3  Discussion

Microarrays represent a very effective technique for the simultaneous screening of many genes. They are ideal tools for the identification of new molecular markers and therapeutic targets in cancer and other diseases. In this study a special custom designed array was used to screen for new genes differentially expressed in prostate cancer. It was possible to identify more than 100 genes over- or underexpressed. Comparing these results with published data sets revealed a general congruence: two thirds of the genes found in this study were also found by other prostate cancer profiling studies [Welsh, 01b; Rhodes, 02b; Pilarsky, 03; Magee, 01; Luo, 01b; Dhanasekaran, 01b]. Only a small fraction (less than 15%) was found in this study, but not in others, although the probesets for the genes were present on their chips. This is an interesting finding as one would expect more discrepancies between the studies based on so many different parameters such as patient material, treatment of the material (microdissected or bulk) and the used chip technique (spotted or synthesized). For example, in this study the very time consuming microdissection of the tissue was performed, which but not in the others. However, still similar results were obtained. This leads to the conclusion, that microdissection may not be as important for prostate cancer profiling as previously expected.

A meta-analysis of gene expression profiles in prostate cancer [Rhodes, 02a] showed that the influence of the chip technique used was also negligible as two spotted studies [Dhanasekaran, 01a; Luo, 01a] showed approximately similar results in comparison with a synthesized study of Welsh et al. [Welsh, 01a]. The main reason why each study found approximately 20% varying differentially expressed genes is the fact that in each study different chips were used, representing different genes. Indeed, TRPM8 could only be identified in this study because it was present on the metaGen Cancer-Chip. None of the other studies performed between 1997 and 2001 represented this gene on their chips. In the future gene profiling studies of cancers (or of other diseases) will not come up with many new genes as the human genome is more or less sequenced and approximately 95% of the genes known are present on whole human chips (HG-U133 Plus 2.0) representing 47,000 genes (Affymetrix, Santa Clara, CA ). The next challenge will be the characterization of these genes in a more detailed way, looking closer at splice variants and in a next step at the protein level.

Chip experiments revealed that TRPM8 was overexpressed in 56% of all prostate tumor patients, ranking on position 4 of the most overexpressed genes. In order to approve the results from the microarray experiments Real Time PCR was performed. The analysis showed that TRPM8 was indeed highly overexpressed in prostate cancer patients (64%). The most striking and important feature concerning the potential of TRPM8 as a drugable target was seen in experiments characterizing its expression among diverse normal tissues. Northern and dot blotting as well as in situ hybridization experiments over a wide range of tissues and patients revealed an exclusive expression of TRPM8 in the prostate, with no expression detectable either in any other normal human tissue, or in any other cancer tissue, with notable exception of the very rare neuroendocrine tumors. These findings stand in contrast to results from Tsavaler et al. [Tsavaler, 01b] who indicated that TRPM8 is expressed at least in trace amounts in normal tissues such as testis, lung, breast, thymus and lung. They also showed that TRPM8 is expressed in different other primary cancers, such as melanoma, colorectal carcinomas and breast carcinoma. The reason for this discrepancy may be due to the different probes used for hybridization. Tsavaler et al. used a 342 bp fragment of which they do not specify the exact localization in the TRPM8 gene. But as they isolated this clone from a cDNA library which was prepared from an mRNA pool it is most likely that the fragment is located near the poly-A tail. In first experiments performed in this study using a short probe derived from the 3’ UTR of TRPM8 it was also observed that transcripts- although different in size- were seen in other organs such as liver and brain (Fig. 46 B). When hybridization was done with a 2.7 kb probe of the open reading frame of TRPM8 the exclusive expression in prostate tissue could be seen as shown in Fig. 46 A. This fact explains also the discrepancies in in situ hybridization experiments seen in this study compared to results from Tsavaler et al. However, it can not be excluded that the expression seen was due to some unspecific binding of the probe. On the other hand results from this study and literature search revealed that often many isoforms of a gene exist which exhibit different expression patterns. In a very recent and interesting finding it could be proven that the differential expression of a gene was only due to just one specific exon [Gandini, 03]. They showed that only exon 4 of the prostate cancer antigen 3 (DD3) is differentially expressed in prostate tumors, whereas exons 1-3 are uniformly expressed in both tumor and normal tissue. Interestingly, this gene was the most differentially expressed gene found in this study by microarray experiments (77% overexpressed; Tab. 1). These findings do not show only how carefully one has to choose the probe for hybridization experiments, but it also indicates that a more detailed experiment design will reveal much better insight in the understanding of the complexity of gene expression.

