4 Discussion

4.1 Identification of the S100A14 cDNA

↓150

This study was undertaken in order to identify tumour-associated genes with potential application in the detection or treatment of the disease. For this purpose, suppression subtractive hybridization (SSH) was performed using cDNA synthesized from normal human bronchial epithelial cells and cells derived from a metastatic small cell lung carcinoma (Difilippantonio et al., 2003).

↓151

SSH was developed by Diatchenko and colleagues (Diatchenko et al., 1996) providing a powerful approach for identification of genes differentially expressed in one cell population compared to another. The technique has been successfully used in many experimental settings and two important applications are differential gene expression in multiple tumour systems and identification of tumorigenesis relevant genes. Although cDNA microarrays are increasingly used for the global analysis of gene expression, SSH is still widely applied. It not only permits identification of tumour-associated genes with known function but also unbiased isolation of novel sequences that are not yet available on microchips. Moreover, it enables the recovery of abundant as well as low-copy-number mRNA transcripts.

Using this approach, we recovered a set of genes differentially expressed in lung tumour cells compared to normal lung cells. These genes then provided the groundwork for identifying unknown transcripts with diagnostic and prognostic value for cancer. In this set of cDNAs, it was then confirmed that one of the under-represented transcripts, with no homology to any known gene, was differentially expressed in a panel of lung tumour cell lines. Hence, we decided to analyse this transcript more extensively.

The isolated full-length cDNA is 1067 bp in length and encodes a putative protein of 104 amino acids. The predicted protein contains the S100-specific EF-hand calcium-binding domain and shares the highest sequence homology with S100A13. We therefore assigned this unknown transcript as a novel member of the S100 gene family and designated it S100A14 (Pietas et al., 2002).

↓152

The S100 protein family forms a growing subfamily of proteins related by Ca2+-binding motifs to the EF-hand Ca2+-binding protein superfamily. The members of this family are involved in Ca2+-, Zn2+-, and Cu2+-dependent regulation of many cellular processes, including tumour development and acquisition of metastatic phenotype (Donato, 2001; Heizmann et al., 2002). The homology to the protein family known to be associated with tumorigenesis prompted us to investigate the involvement of S100A14 in tumour biology.

Further analysis of the cDNA sequence revealed that the S100-specific Nterminal EF-hand of S100A14 consists of 13 amino acids. This is in contrast with the 14 amino acid loop characteristic for the S100 family. However, it should still be a functional calcium-binding domain since the critical Glu45 is present (Kawasaki et al., 1998). The carboxyl terminal canonical EF-hand, also referred to as the high affinity calcium-binding site, seems to be mutated as only 2 out of 6 conserved amino acid residues are present. The presence of the N-myristoylation site at the N-terminus of S100A14 poses the possibility of an interaction with a membrane receptor or with the lipid bilayer itself. In contrast to other S100 protein family members the deduced protein contains an extended hydrophilic loop at the N-terminus spanning at least 17 amino acids. The calculated molecular weight and isoelectric point for the S100A14 protein are in accordance with characteristic features of the S100 family as small acidic proteins. Also the structural organization of the protein based on hydropathy plots is similar to other members of the family.

4.2  S100A14 is Differentially Expressed in Human Tumours

Expression of S100A14 was found in epithelial cells of a variety of tissues including colon, thymus, kidney, liver, small intestine, lung, breast, cervix, ovary, uterus, pancreas, prostate, rectum, stomach and thyroid gland, most of them having mainly an epithelial-parenchymal phenotype. In contrast, most mesenchymal-stromal tissues like skeletal muscle, white blood cells and spleen are negative (Fig. 5). This is largely consistent with our database searches indicating many S100A14-related ESTs present in libraries of normal as well as tumour tissue such as squamous cell carcinoma of the skin, colon, stomach and pancreas adenocarcinoma as well as uterus and lung cancers. Normal tissues were additionally represented by head and neck as well as placenta. In contrast, we did not observe detectable mRNA expression in placenta even after prolonged exposure indicating that the expression level of the protein must be very low (data not shown). We found mouse S100a14 transcript mainly in libraries of embryonal origin.

↓153

We were primarily interested in identifying transcripts differentially expressed in a wide range of tumours. Differential expression of S100A14 was therefore confirmed by Northern blot analysis of various immortalized and tumour cell lines from different tissues. S100A14 was abundantly expressed in cells cultured from normal human epithelial cells, moderately expressed in immortalized epithelial cells, and absent from most of the examined tumour cell lines. The relatively high S100A14 expression in most of the colon tumour cell lines could be due to its high basal expression level in normal colon tissue. These results indicate that altered S100A14 expression occurs in most of the human tumour cell lines. Therefore, we hypothesized that its expression might be also suppressed in primary tumour tissue.

Using the BD Biosciences array with 241-matched tumour/normal cDNA samples from individual patients we determined the expression status of S100A14 in 10 different human tumour types. The geneis preferentially overexpressed in ovary, breast, and uterus tumours, and mainly underexpressed in kidney, rectum, and colon tumours. Most notably, increased rather than decreased expression of S100A14 has been found in lung tumours. The reason for this discrepancy is unclear, although it may be due to differences between the cell model system and the primary tissue. This highlights the need to supplement cell culture-derived transcript abundance data with transcriptional or immunohistochemical analysis of primary material. Thus, using immunohistochemistry, we sought to determine S100A14 protein expression in lung tumour specimens. Additionally, we examined breast tumour samples since this tumour type showed significant S100A14 overexpression in our transcriptional analysis.

