Suppressing the activation of protein kinase A as a DNA damage-independent mechanistic lead for dihydromethysticin prophylaxis of NNK-induced lung carcinogenesis

Abstract

Our earlier work demonstrated varying potency of dihydromethysticin (DHM) as the active kava phytochemical for prophylaxis of tobacco carcinogen nicotine-derived nitrosamine ketone (NNK)-induced mouse lung carcinogenesis. Efficacy was dependent on timing of DHM gavage ahead of NNK insult. In addition to DNA adducts in the lung tissues mitigated by DHM in a time-dependent manner, our in vivo data strongly implicated the existence of DNA damage-independent mechanism(s) in NNK-induced lung carcinogenesis targeted by DHM to fully exert its anti-initiation efficacy. In the present work, RNA seq transcriptomic profiling of NNK-exposed (2 h) lung tissues with/without a DHM (8 h) pretreatment revealed a snap shot of canonical acute phase tissue damage and stress response signaling pathways as well as an activation of protein kinase A (PKA) pathway induced by NNK and the restraining effects of DHM. The activation of the PKA pathway by NNK active metabolite 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) at a concentration incapable of promoting DNA adduct was confirmed in a lung cancer cell culture model, potentially through NNAL binding to and activation of the β-adrenergic receptor. Our in vitro and in vivo data overall support the hypothesis that DHM suppresses PKA activation as a key DNA damage-independent mechanistic lead, contributing to its effective prophylaxis of NNK-induced lung carcinogenesis. Systems biology approaches with a detailed temporal dissection of timing of DHM intake versus NNK exposure are warranted to fill the knowledge gaps concerning the DNA damage-driven mechanisms and DNA damage-independent mechanisms to optimize the implementation strategy for DHM to achieve maximal lung cancer chemoprevention.

Graphical Abstract

Introduction

Lung cancer is the leading cause of cancer-related deaths in the USA and worldwide, among both men and women (1,2). In the USA, one in every 16 people is expected to be diagnosed with lung cancer in their lifetime and ~140 000 people lose lives to lung cancer each year (3,4), exceeding the total mortalities combined from breast, colon and prostate cancers. Moreover, the five-year overall survival of lung cancer patients remains below 20% even with the recent introduction and significant progress in targeted therapies and immunotherapies. Given its high incidence and rather limited success in clinical management, alternative prophylactic strategies targeting initiation or precancerous stages, therefore, are likely public health measures needed to reduce lung cancer risk and improve lung cancer outcome.

Different from the complex and sometimes unknown etiology of other malignancies, tobacco exposure is the dominant risk factor for lung cancer, which is estimated to contribute to 80–90% of lung cancer incidence. Tobacco avoidance and cessation, therefore, are the ideal strategies for reducing the risk of lung cancer (5). Sadly, there are over 30 million adult smokers in the USA (6) and there is little indication that this number will decrease significantly in the near future (7,8). Quitting is also difficult because of the addictive nature of nicotine in tobacco products; <5% smokers succeed in cessation although ~70% of them attempt to quit each year (9,10). Chemoprevention, the use of a synthetic, natural or biological agent to reduce or delay lung cancer occurrence, therefore is needed to help addicted smokers and second-hand smokers reduce lung cancer risk.

Tobacco smoke contains various carcinogens and co-carcinogens, which collectively initiate and promote lung carcinogenesis (5). 4-(Methylnitrosamino)-1-(3-pyridyl)-1-butanone (commonly known as nicotine-derived nitrosamine ketone [NNK]) is a tobacco-specific N-nitrosamine with potent lung carcinogenicity, inducing lung tumor and adenocarcinoma in multiple rodent species (11). Ample evidence suggests that NNK may be key to lung cancer formation among USA smokers. First, the increased adenocarcinomas of lung cancer in the USA over the past few decades appeared to correlate with the increased content of NNK in tobacco products (12,13). Several prospective studies also showed that higher levels of NNK urinary metabolites were associated with increased risk of lung cancer among smokers, linking its exposure to lung cancer risk (14–17). Lastly, a number of studies discovered that polymorphism of multiple genes involved in NNK metabolism is associated with differential lung cancer risks among smokers (18–22), again supporting NNK contribution to lung cancer risk. Agents that can block NNK-induced lung carcinogenesis, therefore, have the potential to help reduce lung cancer risk.

