Transcriptomic Analysis of F344 Rat Nasal Epithelium Suggests That the Lack of Carcinogenic Response to Glutaraldehyde is Due to its Greater Toxicity Compared to Formaldehyde

Abstract

Formaldehyde is cytotoxic and carcinogenic to the rat nasal respiratory epithelium inducing tumors after 12 months. Glutaraldehyde is also cytotoxic but is not carcinogenic to nasal epithelium even after 24 months. Both aldehydes induce similar acute and subchronic histopathology that is characterized by inflammation, hyperplasia, and squamous metaplasia. Because early aldehyde-induced lesions are microscopically similar, we investigated whether transcriptional patterns using cDNA technology could explain the different cancer outcomes. Treatments included 1-, 5-, or 28-day exposure by nasal instillation of formaldehyde solution (400 mM) or glutaraldehyde solution (20 mM). Animals were euthanized and the nasal respiratory epithelium removed for gene expression analysis and a subset of rats treated for 28 days was processed for microscopic examination. RNA was isolated and processed for expression assessment using Clontech ^® Atlas Toxicology II Arrays. Both aldehydes induced hyperplasia, squamous metaplasia, and inflammatory infiltrates with scattered apoptotic bodies in the epithelium covering luminal surfaces of the nasoturbinate, maxilloturbinate, and nasal septum. A subset of 80 genes that were the most variant between the treated and control included the functional categories of DNA repair and apoptosis. Hierarchical clustering discriminated chemical treatment effects after 5 days of exposure, with 6 clusters of genes distinguishing formaldehyde from glutaraldehyde. These data suggest that although both aldehydes induced similar short-term cellular phenotypes, gene expression could distinguish glutaraldehyde from formaldehyde. The gene expression patterns suggest that glutaraldehyde’s lack of carcinogenicity may be due to its greater toxicity from lack of DNA-repair, greater mitochondrial damage, and increased apoptosis.

Introduction

Inhaled chemicals are classified as air toxicants when they have the potential to produce adverse health effects in humans after inhalation exposure. These effects include carcinogenic as well as noncarcinogenic responses such as asthma (Weisel, 2002). Formaldehyde has long been known to be a rodent nasal carcinogen (Kerns et al., 1983; Swenberg et al., 1983). The International Agency for Research on Cancer (IARC) has published several reports on formaldehyde (IARC, 1982, 1987, 1995, 2005), ultimately classifying it as carcinogenic to humans.

The specific mechanism by which formaldehyde induces nasal cancer in rodents and upper respiratory tract cancer in humans is largely unknown. However, formaldehyde exposure induces sustained increases in reparative cell proliferation, DNA-protein crosslinks, and has been shown to be mutagenic in nasal respiratory tissues lining the rat nose at specific sites (Monticello et al., 1996; Morgan, 1990; Morgan et al., 1986a, 1986b; Recio et al., 1992). Because formaldehyde alters cell proliferation rates, it is hypothesized that formaldehyde can also affect cell death rates, which would be consistent with the biological model for formaldehyde-induced rat nasal cancer (Conolly et al., 2003). Evidence has recently emerged indicating that formaldehyde can alter apoptosis in vitro (Szende et al., 1998; Tyihak et al., 2001). Whether a cell dies or survives a chemical exposure is often influenced by its proliferative status, DNA repair capacity, metabolic state, and its ability to induce proteins that can either promote or inhibit apoptosis (Robertson and Orrenius, 2000). These parameters are stringently controlled to prevent loss of overall cell homeostatsis.

Glutaraldehyde is also known to be a respiratory toxicant with most human exposure occurring through occupational settings (ACGIH, 1997). Glutaraldehyde has been investigated extensively in both short- and long-term studies by inhalation or instillation in rodents (Gross et al., 1994; St. Clair, Gross, and Morgan, 1990; van Birgelen et al., 2000). These studies showed that glutaraldehyde caused marked toxicity to nasal epithelium, but was not carcinogenic in this site.

