High throughput gene expression profiling has become an outstanding method for identifying genes differentially expressed in normal and diseased tissue. In order to identify new genes differentially expressed in cancerous tissue, metaGen Pharmaceuticals GmbH designed a customized DNA Cancer-Chip based on Affymetrix technology (Affymetrix, Inc. USA). This oligonucleotide microarray contained 6200 probe sets representing approximately 3,000 genes. The chip design based on a bioinformatic attempt mining approximately 4 million expressed sequence tags (ESTs) of public

http://www.ncbi.nlm.nih.gov/dbEST/

Therefore 52 prostate tumor samples and its corresponding normal tissues were collected at the time of radical prostatectomy (RP) at the Department of Urology at the University Hospital Charité from 1999 to 2000. All samples were microdissected and hybridized to the metg001A microarray. Microdissection was used in order to specifically select tumor areas and normal glands to further increase the number of identified differentially expressed genes, especially as prostate tumors are among the most heterogeneous of cancers [Singh, 02]. The clinical and pathological features as well as the follow-up data of all patients hybridized to the chip are listed in Tab. 7 of the attachment.

Analysis of hybridization experiments was done according to the metaGen criteria, described in “Methods”. Briefly, a gene was called differentially when it was significantly expressed in both normal and tumor tissue (p-value < 0.05) and the quotient (fold change [FC]) of the normalized perfect match quintile (PMQ) value of each patient and probe set (tumor/normal) was > 2. For genes present only in normal or tumor tissue (p-value > 0.05) no fold change was calculated, but marked as differentially expressed (FC was set to 2). Finally, genes were ranked according to their differentially expression between normal and tumor tissue.

Tab. 1 Upregulated genes in prostate cancer patients identified by Affymetrix GeneChip experiments. Genes were sorted by numbers of patients with a fold change > 2 (quotient of PMQ-values of matched tumor and normal sample for each gene), HGN for HUGO Gene Nomenclature, FC for fold change.

Microarray analysis of prostate cancer patients revealed that 26% of the probesets (1434 probesets) present on the microarray showed a differential expression in at least one patient. Two criteria were applied for the identification of possible target genes. First, the gene should be overexpressed in tumors in at least 30% of the patients. Second, this gene should not be downregulated in more than 10% of the patients. Following these criteria 38 genes could be identified as overexpressed in prostate tumor patients. From these genes, seven were even upregulated in more than 50% (27 patients) of the tumors. The list of overexpressed genes in prostate tumor patients is shown in Tab. 1.

Tab. 2 Downregulated genes in prostate cancer patients identified by Affymetrix GeneChip experiments. Genes were sorted by numbers of patients with a fold change < 0.5 (quotient of PMQ-values of matched normal and tumor sample for each gene), HGN for HUGO Gene Nomenclature, FC for fold change.

62 genes could be identified being underexpressed in at least 30% of the patients (FC < 0.5) using the same criteria as for genes downregulated in prostate cancer. From these, 5 genes could be identified to be downregulated in more than 50% of the tumors, listed in Tab. 2.

For a more detailed analysis of this prostate cancer profiling, please refer to the thesis of Christoph Wissmann [Wissmann, 02].

Affymetrix gene chip experiments and electronic Northern analysis revealed that TRPM8 expression is restricted to the prostate. To confirm these high-throughput methods and to analyze the distribution of TRPM8 expression in “wet” experiments, classical Northern blots and dot blots as well as Real Time PCR were performed on a large number of patient and tissues.

Fluorescence in situ hybridization (FISH) analysis of the genomic region of TRPM8 was performed to gain insight into the molecular mechanism of the overexpression of TRPM8 in prostate tumors and LNCaP cells. As TRPM8 is located at the very end of chromosome 2 (2q37.2), the chromosomal region could possibly by either amplified and/or translocated to other chromosomes. Through translocation, the TRPM8 gene might get under the control of an alternative promoter or enhancer resulting in an altered expression profile. Both chromosomal amplification and translocation are known mechanisms often occurring in tumor tissues [Nowell, 97]. In order to address this question FISH experiments were performed on LNCaP cells. This cell line was chosen, as it is the only cell line expressing TRPM8. The BAC AC005538 (2q37.2) was used as a TRPM8 specific probe for hybridization. As a positive control a commercially available probe for identifying MYC on 8q24.12-q24.13 was used (Vysis Inc., Downers Grove; IL, USA). A comparison of hybridization signals of TRPM8 between LNCaP cells to normal human XY specimens showed that TRPM8 hybridized exclusively to the expected position on chromosome 2, in both the LNCaP cell and the XY patient (Fig. 15). No hybridization signal could be detected anywhere else in the chromosomes.

