Cyclin E is a G1 cyclin essential for G1 to S-phase transition of the cell cycle with a profound role in oncogenesis. In tumor cells and tissues, cyclin E is overexpressed and present in its lower molecular weight (LMW) isoforms, and it can be used as a prognosticator for poor patient outcome. In this study, we have examined differences in the processing of cyclin E between normal mammary epithelial and breast cancer cell lines. Five NH2-terminally deleted epitope-tagged (FLAG) cyclin E vectors were constructed spanning the range of LMW forms observed in tumor cells. These constructs were transfected into normal and tumor cells and analyzed for the production of cyclin E-FLAG protein products by Western blot analysis with FLAG and cyclin E antibodies. Our results show that only tumor cells had the machinery to process these cyclin E-FLAG constructs to their LMW forms, whereas normal cells mainly expressed the full-length unprocessed form of each protein. Tumor and normal cells always process the cyclin E-FLAG protein in the same way as endogenously expressed cyclin E. This phenomenon is consistent with all of the cell lines used, regardless of transfection efficiency, time of processing posttransfection, or method of transfection. Furthermore,measurement of FLAG-associated kinase activity in the transfectants revealed that the protein products of the cyclin E-FLAG constructs are 10 times more active in tumor cells than in normal cells. These studies suggest that the LMW forms of cyclin E detected at a much higher level in tumor cells arise from posttranslational action of a protease.
Progression through the cell cycle, the sequence of events between two cell divisions, is governed by the actions of positive and negative regulators in the eukaryotic cell. Cyclins and CDKs3serve as positive regulators, whereas CKIs serve as negative regulators of the cell cycle. To date, 10 different classes of cyclins (A-J, and L), 9 classes of CDKs (CDK1–9), and 2 classes of CKIs (CIP/KIP and INK) have been described, some of which have multiple members (1, 2, 3). In normal cells, the cyclins, CDKs, and CKIs work in concert to ensure a regulated transition from one phase of the cell cycle to the next. The level and pattern of expression of these regulators during the normal cell cycle are critical for the regulated progression through the cell cycle. In tumor cells, this exquisite balance between the positive and negative regulators is not maintained,thus contributing to the transformed phenotype.
The connection between cancer and the cell cycle has been established in part due to the alteration in the expression and function of G1 cyclins in cancer cells and tissues. For example, cyclin D1, a G1 cyclin that forms complexes with CDK4 and CDK6 (4) and whose major function is the phosphorylation of the retinoblastoma gene product pRb (5, 6, 7), was initially cloned as the PRAD1 proto-oncogene in some parathyroid tumors, where its locus is overexpressed as a result of a chromosomal rearrangement (8, 9). Cyclin D1 undergoes gene amplification and/or overexpression in a number of other tumors, including breast, colon, and ovarian cancers (10, 11). Cyclin D1 is also overexpressed in mammary cells of transgenic mice and results in abnormal proliferation of these cells and the development of mammary adenocarcinomas (12).
Cyclin E, another G1 cyclin that forms complexes with CDK2 and is essential for S-phase entry (13, 14), also has a profound role in oncogenesis (15, 16). Cyclin E expression occurs during a brief window of time from late G1 into early S phase, with a peak expression level near the restriction point (17, 18). Kinase activities of cyclin E/CDK2 complexes are also at maximum levels before S-phase entry (19). Functional knockout of cyclin E by injection of anti-cyclin E antibodies into fibroblast cells causes cell arrest in the G1 phase (20). Conversely, the overexpression of cyclin E protein causes acceleration through G1 along with a decreased cell size (20, 21). In addition to its requirement for DNA synthesis, cyclin E also plays a key role in senescence (22), development (23, 24), and modulation of downstream signals involving pRb (7) and E2F (25, 26). Due to the crucial role played by normal cyclin E expression and activity in cell proliferation, any defects in its expression could have a critical effect on oncogenesis.
The linkage between oncogenesis and cyclin E has been reinforced by correlating the altered expression of cyclin E to the loss of growth control in breast cancer (27, 28, 29). Furthermore, several tumor cohort studies (reviewed in Ref. 11) have documented a strong correlation between cyclin E overexpression and poor patient disease-free or overall survival (15, 30) and lack of estrogen receptor expression (31, 32, 33). In addition,patients with high cyclin E levels in their tumors had a significantly increased risk of death and/or relapse from breast cancer, even if they were node negative (30, 32). In our own studies (34),4, in which we examined tumor specimens from 403 breast cancer patients,we observe that cyclin E protein is the most consistent marker for determining the prognosis of early-stage node-negative breast carcinoma. Lastly, examination of the oncogenic potential of cyclin E in transgenic mice under the control of the bovine β-lactoglobulin promoter revealed a corroborating role for cyclin E in mammary tumorigenesis (35). Lactating mammary glands of the transgenic mice contained hyperplasia, and >10% of female transgenic mice also developed mammary carcinomas up to 13 months later (35). Collectively, these data provide strong support for the role of cyclin E overexpression in breast cancer tumorigenesis.
