Molecular Biology of the Nasal Airways: How Do We Assess Cellular and Molecular Responses in the Nose?

Abstract

We summarize studies herein that relate to use of molecular techniques to assess mechanism of toxicant and carcinogen action on the nasal mucosa. Specifically, we present the results of an in vivo mutagenesis assay with the herbicide alachlor, which causes olfactory mucosal tumors in rats following dietary administration. A positive response was found in olfactory mucosa after 3 mo of treatment. There was no increase in mutant frequency in the adjacent nasal respiratory mucosa or in liver, which are both non-target tissues for alachlor carcinogenesis. We also summarize previous findings of gene expression studies. One on these was a GeneChip experiment aimed at elucidating the mechanism of alachlor olfactory carcinogenesis, wherein we found that oxidative stress and gelatinase genes were upregulated early in the carcinogenic process, while genes consistent with activation of Wnt signaling were activated later in the carcinogenic process. The final example presented summarizes the results of a microarray experiment designed to identify novel olfactory genes involved in the plasticity of the olfactory mucosa. Those studies identified novel olfactory mucosal genes including Sgpl1 and Pon1. In each instance, precise sampling is emphasized and proper controls are discussed, and examples of independent means of validation of genomics experiments are presented.

As an introduction to the topic of evaluation of cellular and molecular responses of the nasal mucosa to toxicant-induced damage, we should summarize some of the factors that make the nasal mucosa so susceptible to toxicant-induced damage. The first and probably most obvious factor is the fact that the nasal epithelia are in direct contact with the external environment and can thereby be directly exposed to irritant and cytotoxic gases, vapors, and particles. Second, it has long been known that both the nasal respiratory and olfactory mucosae have a high content of both phase 1 and phase 2 metabolic enzymes. While metabolic enzymes are generally considered to be beneficial and contribute to the detoxification and elimination of toxicants, there are numerous examples of compounds that are bioactivated by nasal enzymes. The effects of metabolism of chemicals by nasal mucosal enzymes has recently been summarized in the context of chemicals that are either detoxified (e.g., cyanide) or bioactivated (e.g., 2,6-dichlorobenzonitrile, coumarin, phenacetin, acetaminophen, and organonitriles) (Ding and Dahl, 2003). Naphthalene (NP), a nasal carcinogen in rats, can be added to the latter category, as the metabolic enzyme that has been demonstrated to bioactivate NP in the lung, CYP2F, has been demonstrated to be highly expressed in olfactory mucosa, as well (Ritter et al., 1991; National Toxicology Program, 2000; Genter et al., 2003).

It is important to realize that various proteins are not uniformly distributed throughout the olfactory mucosa, despite the fact that olfactory mucosa is histologically very similar from region to region throughout the nasal ethmoid turbinate region. This point was dramatically demonstrated in studies that documented the distribution of putative odorant receptors (Ressler et al., 1993; Young et al., 2003). In addition, the distribution of metabolic enzymes varies in olfactory mucosa in different anatomical locations within the nasal cavity, and this difference in distribution directly impacts the distribution of lesion formation in the nasal cavity. For example, the distribution of alachlor-induced olfactory tumors correlates with the distribution of cytochrome P450 2A3 (CYP2A3) in the ethmoid turbinate region of the rat nasal cavity (Genter et al., 2000) (Figure 1). Similarly, 2,6-dichlorobenzonitrile-induced olfactory mucosal damage was found to occur in regions of the nasal cavity with undetectable (by immunohistochemistry) microsomal epoxide hydrolase (Genter et al., 1995), an observation that is consistent with the known metabolic profile for this compound, involving an epoxide intermediate (Ding et al., 1996; Liu et al., 1996).

There are three basic goals for this summary article. The first is to present data from an in vivo mutagenesis assay, in which the olfactory mucosal carcinogen alachlor was administered to Big Blue transgenic rats. To our knowledge, the use of this technique has not previously been applied to the nasal cavity. The second is to summarize our genomic studies of the nasal mucosa; one of these identified novel highly expressed genes in olfactory mucosa, while the other chroniceled the time course of changes in olfactory gene expression in the progression of alachlor-induced olfactory carcinogenesis. Finally, we demonstrate and provide examples of methods to confirm changes in gene expression that are suggested by GeneChip and microarray experiments, emphasizing methods that do not rely on re-analysis of the original RNA pool. In each case, the importance of precise and reproducible tissue sampling will be emphasized.

