Background
Tissue factor pathway inhibitor-2 (TFPI-2) is a 32 kDa Kunitz-type serine proteinase inhibitor synthesized by a variety of cells and directionally secreted into their extracellular matrix (ECM) where it is thought to regulate plasmin-mediated ECM degradation and remodeling (reviewed by Chand et al. [
1]). As matrix degradation is an important step in tumor invasion and metastasis, several, but not all, tumor cells downregulate TFPI-2 expression [
2,
3]. In this regard, overexpression of TFPI-2 in several tumor cells was shown to inhibit their growth, invasiveness, angiogenic potential and metastatic potential [
4‐
9]. The mechanism whereby some tumor cells downregulate TFPI-2 synthesis has been primarily attributed to transcriptional silencing through hypermethylation of CpG sites in the TFPI-2 promoter [
10‐
14], inasmuch as treatment of these tumor cells with a methyltransferase inhibitor, 5'-aza-2'-deoxycytidine, restored TFPI-2 transcription[
14]. In addition, several highly aggressive tumors delete the locus for the TFPI-2 gene in the chromosome 7q region [
15‐
17], resulting in the total loss of TFPI-2 protein expression in these cells. Accordingly, the TFPI-2 gene is becoming increasingly recognized as a tumor suppressor gene, since its down-regulation in several types of cancers allow for enhanced tumor growth and metastasis.
In view of its apparent role in cancer progression, we initiated a study to quantify TFPI-2 transcript levels in total RNA samples from selected normal human tissue, as well as their corresponding tumor tissue. In the course of these studies, we detected a novel, aberrantly-spliced variant of TFPI-2 mRNA derived from TFPI-2 pre-mRNA splicing at exon/intron boundaries, as well as at new sites within exons and introns. The levels of the aberrantly-spliced variant of TFPI-2 were either very low or undetectable in normal tissue, but markedly upregulated in tumor tissues and several tumor cell lines. These findings provide suggestive evidence for an additional mechanism for tumor cells to down-regulate TFPI-2 expression through aberrant splicing.
Discussion
Human TFPI-2, an ECM-associated Kunitz-type serine proteinase inhibitor [
21], is thought to play a significant role in the regulation of plasmin-mediated ECM degradation during tumor cell invasion and metastasis, wound healing, and angiogenesis. Proteolytic degradation of the extracellular matrix, one of the key events in the process of tumor invasion and metastasis, depends on a delicate balance of matrix-degrading proteinases and their inhibitors. The TFPI-2 transcript is abundant in the full-term placenta, and is also present at various levels in several adult human tissues, such as liver, lung, skeletal muscle, heart, kidney, and pancreas [
22]. Conversely, TFPI-2 mRNA levels fall with increasing malignancy in the case of breast, gastric, colon, pancreatic, laryngeal, renal, endometrial, glial and several other human malignant tumors [
3,
23,
24].
In experiments designed to quantify TFPI-2 transcripts in human normal and corresponding tumor tissues using RT-PCR, we observed two TFPI-2 amplicons, using exon II (sense) and exon V (anti-sense) primers. Nucleotide sequencing analysis of both amplicons revealed that the larger
M
r
amplicon (622 bp) corresponded to the full-length TFPI-2 cDNA, while the smaller
M
r
cDNA (289 bp) was an aberrantly-spliced form of the TFPI-2 transcript, which we have provisionally designated as asTFPI-2. Primer walking with a series of primer combinations spanning the entire regions of exon I, intron A and 3' UTR revealed that the complete asTFPI-2 transcript consisted of exon II and exon V, with three unusual spliced regions of nucleotides from exon III, intron C, and exon IV tandemly interspersed (Fig.
1A). Analysis of potential consensus splice sites in the TFPI-2 gene (GenBank accession no.AF217542) utilized for the generation of asTFPI-2 using splice site predictor [
39] revealed no consensus splice sites for subsequent insertion of the observed nucleotide sequences of exon III, intron C and exon IV between exons II and V (Fig.
1C). It is, therefore, highly likely that asTFPI-2 is an aberrant splice product generated from TFPI-2 pre-mRNA in which the splicing of pre-mRNA is misdirected and does not occur solely at
de facto splice sites normally used by the major, U2-dependent spliceosome [
25].
