Different regulation of the human thymidine kinase promoter in normal human diploid IMR-90 fibroblasts and HeLa cells.

Transcriptional activation of the human thymidine kinase (hTK) promoter plays an important role in the cell cycle control of thymidine kinase expression. Using the luciferase reporter cotransfection assay, we found that the activity of the hTK promoter in IMR-90 normal human diploid fibroblasts was increased by the constitutively over-expressed cyclin A or cyclin E but not by cyclin D, suggesting that the former two cyclins may act as positive regulators for the hTK promoter. The sequence responsible for the transcriptional activation by cyclin E was identified to be located between −133 and −92 of the hTK promoter. Regulation of the hTK promoter in HeLa cells appeared to be different from that in IMR-90 fibroblasts. Firstly, the hTK promoter in HeLa was already highly activated and could not be further activated by ectopically expressed cyclin A or E. Secondly, the −133 to −92 region of the hTK promoter was important for the promoter strength in HeLa cells but not in IMR-90 cells. The steady-state levels of cyclins A and E were readily detected in HeLa cells but not in normal IMR-90 fibroblasts. Based on these results, we propose that the cellular environment of the HeLa cell allows the hTK promoter to stay fully activated for transcription regardless of ectopically expressed cyclin A or E and that transcriptional activation of thymidine kinase gene is deregulated in these tumor cells.

Transcriptional activation of the human thymidine kinase (hTK) promoter plays an important role in the cell cycle control of thymidine kinase expression. Using the luciferase reporter cotransfection assay, we found that the activity of the hTK promoter in IMR-90 normal human diploid fibroblasts was increased by the constitutively over-expressed cyclin A or cyclin E but not by cyclin D, suggesting that the former two cyclins may act as positive regulators for the hTK promoter. The sequence responsible for the transcriptional activation by cyclin E was identified to be located between ؊133 and ؊92 of the hTK promoter. Regulation of the hTK promoter in HeLa cells appeared to be different from that in IMR-90 fibroblasts. Firstly, the hTK promoter in HeLa was already highly activated and could not be further activated by ectopically expressed cyclin A or E. Secondly, the ؊133 to ؊92 region of the hTK promoter was important for the promoter strength in HeLa cells but not in IMR-90 cells. The steady-state levels of cyclins A and E were readily detected in HeLa cells but not in normal IMR-90 fibroblasts. Based on these results, we propose that the cellular environment of the HeLa cell allows the hTK promoter to stay fully activated for transcription regardless of ectopically expressed cyclin A or E and that transcriptional activation of thymidine kinase gene is deregulated in these tumor cells.
Thymidine kinase (TK), 1 a crucial enzyme in the salvage pathway of thymidine triphosphate formation, is indirectly involved in DNA replication. The level of TK activity is known to be increased at the G/S phase of the cell cycle (1,2). Several mechanisms, including transcriptional activation (3)(4)(5)(6)(7), posttranscriptional processing (8 -10), and increase of translational efficiency (11)(12)(13), have been proposed to account for the precise timing associated with the induction of TK activity at the G 1 /S phase in normal cells. When the hTK promoter fused with a CAT reporter gene was transferred into Chinese hamster ovary fibroblasts by stable transfection, the sequence between Ϫ109 and Ϫ84 of the hTK promoter was found to be responsible for the TK transactivation during the G 1 /S transition period (14). Furthermore, several complexes containing cyclin A, p107, and p33 cdk2 that would bind to this DNA region were detected in the nuclear extracts isolated from growth-stimu-lated Chinese hamster ovary fibroblasts (15). These results prompted us to investigate the relationship between the expression of G 1 cyclins and the transcriptional activation of the hTK promoter in human cells.