Whilst normal lung and breast epithelial cells express S100A14 at a low level, we observed strong tumour cell-localized cytoplasmic and membranous S100A14 protein expression in over 60% of the analysed tumour samples. Thus, our analysis on the expression of S100A14 in primary tumours differs from the cell culture data by demonstrating a consistent and strong overexpression of this gene in multiple lung and breast tumour samples. We therefore speculated that the expression of the gene might be modulated by extrinsic factors or by the microenvironment in which the cancer cells reside.

↓154

The influence of cell culture conditions on gene expression patterns of cultured cell lines is a longstanding concern. Cells grown in culture have unlimited access to nutrients under conditions most favourable for growth and proliferation and only little exposure to extrinsic factors that modulate growth and differentiation, e.g. cytokines. In contrast, cells in a tumour growing in a host tissue environment face conditions with more limited nutrients and oxygen and are subjected to or benefit from a wide variety of host factors. Recently, global gene expression profiling of tumour cells grown in vitro versus the same cells transplanted into an in vivo environment (nude mice) have been performed (Creighton et al., 2003). This revealed specific up-regulation of one set of genes related to cell growth and proliferation when in culture and a different set of genes related to cell adhesion, extracellular matrix, growth substances, and neovascularization when developing as an in vivo tumour.

Our expression analysis of nine S100A14-negative lung tumour cell lines transplanted into immunodeficient mice as well as spheroides of 2 ovary tumour cell lines did not reveal a general re-expression of the gene (Fig. 12). Thus, our results cannot imply an essential physiological role of the tumour environment for S100A14 expression. The interpretation of this finding remains to be elucidated.

Our expression analysis also suggests that up-regulation of S100A14 is not a universal feature of epithelial cell tumours. A potential mechanism for the apparently contradictory role of S100A14 in renal, colon and rectum cancer could be context-dependent. Since most genes act within networks, differences in S100A14 expression pattern may arise from differences in the expression profile of its interacting proteins as well as differential availability of signalling pathways involved in its regulation in various cell types. A growing number of genes regulated by the relative contributions of specific signalling pathways in a particular cell type have been revealed in the recent years as exemplified by S100A2 and maspin. S100A2 was found to be frequently down-regulated early in lung cancer development (Feng et al., 2001). Supporting these data, a number of expression studies implicated this gene as a breast tumour suppressor gene (Liu et al., 2000; Wicki et al., 1997). However, contrary to the proposed role of S100A2 as a tumour suppressor protein, it was found to be overexpressed in gastric and ovarian tumours (ElRifai et al., 2002; Hough et al., 2001).

↓155

Maspin (SERPIN B5) was also found to be down-regulated in breast and prostate cancer and was suggested to be a suppressor of metastasis (Seftor et al., 1998). Paradoxically, it was reported that the gene was strongly expressed relative to normal tissue in pancreatic tumours (Maas et al., 2001) and lung tumours (Heighway et al., 2002). The seemingly contradictory results of S100A14, S100A2 and maspin suggest that these and other genes perhaps not only behave differently in particular tumours, but also that they may play different roles in various tissue types and perhaps even within distinct components of a tissue.

Significant up-regulation of S100A14 in lung and breast tumours directed our attention towards its clinical significance in these tumours. Most interestingly, we found that S100A14 overexpression was significantly associated with ERBB2 overexpression in breast tumours.

The ERBB family of receptor tyrosine kinases (RTKs) couples binding of extracellular growth factor ligands to intracellular signalling pathways regulating diverse biologic responses, including proliferation, differentiation, cell motility, and survival (reviewed in Prenzel et al., 2001; Olayioye et al., 2000; Marmor et al., 2004)It consists of four receptors: ERBB1 (also called epidermal growth factor receptor – EGFR or HER1), ERBB2 (HER2/neu), ERBB3 (HER3), and ERBB4 (HER4). All the family members have in common an extracellular ligand-binding domain, a single membrane-spanning region and a cytoplasmic tyrosine kinase domain. A family of ERBB ligands, the EGF-related peptide growth factors, have been characterized, including epidermal growth factor (EGF), transforming growth factor-α (TGF-α), amphiregulin, heparin-binding EGF-like growth factor, betacellulin, epiregulin, and neuroregulins (NRG-1,-2,-3,-4). The binding affinity of the EGF-like ligands to various ERBB receptors differs as is their potency to induce signalling. No direct ligand for ERBB2 has yet been discovered. However, increasing evidence suggests that the primary function of ERBB2 is as a co-receptor. In fact, ERBB2 is the preferred heterodimerization partner for all other ERBB family members and plays a role in the potentiation of ERBB receptor signalling.

