Differential effects of selective frankincense (Ru Xiang) essential oil versus non-selective sandalwood (Tan Xiang) essential oil on cultured bladder cancer cells: a microarray and bioinformatics study

Cell culture medium (MEM and DMEM/F-12 (1:1)), fetal bovine serum (FBS), MEM vitamin solution, non-essential amino acids, epidermal growth factor (EGF), insulin-transferrin-sodium selenite (ITS) media supplement, sodium pyruvate, and penicillin-streptomycin were purchased from Life Technologies (Grand Island, NY, USA). Frankincense (B. carterii from Somalia) and sandalwood (S. album from Sri Lanka) essential oils were obtained from Young Living Essential Oils (Lehi, UT, USA) and prepared based on previously reported procedures [7]. An XTT cell proliferation assay kit was obtained from Roche Applied Science (Indianapolis, IN, USA). An RNeasy® Mini Kit was obtained from Qiagen (Valencia, CA, USA).

The essential oil components were identified by gas chromatography–mass spectrometry (GC-MS), using an Agilent 7890A GC system (Agilent Technologies, Santa Clara, CA, USA) equipped with an Agilent 5975C mass selective detector (Agilent Technologies). We used either an HP-1 50 m × 0.32 mm ID × 0.5 μm film column (Agilent Technologies) or a DB-WAX 60 m × 0.32 mm ID × 0.5 μm film column (Agilent Technologies) for sandalwood essential oil, and an Rxi-5 ms 60 m × 0.25 mm ID × 0.25 μm column (Restek, Bellefonte, PA, USA) for frankincense essential oil. Retention indices were calculated by performing injections of C7-C30 normal alkanes (Supelco, Bellefonte, PA, USA) to confirm the MS identification. The detailed procedures for chemical analysis were reported previously [24].

As natural products, chemical constituents of frankincense and sandalwood essential oils varied from batch to batch and from season to season. A batch of distillation was used to establish a baseline relationship between chemical compositions and molecular pathways activation throughout the study.

Human bladder cancer J82 cells, derived from a poorly differentiated, invasive stage 3 transitional cell carcinoma [25], were obtained from ATCC (HTB-1; Manassas, VA, USA). J82 cells were maintained in growth medium consisting of MEM supplemented with 10% FBS, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 2% MEM vitamin solution, and 100 units/mL penicillin-100 μg/mL streptomycin. The UROtsa cell line was isolated from a primary culture of normal human urothelium and immortalized with the SV40 large T antigen [26]. UROtsa cells were provided by Dr. Ricardo Saban (Department of Physiology, University of Oklahoma Health Sciences Center) and cultured in DMEM/F12 supplemented with 10 ng/mL EGF, 1 × ITS media supplement, and penicillin-streptomycin. Both cell lines were maintained in a humidified cell incubator at 37°C and 5% CO₂ and passaged every 3–4 days or when cells reached about 80% confluence.

Bladder cancer J82 cells and immortalized normal UROtsa cells were seeded in 96-well tissue culture plates at a density of 5 × 10³ cells/well in 100 μL of their growth media to determine cell viability following treatment with frankincense and sandalwood essential oils. Following overnight incubation and adherence, an additional aliquot of growth media (100 μL) or varying dilutions of frankincense (1,400–600) (v/v) or sandalwood (16,000–7,000) (v/v) essential oil in the growth media was added to each well in triplicate. Measurement of cell viability following exposure to essential oils was performed using the XTT assay at time 0 and 24 h. During the assay, the growth medium (100 μL) was removed from each well and the XTT labeling mixture (50 μL) was added to each well. The reactions were carried out at 37°C for 4 h and the absorbance of the reaction product was recorded at 450 nm by a μQuant microplate reader (Bio-Tek; Winooski, VT, USA). Absorbance values obtained at 24 h following treatment were normalized to the values obtained at time 0 to calculate fold changes in cell viability [6].

