In order to explore the distribution patterns of SERPINI1 and PDCD10, we performed Northern blot analysis on a human multiple tissue expression array of poly A+ RNAs. As illustrated in Fig. 1, while SERPINI1 expression was observed mainly in different parts of brain, hybridization of the same array with the PDCD10 probe revealed that it was ubiquitously expressed as a housekeeping gene in all tissues analyzed. Moreover, as SERPINI1 and SERPINI2 were both reported to be down-regulated during tumor development 12, it is rational to doubt that the expression of PDCD10, which is located in between the two SERPINI genes at chromosome 3q26, may also be altered in some tumors. Results from the semi-quantitative RT-PCR indicate that while SERPINI1 expression was negatively correlated with tumor progression, the mRNA level of PDCD10 was not much affected in brain tumor cells (Fig. 2). Furthermore, when we investigated the expression patterns of PDCD10 in other normal and cancerous tissues, we found that PDCD10 was up-regulated in various cancers including ovarian, stomach, lung, uterus and small intestine cancers (Fig. 2).

After analyzing the distribution of mRNA transcripts for both SERPINI1 and PDCD10 genes, we then attempted to pinpoint their precise transcriptional start sites in order to determine the length of their intergenic spacer. As revealed by the human genome reference sequence, SERPINI1 [accession number EMBL:BC018043] and PDCD10 [accession number EMBL:AF022385] were mapped in a head-to-head configuration at 3q26 and separated by less than 1 kb. By performing 5'-RACE on human brain RNA, three transcriptional start sites for PDCD10 but only one transcriptional start site for SERPINI1 were identified. The closest transcriptional start site that we identified for PDCD10 and SERPINI1 were at 5 and 15 bp respectively upstream of the sites predicted by NCBI database, thus the shortest intergenic distance was recalculated to be 851 bp (Fig. 3 & 4).

Following the mapping of the transcriptional start sites, the 851-bp intergenic fragment constituting the putative regulatory region for both SERPINI1 and PDCD10 genes was analyzed for its transcriptional activities by driving the firefly luciferase reporter gene. The human cell lines utilized for luciferase reporter assays include brain neuroglioma H4, brain glioblastoma U-87 MG, ovarian adenocarcinoma OVTW-59-4, lung adenocarcinoma CL1-5, and cervical adenocarcinoma HeLa cancer cell lines. The rat adrenal gland pheochromocytoma PC-12 cell line was also used for comparison as it had previously been chosen for the study of mouse Serpini1 promoter 30. The full-length 851-bp fragments in either direction toward SERPINI1 or PDCD10 were cloned into the reporter vector (pGL3-Basic). By measuring the luciferase activity in various cell lysates after transient transfection, the 851-bp intergenic sequence was shown to be capable of driving the reporter gene expression in either direction (Fig. 3). The promoter activities in the direction of PDCD10 in all cells tested were approximately 4 to 10-fold higher than those of SERPINI1. Moreover, the promoter activities toward the SERPINI1 direction were higher in brain tumor H4 and U-87 MG cells than those in other test cells, and they were nearly 3-fold higher than the positive control, the viral SV40 minimal promoter (Fig. 3). These results indicate that the 851-bp intergenic region is able to function as a bidirectional promoter to control SERPINI1 and PDCD10 expressions, and some regulatory elements within this region may asymmetrically regulate the transcription of both genes.

