Expression of G Protein-Coupled Receptor 19 in Human Lung Cancer Cells Is Triggered by Entry into S-Phase and Supports G2–M Cell-Cycle Progression

It has long been known that G protein-coupled receptors (GPCR) are subject to illegitimate expression in tumor cells. Presumably, hijacking the normal physiologic functions of GPCRs contributes to all biologic capabilities acquired during tumorigenesis. Here, we searched for GPCRs that were expressed in lung cancer: the mRNA encoding orphan G protein-coupled receptor 19 (GPR19) was found frequently overexpressed in tissue samples obtained from patients with small cell lung cancer. Several observations indicate that overexpression of Gpr19 confers a specific advantage to lung cancer cells by accelerating transition through the cell-cycle. (i) Knockdown of Gpr19 mRNA by RNA interference reduced cell growth of human lung cancer cell lines. (ii) Cell-cycle progression through G2–M-phase was impaired in cells transfected with siRNAs directed against Gpr19 and this was associated with increased protein levels of cyclin B1 and phosphorylated histone H3. (iii) The expression levels of Gpr19 mRNA varied along the cell-cycle with a peak observed in S-phase. (iv) The putative control of Gpr19 expression by E2F transcription factors was verified by chromatin immunoprecipitation: antibodies directed against E2F-1 to -4 allowed for the recovery of the Gpr19 promoter. (v) Removal of E2F binding sites in the Gpr19 promoter diminished the expression of a luciferase reporter. (vi) E2f and Gpr19 expression correlated in lung cancer patient samples. To the best of knowledge, this is the first example of a GPCR showing cell-cycle-specific mRNA expression. Our data also validate GPR19 as a candidate target when overexpressed in lung cancer. Mol Cancer Res; 10(10); 1343–58. ©2012 AACR.


Introduction
Lung cancer is among the leading causes of cancerrelated deaths worldwide and is strongly associated with smoking. On the basis of pathologic features lung cancers are classified into 2 major groups. Small cell lung cancer (SCLC) accounts for 20% of all cases and displays characteristics of neuroendocrine tumors (1). Non-small cell lung cancer (NSCLC) is further subdivided into squamous cell carcinoma (30%-40%), adenocarcinoma (30%-40%), and large cell carcinoma (<10%). Both types of lung cancer not only differ in their cellular morphology but also in the available treatment and their prognosis-with SCLC exhibiting the most aggressive and metastatic phenotype (2).
The defining features of the family of G protein-coupled receptors (GPCR) are the hydrophobic core of 7 transmembrane-spanning a-helices and their eponymous action, i.e., the activation of their cognate heterotrimeric G protein(s). The human complement of GPCRs consists of more than 800 members (3). GPCRs are intimately involved in the control of virtually all cell types, their structure allows for binding of highly diverse ligands, thus they are considered the most druggable family of proteins (4). However, only about 60 GPCRs are currently targeted by drugs approved for clinical use (5).
Several arguments support the assumption that GPCRs may also be specifically targeted to limit the growth of cancer cells: signaling pathways controlled by GPCRs promote proliferation, survival, cell migration, angiogenesis, inflammation, and subversion of the immune system (4,6). Their uncontrolled exploitation-mostly seen by overexpression of GPCRs-could in turn contribute to the malignant transformation of cells (6,7). It has long been known that illegitimate expression of GPCRs occurs in tumor cells (8) and an involvement in cell-cycle control has been shown for several GPCRs and GPCR-associated proteins. For instance, H4 histamine receptors mediate a reversible cell-cycle arrest in growth factor-induced hematopoietic progenitor cells (9) and GPCR kinase 5 was assigned an essential role in the regulation of prostate tumor growth by affecting cell-cycle control (10).
In the present work, we searched for GPCRs that are aberrantly expressed in lung cancer with a particular focus on SCLC. The underlying rationale was the observation that GPR87 was found to be overexpressed in lung squamous cell carcinoma (11) and contributed to the viability of tumor cells (12). Similarly, the chemokine receptor CXCR4 was proposed to contribute to lung cancer metastasis (13). In SCLC cells, it was suggested that an autocrine loop sustained RhoA activation via activation of a GPCR (14).
Our search identified G protein-coupled receptor 19 (GPR19), which was overexpressed predominantly in samples from patients with SCLC. Currently very little is known about the biologic functions of GPR19 in normal and diseased tissues (15). RNA interference established a link to cell proliferation and transition through the cell-cycle as progression through the G 2 -M-phase was impaired in cells transfected with Gpr19 siRNAs. In fact, we observed that the levels of Gpr19 mRNA varied along the cell-cycle with a peak in S-phase. This might be accounted for by the binding of E2F transcription factors to the promoter region of Gpr19.

Cell culture and cell-cycle arrest
Human lung-derived cell lines were cultured in normal growth medium as suggested by the provider (Supplementary Table S1) and maintained in logarithmic growth phase. They were authenticated using short tandem repeat analysis and tested for mycoplasma contamination on a regular basis. Cells were arrested at different stages of the cell-cycle as follows: 100,000 cells per well were seeded in 6-well culture plates, allowed to adhere for 16 hours in the respective growth medium and subsequently incubated in the presence of hydroxyurea (Sigma Aldrich; 1 mmol/L) or aphidicolin (Sigma Aldrich; 3 mmol/L). After 24 hours, the medium was exchanged for normal growth medium to remove the compounds and to thereby release the cells from the arrest.