↓156

The ERBB RTKs have a broad expression pattern on epithelial, mesenchymal, and neuronal cells. Signalling through these receptors plays a critical developmental role in inductive cell fate determination in many organ systems. This is exemplified by the perinatal (ERBB1) or early embryonic lethality (ERBB2, 3, and -4) of knock-out mice as a result of insufficient heart and nervous system development (Burden and Yarden, 1997). Furthermore, ERBB RTKs are involved in mammary gland development during puberty and pregnancy.

Ligand binding drives receptor dimerization leading to the formation of both homo- and heterodimers. Dimerization consequently stimulates the intrinsic tyrosine kinase activity of the receptors and triggers autophosphorylation of specific tyrosine residues within the cytoplasmic domain. Each ERBB receptor displays a distinct pattern of C-terminal autophosphorylation sites. These phosphorylated residues serve as docking sites for signalling effector molecules containing Src homology 2 (SH2) or phosphotyrosine binding (PTB) domains. Examination of the binding preferences of different effector proteins revealed a great deal of overlap in the signalling pathways activated by the four ERBB receptors. All ERBB family members apparently utilize the mitogen-activated protein (MAP) kinase pathways as the major signalling routes.

A wealth of clinical data demonstrates the importance of ERBB receptors, in particular ERBB1 and ERBB2, in multiple human cancers, including lung, colon, breast, prostate, brain, head and neck, thyroid, ovarian, bladder, gliomas, and renal carcinoma (Yarden and Sliwkowski, 2001). Several phenomena are resposible for hyperactivation of ERBB receptors in tumours, including overexpression, amplification, and constitutive activation of mutant receptors or autocrine growth factor loops. Most notably, ERBB2 is overexpressed in 20-30% of breast and ovarian tumours and its overexpression correlates with tumour chemoresistance and poor patient prognosis, yielding a median survival of 3 years, compared with 6-7 years when unassociated with ERBB2 (Miles et al., 1999; Witton et al., 2003).

↓157

Due to their frequent overexpression in various cancers and their high signalling capacity ERBB receptors are promising targets for therapeutic intervention in human cancers. Monoclonal antibodies raised against several epitopes of the ERBB1 and ERBB2, as well as EGF and TGF-α blocking antibodies, have been validated in the clinic as an ERBB-directed therapeutic approach. Herceptin, a monoclonal antibody that targets the extracellular domain of ERBB2, was the first target-selective drug raised against an oncogenic cell surface receptor and therefore represents the first example of a new era of anti-cancer therapy. This antibody is now applied in the treatment of metastatic breast cancer patients.

The immunohistochemical detection of ERBB2 in our study (29.3%) is in keeping with the reported frequency of the protein overexpression in breast cancer. In contrast to a number of previous reports, however, ERBB2 expression was not related to oestrogen receptor negativity in our study.

Immunohistochemical analysis of primary lung and breast tumours revealed a significant association between the subcellular distribution of S100A14 and the protein abundance. The majority of S100A14-overexpressing tumours displayed both plasma membrane and cytoplasmic localization of S100A14 relative to low-expressing tumours showing predominantly cytoplasmic staining. Also, there was a significant correlation between higher-grade (grade 3) lung tumours and S100A14 cytoplasmic and plasma membrane localization. Moreover, subcellular distribution of S100A14 was related to lung tumour histology. Both cytoplasmic and plasma membrane-localized staining were associated with squamous cell carcinomas compared to adenocarcinomas detecting cytoplasmic-only S100A14.

↓158

In contrast to this tumour cell plasma membrane localization, normal human breast and lung epithelial cells showed exclusively diffuse cytosolic S100A14 expression (Fig. 8 and Fig. 11). These data suggest that translocation of S100A14 to the plasma membrane could be unique to cancer cells over normal lung and breast epithelial cells and could represent the disease-associated state.

Many S100 proteins translocate to different cellular compartments in response to elevated intracellular Ca2+ and Zn2+ level and this process was shown to be cell type specific (Hsieh et al., 2004). Exposure of several different cell lines to elevated Ca2+ and Zn2+ level did not affect the subcellular localization of S100A14 in our study, suggesting that fluctuations of divalent cations do not regulate S100A14 protein distribution. The prevalence and the clinical relevance of the plasma membrane relative to cytoplasmic S100A14 distribution remains to be established in other tumour types. It is conceivable that the subset of lung and breast cancers that show both plasma membrane and cytoplasmic localization define a molecularly different mechanism of regulation on the protein.

Analysis of cultured cells demonstrated predominantely plasma membrane-localized endogenous S100A14 (Fig. 14). Most notably, prominent staining at cell junctions was detected in several tumour cell lines suggesting a potential role in establishing cell-cell contacts. Cellular fractionation of 9442 cells and testing the different fractions with anti-S100A14 antibody confirmed the immunofluorescence data. S100A14 can be detected in both membrane-bound and soluble forms, although the soluble protein is present only in very small amounts (Fig. 15). Analysis of cells transiently transfected with epitope-tagged protein showed predominantely cytoplasmic and perinuclear localization of the S100A14 protein (Fig. 13). This result, however, requires further confirmation by cellular fractionation. A challenge of future investigations will be to address whether the sites of action of S100A14 are in the cytosol or in membranes.