Expression of cyclins A, D, and E has been shown to be an important driving force for the G 1 progression during the cell cycle (for review see Ref. 16). Many studies demonstrated that perturbations in G 1 cyclin expression caused inappropriate cell division, which would lead to the formation of cancer. For example, the cyclin A gene is the site of the integration of a fragment of the hepatitis B virus genome in hepatocellular carcinoma (17). Over-expression of cyclin D1 was shown to be a result of chromosomal rearrangement, translocation, retroviral insertion, and gene amplication in parathyroid tumors, lymphomas, squamous cell tumors, and breast and colorectal carcinomas (18 -24). Cyclin E was recently found to be over-expressed in cultured breast cancer cell lines and in primary breast tumors, and the concentration of cyclin E increased in breast tumor cells as the disease progressed toward severity (25). Presumably, aberrant expression of these G 1 cyclins could propel cells through critical transitions in the cell cycle. Also, the steady-state level of TK mRNA in normal human fibroblast has been shown to be increased in response to serum stimulation and appears to be closely associated with the stringent cell cycle control (26,27). In contrast, in HeLa cells TK mRNA in different phases of the cell cycle is constitutively and highly expressed (13,27). Furthermore, the level of TK activity was often found to be elevated in neoplastic tissues (28); however, it is still unclear whether or not this phenotype in tumor cells is due to deregulation at the level of transcriptional control and related to the abnormal expression of G 1 cyclin. In this study, therefore, we examined the in vivo effect of over-expression of human cyclins A, D1, and E on the hTK promoter activity in normal diploid IMR-90 fibroblasts as well as in HeLa cells, a tumor cell line, and characterized the regulation of the hTK promoter in these two different cell types.

MATERIALS AND METHODS
Cell Culture-IMR-90 human embryonic lung diploid fibroblasts (passage number 5; population doubling level, 12) were obtained from the Institute for Medical Research (Camden, NJ). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone) at 37°C in 5% CO 2 incubator and subcultured to obtain cultures with a population doubling level of 30 -40 remaining.
Transfection and Reporter Gene Assay-Cells were incubated with a mixture of 2.1 g of plasmid DNA and 12 g of lipofectamine (Life Technologies, Inc.) in 1 ml of Dulbecco's modified Eagle's medium for 6 h. The medium was then replaced with fresh Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and incubated further for 24 h. Cells were washed three times with phosphate-buffered saline and lysed with an additional 100 l of reporter lysis buffer (Promega). Cell lysates were centrifuged for 10 min at 10,000 ϫ g to remove the debris. 10 l of the supernatant were added with 100 l of luciferase assay buffer (29). The luminescence was measured in a Packard liquid scintillation counter. Protein concentration was determined directly from the cell lysate using the Bradford assay.
Recombinant Plasmid Construction-The DNA fragment containing Ϫ133 to ϩ33 and Ϫ91 to ϩ33 of the TK promoter sequence, respectively, was inserted in the sense orientation into the BglII site of a luciferase vector, pGL-2-Basic (Promega), to obtain p(Ϫ133/ϩ33)TK-Luc and p(Ϫ91/ϩ33)TK-Luc. Plasmids, pCMV cyclin A, pCMV cyclin D1, and pCMV cyclin E, were constructed by cloning the cDNAs encoding cyclins A (30), D1 (31), and E (31), respectively, into pCDM8 (32) at the XhoI site. To achieve this, the vector DNA and the cyclin cDNA fragments isolated from the corresponding plasmid after EcoRI digestion were all treated with Klenow enzyme to form blunt ends. They were then ligated by the conventional method. The insert orientation in the new recombinant plasmid was determined by restriction enzyme digestion profile.
RNase Protection Analysis-Total RNA was hybridized to a riboprobe at 46°C overnight. The hybridized mixture was exposed to RNases A and T1 at 37°C and analyzed in 4% polyacrylamide-urea gel. The hTK cDNA (33) cloned in pBluescript II SK vector (Stratagene) was linearized at the PvuI site and added to a transcription reaction mixture containing T7 RNA polymerase and [␣-32 P]CTP for the synthesis of the TK riboprobe from the cDNA. This TK riboprobe protected a 410nucleotide transcript derived from hTK mRNA in the assay. The human ␤-actin riboprobe was generated in a transcription reaction containing the template pTRI-␤-actin-Human (Ambion) linearized at the HindIII site. This human ␤-actin riboprobe protected a 250-nucleotide transcript derived from ␤-actin mRNA in the assay.
