraformaldehyde/0.1% glutaraldehyde containing the MR contrast agent Magnevist (gadopentetate dimeglumine, Berlex Laboratories, Wayne, NJ), at a volume concentration of 1 part Magnevist per 500 parts fixative. Each specimen was then placed in a small container (60 ml volume, 42 mm I.D., 59 mm length), and immersed in the same fixative solution.

The use of Magnevist in the fixative solution ensured that acquired MR images had good signal-to-noise and high contrast between the nasal tissue and the nasal airspaces; thereby, facilitating confident segmentation of airway geometry. In preparation for imaging, each specimen was positioned in the specimen container with the rostral aspect of the nasal cavity facing the bottom of the container, and the plane of the palate parallel to the long axis of the container. The specimen was fixed into this position using soaked 2°° × 2°° gauze sponges packed around the caudal aspect as needed. Prior to sealing each bottle, the specimen was then degassed to eliminate any small air bubbles that might otherwise adversely affect MR image quality.

All MR imaging was performed at the Environmental Molecular Sciences Laboratory⁷ (Richland, WA). High-resolution, 3D MR images of fixed nasal tissue were acquired at 2.0 Tesla using a Varian UnityPlus spectrometer (Palo Alto, CA) and an in-house fabricated radio frequency (Alderman-Grant) coil with an inner diameter of 5.5 cm. To visualize complex nasal passages raw image data was collected on a 256 × 128 × 128 matrix using a T₂-weighted, 3D spin-echo sequence with a 100 msec echo-time, a 150 msec repetition time, and 4 averages. Each 3D data set therefore required ~2.7 hours to collect, during which, sample temperature was maintained at ~4°C. After Fourier image reconstruction, all 3D data was stored as a series of 256 2D slices—each showing a squared 4.5 cm field-of-view. Planar resolution within each slice was 350 microns and each slice was 200 microns thick.

Image Analysis and Airway Segmentation

All airway segmentation and mesh generation was performed using methods similar to those already described by Minard et al. (2006). Briefly, Digital Data Viewer (DDV; Computer Geometry Consulting, 〈http://www.compgeomco.com〉) was used to identify and digitally segment the airway geometries from the 3D MR image data of each monkey nasal passage. Segmented airways were subsequently used for the development of geometry databases and 3D mesh reconstruction. To facilitate airway segmentation, each set of 256, 2D slices was first read into DDV. The set of images was then viewed in the x, y, and/or z directions, slice by slice. To digitally segment (extract) the nasal passageway, the area on each of the z-slices associated with the airway was highlighted. A separate file (“overlay”) was then generated from this extracted data and it was used to create an isosurface rendering of the 3D segmentation (Figure 2). During airway segmentation, the x-, y-, and/or z-planes were viewed separately or in any combination with the isosurface rendering (Figure 2) to facilitate comparisons between the segmented airways and the raw MR image data. The isosurface generated from the segmentation files of the MR images, along with the images themselves, could be rotated and viewed from any direction to ensure optimal segmentation. Once completed, this made them ideally suited to serve as guides for subsequent histopathologic and morphometric analyses of nasal passages.

A publicly available mesh generator (NWGrid; 〈http://www.emsl.pnl.gov/nwgrid〉) was used to transform the segmented data into a boundary-fitted, volume-filling tetrahedral mesh that was used for computing surface areas and volumes. The volume meshes were designed to be suitable as the initial finite-element mesh for CFD simulations. NWGrid was also used to generate surface area and volume measurements for each nasal airway specimen.

Tissue Processing for Light Microscopy and Morphometric Analysis

After MR imaging, each left nasal cavity specimen was removed from the fixative with Magnevist and decalcified in 13% formic acid for 10–12 days, then rinsed with distilled water for 2–4 hours. After decalcification, the nasal airways were transversely sectioned at 10 or 12 (based on the size of the specimen) specific anatomical locations, using gross dental and palatine landmarks (Figure 3). This method produced nasal tissue blocks that had cross sectional profiles similar to those described by Kepler et al. (1995), and has been used in prior studies to obtain accurate geometrical coordinates of nasal airways of nonhuman primates for computer dosimetry modeling and nasal airway lesion mapping (Kepler et al., 1998). These locations provided transverse sections through the preturbinate region, main nasal airway chamber, and nasopharyngeal region. Each tissue block was embedded in glycol methacrylate (GMA; Immuno-Bed, Polysciences, Inc., Warrington, PA) for airway epithelial mapping and histopathologic analysis.

