3 Results

To identify new potential cancer diagnostic or therapeutic targets, suppression subtractive hybridization (SSH) transcript profiling was performed between normal human cultured bronchial epithelial cells (HBE) and the metastatic small cell lung carcinoma cell line H526 (Difilippantonio et al., 2003).

The quality of the SSH library was confirmed by Northern blot analysis, indicating that 69% (236/342) of the clones were differentially expressed in the tumour cell line. In an initial evaluation we used a limited panel of cell lines including HBE cells, as well as lung tumour cell lines 2170, H526, and D51, representing the three major subtypes (squamous cell carcinoma, adenocarcinoma, and small cell lung carcinoma). Thirty-one cDNAs scored as differentially expressed in at least two of the tumour cell lines were analysed more extensively on an extended panel of cell lines including small airway epithelial cells (SAE), four immortalized HBE cells, and 32 lung cancer cell lines. The DNA sequences of isolated cDNA fragments were then determined and compared with those deposited in the GenBank database.

Based on our interest in further examining those transcripts with no homology to any known gene, 6 partial cDNAs representing transcripts that were abundantly expressed in normal lung epithelial cells but were absent or under-represented in the majority of lung tumour cell lines were selected for further analysis. These cDNA fragments were sent to the German Resource Center of the Human Genome Project (RZPD, Heidelberg, Germany) for screening of human cDNA libraries enriched in full-length cDNAs. Three of those cDNA fragments represented the same unknown gene. Screening of the human lung, trachea, and primary keratinocyte cDNA libraries yielded 13 positive clones out of which 9 were confirmed positive by Southern blot analysis. Out of 5 clones containing the longest inserts determined by NotI/SalI-restriction digest, clones DKFZp404H1043, DKFZp404B1018, and DKFZp719J2437 were identified as candidate full-length cDNA clones. Clone DKFZp404H1043 (designated H1043) with cDNA insert of 1057 bp was chosen for further characterization. The insert size was in accordance with the transcript size estimated by Northern blot analysis.

On the nucleotide level no homology to any known gene was found. Analysis of the predicted protein sequence revealed an S100-specific EF-hand Ca²⁺-binding domain as well as a high homology to the S100A13 calcium-binding protein. Therefore, we designated the transcript S100A14 and submitted it to the NCBI GenBank database under Accession No NM_020672 (LocusID LOC57402).

To verify the full-length cDNA sequence of S100A14 we performed 5’ RACE on human colon cDNAs since this organ showed the most abundant expression of the transcript in our Northern blot analysis. It provided additional 5’ UTR sequence, not present in the H1043 clone. The predominant cDNA product was 1067 bp in length.

The cDNA was predicted to contain an open reading frame (ORF) of 104 amino acids (nt 109-423), a 5’ untranslated region (UTR) of 108 bp, and a 3’ UTR of 644 bp. The putative ATG initiation codon was embedded in a strong Kozak consensus sequence (CACCATGG) (Kozak, 1996) and was preceded by an in-frame stop codon. A polyadenylation signal (nt 1028-1033) and a poly(A)-tail beginning after nucleotide 1053 oriented the fragment and provided further evidence for the predicted ORF. The nucleotide sequence and the predicted open reading frame of S100A14 are shown in Fig. 3.

Computer-assisted analysis of the deduced amino acid sequence revealed an S100-specific EF-hand Ca²⁺-binding domain at the N-terminus (aa 33-45) consisting of 13 amino acids. The second putative canonical EF-hand motif has been deduced from the secondary structure and hydrophobicity analysis of the protein. It contained only 2 out of 6 conserved residues (Asp76 and Glu80) and might start at Gly72 ending at Ser83. In addition, the protein was predicted to contain a N-glycosylation site (aa 75-78), a protein kinase C phosphorylation site (aa 94-96), 5 casein kinase II phosphorylation sites (aa 16-19, 42-45, 73-76, 7780, 83-86), and a N-myristoylation site (aa 2-7).

The calculated molecular weight of S100A14 is 11 662 Da and the isoelectric point is 4.99. Protein secondary structure analysis indicated two helix-loop-helix structural motifs characteristic of calcium-binding sites. There was an extended hydrophilic loop at the extreme N-terminus in contrast to other S100 family members.

	Fig.   3 Nucleotide and deduced amino acid sequence of the human S100A14 gene. The predicted start codon and polyadenylation signal are underlined. An asterisk indicates the putative stop codon. Numbers to the left correspond to the single-letter amino acid sequence. Numbers to the right correspond to the nucleotide sequence.

The deduced protein of S100A14 shares 68% similarity and 38% identity to S100A13 (Accession No Q99584) being the closest human match. Significant similarity was also found with S100A4 (62% similarity; 30% identity), S100A2 (60% similarity; 30% identity), S100A10 (58% similarity; 31% identity), and S100A9 (55% similarity; 34% identity), as illustrated in Fig. 4 . Furthermore, in the SwissProt database we identified an apparent mouse orthologue, 1110013O05RIK protein (Accession No Q9D2Q8), showing 97% similarity (93% identity) using the human S100A14 amino acid sequence as a query. A comparison between the S100A14 cDNA and RIKEN 1110013O05 (1110013O05Rik) cDNA (GenBank Accession No NM_025393) revealed 87% identity.

	Fig.   4 Alignment of the predicted amino acid sequences of the human S100A14 and the mouse orthologue 1110013O05RIK with the most homologous S100 family members. Homology is represented by shading of the amino acid identities. Identical amino acid residues are indicated as black boxes whereas similar amino acids are indicated as gray boxes.

To characterize S100A14 expression in tumour cell lines from other tissues we extended the expression analysis to normal human cultured mammary epithelial cells, four immortalized epithelial cell lines, and 31 different human tumour cell lines as well as ten mouse and rat cell lines. The complete list of the examined cell lines is given in Table 2 and Table 3.

Table  2 Expression of S100A14 mRNA in human cancer cell lines. S100A14 transcript level was examined by Northern blot and RT-PCR analysis. The expression level was determined by semiquantitative scoring as follows: (+++) strong expression, (++) moderate expression, (+) weak expression, (RT-PCR) transcript not detectable by Northern blot but detected by RT-PCR, (–) negative expression. The tumour cell lines histopathology: SCLC, small cell lung cancer; ADC, adenocarcinoma; SCC, squamous cell carcinoma; LCLC, large cell lung cancer; Ca, carcinoma.

The S100A14 mRNA is most abundant in cells cultured from normal human epithelial cells, but is absent from the vast majority of cancer-derived cell lines. Interestingly, most of the immortalized human cell lines also show significant S100A14 mRNA expression. The S100A14 mRNA level is relatively high in most of the colon tumour cell lines. We did not detect S100a14 in any mouse or rat cell line examined using a mouse S100a14 cDNA sequence as a hybridization probe.

