Blood from rats exposed to each of the five nitrotoluenes, TNT, 2ADNT, 4ADNT, 2,6DNT, and 2,4DNT, for either 24 or 48 h was examined for the presence of parent compound and/or possible nitrotoluene metabolites using HPLC. At 24 h, measurable concentrations of nitrotoluenes were found only in blood of the two highest TNT dosed and 3 highest dosed 2ADNT dosed animals (File S1
). 2ADNT was the only metabolite observed in blood of the two highest dosed TNT treated animals 24 h after exposure. No measurable amounts of any of the five nitrotoluene compounds or their degradation products were found in blood 48 h after the exposure.
Liver tissue of exposed and control rats were examined for the presence of parent chemicals and their metabolites 24 and 48 h after exposure. In livers of exposed rats, only the parent compounds TNT, 2ADNT, and 2,4DNT were detected at 24 h, but only in one replicate (File S1
). Metabolites of TNT were observed in livers of 192 and 384 mg/kg TNT exposed rats at 24 h after exposure where 2,6-diamino-4-nitrotoluene and 2,4-diamino-6-nitrotoluene were detected.
At lower doses of 96 and 48 mg/kg TNT-treatments, 2ADNT was found in liver of one animal for each dose, while no 4ADNT was observed. No measurable concentrations of the other four nitrotoluene parent compounds were found in livers of animals after 48 h (File S1
). Further metabolic products such as azoxytoluenes and benzylalcohols were not measured under the protocol utilized.
Although the upper limit of chemical dose ranges were designed to be one-half the published acute oral LD50
, lethality was observed within the dose range used for the dinitrotoluenes. The TNT dose was based on the publication of Reddy at al., 
. The doses of 2,4DNT and 2,6DNT were based on the publication of Ellis et al.,
. The dose selection for 2ADNT and 4ADNT were based on the report of Ellis et al., 
All treated rats expired at the high dose (398 mg/kg) of 2,6DNT. The high dose of 2,4DNT caused deaths of 2 of 5 and 1 of 5 rats at 24 and 48 h, respectively. Systemic toxicity was also indicated for all compounds by decreases in body weight gain over the 48 h observation period. Body weight gain for vehicle-treated rats was 17±1 g (mean±SEM, n
5) compared to a loss of 6−7 g for 199 mg/kg 2,6DNT and high dose (384 mg/kg) TNT. Weight gains were reduced to between 1 to 4 g for rats treated with 50 and 99 mg/kg 2,6DNT, 398 mg/kg 2,4DNT and 192 mg/kg TNT and to 5−9 g for 48 and 96 mg/kg TNT, 174 and 348 mg/kg 2ADNT, and 374 mg/kg 4ADNT. Body weight loss associated with higher doses of TNT and dinitrotoluenes was possibly related to reduced food consumption as indicated by a coincident decrease in serum alkaline phosphatase (AlkP, ,) 
Clinical Chemistry Outcomes Affected by 2,4- and 2,6-DNT and TNTa.
Clinical Chemistry Outcomes Affected by 2ADNT and 4ADNTa.
Additional differences detected from the comprehensive serum metabolic panel included effects on serum albumin, total protein, glucose and Na (,). Serum albumin was the parameter changed at the lowest level of exposure also known as lowest observed effect level (LOEL). Serum albumin decreased at 48 h after exposure with a LOEL of 99 mg/kg 2,4DNT and LOEL of TNT at 192 mg/kg. Changes in total protein paralleled those of albumin (data not shown). Glucose was elevated at 48 h with 2,4DNT (high dose) and reduced at 48 h with 4ADNT (LOEL 47 mg/kg). Serum sodium levels were decreased at 24 h after 2,6DNT (199 mg/kg) and 2ADNT (348 mg/kg) exposures and at 48 h after 2,4DNT (LOEL 198 mg/kg) and TNT (384 mg/kg) exposures. No significant treatment-related changes were observed for total bilirubin, creatinine or urea (data not shown).