Cell proliferation
The effect of doxorubicin (1 mmol/L) or siRNA (20 nmol/L) on cell proliferation was determined by monitoring the number of viable cells using the alamarBlue detection kit (Life Technologies). Experiments were done in 96-well culture plates. Fluorescence was recorded using a 2030 Multilabel Reader VICTOR X5 (PerkinElmer). In addition, proliferation of cells was assessed by determining cellular confluence using a CloneSelect Imager (Molecular Devices). The surface covered by cells (96-well plate) was recorded on regular time points. Besides, phase contrast microscopy pictures of cells were recorded (20Â magnification).

Flow cytometry and cell-cycle analysis
Cells were fixed with 70% ethanol (-20 C), washed with PBS and subject to RNase A treatment (1 mg/mL; Life Technologies) for 5 minutes followed by propidium iodide DNA staining (50 mg/mL; Sigma Aldrich). Cellular fluorescence was quantified using a FACSCalibur (Becton Dickinson) flow cytometer. Fluorescence histograms were analyzed with FlowJo software 7.5 (Tree Star). The Watson curve-fitting model was chosen for cell-cycle analysis.

RNA isolation and cDNA synthesis
The ABI Prism 6100 Nucleic Acid PrepStation (Life Technologies) was used to isolate RNA from cell lysates following the manufacturer's instructions. Contaminating DNA was removed by DNase treatment. cDNA was synthesized in a GeneAmp PCR System 9700 (Life Technologies) using the iScript cDNA Synthesis Kit (Bio-Rad) following the manufacturer's protocol and diluted with nuclease-free water (1:5-1:8) for direct use in quantitative PCR applications.

Quantification of gene expression by real-time PCR
Reverse transcription quantitative real-time PCR (RT-qPCR) was done on an Mx2005P qPCR machine (Stratagene) using the TaqMan technology (Life Technologies). A total of 10 mL of TaqMan Gene Expression Master Mix (Life Technologies) were mixed with 1 mL of both FAM-(gene of interest) and VIC-labeled (reference gene) TaqMan assays (Life Technologies; Supplementary Table S2) and 8 mL of diluted cDNA. PCR conditions were 2 minutes at 50 C and 10 minutes at 95 C followed by 40 cycles of 15 seconds at 95 C and 1 minute at 60 C (data collection).
The expression of genes of interest was normalized (geometric averaging) against multiple most stably expressed reference genes. Reference genes B2m, Actb, Cypa, Gapdh, Hprt1, Rplp0, Rnase P, and Tfr were tested for stable expression as described (16). The final selection of reference genes was based on this stable expression, a low difference in quantification cycles and a PCR efficiency comparable to the ones for the genes of interest. Pairwise duplexing was done for all genes of interest against all selected reference genes. Relative expression ratios were normalized against a control sample (17). Assay controls (either devoid of template or lacking reverse transcriptase) were included in all experiments and gave negative results.

Immunofluorescence labeling
Cells seeded in black poly-D-lysine 96-well culture plates were fixed with formaldehyde and blocked with 1% goat serum (Dako) and 2% bovine serum albumin in PBS. They were immunostained for cyclin B1 (BD Biosciences) and phosphorylated histone H3 (Ser10; Upstate) proteins. DNA was stained using Hoechst 33342 (Life Technologies).

High-content screening
Fluorescence-based cell detection was done with an ArrayScan VTI HCS reader (Thermo Fisher Scientific). Six wells per reaction condition and 36 fields (maximum) per well were recorded at 20Â magnification. The total number of cells monitored per well was in the range of 300 to 1,000 depending on siRNA treatment and time post-siRNA transfection. For Plk1 siRNA-transfected samples this was only valid on the second day after transfection due to the strong antimitotic effect of the knockdown. The data set was processed and visualized by Spotfire DecisionSite 9.1.2 (TIBCO) and iView 1.0.182.1 (Thermo Fisher Scientific). Cells were detected on the basis of Hoechst 33342 DNA staining. Intensities of immunostaining for both, phosphorylated histone H3 and cyclin B1 were examined by a Kolmogorov-Smirnov goodness-of-fit analysis as described (18).

Transcription profiling and hierarchical clustering
RNA from human NSCLC, SCLC, and normal lung samples was purchased from Origene Technologies. Detailed information on patient samples is listed in Supplementary Table S3. Affymetrix microarray analysis was conducted on the Human Exon 1.0 ST Array platform. Samples were processed following the Affymetrix GeneChip Whole Transcript Sense Target Labeling Assay manual. Raw (single intensity value per probe on a microarray) files were Robust Multichip Average-normalized using the affy package in R/Bioconductor (19). Seven hundred seventy-three genes encoding GPCRs were selected on the basis of Gene Ontology number 0004930. Their expression values in NSCLC, SCLC, and normal lung samples were extracted from the microarray data. Geometric mean values across all normal lung samples were used as reference for the values from NSCLC and SCLC patient samples. Hierarchical clustering (average linkage, Euclidean distance) of the log 2-transformed normalized expression data were conducted as described (20). Gene expression levels of Gpr19 (probe set 207183_at) in various human normal and cancerous tissues were investigated by using the BioExpress database (Gene Logic; 12).