Table  3 Expression of S100A14 mRNA in other mammalian cell lines

We examined the expression pattern and transcript size of S100A14 in normal human tissues by Northern blot hybridization using a cDNA probe corresponding to the complete protein coding region. The transcript of approximately 1.1 kb was expressed most abundantly in colon ( Fig. 5 A). Moderate levels of S100A14 mRNA were detected in thymus, kidney, liver, small intestine, and lung. A very low expression level was detected in heart where a slightly longer transcript of approximately 1.35 kb was observed. No expression could be detected in brain, skeletal muscle, spleen, placenta, and peripheral blood leukocytes. Additionally, S100A14 was expressed in normal breast, ovary, prostate, rectum, stomach, thyroid, and uterus as determined by hybridization to the Cancer Profiling Array ( Fig. 5 B).

To confirm the mRNA expression data, we analysed a panel of normal human tissues by immunohistochemistry using a polyclonal anti-S100A14 antibody raised in rabbit. The S100A14 protein expression was predominantly associated with various epithelia as well as gland cells ( Table 4 ). The expression pattern was virtually consistent with the S100A14 mRNA expression in normal human tissues. The exceptions were brain and placenta, showing a weak expression of S100A14 protein, in contrast to the absence of the transcript in Northern blot analysis.

	Fig.   5 Expression of the S100A14 mRNA in normal human tissues A:A human 12-lane Multiple Tissue Northern Blot (BD Biosciences) containing 2 µg poly(A)+ RNA in each lane was hybridized with a 32P-labelled S100A14 cDNA probe. -actin was assayedas a loading control. B: Expression of S100A14 mRNA in additional normal human tissues examined by hybridization to the Cancer Profiling Array (BD Biosciences).

These results suggest that there is a very low S100A14 expression in these tissues and a not sufficient sensitivity of Northern blot to detect low abundant transcripts.

Table  4 Expression of the S100A14 protein in normal human tissues

Since S100A14 mRNA was not expressed in most of the tumour-derived cell lines, we hypothesized that its expression might be suppressed also in primary tumour tissue. To examine its expression in primary tissue we applied a commercially available Cancer Profiling Array, which contains cDNAs from 241 tumours and corresponding normal tissues from individual patients. This analysis, however, revealed a varied expression of the gene in different tumours ( Fig. 6 A). Significant overexpression was detected in ovary (78.6%), breast (62%), and uterus (45.2%) tumour tissues. Also lung (38.1%), prostate (25%), and thyroid gland (16.7%) tumours showed up-regulated levels of S100A14 transcript. In contrast, marked underexpression was found in kidney (70%), rectum (50%), and colon (40%) tumours and to lesser extent in stomach tumours (18.5%). The expression status in cervix, small intestine and pancreas tumours could not be evaluated because of the limited number of samples available on the array. The schematic presentation of S100A14 expression in normal and tumour tissues is shown in Fig. 6 B.

	Fig.   6 Expression profile of S100A14 in human tumours A: The Cancer Profiling Array containing SMART-amplified cDNA from tumour (T) and corresponding normal (N) tissues from individual patients was hybridized with 32P-labelled S100A14-specific probe. Hybridization to ubiquitin cDNA (top and bottom) and E.coli DNA (middle) is shown on the right side of the array. Human cancer cell lines (no detectable signal) are spotted on the right. B:The schematic presentation of S100A14 expression in normal and tumour tissues determined by hybridization to the Cancer Profiling Array.Data are expressed as percentage of up-regulated, down-regulated and normal samples and normalized to all spots on the membrane to correct for RNA quantity and integrity. The grey bars represent normal expression level, the black and white bars denote up-regulated and down-regulated expression, respectively. Signal intensities were quantified after several exposure times to compensate for differences in mRNA expression between different organs.

Our transcript expression analysis in primary tumour tissues suggested that S100A14 is not a simple suppressor of tumour cell growth as might have been concluded from the cell line model system. It is possible that its altered expression is a result of, or otherwise associated with, tumour progression. To shed light on the association of S100A14 protein with malignant transformation and to verify the RNA expression data we used immunohistochemistry to investigate its expression in tumour tissues.

S100A14 protein was examined in 120 histological sections of human lung cancer and 10 specimens of normal human lung epithelium. The cohort of tumour samples included 62 cases of adenocarcinoma (51.7%) and 58 cases of squamous cell carcinoma (SCC) (48.3%). The small number of samples precluded a separate analysis of large cell lung cancer (LCLC) tumour specimens. We observed a low cytoplasmic S100A14 staining in normal lung epithelium (score 0 to 1+; Fig. 7 ). Therefore, we approved a scoring of 2+ and 3+ as indicative of protein overexpression for our analysis. Scores 0 or 1+ were considered as physiologically normal expression.

	Fig.   7 S100A14 protein is overexpressed in primary lung tumour tissue in comparison to normal lung tissue. Immunohistochemical staining of S100A14 protein in tissue sections of normal lung pneumocytes (left, indicated by an arrow) and lung squamous cell carcinoma (right). Specimens were stained with an anti-S100A14 antibody. Photographs were taken with a 40x objective.

Positive staining was assessed not only as the degree of staining intensity but also as the proportion of cells with staining of both plasma membrane and cytoplasm as compared with those of cytoplasmic staining alone.

S100A14 immunoreactivity in tumour tissue scored positive in 68.3% (82/120) of cases thus validating its up-regulation in this tumour type ( Fig. 7 ). The S100A14 staining was restricted to the cancerous epithelial cells of the tissue. Next, S100A14 expression was analysed in relation to clinicopathological criteria. No significant correlation was found between S100A14 positivity and gender, age, histological subtype, tumour size, number of positive lymph nodes, and histological grade.

We found that 57.3% (47/82) of S100A14-overexpressing lung tumours displayed cytoplasmic and membranous expression relative to exclusively cytoplasmic S100A14 staining in normal lung epithelium (p = 0.006; Fig. 8).

	Fig.   8 S100A14 localizes to plasma membrane in lung tumour tissue. Immunohistochemical analysis of S100A14 protein in sections of normal bronchial epithelial cells (left) and lung squamous cell carcinoma (right). Panels show high power microscopy (magnification, 40x), demonstrating cytoplasmic staining of S100A14 in normal bronchial epithelial cells (left, indicated by an arrow) and distinct plasma membrane-associated staining of S100A14 in lung carcinoma cells (right).

Moreover, we observed a significant association (p = 0.048) between higher-grade tumours (grade 3) and S100A14 subcellular distribution. 50% (27/54) of high-grade tumours displayed S100A14 protein localized both to the cytoplasm and the plasma membrane. Another relation was found between lung tumour histology and subcellular distribution of S100A14. Cytoplasm and plasma membrane-localized S100A14 staining was associated with 61.8% (34/55) of SCC (p = 0.02) compared to 38.2 % (20/58) of adenocarcinoma tumours, the latter detecting predominantly cytoplasmic S100A14 (65.5%; 38/58). S100A14-negative tumours were not included in the statistical analysis of the subcellular distribution of S100A14. These results are summarized in Table 5 .