DNase I Footprint Analysis-Nuclear extracts of cells were prepared by the method of Dignam et al. (34). The 200-base pair EcoRI-ApaI fragment purified from pBluescript II SK vector containing the 160base pair hTK promoter region (Ϫ133/ϩ33) was labeled at one end by Klenow enzyme at the EcoRI site with [␣-32 P]dATP and used in the binding reaction as described previously (35). At the end of the binding reaction, 100 l with 3 units of DNase I (Worthington) in 10 mM MgCl 2 were added and incubated for 60 s at room temperature, and the reaction was stopped by the addition of an equal volume of the buffer containing 8 M urea, 0.5% SDS, and 5 mM EDTA. The digested probe was extracted with phenol/chloroform and precipitated by ethanol. Samples were then analyzed in 7 M urea-polyacrylamide sequencing gel. G ϩ A chemical sequencing reaction of the DNA probe was performed by the method of Maxam and Gilbert (36).
Western Blot Analysis-Cells grown on a 60-mm-dish were washed with phosphate-buffered saline and lysed at 4°C by sonication in TNE buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 2 mM EDTA, and 0.1 mg/ml of each leupeptin and aprotinin). The lysate was centrifuged at 10,000 ϫ g for 15 min to remove the debris and 50 g of protein in the lysate were separated in 10% SDS-polyacrylamide gel, followed by electrophoretic transfer onto a polyvinylidene difluoride membrane (Millipore). After a pretreatment with 5% skim milk, the membrane was incubated with the indicated antibodies. The antibodies and their dilutions were: rabbit polyclonal anti-human cyclin D1 (UBI), 1:200; monoclonal anti-human cyclin A (Santa Cruz), 1:5000; and monoclonal anti-human cyclin E (Santa Cruz), 1:3000. The alkaline phosphatase-conjugated goat anti-rabbit IgG (Promega) was used for the interaction with anti-human cyclin D1 antibody. The alkaline phosphatase color development was performed according to the procedure recommended by the manufacturer (Promega). The horseradish peroxidase-conjugated goat anti-mouse IgG was used for the detection of anti-human cyclins A and E monoclonal antibodies with an enhanced chemiluminescence method (Amersham Corp.). Exposure time was 30 s for cyclin E and 2 min for cyclin A.

The hTK Promoter Can Be Activated by Cyclin A or Cyclin E but Not by Cyclin D1 in IMR-90 Fibroblasts-
In an attempt to test whether or not there was a direct link between the activity of the hTK promoter and expression of G 1 specific cyclins in human cells, recombinant plasmids containing respective cDNA encoding human cyclin A, D1, or E, whose expression would be under the control of the cytomegalovirus early promoter, were constructed. Each of these G 1 cyclin-expressing plasmids was cotransfected into cells with a reporter plasmid, which contained a luciferase gene under the control of either the hTK promoter (Ϫ133 to ϩ33 region, p(Ϫ133/ϩ33)TK-Luc) or the SV40 promoter (pSV40-Luc). The effect of cyclin A, D1, or E in increasing amounts on the hTK promoter in IMR-90 fibroblasts was examined. All data were expressed relative to the luciferase activity from cells transfected with pSV40-Luc and a control expression vector, pCDM8. The results are shown in Fig. 1. Cyclin A enhanced the luciferase activity of p(Ϫ133/ ϩ33)TK-Luc in a dose-dependent manner. Up to a 3.5-fold increase was seen when 0.3 g or more of pCMV cyclin A was used (Fig. 1A). Cotransfection with pCMV cyclin D1 up to 0.6 g did not promote the hTK promoter activity (Fig. 1B). A significantly increased level of cyclin D1 was detected by Western blot analysis in the IMR-90 fibroblast transfected with pCMV cyclin D1, an indication of ectopic over-expression of cyclin D (Fig. 1C). Cyclin E was also found to steadily increase the luciferase activity of p(Ϫ133/ϩ33)TK-Luc reporter, reaching 3-fold when 0.6 g of pCMV cyclin E was used for cotransfection (Fig. 1D). All parallel experiments performed with the pSV40-Luc construct showed some repression when increasing amounts of these three pCMV-cyclins were used for cotransfection. Thus, activation of the hTK promoter by cyclin A or E in IMR-90 fibroblasts appears to be a specific event.