Gene expression data of human lung-derived samples from Hou and colleagues (21) and Rohrbeck and colleagues (22) were extracted from the Oncomine data platform (Compendia Bioscience) and analyzed using Spotfire Deci-sionSite 9.1.2 (TIBCO).
ChIP PCR was conducted in a GeneAmp PCR System 9700 (Life Technologies) using 2 mL of DNA solution and 200 nmol/L of forward and reverse primer, respectively, in a JumpStart REDTaq ReadyMix Reaction Mix (Sigma Aldrich). PCR conditions were 45 seconds at 95 C followed by 32 cycles of 15 seconds at 95 C, 15 seconds at 59 , 40 seconds at 72 C, and completed by 3 minutes at 72 C. Nuclease-free water instead of DNA was used as nontemplate control. PCR products were separated by electrophoresis on 2% agarose gels and visualized by ethidium bromide staining.

Results
Gpr19 mRNA expression is high in SCLC patient samples and cell lines Many GPCRs control signaling pathways that regulate cell growth, proliferation, and differentiation. It has long been known that tumor cells may express GPCRs in an aberrant fashion (8). In an unbiased approach we investigated gene expression levels of all known GPCRs in a series of human SCLC, NSCLC, and normal lung RNA samples. Our survey also included odorant receptors, because these may be expressed in tissues other than the olfactory epithelium; in fact, they may occur in neoplasms such as prostate cancers (27). Hierarchical clustering identified Gpr19 among the GPCRs that showed a message overexpression mostly in SCLC but not in NSCLC samples relative to normal controls (Fig. 1A). High Gpr19 mRNA levels in SCLC samples were verified by a direct comparison with those seen in samples from NSCLC and normal lung (Fig. 1B). This was confirmed by RT-qPCR analysis (Fig. 1C).
We further quantified Gpr19 expression levels by RT-qPCR in a set of NSCLC, SCLC, and normal lung-derived cell lines. The expression pattern of Gpr19 was recapitulated, i.e., cell lines derived from SCLC had higher Gpr19 expression than those derived from NSCLC and from normal lung (Fig. 1D). However, high levels of Gpr19 mRNA were also seen in a NSCLC cell line (NCI-H1703) and an immortalized lung epithelial cell line derived from a cystic fibrosis patient (IB3-1). This may be related to the fact that the IB3-1 cell line was immortalized by the large T antigen (28).
We used the BioExpress database for examination of Gpr19 expression in various human normal and cancerous tissues. This search revealed high Gpr19 mRNA expression predominantly in the central nervous system (cerebral cortex, cerebellum, spinal cord; Fig. 1E) as described (29). In addition, Gpr19 mRNA levels were prominent in samples from SCLC and pancreas islet cell carcinoma.
RNA interference-mediated knockdown of Gpr19 reduces cell proliferation The prominent expression of Gpr19 preferentially in SCLC and in several lung cancer-derived cell lines suggested that this receptor may play a crucial role in cell proliferation and/or survival. We investigated the effect of Gpr19 knockdown on proliferation in lung cancer cell lines NCI-H1703 (NSCLC) and DMS 53 (SCLC). These cell lines were chosen based on high Gpr19 mRNA expression and because they grow as monolayers. Most SCLC cell lines including COR-L88, NCI-H209, NCI-H345, NCI-H446, and SHP-77 failed to grow as monolayers but formed (floating) clusters, which rendered them less amenable to experimental manipulations. GPR19 is an orphan receptor, its physiologic ligand is not known and there is no antagonist. Accordingly, we used RNA interference to examine the role of this receptor in supporting cell growth. Cells were either treated with siRNAs targeting Gpr19 mRNA for destruction (GPR19 #1, GPR19 #2), control siRNA (CTL #1, CTL #2), or transfection reagent only (mock) or left untreated. Additional positive controls included the cytotoxic anthracycline doxorubicin and an siRNA directed against Plk1 (30). We monitored the extent of well surface covered by adherently growing cells (confluence) to quantify cell proliferation. Images of wells covered by NCI-H1703 and DMS 53 cells were captured and analyzed at regular intervals after treatment using the CloneSelect Imager ( Fig. 2A). Well thumbnails (pictures of the well bottom and cellular detection by the selected detection algorithm) were the basis for confluence calculation ( Supplementary Fig. S1A). Measurements at regular intervals started 10 hours after treatment and lasted until day 6 (see figure insets for data points from 3 experiments; Fig. 2A). At this time point, untreated, mock and control siRNA-treated cells had already reached a high extent of confluence. Plk1 siRNA-transfected and doxorubicin-treated cells did not grow (absent gain in confluence over time). For cells transfected with either Gpr19targeting siRNA #1 or #2, the cell growth was inhibited as gain in well confluence was modest.
We confirmed the growth inhibition by siRNAs directed against Gpr19 in an additional proliferation assay (alamar-Blue), which recorded the number of viable, i.e., metabolically active cells. Cells were transfected with siRNAs or treated with doxorubicin and the metabolic activity of cells was determined 3 and 6 days later (Fig. 2B). There was a clear-cut reduction in metabolically active cells transfected with siRNAs targeting Gpr19 compared with untreated, mock and control siRNA-treated samples. In both cell lines the effect was more pronounced on day 6 than on day 3.