Table  5 Association of S100A14 protein expression in lung tumours with clinicopathological factors

A significant up-regulation of S100A14 mRNA in breast tumours prompted us to investigate this tumour type by immunohistochemistry. We examined S100A14 protein expression in 93 breast tumour samples, 14 lymph node metastases and 10 normal mammary epithelia. In our cohort of breast tumours were 77 invasive ductal carcinomas (IDC), 14 invasive lobular carcinomas (ILC) and 2 cases of mixed type (IDC+ILC).

Normal mammary epithelium stained rather weakly for S100A14 protein with cytoplasm-localized staining (score 1+; Fig. 9 ). Therefore we applied the same scoring as for lung tumours, i.e. scores 0 to 1+ were considered as normal physiological expression, whereas scores of 2+ and 3+ were assessed as positive and indicative of protein overexpression. S100A14 positivity was detected in 68.8% (64/93) of breast tumour samples and 71.4% (10/14) of lymph node metastases ( Fig. 9 ). Clinical relevance of S100A14-overexpressing tumours was investigated only in IDC cases due to small sample number of other tumour subtypes. No significant correlation with gender, age, histological subtype, tumour size, number of positive lymph nodes or histological grade was observed.

	Fig.   9 S100A14 protein is up-regulated in primary breast carcinoma relative to normal breast tissue. Immunohistochemical analysis of S100A14 protein in normal breast ductal epithelial cells (left) and ductal carcinoma epithelial cells (right). Specimens were stained with an anti-S100A14 antibody. Photographs were taken with a 20x objective.

Interestingly, we found a significant correlation between a high S100A14 level and ERBB2 overexpression (p = 0.04). 82.9% (29/35) of ERBB2-overexpressing tumours (score 2+ and 3+) had up-regulated S100A14 expression as well ( Fig. 10 ). To further substantiate this finding, we separately examined 31 ERBB2-overexpressing breast tumours of score 3+. We found that 90.3% (28/31) of ERBB2-positive tumours were also S100A14-positive.

	Fig.   10 S100A14 overexpression correlates with ERBB2 overexpression in primary breast tumours. Immunohistochemical analysis of a section of a breast ductal carcinoma specimen stained both with anti-ERBB2 antibody (left) and (the same section) with anti-S100A14 antibody (right). Photographs were taken with a 40x objective.

Oestrogen receptor-α (ER-α) negativity was also significantly associated with elevated levels of S100A14 (p = 0.024). 87.5% (21/24) of ER-α-negative tumour specimens were S100A14-positive. Moreover, a significant association to increased tumour cells proliferation index (MIB) was observed (p = 0.033). 100% (11/11) of tumours assessed as highly proliferating (60-100% of actively proliferating cells) stained positively for S100A14.

Also in this case the subcellular distribution of S100A14 was related to S100A14 protein abundance. 79.7% (59/74) of S100A14-overexpressing tumours displayed prominent plasma membrane and cytoplasmic localization of S100A14 (p < 0.01) relative to 34.8% (8/23) of low-expressing tumours showing predominantly cytoplasmic staining (65.2%; 15/23). In contrast to this tumour cell plasma membrane localization, normal human breast epithelial cells showed solely diffuse cytosolic S100A14 expression (Fig. 11). For the statistical analysis of subcellular localization of S100A14 we omitted the S100A14-negative samples. These results are summarized in Table 6 .

	Fig.   11 S100A14 localizes to plasma membrane in breast cancer tissue. S100A14 immunostaining in sections of normal breast ductal epithelial cells (left) and breast ductal carcinoma sections (right). Panels show high power microscopy (magnification, 40x), demonstrating cytoplasmic staining of S100A14 in normal breast ductal epithelial cells (left, indicated by an arrow) and plasma membrane-associated staining of S100A14 in breast carcinoma cells (right).

Our transcriptional and protein analysis demonstrated that expression of S100A14 in primary tumours differed from that in cell line model; the gene was significantly overexpressed in multiple lung and breast tumour samples. Nevertheless, our data also confirmed the differential expression of the gene in a wide range of primary tumours as well as its potential clinical relevance with regard to the positive correlation with ERBB2 overexpression in breast tumours. Therefore, we decided to further characterize the gene and the molecular determinants of its regulation.

Table  6 Association of S100A14 protein expression in breast tumours with clinicopathological factors

To determine if the absence of S100A14 in the vast majority of cultured tumour cell lines could be due to differences between the cell model system and the primary tissue, we examined 9 mouse xenograft tumours of human cultured lung cancer cell lines. They represented the following histopathological types: SCLC (CPC-N, H526, H446, H82, H209, N417), brain metastasis of SCLC (Colo 668), adenocarcinoma (D54), and LCLC (D97). These cell lines do not express endogenous S100A14 in Northern blot analysis.

RT-PCR analysis did not reveal re-expression of the S100A14 transcript in the majority of the examined xenografts ( Fig. 12 ).

	Fig.   12 S100A14 is not re-expressed in lung cancer cell line xenografts from nude mice. Nine S100A14-negative human lung cancer cell lines (CPC-N, H526, H446, H82, H209, N417, Colo 668, D54, and D97) were transplanted subcutaneously into immunodefficient mice. Mouse xenograft tumours (MT) and the respective lung cancer cell lines (CL) were examined by RT-PCR. GAPDH was used as a loading control.

An enhanced S100A14 mRNA expression was clearly present only in the D54 xenograft tumour as compared with the corresponding cell line. A very weak mRNA expression was detected in the H82 cell line and the corresponding xenograft as well as in the D97 cell line. A modest mRNA expression was observed in the H209 cell line and a weak expression in the corresponding xenograft.

In an attempt to achieve a better understanding of the biology of the S100A14 protein, we examined its subcellular localization by immunfluorescence. Initially, we expressed the protein as carboxyl-terminal fusion of V5-epitope, and determined expression of the fusion protein after transient transfection by confocal laser-scanning microscopy. H157, A549, and COS-7 cells were chosen for their absence of S100A14 transcript in Northern blot analysis.

Distinct subcellular localizations were detected in all analysed cell lines. Immunoreactivity was observed in the cytoplasm with a higher expression around the nucleus in A549 and COS-7 cells ( Fig. 13 A, B). H157 cells displayed a pronounced immunoreactivity in the perinuclear area ( Fig. 13 C). Control incubation with mock-transfected cells was completely negative ( Fig. 13 D). Expression of the protein was also confirmed by Western blot analysis ( Fig. 13 E).

