ABSTRACT
Cyclin E is the regulatory subunit of the cdc2-related protein kinase cdk2 and is a rate limiting factor for the entry into S phase. To date, cyclin E is the only cyclin for which alternative splicing has been described. We report here the isolation of a new splice variant of cyclin E, termed cyclin ET, which has an internal deletion of 45 amino acids compared with the full-length cyclin E protein. Even though cyclin ET contains an intact cyclin box, it is unable to complement a triple cln mutant strain of Saccharomyces cerevisiae or to interfere with rescue by cyclin E, indicating that an intact cyclin box is functionally insufficient. The expression pattern of cyclin ET during cell cycle entry, progression and differentiation differs from that of cyclin E. Thus, ET expression precedes that of the other isoforms during the G0 -> S progression; it shows a sharp peak in early G1 in cells released from a mitotic block and is strongly down-regulated in terminally differentiated myeloid cells. These observations point to different functions for cyclin ET and E and show for the first time that the alternative splicing of cyclin E is a regulated mechanism governed by the cell cycle and differentiation.
Progression through the G1 phase of the cell cycle and entry into S phase in mammalian cells are controlled by a number of proteins, including the cdc2-related kinases cdk2, cdk4 and cdk6, their regulatory subunits, the D and E type cyclins, different inhibitors of cdks such as p15/INK4B, p16/INK4/MTS, p21/WAF1/CDI2/CIP1/SDI1/PIC1 and p27/KIP, the retinoblastoma suppressor protein pRB and its kins p107 and p130 and transcription factors of the E2F/DRTF family (for recent reviews see 1 -7 ). To date, four cyclins (D1, D2, D3 and E) have been described whose functions seem to be associated with, and required for, G1 progression (8 -11 ). The D-type cyclins have been shown to associate with different proteins, such as cdk2, cdk4, cdk6, pRB, the DNA polymerase [delta] subunit PCNA and other gene products (12 -18 ; for reviews see 1 ,3 ,4 ,6 ,7 ,19 ). Complexes between D-type cyclins and pRB have been detected in immunoprecipitates and cdk4-mediated pRB phosphorylation has been observed both in vitro, in insect cells over-expressing cyclin D, cdk4 and pRB and in cyclin D complexes isolated from mammalian cells (14 ,15 ). Cyclin E, on the other hand, appears to be less promiscuous. It is well established now that cyclin E forms complexes specifically with cdk2 and activates its serine-threonine kinase activity shortly prior to entry into S phase (20 ,21 ). This observation suggests that the cyclin E-cdk2 complexes might be directly involved in regulating the G1/S transition. This is supported by the finding that microinjection of cdk2- or cyclin E-specific antibodies results in a G1 block (22 ,23 ). In addition, cells transformed by a cyclin E-expressing retrovirus show a decreased cell size, a decreased requirement for growth factors and a shorter G1 phase of the cell cycle (24 ). In addition, the inducible expression of cyclin E from a tetracycline-regulatable vector has been shown to lead to a shortening of the G1 phase (25 ,26 ).
In contrast to cyclin D, which represents a family of related genes (11 ), only a single gene for cyclin E has been detected (23 ,27 ). On the other hand, cyclin E is the only cyclin gene for which alternative splicing leading to a structurally different protein has been described. We have previously reported the existence of a 43 kDa splice variant of human cyclin E, termed cyclin ES, which lacks 49 amino acids within the cyclin box compared with the known full-length 48 kDa cyclin E. Cyclin ES is expressed at ~10% of the level of full-length cyclin E in all cell lines analysed. The two cyclin E forms functionally differ in that cyclin E, but not cyclin ES, is able to complex with cdk2, to activate the histone H1, pRb and p107 in vitro kinase activity of cdk2 and to rescue a triple cln mutation in Saccharomyces cerevisiae. These findings also indicate that the cyclin box in cyclin E is required for the interaction with cdk2 and the other functions described above, but the biological function of cyclin ES remains obscure. In the present study we describe the identification and analysis of a new splice variant of cyclin E RNA, termed cyclin ET, which has an internal deletion of 45 amino acids relative to the full-length cyclin E protein. Although this deletion leaves the cyclin box intact, cyclin ET is unable to function as a G1 cyclin in yeast, pointing to a crucial functional role for sequences ouside the cyclin box. We also show that cyclin ET expression during the cell cycle precedes up-regulation of the other isoforms and dramatically decreases in terminally differentiated myeloid cells. These observations suggest that cyclin ET and E serve different functions. In addition, this is the first report of an alternative RNA processing event that is controlled by cell cycle progression and differentiation.