Total RNA was isolated from J82 cells for microarray analysis to determine frankincense and sandalwood essential oil-regulated molecular responses at the mRNA level. Briefly, 5 × 10⁵ J82 cells were seeded in 60-mm tissue culture plates, cultured overnight for adherence, and either left untreated or treated with a 1:1,100 dilution (v/v) of frankincense essential oil or a 1:11,000 (v/v) dilution of sandalwood essential oil in growth medium. Total RNA was isolated at 0 h (no treatment) and at 0.5, 1, 2, and 3 h after treatment using the RNeasy® Mini total RNA isolation kit. Total RNA concentration was determined using a nanodrop scanning spectrophotometer, and RNA was qualitatively assessed for degradation based on the ratio of 28S:18S rRNA using a capillary gel electrophoresis system (Agilent 2100 Bioanalyzer, Agilent Technologies).

Normalization of the microarray datasets was performed for each array by plotting a frequency histogram of the raw expression data for all genes [27, 28]. The histograms showed a right-skewed unimodal distribution curve with the mode around zero. A normal distribution curve representing the genes with low levels of expression was then fitted around the mode, mirroring the Gaussian profile of the left part of the histogram. Two parameters mean and standard deviation (SD), were defined for this distribution. The expression values of the rest of genes were then normalized to the SD of the normal distribution curve after subtraction of the mean, and log10-transformed. The arrays were then adjusted to each other by robust linear regression [29, 30]. Genes with expression value < 3.0 (or 0.477 on the log10 scale) were considered to be not expressed; this was equivalent to setting a threshold at three SD above the mean of low levels expression genes.

We defined genes with statistically significant expression levels across the time points when comparing to a reference group [31] as HV genes. The reference group consisted of genes expressed above background levels but with low variability as determined by an F-test, and represented the technical variability of microarray data [27, 32]. Genes whose expression levels varied significantly (P < 1/N, where N is the number of genes expressed above the background) were identified by comparing an individual gene’s variability to that of the reference group using an F-test [27]. The threshold level at P < 1/N is a modification of the Bonferroni correction for multiple hypothesis tests. These genes were further filtered to remove those with variability arising from experimental errors by comparing the variability of the residuals in the replicated group of samples with the same variability obtained after excluding the maximum and minimum, one at a time, using a modification of the ‘leave one out’ method [33]. A statistically significant decrease in variability after excluding one replicate provided evidence of possible error in that particular replicate. Once filtered, the remaining genes were denoted as HV genes [34], and were considered to be a snapshot of the dynamic biological responses following treatment with frankincense and sandalwood essential oils. The HV genes were mean-centered and clustered by hierarchical clustering using the ‘Correlation (uncentered)’ similarity metric. All statistical analyses and clustering were performed in Matlab (Natick, MA, USA).

Significantly over-represented gene ontologies for the identified HV genes were analyzed using the Database for Annotation, Visualization and Integrated Discovery (DAVID, http://​david.​abcc.​ncifcrf.​gov/​) [35]. The genes of interest were uploaded as Illumina Probe IDs to DAVID; and the Illumina HUMANREF-8_V3_0_R1_11282963_A array was selected as a background. Functional annotation clustering was performed with default settings and medium stringency. Statistically significant over-representation of gene ontologies was reflected by the “enrichment score”; details of its calculation can be found on the DAVID website. Gene lists were analyzed for over-represented canonical pathways using Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA). Genes of interest were bound into a network using the “path explorer” tool, and the network was edited using the “path designer” tool in IPA.

Total RNA samples were isolated from J82 cells treated with frankincense and sandalwood essential oils for RT-PCR. For semi-quantitative analysis, first-strand cDNA was synthesized from 2.5 μg of the total RNA with oligo (dT)_12–18 primers (1 μg) (Life Technologies) and MMLV reverse transcriptase (200 U) (Life Technologies) in a total volume of 50 μL; the reaction was performed at 42°C for 2 h. Target mRNA species were amplified using a three-temperature cycle protocol: 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min. PCR amplified products were separated on 1.5% agarose gels (Lonza, Walkersville, MD, USA) and stained with 0.5 μg/mL ethidium bromide (Life Technologies). Ethidium bromide-stained images were captured using a Gel Doc 1000 imaging system equipped with Quantity One® image analysis software (Bio-Rad Laboratories; Hercules, CA, USA).