To directly visualize the effect on proliferation, we recorded pictures of cells from all different experimental conditions on days 3 and 6 after treatment using phase contrast microscopy ( Supplementary Fig. S1B). NCI-H1703 showed a similar amount of dead cells when cells were Gpr19 siRNA-transfected compared with Plk1 knockdown and doxorubicin treatment. This was already the case on day 3 after treatment and became also apparent for DMS 53 cells on day 6. The effective knockdown of Gpr19 mRNA by Gpr19-targeting siRNAs was verified using RT-qPCR (Fig. 2C). We also attempted to confirm these experiments by using antibodies against GPR19. However, neither commercially available antibodies from different providers nor in-house generated antisera toward 9 different GPR19 peptides allowed specific detection on Western blots (data not shown, 31).

Knockdown of Gpr19 induces G 2 -M arrest
Knockdown of Gpr19 reduced cell proliferation in NCI-H1703 and DMS 53 cells. Hence, we examined which phase of the cell-cycle was affected by the loss of Gpr19 mRNA. NCI-H1703 cells were either left untreated or transfected with control (CTL #1, CTL #2), Gpr19 (#1, #2), or Plk1 siRNAs. On days 2, 3, and 4 after transfection flow cytometry was done to analyze the distribution of cells in different stages of the cell-cycle ( Fig. 3A and B). Effective knockdown of Gpr19 mRNA was confirmed by RT-qPCR (Fig. 3C). The number of cells in G 2 -M-phase increased over time when cells had been treated with Gpr19 siRNAs in comparison to untreated or control siRNA-treated cells. This increase in the G 2 -M population occurred at the expense of cells that were in S-phase whereas the proportion of cells in G 1 was less affected. As anticipated, knockdown of Plk1 caused a pronounced G 2 -M arrest (30). This experiment was also conducted in DMS 53 cells using untreated cells or cells transfected with control siRNA CTL #1, Gpr19 siRNA GPR19 #2 (most effective depletion of Gpr19 mRNA), or Plk1 siRNA which yielded similar results (Supplementary Fig. S2A and S2B; knockdown of Gpr19 mRNA documented in Supplementary Fig. S2C).
We verified the observations obtained from flow cytometry by an independent approach that relied on high-content screening of cells stained for the G 2 -M-phase marker cyclin B1 and the mitosis marker phosphorylated histone H3. This experiment was only conducted with NCI-H1703 cells because DMS 53 cells were not suitable for imaging with the ArrayScan VTI HCS reader. NCI-H1703 cells were transfected with control (CTL #1, CTL #2), Gpr19 (#1, #2), or Plk1 siRNAs or left untreated and analyzed on days 2, 3, and 4 after transfection. Knockdown of Gpr19 mRNA by Gpr19-targeting siRNAs was effective as monitored by RT-qPCR (not shown). We used a Kolmogorov-Smirnov test to evaluate the distribution of cells that stained positive for phosphorylated histone H3 and cyclin B1 for each time point after siRNA transfection. This evaluation confirmed the increase in G 2 -M-arrested cells ( Fig. 4A and B). Figure  4C and D show representative pictures from the highcontent screen based on Hoechst 33342 DNA staining and cells stained positive for phosphorylated histone H3 and cyclin B1, respectively, on day 3 after transfection. The level of both, immunoreactivity for phosphorylated histone H3 and cyclin B1 increased in cells transfected with the Gpr19 siRNA. This effect was most prominent on days 3 and 4 after transfection. The G 2 -M block induced by the Plk1 siRNA also resulted in increased levels of phosphorylated histone H3 and cyclin B1. In addition, high-content screening confirmed the G 2 -M arrest and the reduced proliferative capacity of cells treated with Gpr19 siRNA on the basis of Hoechst 33342 DNA staining (data not shown).
Manipulations that interfere with transition of cells through G 2 -M are predicted to result in aberrant cell division. Accordingly, we examined whether transfection with Gpr19 siRNAs elicited morphologic changes that were consistent with impaired chromosomal segregation. This was indeed the case. High-content screening was indicative of abnormal DNA content in cells that had been treated with siRNAs directed against Gpr19. Visual inspection revealed an excess of polylobed and binucleated cells ( Supplementary Fig. S3).