Our earlier work discovered that dihydromethysticin (DHM, Figure 1), a natural compound in kava, holds promise as a chemopreventive agent against NNK-induced lung carcinogenesis in A/J mice (23). Specifically, at human-exposure relevant dosages, DHM (given before carcinogen insult) was able to completely block NNK-induced lung tumor formation in A/J mice (23). Our mechanistic investigation demonstrated that DHM differentially reduces NNK-induced DNA damage in the lung via enhancing the clearance of 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (the major in vivo carcinogenic metabolite of NNK, commonly known as NNAL) through glucuronidation (24,25). The chemopreventive potential of DHM against lung cancer was also supported by the results of a pilot kava trial among current smokers (26), which showed that a 1-week kava exposure resulted in reduced urinary DNA adducts from NNK and increased urinary NNAL clearance. To optimize the treatment regimen of DHM, which is critical for its clinical relevance, we recently characterized the temporal relationship of a single-dose oral gavage DHM given from 0 to 40 h before NNK exposure (27). The temporal effects of DHM on NNK-induced acute DNA adducts and long-term lung tumor formation overall matched (Figure 1). This is consistent with our hypothesis that reduction in DNA adducts by DHM was a key mechanism for its chemopreventive activity. Yet, the complete blockage of tumor formation was achieved in spite of an incomplete protection against lung DNA adducts (i.e. 75–88% reductions between DHM given between −8 h and −3 h). Furthermore, the −40h DHM (given 40 h before NNK) reduced tumor multiplicity by 52% without reducing lung DNA adducts while the 0 h DHM (given concurrently with NNK) reduced tumor multiplicity by only 50% although lung DNA adducts were reduced by 63% (Figure 1) (27). These data indicate that to be an efficacious initiation blocker, the timing of DHM ahead of carcinogen insult must be optimized through targeting both DNA damage reduction and DNA damage-independent mechanisms (27).

Figure 1.

Design schema and summary outcome for evaluating the temporal effects of prophylactic DHM dosing on NNK-induced lung DNA adducts (acute) and tumors formation (long-term) from our previous publication (27). The data implicate DNA damage-driven and -independent mechanisms in NNK-induced lung carcinogenesis and DHM prevention.

Substantial evidence from others also suggests the involvement of DNA damage-independent mechanisms in NNK-induced lung carcinogenesis and prevention. For instance, Hecht et al. used NNK DNA-damaging model compounds to generate similar amounts of DNA adducts in A/J mice and found that the amount of lung tumors induced by the model compounds were substantially lower than that induced by NNK (28). Similarly, Chuang et al. showed that indole-3-carbinol and phenethyl isothiocyanate efficiently reduced NNK-induced DNA adducts (more effective than DHM) but could NOT completely block NNK-induced lung tumor formation (29,30) as did DHM. Indeed, COX-2 activation and inflammation had once been proposed as DNA damage-independent mechanisms for NNK-induced lung carcinogenesis (31–34). There has been little progress in this direction recently, particularly with respect to the in vivo stimuli (NNK or its metabolites), the upstream target(s) and the detailed signaling pathways leading to NNK-associated COX-2 activation and inflammation.

In this study, we leveraged systems biology non-biased approach to interrogate DNA damage-dependent and -independent mechanism(s). Specifically, we performed an RNA seq-based transcriptomic analysis of the lung tissues from the A/J mice following an acute exposure to solvent vehicles, NNK, or DHM/NNK treatment regimen. Such treatment regimens would have resulted in distinct lung carcinogenesis status in the long-term bioassay—A/J mice without NNK exposure (no lung tumor), A/J mice with NNK exposure (high lung tumor multiplicity) and A/J mice with NNK exposure but pretreated with DHM (no lung tumor). The RNA seq results not only identified a few signaling pathways as the mechanistic leads for NNK-induced lung carcinogenesis but also clearly demonstrated the counteracting effect of DHM on these oncogenic processes. One particular pathway, the activation of protein kinase A (PKA) signaling and its targeting by DHM, was substantiated in vivo and in a lung cancer cell culture model via a pharmacological approach.

Material and methods

Caution: NNK and NNAL are potential human carcinogens and should be handled carefully in well-ventilated fume hoods with proper protective clothing and personal protective equipment.

Chemicals, reagents and animal diets

Natural DHM was isolated from a kava product supplied by Gaia Herbs, Inc. (Brevard, NC), following reported protocols (23). NNK was purchased from Toronto Research Chemicals (Toronto, Ontario, Canada). NNAL was synthesized from NNK via a sodium borohydride reduction, following reported protocols (35). Phospho-CREB (Ser133) rabbit mAb (#9198), CREB rabbit mAb (#9197) and anti-rabbit IgG (#7074) were purchased from Cell Signaling (Danvers, MA). Mouse monoclonal anti-beta-actin (#A2228) was purchased from Sigma Aldrich (Burlington, MA). Qiagen RNeasy Mini Kit (#74106) was purchased from Qiagen (Hilden, Germany).