The use of rodent bioassays for cancer detection and human extrapolation is time consuming and expensive. Investigators are evaluating the potential for genomic assays to be useful in the detection of carcinogens with increasing frequency (Afshari, Nuwaysir, and Barrett, 1999; Hamadeh et al., 2002a, 2002b, 2002c). cDNA microarray technology is used to examine profiles of gene expression. This advance is revolutionizing the field of chemical toxicology and is captured by a new term, “toxicogenomics,” that defines the breadth of this new discipline (Farr and Dunn, 1999; Zimmermann et al., 2000). Gene expression changes induced by classical chemical compounds with a known mode of action such as enzyme inducers (phenobarbital) or peroxisome proliferators (clofibrate) can be compared to chemicals with an unknown mechanism such that, if specific genes have altered expression, a mechanism might be inferred (Huang et al., 2001; Bushel et al., 2002). The power of this toxicogenomic approach is amplified when combined with more conventional analytic techniques such as descriptive histopathology. Using both approaches, the genetic basis for a cellular phenotype following a chemical exposure will provide insights into hypothesized toxicity pathways.

In the current study, a low-density, 474-gene rat toxicology array was used to determine the gene profiles induced by formaldehyde and glutaraldehyde instilled into the rat nose. A companion set of animals was examined for histological changes associated with treatment. This study was performed to test the hypothesis that formaldehyde dysregulates apoptosis differently than glutaraldehyde, and that this differential dysregulation may, in part, be responsible for formaldehyde’s ability to induce cancer.

Methods

Animals and Dosing

Sixty-day-old male F344 rats (Charles River Laboratory, Raleigh, NC) (36) were housed 2 per cage and given water and food ad libitum. Experiments were performed according to guidelines established by the EPA Animal Care Committee for the care and use of laboratory animals. Chemical doses were selected based on equivalent induction of reparative cell proliferation as an index of cytotoxicity (St. Clair, Gross, and Morgan, 1990) after instillation. Rats (4 per treatment group) received 40 μl of 400 mM formaldehyde (Sigma-Aldrich, St. Louis, MO) solution, 20 mM glutaraldehyde (Sigma-Aldrich, St. Louis, MO) solution, or distilled water by nasal instillation into each nostril for 1 day, 5 days, or 4 weeks, 5 days per week (light/dark cycle-6 am/6 pm with dose time at 8 am). Day 1 animals were euthanized 24 h following treatment. Five-day animals were euthanized on the 5th day following 4 daily treatments. The final group was euthanized on the 28th day following treatment, or a total of 19 treatments, 5D/wk over 4 weeks with euthanasia 24 h after the 19th treatment.

Experiment Design

This experiment consisted of a multifactorial, balanced design with 3 treatments (formaldehyde, glutaraldehyde, or control) and 3 time points (1 day, 5 days, and 28 days) with 4 animals in each group. The Clontech Rat Toxicology II cDNA Array containing 474 genes was utilized for hybridizations.

Epithelial Cell Extraction From Anterior Rat Noses

F344 male rats were euthanized by CO₂ asphyxiation, and nasal respiratory cells were recovered and processed as described previously (Hester et al., 2002). Briefly, Trizol (Life Technologies, Frederick, Maryland) reagent was instilled into each nostril through polyethylene tubing and incubated for 10 minutes. Using a syringe attached to the tubing, the cellular Trizol ^® solution was removed, placed in a microfuge tube, and immediately frozen in liquid nitrogen for subsequent total RNA isolation and quantification.

Total RNA Extraction

Total RNA was extracted from respiratory cells as previously reported (Hester et al., 2002). Briefly, phase separation was carried out after thawing the cellular/Trizol samples in a warm water bath (65–70°C) for 10 minutes. Chloroform was added, followed by centrifugation steps, to isolate total RNA. The RNA solution was then transferred to siliconized Eppendorf tubes and the concentration (absorbance at 260 nm; 1A₂₆₀ unit of single stranded RNA = 40 μg/ml) was adjusted to 2 μg/μl with diethylene pyrocarbonate (DEPC) water.

Electrophoresis of the RNA

Five μg of the resultant RNA sample was loaded onto a denaturing 1% agarose gel containing formaldehyde according to the protocol described by Lehrach et al. (1977) for electrophoresis. Observation of rRNA subunit bands at 18S and 28S indicated the presence of intact RNA.