There are three main alterations in cyclin E expression that are seen in tumor cells, but not in normal cells: (a) amplification of the cyclin E gene and overexpression of cyclin E mRNA by 64-fold in a subset of breast cancer cell lines (27, 28);(b) cell cycle regulation of cyclin E expression is lost in some tumor cells, leading to constitutive cyclin E expression and activity throughout the cell cycle (16, 36). Such constitutive overexpression and activation of cyclin E also results in the functional redundancy of cyclin E/CDK2 in breast cancer cells because this complex has the ability to phosphorylate pRb under conditions in which cyclin D/CDK4/CDK6 complexes have been rendered inactive by overexpression of p16 (33); and (c)cyclin E expression in tumor cells is commonly characterized by the overexpression of the wild-type form and the appearance of LMW isoforms that are not present in normal cells or tissues (15, 27). The LMW isoforms of cyclin E usually appear as bands migrating between Mr 49,000 and Mr 34,000 as detected by Western blot analysis, whereas the wild-type cyclin E migrates at Mr 51,000 (15, 27). These isoforms are found in breast cancer cell lines as well as tumor tissue specimens from breast cancer patients (15, 29, 32). The expression of these cyclin E isoforms correlates very strongly with the stage, severity, and outcome of breast cancer (15, 32). The LMW isoforms of cyclin E that are linked to poor prognosis are also observed in other tumors such as colon cancer and hematological malignancies (Refs. 37 and 38; reviewed in Ref. 39). However, despite the tumor-specific expression of the LMW forms of cyclin E and the compelling prognostic evidence,very little is known about what gives rise to these isoforms.
We previously reported that the generation of LMW forms of cyclin E is not a result of genomic rearrangements of the cyclin E gene (16, 27). We and others have also identified several alternative splice variants of cyclin E (16, 20, 40). In addition to the authentic cyclin E, referred to as the wild-type form,there are four additional splice variants of cyclin E representing:(a) a variant termed E-L that adds 15 amino acids to the NH2 terminus of cyclin E (20);(b) a rare form of cyclin E termed E-S that lacks 49 amino acids containing the cyclin box motif (40); (c)a 9-bp in-frame deletion at the 5′ end of the gene termed Δ9 variant (16); and (d) a 148-bp deletion in the 3′ end of the gene termed Δ148 resulting in a frameshift of cyclin E (16). Both normal and tumor cells contain these variants;however, cyclin E protein expression clearly shows overexpression of the LMW isoforms in tumor cells, but not in normal cells (16). Because the LMW forms of cyclin E protein are only found in tumor cells, and the five cyclin E splice variants are found in normal and tumor cells, it suggests that the splice variants, by themselves, do not give rise to the cyclin E isoforms.
In this report, we have explored the hypothesis that a posttranslational proteolytic cleavage event is responsible for the LMW forms of cyclin E in tumor cells. We examined the possibility that tumor cells contain the machinery to process a full-length cyclin E protein into LMW forms, whereas normal cells do not. We introduced a full-length cyclin E cDNA that has been COOH-terminally tagged with FLAG sequence into tumor and normal cells and examined its pattern of expression/processing using antibodies against FLAG and cyclin E. These studies show that there is a profound difference in cyclin E processing between normal and tumor cells. Tumor cells process the FLAG-tagged cyclin E into LMW isoforms, identical to the endogenous cyclin E,whereas normal cells were unable to process the FLAG-tagged cyclin E to the LMW isoforms at high levels. Furthermore, the cyclin E processing is taking place at the NH2 terminus, most likely by a protease that is more active in tumor cells.
MATERIALS AND METHODS
Cyclin E-L, cyclin E, and three NH2-terminally truncated FLAG-tagged constructs were engineered by fusing the cloned cyclin E cDNA sequences at the COOH terminus to the sequence coding for the FLAG peptide. FLAG is an 8-amino acid peptide (N-Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys-C) that is immunogenic to a commercially available FLAG antibody (Santa Cruz Biotechnology, Santa Cruz, CA). The 5′ primers used to create the constructs presented in Fig. 1 B are as follows: cyclin E-L,5′-GAGCGGGACACCATGCCGAGGGAGCGCAGG-3′; cyclin E,5′-GAGCGGGACACCATGCGAAGGAGCGGGACA-3′; Trunk 1,5′-GAGCGGGACACCATGGATCCAGATGAAGAA-3′; Trunk 2,5′-GAGCGGGACACCATGGCAGTCTGTGCAGAC-3′; and Trunk 3 5′-GAGCGGGACACCATGTGGAAAATCATGTTA-3′. The 3′ primer containing the FLAG sequence is FLAGCE 5′-GCAAGCTTTTCACTTGTCATCGTCGTCCTTGTAGTCCGCCATTTCCGGCCCGCT-3′. The template used is cyclin E-L cDNA (20). The FLAG-cyclin E PCR product was ligated directly into the TA cloning vector, pCRII(Invitrogen, Carlsbad, CA). For each of the constructs, a cyclin E primer was used that added an identical Kozak ribosome binding site and start codon to allow for equal expression levels in vitroand in cells. Each truncated cyclin E PCR product was cloned directly into TA cloning vector PCRII (Invitrogen). Each FLAG-tagged cyclin E was then cloned into the mammalian expression vector pCDNA3.1(Invitrogen) under the control of a cytomegalovirus promoter. This vector system was used to allow for constitutive, transient expression of high levels of FLAG-tagged cyclin E in mammalian cells.