To understand whether the head-to-head configuration of SERPINI1 and PDCD10 is a unique feature only in human or if it is evolutionary conserved in other mammals, we aligned the homologous regions between these two genes among human, mouse and rat by the Vector NTI program (InforMax, Bethesda, MD, USA). The results revealed that both the gene order and the compact head-to-head arrangement are conserved in different species (data not shown), and that within the intergenic region there is an approximately 400-bp fragment near the vicinity of PDCD10 with the highest percent identity (Fig. 4A). Based on the nucleotide sequence shown in Fig. 4B, the 851-bp intergenic region has a GC content of 65%, which is much higher than the average of the entire human genome (41%). While no TATA-box is observed in this region, two CAAT boxes (consensus GG^T/_CCAATCT) are noted in the most conserved 400-bp fragment with a 100-bp distance from each other. In addition, within this highly conserved fragment, there are one neuron-specific AP-2, one c-Myc and one NFκB binding motives as well as several binding sites for the general transcription factor Sp1 as predicted by MatInspector (Genomatix; http://www.genomatix.de) and TFSEARCH programs (Computational Biology Research Center; http://www.cbrc.jp). The presence and conservation of these putative binding motives suggest that some of these transcription factors must possess an important evolutionary significance and some yet-unknown regulatory demands.

To identify the key regulatory DNA sequences responsible for SERPINI1 and PDCD10 gene expressions, we constructed a series of plasmids containing various intergenic promoter regions derived from the 851-bp sequence. By using the same reporter system addressed in Fig. 3, the luciferase activities of these constructs were assessed by transient transfection into H4 cells. Due to the neuron-specific expression of the SERPINI1 gene, the promoter activity of SERPINI1 in neuron cells is higher than those in cells from other tissues. Thus, H4 cells were used as a vehicle for studying the regulatory mechanism of SERPINI1 promoter. As shown in Fig. 5, removal of the sequence from nt 1 to 175 completely abolished luciferase activity compared to the full-length 851-bp fragment in the SERPINI1 direction, suggesting that this 175-bp fragment is important for the control of SERPINI1 gene expression. This fragment alone was also capable of driving the reporter gene expression in the PDCD10 direction. Further splitting of this bidirectional promoter region into two parts, i.e. the sequence from nt 1 to 107 or from nt 108 to 175, resulted in a dramatic loss of luciferase activity. Surprisingly, deletion of the fragment of nt 176 to 851 increased more than 3-fold of the luciferase activity relative to that of the 851-bp fragment in the SERPINI1 direction but only kept a ~40% remaining of the luciferase activity in the PDCD10 direction (Fig. 5). These results strongly indicate that within the fragment of nt 176-851 there likely exist a repressive element that can regulate SERPINI1 expression and an enhancer that can modulate PDCD10 expression. Based on these findings, we conclude that a 175-bp minimal promoter region from nt 1 to 175 near the vicinity of PDCD10 and a regulatory region outside the 175-bp fragment from nt 176 to 851 may coordinately regulate the gene transcriptions of SERPINI1 and PDCD10.

From the experiment described in Fig. 5, there seems to exist a regulatory fragment from nt 176 to 851 which possesses no promoter activity but appears to contain a repressive element for SERPINI1 and an enhancer for PDCD10. Thus, we cloned this fragment in two opposite directions into the upstream of the SV40 promoter in a pGL3-Promoter vector, and examined their effects on the SV40 promoter. As expected, the fragment of nt 176-851 in the PDCD10 direction was able to increase the SV40 promoter activity for approximately 2.2-fold (Fig. 6A). This observation confirmed the existence of an enhancer for PDCD10 as seen in Fig. 5. However, there was no increase of the SV40 promoter activity in the direction of SERPINI1. This phenomenon was possibly a compromising effect resulted from the coordination between the enhancer and the repressive element within the regulatory region; the extent of such effect may also vary for using different heterologous promoters.

To further identify the enhancer responsible for PDCD10, three intergenic fragments including 1) the nt 176-473 fragment, 2) the nt 474-710 fragment, and 3) the nt 711-851 fragment toward the PDCD10 direction were individually cloned into the upstream of the SV40 promoter and assayed for their promoter activity in H4 cells. The results showed that only the nt 473-176 fragment was able to increase the SV40 promoter activity by 6 to 10-fold (Fig. 6B). The same fragment toward the SERPINI1 direction had no such effect on the SV40 promoter activity (data not shown). Hence, there is an orientation-dependent enhancer located in the fragment of nt 176-473 and acting on the transcription of PDCD10. As for the search of the repressive element for SERPINI1, the same fragments but in opposite direction were individually cloned into a modified pGL3-Promoter vector containing a CMV promoter, a stronger promoter than the SV40 and better for repressive activity studies. The assay results demonstrated that only the fragment from nt 176-473 toward the SERPINI1 direction possessed a repressive ability which inhibited ~70% of the CMV promoter activity (Fig. 6C). Nonetheless, the same fragment but in the opposite orientation (toward the PDCD10 direction) could only enhance the CMV promoter activity for approximately 2-fold (data not shown). These findings signify that a critical fragment from nt 176-473 outside the minimal promoter possesses a strong repressive activity for SERPINI1 and enhancing activity for PDCD10.