Synthesis and Labeling of cDNA

Six μg of total RNA was used to generate a probe for hybridizations as described by (Hester et al., 2002). Briefly, 36 samples (12 control/12 formaldehyde/12 glutaraldehyde) from 3 time-points (1 day, 5 days, or 28 days), cDNA probes were generated by reverse transcribing the RNA with primers in the presence of ³³P. Unincorporated label was removed on a Microspin Sepharose G-50 gel filtration column (Amersham Pharmacia Biotech, Piscataway, NJ). For hybridization of the labeled product, 250,000 Cerenkov cpm (counted with no medium) was used. The probes were hybridized to Clontech Rat Atlas Toxicology II filters overnight at 42°C in MicroHyb buffer (Research Genetics, Carlsbad, California). Arrays were applied to a phosphoscreen for 48 hours, scanned on a Cyclone Phosphoimager (Perkin-Elmer, Boston, Massachusetts) and Clontech software was used to analyze gene expression levels.

Materials for Microscopic Examination

After euthanasia, heads were removed from a separate set of rats (4 control/4 formaldehyde/4 glutaraldehyde) that had been treated for 28 days. The nasal cavity was flushed immediately with neutral buffered 10% formalin (NBF). The mandible was removed and the skin covering the cranium, as well as the eyeballs, were also removed. A labeled cassette was tied to the zygomatic bone. The cranium was immersed in NBF. The nose was then processed by routine methods for paraffin section that included a decalcification procedure. The nose was cut according to the technique described in Mery et al. (1994) with rat sections 2, 4, and 6 provided for microscopic examination after staining with hematoxylin and eosin.

Statistical Analysis

In order to assess data quality on the raw intensities, scatter plots were generated among the replicates of each group (data not shown). The scatter plots revealed 3 arrays with low global intensities and high proportions of undetectable gene expression, which were removed from further analysis leaving 33 arrays. The adjusted signal intensities (ASI) obtained from the Clontech ^® image analysis software were log (base 2) transformed, with a small correction factor, to stabilize the variance across intensity levels (Hester et al., 2002). Scatter plots of replicate arrays were generated as a measure of reproducibility and quality control. Global normalization, to adjust systematic variation across arrays, was achieved through a mixed model proposed by Wolfinger et al. (2001). Once the data were normalized, differential expression across the two factors, chemical treatment and exposure time was analyzed gene-by-gene in a classical 2-way ANOVA model. Further, the data from each chemical were compared individually to the control rats in order to identify sets of differentially expressed genes. Clustering analysis was performed on the normalized intensities from chemical treated groups. Using the software Cluster Genes (Eisen et al., 1998), gene expression data were filtered to a nominal subset of 80 genes that best fit the classical 2-way ANOVA model (using terms for time, treatment, and the interaction of time and treatment) and were then normalized across arrays. Complete linkage hierarchical clustering of correlations generated dendrograms across genes and across arrays, while a heat map displayed the normalized intensities; cluster images were created in TreeView (Eisen et al., 1998). Hierarchical clustering of genes was also performed on sets of genes identified as differentially expressed across chemical treatments.

Results

Histopathology

Controls

One control rat had a focal area of squamous hyperplasia with inflammation along the ventral airway just medial to the nasolachrymal duct in Section 2 and another along the ventral airway also medial to the nasolachrymal duct in Section 6. These were morphologically consistent with the cage contaminant lesion that has been reported in shoebox-cage-housed rats (Bolon et al., 1991), and did not interfere with characterization of treatment-induced changes.

Formaldehyde

Section 2—The maxilloturbinate had mild hyperplasia of the transitional epithelium with rare scattered apoptotic bodies.

Section 4—The nasoturbinate had squamous metaplasia with mild to focally moderate inflammatory cell infiltrate into the epithelial layer and rare apoptotic bodies (Figure 1A). The maxilloturbinate had squamous metaplasia with mild infiltration of inflammatory cells into the epithelial layer and scattered apoptotic bodies (Figure 1B). The epithelium lining the nasal septum had squamous metaplasia with occasional hyperplasia, inflammatory infiltrate, and scattered apoptosis (Figure 1C).