4.3 Identification and Characterization of the Genomic Locus of S100A14

↓159

Numerous studies have described frequent alterations of the chromosomal region 1q21 in tumours where the human S100 gene cluster is located. In order to elucidate the rationale for the differential S100A14 expression in tumours, we defined its chromosomal localization as well as the nature of its genomic locus in size, structure, and sequence of its introns and exons. Furthermore, we analysed its immediately upstream regulatory region to get some insights into the regulation of the gene.

The human S100 gene family encodes a set of structurally related proteins sharing a common genomic organization of three exons and two introns, with the first exon being short and untranslated and the second and third exons containing the coding sequence (Fig. 35). Exceptions are S100A5 consisting of four exons with exon three and four being the coding ones, S100A4 containing an additional alternatively spliced untranslated exon, S100A11 with the coding sequence beginning already in the first exon, and S100A16, which is composed of four exons and three introns with coding sequence in the third and fourth exon and the first and second exon being alternatively transcribed.

Unlike other S100 genes, S100A14 contains an additional intron that interrupts the sequence encoding the N-terminal part of the protein. Lack of nucleotide sequence homology to any of the S100 family members and lack of strong conservation of the genomic structure may indicate that S100A14 is less closely related to other members of the S100 family.

↓160

Fig.   35 Generic S100 gene structure. A typical S100 gene, e.g. S100A1 is composed of three exons (boxes) with exon 1 being not translated (open boxes) and exons two and three containing the coding region (black boxes). Exceptions to this general rule are depicted below with straight lines in S100A4 indicating alternative splicing and arrows indicating the translational start.

An Alu element belonging to the “young” Alu subfamily (AluY) was identified in the direct proximity of the polyadenylation signal in the 3’ UTR reaching beyond the genomic locus of S100A14. This family of repetitive DNA has been implicated in the stimulation of homologous and non-homologous recombination and triggering chromosomal rearrangements (Deininger and Batzer, 1999). The frequent involvement of the highly conserved 26-bp core sequence of Alu elements in gene rearrangements suggested that this sequence might represent a recombinational hotspot (Rüdiger et al., 1995). The Alu element present in the genomic locus of S100A14 contains the 26-bp core sequence and therefore might contribute specifically to gene rearrangements.

We mapped S100A14 to the chromosome band 1q21 where 16 other S100 genes are tightly clustered (Heizmann et al., 2002). This region is known to be involved in structural and numerical aberrations in various human tumours (Weterman et al., 1996; Gendler et al., 1990). Our Southern blot analysis of S100A14-negative as well as S100A14-positive lung cancer cell lines, however, so far indicates that the S100A14 gene itself is not affected by rearrangements. S100A14 was present in all tumour cell lines examined and showed no detectable deletions or gross rearrangements (Fig. 18). Thus, it seems likely that the differential expression observed in tumour cell lines is not driven by dramatic gene deletions or gross rearrangements. This, however, does not completely rule out a possible correlation of S100A14 expression with the frequently observed chromosome 1q21 rearrangements in solid tumours.

↓161

It has been previously shown that the chromosome region 1q21-q22 contains extended regions of high CpG island density (Wright et al., 2001). Consistent with this observation are reports of site-specific methylation associated with transcriptional repression in the promoter region of S100A2 in tumorigenic cells (Wicki et al., 1997; Lee et al., 1992). Examination of the 5’ upstream region as well as introns of S100A14 did not reveal any CpG island. Similarily, we did not observe re-expression of the gene in eight lung cancer cell lines on treatment with 5aza-2'-deoxycytidine as determined by Northern blot and RT-PCR analysis (data not shown).

In the course of our studies on the genomic organization of S100A14 , we identified the S100A14 promoter fragment, located within 511 bp upstream of the transcription initiation site, and containing constitutive promoter elements (TATA box) that might support basal transcription at this promoter.

The primary aim of our promoter analysis was to determine whether the aberrant regulation of S100A14 at the promoter level might be responsible for its altered expression in tumours. Therefore, we chose for the transient expression analysis HEK 293 cells, which are well established in promoter assays but do not express endogenous S100A14 , and Lovo cells, which have a high level of S100A14 mRNA. Another reason for selecting these cell lines was the high transfection efficiency with standard procedures. Using a luciferase reporter system, a panel of deletion mutants covering the entire putative promoter fragment was constructed in the promoterless pGL3 Basic vector.

↓162

Transient luciferase expression experiments indicated the presence of a core promoter located within 196 bp immediately upstream of the major transcription initiation site (Fig. 21). The 4.5-fold activation of the promoter fragment in HEK 293 and Lovo cells compared with the control vector-transfected cells is in accord with many previously published promoter analyses supporting the significance of our data (Perrais et al., 2002; Nichols et al., 2003; Diaz-Guerra et al., 1997; Bordonaro et al., 2002). Besides a TATA box and an incomplete CCAAT box, consensus motifs for transcription factors located within the promoter region included Elk-1, RFX-1, SRF, AP-1, CEBP, CHOP, USF, NFκB, and N-myc, many of which are associated with proliferative response.