In order to assess the relative levels of the TFPI-2 and asTFPI-2 transcripts, we used semi-quantitative RT-PCR and quantitative real-time PCR and observed a dramatic difference in the levels of these two transcripts in normal cells and tumor cells. Quantification of TFPI-2 mRNA and asTFPI-2 levels using real-time PCR was found to be highly reproducible using transcript-specific primers, with low inter-assay variations. By this procedure, normal cells synthesize ~10-fold more wild-type TFPI-2 transcripts than tumor cells, whereas tumor cells synthesized 4–50 times more asTFPI-2 than normal cells. Moreover, the proportion of asTFPI-2 to that of wild-type TFPI-2 was 7–12-fold higher in tumor cells, while in normal cells, the asTFPI-2 transcript was very low to negligible (Fig.
2 and
3).
An unusual feature of the asTFPI-2 is its lack of detectable 5' – and 3' UTRs. As expected, both 5' and 3'-RACE analyses of total RNA from either normal tissue (placenta) or a tumor sample (lung tumor) resulted in a single amplification product containing the 5'-UTR and 3' – poly (A)
+ tail regions of wild-type TFPI-2 cDNA. However, no additional RACE amplification product(s) with either a 5'-UTR or a 3'-poly (A)
+ tail was observed in either lung tumor RNA or placental RNA, indicating that the asTFPI-2 transcript lacked these two elements. In addition, application of HUVEC and Colo-205 total RNA samples to oligo-dt columns with subsequent elution of the poly (A)
+ RNA and RT-PCR analyses of these elutes resulted exclusively in the amplification of the full-length TFPI-2 transcript (data not shown). Thus, either by message enrichment, RACE analyses and primer walking experiments, we conclude that asTFPI-2 does not contain either a 5'-UTR or a 3'-poly (A)
+ tail. Moreover northern blot analysis of total RNA obtained from Colo-205 cells revealed that only the asTFPI-2 transcript (~300 bp) was observed in these cells, further validating the RT-PCR results (Fig.
5). In contrast, two high molecular weight transcripts (1.2 kb and 1.8 kb) for full length TFPI-2 were observed in HUVECs, confirming the results of previous studies [
19].
After treating cells with actinomycin D, the TFPI-2 mRNA exhibited a half-life of ~8 h in HUVECs, whereas the asTFPI-
2 mRNA exhibited a half-life value of ~16 h in Colo-205 cells (Fig.
4). Although it is difficult to directly compare the half-lives of these two transcripts made in normal and transformed cells, the relatively long half-life of the asTFPI-2 transcript is remarkable when one considers that it lacks both a 5'-UTR and 3'-poly (A) tail, two elements that usually offer some protection from exonucleolytic decay [
26]. Inasmuch as asTFPI-2 is apparently not translated and exhibits a relatively long half-life, it is tempting to speculate that this splice variant plays some role in tumor homeostasis and studies to address this possibility are ongoing in our laboratory.
Our results are consistent with several reports on differential alternative splicing between cancer and normal expressed sequence tag libraries where there appears to be a higher proportion of disrupted spliced variants in tumor suppressor genes than in non-cancer-related genes [
27‐
29]. In this regard, aberrant splicing has been reported for mdm2, TSG101 and FHIT pre-mRNAs [
25,
30], exclusively in tumors, and the splicing was characterized by either losses of individual exons, or loss of exons with insertions of unknown DNA sequences without regard for proper splice junctions [
31]. Increased aberrant splicing also represent an additional mechanism for the reduction of the amount of wild type tumor suppressor mRNA without mutation, as shown for the NF1 gene [
32], or alternative splicing, as shown for a brain-specific exon in
Bin1 eliminating the activity of the tumor suppressor in melanoma [
33]. In addition, several cases of aberrant and alternative splicing of different proteins such as transcription factors, signal transducers and extra-cellular matrix components have been related to tumor cell growth [
34]. As a consequence, a number of interventions that either alter or exploit alternative splicing are currently under active investigation for cancer therapy [
35,
36]. Conceivably, detection of asTFPI-2 could find application in tumor diagnosis, or possibly serve as a marker for cancer progression.