Section 6—The maxilloturbinate had squamous metaplasia with occasional hyperplasia, inflammatory cell infiltrate into the epithelium and scattered apoptotic bodies. The nasoturbinate had squamous metaplasia, some areas of slight intraepithelial inflammatory infiltrate and scattered apoptotic bodies. The olfactory epithelium along the dorsal airway between the septum and the nasoturbinate was atrophic with scattered apoptotic bodies. The nasal septum had areas of squamous metaplasia with focal areas of hyperplasia, more frequent and severe, just dorsal to the vomeronasal intumescence. There was mild inflammatory infiltrate into the metaplastic epithelium and scattered to focally frequent apoptotic bodies.

Glutaraldehyde

Section 2—Only mild lesions occurred when present and consisted of focal squamous hyperplasia on the tip of the maxilloturbinate, or nasoturbinate, or along the septum. Apoptotic bodies were present within the septal lesion.

Section 4—The maxilloturbinate had squamous metaplasia with focal to multifocal hyperplasia and inflammatory infiltrate. The middle lateral wall had squamous metaplasia with focal hyperplasia. The nasotubinate had squamous metaplasia with hyperplasia, inflammatory infiltrate and scattered apoptotic bodies (Figure 1A). The septum had squamous metaplasia that tended to be more severe along the superior ventral portion and had focal hyperplasia. In one rat there were also intraepithelial microabcesses and apoptotic bodies.

Section 6—The maxillotubinate had squamous metaplasia with areas of hyperplasia that in some cases projected into the lumen as hyperplastic plaques. There was also an inflammatory infiltrate and scattered apoptotic bodies. The middle lateral wall had squamous metaplasia and scattered apoptosis. The nasoturbinate had squamous metaplasia with inflammatory infiltrate; intraepithelial microabcesses were sometimes present as was focal erosion. Apoptotic bodies were frequently present. The olfactory epithelium between the nasoturbinate and septum was atrophic with apoptotic bodies. The nasal septum had squamous metaplasia with focal hyperplasia, inflammatory infiltrate with occasional microabcesses, and apoptotic bodies.

General observations

The types of lesions present were not qualitatively different between rats treated with formaldehyde or glutaraldehyde. All treated rats had lesions, and they occurred in both airways. In general the distribution tended to be similar, with glutaraldehyde treated rats typically having a more diffuse pattern. In all treated rats, 1 nasal airway had more severe lesions than the contra-lateral airway, and in no case were there bilaterally symmetrical lesions. The lesions became more severe as one progressed caudally, and they were considered to be generally more severe in glutaralde-hyde treated rats as they had progressed to a more chronic phenotype than the formaldehyde-treated rats.

Data normalization and quality assessment

The adjusted signal intensities (ASI) were obtained from the Clontech image analysis software (Atlas Image, Clontech) for 474 genes on the 36 arrays. In order to assess data quality on the raw intensities, scatter plots were generated among the replicates of each group (data not shown). The scatter plots revealed 3 arrays with low global intensities and high proportions of undetectable gene expression, which were removed from further analysis. A box plot of the log (base 2) transformed intensities for control rats (data not shown) demonstrates the need for normalization as the median array intensities varied from 3 to 8. After normalization, as described in the Statistical Analysis section, the mean chip expression was set at 0, with the intensities ranging from −10 to 10.

Differential Gene Expression Analysis

Comparison of aldehydes to control over all time points

Our lab previously reported basal gene expression in the rat nose associated with instillation (Hester et al., 2002). We then conducted instillation experiments in the rat nose with aqueous formaldehyde solution to assess gene expression alterations in formaldehyde-exposed compared to control animals (Hester et al., 2003).

Statistical filtering (p < 0.05) revealed that 40 genes were altered after formaldehyde treatment compared to controls, 30 genes were altered after glutaraldehyde treatment compared to controls, and 3 genes had altered expression after either formaldehyde or glutaraldehyde treatment. Eight of 18 statistically altered genes associated with formaldehyde versus glutaraldehyde (Table 1) were present in the formaldehyde versus control or glutaraldehyde versus control comparisons. These genes were IGF-1, GST1-theta, TCP1, endopepetidase, fibrinogen gamma chain, MAP-kinase5, PON1, and NEDD8.