In Vitro Transcription and Translation.
To transcribe and translate the cyclin E-FLAG constructs cloned in pcDNA3.1 vector, we used the TNT-coupled reticulocyte lysate system (Promega, Madison, WI). Briefly, 1 μg of pCDNA3.1 plasmid containing either the cyclin E-FLAG inserts or no insert was added to rabbit reticulocyte lysate in the presence of T7 RNA polymerase and 1 mm complete amino acid mixture in a total volume of 50 μl and incubated at 30°C for 90 min. One μl of each of the translation products containing both the rabbit reticulocyte lysate and the synthesized cyclin E protein was then separated on a SDS-PAGE gel and subjected to Western blot analysis using either a monoclonal antibody to cyclin E (clone HE-12) or a polyclonal antibody to FLAG (both from Santa Cruz Biotechnology).
Transfection by electroporation was carried out on the tumor cell lines MDA-MB-157 and MDA-MB-436, the normal immortalized cell line MCF-10A, and the normal mortal cell strain 76N. All cells were cultured to 70% confluence in media as described previously (16). Voltages used for transfection were as follows: (a)MDA-MB-157, 0.28 kV; (b) MDA-MB-436, 0.23 kV; (c)MCF-10A, 0.3 kV; and (d) 76N, 0.32 kV (all at 960 μF capacitance). For each cell line, 1 × 107 cells were suspended in 0.5 ml of media, with 40 μg of plasmid (cyclin E-FLAG + pEGFPC-1) in a 0.4 cm gap cuvette. The pEGFPC-1 vector (CLONTECH, Palo Alto, CA) was cotransfected along with the cyclin E-FLAG vectors with a 1:4 ratio of GFP vector:cyclin E vector; the total DNA introduced was maintained at 40 μg. For MDA-MB-157 and MDA-MB-436 tumor cells, transfection was carried out in serum-free α-MEM; for MCF-10A cells and 76N normal cells, transfection was carried out in complete dmedium (41) with 1% serum. After transfection,cells were plated with complete medium and harvested 24 h after transfection for analysis. The transfection time course analysis was performed on MDA-MB-157 and MCF-10A cells. Transfection conditions were identical, except that cells were harvested at time intervals of 12,24, 48, and 72 h after transfection.
Flow cytometry was performed to analyze transfection efficiency by examining GFP expression. After transfection, 1–5 × 106 live cells were harvested by centrifugation at 1000 × g for 10 min and resuspended in PBS. GFP expression was measured on a Becton Dickinson FACScan flow cytometer (Becton Dickinson, San Jose, CA) using an excitation wavelength of 350 nm and absorbance at 485 nm. Data were analyzed using the CellQuest program (Becton Dickinson), and efficiency was measured as a percentage of cells expressing GFP over background fluorescence.
Western Blotting and Kinase Assays.
Cell lysates from transfected cells were prepared and subjected to Western blot analysis as described previously (42). Briefly, 50 μg of protein from each condition were electrophoresed in each lane of a 10% SDS-polyacrylamide gel (SDS-PAGE) and transferred to Immobilon P overnight at 4°C at 35 mV constant voltage. The blots were blocked overnight at 4°C in Blotto [5% nonfat dry milk in 20 mm Tris, 137 mm NaCl, and 0.25% Tween(pH 7.6)]. After six 10-min washes in TBST [20 mm Tris,137 mm NaCl, and 0.05% Tween (pH 7.6)], the blots were incubated in primary antibodies for 3 h. Primary antibodies used were cyclin E monoclonal antibody (Santa Cruz Biotechnology) at 1 μg/ml, anti-FLAG polyclonal antibody (Santa Cruz Biotechnology) at 0.25 μg/ml, and actin monoclonal antibody (Boehringer Mannheim,Indianapolis, IN) at 0.63 μg/ml. All dilutions were made in Blotto. After primary antibody incubation, the blots were washed and incubated with the appropriate goat antimouse or antirabbit horseradish peroxidase conjugate at a dilution of 1:5000 in Blotto for 1 h and then washed and developed with the Renaissance chemiluminescence system as directed by the manufacturer (NEN Life Sciences Products, Boston,MA).