All culture media, except RPMI 1640 medium, were purchased from Gibco Invitrogen (Carlsbad, CA, USA) and supplemented with 10% fetal bovine serum (FBS) (HyClone, Logan, UT, USA). H4 (human neuroglioma) and OVTW-59-4 (human ovarian adenocarcinoma) cells were cultured in DMEM (Dulbecco's modified Eagle's medium) medium, whereas U-87 MG (human brain glioblastoma) and HeLa (human cervical adenocarcinoma) cells were grown in minimum essential medium (MEM). CL1-5 (high-metastatic human lung adenocarcinoma) cells were cultured in RPMI-1640 medium (Hyclone) supplemented with 0.5 mM L-glutamine and 10% FBS. PC-12 (rat adrenal gland pheochromocytoma) cells were cultured in 85% RPMI-1640 supplemented with 5% FBS and 10% horse serum. All cell lines were maintained at 37°C in a humidified atmosphere of 5% CO₂ in air.

Northern hybridizations were carried out mainly as described previously 12. In brief, the probe for SERPINI1 was amplified from the FirstChoice PCR-Ready human brain cDNA (Ambion, Austin, TX, USA) using the forward primer 5'-CTGGAAGTCGCAGTTTAGGCCTGAA-3' and the reverse primer 5'-TGGATGGTCGACAATAACTTGAGGA-3' to produce a 559-bp fragment under the following conditions: 1 cycle of 94°C for 5 min; 32 cycles of 94°C for 20 sec, 55°C for 20 sec, and 72°C for 1 min; and 1 cycle of 72°C for 10 min. The probe for PDCD10 was amplified from the same cDNA using the forward primer 5'-TCCATGGTTTCTATGCCCCT-3' and the reverse primer 5'-TTTGGTGTTCAAGTGCCCTG-3' to produce a 445-bp fragment under the following conditions: 1 cycle of 94°C for 5 min; 35 cycles of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec; and 1 cycle of 72°C for 10 min. All PCRs were performed on a T3 thermocycler (Whatman Biometra, Goettingen, Germany) using Ex Taq polymerase (Takara, Otsu, Singa, Japan) and subjected to sequencing on an ABI PRISM 3700 automated sequencer (Applied Biosystems, Foster City, CA, USA) by the core facility of the National Health Research Institutes (Zhunan, Taiwan). The sequence-verified probes, including a cDNA probe for ubiquitin supplied by the manufacturer and used as a control, were labeled with [α-³²P]dCTP using the Rediprime II DNA labeling system (Amersham/GE Healthcare, Piscataway, NJ, USA) followed by hybridization against either the human MTE multiple tissue expression array or cancer profiling array (BD Bioscience Clontech, Palo Alto, CA, USA) in an ExpressHyb hybridization solution (BD Bioscience Clontech) for 4 h at 65°C. The arrays were then washed and exposed to X-ray film for detection. Each Northern blot analysis was repeated at least three times to confirm reproducibility of the results.