Serum alanine aminotransferase (ALT) activity exhibited a dose-dependent decrease to 33% of vehicle control at 24 h after 4ADNT exposure with doses greater than 5 mg/kg. This effect was not sustained to 48 h post-exposure, but rather an increase was observed for the 94 mg/kg group. A similar trend of lesser magnitude occurred with serum aspartate aminotransferase (AST). Small declines in serum transaminase activities are occasionally observed with some chemicals often due to inhibition of enzymatic utilization of cofactor pyridoxyl-5′-phosphate; however, this effect is thought to be of relatively minor toxicological significance 
One rat in each of the 50 and 99 mg/kg 2,6DNT groups (48 h) had serum ALT values in excess of 3000 IU/ml that thus caused their treatment means to exceed 7 times the control mean with ~200% coefficient of variation (). Histopathology verified that these values truly reflected hepatotoxicity of sensitive rats within their treatment groups (). Livers of these rats contained congested sinusoids, some with sloughed hepatocyte remnants (arrow, ). Segmented neutrophils were frequent, but confined to the sinusoids. Central veins and lining endothelia appeared normal with unaffected proximal hepatocytes in radiating cords. However, several cords become disorganized at midzonal regions where erythrocytes appeared to infiltrate and adjacent hepatocytes exhibited pyknotic nuclei (arrowhead, ) and microvesiculated cytoplasm (asterisks, ). In addition, apoptotic hepatocytes were evident (asterisk, ). Others have shown necrosis and hemorrhage in livers of rats with 50% lethality at 24 h after 50 mg/kg 2,6DNT, i.p., while equivalent treatment with 2,4DNT was without effect 
Pathology of livers from rats treated 48 h earlier with 50 mg/kg 2,6DNT (A,B) or 384 mg/kg TNT (C).
In contrast, hepatocytes of rats treated with the higher dose of 2,6DNT whose serum ALT levels were equivalent to vehicle controls appeared mostly undamaged (), however, sinusoids were markedly congested. Occasional erythrocyte infiltration between adjacent hepatocytes within a cord was noted (arrow, ), but without associated necrosis as seen in the sensitive animals (arrowhead, ). Enlarged nucleoli were evident in several hepatocytes of treated rat liver in agreement with ultrastructural changes after single dose dinitrotoluene reported previously 
Representative photomicrographs of livers from rats treated 48 h earlier with vehicle (5% DMSO in corn oil, A and C) or 199 mg/kg 2,6DNT (B and D).
Sinusoidal congestion was also evident with high dose 2,4DNT. Cytoplasmic amorphous inclusions were noted in hepatocytes of 3 livers and microvesicles in another of the 5 rats treated with 384 mg/kg TNT (). Livers of 4 of 5 rats treated with high dose TNT had regions where sinusoids were congested, but less area was involved than seen with dinitrotoluene-treated rats. Focal hemorrhages were seen in the liver of one rat at 48 h after treatment with 94 mg/kg 4ADNT (), while ballooning degeneration was common in livers from other rats 48 h after 94 and 187 mg/kg 4ADNT (). No pathological lesions were seen in livers of rats treated with 2ADNT.
Relative liver weights of rats treated with 4ADNT increased from 4.12±0.11% to 4.82±0.05%, 4.77±0.05% and 4.76±0.18% (means±SEM, n
5) at 48 h after treatment with 93, 187 and 374 mg/kg, respectively. TNT at 384 mg/kg was associated with an increased liver weight of 4.90±0.23% relative to 4.03±0.17% for controls. The 99 mg/kg 2,4DNT group also had an elevated relative liver weight of 4.59±0.09% compared to concurrent vehicle control (3.85±0.08%), but higher doses (199 mg/kg
4.06±0.25; 389 mg/kg
4.29±0.18) were unaffected. Liver weights of rats treated with 2,6DNT and 2ADNT were unchanged from vehicle controls at 48 h post-treatment.
The most notable finding from the hematology panel was an elevation of several related parameters indicating erythrocytosis associated with the dinitrotoluenes (, ). Erythrocyte count was elevated at 99 mg/kg at 24 h post-exposure and at 199 mg/kg and above at 48 h for both compounds. Increases in hemoglobin concentration and hematocrit paralleled that of erythrocyte count. For all parameters, the magnitude of change was greater for 2,6DNT than 2,4DNT. At 48 h after 199 mg/kg 2,6DNT, hemoglobin and hematocrit reached high values of 20 g/dL and 61% respectively. 2,6DNT, but not 2,4DNT, produced an increase in blood reticulocytes and was associated with mature erythrocytes with Heinz bodies and “bite cells” (). TNT-treated rats exhibited increases in erythrocyte count and hemoglobin 24 h post-exposure at 192 and 384 mg/kg, respectively, but these effects were not sustained to 48 h nor of the magnitude seen with 2,6DNT. In contrast, blood erythrocyte parameters showed downward trends with 4ADNT as evidenced by ~10% declines in hemoglobin and hematocrit at 48 h after treatment with 94 mg/kg and above (, ). Treatment with 2ADNT was without effect.