For histone H1 kinase assay, 100 μg of cell extracts were used per immunoprecipitation with polyclonal antibody to FLAG or CDK2 (42) in lysis buffer containing 50 mm Tris buffer (pH 7.5), 250 mm NaCl, 0.1% NP40, 25 μg/ml leupeptin, 25 μg/ml aprotinin, 10 μg/ml pepstatin, 1 mmbenzamidine, 10 μg/ml soybean trypsin inhibitor, 0.5 mmphenylmethylsulfonyl fluoride, 50 mm NaF, and 0.5 mm sodium orthovanadate. The protein/antibody mixture was incubated with protein A-Sepharose for 1 h, and the immunoprecipitates were then washed twice with lysis buffer and washed four times with kinase buffer [50 mm Tris-HCl (pH 7.5),250 mm NaCl, 10 mm MgCl2,1 mm DTT, and 0.1 mg/ml BSA]. The immunoprecipitates were then incubated with kinase assay buffer containing 60 μmcold ATP and 5 μCi of [32P]ATP in a final volume of 50 μl at 37°C for 30 min. The products of the reaction were then analyzed on a 13% SDS-PAGE gel. The gel was then stained,destained, dried, and exposed to X-ray film. For quantitation, the protein bands corresponding to histone H1 were excised, and the radioactivity of each band was measured by Cerenkov counting.
Generation of LMW Cyclin E Constructs.
A Western blot analysis of cyclin E illustrating the pattern of cyclin E LMW isoforms detected in breast tumor cells and tissue samples from breast cancer patients is shown in Fig. 1,A. The MDA-MB-157 breast cancer cell line and several tumor tissues obtained from breast cancer patients with different stages of the disease (see Fig. 1legend) overexpress the LMW forms of cyclin E. The electrophoretic mobility of the LMW forms of cyclin E is similar in both the MDA-MB-157 cell line and the tumor tissue samples. The normal, human mammary epithelial cell strain 76N, on the other hand, expresses predominantly the full-length form of cyclin E. The appearance of the LMW forms of cyclin E in breast tumor tissues and cell lines is predictive of their physiological roles in tumors but not normal cells or tissues.
To investigate the proteolytic processing of cyclin E in normal and tumor cells creating the LMW isoforms, a series of cyclin E constructs were engineered with the epitope FLAG sequence at the COOH terminus (Fig. 1,B). Five different FLAG-tagged cyclin E constructs were engineered and are schematically presented in Fig. 1,B. These constructs include cyclin E-L-FLAG (the splice variant with a 15-amino acid insertion at the 5′ end; Ref. 20), E-FLAG (the wild-type cyclin E), and three NH2-terminal truncated cyclin E constructs designated Trunk 1–3-FLAG constructs (Fig. 1,B). These five cyclin E constructs were created to serve two purposes: (a)the E-L-FLAG and E-FLAG constructs will be used to determine whether a full-length cyclin E can give rise to the LMW forms in tumor or normal cells. In addition, these two constructs show that the E-L form is the predominant cyclin expressed in both normal and tumor cells (see Fig. 1 A); and (b) the smaller trunks will be used to bracket the LMW forms expressed in tumor cells. Our previous studies suggest that the endogenous LMW isoforms of cyclin E found in tumor cells are a result of NH2-terminal deletions of the protein (Ref. 16; data not shown). By comparing the mobility of the cyclin E trunk forms with the cyclin E LMW forms on a Western blot, the sizes of the isoforms can be more accurately determined.
The expression of the truncated cyclin E cDNAs was first examined by in vitro transcription/translation to analyze the sizes of their respective protein products. FLAG-tagged cyclin E-L, E, and Trunks 1–3 as well as the pCDNA3.1 vector with no insert were synthesized by in vitro transcription/translation and then subjected to Western blot analysis with both cyclin E (Fig. 1,C) and FLAG (Fig. 1,D) antibodies. In both Western blots, Lane 1 represents 50 μg of MDA-MB-157 total cell extract, which was used as a positive control for the cyclin E antibody (Fig. 1,C) detecting endogenous cyclin E. Lane 1 also serves as a negative control for the FLAG antibody (Fig. 1,D). Lane 2 is the negative control for vector pCDNA3.1 with no insert in the transcription/translation reaction and shows no cyclin E protein as detected by the cyclin E or FLAG antibodies. Lanes 3–7 are the in vitrotranslated products of the five different cyclin E-FLAG constructs. The size difference for each trunk can clearly be seen as each lane shows a smaller sized cyclin E band (Fig. 1, C and D). The predicted sizes of the cyclin E-FLAG Trunk forms determined by sequence information of the proteins produced by the constructs are between Mr 48,000 for E-L-FLAG and Mr 34,800 for Trunk 3-FLAG. The FLAG tag adds an additional Mr1,000 to the size of the cyclin E protein (Table 1). However, the actual sizes of the in vitro transcribed trunks, as determined by gel migration, are quite different and range from Mr 52,000 for E-L-FLAG to Mr 34,000 for Trunk 3-FLAG (Table 1). The differences in cyclin E molecular weight determined by Western blotting of the in vitro translated constructs versus those predicted from the amino acid sequence suggest that cyclin E protein migrates anomalously on an SDS-PAGE. In addition,smaller forms of cyclin E are detected on the Western blots, probably as the result of alternate translation start sites present within the cyclin E cDNA. For example, cyclin E-L-FLAG also produced a Mr 46,000 band that comigrates with cyclin E-FLAG at Mr 46,300. In addition, cyclin E-L-, E-, and Trunk-1-FLAG constructs all synthesized a protein migrating at ∼ Mr 41,000; this is most likely a translation start site at bp 136–138, which is close to Trunk 2-FLAG migrating at Mr 41,200. Trunk 3-FLAG produces an additional band at Mr30,600 that may be from a translation start site at bp 358–361 within the middle of the cyclin E gene. The cyclin E constructs used in this study span the entire range of cyclin E LMW forms from about Mr 51,000 to Mr 34,000 detected in MDA-MB-157 (Fig. 1,A, Lane 2 and Fig. 1,C, Lane 1) and tumor tissue samples (Fig. 1 A, Lanes 1–5, 7,and 10).