To confirm that the localization studies performed with epitope-tagged protein corresponded to the subcellular distribution of endogenous protein, we applied the polyclonal antibody against the endogenous S100A14 protein. Patterns of subcellular distribution and expression of S100A14 protein were examined in immortalized lung cells (9442, 2078), colon tumour (Caco-2, HCT-116, Lovo, WiDr), and lung tumour (H322) cells.

9442 and 2078 cells displayed strong plasma membrane immunoreactivity of sharp and clear pattern with the exception of cell protrusions where more diffuse staining was observed ( Fig. 14 A, B). Additionally, a very weak and diffuse cytoplasmic S100A14 staining was detectable. This staining pattern is consistent with our analysis of fractionated proteins in 9442 cells (see below).

	Fig.   13 C ellular localization of S100A14 - V5 protein in human lung tumour and COS-7 cells. A: A549 (lung tumour), B: COS-7 (monkey kidney fibroblasts), and C: H157 (lung tumour) cells were transiently transfected with the S100A14-V5expression vector. Forty-eight hours after transfection immunofluorescent analysis was performed using confocal laser scanning microscopy. Cells were examined for localization of S100A14-V5after indirect staining with monoclonal anti-V5 primary antibody and anti-mouse FITC-conjugated secondary antibody. Magnification using 60x oil immersion objective. D: Control incubation with mock-transfected cells was completely negative. E: Expression of the protein was confirmed by immunoblotting analysis. Vector-transfected cells were used as a negative control. Equal loading was determined by immunoblotting with anti-β-actin antibody.

Moreover, strong S100A14 expression was observed in actively proliferating cells ( Fig. 14 C). Interestingly, staining of distinct vesicles of unknown origin was also observed in these cells ( Fig. 14 D). H322 and HCT-116 cells, which grow in colonies, expressed S100A14 in similar patterns ( Fig. 14 E, F). Immunofluorescent staining in these cells revealed distinct immunoreactivity at cell junctions with occasionally stronger plasma membrane expression relative to neighbouring cells indicating a heterogeneous expression. Weak and diffuse cytoplasmic staining was also observed. In contrast, Caco-2 cells had diffuse rather than sharp staining at cell junctions and moderately strong and homogenous cytosol immunoreactivity ( Fig. 14 G). Prominent membranous staining was observed in Lovo cells with focally stronger expression ( Fig. 14 H). Overall staining was low in WiDr cells showing heterogenous expression at cell junctions (data not shown). Control incubation with preimmune serum was completely negative ( Fig. 14 I).

	Fig.   14 Subcellular localization of the endogenous S100A14 protein in immortalized lung cells, and lung and colon tumour cells Immunofluorescence microscopyof A: 2078, B, C, D: 9442 (immortalized lung epithelial cells), E: H322 (lung tumour), F: HCT-116, G: Caco-2, and H: Lovo. Cells were fixed and stained with the polyclonal anti-S100A14 antibody and with the secondary anti-rabbit FITC-conjugated antibody. Magnification using 60x oil immersion objective. I: Control incubation with preimmune serum was completely negative.

To verify the data obtained by immunofluorescence, we fractionated whole-cell extracts into membrane and solublecomponents. Soluble and crude membrane fractions of 9442 cells were prepared by hypotonic lysis, Dounce homogenization, and differential centrifugation. Western blot analysis with anti-S100A14 antibody revealed that S100A14 was predominantly associated with the membrane fraction, whereas only very little S100A14 protein was detected in the soluble form ( Fig. 15 ). These data indicate that the cells endogenously expressing S100A14 preferentially target the protein to the plasma membrane.

	Fig.   15 The S100A14 protein is distributed through the cytoplasmic and membranous fractions of 9442 cells. Total proteins from 9442 cells were fractionated. For SDS-PAGE, 80 µg of protein extract from each fraction was used for loading. Immunoblotting was performed with a polyclonal anti-S100A14 antibody. The specific S100A14 band was detected in the cytoplasmic (CY) and membranous fraction (ME). α-tubulin and RAS were assayed as loading controls for cytoplasmic and membranous fraction, respectively.

It was of interest to determine whether divalent cations could affect the subcellular localization of S100A14. To characterize the dependence of the subcellular distribution of S100A14 on the intracellular Ca²⁺ and Zn²⁺ levels, we treated 9442, 2078, Caco-2, WiDr, Lovo, HCT-116, and H322 cells with thapsigargin in conjunction with 1 mM CaCl₂ as well as with ZnCl₂. Thapsigargin increases intracellular Ca²⁺ levels by blocking endoplasmic reticulum calcium pumps. We observed predominantly plasma membrane anti-S100A14 staining under all conditions tested, with no detectable alterations in distribution pattern.

Most of the human S100 genes are clustered on the chromosome 1q21, a region which is frequently involved in chromosomal rearrangements and deletions in human cancers. In order to identify the mechanisms of abnormal S100A14 expression in tumours we sought to determine its genomic structure and localization.

To obtain S100A14 genomic sequence data, human RPCI 1,3-5 PAC blood genomic libraries (RZPD) were screened with the purified insert of the full-length cDNA clone of human S100A14 (H1043). We obtained 18 positive clones that we re-screened by PCR using several cDNA-derived primer pairs: H1043f, H1043ex1f, H1043ex2f, H1043ex3f, and H1043r.

We identified three candidate positive clones, RPCIP704 I22230Q2A, RPCIP704 A12752Q2, and RPCI P704 D11609Q2 (designated I22230, A12752, and D11609), and subjected them to a second screening by PCR to confirm their positive status. Following assembly of the genomic sequence of S100A14 based on the Celera’s human genome-unassembled fragments database (Venter et al., 2001), we designed the following PCR primers located in the flanking region of the exon/intron junctions of the relevant putative introns: H1043i1for, and H1043i1rev; H1043i2for, and H1043i2rev; H1043i3for, and H1043i3rev. The three genomic PAC clones were confirmed positive by this screening. Southern blot analysis revealed an approximately 3-kb region of overlap among the three clones that were positive for hybridization with the S100A14 cDNA and intronic probes ( Fig. 16 A). The restriction enzyme-digested fragments covering that region were subcloned and sequenced to identify the genomic organization of S100A14 ( Fig. 16 B).

In total, we obtained 3173 bp of continuous genomic sequence. In addition to the genomic locus of the S100A14 gene, it included part of the putative 5’ upstream promoter sequence and the 3’ downstream region of the gene. It was consistent with the compiled sequence identified in Celera’s human genome-unassembled fragments database using S100A14 cDNA as a query. The sequence was submitted to GenBank under Accession No AF426828.

Comparison of the genomic and cDNA sequence revealed that the gene spanned 2165 bp. It was organized in four exons and three introns ( Fig. 17 ). Exon 1 is untranslated, exon 2 contains the translation initiation codon and the putative Nmyristoylation site, and exon 3 codes for the N-terminal EF-hand motif. The latter exon contains the C-terminal EF-hand motif and the translation termination codon followed by 630 bp of 3’-untranslated sequence. All the intron/exon boundaries confirmed to the GT-AG rule ( Table 7 ).