Three-Dimensional Digital Mapping of Epithelial Distribution

It has been reported that ciliated respiratory epithelium (RE) and nonciliated transitional epithelium (NTE) in the anterior nasal airways are particularly sensitive to ozone toxicity in adult monkeys (Harkema et al., 1987b, 1987c), while other epithelial types remain relatively unaffected. To accurately assess the morphologic response of the nasal mucosa in infant monkeys, we first had to define the type and extent of epithelial populations in the normal infant monkey. To determine the specific distribution of surface epithelial populations lining the nasal airways, and to provide boundary setting information for computer modeling and site-specific tissue dose predictions, we developed a method to generate a detailed, 3-dimensional digital epithelial map of the entire nasal airspace of a 180-day-old, infant, rhesus monkey. The details of this method are described next.

Morphological Identification of Nasal Epithelial Types

There are 5 distinct epithelial populations in the nasal airways of adult macaque monkeys. These include the squamous epithelium (SE), ciliated respiratory epithelium (RE), nonciliated transitional epithelium (NTE), olfactory epithelium (OE), and lymphoepithelium (LE). The morphological features used to identify each epithelial type for 3D mapping are described here. The reader is referred to a recent review article for a more thorough discussion of the structure and function of these epithelial types and their relevance to nasal toxicologic pathology (Harkema et al., 2006).

Squamous Epithelium

The nasal vestibule is lined with keratinized, stratified squamous epithelium. This nasal epithelial type is composed of basal cells lining the basal lamina, and several layers of squamous epithelial cells, which become progressively flatter toward the luminal surface of the epithelium. The epithelium of the vestibule is contiguous to the epidermis of the skin, and both serve to protect the underlying tissues from exposure to potentially harmful agents in ambient air. In adult monkeys, the squamous epithelium of the nasal vestibule forms an abrupt border with the transitional epithelium that lies immediately distal to it in the main chamber of the nasal airways (Harkema et al., 2006).

Transitional Epithelium

Adjacent to the stratified squamous epithelium of the nasal vestibule, and proximal to the ciliated respiratory epithelium within the main nasal chamber, is a narrow zone of nonciliated transitional epithelium (NTE). In monkeys, this epithelium is stratified (4–5 cell layers thick), nonkeratinized, and consists primarily of cuboidal and columnar epithelial cells with a paucity of mucous goblet cells. In adult monkeys, this epithelium is morphologically distinct from the keratinized squamous epithelium anterior to it, and the RE posterior to it.

Respiratory Epithelium

Most of the main nasal chamber in adult macaque monkeys is lined with ciliated respiratory epithelium. This pseudostratified to columnar epithelium is comprised primarily of ciliated cells, with lesser numbers of narrow nonciliated (serous) cells, various numbers of mucous (goblet) cells, and basal cells.

Olfactory Epithelium

The olfactory epithelium covers a small region of the dorsal meatus and the dorsal half of the ethmoturbinate in monkeys. The 3 principal cell types that compose the olfactory epithelium are the olfactory sensory neurons, sustentacular cells, and basal cells. The supporting, sustentacular cells form a prominent apical row of nuclei. The olfactory sensory neurons are bipolar neuronal cells interposed between the sustentacular cells. The dendritic portions of these neurons extend above the epithelial surface and terminate into a bulbous olfactory knob, from which extend 10–15 immotile cilia per cell. Basal cells line the basal lamina.

Lymphoepithelium

Lymphoepithelium is a specialized epithelium, found primarily in the walls of the nasopharynx. This epithelium covers discrete, focal aggregates of nasal-associated lymphoid tissue (NALT) in the underlying lamina propria. The lymphoepithelium is composed of cuboidal ciliated cells, few mucous cells, and numerous nonciliated, cuboidal cells with luminal microvilli, called M-cells. These latter cells are similar in structure and function to the analogous immune surveillance cells found in gut- and bronchus-associated lymphoid tissues (GALT and BALT, respectively).