Differential Processing of Cyclin E in Normal versusTumor Cells.
The in vitro translation of the cyclin E constructs generated only the full-length and the alternative start site protein products (Fig. 1). However, the pattern of cyclin E expression in breast cancer cells and tissues is indicative of a more complex processing of cyclin E (Fig. 1,A). To compare normal and tumor processing, we transfected two different sets of normal and tumor cells with the cyclin E-FLAG constructs and examined the expression and associated kinase activity of the protein products from these constructs (Fig. 2).
The two normal cell lines and the two tumor cell lines used for these study are the tumor cell lines MDA-MB-157 and MDA-MB-436 and normal cell lines MCF-10A (immortalized) and 76N (mortal) cell strain. All cells were transfected with either the vector backbone, cyclin E-L-FLAG, or cyclin E-FLAG constructs. Because differences in transfection efficiencies could account for differential processing of cyclin E between cells, we monitored transfection efficiencies by cotransfection with the GFP-expressing vector pEGFPC-1, and the percentage of efficiency was determined by measuring GFP expression by flow cytometry (see “Materials and Methods”). Table 2 lists the transfection efficiencies of the four cell lines used in this study.
After transfection with the indicated vectors and determination of transfection efficiency, cell extracts were prepared and subjected to Western blot analysis with FLAG antibody (Fig. 2, A and B). The results revealed that in both MDA-MB-157 and MDA-MB-436 tumor cells, cyclin E-L-FLAG and cyclin E-FLAG were processed into several LMW isoforms as well as the full-length form,generating a pattern of cyclin E-FLAG expression similar to endogenously expressed cyclin E (Fig. 2, A and B, Lanes 3 and 4). [The generation of these LMW forms of cyclin E in tumor cells is independent of the method of cell lysis, protein, extraction, or buffer used to lyse cells before Western blotting (data not shown)]. This result suggests that a single full-length cyclin E has the ability to give rise to LMW forms in tumor cells. Moreover, these forms most likely arise from a posttranslational processing event, because the cyclin E cDNA used was isolated from a mature mRNA that was already spliced (17, 20). The LMW forms produced from the full-length cyclin E-L-FLAG and cyclin E-FLAG range from Mr 51,000 to Mr 34,000, which is the same range as endogenous LMW forms detected in nontransfected cells.
The pattern of expression of cyclin E-L-FLAG and cyclin E-FLAG in normal cells is quite different than that observed in tumor cells (Fig. 2, A and B). Transfection of either cyclin E-L-FLAG or cyclin E-FLAG into either MCF-10A or 76N cells results mainly in the expression of the full-length protein products (Fig. 2, A and B, Lanes 7 and 8). There was a slight expression of LMW forms in normal cells after transfection with cyclin E-FLAG; however, the levels of these proteins were much lower than those of the LMW forms expressed by tumor cells. The inability of normal cells to express high levels of LMW forms of cyclin E-L-FLAG or cyclin E-FLAG proteins from the full-length form shows that the LMW processing is much less active in normal cells than in tumor cells. The similar Western blot pattern of expression between endogenous and transfected cyclin E in normal cells again shows that the transfected cyclin E-FLAG constructs are being expressed similarly to the endogenous cyclin E. The levels of expression of the LMW forms of both the endogenous and transfected cyclin E in normal cells are always much lower than the levels exhibited by tumor cells(Fig. 1 A, Lanes 1 and 2). This suggests that the machinery to process cyclin E may be present in both normal and tumor cells; however, this processing machinery is more active in tumor cells.
We found that transiently expressed cyclin E-L-FLAG and cyclin E-FLAG can activate CDK2, and the kinase activation is greater in tumor cells(Fig. 2, C and D). Cyclin E-associated kinase activity was measured by the phosphorylation of histone H1 in immunoprecipitates prepared from transfected normal and tumor cells using an antibody to FLAG. This analysis revealed that cyclin E-FLAG-associated kinase activity was 10-fold higher in tumor cells than in normal cells. This was surprising because equal amounts of cyclin E-L-FLAG or cyclin E-FLAG constructs were transfected into the cell lines as determined by efficiency studies (Table 2). The increased kinase activity associated with FLAG in tumor cells may be due not only to the processing of cyclin E-FLAG into its LMW forms but also to its overexpression. Furthermore, the high kinase activity in tumor cells suggests that the LMW isoforms of cyclin E-FLAG products can activate the kinase and that there is more CDK2 present in tumor cells. We also measured CDK2-associated kinase activity in both normal and tumor cells and found that tumor cells harbor a higher level of CDK2 activity than normal cells (Fig. 2, C and D). Transfection of cyclin E-L-FLAG and cyclin E-FLAG in normal and tumor cells increases cyclin E-FLAG-associated kinase activity in tumor cells due to the increased processing of these constructs and the greater amount of CDK2 in tumor cells.