	Fig. 46 TRPM8 Northern blot using 2 different probes for hybridization. A) Northern blot using a 2.7 probe from the 5’-end of the TRPM8 gene. B) Hybridization with a probe from the 3’ end of TRPM8. All other conditions were exactly the same.

TRPM8 belongs to the transient receptor potential family. It has a six membrane spanning domain with both COOH and NH2 termini located intracellularly. In FRET experiments it was possible to demonstrate for the first time that TRPM8 subunits homomultimerize. Most likely they form tetramers as it was shown for several TRP channel such as TRPV1, TRPC4 and TRPV5 [Schaefer, 02b; Strubing, 01]. Coexpression of C-terminal fusion proteins (TRPM8-CFP and TRPM8-YFP) yielded FRET efficiencies of nearly 16%. Comparing these data with published FRET efficiencies, the interaction of TRPM8 subunits is quite strong. For example FRET efficiencies of approximately 9% and 8% were detected for TRPC4α and TRPC4ß, respectively [Schaefer, 02c]. The TRPV1 altered FRET efficiencies of 18.5%.

Two studies analyzing the TRPM8 orthologs in mouse [Peier, 02c] and rat [McKemy, 02c] showed that TRPM8 is activated by cold stimuli and cooling agents such as menthol and icilin. Following these findings it was possible to show that the human TRPM8 could be activated by a cooling agent, too. The human TRPM8 channel could be activated strongly with icilin, which was followed by a large Ca2+ invard current into the cell. The mechanism of TRPM8 activation by icilin is not yet known. Either it is directly activated through conformational changes or by activation through a second messenger pathway [Peier, 02b]. The observations made by Peier et al. and also in this study reveal that TRPM8 is activated immediately after exposure to the agent suggesting a direct gating mechanism.

TRPM8 is the only member of the TRPM8 subfamily for which activation by cooling agents and by cold has been shown. Although all TRPM proteins are Ca2+ permeable channels, their mode of activation is different. For example TRPM2 is activated by H2O2 [Kraft, 04; Wehage, 02], TRPM1 by switching the cells from Ca2+-free to Ca2+ containing medium [Xu, 01c]. TRPM5 could be activated by the depletion of intracellular calcium stores [Perez, 02].

Analysis of TRPM8 expression among different cell lines revealed exclusive expression in the LNCaP cell line. Neither any of the 4 additional prostate cell lines nor any of the other cell lines derived form various human tissues express TRPM8. One of the profound features of the LNCaP is their androgen dependency. Thus it was analyzed whether TRPM8 was regulated by androgens or not. Indeed, Real Time PCR experiments performed on LNCaP cells incubated with different concentrations of the androgen R1881 revealed a 90 times upregulation of TRPM8. This is a new and interesting finding especially as KLK3, the gene coding for PSA, could just be activated by androgens up to 30 times. However, absolute KLK3 mRNA levels in untreated and in treated cells were higher than TRPM8 mRNA levels. This could be many reasons for these findings: 1. KLK3 is additionally co-activated by other factors, 2. the AR activation of transcription of the KLK3 gene is more pronounced, due to a higher specific transcription factor binding site to the KLK3 promoter, 3. Expression of TRPM8 is regulated by specific other transcription factors, or 4. the AR activates the TRPM8 gene to a lesser extent which might be due to less specific TF- binding sites in the TPM8 promoter. A study from this summer supports the finding that TRPM8 is androgen regulated [Henshall, 03b]. They grew the androgen prostate cancer xenograft LuCaP-35 subcutaneously in nude male mice. Tumor bearing animals were castrated and tumors were harvested at several time points after castrations (0 -100 days). They showed that TRPM8 expression levels were high in mice on day 0 – 2 after castration but not 5 -100 days post castration. Further, TRPM8 mRNA expression correlated significantly with KLK3 expression in the same mice (Person P = 0.80).