	Fig.   16 Southern blot analysis of the genomic PAC clones positive for S100A14 Southern blot analysis ofthe candidate S100A14-positive clones, RPCIP704 I22230Q2A, RPCIP704 A12752Q2, and RPCI P704 D11609Q2. A: Restriction analysis with BamHI, XbaI, KpnI, XhoI, and SalI restriction endonucleases revealed a ~3-kb region of overlap (rectangle) among the three clones that were positive for hybridization with the S100A14 cDNA and intronic probes. B: The BamHI-digested fragments covering that region (rectangles) were subcloned and sequenced. Genomic PAC clone K11781, which was negative by PCR, was assayed as a negative control. λHindIII-digested DNA was used as a molecular size marker.

We also identified a candidate TATAA box, lacking the third and fourth adenosine residues (Bucher, 1990), 26 bp upstream from the transcription initiation site determined by RACE. No other consensus promoter sequences were identified in the entire 5’ region sequenced, suggesting that the 5’-end of the cDNA sequence is in fact the major initiation site for the gene.

	Fig.   17 Genomic structure of the human S100A14 Schematic representation of the exon/intron organization. Boxes represent exons, and intervening lines represent introns. The hatched region within the boxed exons denotes the coding sequence.

Sequence analysis of the entire genomic locus of S100A14 did not reveal any CpG island. This was in accordance with our DNA hypermethylation analysis in 8 lung cancer cell lines. Treatment with 5-aza-2‘-deoxycytidine (20 μM) did not induce re-expression of the gene either.

A search for repetitive elements revealed AluY subfamily repeat of 283-bp (nt 2450-2732) located near the polyadenylation site as well as (TG)_n and (GA)_n simple repeats in the putative promoter region.

We verified the human S100A14 genomic sequence by Southern blot hybridization to examine the possibility of deletion or gross rearrangement in the gene. The 1628-bp (SacI; spanning all introns and coding exons) and 500-bp (EcoRI; spanning exon 3 and part of exon 4) bands were found to be informative for the verification of the assembly. Using the cDNA sequence that covers the coding region of S100A14 as a probe for the hybridization, we observed the same restriction pattern for each enzyme in both S100A14-negative and S100A14-positive tumour cells. A single band was detected in digests of 25 tumour cell lines ( Fig. 18 ).

Table  7 Exon/intron boundaries of the human S100A14 gene. The exon and intron sequences are shown in upper- and lowercase letters, respectively. Splice acceptor and donor sequences are shown in boldface type.

	Fig.   18 Southern blot analysis of the S100A14 gene in lung cancer cell lines. Southern blot analysis ofthe representative S100A14-positive (2078, 9442, H2170) and S100A14-negative (Colo 677, H209, Colo 668, H157, N417, A427, H526) cell lines. Genomic DNA (30 µg) isolated from 25 lung tumour cell lines was digested with SacI and EcoRI (300 U). The 1628-bp (SacI) and 500-bp (EcoRI) bands were detected in digests of all examined cell lines. S100A14 cDNA sequence that covers the protein coding region was 32P-labelled and was used as a probe for hybridization.

Next, the three PAC cloneswere used to determine the chromosomal localization of S100A14 by fluorescence in situ hybridization (FISH) on normal human metaphase chromosomes. All clones revealed a strong signal at chromosome band 1q21 with the clone I22230 being the most specific ( Fig. 19 ).

	Fig.   19 Chromosomal localization of S100A14 by FISH. The FISH analysis was performed on metaphases of normal human lymphocytes using biotin-16-dUTP-labelled genomic PAC clone I22230. White arrowheads indicate the specific signal on chromosome 1q21.

Clones A12752 and D11609 showed additional weaker signals on other different chromosomes. Due to the presence of these signals the mapping result was confirmed and refined by hybridizing the S100A14 cDNA to the established 6 Mb YAC contig and to the genomic restriction map of region 1q21 (Marenholz et al., 1996, 2001). The specific probe detected two of the 24 YACs of the contig, 100_f_3 and 950_e_2, which were already known to contain several other genes of the S100 gene cluster in chromosomal region 1q21 (Schäfer et al., 1995, Marenholz et al., 1996) ( Fig. 20 A). S100A14 was mapped on the same SalI fragment as S100A1 and S100A13, the most telomeric genes of the established S100 cluster ( Fig. 20 B).

	Fig.   20 S100A14 is localized within the S100 gene cluster on human chromosome 1q21. A Southern blot that contains SalI-digested DNA of 24 YACs covering chromosomal region 1q21 was hybridized with the 32P-labelled S100A14 cDNA. A: Four YACs of the contig that cover the distal S100 genes including the adjacent regions, are shown. Two YACs were positive for the S100A14-specific probe (right) that detected the same restriction fragments as the S100A1 probe (left). The second fragment identified in YAC 100_f_3 is probably due to an additional, rearranged clone in the original YAC culture. B: SalI restriction map of YAC 100_f_3. A vertical bar represents the YAC. Boxesrepresent SalI restriction fragments that were detected by the genes above. The order of genes has been resolved previously (South et al., 1999). The position of S100A14 relative to S100A13 and S100A1 could not be resolved. C: Ideogram of human chromosome 1.

We additionally confirmed localization of the gene in the region 1q21 on the genomic restriction map of the H2LCL cell line (Marenholz et al., 1996), on which the S100A14 probe detected the same NotI-, NruI-, MluI-, and BsiWI-fragments as the adjacent S100 genes (data not shown).

To further investigate the reasons for deregulated S100A14 expression in tumours, we examined regulation of its putative promoter. As cell model system we chose HEK 293 cells, which do not express endogenous S100A14, and Lovo cells, which express the gene at a high level.

Isolation of S100A14 genomic DNA provided 511 bp of the 5’ upstream region of the major transcription initiation site of the gene. In order to identify potential regulatory sequences within this region, a panel of deletion mutants, covering the region of 495-bp immediately adjacent to the transcription start site, was generated and linked to a promoterless luciferase reporter gene. The presence of endogenous SacI and PstI sites was used to aid the subcloning. The p500-luc luciferase reporter driven by the longest promoter fragment showed a significant luciferase activity in extracts from HEK 293 and Lovo cells ( Fig. 21 A, B). Thus, we generated deletion mutants of this fragment to locate the minimal promoter region responsible for the observed activity.

The p300-luc promoter construct that covers the proximal region of the promoter (313/-12) conferred a 4-fold activation. The other promoter fragment (p250-luc) covering the -495/-313 region showed a very low level of activity, similar to the level obtained with the control vector pGL3 Basic. This indicated that the -313/-12 region of the S100A14 promoter is sufficient to induce a potent luciferase activity. Therefore, we investigated deletion mutants of this promoter fragment. p2A-luc and p2B-luc luciferase reporters showed activity at a level approximately equal to that of the control vector. On the contrary, transfection of a p2C-luc construct, covering the most proximal promoter fragment (-208/-12) of 196-bp, resulted in an enhanced luciferase activity that could be indicative of the presence of possible transcription factor binding sites.