Our laboratory, as well as others, have previously reported that overexpression of candidate tumor suppressor TFPI-2 gene in several highly metastatic tumor cells markedly inhibits their growth and metastasis
in-vivo by regulating pericellular extracellular matrix (ECM) remodeling and angiogenesis [
4‐
10]. Conversely, hypermethylation of tumor suppressor genes is a major epigenetic change that contributes to tumor progression and is correlated with transcriptional silencing. Hypermethylation of the TFPI-2 gene promoter has been described recently as a possible mechanism for gene silencing in a limited number of tumor cell lines [
3,
10‐
12,
37]. However, the prevalence of hypermethylation of the TFPI-2 promoter in primary tumors has not been systematically investigated. In many cancer cells, the hypermethylation of the TFPI-2 promoter does not appear to be the sole basis for TFPI-2 gene silencing, as one or more components of the cellular transcription machinery that regulates TFPI-2 expression may also be hypermethylated leading to their silencing as well [
13]. The existence of an aberrantly-spliced variant of TFPI-2 that is untranslated, and appears to be highly upregulated in tumor cells, suggests an additional, novel mechanism for further downregulation of TFPI-2 expression in tumor cells.
Methods
Materials
The human tumor cell lines J-82 (bladder cell carcinoma), Capan-2 (pancreatic adenocarcinoma), HepG2 (hepatocellular carcinoma), HT-1080 (fibrosarcoma), Colo-205 (colon adenocarcinoma), and HEK 293 (primary embryonal kidney), were obtained from American Type Culture Collection (Manassas, VA). The human embryonic fibroblast cell line WI-38 was kindly provided by Dr. LVM Rao (University of Texas, Tyler, TX). Human umbilical vein endothelial cells (HUVECs) were obtained from Cambrex (Walkersville, MD). Dulbecco's minimal essential medium (DMEM), penicillin, and streptomycin, actinomycin D and mouse monoclonal anti-tubulin antibody were purchased from Sigma-Aldrich (St Louis, MO). Fetal bovine serum was purchased from Hyclone (Ogden, UT). RNEasy® RNA extraction kit, Oligotex mRNA mini kit, and Plasmid Spin Miniprep kit were purchased from Qiagen (Valencia, CA). Normal human lung, liver and colon RNAs, as well as their corresponding tumor RNAs, were purchased from Chemicon Inc. (Temecula, CA). Human placental total RNA and SMART® RACE kit was purchased from BD Biosciences Clontech (Palo Alto, CA). RT-PCR kit TaqMan® and SYBR®-green PCR Master Mix were purchased from Applied Biosystems (Foster City, CA). AccuPrime™ Pfu Polymeras and DNase-I were purchased from Invitrogen (Carlsbad, CA). Nylon membrane, Nitrocellulose (NC) membranes and goat anti-rabbit IgG-HRP were purchased from Bio-Rad (Hercules, CA). Random primed labeling kit was purchased from Roche diagnostics (Indianapolis, IN). PCR-Script™ AMP cloning kit and E. coli XL1 Blue cells were from Stratagene (Cedar Creek, TX). Chemiluminescent HRP substrate was purchased from Millipore Corporation (Billerica, MA).
Cell culture and total RNA preparation
All cell lines were maintained in Dulbecco's minimal essential medium (DMEM), supplemented with 10% fetal bovine serum and penicillin-streptomycin. The cells were cultured at 37°C in a humidified atmosphere containing 6% CO2, and approximately 5 × 106 cells were processed for total RNA isolation. The total RNAs were purified using the RNEasy® RNA extraction kit according to the manufacturer's instruction. All purified RNA samples were further digested with to ensure complete removal of any contaminating genomic or mitochondrial DNA. The DNase-I was subsequently inactivated by incubation with 2 mM EDTA at 65°C for 15 min. Total RNA samples from HUVECs and Colo-205 cells were also subjected to further purification on oligo-dt columns (Oligotex) to obtain poly (A)+ RNA.
