Regulation of gene expression in ovarian cancer cells by luteinizing hormone receptor expression and activation

The parent SKOV-3 ovarian cancer cell line was chosen as a control in this study since it does not express LHR [14, 19, 20], and, following transfection, the LHR+ cells serve to determine the alterations in gene expression elicited by LH. The LHR+ cells bound [¹²⁵I]-human chorionic gonadotropin with a K_d of 0.3 nM (human chorionic gonadotropin and LH utilize the same G protein-coupled receptor, LHR), consistent with the binding affinity using ovarian reproductive cells, and responded to LH with increased intracellular levels of cAMP and inositol phosphates. In total, six groups of SKOV-3 cells (LHR-, LHR+, and LHR+ incubated with LH for various times: 1, 4, 8, and 20 h), each with three independent replicates, were used for examining the cell response. These times were chosen to provide temporal information on the early, intermediate, and later response genes altered by LH-mediated LHR activation.

Total RNA was extracted from the above 18 SKOV-3 samples [20] and was amplified using the NuGEN™ Ovation™ RNA Amplification System V2. The resultant fragmented and labeled cDNA was added to the hybridization cocktail in accordance with the NuGEN guidelines for hybridization onto Affymetrix human genome U133 Plus2 Arrays. Following hybridization for 18 h at 45°C, the array was washed and stained on the GeneChip^® Fluidics Station 450 using the appropriate fluidics script, before being inserted into the Affymetrix autoloader carousel and scanned using the GeneChip^® Scanner 3000. The microarray analyses were done by Almac Diagnostics, Durham, NC. The raw data has been deposited to GEO database (Accession ID: GSE27328).

Poly(A)+ RNA was extracted from the cells and equivalent amounts were converted to cDNA, which was then analyzed by qRT-PCR. 23 genes are tested, which are mostly associated with cell growth and invasion (Additional file 1 Table S1). The amounts of cDNA for each gene were determined in duplicate by qRT-PCR with the SYBR Green detection system, and the relative gene expression was calculated from the Ct values, where Ct is the cycle at which the threshold (i.e. the number of cycles at which the earliest measurable fluorescence signal reaches 25X baseline) can be detected in a qRT-PCR assay [21]. The relative gene expression is given as a ratio of Ct of the gene of interest to that of the housekeeping gene, GAPDH, taken as a reference gene [GAPDH-287F: GAAATCCCATCACCATCTTCC

AG; GAPDH-599R: CTTTGGTATCGTGGAAGGACTCAT] and distilled water as a negative control,

QC was performed for each hybridized array by assessing quality metrics comprehensively, and hierarchical clustering and principal components analysis were employed for data QC assessment by using 26,821 transcripts that passed the background filter, i.e., three times the standard deviation of the average background intensity of the 18 samples, as shown in Additional file 2 Fig. S1. The results of clustering and data reduction were assessed comprehensively to ascertain the suitability of the results for further analysis.

Subsequently, statistical analyses were performed to identify the differentially expressed genes between any two groups, especially at the transition points when LHR and LH are introduced. The ANOVA [22] and Mann-Whitney tests were initially applied, and the geometric mean of gene expression was calculated within the triple duplicates. Given the different tests underlying the individual significance, differential expression was assessed by applying p-value < 0.05 (restraining FDR < 0.1) and fold-change ≥ 2.0. More rigorously, we only focused on those expression changes consistently observed at the transition points, which means the expression levels of the triplicate measures of group A are all higher (or lower) than those of group B. Overall, the experimental design, coupled with the statistical significance and fold change criteria employed, engender high confidence of selecting reliable differential expressions. Both hierarchical clustering [23] and self organization maps (SOMs) [24] were applied to extract co-expression patterns associated with LHR expression and LH-mediated activation, especially to identify the significant functional clusters among the profiles. Enrichment analyses on functional families and pathways have been carried out according to Gene Ontology (GO) and KEGG curation [25], respectively.

Public normal HOSE expression dataset (GSE14001) [26] was downloaded from the GEO database for comparative purposes, which were collected using the identical Affymetrix platform. The microarray analysis was done on RNA obtained from short-term cultures of three different normal human ovarian surface epithelial cells that were initiated from the surface scraping of normal ovaries [26]. The same RMA algorithm was applied for gene expression summarization; no further normalization was conducted between different cell types to retain the variance of overall mRNA expression.