NNK and DHM treatment in A/J mice

Detailed procedures for the animal experiments, characterizing the temporal effects of DHM pretreatment (−40 h to 0 h) on NNK exposure induced acute DNA adducts (2 h) and long-term tumor outcome (17 weeks) respectively, have been published (27). Figure 1 summarizes the DNA adducts and tumor results. Briefly, female A/J mice (5–6 weeks of age) from the Jackson Laboratory (Bar Harbor, ME) were maintained in the specific pathogen-free facilities, according to animal welfare protocols approved by the Institutional Animal Care and Use Committee at the University of Florida. After 1-week acclimation, they were weighed and randomized to different groups: mice in Control group (no DHM, no NNK) were given PEG-400 gavage (vehicle for DHM) and saline i.p. injection (vehicle for NNK), respectively. Mice in the NNK group were given a single i.p. injection of NNK in saline (100 μl, 100 mg/kg of body weight) after receiving PEG-400 (200 µl) via gavage for a duration matching the DHM pretreatment. Mice in the NNK + DHM groups were given a single dose of 0.8 mg DHM pretreatment in PEG-400 (200 µl) via oral gavage for a duration of 40 h (−40 h) to 0 h (concurrent) followed by a single i.p. injection of NNK in saline (100 μl, 100 mg/kg of body weight). Several mice from each treatment regimen (n = 2–4) were sacrificed 2 h after NNK. A portion of each lung was immediately immersed in RNAlater solution until RNA isolation. The rest of the lungs were frozen on dry ice during sample collection and later stored in −80°C freezer until DNA adduct analyses (27) and additional molecular characterization. Such lung tissues from the following three groups were also used for the RNA seq transcriptomics analyses: Control (n = 2): mice were given PEG-400 and saline, respectively, with an 8-h time gap; NNK (n = 4): mice were given a single i.p. injection of NNK in saline (100 μl, 100 mg/kg of body weight) 8 h after PEG-400 (200 µl) gavage; NNK + DHM (n = 3): mice were given a single dose of 0.8 mg DHM in PEG-400 (200 µl) via oral gavage followed by a single i.p. injection of NNK in saline (100 μl, 100 mg/kg of body weight) 8 h after DHM treatment. Majority of the mice were maintained for 17 weeks for enumeration of the tumor nodules on the lung surface (27).

RNA isolation and quantification

The Qiagen RNeasy Mini Kit was used, following the protocol provided by Qiagen. Briefly, 20 mg of mouse lung tissues preserved in RNAlater RNA stabilizing solution were disrupted in 600 μl of Buffer RLT and homogenized using a Qiagen TissueRuptor. One volume of ethanol was then added to the lysate. The sample was then applied to the RNeasy Mini spin column and washed with RPE and RW1 sequentially. Finally, RNA was eluted in RNAse-free water. Its quantity (in ng/μl) and quality (260/280) were measured. Of the isolated RNA, 420 μg per sample was used for RNA sequencing analyses and the remaining samples were used for quantitative real-time PCR (qRT-PCR) analyses.

RNA sequencing

RNA sequencing was performed by BGI Americas Corporation (Cambridge, MA) following its standard protocols. Briefly, mRNAs were purified from total RNA using oligo(dT)-attached magnetic beads, and were fragmented into small pieces using fragmentation reagent. cDNA was synthesized with first strand generated using random hexamer-primed reverse transcription, followed by a second-strand cDNA synthesis. The synthesized cDNA was subjected to end-repair and then was 3ʹ adenylated. Adapters were ligated to the ends of these 3ʹ adenylated cDNA fragments. Following standard PCR amplification, the cDNA library was validated by the Agilent Technologies 2100 bioanalyzer. The double-stranded PCR products were heat-denatured and circularized by the splint oligo sequence. The single-strand circle DNA (ssCir DNA) was formatted as the final library. The library was amplified with phi29 to make DNA nanoball, which had more than 300 copies of one molecule. The DNA nanoballs were loaded into the patterned nanoarray and single end 50-base (pair end 100) reads were generated in the way of sequenced by synthesis.

Bioinformatics and data preprocessing

The raw reads of the RNA-sequencing data were aligned to the Mus musculus (house mouse) genome assembly GRCm38 (mm10) using the HISTA2 software (36). Deduplicated reads were marked and removed by using the Picardtools software (37). Gene expression count data was extracted using the HTseq software (38). The count data were normalized to counts per million reads. The data were further filtered by removing genes with mean counts per million value less than 1.

Dimensional reduction

To visualize the similarity of the RNA-sequencing samples, we performed principal component analysis, which reduced the dimensionality of thousands of genes onto a two-dimensional space, while retaining the relative similarity of all pairs of samples.

Differentially expressed genes

The differential expression analyses comparing (i) NNK versus Control and (ii) NNK + DHM versus NNK were performed by using the “edgeR” package in R (39,40), which utilizes a negative binomial model for modeling count data. To control for multiple testing, we employed the Benjamini-Hochberg method to convert P-values to q-values (FDR-adjusted P-values), where FDR represents the false discovery rate (41).