WI-38 cells (28 ), obtained from the American Type Culture Collection (ATCC), and HaCat cells (29 ) were cultured in Dulbecco-Vogt modified Eagle's minimum essential medium (DMEM) supplemented with 10% fetal calf serum (FCS). WI-38 cells were seeded at a density of 104 cells/cm2 24 h prior to starvation in serum-free medium for 72 h and stimulated with 10% FCS for the indicated times. In some experiments transcription was inhibited by the addition of actinomycin D (5 [mu]g/ml; Sigma) to the medium. HaCat cells were synchronized in mitosis by a combined hydroxyurea/nocodazole block. After 14 h incubation with 0.1 M hydroxyurea (Sigma) cells were cultured for 9 h in fresh medium, followed by another 5 h incubation in the presence of 30 ng/ml nocodazole (Sigma). The mitotic cells were subsequently detached from the plate by gentle shaking, washed twice with phosphate-buffered saline (PBS) and replated into fresh medium. HL-60 cells were grown in RPMI 1640 supplemented with 10% FCS. Differentiation of HL-60 cells by 10-7 M 12-O-tetradecanoylphorbol-13-acetate (TPA; Sigma) or 10-6 M retinoic acid (RA; Sigma) was induced as described (30 ,31 ), except that the TPA was removed after 30 min. Native granulocytes and monocytes were isolated from peripheral blood samples by Ficoll density gradient centrifugation as previously described (32 ).
For FACS analysis, cells were washed once with PBS and fixed in ice-cold 75% ethanol overnight at 4oC. After washing once with PBS, the fixed cells were stained for 15 min in Hoechst 33258 staining buffer (100 mM Tris, pH 7.4, 154 mM NaCl, 1 mM CaCl2, 0.5 mM MgCl2, 0.1% v/v Nonidet P-40, 0.2% w/v bovine serum albumin, 2 [mu]g/ml Hoechst 33258). Stained cells were analysed in a FACS-STAR Plus (Becton-Dickinson) using UV laser excitation at 325 nm. The fluorescence was amplified linearly. Cell cycle distributions were calculated using the LYSIS II program (Becton Dickinson).
RNA isolation, reverse transcription into cDNA and PCR analysis were performed as described (33 ). The RT-PCR established in our laboratory (33 ) yields results that are basically identical to the data obtained by Northern blotting, as previously shown by a direct comparison (34 ). The experimental strategy included the following precautions: (i) the number of PCR cycles was kept low in order to obtain a linear relationship between the levels of specific input RNA and PCR product, which was achieved by the incorporation of radioactive precursor nucleotides and evaluation by [beta]-radiation scanning using a Molecular Dynamics PhosphorImager; (ii) all results were standardized using the signal obtained with GAPDH or [beta]-actin, whose expression is independent of cell proliferation or differentiation; (iii) all experiments were performed with at least two independent RNA and cDNA preparations and yielded results that showed <20% deviation after standardization; (iv) to control for DNA contamination, primers were designed that spanned at least one exon-intron boundary. The following cyclin E primer pairs were used for PCR amplification (8 ): 5'-Primer, 5'-GC
The open reading frames of the three cyclin E isoforms were cloned into the eukaryotic expression vector pBK-RSV (Stratagene) and N-terminally fused to a three times repeated nine amino acid 12CA5 hemagglutinin (HA) epitope (35 ). Expression of the tagged cyclin E proteins was controlled by in vitro translation using the Promega TNT T3 reticulocyte lysate system following the instructions of the manufacturer. Detection of in vitro translated HA-cyclin E proteins was monitored by immunoprecipitation using a monoclonal [alpha]-HA antibody (Babco, Berkeley, CA). Four microlitres of in vitro translated HA-cyclin E proteins were incubated with 4 [mu]l affinity purified [alpha]-HA antibody for 30 min at 4oC followed by incubation for 20 min with 40 [mu]l protein G-Sepharose 4 Fast Flow (Pharmacia). Immune complexes were centrifuged and pellets were washed three times in IPB buffer (50 mM Tris-HCl, pH 7.6, 2 mM NaCl, 0.5% w/v NP-40, 0.02% aprotinin, 2 mM PMSF, 1 mM DTT), resuspended and boiled in SDS sample buffer.