Collectively, these data suggest that TRPM8 is regulated by androgens, a mechanism by which it could also be regulated in human prostate tumors.

Taking advantage of the specific expression of TRPM8 in the prostate, it was analyzed whether the TRPM8 promoter could be used for gene therapy. The strategy was to clone the TRPM8 promoter in front of the sequence of a certain toxin, such as diphtheria toxin A into a viral vector used for gene therapy. Expression of the toxin would thus be restricted to the prostate as the transcription of the toxin would be under the control of the tissue specific TRPM8 promoter.

In order to analyze whether the promoter could be used following this approach the 1.9 kb fragment of the 5’ flanking region of the transcription start site of TRPM8 was analyzed in silico. It was possible toidentifying multiple tran­scription factor binding sites such as PRX2, NKX3-1, NKX2-5, USF1 and MYCMAX. The most striking feature is the high abundance of binding sites for homeobox genes such as PRX2, NKX3-1and NKX2-5.

In reporter assays cloning the TRPM8 promoter in front of a luciferase reporter gene it could be demonstrated that the 1.9 kb-promoter of TRPM8 was able to activate the transcription significantly, ranging from 3.9 times in HEK293 cells to nearly 14 times in DU145 cells. In LNCaP cells the activa­tion level was about 9 times. This was an unexpected result as in other ex­periments it could be shown that expression of TRPM8 was restricted to the cell line LNCaP. Therefore it was expected that the promoter would be strongest activated in this cell line. Concluding, the identified promoter region does not represent the functional element for prostate specific expres­sion, especially as androgens, in this case R1881 was not able to enhance the basal transcription activation. But as androgens definitely raised transcription of TRPM8 up to 90 fold in R1881 treated LNCaP cells, there must be enhancement through other mechanisms. Either, activation of TRPM8 is regulated by additional enhancer elements upstream or downstream from the transcription start site or the activation is regulated by other mechanism than on the transcriptional level. Most likely the first assumption is correct: It could be shown for the FLOH1, which is largely regulated by androgens, that the basal activation of the promoter was greatly enhanced by a DNA fragment found in intron 3 of the FLOH1 gene[Warren D.W., 03]. Unfortunately they neither describe the basic promoter nor the DNA-piece from intron 3 precisely, but it indicates that activation of androgen responsive genes can be regulated by elements more than several 1000 base pairs away from the transcription start site.

Additionally, in silico analysis of the 30 kb upstream and downstream of the transcription start site of the TRPM8 gene showed many potentially androgen responsive elements (AREs). Although it was not possible to find any classical consensus sequences for AREs, several sequence altering the core base pairs for potential AREs could be identified. Additionally, some of the AREs were present in conserved regions which stress their importance. Further promoter studies including these binding sites will give further insights into the transcriptional regulation of TRPM8.

It was further possible to identify a highly conserved 172-bp element within the TRPM8 promoter, which functions as a transcriptional repressor. This is consistent with previous reports demonstrating that a NKX3-1 element can repress the activity of a basal promoter containing a multimerized NKX3-1 binding site when it was expressed in TSU-Pr prostate cells [Steadman, 00a]. However, this observation was only made when artificial NKX3-1 was co-transfected: When the multimerized NKX3-1-reporter construct was transfected alone, the promoter activity was even enhanced. In this study the repression was independently of exogenous TF indicating that this element is a strong repressor; especially as repression could also be seen in non-prostate cells such as the HEK293 kidney cells. Experiments co-expressing the 172-pGL3 construct with NKX3-1 will show if this could further reduce the activation potential. Another possibility to analyze the function of this repressor would be to exclude the 172-base pair fragment from the 1.9 kb construct. It is likely that this would enhance the activation potential of the promoter.