	Fig.   21 T ranscription activities of S100A14 promoter constructs. A: Effects of deletions of putative cis-acting regulatory elements on luciferase activity in HEK 293 cells. A series of 5’ deletion mutants were fused with the firefly luciferase reporter gene (black box) and analysed by transient transfection. Native restriction sites used for cloning are indicated. The relative sizes of the constructs are to scale. The numbers in parentheses indicate the position of the tested promoter region within the 5’ upstream region of the S100A14 gene. Firefly luciferase activity was normalized to Renilla luciferaseactivity, which was measured to control for transfection efficiency. Means and standard deviations are indicated to the right. The data shown are representative of three experiments, each performed in triplicate. B: Transcription activities of the S100A14 promoter constructs in HEK 293 and Lovo cells. Relative luciferase activity is indicated ± the standard deviation. The mean values for Lovo cells were uniformly scaled by the factor 3 to allow comparison of the two cell lines. The results represent the mean of triplicate experiments.

Computer-assisted analysis of the minimal promoter fragment revealed a TATA box, multiple potential transcription factor binding sites for Elk-1, and RFX-1, as well as single binding sites for SRF (serum response factor), AP-1, CEBP, CHOP, USF, NFκB, and N-myc.

In search for conserved elements in the human S100A14 promoter, we retrieved 3460 bp in the sequence 5’ upstream of the mouse S100a14 gene. Comparison of the region with the 511 bp of the human S100A14 promoter revealed presence of two fragments immediately upstream of the transcriptionstart site that were conserved in the human gene ( Fig. 22 ). One of the fragments (7/48) showed 97% identity with the human counterpart and encompassed the TATA box. The other fragment (-76/-135) showed 89% identity with the human sequence and contained a CAAT sequence. This sequence, however, lacked significance to the consensus CCAAT box in our promoter prediction analysis. Within the conserved fragment from -76 to -135 bp two possible transcription factor binding sites for CEBP and C/EBP homologous protein (CHOP), also known as growth arrest- and DNA damage-inducible gene 153 (GADD153) were detected. Both of them belong to the CCAAT/enhancer binding protein (CEBP) family of transcription factors.

Additionally, a potential NFκB consensus motif was conserved in the mouse and human DNA within the minimal promoter fragment but further upstream of the two strongly conserved promoter fragments. This putative consensus site was located in equivalent position in alignment between the two orthologous sequences. No significant homologyof sequences further upstream could be detected.

To determine if the potential NFκB binding site represents a functional NFκB response element, we examined it in more detail. Therefore, it was evaluated whether the S100A14 promoter was subjected to transactivation by wild type p65 – the DNA binding subunit of NFκB. HEK 293 cells were co-transfected with p65/Flag and the p2C-luc luciferase reporter or with empty vectors. Measurement of luciferase activity revealed a weak decrease in the p2C-luc luciferase reporter activity following transient transfection with an empty vector for p65, indicating an inhibitory effect of pcDNA3/Flag vector (data not shown). Nevertheless, no increase of activity was detected upon co-transfection of HEK 293 with p65 compared with the co-transfection with the empty vector (data not shown). Treatment with TNF-α also did not induce any significant transactivation of the promoter in any concentration tested (5-25 ng/ml) thus confirming our Northern blot results (data not shown).

	Fig.   22 Comparison of nucleotide sequence and potential regulatory elements of the S100A14 promoter fragment in mouse and human genomic DNAs. The consensus sequences of the potential transcription factor-binding sites are underlined or in bold. Nucleotide identities are indicated by asterisks. The predicted TATA box and CAAT sequence are represented by shading the consensus sequence. The major transcription initiation site identified in the human S100A14 is indicated by boxes.

Preliminary experiments with H322 lung tumour cells revealed that stimulation of the cells with serum induced S100A14 transcript ( Fig. 23 ). Furthermore, our initial analysis by immunohistochemistry of breast tumours demonstrated a significant correlation between a high S100A14 level and ERBB2 overexpression (p = 0.04). This prompted us to investigate whether S100A14 could be subjected to epidermal growth factor (EGF) regulation.

	Fig.   23 Up-regulation of S100A14 expression in H322 cells after stimulation with serum. Northern blot analysis of S100A14 expression in H322 cells was performed after stimulation with 20% serum for the indicated times. S100A14 cDNA sequence that covers the protein coding region was 32P-labelled and used as a probe for hybridization. 18S RNA was assayedas a loading control.

Eight human cell lines were treated with recombinant human EGF and S100A14 mRNA expression was measured by Northern blot analysis. The eight cell lines included immortalized bronchial epithelial cells (9442) and immortalized mammary epithelial cells (HMEB), which express S100A14 as well as six S100A14-negative tumour cell lines: three breast (SK-BR-3, MCF-7, MDA-MB-231) and three lung (H157, DMS-114, A549) tumour cell lines. Expression of endogenous S100A14 mRNA was remarkably induced by EGF in S100A14-positive 9442 and HMEB cells, but not in S100A14-negative tumour cell lines ( Fig. 24 ).

	Fig.   24 Treatment of 9442 and HMEB cells with EGF leads to the induction of S100A14 9442 and HMEB cells were grown for 3 days in regular culture medium and then treated for 12 hours with 50 ng/ml of EGF without medium change. 10 µg of total RNA was size-fractionated followed by Northern blot analysis. S100A14 cDNA sequence was applied as a probe for hybridization. 18S RNA was used as a loading control.

Based on this result, we decided to further characterize regulation of the gene by EGF using 9442 cells as a cell model. The 9442 (BET-1A) cells are derived from normal human bronchial epithelial cells that were immortalized with the SV40 early region and represent non-tumorigenic cells.

First, we determined the kinetics of S100A14 induction by EGF in 9442 cells. Following the growth factor treatment, S100A14 levels were induced at 3 hours post-treatment and then gradually increased, reaching a maximal level at 12 hours ( Fig. 25 ). The observed sustained response remained for up to 24 hours, with a subsequent gradual decrease to slightly above pre-stimulation level. These results clearly demonstrated an EGF-dependent induction of endogenous S100A14 expression and indicated that S100A14 is an EGF target gene.

	Fig.   25 S100A14 is induced by EGF in a time-dependent manner in 9442 cells. 9442 cells were grown for 3 days in regular culture medium and then treated for the indicated times with 50 ng/ml of EGF without medium change. Northern blot analysis was performed using specific 32P-labelled S100A14 cDNA probe for the detection. 18S RNA was used to correct for equal loading.