Clusters

Formaldehyde versus glutaraldehyde time course

The gene expression filtering strategy utilized (Figure 2) identified a subset of 80 genes that best fit the 2-way ANOVA model. The set of 80 significantly altered genes was analyzed to identify to which functional categories they belong by using GenMAPP software. A single complete pathway was not identified but rather several pathways, each containing 1 to 3 genes per pathway (data not shown) were identified. These pathways included inflammation, cell proliferation, and apoptosis. That finding is consistent with previous literature describing the pathogenesis of acute exposure of the rat nose to aldehydes, that includes increased cytotoxicity and reparative cell proliferation along with inflammatory cell in-filtrates as the primary tissue responses (St. Clair, Gross, and Morgan, 1990; Gross et al., 1994; Monticello et al., 1996).

Figure 3 depicts a hierarchical tree showing that arrays segregate by length of exposure. After 28-days of exposure to either compound, a subset of genes had similarly increased expression (Figure 3, bracket A) when compared to either 1 or 5 days of exposure or decreased expression (Figure 3, bracket B). Another pattern of expression is present (Figure 3, bracket C) that represents a transcriptional transition. This pattern shows that after 5 days of glutaraldehyde exposure the expression profile is similar to the 28-day pattern. In contrast, after 5 days of formaldehyde treatment, the pattern is more similar to the 1-day pattern. This temporal separation in gene expression is evidence that transcriptional pro-files can differentiate between different chemical exposures that induce similar tissue responses. The chemical treatments were futher differentiated using 1-dimensional unsupervised clustering (Figure 4). To more closely examine the genes that distinguish between formaldehyde and glutaraldehyde exposure, we generated a 1-dimensional unsupervised clustering of all genes that showed significant differential expression between formaldehyde and glutaraldehyde (p < 0.05). Six gene clusters were identified as showing the most differentiation in the 5-day samples (Table 1). Three of the clusters contained a total of 10 genes that had increased expression after 5 days of glutaraldehyde treatment and 3 clusters of a total of 8 genes had increased expression after 5 days of formaldehyde treatment.

Apoptosis Genes

Cluster analysis of DNA repair and apoptosis genes

Additional cluster analysis was performed on the expression of the genes that control DNA repair and apoptosis. The heat map of the 28 DNA-repair genes on this array (Figure 5) shows that after 28 days a subset of genes are induced after either treatment. However, a different subset of DNA repair genes are induced, by both aldehydes after 1 or 5 days of treatment (Figure 5). There were 8 genes differentially expressed by the 2 treatments that represent different pathways for DNA repair and include recombination, base excision repair, and nucleotide excision repair. Within this group there were 2 genes with a 2-fold induction in foramldehyde compared to glutaraldehyde (replication protein A70 and DNA excision repair ERCC1) and this differential expression after 5 days of treatment could specifically distinguish glutaraldehyde from formaldehyde (Figure 5). These same 2 genes have increased expression after 1 day of treatment by either aldehyde and only differentiate after 5 days. Both these genes function to recognize and remove damaged DNA bases. This finding suggests that both adehydes or their metabolites may directly damage DNA bases but that formaldehyde-exposed cells may function more efficiently to remove damaged bases than glutaraldehyde exposed cells.

After 1 day of treatment both formaldehyde and glutaraldehyde induced the same 3 proapoptotic genes (Figure 6). However, a different pattern is present after 5 and 28 days of glutaraldehyde treatment. Five proapoptotic and 2 anti-apoptotic genes were induced by glutaraldehyde. There was a greater induction of proapoptotic genes after treatment with glutaraldehyde compared to formaldehyde suggesting enhanced cell death after glutaraldehyde treatment as compared to formaldehyde treatment.

The results from the current study were compared to our previous work (Hester et al., 2003). Both studies had consistent changes of expression of the genes the 2 array platforms had in common. A representative sample of 12 genes from several functional categories in the present work were selected for specific comparison and showed that the direction of gene expression was in agreement for 10 out of the 12 genes after 1-day of treatment with formaldehyde.

Cluster Analysis of Other Gene Categories

Examination of gene expression patterns identified an additional cluster of statistically significant genes (Figure 3). The subset of genes involved in cell cycle regulation, inflammation, cell growth, energy production, and cell structure, were differentially expressed after 28 days of treatment with formaldehyde or glutaraldehyde. These transcriptional alterations are consistent with the histopathology described previously, including the presence of inflammatory cells, hyperplasia, and squamous metaplasia such that the transcriptome correlated with the phenotype.