Processing of Cyclin E in Tumor Cells Is Independent of Time,Transfection Efficiency, Cell Lines Used, or Method of Transfection.
To examine changes in the pattern of cyclin E processing in tumor and normal cells over time, cyclin E-L-FLAG and cyclin E-FLAG constructs were transfected into MDA-MB-157 and MCF-10A cells. Protein was extracted at 12, 24, 48, and 72 h after transfection and subjected to Western blot analysis with antibodies to cyclin E and FLAG (Fig. 3). The decreasing expression of cyclin E-L-FLAG and cyclin E-FLAG can be seen over the time course examined in both MDA-MB-157 (Fig. 3,A) and MCF-10A (Fig. 3,B) cells. The maximum level of expression peaks between 12 and 24 h and then drops steadily in both cell lines. Again, fewer cyclin E LMW forms are found in the MCF-10A cells (Fig. 3,B) than in MDA-MB-157 tumor cells (Fig. 3 A). Tumor cells process cyclin E into LMW forms as early as 12 h after transfection. Furthermore, at every time interval examined, the relative levels of the isoforms and the full-length form do not change in the tumor cells, revealing a persistent pattern of cyclin E-FLAG overexpression over the time of transfection. These results suggest that there is a balance between steady-state synthesis and proteolysis of cyclin E over time and that this balance is maintained regardless of time or the degree of transfection.
The overexpression of transfected cyclin E-L-FLAG and cyclin E-FLAG in both cell lines is evident in the total cyclin E expressed(i.e., endogenous plus transfected) using an antibody to cyclin E that reacts to both forms of cyclin E (Fig. 3, Aand B, middle panel). The vector alone(i.e., mock-transfected) lanes in Fig. 3, A and B, establish the endogenous baseline amount of cyclin E in each cell line. Although cyclin E-FLAG is significantly overexpressed transiently in normal cells, it does not give rise to the LMW isoforms. The amount of cyclin E overexpressed is similar between normal and tumor cells, as reflected by the similar transfection efficiencies in these two cell lines (55–68%; Table 2). The processing of cyclin E in tumor cells is independent of transfection efficiency and of the expression level of cyclin E. However, it is possible that although the same amount of plasmid was delivered into each cell type, tumor cells may express cyclin E-FLAG at a higher level than normal cells. Nonetheless, we have found that even when very low amounts of plasmid are delivered to tumor cells (i.e., 10–15% transfection efficiency), the tumor cells process cyclin E-FLAG into its LMW forms(data not shown). In fact, we have investigated the effect of transfection efficiency on the processing of cyclin E in tumor and normal cells and have found that tumor cells with a wide range of transfection efficiencies (i.e., 10–70%) always process cyclin E into its LMW forms, whereas normal cells, even those with the highest transfection efficiency (i.e., 80%) never process cyclin E into its LMW forms (data not shown). Therefore the appearance of the LMW form of cyclin E is independent of the amount of cyclin E-FLAG expression. In Fig. 2, we also show that the processing of cyclin E occurs similarly in tumor cell lines MDA-MB-157 and MDA-MB-436. The transfection efficiency in these two cell lines is very different (Table 2); however, the pattern of cyclin E processing is always similar to the pattern of endogenous cyclin E, suggesting that the transfected cyclin E is treated the same as the endogenous form. Lastly, the processing of cyclin E is independent of the method of transfection used. For all of the experiments depicted in this study, we have used the electroporation method of transfection. However, identical patterns of cyclin E processing in tumor cells were seen when other methods of transfection were used; such as lipofection,calcium phosphate, or GenePORTER (data not shown). Collectively, our data suggest that the processing of cyclin E in tumor cells is independent of time, transfection efficiency, cell lines used, or method of transfection.
Differential Processing of All Cyclin E-FLAG Constructs in Normal versus Tumor Cells.
Because tumor cells are able to process the full-length cyclin E into LMW forms, the next step was to determine more precisely which region of the cyclin E protein is subject to the processing. Transfection of all five cyclin E-FLAG vectors (see Fig. 1,B) into normal and tumor cells will bracket the endogenous LMW forms of cyclin E found in tumor cells and define their approximate mass. MDA-MB-157 and MCF-10A cells were transfected with each of the five cyclin E-FLAG constructs (presented schematically in Fig. 1,B) and the vector backbone, harvested 24 h after transfection, and subjected to Western blot analysis with FLAG and cyclin E antibodies (Fig. 4).