Microarray analysis of lung cancer patients revealed that one lung carcinoma sample expressed TRPM8. The analysis of the clinical data of all 172 lung cancer samples indicated that this patient was the only one with a 10% neuroendocrine infiltration. As this finding could be pure coincidence, Real-Time PCR was performed including a 100% neuroendcrine tumor (not used in chip experiments) of the lung and as a negative control an adenocarcinoma of the lung. The result was convincing: TRPM8 was more than 60 and 900 times overexpressed in the 100% and 10% neuroendocrine tumors, respectively. Most likely that the normal tissues did not have any expression at all and the trace expression seen was just due to contamination. RT-PCR experiments with cell lines of neuroendocrine origin supported these results: TRPM8 was expressed also in LCC18 (colon), QGP1 (pancreas) and BON1 (pancreas) cells.

The finding that TRPM8 is expressed in neuroendocrine cells (NE) and prostate tissues raises the question, whether both cell types have something in common. For example, is it possible that neuroendocrine (NE) cells are present in the prostate? The answer is yes. NE cells represent, beside the basal and the secretory cells, the third form of epithelial cells of normal prostate epithelium [Abrahamsson, 96]. They are located in all regions in the prostate and are present in normal, hyperplastic and dysplastic prostate tissue. NE cells have a complex appearance with irregular dendrite like processes extending between adjacent epithelial cells. Abrahamsson et al. summarizes the cytological and histological patterns of NE cells as follows:” Ideally, a NE cell is defined as a cell of neuronal or epithelial type that fulfills all or most of the following criteria: it contains secretion granules; its secretion is essentially derected towards the blood, …, and is immunoreactive to antisera against neurone-specific enolase or chromogranin A or other NE markers”.

This “nerve like” appearance could also be observed when LNCaP cells were cultured for a period longer than 10 days in steroid deprived medium (own observation). Therefore it might be possible that TRPM8 is expressed in the prostate from neuroendocrine cells. But, are neuroendocrine cells amplified in prostate tumors, which may solve the question why TRPM8 is overexpressed in neoplastic tissue? The answer is again yes. It was shown that neuroendocrine differentiation in prostatic adenocarcinomas is associated with a poor prognosis [Bostwick, 02]. The most common form observed in prostate carcinomas is a focal neuroendocrine differentiation, which may be pronounced in approximately 10% of adenocarcinomas. Further, it could be shown that NE are positive for a nuclear located (and thus fuctional) androgen receptor [Nakada, 93]. In another study, Singh et al. showed that Cromogranin A, which is a marker for NE cells, is significantly overexpressed in prostate tumors. These findings support the speculation that TRPM8 might be expressed in neuroendocrine cells of the [Seite 62↓]prostate, but double staining of TRPM8 and neuroendocrine markers such as Chromogranin A or serotonin will have to be performed in the future.

The mouse and the rat orthologes of TRPM8 were isolated from RNA of DRG (dorsal root ganglia) and from trigeminal neurons of newborn rats, respectively [Peier, 02a; McKemy, 02d]. In humans TRPM8 it is expressed in the prostate with elevated levels in early stages of prostate cancers. All three orthologes (mouse, rat and human) can by activated by cooling agents such as icilin, menthol or by temperatures below 28°C. Looking at these characteristics, is there a regulation link between expression in prostate cancers and neuronal cells? Most likely, cold or cooling agents are not the biological stimulus of TRPM8 in the prostate. But are androgens expressed in neuronal cells? Most likely they are not. One possibility is that the transcriptional regulation of TRPM8 is completely different in these three species. For example it could be shown in this study that the mRNA of the mouse orthologe misses the first exon of human TRPM8, but alters 4 additional exons at the 5’ end, which so far were not seen in humans. Thus the promoter regulating the TRPM8 gene transcription in human and mouse might be completely different leading to different activation of the transcription.