Gpr19 mRNA expression is cell-cycle dependent and peaks during S-phase Observations from Gpr19 mRNA knockdown experiments are consistent with the hypothesis that this GPCR plays a role during the G 2 -M-phase of the cell-cycle. We therefore investigated whether Gpr19 mRNA was differentially expressed during the course of the cell-cycle.  NCI-H1703 and DMS 53 cells were treated with hydroxyurea for 24 hours to synchronize them at the G 1 -S transition. Samples were taken at various time points post release from cell-cycle arrest for DNA content analysis by flow cytometry. Upon removal of hydroxyurea, DMS 53 (Fig. 5A, left panel) and NCI-H1703 cells (Fig. 5B, left panel) moved subsequently through the cell-cycle in an essentially synchronous fashion: the number of cells in Sphase rapidly accumulated and the DNA content shifted from 2n to 4n within 8 hours. After 12 hours, the vast majority of cells had reached the G 2 -M-phase of the cellcycle. At later time points (24,36, and 48 hours) the effect of hydroxyurea was lost and the cell-cycle distribution in the population approached that seen in an asynchronously growing culture. RNA was also extracted at these time points and the message levels of several genes were determined by RT-qPCR ( Fig. 5A and B, right-hand panels). The levels of mRNA encoding cyclin E1 were elevated although cells were at the G 1 -S boundary and subsequently declined to reach a nadir after 12 hours in both, DMS 53 and NCI-H1703 cells. In contrast, a peak in cyclin B1 mRNA expression was seen after 12 hours, i.e., coincident with the accumulation of cells in the G 2 -M-phase of the cell-cycle. These results were anticipated as cyclin B1 and cyclin E1 show a known cyclic pattern of mRNA expression over the course of the cell-cycle. Expression levels of cyclin E1 are highest before S-phase entry, fall during S-phase, and remain low during G 2 -Mphase whereas cyclin B1 message is low in G 1 -phase, increases during S-phase and peaks at G 2 -M-phase (32). Both patterns were recapitulated in our experiments. The mRNA of Rpl32 encodes a ribosomal protein of the 60S large ribosomal subunit and does not vary through the cell-cycle (32). Accordingly, we used Rpl32 as an internal control. In addition, we examined the cell-cycle-dependent changes in mRNAs encoding 2 different GPCRs, namely Lpar1 for NCI-H1703 cells and Chrm3 for DMS 53 cells. These  For each day and cell-cycle phase, differences across all experimental groups were tested for significance using one-way ANOVA followed by Tukey test (B). GPR19 #1 and GPR19 #2 P values refer to the least significant one from the comparison with CTL #1 and CTL #2 ( Ã , P < 0.05; ÃÃ , P < 0.01). C, relative Gpr19 mRNA levels on day 2 after transfection (normalized against reference genes Cypa and Hprt1, geometric mean of triplicates per condition; data are means þ SD of 3 experiments).
mRNAs remained essentially constant at the time points examined. The mRNA encoding Gpr19, however, was subject to cell-cycle-dependent regulation: Gpr19 mRNA levels rose when the synchronized cells moved through Sphase, peaked after 6 to 8 hours, and started to decline as cells reached G 2 -M-phase, i.e., after 10 hours ( Fig. 5A and B, right panels). At later time points (24, 36, and 48 hours) mRNA levels did not differ appreciably from those seen in untreated control cells (open bar in the right-hand panel of Fig. 5A and B), regardless of which mRNA was examined. This is consistent with the fact that the cell population grew again asynchronously at this stage. We addressed the possibility that differential Gpr19 mRNA expression was because of an off-target effect of hydroxyurea, i.e., unrelated to its effect on the cell-cycle. For this reason we repeated the synchronization study with DMS 53 cells using the DNA replication inhibitor aphidicolin. Cell-cycle profiles and the degree of synchronization were comparable to those of hydroxyurea-treated cells (Fig. 5C,  left panel). Similarly, mRNA profiles were also comparable (Fig. 5C, right panel). Most importantly, following aphidicolin-induced synchronization, the time-dependent changes in Gpr19 mRNA levels recapitulated those seen in hydroxyurea-treated cells.
We explored the correlation between Gpr19 expression levels and the relative number of cells in G 1 -, S-, and G 2 -Mphase for both hydroxyurea-synchronized DMS 53 and NCI-H1703 cells (Fig. 6). The proportion of cells at different stages of the cell-cycle was determined using the Watson curve-fitting model from FlowJo. As hydroxyurea arrested cells at the G 1 -S transition, it was impossible to precisely discriminate between G 1 -and S-phase populations at 0 and 2 hours after release of the hydroxyurea-induced block. These data points were therefore omitted. In both cell lines Gpr19 message levels correlated with the relative number of cells in S-phase. Conversely, Gpr19 mRNA levels were inversely related to the fraction of G 1 cells. In contrast, we did not observe any correlation between the amount of cells in G 2 -M-phase and Gpr19 mRNA expression.
The data in Figs. 5 and 6 suggested that accumulation of the mRNA encoding Gpr19 was initiated as soon as cells progressed from G 1 -to S-phase. Members of the E2F family of transcription factors are known to boost transcription from E2F-targeted promoter sites during late G 1 -or early Sphase (33), e.g., the component of the prereplicative complex cell division cycle 6 (Cdc6; 34). The Gpr19 promoter region (5 kb of the 5 0 upstream sequence ahead of the open reading frame) was examined for prospective E2F binding sites using ConSite and the University of California, Santa Cruz (UCSC) genome browser. Four potential E2F binding sites (À15, À185, À3184, and À3769 bp upstream of the Gpr19 open reading frame) were retrieved from the ConSite search. In contrast, only the one at À15 was identified by the UCSC genome browser search. We confirmed the presence of E2F binding sites in the Gpr19 promoter by ChIP. ChIP assays identified the presence of E2F-1, E2F-2, E2F-3, and E2F-4 in both NCI-H1703 and DMS 53 cells (Fig. 7). The promoters of Cdc6 and albumin served as positive and negative controls (34,35), respectively. Chromatin fragmentation was achieved by sonication and shearing resulted in DNA fragments of 300 to 600 base pairs. It was therefore impossible to discriminate between recruitment of E2F-1, E2F-2, E2F-3, and E2F-4 to either binding sites À15 or À185 and/or to both of them. Primer pairs for PCR amplification of all predicted E2F binding sites were chosen in a way that the PCR product did not contain any other site-except for primer pair GPR19 (À15; À185). With this primer pair the resulting PCR product covered both these predicted E2F binding sites of the Gpr19 promoter.