Comparison of differentially expressed genes from NNK versus Control and NNK + DHM versus NNK

We first visualized the differentially expressed genes (DEGs) from both (i) NNK versus Control and (ii) NNK + DHM versus NNK using heatmaps with a P-value ≤ 0.05 and a q-value ≤ 0.05 cutoffs. Since such analyses only cover DEGs with certain significance cutoffs, and the results are expected to vary as the significance threshold changes, we further deployed the Rank-Rank Hypergeometric Overlap test, a threshold-free algorithm (42), to evaluate trends of overlap between two ranked gene lists. Briefly, the gene lists from both datasets (i.e. (i) NNK versus Control and (ii) NNK + DHM versus NNK) were ranked by the –log10 P-values signed by the effect size direction. By iterating all possible rank values as the cutoffs for both datasets, a hyper-geometric test was used to calculate the significance level of the overlap at each pair of cutoffs. The resulting significance levels of all possible cutoffs were presented using a heatmap.

qRT-PCR validation

The same RNA samples used for RNA seq analyses were reversely transcribed with qRT-PCR performed for representative gene candidates. Primer sequences used were listed in Supplementary Table S1, available at Carcinogenesis Online.

Pathway analyses

Pathway enrichment analyses were performed to identify the canonical pathways and upstream regulators by using Ingenuity Pathway Analysis (IPA) (43) (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuitypathway-analysis).

Human H1299 cell line monolayer cell culture with NNAL, isoproterenol and DHM treatment

H1299 cell line was purchased from ATCC (Manassas, VA) and authenticated via the Cell Line Authentication Service provided by Genetica DNA Laboratories (Burlington, NC). H1299 cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (Gibco). For experiments, after starvation with 0.5% fetal bovine serum RPMI 1640 medium overnight, H1299 cells were treated with NNAL or isoproterenol (ISO) for 30 min. Pretreatment with DHM was started 1 h before NNAL or ISO treatment to evaluate its effects on NNAL or ISO-driven signaling.

Western blotting analysis

Whole-cell lysates from H1299 cells were prepared in RIPA lysis buffer. Protein lysates from mouse lung tissues were prepared similarly. Briefly, 20 mg of lung tissue was homogenized in 250 µL RIPA buffer. The supernatant was collected after centrifugation at 13 000 g for 15 min at 4°C. The concentration of protein in each lysate sample was quantified using bicinchonicic acid (BCA) assay. The lysate samples with 40 – 60 µg of total protein from each sample were denatured in SDS-PAGE sample buffer and resolved on 4–12% Bis-Tris PAGE gels. The separated proteins were transferred onto polyvinylidene difluoride membrane followed by blocking with 5% non-fat milk powder (w/v) in Tris-buffered saline (10 mM Tris–HCl, pH 7.5, 100 mM NaCl, 0.1% Tween-20) for 1 h at room temperature. After blocking, the membranes were incubated with cAMP response element-binding protein (CREB) primary antibodies (1:1000) overnight at 4°C followed by an appropriate peroxidase-conjugated secondary antibody (1:5000) for 2 h at room temperature and visualized by the Bio-Rad ChemiDoc Imaging system. To correct for protein loading difference, each membrane was stripped with Restore Western Blot stripping buffer (Thermo Scientific) and re-probed with β-actin antibody (1:10 000). Semi-quantification of the bands was performed using ImageJ software. After background subtraction, the expression level of each protein was normalized to β-actin abundance.

Statistical analyses

Data on lung O⁶-mG and tumor multiplicity were reported as mean ± SD (n = 3 – 5) as detailed in our earlier publication (27). Data on qRT-PCR-based mRNA were reported as mean ± SD (n = 2 – 4). One-way analysis of variance was first used to compare means among NNK and the other groups. Dunnett’s test was then used for comparisons of NNK and the other treatment groups when analysis of variance analysis was statistically significant. P-value ≤ 0.05 was considered statistically significant. For qRT-PCR and Western blot analysis, at least three biological replicates were performed for quantitative analyses with one representative data shown. All analyses were conducted in GraphPad Prism 4 (GraphPad Software, Inc. La Jolla, CA).

Results

RNA seq bioinformatics

The alignment rate for each sample is > 95% (Supplementary Table S2, available at Carcinogenesis Online), confirming the quality of the RNA preparation and sequencing procedure. A total of 55 291 genes were obtained from the gene expression matrix extraction. Among them, 15 438 genes passed the filtering criteria (counts per million  ≥ 1), which were used for further analyses.