Another point to be discussed is the question why TRPM8 expression is regulated by androgens, but channel activation occurs through cooling agent such as icilin? This is another point why the TRP family has been denoted as “a very interesting and versatile family” [Montell, 02c] .

In situ experiments on prostate cancer patients in this study revealed that TRPM8 is expressed moderately in all normal prostate cells, strongly enhanced in PIN and in adenocarcinomas. However it was not possible to correlate in situ expression with disease progression; neither the Gleason Grading System nor the TNM System showed any significant correlation to TRPM8 expression. Locally TRPM8 is predominantly expressed in basal epithelial cells, which is conform to results from Tsavaler et al. [Tsavaler, 01a].

Affymetrix gene chips experiments were analyzed for correlation of TRPM8 expression to Gleason grading or TNM staging. Results showed that TRPM8 mRNA expression increases with Gleason sum linearly from normal tissue up to a Gleason sum of 8, but mRNA levels dropped significantly to nearly normal levels in highly undifferentiated tumors (Gleason sum of 9). Correlating expression of TRPM8 to the TNM staging system showed similar results: Expression of TRPM8 first rise in tumors of early stages (N - T2), but when the tumor has extended through the prostate capsule into seminal vesicles (T3), TRPM8 levels fall. In cancers stage 4, when the tumor has spread further into the bladder neck or other nearby tissues, expression drops even more. Collectively, [Seite 63↓]TRPM8 increases at the beginning of the disease, but in very undifferentiated tumors which have extended through the prostate capsule expression decreases significantly. In situ hybridization further supported these findings as it was observed in cases where high grade tumors and low grade tumors were present in the same specimen, that TRPM8 was lost in undifferentiated tumor cells (GGs 9-10).

The next question to be answered was whether TRPM8 also correlated with PSA-relapse after radical prostatectomy. Interestingly this was not the case. In July this year a study was published showing that TRPM8 is a good prognostic marker of PSA-relapse [Henshall, 03a]. They analyzed (using also Affymetrix GeneChip) 17 patients with a PSA relapse after radical prostatectomy, compared to 55 patients who remained free of PSA relapse after surgery. It indicates that patients with low TRPM8 expression prior to radical prostatectomy have a higher risk of getting a relapse of PSA. According to this study until now 8 patients out of 52 had a relapse of PSA, but it was not possible to correlate this to TRPM8 expression. It may be assumed that the number of patients was too small and/or the follow-up time too short in order to find any correlation, but the other study was not much bigger for statistical significance. Thus it remains to prove in larger studies whether TRPM8 is a prognostic marker for PSA relapse.

Prostate specific antigen (PSA) and the digital rectal examination (DRE) are the two major screening parameters for prostate cancer. But neither PSA nor any of the other prostate cancer specific markers available today meet the requirements of a good marker: sensitive, specific, providing prognostic information, and indicating post treatment progression or cure. For example prostatic acid phosphatase (PAP) has been used extensively for diagnosis, staging and monitoring of prostatic cancer in the last century, but it is ineffective for screening of prostate cancer as it has a low positive predictive value, a low specificity and sensitivity. Also PSA, it is specific for the prostate, but not for prostate cancer as it is expressed in all stages of cancer [Caplan, 02a]. It tends to increase with age and rises in men with evidence of benign prostatic hypertrophy. Additionally, preoperative PSA cannot be used to predict capsular penetration or seminal vesicle invasion. Neither PSA nor PAP are able to predict progression in adenocarcinomas of the prostate following radical prostatectomy [Sauvageot, 98a]. But is TRPM8 at better diagnostic marker than PSA? Most likely it is not. 1. TRPM8 mRNA expression rises and falls with progression of the disease, which makes it difficult for the pathologist to distinguish (at least alone) whether the patient alters a tumor of very high or very low grading or staging. 2. The findings in this study were based on RNA data, thus pathological classification would be very time consuming and expensive. Whether TRPM8 expression could be useful for immunhistochemical classification remains to be demonstrated. 3. As TRPM8 is not a secreted protein, it will not be measurable in the blood, excluding it for the usage as a general screening parameter.