Transfection of each of the cyclin E-FLAG constructs into tumor cells resulted in their processing into LMW forms that fell within the range of Mr 52,000 to Mr 36,000 (Table 1). The largest protein produced from each construct migrated with a mobility similar to that observed by in vitro translation (Table 1). However,MDA-MB-157 cells were able to further generate LMW forms from all of the constructs. For example, cyclin E-L-FLAG expressed in vitro generated one major protein migrating at Mr 52,000 and two minor bands at Mr 46,000 and Mr 41,800, detected by a FLAG antibody(see Fig. 1, C and D; Table 1). However, the same construct transfected in tumor cells generated six major proteins migrating between Mr 37,000 and Mr 52,000 (Fig. 4,A; Table 1). This result indicates that alternate translational start sites active in the in vitro translation system do not account for all of the LMW forms of cyclin E produced by tumor cells. Additionally,each of the constructs transfected in the tumor cells is also processed into the next series of LMW forms, and the LMW isoforms common between all constructs comigrate (Fig. 4,A, Lanes 2–6;Table 1). Because the FLAG tag is on the COOH end of each construct,these data provide additional evidence that the processing of cyclin E occurs from the NH2-terminal end of the cyclin E protein and that the COOH terminus remains intact.
Transfection of MCF-10A cells with each of the cyclin E-FLAG constructs results mainly in the expression of the full-length form of each construct with minimal processing (Fig. 4,B, Lanes 2–6). This is consistent with the results obtained in the transfection of cyclin E-L-FLAG and cyclin E-FLAG constructs into MCF-10A and 76N cells (Fig. 2). The protein products of each MCF-10A transfected cyclin E-FLAG construct migrated with the same mobility as the in vitro synthesized form as well as the longest form of each construct after transfection in MDA-MB-157 cells (Table 1). Based on these comparisons, the sizes of all of the cyclin E processed proteins detected in MDA-MB-157 cells were estimated within 6 amino acids and listed in Table 1.
Lastly, expression of total cyclin E (i.e.,endogenous + transfected), as detected by cyclin E antibody,highlights the levels of overexpression of transfected cyclin E-FLAG in both cell lines (Fig. 4, middle panels). These results show that although the cyclin E-FLAG constructs were overexpressed in both normal and tumor cells to the same extent, only tumor cells can process cyclin E into LMW isoforms, whereas normal cells have a reduced capacity for further processing.
In this study, we provide evidence that a tumor-specific proteolytic activity generates the LMW forms of cyclin E detected in tumor cells. The physiological relevance of the LMW forms of cyclin E unique to tumor cells has been substantiated by their role in the prognosis of breast cancer and their constitutive expression and activation throughout the tumor but not the normal cell cycle (see“Introduction”). The generation of LMW forms of cyclin E could potentially occur at several different levels including alternative splicing of the cyclin E gene. Although five different splice variants of cyclin E have been identified to date (16, 20, 40), all splice variants are present at similar levels in normal and tumor cells and cannot account for the generation of the LMW isoforms of cyclin E protein detected in tumor cells and tissues. Other translational or posttranslational events that could give rise to the LMW isoforms of cyclin E include alternative translation start or stop sites and proteolytic cleavage of cyclin E. Analysis of the cyclin E sequence reveals that there are simply not enough start codons that lie within the size range of the gene, which could give rise to all LMW forms of cyclin E between Mr 51,000 and Mr 35,000. Premature termination of translation of cyclin E would give rise to LMW forms of cyclin E but would not contain the COOH terminus sequence required for antibody detection (43). Although posttranslational modification of cyclin E such as glycosylation or phosphorylation could potentially account for the LMW forms of cyclin E, such a possibility is unlikely. Glycosylation of cyclin E has never been detected and cannot result in faster migration of proteins. The full-length form of cyclin E is already hyperphosphorylated, and although autophosphorylation of cyclin E at Thr380 occurs (44), it cannot account for a Mr 16,000 difference in molecular weight (i.e., between Mr51,000 and Mr 35,000) forms of cyclin E detected in tumor cells. Posttranslational proteolytic cleavage seems reasonable for generating tumor cell-specific LMW forms of cyclin E.
As a strategy to study the processing of cyclin E in normal and tumor cells, we introduced full-length cyclin E cDNA tagged with FLAG into normal cells and tumor cells and examined the transfected cyclin E-FLAG for processing into LMW forms. This strategy was used to determine whether the processing occurs posttranslationally and to reveal any processing differences between normal and tumor cells. We found that tumor cells processed the full-length cyclin E-FLAG into its LMW isoforms, and normal cells predominantly expressed the full-length form. Furthermore, cyclin E-FLAG protein products have a much higher associated kinase activity in tumor cells than in normal cells. This increased cyclin E activity is due to increased free CDK2 in tumor cells (data not shown) and the increased ability of the LMW forms of cyclin E to form active complexes with CDK2.
The processing of the transfected cyclin E-FLAG seen in normal and tumor cells always reflects the processing of endogenous cyclin E. This similarity suggests that the same machinery that processes the endogenous cyclin E is also processing the transfected cyclin E. Furthermore, the processing occurs at a low level in normal cells,which suggests that it is an overactive mechanism in tumor cells,acting to cleave the full-length cyclin E into its LMW forms. Although tumor cells express high levels of cyclin E, the processing is not a function of overexpression because normal cells do not process cyclin E, even when it is transiently overexpressed to the same extent as in tumor cells. These results suggest that overburdening the cell with cyclin E does not, by itself, result in the generation of the LMW forms. Both 76N and MCF-10A cells can accommodate overexpression of cyclin E proteins without processing them to their LMW forms. Hence,the difference in processing of cyclin E detected between tumor and normal cells is not a result of overexpression of cyclin E but is more likely due to the action of proteases active in tumor cells that can process cyclin E to its LMW forms.