Site conservation is a good indicator of its functional importance and comparative genome sequence analysis (phylogenetic footprinting) can eliminate up to 90% of false binding-site predictions (Wasserman and Sandelin, 2004). Therefore, we compared the region immediately upstream of the mouse orthologue with the 511-bp fragment of the human S100A14 promoter. The analysis revealed the presence of two fragments immediately upstream from the transcriptionstart site that were conserved in the mouse and the human gene (Fig. 22). They encompassed the TATA box and the CAAT sequence. Additionally, a potential NFκB consensus motif was conserved in the mouse and the human DNA within the minimal promoter fragment but further upstream of the two highly conserved promoter fragments. No other sequences matching the further upstream consensus transcription factor binding sites related to mitogenic response could be detected.

Thus, the strong conservation between the mouse and the human S100A14 promoter fragment concerns only constitutive promoter elements that probably contribute to the basal activity of the promoter. It is therefore conceivable that the respective homologous promoter fragment is present further upstream in the mouse promoter or that the regulation of the mouse S100a14 is different from the regulation of the human counterpart.

↓163

A potential NFκB transcription factor binding site was conserved in the mouse and the human promoter fragment. NFκB has been reported to be involved in TNF- α -induced expression of other S100 genes (Joo et al., 2003). Therefore, we have analysed the potential role of NFκB in the regulation of the S100A14 transcription. We found that NFκB could not transactivate S100A14 by co-transfection experiments with the NFκB subunit p65 and the S100A14 luciferase reporter. Similarly, treatment of cells with TNF-α, which induces degradation of IκB and translocation of NFκB to the nucleus, did not enhance the S100A14 promoter-driven reporter gene transcription. Moreover, TNF-α did not induce S100A14 expression at the mRNA level in Northern blot analysis. Thus, we found no evidence that NFκB transactivates the S100A14 promoter fragment.

Taken together, our analysis of two cell lines although differing in S100A14 expression yielded the same activation pattern of the analysed promoter fragment. This means that the transcription regulatory region that we identified is most probably not the critical determinant of S100A14 expression in tumours. Based on these findings, we suggest that a further distal promoter or enhancer confers induction to S100A14 and possibly influences its expression in tumours.

4.4 Oncogenic Signalling Pathways Mediate S100A14 Transcriptional Induction

S100A14 is differentially expressed in many human tumours and its expression is enhanced in lung and breast tumours (Pietas et al., 2002). No data pertinent to the molecular mechanisms responsible for the regulationof the S100A14 gene are available.

↓164

Based on our initial finding that serum stimulation enhanced S100A14 expression in H322 cells, we reasoned that this effect could be mediated by growth factors. In this context, the additional finding of positive correlation with ERBB2 overexpression in breast tumours raised the possibility that epidermal growth factor (EGF) might function as a positive regulator of the gene. To elucidate mechanisms whereby S100A14 expression is enhanced in lung tumours, we studied the effects of EGF on S100A14 expression in 9442 human immortalized bronchial epithelial cells as well as the signal transduction pathways that trigger its expression.

EGF induced a significant increase in S100A14 mRNA expression in a time- and dose-dependent manner in 9442 cells but not in S100A14-negative cell lines (Fig. 25 and Fig. 26). This result implies that a low EGF concentration in culture media cannot account for the absence of S100A14 transcript in the examined negative cell lines. The EGF effects on S100A14 expression were also detected in another cell line examined in this study – HMEB, indicating similar transcriptional regulation in immortalized breast epithelial cells (Fig. 24).

Moreover, treatment of 9442 cells with transforming growth factor-α (TGF-α), a member of the EGF ligand family that binds to and activates the ERBB receptor, also enhanced S100A14 transcript with similar kinetics to EGF, thus confirming the involvement of ERBB receptor signalling in growth factor-induced S100A14 expression (Fig. 27).

↓165

TGF-α is an essential mediator of oncogenesis and malignant progression. Its overexpression in transgenic mice leads to hyperplasia as well as malignancy in pancreas and breast tissues (Sandgren et al., 1990). TGF-α also acts as a strong collaborator in promoting carcinogenesis by other oncogenes (Sandgren et al., 1993) and chemical carcinogens (Takagi et al., 1993). Among the ERBB family members, TGF-α binds only to the ERBB1 receptor and is best known as an autocrine stimulatory growth factor (Riese et al., 1996). Strong expression of TGFα has been found in 67% of pulmonary adenocarcinomas (Tateishi et al., 1991) pointing to an important role in non-small cell lung cancer (Rusch et al., 1997; Fontanini et al., 1998).

S100A14 was also induced in response to fresh medium or “stimulation medium” showing, however, an earlier and less potent maximal response relative to the response induced by EGF (Fig. 28). The less potent response in comparison to that induced by stimulation with EGF could be due to the lower EGF concentration present in the medium than the concentration applied in our EGF-stimulation experiments (50 ng/ml). In addition, other unknown growth factors could contribute to this more rapid stimulation.

Regulatory promoter regions responsive to EGF or PMA could not be identified in this study, neither in the 511-bp fragment nor in the minimal S100A14 upstream regulatory fragment. This suggests the existence of other cooperative elements present at more distant site or in the intervening sequences of the S100A14 gene.

↓166

The analysis of S100A14 expression at the protein level did not reveal any induction following EGF treatment (Fig. 29). Moreover, treatment with U0126 and AG1478 inhibitors also had no impact on the S100A14 protein level as determined by Western blotting of whole cell- and fractionated cell-extracts. These findings raise the possibility of an additional post-transcriptional regulation of the gene.