Alachlor is one of a class of herbicides known as the chloracetanilides; these compounds (e.g., acetochlor, butachlor, metolachlor, and propachlor), together with alachlor, have a range of toxic and carcinogenic responses in rodents. In rats, these compounds are associated with olfactory mucosal, thyroid, stomach, and/or liver tumors; mice appear to be refractory to the development of these tumors, but instead have an elevated incidence of lung tumors following chronic exposure to alachlor (United States Environmental Protection Agency 1985, 1998; Genter et al., 2004). We hypothesized that target tissue bioactivation to a mutagenic metabolite was critical to alachlor carcinogenesis. In vitro mutagenesis studies (Ames assay and mouse lymphoma assay) were weakly positive (Wetmore et al., 1999). In recognition that one or more circulating alachlor metabolite was likely to be responsible for the carcinogenic activity, not the parent compound itself (Galati et al., 1998), we undertook an in vivo mutagenesis assay, in which alachlor was administered in the diet at 126 mg/kg/d for 0–12 wk to Big Blue rats (Stratagene, LaJolla, CA).

Mutagenesis was evaluated in the cII gene in olfactory mucosa, nasal respiratory mucosa and liver (with the latter 2 tissues serving as negative control tissue, given that they are not targets for alachlor carcinogenesis). Mutant frequency was assayed using the cII assay following isolation of high molecular weight DNA using the Recoverease (Stratagene) method. The ethmoid turbinates, but not the nasal septum, were used as the source of olfactory mucosal DNA, as our data showed that the nasal septum is not a site of alachlor-induced tumors and we thereby did not dilute out mutant DNA with non-target DNA. The nasoturbinates, maxilloturbinates, and anterior one-half of the nasal septum served as the source of nasal respiratory DNA. Packaging, plating, and determination of mutant frequency were carried out following the manual for the λ Select-cII Mutation Detection System for Big Blue rodents (Stratagene; Revision #028001).

Mutant plaques were selected by incubation at 24°C for 40 hr, while total titers were determined following incubation at 37°C for 20–24 hr. Figure 2 shows the outcome of these studies, with a significant increase in mutant frequency in olfactory mucosa following 12 wk of treatment, but no significant increase in mutant frequency in olfactory mucosa at earlier time points, and there was no significant difference in mutant frequency in nasal respiratory mucosa or liver when the 3 mo-treated animals were compared to age-matched untreated controls. The positive response at 3 mo in olfactory mucosa was particularly interesting, as this is the time point at which histological changes were first observed (Genter et al., 2002a). These data support our hypothesis that target tissue bioactivation, likely of the intermediate metabolite 2,6-diethylaniline (Feng et al., 1990; Galati et al., 1998), is critical in the mutagenesis and carcinogenesis of alachlor.

DNA from mutant plaques was sequenced to determine whether alachlor-induced cII mutations were random events, or whether a “hot spot” for mutagenesis could be identified. Single, well-isolated individual plaques from confirmation plates were cored, were used as the source of DNA for sequence analysis. Forward and reverse primers were as follows, respectively: 5′CCGCTCTTACACATTCCAGC-3′ (designated the ‘alternative sequencing primer’ in the Stratagene λ Select-cII Mutation Detection System for Big Blue rodents manual [Revision #028001]) and 5′GGAGACGGCTTCAACTCATA-3′. The PCR reaction was performed with the following parameters: a 3 min denaturation at 95°C, followed by 40 cycles of 30 sec at 95°C, 1 min at 63.5°C, and 1 min at 72°C, followed by a final extension step of 10 min at 72°C. PCR products were purified using Centri-Spin –20 spin columns (Princeton Separations, Adelphia, NJ; Cat. CS-201). cII mutant DNA was sequenced with an ABI Prism Big Dye Terminator Cycle Sequencing version 2.0 kit, using the same primers as those used for PCR, and an ABI GeneAmp PCR System 2700. Sequencing products were purified using Sephadex G-50 resin. The purified products were sequenced on an ABI 3100 Genetic Analyzer. Sequence data were analyzed with the aid of BLAST 2 sequences program 〈http://www.ncbi.nlm.nih.gov/blast/bl2seq/bl2.html〉. No “hot spots” for mutagenesis were identified in the cII gene.

Alachlor-induced olfactory mucosal tumor formation occurs with a very orderly series of histological changes (Genter et al., 2002b). Up to 1 mo of exposure (126 mg/kg/d in the diet) is not associated with any histological alterations or changes in cell proliferation (the latter was assessed by bromodeoxyuridine incorporation; Wetmore et al., 1999; Genter et al., 2000). It is important to emphasize that at no time point in the pathogenic process was there any evidence of frank cytotoxicity, as is characteristic of other rodent olfactory mucosal carcinogens such as phenacetin or hydrazine (Bogdanffy et al., 1989; Latendresse et al., 1995). Histological changes, involving focal loss of the characteristic organization of the olfactory mucosa and the appearance of ciliated cells resembling of those found in respiratory mucosa, were first noted after three months of exposure (Genter et al., 2002b). The first olfactory neoplasms were noted after 5 mo of exposure (Genter et al., 2002a), and following 6 mo of exposure, approximately half of alachlor treated rats had one or more neoplasms that were characterized as polypoid adenomas (Genter et al., 2000).