Translation of Light Microscopic Data to 3D Epithelial Map

One nasal airway specimen from a filtered air-exposed monkey was selected for complete epithelial mapping. Initial slides were taken from the anterior face of each tissue block and stained with hematoxylin and eosin (H&E) for routine histopathologic examination. Slides were also stained with AB/PAS for detection of acidic and neutral intraepithelial mucosubstances. Additional unstained slides were taken for use in future immunohistochemical analyses (the analyses of intraepithelial mucosubstances and immunohistochemistry are not presented in this report). The remainder of each tissue block was completely sectioned at 50 μm increments, and slides were stained with H&E for routine histopathologic analysis.

Each z-slice of the MR image file was 200 μm thick. Sectioning the block at 50 μm increments resulted in 4 H&E slides for each image. The H&E slide that best corresponded to the shape of the nasal airspace for each image was selected and examined by light microscopy. Each epithelial type was coded to a color in DDV. The location of nasal epithelial types (squamous epithelium, nonciliated transitional epithelium, respiratory epithelium, olfactory epithelium, lymphoepithelium) in each section was recorded on the corresponding transverse MR image by drawing a color-coded line along the air-tissue interface at the corresponding site (Figure 4A). This process was repeated in all transverse images. These data were saved as a DDV overlay file, and were used for generation of a 3-dimensional nasal epithelial map as explained below.

In computational fluid dynamics (CFD) models, different epithelium-specific gas uptake properties may be accounted for by setting different boundary conditions for gas uptake at epithelium-specific locations on the nasal surface. This requires a seamless, single surface where different epithelial types are defined as regions of the surface. Our approach for producing this surface was to construct separate 3D reconstructions of each epithelial type as well as the entire nasal surface in DDV and import these reconstructions as stereolithography (STL) files into ICEM-CFD (Ansys, Inc., Canonsburg, PA), which is a computer-aided design (CAD) software used for meshing. An STL file defines a surface enclosing a given volume.

Therefore, when exporting the epithelial reconstructions created in DDV, we actually exported a volumetric blob that corresponds to the volume defined by the voxels (3D pixels) drawn in DDV while delimiting each epithelial surface (Figure 4B). The surfaces confining each of these volumetric blobs were intersected with the whole-nose surface in ICEM-CFD to determine the 3-dimensional curves separating the different epithelia. These curves were then used to divide the surface of the model into epithelial regions. The final result was a smooth, seamless, single surface of the nasal airway composed of different parts, each representing a particular epithelial type (e.g., squamous, respiratory) (Figure 5). This 3D reconstruction was used to visualize the locations of the major epithelial types and to estimate epithelium-specific surface areas.

Morphometric Quantitation of Ozone-Induced Nasal Lesions

Light Microscopic Morphometric Analysis

Nasal airway tissues were processed, decalcified, and embedded as described previously. Slides were prepared from the anterior face of each of the 10–12 tissue blocks at a 1–2 μm thickness. Slides were stained with H&E for routine histopathologic examination and morphometric evaluation. Slides from each section were also taken for AB/PAS staining and immunohistochemistry as described here.

A board certified veterinary pathologist (JRH) examined the transverse microscopic sections of each nasal specimen. The location, nature, and extent of the specific mucosal lesions in each transverse microscopic section were characterized for each animal. The nasal epithelial lesions identified in the current study were confined to the anterior nasal cavity, and were primarily associated with the ciliated respiratory epithelium. To quantitate the loss of respiratory epithelium in this region following ozone exposure, we used standard morphometric techniques to measure the thickness of the respiratory epithelium lining the dorsal surface of the anterior maxilloturbinate (Figure 6A).

Severity of respiratory epithelial cell loss was quantified by measuring the average height of the epithelium overlying the dorsal surface of the first 3 transverse sections of the maxilloturbinate. Thickness of the RE within this region was morphometrically evaluated as previously described for airway epithelium (Hyde et al., 1990, 1991; Plopper et al., 1994). All measurements were obtained at a final magnification of 1,710x using a light microscope (Olympus BX40; Olympus America, Inc., Melville, NY) coupled to a 3.3-megapixel digital color camera (Q-Color 3 Camera; Quantitative Imaging Corporation, Burnaby, British Columbia, Canada), and a personal computer (Dimension 8200; Dell, Austin TX). The morphometric analyses were performed using a 135-point cycloid grid overlay with an automated software package for counting points and intercepts within the grid (Stereology Toolbox; Morphometrix, Davis, CA) (Hyde et al., 1990, 1991). The percent volume density (the proportion of the total epithelial volume), V _ν, of cytoplasm, nuclei, and intraepithelial mucosubstances, was determined by point counting and calculated using the following formula:

$$V_v=P_p=P_n/P_t$$(1)

where P _p is the point fraction of P _n , the number of test points hitting the structure of interest (e.g., cytoplasm), divided by P _t , the total number of points hitting the reference space (i.e., respiratory epithelium). The volume of each epithelial component of interest per unit of basal lamina length (S _ν) was determined by point- (epithelial component) and intercept-(basal lamina) counting and was calculated using the following formula:

$$S_{\nu }=2I_{\text{o}}{L}_{\gamma }$$(2)

where I _o is the number of cycloid intercepts with the object (epithelial basal lamina), and L _γ is the length of the test line in the reference volume (epithelium within the test grid). To determine the thickness of the RE, a volume per unit area of basal lamina (cubic micrometers per square micrometer) was calculated using the following formula for arithmetic mean thickness (τ):

$$\tau =V_{\nu }/S_{\nu }$$(3)

Statistical Analyses

Statistical analyses for all morphometric parameters were performed using one-way analysis of variance (ANOVA). Pairwise comparisons were selected a priori, and were performed using Student–Newman–Keuls multiple comparisons test (p ≤ 0.05). Statistical outliers were detected using Grubb’s test (p ≤ 0.05).

Three-Dimensional Digital Mapping of Ozone-Induced Nasal Injury

We used a combination of standard histopathologic methods, MR imaging, and computer-assisted, 3D airway modeling to generate subject-specific, 3D maps of ozone-induced nasal lesions. Using DDV, the frontal profile of each transverse section was visually matched to its corresponding z-plane slice (2D image) in the MRI segmentation (Figure 6). The locations of ozone-induced lesions identified on the nasal tissue sections were recorded on the corresponding sites on the z-plane MR images by highlighting the tissue-airspace interface at that site (Figure 7C). The entire z-plane slice, including the highlighted lesion site, was subsequently superimposed over the isosurface rendering of the entire nasal airspace to provide 3-dimensional lesion localization for that specimen (Figure 7D).

Using the slice plane clip sliders in DDV, the specific Cartesian (x-, y-, and z-plane) coordinates for epithelial lesions can be recorded. These coordinates can be read into NWGrid and translated into the corresponding sites on the surface of the computational volume mesh for that animal. The individual cell faces of the volume mesh corresponding to the site of injury can be digitally highlighted using the General Mesh Viewer (GMV, Los Alamos National Laboratory, 〈http://www-xdiv.lanl.gov/XCM/gmv/GMVHome.html〉) (Figure 7E).

Results

Nasal Cavity Surface Area and Volume Calculations

Calculations of total surface areas and volumes of the left nasal passages of 2 FA-exposed and 2 acute ozone-exposed 180-day-old rhesus monkeys are presented in Table 1. It should be noted here that the calculations of surface area and volume, as well as the data used in epithelial mapping, include the nasal vestibule, starting at the first enclosed nasal transverse section, and the nasopharyngeal meatus, caudally to the point at which the left and right sides join to form the nasopharynx. The volume and area measurements presented here also include the maxillary sinus. The data presented here show that there was consistency in the nasal surface areas and volume measurements among the age-matched infant monkeys. There is no significant effect of ozone exposure on the nasal surface area and volume measurements obtained in these monkeys. This finding of consistency in nasal size in infant monkeys is in contrast to the significant interindividual variability in results reported previously for adult monkeys (Schreider, 1986; Gross et al., 1987; Kepler, 1995).

Digital Epithelial Mapping and Surface Area Calculations

Complete epithelial mapping of one-half of the nasal cavity was performed in one 180-day-old, filtered-air exposed rhesus monkey. Estimated surface areas of the five major epithelial types are presented in Table 2. The majority of the nasal airspace in this monkey (65.1%) was lined with ciliated respiratory epithelium, including the ventral half of the main nasal chamber, most of the nasopharyngeal meatus, and the entirety of the maxillary sinus, including the duct. Our results indicated that approximately 16% of the nasal cavity surface area in this infant rhesus monkey was lined by olfactory epithelium. This is consistent with an earlier report estimating that ~14% of the adult rhesus monkey nasal cavity is lined with olfactory epithelium (Gross et al., 1987). We found that the NTE in this infant monkey was confined to a narrow band in the preturbinate region, between the squamous epithelium of the nasal vestibule and the respiratory epithelium of the main nasal chamber. This localization, as well as the finding that NTE comprised approximately 1.4% of the total nasal epithelial surface area in this monkey, is also similar to results obtained from adult rhesus monkeys (Gross et al., 1987). The lymphoepithelium was primarily localized over 2 discrete aggregations of NALT located in the septal aspect of the nasopharynx. A small island of LE was also found in the ventral aspect of the preturbinate region.