Discussion

We reported previously that 24 hours following a single nasal instillation of formaldehyde, changes in expression of genes involved in xenobiotic metabolism, cell cycle control, apoptosis, and DNA repair were found (Hester et al., 2003). The previous experiments utilized a filter array containing 1185 cDNA fragments (representing individual genes), had a limited number of animals, and only 1 time point (24 hours). The present studies were conducted to extend the previous work, compare two aldehydes, and correlate transcriptional alterations with the resulting histopathology. In the present study we utilized a filter array containing only 474 genes, but enriched in the functional categories that were previously identified as having significantly altered expression after formaldehyde treatment.

Comparing gene expression profiles from the carcinogenic aldehyde, formaldehyde, and the noncarcinogenic aldehyde, glutaraldehyde, helped explain the different cancer response associated with the toxicity of these two aldehydes. Formaldehyde induced an alteration of apoptosis gene expression in nasal epithelium that increased with length of treatment. Altered expression peaked after 5 days of treatment and apoptotic gene expression signaling was greater in nasal epithelium from formaldehyde-treated rats than control (data not shown), but less than glutaraldehyde-exposed rats. Formaldehyde-induced increased expression of DNA repair genes to a greater extent than glutaraldehyde, suggesting either greater ongoing formaldehyde-induced DNA damage or a greater ability to repair DNA damage.

Both formaldehyde and glutaraldehyde are toxic to the nasal epithelium and induce similar histopathology that was consistent with previous reports. However, glutaraldehyde-induced lesions at doses employed in the present study had progressed to a more chronic phenotype than those induced by formaldehyde and so would be considered more severe. The lesions that develop after persistent aldehyde exposure are consistent with an adaptive or protective response to persistent exposure to an irritant chemical. In response to a persistent cytotoxic injury, the nasal epithelium undergoes a series of changes, including epithelial hyperplasia, metaplasia, and associated inflammatory cell infiltration. In addition to the above observed changes and consistent with the hyperplasia is the common feature of increased cell proliferation. The transcriptional profiles after 5 and 28 days associated with treatment had alterations in gene expression that correlated with the observed phenotypic changes. The changing pattern of expression represented in the Heat Map after 5 days of glutarladehyde treatment were more like the 28-day exposure pattern whereas the 5-day formaldehyde expression was more like the 1-day pattern.

Rats exposed to 6 ppm of formaldehyde or greater by inhalation can have a location dependent increased nasal cell proliferation as high as 25-fold over control that can be sustained after many weeks of exposure (Monticello et al., 1991). Similarly, nasal lesions associated with instilled aldehyde solutions resembled both qualitatively and quantitatively, changes induced by inhalation in rats exposed to carcinogenic levels of formaldehyde gas (St. Clair et al., 1990). In the present study, transcriptional changes consistent with increased cell proliferation (data not shown) included significantly (p < 0.05) increased expression of PCNA, cell division protein 25B, DNA topoisomerase IIB, and cyclin-dependent kinases 2 and 7. The expression pattern of these genes showed an increase after 1 day of treatment that peaked after 5 days and declined after 28 days. This transcriptional pattern is similar temporally to the pattern of nasal cell proliferation that was described in rats exposed to formaldehyde by inhalation, where proliferation increased after 1-day of exposure, peaked after 4 days and then was decreased after 6 weeks of exposure although still greater than control (Monticello et al., 1991).

Formaldehyde-induced tissue response has been mapped to specific sites within the nasal cavity and correlated with pre-neoplastic and neoplastic changes (Monticello et al., 1996). These responses, as measured by increases in cell proliferation, were located in the walls of the lateral meatus, the nasal septum, and the medial region of the maxilloturbinate. Similarly, previous studies have correlated formaldehyde-induced DNA-protein crosslinks with cell replication at specific sites within the rat nose after exposure to formaldehyde treatment for 12 weeks (Casanova et al., 1994). The present studies showed that squamous metaplasia with moderate infiltration of inflammatory cells and scattered apoptotic bodies were histologic features associated with both formaldehyde and glutaraldehyde treatment and that they were present in the mid-septum and lateral meatus after 28 days of exposure, which is consistent with the locations described by Monticello et al. (1996). Altered transcriptional profiles consistent with the histologic features described after both aldehyde exposures included increased expression of retinoic acid binding protein II, c-myc-responsive protein, inhibitor of DNA binding 1, and involucrin. In addition, consistent with the inflammatory cell infiltration, genes associated with inflammation, such as eosinophil granule major basic protein and macrophage inflammatory protein 2, were significantly upregulated. Consistent with the histologic presence of apoptotic bodies, TP53 and caspase 2 and 7, known to be involved in apoptosis, had a significantly increased expression after formaldehyde and glutaraldehyde exposure.