Hematology Outcomes Affected by 2,4- and 2,6-DNT and TNTa.
Effect of 2ADNT and 4ADNT on Hematology Outcomesa.
Morphology of erythrocytes from rats treated 24 h earlier with 2,6DNT.
Marked elevation of blood granulocytes were noted at 48 h after high dose 2,4DNT and TNT and 199 mg/kg 2,6DNT (3). Earlier elevations occurred with 2,4DNT at doses of 99 mg/kg and above and with 384 mg/kg TNT. Granulocytosis at 24 h for TNT and at 48 h for 2,4DNT tended to parallel a decrease in lymphocytes as often occurs with corticosteroid-mediated stress 
. In contrast, leukocyte counts of rats treated with the amino-dinitrotoluenes were unaffected by treatment at either time point ().
Dose responsive differentially expressed gene numbers in liver affected by nitrotoluenes
A strong dose responsive change in the total number of differentially expressed genes was observed at both time points for all compounds (). The dose responsive change suggests that the doses are in an appropriate linear range for the microarray experiment. The number of statistically regulated transcripts was consistently larger at 24 h than 48 h for all compounds except TNT. TNT was reversed in comparison to the other 4 compounds in that it produced a higher number of altered transcripts at 48 h. In light of the chemistry data, metabolism of TNT into 2,6DNT and 2,4DNT in liver may cause a lag in a decrease of observed impacts upon gene expression.
The dose and time dependent numbers of transcripts regulated by 2,4DNT, 2,6DNT, 2ADNT, 4ADNT or TNT.
While the dose responsive pattern of the number of gene expression was similar for the compounds examined (), there was a large difference in the total number affected by the chemicals (). We observed a remarkable difference in the number of altered genes for both time points together (24 h+48 h) induced by the 5 compounds (). 2,4 DNT displayed the highest changed number (close to 6000) at 24 h+48 h, followed by 2,6 DNT (nearly 5000), 4ADNT (over 1000), TNT (nearly 700) and 2ADNT (110) (Tables S1
). The identification of very few differentially expressed transcripts by 2ADNT relative to the other nitrotoluenes is consistent with observed clinical effects suggesting that reduction of the nitroso group to an amino group at ring position C-2 dramatically reduces the biological effects of nitrotoluenes.
Total transcript numbers regulated by the five compounds.
Next, we determined the amount of up-regulated or down-regulated transcript numbers for the compounds. To do that, we compared treated and control samples at 24 h or 48 h or both time points together for each compound. Based on significantly regulated transcripts, if the averaged normalized intensity of a gene was higher in the treated group than the control group, we considered the gene relatively up-regulated; and if lower, down-regulated. According to this criterion, we found another common phenomenon: there were much more up-regulated transcripts than down-regulated transcripts at 24h and both time points together (24 h+48 h) () for 2,4DNT, 2,6DNT, 2ADNT, 4ADNT and TNT. Interestingly, at 48 h, there were far more down-regulated transcripts than up-regulated transcripts for the 4 compounds 2,4DNT, 2,6DNT, 2ADNT, 4ADNT. TNT showed a reverse trend once again in that it induced more up-regulated transcripts at 48 h than down-regulated ones. Our results indicate that not only was a trend conserved for the dose responsive effects, but also a time –responsive up and down-regulated trend among these compounds was conserved, implicating that a common underlying mechanism may be shared by these compounds.
Comparison of general gene expression profiles induced by the 5 compounds
The overall gene expression profiles induced by the 5 chemicals with various doses and different times were compared by two-way hierarchical clustering of averages of replicates within each of the 40 conditions (). The dendrogram was clearly divided into 3 groups. Eight conditions formed the first group containing 3 conditions from 2,4DNT, all 4 conditions from 2,6DNT at 24 h, and one from 2ADNT at 48 h. The second group included 22 conditions. Interestingly, this group was almost entirely composed of conditions from 2ADNT, 4ADNT and TNT except two conditions from 2,4DNT, 48 h. The third group contained 10 conditions in which two distinctive subgroups were formed. One subgroup contained 3 conditions that were from 2ADNT at 48 h, while the other subgroups included conditions from 2,4DNT at 48 h and 2,6DNT at 48 h. Experimental conditions with the same chemical dose and time were likely to cluster together. Conditions within the same dose and the same compound from different time points (24 h or 48 h) seldom clustered together in a subgroup indicating that the time effect is much stronger than the dose effect for all the compounds.