Ozone-Induced Nasal Epithelial Injury and Morphometry

All monkeys acutely exposed to ozone for 5 days had a moderate to marked necrotizing rhinitis that was restricted principally to the nasal mucosa lining the middle meatus in the main nasal chamber. There were focal regions of epithelial exfoliation, and these lesions were locally extensive in the respiratory epithelium lining the dorsal surface of the anterior maxilloturbinate. There was an anterior to posterior decrease in the severity of these intranasal ozone-induced lesions. In 90-day-old monkeys acutely exposed to ozone, the most severely affected areas of the nasal mucosa had near full-thickness necrosis and exfoliation of the surface epithelium with numerous neutrophils, and lesser numbers of eosinophils, infiltrating the subepithelial lamina propria and the remaining surface epithelium (Figure 8). Ninety-day-old monkeys that were episodically exposed to 5 cycles of ozone also had ozone-induced rhinitis and epithelial necrosis in relatively the same intranasal locations as the ozone-induced lesions in the acutely exposed animals.

Morphometric analyses of these ozone-induced nasal lesions indicated that 90-day-old monkeys that were acutely exposed to ozone (0.5 ppm; 8 hours/day for 5 days) had a 65% reduction (compared to filtered air controls) in the mean thickness of the nasal epithelium lining the dorsal surface of the anterior maxilloturbinate due to toxicant-induced necrosis and exfoliation of this targeted site of nasal airway epithelium. Furthermore, this acute ozone-induced atrophy of the nasal epithelium resulted in an 88%, 64%, and 59% loss of the volume densities of airway cilia, epithelial cytoplasm, and epithelial nuclei, respectively. Interestingly, the character, severity, and distribution of the ozone-induced nasal epithelial lesions in episodically exposed monkeys were similar to those in the acutely exposed infant monkeys of similar age. The loss of mean airway epithelial thickness and the volume densities of airway cilia, epithelial cytoplasm, and epithelial nuclei were only slightly less in infant monkeys that were exposed to episodic ozone (Figure 9).

Discussion

This is the first report to describe in detail the 3-dimensional representation of the normal epithelial distribution in the nasal airways of an infant, nonhuman primate model. This technique was designed to develop age-specific epithelial maps of the nasal airways in experimental animal models for use in toxicant-induced lesion mapping and CFD simulations of site-specific tissue dosimetry. Our results in this infant monkey, when compared to a previous report in adults, suggest that the relative amounts of the five epithelial types in the nasal airways of monkeys remained consistent between infancy and adulthood (Gross et al., 1987).

However, because there is considerable change in the size and shape of the head of the rhesus monkey from the neonatal to the adult stage, we estimated that there would be significant age-related differences in the location of susceptible epithelial cell populations. These differences in airway geometry and epithelial lining may both contribute to differences in the responses to oxidant pollutant exposure by altering the dose delivered to susceptible epithelial types. This technique allowed for the determination of the specific distribution of susceptible epithelial populations (i.e., RE, NTE), and ultimately, will serve to reduce uncertainty in CFD dosimetry modeling and extrapolation of data from experimental animal models to humans (Kimbell, 2006).

It is also crucial to understand the distribution of the normal epithelial populations in infant monkeys to be able to adequately distinguish age-related epithelial differences from epithelial alterations induced by toxicant exposure. For example, both squamous metaplasia and respiratory metaplasia of the olfactory epithelium have been reported as adaptive sequelae to toxicant exposure (Harkema et al., 2006). Prior knowledge of the normal distribution of these epithelial types in the infant nasal airways is needed to make this distinction. A clear limitation to the data reported here is that this map represents the nasal airway from a single, FA-exposed, infant rhesus monkey. In future studies, we plan to complete additional epithelial maps for 90- and 180-day-old FA-exposed animals.

The utility of 3-dimensional imaging techniques in laboratory animals for toxicant-induced lesion screening and dosimetry modeling has been previously reviewed (Menache et al., 1997; Wiethoff et al., 2001; Robinson et al., 2003). These approaches used semi-quantitative or subjective assessments of airway lesions applied to models of the nasal airspaces of rodents. The use of morphometric techniques for the quantification of respiratory structures has also been the subject of a recent review (Hyde et al., 2006).