Regulatory elementsmodulating mRNA half-life are often found within 3' untranslatedregions (3' UTR) (Chen et al., 1995; Hollams, 2002). They include AU-rich elements(AREs) that are associated with stabilization-destabilization of themRNA or hairpin (stem-loop) structure. Binding of a trans-acting factor to these elements influences the turnover of the mRNAor its translational efficiency (Baker et al., 2000; Ranganathan et al., 2000). AREs are found in the 3' UTR of many mRNAs that code for proto-oncogenes, nuclear transcription factors, and cytokines e.g. COX-2, p21Waf1, TNF-α, interferon-β, GM-CSF, c-myc, and c-fos (Dixon et al., 2000; Giles et al., 2003; Wang et al., 1997; Kruys et al., 1989; Han et al., 1990; Levine et al., 1993). The 644-bp 3' UTR of S100A14 accounts for over 60% of the transcript length suggesting that it might be a target of post-transcriptional regulation.

Although, long tracts of the highly conserved AREs could not be found in the 3’ UTR of S100A14, we cannot exclude that the few poorly conserved AREs found in the 3’ UTR of S100A14 confer post-transcriptional control to its mRNA. Moreover, it is possible that sequestrationof some factors involved in translation that are present in limiting amounts may be responsible for the observed inhibition of translationof the S100A14 mRNA.

↓167

The development of efficient and selective inhibitors against protein kinases involved in signalling has made significant progress. These inhibitors provide a suitable alternative to transfection experiments using dominant-negative mutants of protein kinases, in particular for cells which cannot be transfected efficiently in culture with standard procedures. In the present study, we used several inhibitors to characterize the signalling events involved in EGF-induced expression of S100A14 ( Fig. 36 ).

Fig.   36 Pharmacological modulation of the ERBB-induced signalling pathways

In order to determine the effects of the inhibitors on the signalling cascades we first carried out an analysis of the kinetics of the mitogen-induced response for the three MAPK signalling pathways (ERK1/2, p38, JNK) and the PI3K cascade in 9442 cells.

↓168

Growth factor stimulation caused a significant increase in phosphorylation and activation of ERK1/2, p38, and JNK kinases but had no effect on the activation of AKT, suggestive of a constitutive activation of the PI3K pathway in these cells (Fig. 30). The EGF-induced S100A14 expression was inhibited by mitogen-activated protein (MAP) kinase kinase (MEK) inhibitor U0126 but not by p38, JNK, and PI3K pathway inhibitors (Fig. 32). Inhibition of ERBB receptor tyrosine kinase activity by AG1478 inhibitor also resulted in a reduction of S100A14 mRNA levels to those seen in unstimulated cells, thus confirming the results obtained with the MEK1/2 inhibitor. Taken together, these inhibitor studies demonstrate that the activation of ERBB tyrosine kinase by its ligands EGF and TGF-α, leads to the synthesis of S100A14 mRNA via the activation of the RAS-RAF-MEK-ERK signalling pathway.

The importance of constitutively activated mitogenic signalling pathways in oncogenesis is well established. Binding of extracellular growth factor ligands couples the ERBB family of receptor tyrosine kinases to intracellular signalling pathways regulating diverse biologic responses including proliferation, differentiation, cell motility, and survival, all of them implicated in tumorigenesis. The RAS-RAF-MEK-ERK pathway is at the heart of signalling networks that govern these processes. It may be noted that about 30% of human tumours carry RAS gene mutations that may prolong the activation of mitogenic signalling pathways (Khosravi-Far and Der, 1994). Of the three genes in this family (composed of KRAS, NRAS, and HRAS), KRAS is the most frequently mutated member in human tumours, including adenocarcinoma of the pancreas (~70-90% incidence), colon (~50%), and lung (~25-50%) (Pellegata et al., 1994; Bos et al., 1987; Mills et al., 1995).

Thus, multiple components of mitogenic signal transduction pathways are either overexpressed (ERBB receptor family and ligands, and cyclin D1) or mutated (KRAS) in cancer, leading to excessive activation of these important growth-modulating cascades. We therefore suggest that the involvement of ERBB receptor signalling and the RAS-RAF-MEK-ERK pathway in S100A14 transcriptregulation links the gene to oncogenic signalling pathways in cancer. Identification of target genes of ERBB signalling such as S100A14 may, therefore, prove important for a clearer understanding of the output of the ERBB module and consequently contribute to future ERBB-directed approaches that are more beneficial for cancer treatment.

↓169

S100 family members have been specifically related to the ERBB status in breast tumorigenesis including S100A4 and S100P (Perou et al., 2000; Rudland et al., 2000; Sorlie et al., 2001; Mackay et al., 2003). S100A4 is a direct target of ERBB2 signalling in medulloblastoma cells and levels of ERBB2 and S100A4 are tightly correlated in samples of primary medulloblastoma (Hernan et al., 2003). The induction involves phosphatidylinositol 3-kinase, AKT and ERK1/2 kinase.