Dimensional reduction

As shown in Figure 2A, the three treatment groups (Control, NNK and NNK + DHM) were well separated and clustered. Interestingly, along the first principal component (PC1, accounting for 28.2% variance), the NNK samples were to the right of the Control samples, while the NNK + DHM samples were to the left of the NNK samples, implying the opposite effects of DHM and NNK treatment, consistent with DHM’s efficacy in blocking NNK-induced lung carcinogenesis.

Figure 2.

The global views of the NNK effect (NNK versus control) and the DHM effect (NNK + DHM versus NNK) on mRNAs in A/J mouse lung tissues. (A) Principal component analysis (PCA) plot of the RNA-sequencing data that captures the similarity of the samples from Control (red), NNK (blue) and NNK + DHM (green) groups. Each dot represents an RNA-sequencing sample from an individual mouse projected onto a two-dimensional space. The variances explained by the first principal component (PC1) and the second principal component (PC2) are marked in the axis labels. (B) Heatmaps of the 984 genes that are differentially expressed for both (i) NNK versus Control; and (ii) NNK + DHM versus NNK with P < 0.05. The color key indicates the z-score of a gene expression value, with red color indicating higher gene expression value and blue color indicating lower gene expression value. The red, blue and green color on top of the heatmap indicate the Control, NNK and NNK + DHM groups, respectively. (C) Schematic interpretation of the Rank-Rank Hypergeometric Overlap (RRHO) plot: a hot spot in the top left quadrant indicates significant overlap in genes showing upregulation in NNK effect and countered by DHM effect (Type A genes); a hot spot in the bottom right quadrant indicates significant overlap in genes showing downregulation in NNK effect and countered by DHM effect (Type C genes). (D) RRHO plot showing the trend of overlap between genes showing opposite direction of the NNK effect and the DHM effect. Each pixel in the heatmap indicates the significance level of the overlap under a pair of cutoffs (the cutoff for the NNK effect is on x-axis, and the cutoff for the DHM effect is on the y-axis). The RRHO plot iterates all possible pairs of cutoffs, thus showing the global trend of overlap between two gene lists.

Global analyses of DEGs

To further evaluate the opposite effects of DHM and NNK, DEGs among the three treatments with a statistical threshold of P ≤ 0.05 for group differences were visualized. Figure 2B shows the heatmap of individual mouse lung levels of 984 DEGs that met the cutoff from the three treatment groups. As schematically illustrated in Figure 2C, Type A genes were upregulated by NNK but were prevented or restrained from such upregulation by DHM pretreatment. Conversely, Type C genes were downregulated by NNK, but were prevented from such downregulation by DHM pretreatment. Genes in these two categories would be the most meaningful to account for DHM preventive efficacy against NNK carcinogenesis. They were used for IPA analyses to identify potential canonical pathways significantly altered by NNK and normalized by prophylaxis with DHM gavage.

A small fraction of the DEGs showed the same direction of change between the NNK effects and the DMH effects (Figure 2B). Type B genes were upregulated by NNK and either was not suppressed by DHM or even further upregulated. Type D genes were downregulated by NNK and either was not modified by DHM or even further suppressed. As genes in these two categories further exacerbated the transcriptional effects of NNK, they would be counter-balancing or at best irrelevant to the DHM efficacy. Overall, the genes in these two categories account for a small portion of the total DEGs.

With a more stringent statistical cutoff of q-value ≤ 0.05, 328 DEGs were differentially expressed for both (i) NNK versus control and (ii) NNK + DHM versus NNK. Similar to the larger DEG dataset from P = 0.05 cut off, 105 of the 328 DEGs could be considered Type A genes and 193 Type C genes. Only less than 10% of the DEGs maintained the same direction of changes between the NNK effect and the DMH effect (6 Type B genes and 24 Type D genes). The odds ratio of showing opposite effect size direction compared to showing the same effect size direction is 140.73 with a P = 8.8 × 10⁻⁵⁵, strongly arguing that DHM pretreatment overall neutralized the signaling perturbations induced by NNK. The complete lists of DEGs with q ≤ 0.05 for NNK versus Control and for NNK + DHM versus NNK are summarized in Supplementary Tables S3 and S4, available at Carcinogenesis Online, respectively.

To further globally examine the potential opposite effect of NNK and DHM, we adopted the rank-rank hypergeometric overlap test as a threshold-free approach to examine the extent of overlap between the two sets of DEGs (Figure 2D). With hot spots on the top-left quadrant and the bottom-right quadrant of the heatmap, the rank-rank hypergeometric overlap results clearly indicate a significant trend of overlap genes with opposite effect size direction between NNK and DHM, again supporting the effect of DHM to neutralize the global transcriptomic effects induced by NNK.