A total of 54,671 transcripts were originally profiled, among which 2,373 genes exhibited at least 2-fold differential expression between any two experimental groups (see detailed statistics in Additional file 1 Table S2), including 1,783 genes differentially expressed in LH-treated cells. Out of the 23 differentially expressed genes analyzed by qRT-PCR in this study and earlier [20], we found that 22 genes exhibit consistent expression pattern between microarray and qRT-PCR data (Additional file 1 Table S1), which indicate that majority differential information derived from microarray is reliable. According to the IPA annotation [27], 689 differential genes are cancer-related, and 265 genes are highly expressed in the ovary (see Additional file 1 Table S3). Five major functional families were found to be significantly enriched by the differentially expressed genes, including growth factors, translation regulators, transporters, G-protein coupled receptors, and ligand-dependent nuclear receptors (Figure 1). Generally, these differentially expressed genes participate in pathways involved in the cell cycle, focal adhesion, cytokine-cytokine receptor interaction, regulation of the actin cytoskeleton, purine metabolism, and a number of key signaling pathways such as MAPK, TGF-β, p53, and Jak-STAT.

The 2,373 genes were subject to hierarchical clustering [23] for identification of distinct gene-expression patterns across all sample groups. In Figure 2A, different expression patterns were observed across the five different transitions, i.e. LHR-/LHR+ and LHR+/LHR+ plus LH for each of the four time points, clearly supporting the hypothesis that LHR expression and LH-mediated receptor activation impose significant effects on gene expression in ovarian cancer cells.

In total, 12 highly correlated expression patterns were identified from the differentially expressed genes, by using a self organization map (SOM) [24] (Table 1). The gene lists of each cluster are given in Additional file 1 Table S4. Seven clusters (C1-C7) represent the up-regulated genes, while the other five show down-regulated genes concomitant with LHR expression in the cells. After incubation of LH with the LHR+ cells, the genes of each group became more diverse in terms of their expression level changes, either being up-regulated, down-regulated, or unchanged, falling into five categories. Within each cluster, enriched GO and pathways were identified (Table 1), reflecting the major involved functional groups or cellular processes, as discussed in the next two sections.

Of the 414 genes that were differentially expressed significantly when LHR was introduced, 144 were up-regulated and 270 down-regulated. A few pathways, including gap junction, purine metabolism, calcium signaling, and actin cytoskeleton regulation, are associated with the up-regulated genes, perhaps indicating a moderate activation of these processes. Since one of the objectives is to examine the regulation of the genes that may promote or inhibit tumor growth, migration, and invasiveness in LHR+ tumors, the up-regulated TUBAL3, TUBB2B, and GUCY1B3 genes involved in gap junction formation and function may indicate a reduced tumor progression and metastasis [28]. Opposing these increased expressions, LHR+ cells exhibit significant down-regulation of genes associated with cellular processes such as cell communication, ECM-receptor interaction, regulation of vesicle fusion, and focal adhesion, for example genes encoding extracellular matrix structural constituents (KRT7, DSC3, KRT16, TNC, LAMB3), collagens (COL3A1, COL6A3, COL4A1, COL1A2), and matrix metalloproteinases (MMP1, MMP2, MMP13, MMP14). The negative effect on cell communication and ECM interaction is consistent with a reduced invasive activity of the cancer cells, thus inhibiting cancer progression. In addition, other down-regulated genes are found relevant to apoptosis (PIK3R3, IL1R1, FAS, TNFSF10) and major signaling pathways (P53, TGF-β, ERBB HER-2).

A total of 1,783 genes were found to be regulated by LH, when compared to gene expression in LHR+ cells (including all clusters in Table 1 except for C4 and C10). The up-regulated genes are enriched in 21 pathways (Table 2), including VEGF signaling, gap junction, and immune responses (the Toll-like receptor signaling and the B cell receptor signaling pathways). Besides the immune responses that are generally observed in most types of cancers, the activation of genes involved in VEGF signaling may be hypoxia responsive and introduce a positive effect on cancer growth, while those involved in gap junction and Notch signaling accelerate cell-cell communication and influence several key aspects of the normal development by regulating differentiation, proliferation, and apoptosis [29].