In this context, it should be noted that signalling by ERBB receptors is accompanied in many cell types by a transient increase in the cytosolic concentration of Ca2+. This process is mediated by phospholipase Cγ-protein kinase C (PLCγ-PKC) pathway (Marmor et al., 2004). PLCγ phosphorylation by ERBB1 and ERBB2 results in its activation. PLCγ hydrolyzes phosphatidylinositol 4,5-biphosphate to generate the second messengers diacylglycerol and inositol triphosphate. Binding of inositol triphosphate to receptors on the endoplasmic reticulum results in Ca2+ release, which lead to the activation of calcium/ calmodulin-dependent protein kinases and phosphatases, including Pyk2 and calcineurin. Furthermore, Ca2+ and diacylglycerol activate protein kinase C (PKC), resulting in the phosphorylation of a large variety of substrates. PKC also mediates a feedback attenuation of the signalling by ERBB receptors due to its capacity to phosphorylate ERBBs and to inhibit the receptor’s tyrosine kinase activity (Welsh et al., 1991).

ERBB receptors also play an important role in the regulation of cell motility and cytoskeletal reorganization mainly by the recruitment of Ca2+ and phosphatidylinositol 4,5-biphosphate (Feldner and Brandt, 2002). Many experimental data support the notion that PLCγ is the most important molecule in the ERBB2-mediated migratory/invasive ability of certain tumour cells (Brandt et al., 1999). Interestingly, ERBB1 homodimers and ERBB1:ERBB2 heterodimers differentially modulate the time course of PLCγ activation thereby generating different patterns of oscillations in Ca2+ level (Dittmar et al., 2002). EGF-induced breast tumour cell migration was attributed to a transient, rather than sustained, activation of PLCγ due to ERBB2 signalling.

↓170

There are multiple ways how EGF signalling could have an effect on the expression level of S100A14: 1) direct transcriptional regulation, 2) influence on the mRNA stability or 3) indirectly through other transcription factors/signalling pathways. We wished to determine which of these mechanisms is responsible for EGF-mediated S100A14 transcriptional activation. Although the effect of EGF on the stability of S100A14 mRNA was not examined in this study, cycloheximide – an inhibitor of protein synthesis – partially blocked S100A14 induction by EGF, indicating requirement for de novo protein synthesis (Fig. 33). Therefore, we suggest that S100A14 is an indirect transcriptional target of EGF signalling.

Treatment with the phorbol ester PMA enhanced S100A14 expression in 9442 cells (Fig. 34A). PMA is a potent tumour promoter as well as a growth regulator of many different cell types (Hunter and Karin, 1992). It activates protein kinase C (PKC), a ubiquitous family of serine/threonine kinases. These kinases play key regulatory roles in a multitude of cellular processes, including proliferation, apoptosis, differentiation, cell migration, and adhesion (Mellor and Parker, 1998; Lafon et al., 2000; Zhao et al., 2000). The PKC family consists of at least 11 isoforms. Specific isoforms are activated, depending on isoform, by Ca2+, phospholipids or diacylglycerol generated by phospholipase Cγ (PLCγ) or phospholipase D (PLD) from phosphatidylinositol 4,5-biphosphate.

We used bisindolylomaleimide I which acts as a competitive inhibitor for the ATP-binding site of PKC to inhibit PKC in 9442 cells. Bisindolylmaleimide I shows high selectivity for PKC-α, -β1, -β2,-γ, -δ, and -ε isozymes with a ranked order of potency α>β1>ε>δ (Martiny-Baron et al., 1993). Pretreatment with the inhibitor did not significantly affect PMA-induced S100A14 expression in 9442 cells (Fig. 34B). The reason for this is unclear considering that the concentration of the inhibitor we applied was shown by others to block completely the activity of PKCs in bronchial epithelial cells (Reibman et al., 2000; Graness et al., 2002). Furthermore, the range of phorbol ester-sensitive isoforms of PKC (PKC-α, -β1, β2, -γ, -δ, -ε, -η, -θ) virtually matches that of isoforms inhibited by bisindolylmaleimide I. Nevertheless, we cannot exclude that the PKC isotypes that are not inhibited or only weakly inhibited by the inhibitor are responsible for S100A14 induction.

↓171

We next examined whether EGF-induced stimulation of S100A14 was mediated by PKC. Preincubation of cells with bisindolylomaleimide I does not prevent the EGF-induced stimulation of S100A14 indicating that this induction is not mediated by PKC, in agreement with the finding that EGF activates the ERK1/2 cascade in a PKC-independent manner, via the RAS-dependent pathway (Boulikas, 1995), see Fig. 34D.

Previous studies have demonstrated that acute treatment with phorbol esters leads to a rapid activation of ERK MAPK in most cell types (Rossomando et al., 1989). Since PKC is the major target for these tumour promoters, it has been implicated in the activation of the ERK MAPK pathway and the consequent triggering of cellular responses such as differentiation and proliferation (Mischak et al., 1993; Murray et al., 1993). Accordingly, it has been demonstrated that members of all three groups of PKC isoforms (conventional, novel, and atypical) are able to activate upstream elements of the ERK MAPK pathway, including RAS and RAF-1, and the mechanism of activation shows some isotype specificity (Schönwasser et al., 1998; Marais et al., 1998).