The cyclin E processing event seems to involve a cleavage of the NH2-terminal end of cyclin E, yielding LMW isoforms that span the region of Mr51,000 to Mr 35,000. There is already some evidence that the processing event is occurring at the NH2 terminus of cyclin E. The antibodies used to detect the LMW forms of cyclin E (i.e., clone HE-12 or FLAG)both recognize the COOH terminus of cyclin E, suggesting an intact COOH terminus. Based on these findings, we hypothesize that tumor cells may overexpress or activate a protease that could cleave the NH2-terminal region of cyclin E at several specific sites to generate the LMW forms. Additionally, the NH2-terminal secondary structure of cyclin E is not critical for the processing of cyclin E. The results from Fig. 4clearly show that transfection of Trunk 3-FLAG, the smallest construct made (Fig. 1 B), results in its cleavage very close to its NH2 terminus. The proximity of the Trunk-3 cleavage site to its NH2 terminus is not likely to allow for a significant NH2-terminal secondary structure.
Based on the unique pattern of cyclin E processing observed in tumor cells but not in normal cells and the definitive role of cyclin E for DNA replication, we propose that tumor cells harbor proteases that cleave cyclin E into its LMW forms. These proteases have to be localized in the nucleus and act independently of the proteasome. Localization of the cyclin E protease to the nucleus is necessary because cyclin E immunostaining with antibodies that detect the LMW forms always shows nuclear localization of the cyclin E signal (Refs. 20 and 29; data not shown). The appearance of both the LMW forms and the unprocessed full-length forms of cyclin E in tumor cells provide evidence for a non-proteasome-mediated cleavage of cyclin E. Although the proteasome pathway has been implicated for the degradation of cyclin E (44, 45), it is not likely to be responsible for the generation of the LMW forms of cyclin E observed in tumor cells. The manner by which the proteasome proteolytic machinery degrades proteins is either all or none. Once a protein has been tagged for proteasome degradation, it is completely degraded. Because the LMW forms of cyclin E are present constitutively in the tumor cells (16), they do not represent the intermediate proteolytic products of degradative machinery. These LMW forms of cyclin E are more likely due to the action of a protease. In fact, the analysis of the molecular weights generated by the different cyclin E-FLAG constructs used in this study helped to identify two regions in the NH2 terminus of cyclin E that are cleaved to give rise to the LMW forms of cyclin E detected in tumor cells. These two motifs in the cyclin E sequence are potential protease cleavage sites. Current biochemical and molecular approaches have identified these sites as target sequences for a serine protease with high activity in tumor cells.5
In summary, we show that tumor cells contain the machinery to process epitope-tagged cyclin E-FLAG constructs into LMW isoforms, identical to the endogenous cyclin E, at a much higher degree than normal cells. The processing of cyclin E is independent of the amount of cyclin E, takes place at the NH2 terminus, and is most likely performed by the action of a nonproteasome nuclear protease with high activity in tumor cells.
We thank Dr. Andrew Koff for the cDNA to cyclin E-L. We also gratefully acknowledge the use of Wadsworth Center’s Immunology,Tissue Culture, and Photography/Graphics core facilities.
Supported in part by Grant DAMD-17-94-J-4081 from the United States Army Medical Research Acquisition Activity and by Grant R29-CA666062 from the National Cancer Institute (both to K. K.). R. M. H. was supported by a fellowship (BC980981) from the United States Army Medical Research Acquisition Activity.
The abbreviations used are: CDK,cyclin-dependent kinase; CKI, CDK inhibitor; LMW, lower molecular weight; GFP, green fluorescent protein; TNT, in vitrotranscription and translation.
S. Bacus, M. Lowe, T. Herlizek, C. Danas,W. Toyofuku, and K. Keyomalsi, Cyclin E, a novel predictor of metastasis for low-stage node-negative breast carcinoma, manuscript in preparation.
D. C. Porter, C. Danes, and R. H. Harwell. Elastase proteolysis of cyclin E in breast cancer cells,manuscript in preparation.
|Constructs .||Predicted Mr .||In vitro translated Mr .||MDA-MB-157 transfected Mr .||MCF-10A transfected Mr .|
|Constructs .||Predicted Mr .||In vitro translated Mr .||MDA-MB-157 transfected Mr .||MCF-10A transfected Mr .|
|Cell lines .||% GFP expression .|
|MDA-MB-157||55% (n = 10)|
|MCF-10A||68.4% (n = 10)|
|MDA-MB-436||14% (n = 5)|
|76N||13% (n = 5)|
|Cell lines .||% GFP expression .|
|MDA-MB-157||55% (n = 10)|
|MCF-10A||68.4% (n = 10)|
|MDA-MB-436||14% (n = 5)|
|76N||13% (n = 5)|