Fig. 15 FISH mapping of the genomic region of TRPM8 and MYC in normal human XY patients and LNCaP cells. Hybridization signals for TRPM8 (AC005538 on 2q37.2.) are shown in green; hybridization signals for MYC (8q24.12-q24.13) are indicated in red. Picture A1 to C1 show metaphase chromosomes. Pictures A2 to C2 show interphases of the same sample. A) 1 and 2 show hybridization signals specific for TRPM8 in a healthy XY-person. B) 1 and 2 show mapping of TRPM8 in LNCaP cells. C) 1 and 2 show the hybridization signal of MYC to chromosome 8.

The finding that chromosome 2 is tripled in LNCaP cells and chromosome 8 (MYC-specific binding) is quadruplicated, is in coherence with the literature [Augustus, 03]. A spectral karyotyping of LNCaP cells shows a predominantly tetraploid karyotype of LNCaP cells. Only chromosome 2 (triploid), 6 (diploid), 19 (triploid) and 21 (triploid) differ from that pattern (Fig. 16).

Fig. 16 Spectral karyotyping (SKY) of LNCaP cells taken from [Augustus, 03]. Chromosome 2 and 8 are boxed as these are the chromsomes where TRPM8 and MYC are localized, respectively.

Additionally, the signal strength was not enhanced in LNCaP cells compared to normal XY patients. Therefore, it can be concluded that the overexpression of TRPM8 in LNCaP cells and thus most likely also in prostate tumor patients is not due to chromosomal amplifications or rearrangements.

TRPs are known to form either homo- or hetero-tetrames. In order to prove direct interactions between TRPM8 subunits, channel multimerization was assessed using the fluorescence resonance energy transport (FRET) technology. Therefore TRPM8 was terminally fused to CFP and YFP vector construct. These plasmids where transiently coexpressed in HEK293 cells and the proximity of the homomerization was measured with FRET. FRET signal was obtained by measuring the increase in fluorescence of donor (CFP) emission during photobleach of the acceptor (YFP). The recovery of donor fluorescence emission was then monitored at 480 nm and was expressed as percentage of CFP emission after acceptor bleach. The FRET analysis showed a strong interaction of TRPM8 homomultimers which is shown in FRET efficiencies of 15.3% (Fig. 17, A).

Fig. 17 FRET analysis of transiently transfected HEK293 with TRPM8-CFP and TRPM8-YFP after 24 h. A) and B) Recovery of fluorescence intensity of the FRET donor (∆FCFP) during disruption of energy transfer by photobleaching of the acceptor (FYFP). The acceptor was selectively bleached at ג = 515 nm. The relative increase of CFP (ΔFCFP (%) intensities compared to initial levels and YFP fluorescence intensity decrease (FYFP (%)) over time. The increase of CFP fluorescence intensity of 15, 3% is a direct evidence of FRET. Data shown are representative for several FRET analyses.

It was shown by McKemy et al., that the TRPM8 rat orthologue CMR1 can be activated by cold inducing agents such as icilin, menthol and eucalyptol as well as by temperatures below 28 °C resulting in an increase of intracellular calcium levels. The mouse orthologue of TRPM8 could also be activated by menthol and temperatures ranging from 25°C to 15°C [Peier, 02d]. In order to find out whether human TRPM8 could also be activated with different cooling agents, TRPM8-stable expressing HEK293 cells were loaded with the Ca2+- indicator Fura-2 and exposed to 50 µM of Icilin. These experiments were done in cooperation with Stefan Mergler from the Charité in Berlin.

As displayed in Fig. 18 (A, B) TRPM8 expressing cells showed a threefold increase in intracellular calcium upon stimulation with Icilin. That response was not observed in nontransfected and empty vector transfected cells. The icilin mediated calcium influx occurred within msec. Washout of icilin induced recurrence of [Ca2+]i back to levels of unstimulated cells (recovery effect). The response was dependent on Ca2+ in the buffer, because removal of extracellular calcium suppressed the Icilin response. The result indicated that TRPM8 is localized in the plasma membrane, although transmembrane localization could not be seen in most of the TRPM8-expressing HEK293 cells. This is most likely due to limitations of sensitivity in the cytochemistry procedures.