Hierarchical clustering of experimental conditions.
From , most of the gene expression profiles of 2,4DNT and 2,6DNT clustered together and most of the 2ADNT, 4ADNT and TNT induced gene expression profiles grouped together. From supplementary (File S1
), we noticed that gene expression profiles of 2,4DNT and 2,6DNT clustered together and 2ADNT, 4ADNT and TNT induced gene expression profiles grouped together. We further made averaged conditions based on compound treatments and ignored dose and time effects. A simple condition tree using the same flagged transcripts is shown in supplementary . 2,4DNT and 2,6DNT formed in one separate group, 2ADNT, 4ADNT and TNT were located in the other group in which 4ADNT and TNT were clustered together. Since 2ADNT and 4ADNT are metabolites of TNT, it is easy to understand that the 3 compounds form in a group. The expected results provide proof that this high-throughput experiment is reliable.
Common gene expression patterns of significantly differentially expressed genes due to nitrotoluene exposure
We also performed a hierarchical clustering per treatment. Because the two very strong gene regulators, 2,4DNT and 2,6DNT have over 2000 genes that were affected, in order to identify most significantly regulated genes, a more stringent false discovery rate (FDR) with adjusted p-value less than 0.002 was applied resulting in 1418 and 1241 significantly differentially expressed transcripts for 2,4DNT and 2,6DNT, respectively. For the other three compounds, no more stringent FDR was applied because their changed gene numbers were much less than 2000. One-way hierarchical clustering was performed across various doses and times per chemical. Interestingly, the 5 compounds share similar gene expression patterns (). The genes were primarily divided into two distinctive expression patterns: early up-regulated (24 h) and early down-regulated (24 h). Other common patterns shared by the 5 compounds include a clear dose response for both up-regulated and down-regulated transcripts at 24 h. We also noticed that there were more up-regulated than down-regulated genes at 24 h for all the compounds which is consistent with . For all the compounds but 2ADNT, transcripts generally up-regulated at 24 h were also up-regulated at 48 h as well. The common gene expression patterns shared by the compounds imply that these compounds may share some common mechanisms.
Hierarchical clustering of significantly regulated genes.
Overlapping functional terms of differential genes among the 5 compounds
To gain insight into the functional categories of these significantly changed genes, we performed gene ontology (GO) analyses based on biological processes. Early up-regulated and early down-regulated gene groups for each compound across all doses () underwent GO analyses separately. We first examined the similarity of functional terms from the up-regulated gene groups between compounds (). The greatest overlap occurred between 2,4DNT and 2,6DNT. 4ADNT also had a relatively high similarity of functional terms with 2,4DNT and 2,6DNT (). These results are consistent with the comparison of general gene expression profiles ( and File S1
). Consistent with observations of little effect on biological parameters and differentially expressed genes, 2ADNT had very few significantly affected GO terms. Four significantly up-regulated functional terms were shared by all 5 compounds (). The top significantly changed GO terms regulated by all 5 compounds were “metabolic process” and “cellular metabolic process” (a child term under metabolic process), protein folding, and “xenobiotic metabolic process” (a child term of “response to xenobiotic stimulus”).
Comparison of significantly functional terms up-regulated by the 5 compounds.
Commonly most significantly up-regulated functional terms based on biological process of Gene Ontology in response to the compounds.
Twenty significantly up-regulated GO functional terms were shared by 2,4DNT, 2,6DNT, 4ADNT and TNT, among which at least half were child terms of “metabolic process” and “cellular metabolic process”. Primarily, these terms could be categorized as “macromolecule metabolic process”, “RNA metabolic process” and “cellular biosynthetic process”. The major term under macromolecule metabolic process was cellular protein metabolic process, under which a significant common term was protein folding.