We report the first use of MR imaging, computational biological methods, and standard morphometric techniques to provide a quantitative and site-specific assessment of the nasal responses to inhaled respiratory toxicants in a non-human primate model. The methods described in the current report demonstrate the use of stereologic methods to estimate volume density (Vs) of the airway epithelium as a quantitative assessment of the epithelial response to ozone exposure in the nasal airways of infant monkeys. In addition to the Vs methods described here for measurement of airway epithelium, other morphometric assessments of the epithelium could be combined with 3D mapping to provide additional information about the response to toxicant challenge.

Morphometric techniques may be used to measure the numeric cell density of specific epithelial cells per length of basal lamina (i.e., volume of cells per unit area of basal lamina, μm³/μm²) as an assessment of hyperplastic or metaplastic responses. These methods may also be combined with immunohistochemistry to quantitate other responses, including inflammation (e.g., # of neutrophils per mm basal lamina) and cell proliferation data (e.g., proliferating cell nuclear antigen, PCNA). By providing additional depth to the characterization of the tissue response to toxicant exposure, the measurement of these types of responses is crucial to the establishment of exposure-dose-response relationships in the nasal airways of infant monkeys. Morphometric quantitation of these epithelial responses, combined with site-specific 3D mapping, provides objective comparisons of the responses of monkeys to other species, facilitates comparisons to other toxicants, and provides an important basis for extrapolation of experimental data to human risk assessment.

3D mapping of toxicant-induced nasal airway lesions is unique in that the pathology from each individual animal can be applied directly to a 3-dimensional model of its own nasal airways for use in CFD simulations. The nasal airway surface area and volume measurements presented in this study, when compared to a previous characterization in adult monkeys (Gross et al., 1987) demonstrated that significant postnatal growth occurs in the nasal airways of monkeys from infancy to adulthood (Table 1). This suggests that the nasal deposition patterns of inhaled toxicants would exhibit age-dependent variation, potentially resulting in exposure of different epithelial types to peak doses of toxicant.

By providing estimates of toxicant dose relative to airway geometry, CFD simulations provide an effective way to explain the contribution of airway geometry to site-specificity. However, the effectiveness of CFD simulations is limited by the accuracy of the initial model geometry. The 3-dimensional models generated from the MR image segmentations provide a logical framework not only for localizing the airway injury and for correlating injury to susceptible epithelial types, but also to serve as the basis for the initial finite-element meshes to be used for CFD simulations.

The Cartesian coordinate system relationships used to record the specific locations of the epithelial lesions in DDV are maintained in the finite element meshes used for predictive dosimetry modeling. This allows for greater accuracy in the correlations between experimentally induced and predicted airway injury, and also allows for more specific testing of hypotheses about transport mechanisms and subsequent extrapolation of animal data to humans. CFD airflow and site-specific tissue dosimetry simulations using this model are in progress.

The nasal injury in the 90-day-old infant monkeys was particularly severe in the ciliated respiratory epithelium along the dorsal surface of the maxilloturbinate. This is consistent with the report that adult monkeys exposed acutely to ozone (0.3 ppm, 8 hours/day for 6 days) also experienced necrosis and exfoliation of ciliated cells in the anterior nasal cavity (Harkema et al., 1987c). Despite differences in the ozone concentration and duration of exposure, our results indicate that the character and location of the ozone-induced nasal lesions in the acutely exposed 90-day-old monkeys reported here are similar to those previously reported for adult monkeys acutely exposed to ozone.

However, while the character and location of toxicant-induced injury can be compared with previously described methods, the imaging and 3D modeling techniques described in the current study also allow for the assessment of the extent of airway lesions. In future studies, we will section the complete nasal airways of ozone-exposed monkeys, in a manner similar to the sectioning described here for epithelial mapping. This will allow for the calculation of surface areas for ozone-induced lesions, and facilitate direct comparisons across age groups and ozone doses.

It is possible that the age-related changes in the surface epithelial cell populations alter the susceptibility of the nasal epithelium to ozone toxicity. Another possibility is that age-related differences in airway geometry alter the local delivery of ozone to susceptible epithelial populations within the nasal cavity. Future CFD simulations will provide valuable information on the relative importance of each of these factors as they relate to the location and extent of ozone-induced nasal airway injury in infant monkeys, and ultimately, in children.