Several facts suggest that the binding of E2F-1, E2F-2, E2F-3, and E2F-4 to the promoter sites at position À15 and/or À185 of the Gpr19 gene were specific: (i) immunoprecipitation of these E2F-family members    NCI-H1703 DMS 53 Figure 7. Binding of E2F-1 to -4 to the promoter of Gpr19. Chromatin immunoprecipitation was conducted with antibodies against E2F family members (E2F-1, 2, 3, 4, 5, 6, 7) and IgG isotype control. As an additional negative control the immunoprecipitation was done in the absence of any antibody (no AB). DNA from the sheared chromatin input was serially diluted with water (undiluted, 1:5, 1:25, 1:125). Internal controls also included amplification of the promoter of a known E2F target gene (Cdc6) and of an E2F-independent gene (albumin) and to a region lying in the coding sequence (CDS) of Gpr19. Primer pair GPR19 (À15; À185) covered both predicted E2F binding sites of the Gpr19 promoter in the PCR product. Shown are the amplicons from a representative experiment, which was repeated twice with similar results. Full-length agarose gels for NCI-H1703 and PCR primer sequences are presented in Supplementary Fig. S4. of a firefly luciferase reporter gene to evaluate its influence on gene expression (Fig. 8). In addition, the putative E2F transcription factor-binding sites at positions À15 and/or À185 had been selectively removed. The removal of the putative E2F binding site at position À15 alone or in combination with the one at position À185 negatively affected reporter gene expression in HEK-293 and NCI-H1703 cells (Fig. 8, left and centerfold graphs). In contrast, the knockout of the putative E2F binding site at position À185 alone did not diminish luciferase expression. The opposite was true for DMS 53 cells: Removal of the E2F consensus site at position À185 from the Gpr19 promoter sequence diminished luciferase expression here (Fig. 8, right graph). This signal reduction was further augmented when the E2F consensus site at position À15 had also been erased. Hence, the consensus motif for the recruitment of E2F transcription factors to the promoter of Gpr19 at both positions À15 and À185 upstream of the Gpr19 open reading frame seemed to be important for luciferase reporter gene expression. The preferred usage of each position might be cell line dependent.

E2F transcription factor gene expression correlates with
Gpr19 expression in human lung cancer samples.
The recovery of the Gpr19 promoter in E2F ChIP assays and the results from Gpr19 promoter fragment-controlled luciferase reporter expression indicated a cause-and-effect relation between the action of E2F transcription factors and the expression of Gpr19. We addressed the question whether coexpression of E2F transcription factors and Gpr19 could be observed in patient samples as well. Indeed, on the basis of 2 published studies (21,22) and our in-house study described in Fig. 1A-C, a good correlation of the expression of Gpr19 and E2F transcription factors E2f-1 to E2f-3 could be detected (Fig. 9). The in-house study revealed a correlation coefficient of 0.863 for E2f-1, 0.682 for E2f-2, and 0.554 for E2f-3 whereas the correlation coefficient for E2f-4 with Gpr19 was 0.350. Similar correlations were seen in the other 2 studies.
The study by Rohrbeck and colleagues and our in-house study identified high Gpr19 expression predominantly in SCLC patient samples ( Fig. 9B and C). SCLC samples were not included in the study described by Hou and colleagues but high Gpr19 expression was observed in a subset of NSCLC samples here, namely large cell lung carcinoma (Fig. 9A). This was also true for a few lung cancer samples classified as adenocarcinoma or squamous cell carcinoma.

Discussion
The orphan receptor GPR19 was first identified in the central nervous system (29,37); it is also found in ovary and testis (37) and abundantly expressed in human embryonic stem cells (38). The gene encoding human Gpr19 resides on the short arm of chromosome 12 in a region that is frequently rearranged in childhood leukemia and to a lesser extent in other neoplasms; accordingly, GPR19 has been postulated to play a role in cancer development (39). In fact, high levels of mRNA coding for GPR19 were found in metastatic melanoma (7,40). Here we showed that GPR19 is associated with lung cancer. This conclusion is based on (i) an unbiased approach that surveyed GPCRs specifically for their overexpression in SCLC, (ii) database mining, and (iii) the presence of high Gpr19 mRNA levels in several human lung cancer-derived cell lines. High levels of Gpr19 mRNA were not only detected in patient samples from SCLC but also from pancreas islet cell cancer. Both these carcinoma types are often characterized by the presence of neuroendocrine markers (41,42). This is also true for a subset of NSCLC called large cell neuroendocrine carcinoma (2) and approximately 30% of SCLC is associated with NSCLC elements (43). In fact, an increase in Gpr19 message was observed in large cell lung carcinoma relative to normal lung samples when using the BioExpress database (Fig. 1E). In addition, high levels of Gpr19 expression could also be attributed to some NSCLC patient samples (mainly large cell lung carcinoma) in the lung cancer gene expression study described by Hou and colleagues (21; Fig. 9A). These data sets did not discriminate between neuroendocrine and nonneuroendo-crine large cell lung carcinoma. The NSCLC cell line NCI-H1703 also showed high levels of Gpr19 mRNA expression. It is not clear whether this reflects a neuroendocrine phenotype or de novo illegitimate expression. Thus, the discrimination between Gpr19 high and Gpr19 low expression lung cancers might not strictly follow histologic boundaries. In addition, we confirmed the presence of high Gpr19 mRNA expression in melanoma (Fig. 1E).