Rapid amplification of cDNA 5' and 3'ends (5'-and 3'-RACE)
The corresponding first-strand cDNAs were prepared from 1 μg of total RNA, using the PowerScript reverse transcriptase, 5'-CDS® primer (modified oligo-dT primer) and BD™ SMART IIA® primer using the BD™ SMART® RACE cDNA amplification kit. Dilutions of each 5' and 3' RACE-ready cDNAs were used in PCR amplification reactions with the SMART® RACE kit universal primer mix and either gene-specific antisense exon IV primers (5'-TAC TTT TCT GTG GAC CCC TCAC-3'), to amplify TFPI-2, or exon V primer (5'-TTG CTT CTT CCG AAT TTT CCG G-3') to amplify 5' ends of both TFPI-2/asTFPI-2. Alternatively asTFPI-2 specific primer (5'-CTT TTG CGG ATT CAA C-3') was also used in combination with the universal primer mix. The exon II sense primer was used to amplify 3' ends (5'-AAC GCC AAC AAT TTC TAC ACC T-3') in combination with the universal primer mix. The 5' and 3' RACE cDNAs syntheses and PCR amplifications were performed according to the manufacturer's instruction.
Nucleotide sequencing analysis
A series of oligonucleotide primers spanning from the cap region (FT2-CAP), exon I (FT2-SP1, FT2-Ex1A, FT2-Ex1B, FT2-Ex1C), intron A (FT2-Int1A, FT2-Int1B), exon II (FT2-Ex2), and exon V (RT2-R1) and 3'-UTR (RHT-OD, RT2-PAT), of the TFPI-2 gene were designed to identify the 5' and 3' boundaries of the asTFPI-2 transcript (Table
1). The PCR amplicons obtained from the primer set (FT2-Ex2 as sense and RHT-OD as antisense primers) were ligated into pPCR-Script™ AMP cloning vector and transformed into competent XL1 Blue cells. Plasmid DNAs were prepared from individually selected clones, using a Plasmid Spin Miniprep kit, and vector primers were used to for sequencing the inserts. The nucleotide sequence was determined by the dideoxy chain termination method using the BigDye Terminator Cycle Sequencing Kit in a Model 377 Sequencing System (Applied Biosystems).
Table 1
Oligonucleotide primers used to amplify asTFPI-2
FT2-CAP | GAA AGC CGC GCA CCT CCT |
FT2-SP1 | CTG CAC CAT GGA CCC CGC T |
FT2-Ex1A | TGT CGA TTC TGC TGC TTT TCC TG |
FT2-Ex1B | CTG CAC TGG GCG ATG CTG CT |
FT2-Ex1C | CTG CTC AGG AGC CAA CAG |
FT2-Int1A | AGC CTC GCT TTC TCC AGG TCC |
FT2-Int1B | TAG GAG CTA CGC CTG ACC ACT T |
FT2-EX2 | AGG AAA TAA CGC GGA GAT CTG TCT |
RT2-R1 | TTG CTT CTT CCG AAT TTT CCG G |
RHT-OD | TTAA AAT TGC TTC TTC CGA AT |
RT2-PAT | GTC ATA TTA TTC TTC AGA TAC |
Semi-quantitative RT-PCR of TFPI-2 transcripts
The first strand cDNAs were prepared from 2 μg of total RNA using random hexamers and a reverse transcriptase-PCR (RT-PCR) kit in a total volume of 50 μl. The TFPI-2 and asTFPI-2 transcripts were amplified using a common sense primer (5'-AGG AAA TAA CGC GGA GAT CTG TCT-3') and a common antisense primer (5'-TTAAAATTGCTTCTTCCGAAT-3'). The constitutively expressed endogenous housekeeping gene, cyclophilin A (using sense primer 5'-GTCTCCTTTGAGCTGTTTGC-3' and antisense primer 5'-AAGCAGGAACCCTTATAACC-3') was also amplified in separate tubes. The controlled semi-quantitative RT-PCR assays were performed using 1 μl of first-strand cDNA product and AccuPrime™ Pfu polymerase in a total volume of 50 μl. PCR conditions were 1 cycle of 94°C for 2 min; 30 cycles of 94°C for 30 sec (denaturation), 60°C for 30 sec (annealing) and 68°C for 1 min (extension) followed by 1 cycle of 68°C for 5 minutes (final extension). Aliquots of all PCR products were analyzed in 2% agarose gels and visualized with 0.1 mg/ml ethidium bromide staining. The relative intensity of each amplicon was quantified using Quantiscan software (Biosoft™) and intensities are reported in arbitrary units (A.U.). The intensity of each amplicon was normalized to 10,000 A.U. levels of cyclophilin A amplified from each sample.