We found that PMA indeed induces activation of the ERK1/2 signalling pathway in 9442 cells (Fig. 34C). Moreover, MEK1/2 inhibitor – U0126 inhibited PMA-induced S100A14 expression indicating that the ERK1/2 MAPK pathway is the signalling pathway that targets S100A14 expression in response to PMA (Fig. 34B).

↓172

Protein kinase C overexpression is associated with increased tumorigenicity and metastatic potential in several experimental models, and its activity is increased in tumours of breast and lung as compared with their normal counterparts (Blobe et al., 1994; O’Brian et al., 1989; Clegg et al., 1999). The essential role of PKC in processes relevant to neoplastic transformation and tumour invasion renders it a potentially suitable target for anticancer therapy (Basu, 1993). PKC-α, an important tumour-promoting factor is currently being studied as a target in treating patients with cancer.

In summary, our results demonstrate that PMA exerts stimulation of S100A14 via PKC activation in 9442 cells. We also show that the PMA-induced S100A14 expression in 9442 cells requires activation of the ERK1/2 MAPK pathway. Overexpression of PKC in lung and breast tumours could therefore contribute to the enhanced expression of S100A14 in these tumours. Our data also indicate that both PKC- and ERBB-dependent signalling converge on the ERK1/2 cascade to regulate the expression of S100A14 and are probably necessary for its full activity (Fig. 37).

Fig.   37 Schematic representation of the signalling pathways leading to S100A14 up-regulation in response to EGF and PMA in 9442 cells

↓173

A recent study of the protein arylation targets in human bronchial epithelial cells dosed with 1,4-benzoquinone (BQ) and 1,4-naphthoquinone (NQ), common tobacco smoke and environmental pollutants, revealed that the S100A14 protein was one of the major cellular targets (Lamé et al., 2003).

Many quinonoid compounds are reactive electrophiles capable of causing cellular injury by two mechanisms: 1) direct disruption of the function of critical proteins or regulatory pathways, or 2) secondary activation of the immune system by the modified protein. For quinonoid compounds, a third mechanism might be considered where the adducted protein acts as a platform for quinone redox cycling producing reactive oxygen species that in turn damage the protein and its surrounding environment, including oxidative DNA damage. The toxicity of arylating quinones can be magnified by the redox cycling properties of quinone metabolites, e.g. glutathione (GSH) conjugates.

Specifically, Lamé and his collegues, found that S100A14 protein formed substantial amounts of adducts with BQ and NQ, as well as with GSH-BQ on two cysteine containing peptides (residues 1-5 and 68-78), providing evidence for a cycle of oxidation of the resulting BQ-GSH conjugate metabolite followed by attachment to the protein. Formation of such protein adducts is expected to increase the residence time of redox cycling substances and therefore induce single and double strand breaks in DNA.

↓174

The finding of the modification of S100A14 protein by quinones is the first report, to our knowledge, implicating one of the putative functions of this protein. Future research aimed at identifying target proteins of S100A14 needs to be done to determine further functions of this novel gene and elucidate its role in neoplastic transformation.

In summary, by analysing a human lung tumour cell line subtraction cDNA library, we have identified and characterized a novel member of the human S100 gene family that we designated S100A14. It encodes an mRNA that is ubiquitously expressed in normal human tissues of epithelial origin. We demonstrated that S100A14 transcript is down-regulated in many immortalized and tumour cell lines from different tissues. In contrast, studies on human primary tumours, including lung and breast, revealedpredominantly up-regulation of the gene at the mRNA and protein level.

We mapped the S100A14 gene to a region of chromosomal instability on human chromosome 1q21 and subsequently resolved the gene structure of S100A14 in human by demonstrating its organization of four exons and three introns spanning a total of 2165 bp of genomic sequence. By analysing the 5’ upstream proximal region of the S100A14 gene, we identified and characterized the minimal promoter fragment of the gene.

↓175

Studies on regulation of S100A14 in 9442 immortalized bronchial epithelial cells identified it as a growth factor-inducible gene that is induced by epidermal growth factor (EGF) and transforming growth factor-α (TGF-α). EGF-mediated transcriptional induction of S100A14 depends on extracellular signal-regulated kinase (ERK1/2) signalling and requires de novo protein synthesis. In support of these findings, we demonstrated co-existence of ERBB2 overexpression and S100A14 protein accumulation in primary breast tumours.

Further studies identified phorbol ester 12-myristate 13-acetate (PMA) as an activator of S100A14 in 9442 cells suggesting regulation by protein kinase C (PKC). The PMA-induced S100A14 expression is mediated by activation of the ERK1/2 signalling cascade indicating that both PKC- and ERBB-dependent signalling converge on the ERK1/2 cascade to regulate the expression of S100A14. Considering the frequent incidence of aberrant activation of the ERK1/2 and PKC signalling pathways in tumours as well as their oncogenic potential we suggest that it is the impaired regulation of these signalling cascades that couples S100A14 to malignant transformation.

The challenge of future investigations will be to define the binding partners of S100A14 to determine the function of this novel gene in cancer, and to further elucidate the functional importance of ERBB- and PKC-dependent signalling for the S100A14 gene regulation.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
XDiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
04.03.2005