Total RNAs were extracted from two human brain tumor cell lines, H4 and U-87 MG, using TRIZOL reagent (Gibco Invitrogen) followed by DNase (Ambion) treatment to eliminate potential genomic DNA contamination. For control, total RNA of normal brain tissue was obtained from Clontech. Complementary DNA (cDNA) was synthesized from 5 μg of total RNA using oligo(dT) priming (Roche Molecular Biochemicals, Indianapolis, IN, USA) and MMLV reverse transcriptase (Epicentre Technologies, Madison, WI, USA). After a 1-h incubation at 37°C, the reaction mixtures were heat inactivated at 92°C for 10 min and then applied as templates to amplify cDNAs specific for SERPINI1 (amplified with the forward primer 5'-CTTGCAATGGGAATGATGGAACTTG-3' and the reverse primer 5'-CGAAATCATGTCCACTTGTGTTCAT-3' at 1 cycle of 94°C for 5 min; 35 cycles of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 90 sec; and 1 cycle of 72°C for 10 min) and for PDCD10 (amplified with the forward primer 5'-TCCATGGTTTCTATGCCCCT-3' and the reverse primer 5'-TTTGGTGTTCAAGTGCCCTG-3' at 1 cycle of 94°C for 5 min; 35 cycles of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec; and 1 cycle of 72°C for 10 min). GAPDH serving as internal control was amplified using the primer set 5'-CCTGCACCACCAACTGCTTA-3' and 5'-ACCCTGTTGCTGTAGCCAAA-3' to allow comparison of RNA levels among different samples. The PCR conditions for GAPDH were the same as those for PDCD10.

The 5'-RACE analysis was performed using the FirstChoice human brain RNA ligase-mediated (RLM)-RACE kit (Ambion). Two RNA adaptor primers were employed as forward primers according to the manufacturer's instruction. The reverse primers for the 5'-RACE of SERPINI1 were 5'-TAGGCTGTCATATCCCATTGAGTGG-3' for the outer primer and 5'-CAAGTTCCATCATTCCCATTGCAAG-3' for the inner primer. The reverse primers for the 5'-RACE of PDCD10 were 5'-CTCAGCTTCATTCTTCATCTCTTC-3' for the outer primer and 5'-CCTCATTCAAAAGCCAACTACAG-3' for the inner primer. Both the first and second rounds of PCR were performed with 1 cycle at 94°C for 3 min; 35 cycles at 94°C for 30 sec, 60°C for 30 sec, and 72°C for 30 sec; and 1 cycle at 72°C for 7 min. The PCR products were cloned into a T-tailed vector pGEM-T-Easy (Promega, Madison, WI, USA) followed by sequencing to determine the 5'-ends of SERPINI1 and PDCD10 transcripts.

All restriction enzymes were purchased from Takara. Genomic DNA was purified from human whole blood samples using the QIAamp Blood kit (Qiagen, Hilden, Germany). The various lengths of the intergenic segment between SERPINI1 and PDCD10 were obtained by PCR using the primers listed in Table 1 under the following condition: 1 cycle of 94°C for 5 min; 35 cycles of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec; and 1 cycle of 72°C for 10 min. The amplified DNA fragments were then cloned into one of the three vectors: pGL3-Basic, pGL3-Promoter, or pGL3-Promoter with the original SV40 promoter being replaced by a CMV promoter.

Cells were first seeded in 12-well culture plates at a density of 1.8 × 10⁵ cells per well. Cells were then co-transfected with 3 μl of Lipofectamine 2000 (Gibco Invitrogen) and a mixture of 525 ng of DNA consisting of 500 ng of the tested pGL3 promoter constructs and 25 ng of the pRL-TK Renilla luciferase reporter plasmids (Promega) in a serum-free medium. After a 3-h incubation at 37°C, the transfection medium was replaced with fresh medium supplemented with 10% FBS and cells were incubated for another 24 h. After incubation, cells were washed twice with PBS and lysed by reporter lysis buffer (Promega). Cellular debris was removed after centrifugation at 10,000 × g for 5 min at 4°C. Cell extracts were then assayed for luciferase activity using a SIRIUS luminometer (Berthold, Bad Wildbad, Germany). In each experiment, cells were transfected with pGL3-Basic and pGL3-Promter plasmids that served as negative and positive control, respectively. To normalize transfection efficiency, the pRL-TK plasmids were co-transfected as an internal control.