Next, p(Ϫ91/ϩ33)TK-luc, with a region from Ϫ133 to Ϫ92 deleted, was introduced by cotransfection either with pCMV cyclin A or pCMV cyclin E into IMR-90 fibroblasts to test whether or not the upstream sequence is involved in the activation. In the cells with this deletion construct, no significant increase in luciferase activity was seen with pCMV cyclin E, whereas a stimulation in response to pCMV cyclin A was still observed with this deletion construct (Fig. 2), suggesting that the region between Ϫ133 and Ϫ92 contained the element involved in the activation by cyclin E but not by cyclin A. Thus, the activation mechanism elicited by these two cyclins may be different. Furthermore, the upstream sequence did not seem to play an important role in the strength of the promoter activity in this normal cell strain, because no decrease in luciferase activity in cells transfected with p(Ϫ91/ϩ33)TK-Luc was found when compared with that in cells with p(Ϫ133/ϩ33)TK-Luc.
Little Enhancement of the hTK Promoter Activity by Cyclin A, D, or E in HeLa Cells-The in vivo effect of G 1 cyclins on the hTK promoter was also examined in HeLa cells. All data were similarly expressed relative to the luciferase activity under the control of the SV40 promoter in HeLa cells. In contrast to the results observed in IMR-90 fibroblasts, none of cyclins A, D, and E significantly enhanced the hTK promoter activity in this tumor cell line (Fig. 3, A, B, and C). Clearly, the luciferase activities expressed from the hTK relative to the SV40 promoters were already high without ectopic expression of cyclins in HeLa cells. In fact, the activities of either the hTK or SV40 promoters steadily decreased as the cells were cotransfected respectively with increasing amounts of the plasmids that express these three G 1 cyclins.
Correlation of the Level of the hTK Promoter Activity and the Amount of TK mRNA in HeLa Cells-The basal activity of the hTK promoter appeared to reach nearly 60% of the activity expressed from the SV40 promoter in HeLa cells, whereas it reached only 17% in IMR-90 fibroblasts, indicating that the hTK promoter was more active in HeLa cells. Table I summarizes the luciferase activities (in light counts) expressed by the SV40 promoter and the hTK promoter, respectively, in these two cell types. To verify whether or not the dramatic difference in promoter activity can be correlated with the steady-state level of TK RNA, the quantity of TK RNA in proliferating HeLa and IMR-90 cells was compared using RNase protection assay. As expected, the amount TK RNA in HeLa cells was clearly much more than that in IMR-90 fibroblasts (Fig. 4).
We then transfected p(Ϫ133/ϩ33)TK-Luc and p(Ϫ91/ ϩ33)TK-Luc, respectively, into HeLa cells to examine whether or not there was a difference in luciferase activity expressed from these two reporter plasmids. As shown in Fig. 5, a 50% reduction of luciferase activity with the deletion construct was clearly seen, suggesting that an activation process through the Ϫ133/Ϫ92 sequence is required for the maintenance of the maximum promoter activity in HeLa cells.
Footprint Analysis of the Human TK Promoter-Because the Ϫ133 to Ϫ92 region of the hTK promoter contributed to the promoter strength in HeLa cells but not in IMR-90 cells, it is possible that the capacity of cognate factors binding to the promoter region was different between HeLa and IMR-90 cells.
To determine this, the DNase I footprint analysis was carried out. When the nuclear extract prepared from serum-stimulated IMR-90 fibroblasts was used, the protected region was seen to span from Ϫ29 to Ϫ84, including the distal and proximal CCAAT boxes (Fig. 6, lanes 2 and 3). The region protected by nuclear protein binding disappeared in the presence of 40-fold excess unlabeled DNA fragment, indicating the specificity of the interaction between the DNA probe and nuclear factor(s). Under the same conditions, the DNA region that interacted with proteins from the HeLa extract appeared to be more broadly extended, covering the area upstream from Ϫ84 (Fig. 6,   6 and 7). Apparently, the protein binding pattern with the hTK promoter was quite different in these two cell types. Unlike with the extract of HeLa cells, there seemed to be no stable proteins bound to the 5Ј-flanking sequences upstream from Ϫ84 of the hTK promoter in the extract of IMR-90 cells. To some extent, this may explain that this upstream region did not contribute to the promoter activity in this normal cell strain.