The functional term “cellular biosynthetic process” was more enriched in response to TNT, 4A DNT than 2,4DNT and 2,6DNT treatments (). NQO1, represented in “xenobiotic metabolic process” and in the “cellular biosynthetic process”, was up-regulated by all 5 compounds indicating this gene may play an important role in the response to all 5 compounds.
Under the functional term “RNA metabolic process”, pirin (PIR) was up-regulated by 2,6DNT, 2ADNT, 4ADNT and TNT. Tryptophanyl-tRNA synthetase (WARSs) was induced by 2,6DNT, 4ADNT and TNT; small nuclear ribonucleoprotien polypeptide A (SNRPA), NHP2-like protien1 (NHP2L1) was up-regulated by 2,4DNT, 4ADNT and TNT. 2,4DNT, 2,6DNT and 4ADNT shared more induced genes of “RNA metabolic process” which included the gene for absent, small, or homeotic-like gene (ASH2), general transcription factor II H, polypeptide 1 (GTF2H1), EBNA1 binding protein 2 (EBNA1BP2), nucleolar protein 5A (NOL5A), peptidylprolyl isomerase (cyclophilin)-like 3 (PPIL3) and tRNA splicing endonuclease 2 homolog (Saccharomyces. cerevisiae) (TSEN2).
Other shared up-regulated functional terms included “mitochondrion organization and biogenesis” as well as “protein transport” (). Interestingly, some genes present in earlier categories such as P53, HSPD1 also fit into both categories. The gene translocase of inner mitochondrial membrane 9 (TIMM9) found under both categories, was more strongly induced by 2,4DNT and 2,6DNT.
Further comparison of up-regulated GO functional terms of 2,4DNT, 2,6DNT and 4ADNT revealed a high number of shared terms, 65 terms. “DNA metabolic process,” a child term of “nucleobase, nucleoside, nucleotide and nucleic acid metabolic process” and “biopolymer metabolic process”, was a key term shared by the 3 compounds.
When down-regulated GO biological process terms were compared, 2,4DNT and 2,6DNT were more similar than other compounds (). This high similarity is consistent with the up-regulated GO biological process terms and gene expression profiles as described earlier. Such correlation suggests that these two compounds have highly similar molecular modes of action. “Lipid biosynthetic process”, which is highly related to “lipid metabolic processes”, was affected by all compounds (). Two critical downstream GO terms of “lipid metabolic process” were “fatty acid metabolic process” and “sterol metabolic process”. “Amino acid metabolic process” and/or its child terms, “organic acid metabolic process” and “carboxylic acid metabolic process”, were significantly down-regulated by 2,4DNT, 2,6DNT, 2ADNT and TNT.
Comparison of significantly functional terms down-regulated by the 5 compounds.
Commonly most significantly down-regulated functional terms based on biology process of Gene Ontology in response to the compounds.
The second common significantly down-regulated GO biological process term was “immune system process” under which “immune response” was also affected. Both terms were more enriched in 4ADNT and TNT exposures than with exposure to the other 3 compounds(). Our results suggest that these chemicals affect immune response and chemokine signaling.
Most significantly regulated genes involved in immune response for the 5 compounds.
Pathways influenced by all the 5 compounds
To further explore the molecular mechanisms, we used the most significantly differentially expressed transcripts of each chemical exposure to perform canonical pathway analyses. In general, a number of pathways were common between any compound pair (). Five pathways were strongly affected by all 5 compounds (File S1
). The common pathways were also generally present in the top significant pathways affected by the compounds (). The most significantly regulated genes are listed in File S1
for each pathway regulated by each compound. There were 5 pathways that were significantly regulated by 4 of the 5 compounds (File S1
). Xenobiotic metabolism signaling and metabolism of xenobiotics by cytochrome P450 pathways were highly enriched in 2,6DNT, 4ADNT, 4ADNT and TNT exposure gene expression profiles. Propanoate metabolism, fatty acid metabolism and valine, leucine and isoleucine degradation pathways were strongly affected by 2,4DNT, 2,6DNT, 2ADNT and TNT. Nearly half of these 10 commonly impacted pathways (File S1
), belong to amino acid metabolism, suggesting that amino acid metabolism plays a critical role in the compounds' effects which is consistent with Gene Ontology analysis.
Comparison of significantly canonical pathways regulated by the 5 compounds.
Most significantly canonical pathways regulated by the 5 compounds.