The genes that are down-regulated by LH represent 22 pathways (Table 2). Besides the continuous inhibition of the cell cycle, p53 signaling, and the complement and coagulation cascades, LH also seems to impose a negative and delayed effect on a few metabolic pathways related to pyrimidine, glycerolipid, methionine, androgen, and estrogen metabolism. These results indicate an LH-mediated reduction in certain aspects of nucleic acid, lipid, and amino acid metabolism. Since the epithelial cells are not steroidogenic, the down-regulation of androgen and estrogen pathways may relate more to sex steroid action [30, 31].

Table 2 shows 34 additional pathways consisting of both up- and down-regulated genes to different extents, among which the overall effects on tumor growth and apoptosis cannot be evidently inferred. For a few, one particularly interesting observation is the substantially increased expression of the tumor necrosis factor member 10 gene (TNFSF10), involved in natural killer cell-mediated cytotoxicity (Figure 3). TNFSF10 encodes the cytokine tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) that binds to TNF and induces apoptosis, primarily in tumor cells [32].

To exclude the possibility that some effects of LH on cell growth and apoptosis were masked by the extremely high levels of gene expression in SKOV-3 cells, the most highly-expressed genes (top 5%, ~3000 transcripts) in each group were examined and compared with those from normal HOSE cells. Figure 4 shows that the genes most highly expressed in SKOV-3 cells and normal HOSE cells are largely different, with only 1,726 out of the 3,000 transcripts in common. The 1,056 unique genes, specific in the LHR-SKOV-3 cells, are participating in the regulation of translation, cell division, chromosome partitioning, post-translational modifications, protein turnover, chaperones, and signal transduction mechanisms, indicating possible alterations of these processes in cancer compared to normal HOSE cells, such as an increase in the overall rate of protein synthesis and translational activation of the mRNA molecules involved in cell growth and proliferation. The majority (2,748) of the highly expressed genes in the LHR- SKOV-3 cells continue to be expressed in the LHR+ cells.

In contrast, the highly-expressed genes are quite consistent in the LHR+ cells incubated with LH, where 3,560 genes were involved. The 689 genes specifically introduced by LH-mediated LHR activation, compared to LHR+ cells, reveal the involved cellular processes such as coenzyme metabolism, post-translational modifications, nucleotide transport, DNA replication and repair, intracellular trafficking, and secretion.

In addition, the most differentially expressed genes were examined (Additional file 1 Table S5), and a few were found to be altered significantly by LHR expression, such as ERBB4 (↓46) and CASP1 (↓44). Down-regulation of ERBB4 is deemed to be beneficial as its over-expression may promote cell proliferation, while down-regulation of CASP1 may result in a suppressive effect on cell apoptosis [33]. Following LH activation of LHR, the most highly up-regulated genes, e.g., PDE4B, TNFSF10, FOSB, and the highly induced early and intermediate response genes, e.g., THBS1, CCl20, DUSP1, are found to occupy a gene network connecting transcriptional regulation, cell proliferation and differentiation, apoptosis, and multiple signaling transductions such as MAPK, Erk1/Erk2 MAPK, Jak-STAT, VEGF, and the TGF-β signaling pathway. Thus, from some of the results one could argue that LH may serve as a positive regulator on cancer growth and invasion through overexpression of CCl2 and FOSB. However, the large increase in the expression of TNFSF10 can act to increase apoptosis. The high level of up-regulation of PDE4B is interesting since the enzyme, a cyclic nucleotide phosphodiesterase up-regulated by cAMP [34], is responsible for inactivating cAMP and thus rendering the cells refractory to additional LH signaling for an extended time.

Ovarian carcinogenesis is a complicated process that involves the deregulation of multiple signaling pathways. In this study, proteins and signaling pathways involving Wnt signaling, p53 tumor suppressor, APC/β-catenin signaling, K-Ras concogene, and EGFR tyrosine kinase were found to be affected by LH activation of LHR, either positively or negatively. Of particular interest was the result demonstrating that LH dramatically activates the expression of the interleukin-6 gene (IL-6) (33↑), a pleiotropic cytokine that is assumed to be involved in ovarian carcinogenesis and may induce signaling pathways such as toll-like receptor, NOD-like receptor, cytosolic DNA-sensing, and Jak-STAT [35, 36], which, in this sense, may indicate a potential therapeutic target for treating ovarian cancer. Moreover, an increase in the production of potent growth factors like IL-8 may facilitate tumor growth and angiogenesis [37].