We examined if expression of Gpr19 conferred a growth advantage in lung cancer cells. The results were unequivocal: regardless of the histologic classification of the cell line (SCLC or NSCLC), knockdown of Gpr19 mRNA depressed cell proliferation. This finding is consistent with earlier  For the generation of the row dendrogram, the following settings were used: clustering method, unweighted pair group method with arithmetic mean (UPGMA); distance measure, Euclidean; ordering weight, average value; normalization, none; and empty value replacement, constant value: 0. For the generation of the column dendrogram, the same settings were applied except for that correlation was used as the distance measure. Max, Average, and Min refer to the expression level of the respective gene. The probe set for each gene used is indicated on the X-axis, the various tumor types are indicated by a color code in the column tissue type.
observations that indicate a role of GPR19 during embryogenesis, in particular in the developing brain: high levels of Gpr19 mRNA are initially observed in germ cell layers of the embryo. Subsequently Gpr19 mRNA accumulates in the neural plate, the subventricular zone and other sites, from which neuronal cells emerge in the differentiating central nervous system (15).
GPCRs control signaling pathways that are typically associated with recruitment of quiescent cells into the cell-cycle (G 0 -G 1 transition) or with accelerated progression through G 1 (4,6). Surprisingly, knockdown of Gpr19 affected progression through the cell-cycle at a later stage, namely in G 2 -M. This conclusion is based on 2 independent lines of evidence, namely the determination of DNA content by flow cytometry and the quantification of the G 2 -M marker cyclin B1 and the mitosis marker phosphorylated histone H3 by immunocytochemistry. Cyclin B1 exerts its function in conjunction with cyclin-dependent kinase 1 during late G 2 -phase and early mitosis (44). Histone H3 phosphorylation at Ser10 correlates with chromosome condensation during mitosis with phosphorylation levels starting to increase in late G 2 -phase (45). Thus, our observations are consistent with a model where GPR19 impinges on checkpoint controls that allow for transition through G 2 and entry into mitosis or for initiation of the separation of daughter cells. Earlier findings also suggest that GPR19 may play a role in cell division: Gpr19 gene expression levels are high in murine spermatocytes undergoing meiotic cell division (46). In this context it is of interest that NCI-H1703 cells treated with siRNAs directed against Gpr19 exhibited a binuclear or polylobed nuclear phenotype ( Supplementary  Fig. S3).
The mechanistic link between a GPCR and cell division during G 2 -M is not intuitively evident. It is, however, worth pointing out that heterotrimeric G proteins of the G ai /G aosubfamily are involved in the control of the mitotic spindle; they are thought to regulate the mitotic force generator and thus to promote chromosomal segregation (47). The expression of a GPCR does not per se render a cell susceptible to regulation by the receptor-the ligand must also be present. Alternatively, the receptor has a high level of basal activity and engages its cognate G protein(s) in the absence of an agonist (48). The growth inhibitory action of Gpr19-directed siRNAs could also be accounted for by these 2 scenarios that are not necessarily mutually exclusive: (i) when overexpressed, GPR19 displays strong constitutive activity. (ii) GPR19-expressing cells also synthesize and release the agonist; knockdown of the mRNA encoding the receptor disrupts the resulting autocrine loop. GPR19 was previously proposed to couple to G i because it engaged a fusion protein composed of G aq and the last 5 amino acids of G ai (49). On the basis of sequence alignments of the transmembrane core, GPR19 was classified as a class A (rhodopsin-like) receptor in spite of the absence of an NPxxY-motif in transmembrane helix 7; the cognate ligand was proposed to be a peptide (37). We attempted to verify G i -dependent inhibition of cAMP formation via GPR19. However, we failed to detect an effect of pertussis toxin on cAMP accumulation in forskolin-stimulated NCI-H1703 cells that highly express Gpr19 mRNA (data not shown). Because pertussis toxin-catalyzed ADP-ribosylation of G ai /G ao blocks access of receptors to their C-terminus, this manipulation ought to have unmasked constitutive inhibition of cAMP accumulation. Similarly, we failed to find any effect of conditioned medium (supernatants from cell lines NCI-H1703 and NCI-H345 which highly express Gpr19 mRNA) on cAMP accumulation in HEK-293 cells that had been transfected with Gpr19 expression plasmids (data not shown). Thus, it seems unlikely that the action of GPR19 in the G 2 -M-phase of the cell-cycle arises from coupling to G i /G o . Further progress in this area is contingent on the identification of the cognate agonist(s) of GPR19. Finally, a function other than direct signaling via G proteins might also be conceivable for GPR19, e.g., recruiting G protein-independent pathways via b-arrestins (50), via direct binding of tyrosine kinases (51), or by acting as a scaffolding protein (52).
Our experiments showed that the expression of Gpr19 was regulated during the cell-cycle. This finding provides additional-albeit circumstantial-evidence for a role of GPR19 in the regulation of the cell-cycle. The levels of mRNA encoding GPR19 peaked when most cells were in S-phase. For soluble proteins, this typically implies that the proteins are required for S-phase transition. GPR19, however, is a membrane protein, which must mature through the secretory pathway to reach its presumed site of action, the cell surface. Although transcription and translation are rapid (less than 1 minute for a protein of 400 amino acids), folding in the endoplasmic reticulum imposes a time lag of up to several hours (53). Besides, subsequent trafficking through and maturation steps in the secretory pathway result in an additional delay before insertion of the protein into the cell membrane. Pharmacochaperone-triggered folding of the A1-adenosine receptor suggests that more than 5 hours are required for the receptor to reach the cell surface (54). Thus, a delay of several hours is to be anticipated between induction of mRNA expression and accumulation of GPR19 at the cell surface. Given this inherent delay, expression of Gpr19 is timely for a membrane protein required for the G 2 -M-phase.