Next, we examined the dose dependency of the S100A14 mRNA induction in 9442 cells. The response appeared to be maximal at a concentration of 50 to 100 ng/ml of EGF ( Fig. 26 ). Therefore, 50 ng/ml of EGF treatment for 12 hours was used in this study.

	Fig.   26 Induction of S100A14 is EGF dose-dependent. 9442 cells were grown for 3 days in regular culture medium and then treated with 1-100 ng/ml of EGF for 12 hours without medium change. S100A14 was detected by Northern blot analysis using a specific 32P-labelled cDNA probe. 18S RNA was used as a loading control.

Treatment of the cells with transforming growth factor-α (TGF-α), another ERBB receptor ligand, also resulted in an increase of the S100A14 transcript level ( Fig. 27 ). The kinetics of the TGF-α-induced response was similar for the EGF and TGF-α ligands with a delayed time course relative to the EGF-induced response. The induction was transient, starting at 6 hours post-treatment, reaching maximal levels 12 hours post-stimulation and remaining for up to 48 hours, with levels returning to basal by 72 hours.

	Fig.   27 TGF-α induces S100A14 mRNA in 9442 cells. 9442 cells were grown for 3 days in regular culture medium and then treated for the indicated times with 20 ng/ml of TGF-α without medium change. 10 µg of total RNA was size-fractionated followed by Northern blot analysis. S100A14 cDNA sequence was 32P-labelled and used as a probe for hybridization. 18S RNA was used as a loading control.

S100A14 was also induced in response to fresh medium or “stimulation medium” showing a maximal level at 6 hours ( Fig. 28 ). Notably, the induction was less potent than in response to EGF.

	Fig.   28 Up-regulation of S100A14 expression in 9442 cells after stimulation with fresh medium. 9442 cells were grown for 3 days in regular culture medium and then incubated for the indicated times with fresh medium. Northern blot analysis was performed using 32P-labelled S100A14 cDNA probe for the detection. 18S RNA was used to correct for equal loading.

It was of interest to determine whether the minimal positive promoter element, which we identified, confers EGF responsiveness to the luciferase reporter gene. Treatment of the p2C-luc-transfected HEK 293 and Lovo cells with EGF revealed no significant induction of the promoter fragment (data not shown).

The up-regulation of S100A14 by EGF was examined at the protein level by Western blotting of the whole cell lysates. The cells were collected 24 hours after stimulation with EGF. We did not, however, detect any substantial changes in S100A14 protein level even after increasing the EGF concentration (Fig. 29).

	Fig.   29 S100A14 protein is not induced following treatment with EGF. 9442 cells were grown for 3 days in regular culture medium and then treated with 1-100 ng/ml of EGF for 24 hours without medium change. 10 µg of whole protein extracts were subjected to SDS-PAGE followed by Western blot analysis using anti-S100A14 antibody. β-actin was used as a loading control.

The major downstream signalling pathways triggered by activated ERBB receptors are ERK1/2 mitogen-activated protein kinase (MAPK), stress-activated protein kinase/c-JUN NH₂-terminal kinase (SAPK/JNK), p38 MAPK, and phosphatidylinositol 3’-kinase (PI3K) pathways. To assess which signalling pathway(s) mediated the stimulatory effect of EGF on S100A14 we applied several pharmacological agents to inhibit these pathways.

First, kinetics of the growth factor-induced response was determined for the three MAPK signalling pathways (ERK1/2, p38, JNK) and the PI3K cascade in order to establish the stimulation efficacy and the optimal treatment time. Immunoblotting with the respective antiphospho antibodies revealed that growth factor stimulation caused a significant increase in ERK1/2 activation which remained on a steady level over a period of 120 min ( Fig. 30 A). The p38 signalling pathway was also markedly activated reaching a maximal induction level between 5 and 30 min after treatment with levels returning to basal by 120 min ( Fig. 30 B). Mitogen exposure also enhanced the phoshorylation and activation of the JNK MAPK cascade in these cells with maximal response at 5 to 15 min after stimulation ( Fig. 30 C). Remarkably, growth factor treatment had no effect on the activation of AKT suggestive of a constitutive activation of the PI3K pathway in these cells ( Fig. 30 D).

	Fig.   30 Kinetics of the growth factor-induced response for the ERK1/2, p38, and JNK MAPK pathways and the PI3K cascade. 9442 cells were grown for 3 days in regular culture medium. The medium was then replaced with the medium that was supplemented with a double set of growth factors (referred to as “stimulation medium”; MS). The cells were cultured for the indicated times and harvested. 20 µg of whole protein extracts were subjected to SDS-PAGE followed by Western blot analysis using A: anti-phospho ERK1/2, B: anti-phospho p38, C: anti-phospho SAPK/JNK, and D: anti-phospho AKT antibodies. β-actin, p38, SAPK/JNK, and AKT were used as loading controls, respectively.

Next, we examined the inhibitory efficacy of various pharmacological agents on activation of the intracellular signalling pathways triggered in response to EGF. U0126 (MEK1 and MEK2 inhibitor) markedly suppressed growth factors-induced ERK1/2 activation in 9442 cells ( Fig. 31 A). AG1478 (ERBB receptor tyrosine kinase inhibitor) also markedly attenuated ERK1/2 phosphorylation ( Fig. 31 B). The suppressive efficacy of SB203580 (p38 kinase inhibitor) on the p38 MAPK pathway and SP600125 (JNK kinase inhibitor) on the JNK pathway was confirmed by the ability to block the activation of their downstream targets: HSP27 and cJUN, respectively. This was determined by immunoblotting with the respective antiphospho antibodies ( Fig. 31 C, D). Pretreatment with LY294002 (a PI3K inhibitor) only partially inhibited the response to growth factors ( Fig. 31 E). Increasing concentration of the inhibitor failed to alter the inhibitory response in these cells.

	Fig.   31 The inhibitory efficacy of various pharmacological agents on activation of the ERK1/2, p38, and JNK MAPK pathways and the PI3K cascade9442 cells were grown in regular medium for 3 days and then preincubated for 1 hour with A: 20 µM U0126, B: 10 µM AG1478, C: 40 µM SB203580, D: 40 µM SP600125, and E: 40 µM LY294002, respectively, without medium change. Next, the cells were stimulated with the medium that was supplemented with a double set of growth factors (referred as “stimulation medium”; MS) for 10 min (A), 10 min (B), 5 min (C), 5 min (D), and 10 min (E). As negative controls, cells were treated with the vehicle: DMSO (A, B, C, D) or ethanol (E). 20 µg of whole protein extracts were subjected to SDS-PAGE followed by Western blot analysis using anti-phospho ERK1/2 (A, B), anti-phospho HSP27 (C), anti-phospho c-JUN (D), and anti-phospho AKT (E) antibodies. β-actin, p38, c-JUN, and AKT were assayed as loading controls.