Because of this orderly progression of the lesions, we undertook a genomic analysis of the evolution of olfactory tumors with the goal of identifying genes that characterized each stage of the tumorigenic process. To do this, we treated rats with alachlor (126 mg/kg/d in the diet) for time points ranging from 1d to 18 mo and prepared RNA (using TriReagent, MRC, Cincinnati, OH) from the ethmoid turbinates of rats treated for 1 d –5 mo, and from grossly observable tumors from the 18 mo-treated rats. For most treatments, two rats were used for RNA isolation, and we confirmed the histological stage of treatment-matched samples. Gene expression was assessed using the Affymetrix (Santa Clara, CA) U34 GeneChip (allowing evaluation of >8,000 genes simultaneously), with labeling and hybridizations performed by the University of Cincinnati-Cincinnati Children’s Hospital Affymetrix core facility according to the manufacturer’s protocols. Results have been previously published (Genter et al., 2002b), and will be summarized briefly here: 4,777 probe set elements were designated as “Present” by Affymetrix algorithm on ≥1/26 chips; 988 probe set elements were over expressed by ≥1.8X in at least 2 samples; 584 were downregulated ≥0.5X in at least 2 samples. Initial clustering of these 1,392 genes eliminated 125 genes, resulting in 1,265 alachlor-regulated genes. Examination of specific genes regulated at various stages of the carcinogenic process revealed that oxidative stress-related and metalloproteinase genes were among those up-regulated following relatively acute alachlor exposure durations. In addition, genes characteristic of fully-differentiated olfactory mucosa (e.g., CYP2A3) were downregulated in the olfactory mucosa of alachlor-treated rats (Figure 3). The progression of alachlor-induced tumors to more malignant phenotype was associated with up-regulation of genes, such as Axin2, that suggested activation of the Wnt signaling pathway (Jho et al., 2002). Increased cytoplasmic accumulation and nuclear localization of β-catenin (Figure 4) confirmed gene expression changes associated with Wnt activation (Goss and Groden, 2000), although the mechanism of Wnt activation in the alachlor-induced tumorigenesis progress has not yet been elucidated. Importantly, early epithelial alterations and adenomas did not display cytoplasmic accumulation or nuclear localization of β-catenin (Genter et al., 2002b), implicating other factors in the initiation of alachlor-induced olfactory mucosa tumors and Wnt activation as a late event in tumor progression.

Another series of experiments was designed to discover novel, highly-expressed olfactory mucosal genes in the C57BL/6J mouse. Because the olfactory mucosa is the main site of programmed, continuous neurogenesis in the nervous system in adult animals (Graziadei and Graziadei, 1979), we reasoned that genes related to neurogenesis might be among those highly expressed in olfactory mucosa compared to the adult mouse brain and other mouse tissues. This experiment was part of a very extensive collaborative effort among a large number of investigators at the University of Cincinnati and Children’s Hospital, Cincinnati. Briefly, mRNA was isolated from >90 mouse tissues (normal adult, various developmental stages, diseased/manipulated tissues) using the Qiagen RNAeasy kit (Qiagen, Valencia, CA). For this experiment, olfactory mucosa from both the ethmoid turbinates and the caudal one-third of nasal septum served as the tissue source of RNA. All mRNAs were hybridized vs. a universal mRNA pool consisting of total RNA from a postnatal day 1 mouse. Probe preparation and microarray hybridization were performed by Incyte Genomics (Palo Alto, CA) using the GEMBright random primer reverse-transcription labeling kit. Labeled cDNA was prepared from each poly A⁺ RNA sample using nucleotides labeled with the fluorescent dye Cy5, and the universal control mRNA was labeled with Cy3. Genes and expressed sequence tags with expression ≥1.7X over control were subjected to validation and further analysis. This approach yielded 269 genes for further consideration.

The gene list has previously been reported and validation of several genes of interest was performed (Genter et al., 2003). Among the previously unreported genes that we found to be highly upregulated in olfactory mucosa was the gene encoding sphingosine phosphate lyase (Sgpl1). The SGPL1 protein is involved in the sphingolipid signaling pathway, which has not been previously described in olfactory mucosa. In order to confirm that high Sgpl1 expression was associated with demonstrable protein and activity in olfactory mucosa, we immunolocalized to mature olfactory neurons (Genter et al., 2003). We also found enzymatic activity in rat olfactory mucosa was high compared to most other rat tissues (Figure 5). Given that the substrate for SGPL1, sphingosine-1-phosphate (S1P), is regarded by some as a ‘survival factor’ (Edsall et al., 2001), we hypothesize that SGPL1 in mature olfactory neurons degrades S1P and promotes apoptosis of this population of cells. This hypothesis is consistent with recent observations that SGPL1 enhances stress-induced ceramide generation and apoptosis (Reiss et al., 2004). Therefore, we have validated the high expression of Sgpl1 in olfactory mucosa and have begun to investigate the role of sphingolipid signaling in olfactory mucosal homeostasis.