The Steady-state Levels of Cyclins A, D1, and E in HeLa Cells and IMR-90 Fibroblasts-Because deregulation of each G 1 cyclin has been suggested to cause derangement of the cell cycle (16), here we investigated whether or not the steady-state level of cyclin A, D1, or E was correlated with the level of the hTK promoter activity in these two cell types. Lysates containing the same amount of protein prepared from semi-confluent cultures of IMR-90 fibroblasts and HeLa cells, respectively, were subjected to SDS-polyacrylamide gel electrophoresis, followed by Western blot detection with anti-cyclin A, cyclin D, or cyclin E monoclonal antibodies, respectively. The results showed that cyclin A was almost undetectable in IMR-90 fibroblasts but readily detectable in HeLa cells (Fig. 7A), whereas the level of cyclin D1 was rather similar in both cell types (Fig. 7B). As to RNase protection experiments were carried out with total RNA isolated, respectively, from HeLa cells and IMR-90 fibroblasts at semi-confluency. 2 g of RNA sample were hybridized with the ␤-actin probe, which protected a 250-nucleotide transcript from ␤-actin RNA. 5 g were used in hybridization with hTK-probe, which protected a 410-nucleotide transcript from TK RNA. The hybridized sample after RNases A and T1 digestion was electrophoresed in 4% polyacrylamide-urea gel (see "Materials and Methods"). The autoradiographic exposure times for ␤-actin probe and hTK probe were 4 and 24 h, respectively.

FIG. 5. Comparison of luciferase activity from p(؊133/؉33)TK-Luc and p(؊91/؉33)TK-Luc in HeLa cells.
HeLa cells were transfected with p(Ϫ133/ϩ33)TK-Luc or p(Ϫ91/ϩ33)TK-Luc (1 g) together with equal amount of the RSV-CAT plasmid, which contained the CAT gene under the control of the Rous sarcoma virus long terminal repeat. Extracts were prepared 24 h after the transfection and assayed for luciferase activity as well as CAT activity. Luciferase activity was normalized to the CAT activity in the same sample and expressed as percentage of that from p(Ϫ133/ϩ33)TK-Luc. cyclin E, there were two polypeptides detected in HeLa cells with molecular sizes of 62 and 52 kDa, of which only the former (62 kDa) was observed in IMR-90 fibroblasts (Fig. 7C). When IMR-90 fibroblast transfected with pCMV cyclin E, polypeptides of sizes 52 and 42 kDa were specifically over-expressed (Fig. 7D). The detection of the 62 kDa polypeptide also appeared in cells transfected with pCDM8, indicating that it was a nonspecific signal. Thus, cyclin E (the 52-kDa band) was readily detectable in HeLa cells but not in IMR-90 fibroblasts. Taken together, it seemed to show that unlike in HeLa cells, the amount of endogenous cyclins A and E was rather limited in the proliferating IMR-90 fibroblasts. DISCUSSION The results presented here establish four points: (i) cyclin A or cyclin E may act as the positive modulator of hTK promoter activity in normal IMR-90 human diploid fibroblasts; (ii) the transcriptional activation stimulated by cyclin E is via a region between Ϫ133 and Ϫ92 of the hTK promoter; (iii) the activity of the hTK promoter is much higher in HeLa cells and cannot be further induced by ectopic expression of cyclin A or E; and (iv) the region between Ϫ133 and Ϫ92 is required for the maximum promoter activity in HeLa cells but not in IMR-90 fibroblasts. These data also suggest that regulation of the hTK promoter in tumor cells is different from that in normal human cells and that the loss of transcriptional control of the TK gene is one of the events relevant to the deregulation at the G/S transition of the cell cycle in tumor cells.