GPR19 is not the only GPCR whose expression is cellcycle-dependent. In primary cultures of human microvascular endothelial cells, the protein levels of the chemokine receptor CXCR3 were found to be highest when cells exited S-phase and entered G 2 -M (55). We stress that our data refer to mRNA levels of Gpr19 rather than accumulation of the protein at the cell surface. CXCR3 was also shown to be directly involved in the proliferation of endothelial cells: its ligands interferon-g-inducible protein of 10 kDa (IP-10) and monokine induced by interferon-g (Mig) stalled cell proliferation. This effect could be reversed by a CXCR3specific antibody.
The promoter region of the gene encoding GPR19 contains several candidate-binding sites for E2F transcription factors. We provide conclusive evidence that at least one of these was in fact occupied by E2F family members E2F-1, E2F-2, E2F-3, and E2F-4 in both DMS 53 and NCI-H1703 cells. E2F transcription factors are known to activate the expression of genes at the G 1 -S transition necessary for cell-cycle progression (33,35) including, e.g., Cdc6, Cdc25 (34), thymidine kinase, or dihydrofolate reductase (35). The family of E2F transcription factors has been subdivided into activators (E2F-1, E2F-2, E2F-3) and repressors (E2F-4, E2F-5, E2F-6, E2F-7, E2F-8) of gene transcription (33,36). Transactivation and repression result from interactions with different cofactors. E2F-4 and E2F-5 both prevent quiescent cells from entering G 1 -phase. Upon mitogenic stimulation, E2F-1, E2F-2, and E2F-3 accumulate in late G 1 -phase and cause a pattern of gene expression that drives cells into S-phase. During late S-phase this signal is attenuated by E2F-6, E2F-7, and E2F-8 (36). Classification of E2Fs into activators and repressors of gene transcription is an oversimplification, because some isoforms can mediate both transactivation and repression. However, we documented binding of 3 activating E2F isoforms to the Gpr19 promoter. Furthermore, the cloning of Gpr19 promoter fragments of different length with and without E2F binding site integrity at positions À15 and/or À185 upstream of a luciferase reporter identified the importance of both sites for gene expression. It might be possible that both E2F binding sites are in use for the activation of Gpr19 gene expression in a cell-specific manner. In all 3 cell lines tested, the combined abolition (positions À15 and À185) led to the most pronounced reduction of luciferase reporter expression.
Besides, the expression of activator E2F transcription factors E2f-1, E2f-2, and E2f-3 could be correlated with the expression of Gpr19 in human lung cancer samples from independent lung cancer gene expression studies. Only a weak correlation was observed between Gpr19 and E2f-4. The observed Gpr19 promoter binding of the repressor E2F-4 might reflect its recruitment to the promoter during early G 1 -phase and subsequent displacement with activator E2Fs in late G 1 -and early S-phase leading to Gpr19 transcription. Thus, it seems justified to conclude that there might be a cause-and-effect relation between E2F binding to the Gpr19 promoter and the observed accumulation of the mRNA coding for GPR19 during S-phase. Our conclusion is further supported by the observation that Gpr19 mRNA was present at high levels in the immortalized cell line IB3-1. This cell line was derived from the lung epithelium of a patient suffering from cystic fibrosis with the help of a hybrid virus. Accordingly, IB3-1 cells contain high levels of SV40 large T antigen (28), which inactivates the retinoblastoma protein and hence derepresses E2F-dependent transcription.
When the Gpr19 message in Gpr19-expressing lung cancer-derived cell lines was attenuated by the use of RNA interference, proliferation of these cells was diminished (Fig.  2). Thus, GPR19 might be a vulnerable target, whichwhen overexpressed in lung cancer cells-is involved in cell proliferation. Transiently transfected GPR19 expression did not result in gain of proliferation in neither HEK-293 nor NCI-H1703 cells (data not shown). However, these studies have not been conducted in nonmalignant cells. Whether this indicates that GPR19 might lack the ability to transform cells into a malignant phenotype needs further investigation. Hence, the present data do not allow the categorization of Gpr19 as an oncogene (56). Furthermore, its overexpression particularly seen in lung cancer-derived cell lines and SCLC patient samples might be accounted for by the deregulated action of members from the E2F family of transcription factors, which is very often observed in lung cancer (57,58).
Taken together, our observations highlight GPR19 as a candidate drug target for the treatment of a subset of lung cancers. To the best of our knowledge, GPR19 is the first GPCR whose mRNA has been shown to be expressed in a cell-cycle-dependent manner and to impinge on G 2 -M transition. Retinoblastoma-and E2F-dependent transcription may not only regulate S-phase-specific genes but also control the expression of genes that are required at later stages, e.g., in G 2 -M, particularly if these are membrane proteins subject to intracellular trafficking. Finally, we believe that the insights from our current work should also be of relevance to understand the physiological role of GPR19 during embryogenesis and brain development.