Examination of the most significantly regulated genes in pathways reveals that a number of genes from a few different gene families play a repeated role in multiple pathways that were regulated by chemical exposures. These gene families included aldehyde dehydrogenase 1 (ALDH1), gluathione S-transferase (GST), heat shock protein (HSP), and cytochrome P450 (CYP) families. ALDH1 family is involved in all the common pathways except NRF2-mediated oxidative stress response. The most significant differentially expressed genes under this family included ALDH1A1, ALDH1A3, ALDH1B1, ALDH1L1, and ALDH1L2. They were strongly up-regulated by at least two compounds, with some members affected by 3 compounds. For example, the family member ALDH1B1 was regulated by 2,4DNT, 2ADNT, and 4ADNT while ALDH1L1 was significantly affected by 2,4DNT, 4ADNT and TNT. Several genes in the GST family including GSTM1 (muscle), GSTM2, GSTM3, GSTM4 or GSTM7-7, GSTM5, GST pi1(GSTP1) and GST theta 1(GSTT1) participate in at least half of the commonly affected pathways (File S1
). These genes were more likely to be up-regulated by the compounds.
The most significantly regulated genes shown in the commonly affected pathways in the HSP family included HSP 90KDa alpha (cytosolic), class A member 1(HSP90AA1), heat shock 27kDa protein 1 (HSPB1), HSP 90kDa alpha (cytosolic), class B member1 (HSP90AB1), and heat shock 22kDa protein 8 (HSPB8). These genes were usually up-regulated and involved in the following common pathways: aryl hydrocarbon receptor signaling, xenobiotic metabolism signaling, and NRF2-mediated oxidative stress response. We have found that HSP genes were highly enriched in the GO biological process term, “response to xenobiotic stimulus”. Since xenobiotic metabolism signaling also encompasses aryl hydrocarbon receptor signaling, and NRF2-mediated oxidative stress response, the pathway analysis results are consistent with the GO analysis indicating that heat shock proteins play an important role in the response to exposure to the 5 compounds.
A large number of differentially expressed genes belonged to the cytochrome P450 family (CYP1A1, CYP1A2, CYP3A5, CYP3A43, CYP4A11, CYP7A1, CYP51A1, CYP2C7, CYP2C44, CYP2C70, CYP2D6, CYP2D12, CYP2D26, CYP4F8, and CYP2J9. They mainly took part in the following common pathways: aryl hydrocarbon receptor signaling, LPS/IL-1 mediated inhibition of RXR function, xenobiotic metabolism signaling and metabolism of xenobiotics by cytochrome P450. These CYP family genes were primarily up-regulated by all the compounds.
Although there was no common significant cell death signaling pathways involved in the expression profiles induced by these compounds, different pathways associated with cell death signaling were strongly impacted. Apoptosis and/or P53 signaling were evidently affected by two and more of these compounds. Three genes which were commonly regulated by two or more compounds were P53, Bcl2-associated X protein (Bax), and caspase 3(CASP3). Some pathways such as estrogen receptor pathway and nucleotide excision repair pathway were only significantly regulated by 2,4DNT and 2,6DNT. Antigen presentation pathway was only evidently affected by TNT.
Nitrotoluenes impact common gene networks
To further understand the common mechanisms, we looked for differentially expressed genes in common with all five chemical exposures. Fifty-four transcripts corresponding to 47 non-redundant genes were significantly affected by at least four of the 5 compounds (Table S11
). We constructed several networks using the 47 genes. Interestingly, one significant network was associated with hepatic system disease and liver cholestasis () suggesting this network could present a common mechanism for possible induction of hepatoxicity seen with exposure to 2,4DNT, 2,6DNT and TNT. Although no adverse effects were seen with the chemicals 2ADNT and 4ADNT, the shared network suggests the potential for some effects similar to the other chemicals under prolonged exposure conditions. Several shared up-regulated genes including ABCC3, PRMT7, GSS, SNRPA, PANX2, MRPL24, and ALDH1L1 all connected with hepatocyte nuclear factor 4 alpha (HNF4A). While HNF4A was not a common altered gene, as a central transcription factor in this network, it may be important in modulating the effects caused by these chemicals. Genes involved in NRF2-mediated oxidative stress response pathway were also found in the network.
Common and specific gene networks related to liver toxicity induced by the compounds.