On the other hand TRPM8 is an extremely good target for the development of a drug. It was shown in this study that a drug against TRPM8 would only affect prostate tissues and neuroendocrine tumors, as TRPM8 is exclusively expressed in these tissues. The data from this work at least proves that TRPM8 mRNA could be a good target for an antisense drug. Although it was not part of the study to analyze the expression of TRPM8 on the protein level it could be proven in experiments done in parallel within the company that TRMP8 is also highly overexpressed on the protein level in prostate cancer patients. Further, mRNA expression correlated with protein expression in prostate cancer patients. Thus therapeutic approaches using antibody and small-molecule are also possible. But would it be enough to develop a therapeutic that inhibits the function of TRPM8 and thus prevents the cancer from growing or even better diminishes it? Is TRPM8 a promotor of tumor growth? Most likely it is not. In experiments overexpressing TRPM8 in cell lines, the division of the cells was neither enhanced nor did the cells have a different morphology. Thus the development of a therapeutic must focus on the design of a drug which binds to and destroys the tumor cells. That could be achieved either by using modified antibodies delivering toxic or modulatory payloads (small-molecules, radionuclides and enzymes) to the cancer cell or by the specific delivery of a radioactive payload carried on a small-molecule[Chang, 02]. That this approach is feasible has been shown over the years since it was introduced more than 20 years ago

In this study more than 10 splice variants (SV) of TRPM8 could be identified using an in silico approach and RT-PCR experiments. All of them were as differentially expressed as TRPM8 itself, some of them exhibited even a significantly higher differentially expression in prostate tumors. Especially the regulatory RNA, located on the opposite strand of TRPM8, exhibited an 80% overexpression of TRPM8 in prostate tumors as analyzed by RT-PCR experiments. The identification of splice forms of a member of the transient receptor potential family is not unique. It is the fourth TRPM family member for which isoforms have been identified. Up to now splice variants of TRPM1 (MLSN) [Xu, 01b], TRPM2 [Zhang, 03b] and TRPM5 (MTR1) [Prawitt, 00] have been described. In this study splice variant 16b was more closely examined. It has a truncated C-terminus leading to truncated transmembrane domains with loss of the functional pore. Interestingly, in addition to 16b of TRPM8, isoforms of TRPM1 and TRPM2 also have a deletion of the C-terminus. The three short forms alter different numbers of transmembrane domains. TRPM1 is devoid of all, 16b has one and TRPM2 has two transmembrane segments. The short form of TRPM1 is uniformly distributed in the cytoplasm, whereas the long form localizes in the plasma membrane. Published data show that the short form of TRPM1 suppresses the function of the long form by inhibiting the transportation of the long form to the plasma membrane [Xu, 01a]. By contrast the TRPM2 short form does not alter the localization of the long form. It inhibits the function of the long form by an unknown mechanism [Zhang, 03c]. The data in this study [Seite 65↓]show similar results as shown for TRPM2. 16b does not alter the location of TRPM8 (at least not visible in confocal microscopy experiments), but shows -at least partially- an inhibiting effect on the activation of TRPM8 by icilin.

All of the study discribed above were performed in vitro overexpressing both, the long form and the short form through expression-vectors. No study is available which discribes functions of endogenously expressed splice forms of TRP channels. There will be further studies necessary in order to prove that the truncated transcripts are a) endogenously translated into proteins and b) expressed by the cell in a sufficient amount to alter the functions of the longer form. One reason why no publication is available might be the difficulty in finding working antibodies which selectively recognize only the truncated forms. In most cases the splice variants do not differ in a large number of amino acids. Thus the availability of a good epitope is very limited.