Fig. 18 Analysis of calcium influx of HEK293 cells expressing TRPM8 using Fura-2. Representative measurement data were taken from calcium experiments of transfected and nontransfected HEK293 cells loaded with Fura-2. WT = wild type

As prostate was the first entity to be hybridized to Affymetrix GeneChips at metaGen, it was interesting to check two years later the expression on TRPM8 in entities other than the prostate. Until today 6 other tissues have been hybridized to several Affymetrix chips (either proprietary or custom designed). Bladder, ovary, mammary gland and pancreas were hybridized to the metg001A Cancer-Chip; lung and colon were hybridized to the U133A and B Affymetrix GeneChip. The probe sets on the U133B were not identical to the ones on the metg001A chip but adequate, as described in detail in “Methods”. Bioinformatic analysis of these 7 tissues revealed that TRPM8 expression was restricted to the prostate with two exceptions. One bladder-tumor sample (out of 79) and one lung cancer (out of 172) expressed TRPM8 (Fig. 19).

Fig. 19 Expression of TRPM8 in Affymetrix GeneChip experiments of 7 human tissues. Hybridization experiments were performed with 123 prostate -, 21 ovary -, 90 mammary gland -, 78 bladder -, 102 colon -, 11 pancreas - and 172 lung –specimens of normal and cancer tissue.

These results were quite unexpected as dot blot and Northern blot experiments showed exclusive expression of TRPM8 in the prostate. The bladder and the lung cancer sample stoud out very clearly from the other samples. Looking closer at the bladder cancer patient, this sample did not only express TRPM8, but also Kallikrein (PSA), Folathydrolase1 (PSMA) and the Acidic Phosphatase Prostate (ACPP) all of which are prostate specific markers (Fig. 20). None of the other 71 bladder samples did express any of these genes. In general these 4 genes were expressed conjointly in nearly 100% of all prostate samples, but very rarely (and never all 4 gene together) in any non-prostatic tissue tested (data not shown). As it was most unlikely that this patient really expressed prostate specific genes, the pathologist at the University of Regensburg (from where the tissue came from) was asked to diagnose this sample again. In response, she told us that this patient was indeed a bladder cancer patient, but that some prostate glands had infiltrated into the bladder. Unfortunately these glands were also microdissected and thus amplified and hybridized on the chip. Consequently this sample was taken out of further analysis of the bladder-experiments at metaGen. The most important aspect of this feature was not the fact that the sample was no pure bladder sample, but the confidence it provided regarding the data gained from the microarray experiments.

Fig. 20 Affymetrix GeneChip analysis of prostate specific genes in bladder cancer patients. Expression values are shown as PMQ for TRPM8, KLK3, FLOH1 and ACCP.

Fig. 21 GeneChip analysis of prostate specific genes in lung cancer patients. Expression values are shown as PMQ for TRPM8, KLK3, FLOH1 and ACCP.

Strikingly, in lung cancer patients TRPM8 was the only prostate specific gene expressed (Fig. 21). Neither Kallikrein nor Folathydrolase1 nor acidic phosphatase were expressed in this sample. A contamination with prostatic tissue or a prostatic metastasis in the lung could thus be excluded. In order to find an answer to this result the patients data sheets were checked in collaboration with the pathologist and an interesting observation could be made: Lu93 was the only patient run on microarray chips which had a fraction of 10% neuroendocrine tumor cells.

Real-Time PCR was performed to evaluate the assumed expression of TRPM8 in the patient with a 10% neuroendocrine tumor of the lung (Lu93). Additionally, a 100% neuroendocrine tumor from the lung and a lung adenocarcinoma were analyzed. The mRNA of normal and tumor material of these patients was isolated and quantitative PCR was performed. Results shown in Fig. 22 indicate that TRPM8 is highly overexpressed in both the 10% and the 100% neuroendocrine tumors, but not in the andenocarcinoma of the lung.

Fig. 22 Strong overexpression of TRPM8 in neuroendocrine tumors. Real-Time PCR of TRPM8 shown as relative expressions to corresponding normal tissue of each patient. A) Pure adenocarcinoma of the lung (panel 1), a lung adenocarcinoma with 10% of neuroendocrine cells (panel 2) and a 100% neuroendocrine tumor also located in the lung (panel 3). B) Expression of TRPM8 in neuroendocrine cell lines. RT-PCR results are shown as relative expression levels to the normal prostate epithelial cell line PrEC.