To sum up the major impact of LH, we have grouped all involved pathways (Additional file 1 Tables S6-7) according to their cellular functions (Table 3). A few of them were selected for a closer examination (cf. Discussion), which include the cell cycle, the MAPK pathway, apoptosis, the Jun and Fos family of transcription factors, and other signaling pathways.

A comparison of the gene expression profiles between normal OSE cell and SKOV-3 cells (including all LHR+ cancer cells) has the potential to identify a group of genes that can discriminate between normal and cancer cells regardless of LHR expression and LH action. Two lists of genes have been identified as up-regulated (185 genes) and down-regulated (248 genes) in all cancer versus normal cells, whose expression profiles are shown in Figure 5 (see names in Additional file 1 Table S9). Functional analysis reveals that the up-regulated genes are involved in cell communication, ECM-receptor interaction, and focal adhesion, especially functioning in cell division and chromosome partitioning, as well as carbohydrate transport and metabolism, which are fundamental processes for cancer growth. We have conducted the specificity analyses of the identified markers against public microarray gene-expression data for other human diseases (http://​bioinfosrv1.​bmb.​uga.​edu/​DMarker/​) and obtained 106 genes whose differential expression are specific to ovarian cancer. Among these genes, nine have been reported with the same expression changes in a newly-developed YDOV-157 cell line versus HOSE (Additional file 1 Table S9), which illustrate some consistency between different cell lines. These results engender confidence in proposing some genes as potential molecular markers to discriminate between ovarian epithelial carcinoma cells and normal OSE cells. Based on a recently developed approach from this laboratory [38], 103 of these genes (Additional file 1 Table S9) were predicted in which their protein products may be secreted into the bloodstream, thus providing another important pool of potential serum markers for further investigation. According to the proteomic reports from the Plasma Proteome Project (PPP) [39] and a literature search for diseased protein markers [38], we know that 22 of these proteins have been identified as secreted proteins in normal or diseased blood (Additional file 1, Table S9). While it is unlikely that just one marker would emerge with good specificity and sensitivity, combinations of two or more may prove highly useful. Some of the predicted proteins could be peptides/fragments derived from extracellular matrix proteins and membrane receptors, many are readily soluble and assayable, e.g. chemokine ligands 1, 5, 9, 10, 11, and 18, placental growth factor (PGF), and growth hormone secretagogue receptor ligand (GHRL), to mention but a few.

Our literature search against the Therapeutic Target Database (TTD) [14] found that 48 therapeutic targets were reported to be ovarian cancer-associated, including 18, 20, and 12 targets in three categories, successful, clinical trials and, research, respectively (detailed list in Additional file 1, Table S10). Our data cover 39 of the 48 therapeutic targets, some of which are significantly regulated by LHR activation. Table 4 lists four of these targets with the greatest changes in gene expression.

Endothelin-1 (ET-1)-mediated activation of the endothelin-1 receptor is known to result in vasoconstriction and smooth muscle cell proliferation and is implicated in the pathogenesis of hypertension, coronary vasospasms, and heart failure. More recently, it has been shown that ET-1, acting through its G protein-coupled receptor, ET_AR, is an important component of ovarian cancer initiation and progression [40‐42]. These findings have led to an interest in the development of endothelin-converting enzyme-1 inhibitors and small interfering RNAs as a new therapeutic agent for ovarian cancer [43]. Interestingly, the LHR+ cells respond to LH with a 10-fold increase in ET-1 gene expression, peaking at 1 h and remaining slightly elevated up to 20 h. The LH-mediated increase in ET-1 gene expression was confirmed by qRT-PCR (Figure 3). ET_AR expression is also increased about 2-fold in response to LH, while there are no significant effects on expression of the genes for endothelin-converting enzyme-1 and the endothelin B receptor. These results alone could indicate a possible enhancement of cell proliferation in response to LH. LH-mediated LHR activation also significantly up-regulates the stromal cell-derived factor 1 (SCD-1) and insulin-like growth factor II (IGF2) genes. The former has been reported to increase the invasiveness and migration of breast cancer cells [44], and the latter is known as a fetal promoter of cell proliferation that is involved in various forms of cancer [45]. The up-regulation of just these genes could suggest that LH exerts positive effects on tumor growth and metastasis. We know, however, from the experimental evidence that the up-regulation associated with these growth-promoting genes is not manifested in LH-activated LHR+ cells, and thus expression of the other negative regulators, e.g. c-JUN, TNFSF10, and MMPs, must assume a dominant role in relating gene expression and tumor cell properties.