We report here that the loss of mean airway epithelial thickness and the volume densities of airway cilia, epithelial cytoplasm, and epithelial nuclei, were not significantly different between infant monkeys receiving acute or subchronic, episodic ozone exposures (Figure 9). Neither group of ozone-exposed infant monkeys exhibited mucous cell metaplasia in the region of airway injury. We have previously reported that adult monkeys exposed continuously (i.e., daily) to high ambient concentrations (0.3 ppm) of ozone undergo epithelial hyperplasia and mucous cell metaplasia (MCM) in the nasal epithelium.

Furthermore, this epithelial remodeling was associated with attenuation of the epithelial necrosis and inflammation in response to subsequent acute ozone exposure. This ozone-induced epithelial adaptation in adult monkeys may render the epithelium less susceptible to tissue injury in response to subsequent ozone challenge (Harkema et al., 1987b, 1987c). It is important to note, however, that the exposure regimen used in adult monkey studies involved chronic (90 days), daily exposures, while the infant monkeys presented in this report were exposed subchronically (70 days) and episodically.

Nevertheless, these findings suggest that infant monkeys episodically exposed to ozone do not undergo nasal airway epithelial remodeling or adaptation with concurrent attenuation of toxicant-induced epithelial necrosis and inflammation. This also suggests the disturbing possibility that infants may develop a persistent necrotizing rhinitis following episodic exposures without any protective molecular adaptation or morphologic remodeling of the surface epithelium that would make the nasal airways less susceptible to ozone toxicity. The long-term consequences of this ozone-induced nasal injury are yet to be determined.

In addition to anatomic and histologic factors, age-related biochemical and molecular changes may also contribute to differential susceptibility to ozone toxicity. For example, antioxidants in the surface lining fluid and epithelium are considered to provide the first line of defense against inhaled oxidant challenge (Housley et al., 1995; Mudway and Kelly, 1998). However, in some in vitro models, the presence of antioxidants in the surface lining fluid serves to augment oxidant pollutant-induced cell membrane damage (Velsor and Postlethwait, 1997; Ballinger et al., 2005). This differential role of lining fluid antioxidants appears to be concentration-dependent. Age-related differences in the regulation of airway antioxidants, as well as age-related differences in the regulation of inflammation, mucus synthesis and secretion, and epithelial repair at the molecular level, may contribute to heterogeneity in the responses of immature models to ozone exposure.

By splitting the nasal cavity and preserving the right side of the airway for biochemical and molecular analyses, we attained the ability to directly correlate site-specific airway injury and repair with site-specific indicators of tissue susceptibility to ozone toxicity on an individual animal basis. Furthermore, this dissection technique allows for the direct comparison of susceptible and resistant epithelial cell populations within the same animal (e.g., RE vs. OE). The use of morphometry permits correlation of quantified structural features with physiological and biochemical data (Tyler et al., 1985).

Using the 3-dimensional mapping technique, this information can be applied to predictive dosimetry models, increasing the specificity of the results obtained from CFD simulations. Employing this technique, however, assumes that the toxicant-induced nasal injury is bilateral and symmetrical, and that it is not significantly influenced by the nasal cycle, the physiologic alternating of airflow from one nasal passage to the other due to periodic cycling of mucosal congestion and decongestion (Flanagan and Eccles, 1997).

Summary

The site-specificity of ozone-induced nasal injury necessitates that histopathologic assessment be applied with the same site-specificity. By using the stereologic methods described here, we obtain reliable measurements of the airway epithelium, its individual components, and the components of the underlying basal lamina and lamina propria. These measurements provide a quantifiable assessment of the epithelium and its response to inhaled pollutants. The use of an infant rhesus monkey, while providing data from nasal airways that are grossly and microscopically similar to the nasal airways of children, also presents ethical challenges requiring efficient, robust, and pragmatic experimental design and methods. Combining morphometric techniques with 3D mapping provides the advantage of applying information about the character, distribution, and magnitude of nasal airway injury into a single model (Figure 10). Quantitation of these epithelial responses facilitates the comparisons of susceptible and resistant epithelial populations within the same animal. More importantly, this also provides an important method of providing comparisons between experimental animals, and facilitates extrapolations between species with different airway geometries.