TRPM8 is located on chromosome 2q37.2 TRPM8 distributed over 100kb at the very end of chromosome 2. The gene consists of 26 exons resulting in 5641 base pairs (Fig. 23). The open reading frame (ORF) has 3312 basepairs resulting in an ORF of 1104 amino acids. The ion pore of TRPM8 consists of 6 transmembrane spanning domains located between exon 16 and 20. Both N and C-termini are located in the cytoplasm.

Fig. 23 Genomic structure of TRPM8 on Chromosome 2q37.2.

TRPM8 splice variants could be identified while cloning the TRPM8 full length gene. The TRPM8 gene was amplified from a normal prostate mRNA pool (Ambion, Huntington, UK). A polyT-T7 primer was used in the reverse transcription RT-reaction which was followed by a PCR reaction using TRPM8 specific primer which bound to the first and last exon of TRPM8. It was interesting to observe that after cloning and sequencing of the constructs a short variant of TRPM8 could be identified. Exons 1-6 of this variant were identical to TRPM8, but exon 6 was elongated by 245 bp ending with a polyadenylation signal (SV 6b). Cloning of the product was possible as the primer used for the cDNA synthesis functioned as the reverse primer in the PCR reaction. This finding was in accordance with the results gained in the Northern blot experiments where two other isoforms of TRPM8 could be detected, although splice variant 6b was smaller than the shortest fragment (4.1kb) seen in the Northern blot experiments (Fig. 11).

In the course of performing genome database search using public

http://www.ncbi.nlm.nih.gov/dbEST/

Fig. 24 Six splice variants of TRPM8. Alternative exons of TRPM8 are indicated as black boxes. Exons of TRPM8 are shown in grey and regions which are not transcribed are shown in stripy.

Splice variant 20b has the SEQ ID NO: 4. The sequence contains of 26 exons. Exons 1-19 and 21-25 are identical to TRPM8. Distinct from TRPM8 splice variant 20b contains an elongated exon 20 with an addition of 127 basepairs immediately beginning after exon 20; exon 26 is truncated: The poly-A signal starts already at basepair 1136. The splice variant avant25 of TRPM8 has the SEQ ID NO: 5. This sequence contains of at least 3 exons with an unknown 3’ end. Avant25 contains an exon of 570 basepairs not found in TRPM8. This exon extends into in exon 25 and 26 but exon 26 is truncated as in splice variant 20b with a poly-A signal beginning at basepair 1136. The splice variant avant13 has the SEQ ID NO: 6. The sequence contains of at least 6 exons with an unknown 3’ end. Avant13 has an exon of 272 basepairs not found in TRPM8. This exon extends into exon 21 but only 20 bases of the 3’end of this exon are within the transcript. The last 4 exons are identical to TRPM8 with a different splice pattern (exons 22, 23, 24 and 26, but exon 26 is truncated with a poly-A signal beginning at basepair 654).

Though it was possible to identify more splice variants, but as they were less abundant and less differentially expressed than those 5 they will not be further discussed here.

The structure of TRPM8 isoforms 16b and 20b is shown in Fig. 25. 16b has one transmembrane spanning domain which results in an extracellular C-terminus. Isoform 20b consists of the same transmembrane spanning domain as TRPM8 but the C-Terminus ends a few amino acids after the last transmembrane spanning domain.

Fig. 25 schematic structure of TRPM8 its isoforms 20b and 16b. Number boxes (grey) indicate tranmembrane spanning domains. The blue horizontal beam represents the cell membrane.

Affymetrix gene chips experiments were analyzed for correlation of TRPM8 expression to the Gleason grading and the TNM staging system. Patients’ PMQ values for TRPM8 and PSA were grouped by staging and grading and the median of each group was calculated. Results showed that TRPM8 mRNA expression increases with Gleason sum from normal tissue to a Gleason sum of 8, but mRNA levels fall to nearly normal levels in high grade tumors (Gleason sum of 9) (Fig. 45 A and B). Correlating TRPM8 expression to the TNM staging system showed similar results. Expression of TRPM8 first rise in early stage tumors compared to normal tissues, but when the tumor has extended through the prostatic capsule into the seminal vesicles (T3) TRPM8 levels fall. Stage 4 cancers, which are characterized by the spreading of the tumor into the bladder neck, show further reduced expression of TRPM8 compared to T3 tumors (Fig. 45 C and D). On the other hand, PSA levels do not show any correlation to either staging or grading.

Fig. 45 Microarray results of TRPM8 and KLK3 in correlation to Gleason sum and TNM staging (pT). Data is shown as the median of PMQ for each patient group. At least 3 patients represent one group.