Based upon evidence that the nitrotoluenes impacted a common gene network related to liver effects, we next examined each chemical exposure for specific gene networks related to liver function. To fulfill this goal, we employed the significantly changed genes by each compound that have been known to be associated with liver function to build gene networks based on Ingenuity knowledge base. Specific significant gene networks that were associated with hepatic system function, liver cholestasis and/or liver damage were generated for each compound (). Several transcription factors such as NFE2L2 (NRF2), EGR1, JUND, and SREBF1 were highly connected in the 2,4DNT affected gene network, indicating these genes may play a role in 2,4DNT impacts on liver (). In the 2,6DNT regulated gene network, the genes CDKN1A, PAK1, ITGAV, and NFE2L2 (NRF2) were highly connected () suggesting these genes are important in 2,6DNT effects on liver. Interestingly, the highly connected genes in the TNT regulated gene network () were largely involved in immune response. These genes consisted of CXCL10, CCL13, TNFSF10, NOS3 and CCND1 and may play a major role in TNT adverse effects on the liver.
In the 2ADNT affected gene network, CDKN1A and ABCC3 were greatly connected (), indicating possible roles in 2ADNT affected potential liver function. Genes including HMOX1, PRKCz, PKC, Raf, and RAF1 were highly connected in the gene network responsive to 4ADNT (). These genes could play more important roles in potential liver effects regulated by 4ADNT.
Despite the different topologies and gene compositions of the networks, a number of common pathways that were modulated by all compounds and were also present in all sub networks. These included NRF2-mediated oxidative stress response, aryl hydrocarbon receptor signaling, xenobiotic metabolism signaling, LPS/IL-1 mediated inhibition of RXR function, and xenobiotic metabolism signaling pathways.
Verification of microarray responses using real time QRT-PCR
To verify the credibility of microarray data, we first selected 96 genes that are responsible for oxidative stress (Table S12
) to perform real time quantitative reverse transcriptase PCR (QRT-PCR). Real time qPCR studies were conducted using vehicle control samples and 199 mg/kg 2,4DNT samples at either 24 or 48 h time points. The same RNA was used for QRT-PCR as was used for microarray assessment. We found a surprising consistency between the two approaches. The significantly regulated gene numbers by 2,4DNT were exactly the same for both QRT-PCR and microarray, 32 for 24 h and 34 for 48 h (File S1
). Moreover, both methods were identical for the positive and negative directions of all the genes between both time points (File S1
). A regression coefficient of 0.93 between the two approaches for both time points was achieved (File S1
Previous studies by Wintz et al.
found that 24DNT impacted oxygen transport and lipid metabolism in livers of the fish fathead minnow (Pimephales promelas) via interaction with hemoglobin and peroxisome proliferator activated receptor alpha (PPARα) pathways. We examined whether similar impacts are observed in rats exposed to nitrotoluenes. To further verify gene expression profiles, genes peroxisome proliferator activated receptor alpha (PPARα), transferrin (TF), apolipoprotein B (APOB), and alpha Hemoglobin (aHG) were chosen to carry out QRT-PCR across all samples including different compounds, doses, and time points (File S1
). Correlations of 0.71 and 0.85 were observed between the results of QRT-PCR and microarrays (File S1
). In fathead minnows, the gene TF was down-regulated in livers of fathead minnows exposed to 24DNT. Here, expression of TF was also attenuated by 2,4DNT at 24 h, in addition to 2,6DNT at 24 and 48 h with clear dose responses for both QRT-PCR and microarray with R2
0.79 (File S1
). The gene PPARα has been shown to be repressed by 2,4DNT in liver of exposed fathead minnow. PPARα was also down-regulated by 2,4DNT at 24 h in rat with a strong dose response. PPARα was consistently suppressed by 2,6DNT at 24 h and 48 h but up-regulated by 2ADNT and TNT at 48 h. Microarray and QRT-PCR data had an R2
0.84. Other genes such as APOB were reported to be down-regulated in fathead minnows exposed to 2,4DNT while alpha Hemoglobin (aHG) was up-regulated. Hepatic nuclear factor 4, alpha (HNF4A) was not affected by 2,4DNT in fathead minnows. HNF4A had no significant change with exposure to the five chemicals (not shown).We saw similar impacts on these 3 genes by 2,4DNT in rat livers by both QRT-PCR and microarray (File S1