It is becoming increasingly important, given the dramatic increase in the use of gene expression studies by researchers, to develop means of independent verification of preliminary gene expression results. There are, admittedly, many difficulties associated with independent verification of altered gene expression. High quality antibodies are available for only a small fraction of the many genes represented on commercial chips and arrays. To further complicate validation by immunohistochemistry, the methods that are considered routine by many laboratories for nasal cavity histological preparation, namely formalin fixation and formic acid decalcification, can be too harsh and thereby compromise the integrity of many olfactory mucosal antigens. This problem has been circumvented by light (e.g., 2%) paraformaldehyde fixation and decalcification with EDTA (e.g., 0.3M, pH 7.4); the latter procedure requires significantly more time (e.g., 1 wk for mouse nasal cavities, up to a month for rat nasal cavities with frequent changes of the EDTA solution, in our experience) and is more costly than formic acid decalcification. In situ hybridization is another excellent method for confirming gene expression results for presence or absence of expression and even localization in tissues with multiple cell types, but is not very useful, typically, in confirmation of fold of gene expression changes. In situ hybridization procedures also tend to be expensive and tedious, and require dedicated laboratory space, as RNAse free conditions should be maintained, and radiolabeled probes still tend to be the most widely used.

Another hindrance to confirming gene expression results is that unequivocal substrates are not available for many enzymatic activities, making validation of changes in enzyme expression difficult in some circumstances. In the studies summarized above, we used such methods as western blot analysis (to correlate the decrease in rat CYP2A3 expression with a decrease in the protein in olfactory mucosa (Figure 3), in situ hybridization (to localize Sgpl1, Pon1, and Dmbt1) (Figure 5), enzyme activity (SGPL1) (Figure 5), immunohistochemistry (CYP2F2, DMBT1, and SGPL1) (Genter et al., 2002b, 2003), and zymography (MMP2 and MMP9 in olfactory mucosa (Figure 5).

While many investigators use reverse transcription or real time polymerase chain reactions on the same pool of RNA that was used for the expression studies to validate the results of GeneChip or microarray experiments, there are instances in which this approach could perpetuate false results. Therefore, the importance of independent verification cannot be overemphasized. For example, in our mouse olfactory gene expression experiment described here, crp ductin (Dmbt1) was initially among the highly-expressed olfactory mucosal genes. We acquired an antibody that recognized DMBT1 (from Dr. Qais Al-Awqati, Columbia University; described in Takito et al., 1999) and found by immunohistochemistry that DMBT1 was not present in olfactory mucosa, but was detectable in nasal respiratory mucosa (Genter et al., 2002b). (Human DMBT1, rabbit hensin, mouse crp ductin and rat ebnerin are highly conserved proteins in the respective species and have been associated with cell polarity (Takito et al., 1999)). We similarly found in in situ hybridization experiments, that an antisense Dmbt1 probe did not hybridize in olfactory mucosa, but was extremely highly expressed in the lateral nasal gland (Genter et al., 2002b; Figure 5). Because there was significant lack of overlap between other highly expressed nasal respiratory genes and highly expressed olfactory mucosal genes (but many genes also co-highly expressed), we reasoned that we had not significantly contaminated our olfactory RNA pool with nasal respiratory RNA; instead, we concluded that our gene expression studies were inaccurate with respect to DMBT1 in olfactory mucosa due to contamination of the RNA pool with RNA from the lateral nasal gland. We believe that the proximity of the gland to the lateral wall of the nasal cavity, and the speed with which dissections had to be performed in order to rapidly freeze tissues to assure isolation of high quality RNA resulted in breach of the lateral wall of the nasal cavity in one or more of the mice used for RNA isolation and inadvertent collection of lateral nasal gland tissue. This example demonstrates the need for independent verification of gene expression results by means that do not depend on re-analysis of the original RNA pool.

In summary, given that the nasal cavity is a target for toxicity for many inhaled as well as systemically administered toxicants, it is important to develop and expand state-of-the art molecular and cellular methods to assess toxic end points. Novel methods, such as imaging techniques that can identify nasal lesions in an intact animal, will likely emerge and prove to be of tremendous value in the future.