Here, cyclin D expression seemed to exert little effect on the activation of the hTK promoter in both IMR-90 and HeLa cells. Because the activity of cyclin D-cdk4 complex occurs rather early during the G 1 progression (37), it is conceivable that cyclin D may not be involved in the transactivation of the hTK promoter directly. Furthermore, the steady-state level of cyclin D, unlike that of cyclin A or cyclin E, can readily be detectable in IMR-90 fibroblasts, suggesting that the expression of cyclin D is not as limited as that of cyclin A or cyclin E in this normal cell strain. In a normal cell cycle, cyclin E-cdk2 association is FIG. 6. Identification of the hTK promoter region covered by nuclear factor binding using DNase I footprint analysis. The DNA fragment containing 160-base pair (Ϫ133 to ϩ33) hTK promoter sequence was labeled at the coding strand and incubated with nuclear proteins (40 g) prepared from HeLa cells (lanes 6 and 7) or IMR-90 fibroblasts that have been serum-deprived for 48 h, followed by 16 h of serum stimulation (lanes 3 and 4), for DNase I footprint analysis. For lanes 4 and 7, the incubation mixture contained a 40-fold molar excess of unlabeled homologous DNA fragment. For lanes 2 and 5, the reaction mixture did not contain nuclear extract protein. required for the G 1 /S transition (38,39), whereas cyclin A-cdk2 association is needed for the S phase (40,41). Apparently, the trans-activation of the human TK promoter is well coordinated with theses two temporally coupled events at the G 1 /S transition during the cell cycle progression. Data presented here would support the model that the transcriptional complex formation involving cdk2, cyclin A, or cyclin E is critical for the transcriptional activation of cell cycle-regulated genes at the G 1 /S and S phases. Luciferase activities expressed from p(Ϫ133/ϩ33)TK-Luc and p(Ϫ91/ϩ33)TK-Luc showed little differences in IMR-90 fibroblasts (Fig. 2), suggesting that the deleted region (Ϫ133/Ϫ92) is not involved in the negative control; it is, however, needed for positive control by cyclin E. The sequence alteration in Ϫ84 to Ϫ109 of the hTK promoter has been shown to abolish G 1 /S phase regulation of the reporter gene that was under the control of the hTK promoter in the stably transfected Chinese hamster ovary cells (14). This Ϫ84 to Ϫ109 region contains two potential binding sites for transcription factor E2F and two Yi-like binding motifs (42). It remains to be seen whether or not ectopically expressed cyclin E, by forming a transcriptional complex with E2F or Yi-like factors, activates the hTK promoter via this upstream sequence. The activation mechanism by cyclin A, on the other hand, seems to differ from that by cyclin E, because the deletion of the Ϫ133 to Ϫ92 region did not affect the stimulation by cyclin A.
Our results indicated that the hTK promoter can be fully activated in IMR-90 only when there is a sufficient amount of cyclin A or cyclin E by ectopic expression (Fig. 1). In contrast, in HeLa cells it seems that all factors are already present for the transactivation. In other words, the cellular environment of this human papilloma virus-transformed tumor cell allows the hTK promoter to stay fully activated for transcription. The human papilloma virus E7 protein was shown to disrupt the interaction between E2F and the Rb protein or p107. This may in turn impair negative control of the cell cycle-dependent gene (43)(44)(45). We showed that the activity of the hTK promoter without the Ϫ133 to Ϫ92 region was significantly decreased in HeLa cells. However, we did not find a DNA-protein complex formation of this DNA sequence with E2F in HeLa nuclear extract. 2 Therefore, we do not know whether or not the greater activity of the hTK promoter and its unresponsiveness to cyclins A and E in HeLa cells are due to the lack of negative control on the promoter. Alternatively, it is possible that the level of endogenous cyclin E in HeLa cells is high enough to maintain stable promoter activity via the region Ϫ133 to Ϫ92. Our previous study has shown that human tumor cells exhibit constitutive binding activity to the distal CCAAT box located in the cell cycle-controlled region (Ϫ64/Ϫ133) of the hTK promoter (35). The TK-CCAAT-binding protein was later identified to be NF-Y (46), whose binding to the CCAAT elements in the hTK promoter may affect the promoter activity (47). Here, we show, in addition, that HeLa cells contain higher levels of cyclins A and E, which may act as positive modulators for the hTK promoter. Taken together, it is conceivable that the celluar levels of NF-Y, cyclin A, and cyclin E in HeLa cells may contribute to the loss of stringent transcriptional regulation of the hTK gene in HeLa cells, representing deregulation at the G 1 /S transition control.