). While APOB was up regulated by all chemicals, aHG was up regulated by 24DNT but it was primarily down-regulated by other compounds (File S1
). Microarray and QRT-PCR data for APOB and aHG had R2
values of 0.75 and 0.71 respectively (File S1
QRT-PCR analysis of genes found in networks and affected pathways
QRT-PCR analysis of select genes present in gene networks and affected pathways were performed including GSTM4, NRF2, ALDH1A1, and fatty acid binding protein 2 (FABP2) had R2
values of 0.85, 0.83, 0.73, and 0.72, respectively, when compared to microarray data across all samples (File S1
). Differential effects were seen on NRF2 expression with up-regulation at 24 h with 2,4DNT and 2,6DNT while it was down-regulated at 24 h with 2ADNT, 4ADNT and TNT. At 48 h, NRF2 expression was up-regulated with 2,4DNT, 2ADNT, 4ADNT and TNT while no significant effect was seen with 2,6DNT exposure. GSTM4 expression was consistently up regulated at 24 h by all five chemicals. Differential effects due to nitrotoluene exposure were observed on expression of ALDH1A1. 2,4DNT up-regulated ALDH1A1 at both 24 and 48 h. 2,6DNT down-regulated ALDH1A1 at 24 h while producing little consistent effect at 48 h. 2ADNT up-regulated expression at 24 h and down-regulated expression at 48 h. 4ADNT up-regulated expression at 24 and 48 h. TNT had a slightly inconsistent effect but generally up-regulated ALDH1A1 expression at both 24 h and 48 h. Fatty acid binding protein 2 (FABP2) was up-regulated by all chemicals except 2,6DNT which only up-regulated FABP2 at a single dose and time.
Nitrotoluene exposure adversely impacts lipid metabolism
The enrichment in GO biological process related to lipid metabolism and pathways including fatty acid metabolism and NRF2-mediated oxidative stress response in addition to data from other species suggests that nitrotoluenes exposure would significantly impact lipid biosynthesis in liver. To validate this we analyzed lipid metabolites in livers of the exposed rats. Two doses were selected at 24 h to measure lipids levels, the second lowest and the highest, and vehicle control of the 5 compounds. A total of 340 lipid species (Table S13
) belonging to 12 lipid classes were measured using a mass spectrometry approach 
. One-Way ANOVA with a multiple adjustment (p value <
0.05) was applied to identify differentiated lipid species for each compound. 2,4DNT, 2,6DNT and TNT had stronger impact on the regulation of lipid profiles than 2ADNT and 4ADNT. Overall, more lipid species were down-regulated by the compounds. Thirty-six lipid species were commonly affected by at least 3 of the 5 compounds (). Of them, 21 lipid species were commonly down-regulated in response to 2,4DNT, 2,6DNT and TNT: Nine PC with one ether-linked (alkyl or alkenyl) chain (ePC), 8 phosphatidylcholine (PC), and 4 phosphatidylethanolamine (PE). Phosphatidylinositol (PI 38
5; phospholipid species are indicated by total acyl carbons
total acyl carbob-carbon double bonds) and lysoPC20
2 were down-regulated by all the compounds. Only 3 lipid species, PC 34
2, PE 34
2 and PE 36
2 were up-regulated by all the compounds. The lipid species lysoPC 18
3 was decreased by 2,4DNT, 2,6DNT, 2ADNT and 4ADNT but increased by TNT. PS 38
5 was down-regulated by 2,4DNT, 2,6DNT, 2ADNT, and TNT but up-regulated by 4ADNT. Interestingly, no matter whether a lipid species was up or down-regulated, a strong dose effect was observed (). We also compared total lipid species differences in each lipid class. There were more total lipids that were down-regulated by 2,4DNT, 2,6DNT, 2ADNT and TNT except 4ADNT (). Total ePC, ePE were commonly down-regulated by all the compounds. There was no single lipid class that was commonly up-regulated by all 5 compounds. The amount of total PA was reduced in response to 2,4DNT, 2,6DNT, 2ADNT and TNT but was elevated in response to 4ADNT. Total PS was up in response to 2,4DNT, 2,6DNT and 4ADNT but down to 2ADNT and TNT. Lipidomics results of an overall down-regulated lipid profile are well correlated with the gene expression profiles where lipid metabolism was predicted to be inhibited.
Hierarchical clustering of commonly significantly regulated lipid species in response to 2,4DNT, 2,6DNT, 2ADNT, 4ADNT and TNT.