Although histologically formaldehyde and glutaraldehyde induced similar lesions after 28-days of exposure, the transcriptional patterns after formaldehyde and glutaraldehyde exposure were different suggesting different subcellular targets (Figure 7). For example formaldehyde-induced apoptosis was associated with induction of Fas-ligand and TNF receptor (membrane bound) with subsequent induction of downstream caspases. In contrast, glutaraldehyde strongly induced genes associated with the mitochondrial bcl2- family, BAD and Bax. In addition, glutaraldehyde, but not formaldehyde, caused a 12-fold induction of an organic ion transporter, OCTN2, known to transport carnitine to the mitochondria for energy production.

These data suggest that glutaraldehyde altered apoptosis by a different pathway than formaldehyde and that glutaraldehyde also selectively targets the mitochondria as part of its toxicity pathway. This suggests a rationale why glutaraldehyde is more toxic at a comparable concentration that produces the same level of cell proliferation as formaldehdye. Glutaraldehyde treatment resulted in greater expression of the proapoptotic gene BAD and greater induction of genes related to mitochondrial enzymes that control cell respiration. Mitochondria have a complex role in the cell which includes ATP generation, sequestering calcium, and generating as well as detoxifying reactive oxygen species (Alberts, 1994). Glutaraldehyde is oxidized quite rapidly by rat liver mitochondria, and oxidation is energy-linked (Packer and Greville, 1969).

Formaldehyde is also oxidized, but considerably less so than glutaraldehyde, by isolated rat mitochondria (Teng et al., 2001). The inhibition of mitochondrial respiration could have a significant effect. By reducing energy needed for DNA repair (Oei and Ziegler, 2000) and apoptosis (Eguchi, Shimizu, and Tsujimoto, 1997), ATP may be diverted away, resulting in further cellular damage by impairing ATP-dependent ion pumps that keep intracellular calcium at low cytosolic concentrations (Carafoli, 1987). These pumps are important to maintain internal potassium concentrations and can have a role in apoptosis suppression (Bortner, Hughes, and Cidlowski, 1997; Hughes et al., 1997). These data suggest that the loss of cell energy with resulting cell death could underlie the differential effect of these aldehydes, with glutaraldehyde’s transcriptional response indicating greater cellular distress.

Formaldehdye also induced the expression of DNA repair genes to a greater degree than glutaraldehyde treatment. These findings suggest that although formaldehyde may persistently damage DNA as compared to glutaraldehyde, increased repair capability can also result in increased misrepair.

Although acute treatment by the aldehydes generated gene profiles associated with cellular proliferation, cellular stress, and xenobiotic metabolism, longer exposures induced a different subset of genes suggesting a change occurs that is associated with persistent toxicity. This pattern included genes associated with activation of DNA repair and induction of growth arrest. Less induction of DNA repair genes by glutaraldehyde may be part of the reason for its greater toxicity and thus the lack of cancer development. Whereas after formaldehyde exposure, incompletely repaired cells that experience a growth arrest could persist and accumulate further insults. These cells could pass on genetic damage whereas a lack of repair, as is suggested for glutaraldehyde, would stimulate cell death.

In summary, this report provides the first genomic comparison of a carcinogenic to a noncarcinogenic aldehyde in the rat nose. These treatments could be discriminated through generating unique gene expression profiles. Hierarchical clustering revealed different patterns of expression in DNA repair and apoptosis. Although glutaraldehyde produces a similar pattern of responses in genetic toxicity tests, presence of DNA-protein cross-links, and histopathology, the pattern of gene expression suggests that glutaraldehyde’s lack of carcinogenicity may be due to its greater toxicity, through lack of DNA-repair, mitochondrial damage and increased apoptosis. The authors recognize that while these results may suggest a potential mechanism, confirmation of these data awaits additional studies by others using the same or different platform.