SYBR Green-based quantitative PCR was used for quantifying the expression levels of target genes that are either commonly or differentially regulated by frankincense and sandalwood essential oils. Isolated total RNA was subjected to RNase-free DNase I (Qiagen, Valencia, CA, USA) (1 unit) digestion. PCR was performed using the above described three-temperature cycle with 30 s for each temperature reaction using a Bio-Rad iCycler real-time PCR detection system (Bio-Rad Laboratories) Melt-curve analysis was conducted after the final cycle to ensure the amplification of single species of PCR products in each sample; and agarose gel electrophoresis was used to confirm the presence of a single PCR band with the expected size in each reaction. We used β-actin as a reference gene to correct for sample-to-sample variations. Relative changes in JUN, DUSP1, PMAIP1, and ZNF311 expression after frankincense and sandalwood essential oils treatments were calculated based on Livak et al.[36].

The results were expressed as means (SD) from at least 3 repeats. Comparisons of cell viability following frankincense oil treatment were made using one-way analysis of variance followed by a post hoc Dunnett's test. The Student’s t-test was used to compare levels of target gene expression between frankincense and sandalwood essential oil-treated J82 cells. P < 0.05 was considered statistically significant.

Frankincense and sandalwood essential oils were characterized by polarimetry and GC-MS. One of the best chemical signatures for Boswellia carterii is alpha-pinene ranging between 33.9% and 62.5% with a distinctive negative optical rotation (−13.3° ± 4.9°) as we previously reported [24].

Sandalwood essential oil (n = 23) exhibited unique chemical compounds cis-alpha-santalol and cis-beta-santalol at high concentrations, 41.1–47.9% and 15.9–21.3%, respectively. GC-MS with nonpolar HP-1 and polar DB-WAX qualified the purity of sandalwood essential oil as the method separated these two chemical components from similar compounds [37]. It is extremely difficult to establish standards for natural products. Even though standard operating procedures are established for harvest, storage, and processing, variable including weather, soil condition, and water quality all have significant impacts on the chemical compositions of all herbal products (i.e., essential oils in this case). Therefore, this industry and scholarly publications present the chemical compositions of essential oils using a percentage range. However, with the introduction of analytical chemistry and molecular biology, we would like to relate and translate such descriptions of chemical compositions into descriptions of molecular responses, starting with the present article. Any deviation in molecular responses from the current observations will be related to changes in chemical compositions in the future. We hope that we will be able to create a “standard” for medical applications of essential oils or for all herbal products in the future.

Both frankincense and sandalwood essential oils suppressed the viability of human bladder cancer J82 cells but with different morphologies (Figure 1A). At high concentrations of both essential oils (low dilution factors), no viable cells were detected, while frankincense and sandalwood essential oils significantly suppressed J82 cell viability at 1:1,100 (P = 0.008) and 1:11,000 (P = 0.012) dilutions (v/v), respectively (Figure 1B).

Based on the results of cell viability assays, J82 cells were more sensitive than UROtsa cells to frankincense essential oil-suppressed cell viability, similar to the findings in our previous study [6]; the IC₅₀ values were 1:1,250 and 1:600 dilutions (v/v) for J82 cells and UROtsa cells, respectively. In contrast, both J82 and UROtsa cells responded similarly to sandalwood essential oil treatment, with an IC₅₀ value around 11,000 dilutions (v/v) for both cell lines (Figure 2).

Temporal changes in gene expression patterns were analyzed in total RNA samples isolated from untreated J82 cells and cells that received either frankincense or sandalwood essential oil for between 0.5 and 3 h. Full microarray data have been deposited in Gene Expression Omnibus (GEO) with the accession number GSE53171, and are accessible on the GEO web-site (http://​www.​ncbi.​nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GSE53171).

A total of 139 genes were identified to be HV genes in frankincense essential oil-treated cells, whereas 206 HV genes were identified to be regulated by sandalwood essential oil (Additional file 1: Table S1). Among the HV genes, 51 genes were commonly regulated by both essential oils with a gradual increase in their expression levels over the experimental period (Figure 3A). Although the remaining HV genes also showed a gradual increase following stimulation with both essential oils, these genes responded differently between frankincense and sandalwood essential oils. A total of 88 HV genes were specifically regulated by frankincense essential oil, but their expression levels remained relatively constant following sandalwood essential oil treatment (Figure 3B). Another 155 HV genes were classified as sandalwood essential oil-responsive genes (Figure 3C).