Splice variant-specific quantitative RT-PCR
A two-step, quantitative real-time RT-PCR was performed in triplicate using an ABI Prism 7000 Sequence Detection System (Applied Biosystems) with standard temperature protocol and 2x SYBR Green PCR Master Mix reagent in a total volume of 25 μl. The first-strand cDNAs were prepared from 1 μg of total RNA in a final volume of 50 μl using random hexamers and a reverse-transcriptase PCR kit. For absolute quantitation, the PCRScript-TFPI-2 and PCRScript-asTFPI-2 plasmid serial dilutions, reflecting copy numbers (1 × 100-1 × 106), were used to obtain a standard curve correlating the copy number with the threshold cycle number (Ct values). Primers used for amplification were as follows: TFPI-2, sense (5'-AAC GCC AAC AAT TTC TAC ACC T-3') and antisense (5'-TAC TTT TCT GTG GAC CCC TCAC-3'); asTFPI-2, sense (5'-AAC GCC AAC AAT TTC TAC ACC T-3') and antisense (5'-CTT TTG CGG ATT CAA C-3'). The cDNA products were quantified from each sample in terms of the TFPI-2 and asTFPI-2 copy numbers. As a control, cyclophilin A copy numbers were obtained in a similar manner, and the relative copy numbers of TFPI-2 and asTFPI-2 were calculated after normalizing to 102 copies of cyclophilin A transcript for each sample. To assure the primers specificity, the amplicons were subjected to melting curve analysis, agarose gel electrophoresis and nucleotide sequencing. Statistical analysis was performed using GraphPad Prism software (San Diego, CA) and a P-value < 0.05 was considered significant.
Determination of mRNA stability
The relative stabilities of the wild-type TFPI-2 and as TFPI-2 transcripts were assessed in HUVECs and Colo-205 cells, respectively, following treatment of the cells with the transcriptional inhibitor, actinomycin D. Briefly, subconfluent cells were exposed to actinomycin D (5 μg/ml), and the cells harvested for total RNA extraction at selected times (0–24 h) following the addition of actinomycin D [
38]. RNA extraction, cDNA synthesis, and splice variant-specific quantitative real-time PCR analysis were performed as described above. The half-lives of TFPI-2 and asTFPI-2 transcript was determined by calculating linear regressions from a log-linear plot of mRNA expression versus time (SigmaPlot).
Northern Blotting
Total RNA from HUVEC and Colo-205 cells was prepared as described above. RNA (20 μg) was electrophoresed in a 1.2% agarose-formaldehyde gel, and transferred to a nylon membrane overnight by capillary action. The membrane was pre-hybridized at 65°C for 3 h and hybridized at 42°C overnight with an asTFPI-2 cDNA probe labeled with 32P-deoxycytidine triphosphate by random primed labeling (Roche) according to the manufacturer's instruction. The membrane was washed sequentially with 2% SSC containing 0.1% SDS for 15 minutes at room temperature, 0.5% SSC containing 0.1% SDS for 20 min at room temperature, and 0.1% SSC containing 1% SDS for 10 min at 65°C. The blot was then exposed to X-ray film at -70°C for 24–72 h.
Western blot analysis
Various cells (~1 × 10
6) were lysed by sonication in 500 μl of 125 mM Tris-HCl (pH 6.8) containing 2% SDS, 10% glycerol, 50 mM sodium phosphate, 10 mM benzamidine and 1 μg/ml aprotinin, centrifuged for 15 min at 10,000 × g at 4°C, and the supernatant recovered. The supernatants were boiled for 3 min and subjected to SDS-PAGE in 4–20% polyacrylamide gradient gels. Following electrophoresis, the proteins were electrotransferred to nitrocellulose (NC) membranes and subsequently blocked with 5% calf serum/TBS/0.1% Tween 20 at 37°C for 2 h. The NC membranes were then probed with rabbit anti-TFPI-2 IgG [
21], washed, and incubated with diluted goat anti-rabbit IgG-HRP. Immunoreactive proteins were identified using a chemiluminescent reagent system. Separate blots were treated with mouse monoclonal anti-alpha tubulin antibody in order to demonstrate equal loading of the proteins in the gel.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
P.K., H.S.C., and W.K. designed the research; P.K. discovered asTFPI-2 and performed the research; P.K., H.S.C., and W.K. analyzed the data; P.K. and W.K. wrote the manuscript.