Gene validation by qRT-PCR

Eight genes (COX-2, ADRB2, SAA3, FKBP5, PDK4, ATF3, HMGCS and SEMA7A) were analyzed by qRT-PCR to validate the RNA seq results. These eight genes were selected because their RNA seq results have varied levels of statistical significance among the three treatment groups. As shown in Figure 3, there was complete concordance between the RNA seq quantification and the qRT-PCR quantification, further substantiating the quality of the RNA seq results.

Figure 3.

Validation of RNA seq expression outcome with qRT-PCR. The levels of mRNA of eight genes were quantified through RNA seq and qRT-PCR among the three different treatment groups (control, NNK and NNK + DHM). One-way analysis of variance (ANOVA) was first used to compare means among NNK and the other groups. Dunnett’s test was then used for comparisons of NNK and the other treatment groups when ANOVA analysis was statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

The effects of prophylactic dosing of DHM on the main signaling pathways perturbed by NNK in A/J mouse lung

To explore and identify the signaling pathway candidate(s) that might have contributed to NNK-induced lung carcinogenesis, types A plus C genes (P ≤ 0.05) were subjected to the IPA pathway analysis. Using z-score cutoff of ≥1 (activation) or ≤−1 (inhibition) and –log(P) cutoff of 2 (P ≤ 0.01, i.e. statistical assessment of probability of DEGs involvement in a given pathway versus random occurrence), Table 1 lists the top canonical pathways ranked by −log (P). Thirteen pathways were NNK activated (positive z-score), but their extent of activation was effectively inhibited or normalized by DHM prophylaxis. Conversely, DHM prophylactic dosing normalized the suppression of 10 pathways by NNK (negative z-score).

Table 1.

The top canonical pathways perturbed by NNK based on Ingenuity Pathway Analysis, which were restored upon DHM treatment

Ingenuity canonical pathway	NNK versus Control
	−log (P-value)	Ratioa	z-scoreb
Acute phase response signaling	4.46	0.097	2.673
Protein kinase a signaling	4.16	0.071	1.633
Colorectal cancer metastasis signaling	3.24	0.074	1.213
Coagulation system	3.08	0.171	1.633
Endocannabinoid cancer inhibition pathway	3.05	0.090	2.309
IL-17 signaling	2.49	0.075	1.604
Retinol biosynthesis	2.45	0.130	2.236
PTEN signaling	2.44	0.080	1.897
ERK/MAPK signaling	2.36	0.070	1.387
p38 MAPK signaling	2.30	0.085	1.897
ERK5 signaling	2.06	0.097	2.449
CXCR4 signaling	2.04	0.071	1.000
CD27 signaling in lymphocytes	2.00	0.105	1.342
Pulmonary fibrosis idiopathic signaling pathway	5.51	0.086	−1.890
Factors promoting cardiogenesis in vertebrates	4.48	0.106	−1.500
Aryl hydrocarbon receptor signaling	4.21	0.101	−1.387
GP6 signaling pathway	4.18	0.110	−2.138
Heparan sulfate biosynthesis	3.99	0.128	−1.000
HOTAIR regulatory pathway	3.55	0.092	−1.069
Phagosome formation	2.94	0.055	−1.298
Apelin liver signaling pathway	2.88	0.192	−1.342
VDR/RXR activation	2.43	0.103	−1.134
Synaptogenesis signaling pathway	2.15	0.061	−1.698

Ratio: a ratio of the number of differentially expressed genes that map to a pathway divided by the total number of genes that map to the same pathway.

z-score: the activation z-score is used to infer activation states of a canonical pathway compared to a model with random regulation directions. A positive z-score indicates a pathway is activated; a negative z-score indicates a pathway is inhibited.

In the activated pathways, acute phase response signaling was the top candidate, both by z-score and statistical probability. This was not a surprise since NNK and its metabolites would have caused acute damage (2 h exposure) to DNA, proteins, and cellular organelles in the lung tissue, eliciting various stress responses. As such, ERK/MAPK, P38MAPK and ERK5 signaling activations were the most logical due to stress responses, so were inflammation, immune and cytokine signaling activations (IL-17 signaling, CXCR4 signaling and CD27 signaling in lymphocytes). Along with the above ERK/MAPK signaling also likely contributing to cell proliferation, the activation of PKA signaling as well as colorectal cancer metastasis signaling plus Pten signaling (inactivation of AKT survival) and endocannabinoid cancer inhibition pathway could reflect a competition of signals in NNK-induced death from chemical damage versus oncogenic proliferation.