The 139 and 206 HV genes regulated by frankincense and sandalwood essential oils, respectively, were subjected to DAVID analysis with Gene Ontologies (GO) of the over-represented HV genes. Four groups of genes were categorized into the GO terms of transcription factor activity, histone genes, negative regulation of biological process, and apoptosis (Table 1). Transcription factor activity was the most significantly over-represented class of genes regulated in frankincense and sandalwood essential oil-treated J82 cells. The expression of some transcription factors, such as ATF3, DDIT3, EGR1, FOSB, JUN, JUNB, and MYC, was modulated by both essential oils. Other transcription factors such as FOS (Finkel–Biskis–Jinkins murine osteogenic sarcoma virus) were specifically modulated by frankincense essential oil, whereas sandalwood essential oil modulated the expression of several inhibitors of DNA binding (ID1, ID2, and ID3), along with members of the zinc finger family. In addition, all identified histone core genes, except for HIST1H3D, were commonly modulated by both essential oils.

Negative regulation of biological processes categorized by IPA was more pronounced in sandalwood essential oil-treated J82 cells. Among the most notable HV genes, growth arrest genes (GADD45A, GADD45B, and PPP1R15A) and pro-apoptotic genes (CASP9 and ING5) were specifically up-regulated by sandalwood essential oil. As part of the negative regulation of biological processes, several pro-inflammatory interleukins (IL1A, IL6, and IL8) were up-regulated by both essential oils.

The 51 genes commonly regulated by both essential oils were excluded to identify over-represented genes specifically modulated by either frankincense or sandalwood essential oil in J82 cells. The remaining 88 and 155 HV genes specifically modulated by frankincense and sandalwood essential oils, respectively (Figure 3B and C), were analyzed by DAVID. Heat shock proteins (HSP), including three members of DNAJ family (HSP40 homolog), were the main over-represented call of HV genes following frankincense essential oil treatment (Table 2). In addition to the histone core family genes (Table 1), three other members of this family were significantly over-represented among the HV genes identified in frankincense essential oil-treated J82 cells.

Transcriptional regulation was identified as the most over-represented class of genes that were specifically modulated by sandalwood essential oil in J82 cells. Among these transcription factors, members of the zinc finger family of proteins were predominantly identified (Table 2). In addition, cell death-related genes were over-represented in sandalwood essential oil-treated cells, specifically epithelial membrane protein 1, immediate early response 3, and myeloid-associated differentiation marker.

The HV genes modulated by frankincense and sandalwood essential oil treatments in J82 cells were analyzed by IPA to identify canonical pathways through connecting both essential oil-modulated genes in a single network (Figure 4). Consistently, cell death genes (67/84 genes, P = 8.27E-33) and apoptotic genes (62/84 genes, P =1.69E32) represented significantly enriched ontological categories.

Pathways including the p38 MAPK, p53, IL6, and HMGB1 signaling pathways were commonly represented by HV genes modulated by both essential oils (Table 3). IPA also identified canonical pathways that were specific for either frankincense or sandalwood essential oil treatment. For example, aryl hydrocarbon receptor (AhR) signaling and NRF2-mediated oxidative stress response genes were the two main canonical pathways over-represented by the HV genes regulated by frankincense essential oil. In contrast, ATM signaling was uniquely over-represented by the HV genes identified in J82 cells treated with sandalwood essential oil.

Microarray and RT-PCR analysis provided consistent results for these genes. Although both JUN and DUSP1 were up-regulated by both essential oils in J82 cells (Figure 5A), the levels of DUSP1 were significantly elevated in frankincense essential oil-treated cells (P = 0.011), whereas there was no statistical significance for JUN (P = 0.282) at 2 h following treatment (Figure 5B). DNAJB4 was up-regulated by frankincense essential oil (P = 0.043), whereas PMAIP1 (P = 0.032) and ZNF311 (P = 0.035) transcripts were specifically up-regulated in sandalwood essential oil-treated J82 cells at 2 h after stimulation.