To delineate the pathways involved in S100A14 mRNA induction we treated 9442 cells with the MAPK and the PI3K pathway inhibitors prior to EGF stimulation. Treatment with LY294002 does not abrogate growth factor-dependent induction of S100A14 mRNA ( Fig. 32 A). Its expression was also unaffected by SB203580 and SP600125 inhibitors. In contrast with this result, U0126 completely abrogated EGF-induced S100A14 mRNA expression at the time of optimal gene induction (after 12 hours) and this treatment reduced mRNA levels to those seen in unstimulated cells ( Fig. 32 B). To examine whether the intact tyrosine kinase activity of the EGF receptor was necessary for the EGF-induced S100A14 expression, cells were preincubated with AG1478. AG1478 completely prevented EGF-induced S100A14 up-regulation ( Fig. 32 B). None of the concentrations of the inhibitors that were used in this study caused cell detachment or cell death.

	Fig.   32 ERK1/2 MAPK signalling pathway determines S100A14 induction following stimulation with EGF9442 cells were grown in regular medium for 3 days and then preincubated for 1 hour with A: 40 µM SP600125, 40 µM SB203580, 40 µM LY294002, B: 20 µM U0126, and 10 µM AG1478, respectively, without medium change. As negative controls, cells were treated with the vehicle: DMSO or ethanol. Next, the cells were stimulated with 50 ng/ml of EGF for 12 hours and harvested. 10 µg of total RNA was size-fractionated followed by Northern blot analysis using 32Plabelled S100A14 cDNA probe for the detection. 18S RNA was used to correct for equal loading.

These results demonstrated that ERK1/2 MAPK signalling plays a prominent role in regulation of S100A14 transcript in response to growth factors.

No inhibition of S100A14 protein was detected following treatment with the U0126 and AG1478 inhibitors (data not shown). Experiments included both whole cell lysates as well as subcellular fractionated lysates (membrane and soluble fraction). Similarly, no shift in the S100A14 subcellular localization was observed and S100A14 remained compartmentalized to crude membrane and soluble fractions upon induction by EGF, as determined by Western blotting with anti-S100A14 antibody (data not shown).

To further investigate the mechanisms of EGF-mediated S100A14 transcriptional activation, we employed cycloheximide (CHX) – an inhibitor of protein synthesis. Pretreatment of cells with CHX led to a partial reduction in the EGF-induced S100A14 mRNA levels in a dose-dependent manner ( Fig. 33 ). Based on this result, we concluded that the induction of S100A14 following EGF treatment requires a new protein synthesis. Therefore, S100A14 is likely to be an indirect transcriptional target of EGF signalling and it is conceivable that the synthesis of other protein(s) is involved in its transcriptional activation.

	Fig.   33 EGF-induced S100A14 expression is dependent on de novo protein synthesis. 9442 cells were grown in regular medium for 3 days and then preincubated for 1 hour with 2, 5, and 10 µg/ml of cycloheximide (CHX), respectively, without medium change. As negative controls, cells were treated with the vehicle DMSO. Next, the cells were stimulated with 50 ng/ml EGF for 12 hours and harvested. 10 µg of total RNA was size-fractionated followed by Northern blot analysis using 32Plabelled S100A14 cDNA probe for the detection. 18S RNA was used as a loading control.

Screening for transcriptional modulators of S100A14 revealed phorbol ester 12myristate 13-acetate (PMA) as a potential activator of the gene. PMA acts as a specific agonist of both conventional and novel protein kinase C (PKC) isoenzymes activation. The phospholipase Cγ – PKC pathway is well known to be coupled to activation of ERBB receptors and PKC has often been implicated as a mediator of ERBB receptor transactivation. The involvement of PKC was therefore investigated to determine whether PKC could mediate S100A14 mRNA induction.

PMA induced S100A14 mRNA in 9442 cells reaching a maximal level by 12 hours and returning to a level slightly above basal by 24 hours ( Fig. 34 A). The kinetics of this response paralleled the time course of the EGF effect, although PMA was less potent than EGF in inducing S100A14.

To confirm that S100A14 induction in response to PMA was PKC-dependent, we tested the ability of the PKC inhibitor bisindolylomaleimide I to block PMA-stimulated S100A14 induction. As demonstrated in Fig. 34 B, bisindolylomaleimide I (5 μM) did not abrogate S100A14 induction in response to PMA.

The ability of PMA to cause the phosphorylation and activation of ERK1/2 MAP kinases is well established and has been shown to depend upon PKC-mediated activation of upstream elements of the ERK1/2 MAPK pathway, including RAS and RAF-1. We therefore addressed the question whether PMA had the capacity to activate the ERK1/2 MAPK pathway in 9442 cells. PMA induced a rapid increase in the amount of phospho-ERK1/2 in 9442 cells and this activation continued for at least 60 min ( Fig. 34 C). To determine whether the PMA-induced increase in S100A14 mRNA also depends upon the activation of ERK1/2, we stimulated 9442 cells with PMA and tested by Northern blot analysis the ability of U0126 to inhibit S100A14 mRNA. In the presence of U0126, the S100A14 mRNA level decreased to slightly above pre-stimulation level ( Fig. 34 B).

The ability of both EGF and PMA to stimulate ERK1/2 activity in 9442 cells could suggest that the capacity of EGF to stimulate S100A14 might be PKC-dependent. We therefore tested the effect of the PKC inhibitor on S100A14 induction in response to EGF. Pretreatment of cells with bisindolylmaleimide I did not significantly influence S100A14 up-regulation in response to EGF suggesting that the induction is not mediated by PKC ( Fig. 34 D).

	Fig.   34 PMA exerts stimulation of S100A14 via PKC activation. 9442 cells were grown for 3 days in regular culture medium. A: The cells were then treated without medium change with100 nM of PMA for the indicated times and harvested for Northern blot analysis. B: The cells were preincubated for 1 hour with 5 µM of bisindolylomaleimide I (BIS) or 20 µM of U0126, respectively, without medium change. As negative controls, cells were treated with the vehicle DMSO. Next, the cells were stimulated with 100 nM of PMA for 12 hours and harvested. 10 µg of total RNA was size-fractionated followed by Northern blot analysis using 32P-labelled S100A14 cDNA probe for the detection. 18S RNA was used as a loading control. C: The cells were stimulated without medium change with 10 and 100 nM of PMA for the indicated times and harvested. 20 µg of whole protein extracts were subjected to SDS-PAGE followed by Western blot analysis using anti-phospho ERK1/2 antibody. β-actin was used as a loading control. D: The cells were preincubated for 1 hour with 5 µM of bisindolylomaleimide I (BIS) without medium change. Next, the cells were stimulated with 100 nM of PMA or 50 ng/ml of EGF, respectively, for 12 hours and harvested. As negative controls, cells were treated with the vehicle DMSO. Northern blot analysis was performed using 32P-labelled S100A14 cDNA probe for the detection. 18S RNA was used as a loading control.