Among the NNK-repressed pathways, pulmonary fibrosis idiopathic signaling, GP-6 signaling, long non-coding RNA HOTAIR regulatory pathway as well as Apelin liver signaling would be related to disarming lung damage repair and healing (collagens) in response to NNK exposure and DHM pretreatment was able to prevent their suppression. Interruption by NNK of factors promoting cardiogenesis in vertebrate and synaptogenesis signaling pathway probably reflected an impaired innervation between nerve and neuroendocrine cells with the lung cells (pneumocytes). Downregulation of aryl hydrocarbon receptor signaling and VDR/RXR activation by NNK probably reflected a compromise on their usual role in cellular protection through phase II metabolism (e.g. GSTA5, GSTM6, MGST2 and NQO1) and cell cycle arrest (CDKN1A/P21Cip1, CDKN2A/P16Ink4a) to reduce and repair damages. Some of the phase II enzymes may modulate the detoxification of NNK or NNAL.

The PKA signaling as a potential mechanistic lead

Among the 13 NNK activated signaling pathways, the PKA signaling pathway is of particular promise as a mechanistic lead to the DNA damage-independent signaling impacted by DHM for its prophylactic efficacy on NNK-lung carcinogenesis. IPA analysis also predicted that the mRNA changes caused by NNK were similar to those changes caused by forskolin, which is the gold standard chemical that activates the PKA signaling pathway (Table 2, top No. 4). In addition, there is ample literature supporting PKA activation as a logical mechanistic lead in NNK-induced lung carcinogenesis and DHM intervention. Specifically, the PKA pathway is typically activated by stress hormones, such as norepinephrine, through binding to and activating the stress hormone receptors, particularly the β-adrenergic receptor (β-AR). At the same time, NNK has been reported as an agonist for β-AR (44). PKA activation has also been reported to induce anxiety in several animal models (45–48). Kava, from which DHM originates, is well-known for its anxiolytic (anti-anxiety) property (49). In addition, probably most importantly, PKA activation has been demonstrated to promote primary tumorigenesis (50–52), including lung cancer. Indeed, the blood levels of PKA were found to be elevated in patients with cancers (53–57). We, therefore, focused on characterizing the PKA pathway as a mechanistic lead next.

Table 2.

The top 10 predicted upstream regulator/ligand, mimicked by NNK, based on Ingenuity Pathway Analysis, which were neutralized by DHM pretreatment

Upstream regulators	NNK versus Control
	Molecule type	P-value	Activation score
Lipopolysaccharide	Chemical drug (bacterial toxin)	7E-41	5.54
IL1B	Cytokine (inflammation)	2.08E-27	5.237
Dexamethasone	Chemical drug (stress hormone)	2.02E-43	5.098
Forskolin	Chemical drug (PKA activator)	4.06E-24	4.416
Methylprednisolone	Chemical drug (stress hormone)	9.06E-14	4.391
PDGF BB	Protein Complex (RTK ligand protein)	8.67E-24	4.228
Prexasertib	Chemical drug (Chk1 inhibitor, apoptogenic)	1.29E-10	4.123
IL6	Cytokine (inflammation)	9.78E-23	3.983
Poly rI:rC-RNA	Biologic drug (viral infection)	2.34E-11	3.863
GnRH-A	Chemical reagent (RTK ligand protein)	1.88E-08	3.805

The classical signaling process of the PKA pathway has been well established: upon stress hormones binding to and activating β-AR, cAMP is generated. cAMP then binds to the regulatory subunit of PKA and activates protein kinase cAMP-activated catalytic subunit alpha (PRKACA, the catalytic subunit of PKA). One dominant downstream event of activated PRKACA is to phosphorylate CREB and induce pCREB-mediated transcription. This would result in the activation of a range of oncogenic pathways involved in cancer initiation, transformation and progression, including the upregulation of COX-2. Indeed, Enricher analyses (58) of the DEGs modulated by NNK in the RNA seq data (P < 0.05) predicted CREB as the top transcription factor to be activated by NNK treatment. In addition, both β-AR and COX-2 have been upregulated in the A/J mouse lung tissues upon NNK exposure and suppressed by DHM treatment (Figure 3). These data overall support the PKA activation in A/J mouse lung tissues upon NNK treatment while DHM pretreatment suppressed such an activation.

We next characterized the phosphorylation status of CREB in the A/J mouse lung tissues upon NNK treatment with/without DHM pretreatment. As expected, NNK treatment resulted in increased pCREB while DHM pretreatment for 8 h significantly reduced such an increase (Figure 4A). We then evaluated whether NNAL, the major in vivo metabolite of NNK, could induce CREB phosphorylation in H1299 cells, a non-small cell lung adenocarcinoma in vitro model. At a physiologically relevant concentration (10 nM) that did not increase DNA adducts (data not shown), NNAL rapidly induced CREB phosphorylation (Figure 4B). DHM pretreatment for 1 h effectively and concentration-dependently suppressed NNAL-induced CREB phosphorylation (Figure 4C), consistent with the in vivo results (Figure 4A). In addition, the effect of NNAL on CREB phosphorylation could be effectively blocked by propranolol (a β-AR antagonist) and H89 (a PKA inhibitor) but not by yohimbine (an α-AR antagonist) (Figure 4D), substantiating that NNAL-induced CREB phosphorylation is mediated through both β-AR and PKA. To further confirm that the DHM effect is also mediated through blocking β-AR, a classical β-AR agonist (ISO) was used to induce CREB phosphorylation in H1299 cells, which was blocked by DHM pretreatment in a concentration-dependent manner (Figure 4E). These in vitro and in vivo data overall provided additional and cohesive support for the activation of the PKA pathway as a DNA damage-independent mechanistic lead in NNK-induced lung carcinogenesis, which was effectively suppressed by DHM to contribute to its outstanding chemopreventive efficacy in blocking NNK-induced lung carcinogenesis.