What might be the function of all the other identified splice variants? Do they have any functional properties? It is quite difficult to believe that all of the identified transcripts have a distinct function. But if so, does the abundance of spliced transcripts contribute to the malignant transformation? Most likely only a few of them have functional properties. The great majority might be overexpressed just because of diverse disregualtions in cancers.

Interestingly, most of the identified splice variants in this study were isolated from mRNAs derived from tumor tissues. This might indicate that not only the overexpression, but also the diversity of splice variants in tumors is increased. Real-Time PCR experiments as well as dot blot experiments revealed that some of these isoforms are much stronger differentially expressed as TRPM8 itself. That makes them excellent markers for the detection of prostate cancer.

Alternative splicing has been shown to be tightly regulated in a tissue and developmental- specific manner [Nissim-Rafinia, 02b]. Therefore, changes in the absolute or relative expression of isoforms are expected to effect cellular functions which might be a contributing factor or cause to the development, progression or maintenance of cancer. Indeed, for many genes dramatic changes in alternative splicing patterns are associated with neoplasia and metastasis [Nissim-Rafinia, 02a; Philips, 00]. In example WT1 (Wilms tumor), CD44 (renal, lung, gastric and urothelial cancers), BCL2 (prostate, lymphoma and gastric cancers) and FGFR2 (prostate cancers) are alternatively spliced genes found in cancers. BCL2L1 (formerly Bcl-x) a member of the BCL2 family, produces two alternative splice forms, one having a pro-apoptotic and one having an anti-apoptotic effect [Boise, 93]. It remains to find out what the reason for the upregulation of TRPM8 splice forms in prostate cancers could be. Four explanations are possible: First, mutation in the basal cis-splicings sides might be mutated. Second, mutation in auxiliary ISE/ISS or ESE/ESS elements might lead to the inappropriate expression of the isoforms. Third and fourth, trans-acting factors of the basal and the auxiliary splicing machinery are defective.

Most likely the overexpression of TRPM8 splice variants is due to changes in trans-acting factors of the auxiliary splicing machinery as mutations in cis-elements of the splicing machinery can not explain the immense diversity of [Seite 66↓]splice variants as demonstrated especially in splice variant 16b. Point mutations in the basal or auxiliary cis-elements would only lead to a few alterations in splicing, thus only a handful of new isoforms would be expected to be seen. Mutations in proteins functioning in the basal trans-system of the splicing machinery would cause all genes to be alternatively spliced, which is not very likely as this mutation would be lethal for the organism. Support for the theory of the alterations in the auxiliary trans-splicing machinery is coming from literature. It has been shown that tumor cells express high levels of a broad spectrum of a group of splicing factors which are members of a conserved family of proteins. These factors bind to the active sites of RNA polymerase II transcription and thus function as key regulators of alternative RNA splicing, whereas preneoplasias often express only a sub-set of the family [Stickeler, 99]. The major group of these proteins belongs to Serine/arginine-rich (SR) proteins [Zahler, 92b; Fu, 95a; Graveley, 00b]. They have dual functions and serve as splicing enhancer or splicing repressor proteins, depending on where they bind in the pre-mRNA reviewed in [Akusjarvi, 03].

Although the majority of the isoforms of TRPM8 expressed in prostate tumors are most likely due to splicing defects resulting in unnatural splicing with no biological relevant functions, some of those splice variants might be of biological significance. In Northern blot experiments two isoforms of TRPM8 could be identified: one longer transcript as TRPM8 with a length of 7.3 kb and a shorter one with a length of 4.1kb, but neither of these matches the length of the transcripts identified in prostate tumors.

In conclusion it can be said that the splice variants found in this study are extremely good markers for the detection of prostate cancer. They would be good targets for a pay load based therapeutic approach. Their potential for being a target for a small-molecule has to be analyzed in further studies, as their contribution to malignant transformations has not yet been proven.