Figure 4.

Experimental data supporting the protein kinase A pathway activation as a mechanistic lead. (A) NNK treatment (2 h) resulted in the phosphorylation of CREB in the lung tissues in A/J mice while DHM pretreatment (8 h prior to NNK) suppressed such an effect (n = 3–6). (B) Exposure to NNAL, the major in vivo metabolite of NNK, resulted in a time-dependent phosphorylation of CREB in H1299 cells. (C) DHM pretreatment for 1 h prevented NNAL-induced CREB phosphorylation in H1299 cells in a concentration-dependent manner. (D) The effects of propranolol (β-AR antagonist), yohimbine (α-AR antagonist) and H89 (PKA inhibitor) on NNAL-induced CREB phosphorylation in H1299 cells. (E) The concentration-dependent blocking effects of DHM on Isoproterenol (ISO)-induced CREB phosphorylation in H1299 cells. For (B–E), at least three biological repeats were performed. The data were quantified via Image J as detailed in the experimental section. One-way analysis of variance (ANOVA) was first used to compare means among NNK/NNAL and the other groups. Dunnett’s test was then used for comparisons of NNK/NNAL and the other treatment groups when ANOVA analysis was statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Discussion

Our RNA seq results strongly imply that β-AR-mediated NNK activation of the PKA signaling pathway in A/J mouse lung tissue could collectively promote lung carcinogenesis initiated by DNA damage-driven mechanisms, whereas DHM repressed such activation. The RNA seq data (Table 1) also suggest that DHM may suppress NNK-induced DNA adducts by enhancing phase II detoxification of NNAL through glucuronidation and potentially glutathionylation.

Two recent publications with respect to profiling NNK-induced changes in the mouse lung carcinogenesis with “omics” approaches are noteworthy. Hudikar et al. reported metabolomic, DNA CpG methylomic and RNA seq transcriptomic analyses of A/J mice lung tissues at 2 and 4 weeks after an NNK i.p. injection exposure versus vehicle-treated control mice (59). Jin et al. reported DNA CpG methylomic profiling of mouse lung tissue after twice weekly intratracheal administration of NNK for 8 weeks (60). These studies therefore likely inform lung tissue changes weeks following the initial NNK events. Our study complements these papers by highlighting the acute (2 h) transcriptional changes following NNK exposure. Future detailed characterization of the mechanisms involved in both DNA damage-driven and -independent pathways in NNK-induced lung carcinogenesis and DHM-based chemoprevention by systems biology approaches, such as transcriptomics, methylomics, metabolomics, proteomics as well as phosphoproteomics, are warranted.

Limitations

The RNA seq profiling results from this study provided only a snap shot of the mechanistic leads into DHM prophylactic dosing for 8 h before NNK exposure for 2 h (acute transcriptional impacts). The lack of a DHM-only control group and the lack of temporal dynamic sampling of DHM induced transcriptome landscape changes limited our ability to seek rigorously causal connections. Future time-course studies are warranted to address these mechanistic gaps to provide a more comprehensive understanding of the DNA damage-driven and -independent mechanisms of DHM to prevent NNK-induced lung carcinogenesis. Such studies could guide the rational implementation of prophylaxis with DHM or kava for lung cancer risk reduction.

Glossary

Abbreviations

β-AR

β-adrenergic receptor

DEGs

differentially expressed genes

DHM

dihydromethysticin

IPA

Ingenuity Pathway Analysis

NNK

nicotine-derived nitrosamine ketone

PKA

protein kinase A

q-RT-PCR

quantitative real-time PCR

Funding

The research reported in this publication was financially supported in part by the grants R01 CA193278 (C.X.) from National Institutes of Health, Lung Cancer Research Foundation Research Grant on Disparities in Lung Cancer (C.X. and Z.H.), Frank Duckworth Endowment College of Pharmacy University of Florida (C.X.) and Startup Fund University of Florida Health Cancer Center (C.X.). The content is solely the responsibility of the authors and does not necessarily represent the official views of any funding agencies.

Conflict of interest statement

No potential conflicts of interest were disclosed by authors.