The yeast strain DL1 (MAT[alpha], ade1, his2, leu2-3, 112, trp1-1a, ura3, cln1::TRP1, cln2, cln3, leu2::GAL1-CLN2) was kindly provided by Steve Reed (Scribbs Research Institute, La Jolla, CA). Yeast cells were grown on either YPGR (1% yeast extract, 2% Bactopeptone, 3% galactose, 3% raffinose) or SGR medium (36 ) with or without 1 mM methionine. For repression of the GAL1 promoter 2% glucose was added. All S.cerevisiae expression vectors used have been described (37 ,38 ) and transformations were carried out according to Gietz et al. (39 ).
We have recently described an alternatively spliced variant of human cyclin E, which we termed cyclin ES (27 ). A detailed analysis of human cyclin ES expression by RT-PCR with specific primer pairs revealed the existence of an additional cyclin E isoform in WI-38 lung fibroblasts that is only slightly larger than cyclin ES (Fig. 1 ). Expression of this new isoform of human cyclin E, which we have called cyclin ET (for third isoform of cyclin E), was also detected in a variety of other human cell lines, such as promyelocytic HL-60 cells, HaCaT keratinocytes and the cervical carcinoma HeLa cell line, and is therefore obviously not restricted to a certain cell type. In all these cell lines the expression of cyclin ET was similar to that of cyclin ES, although there may be subtle cell type-specific differences (Fig. 1 ).
In order to obtain more information about the nature of cyclin ET, the corresponding cDNA was cloned via cyclin E-specific PCR primers. Subsequent sequence analysis showed that the difference in size between cyclin ET and cyclin ES is only 12 bp. Despite their nearly identical size, however, both encoded proteins clearly differ in their molecular structure (Fig. 2 a). Whereas cyclin ES has an internal deletion of 49 amino acids within the cyclin box (27 ), cyclin ET contains an intact cyclin box but has undergone an in-frame deletion of 45 amino acids immediately C-terminal of the cyclin box (amino acids 221-265 of the full-length cyclin E protein, subsequently referred to as cyclin EL, for long isoform of cyclin E). Apart from this deletion of 135 bp the nucleotide sequence of the cyclin ET cDNA is identical to that of cyclin EL (data not shown; 8 ,9 ). Cyclin E is a single copy gene in humans (23 ,27 ) and the nucleotide sequences around positions 717 and 853 (8 ) show significant homologies to splice donor and acceptor consensus sites:
consensus:5'-AG/GT-----AG/G-3'
cyclin ET: 5'-716AG/GC-----AG/C853-3'.
These results therefore strongly suggest that cyclin ET is generated by alternative splicing of the primary cyclin E transcript, as is cyclin ES.
Cyclin E was originally discovered by virtue of its ability to complement a G1 cyclin deficiency in S.cerevisiae (8 ,9 ). In order to analyse the functional properties of cyclin ET, its cDNA as well as that of cyclin EL and ES were cloned into the yeast expression vector pRS426Met25 (37 ) and subsequently transformed into the S.cerevisiae strain DL1 (kindly provided by Dr S.Reed, Scribbs Research Institute, La Jolla, CA). DL1 yeast cells lack the major G1 cyclins CLN1, -2 and -3, but harbour an exogenous CLN2 gene under the control of the inducible GAL1 promoter. As a consequence these cells can only grow in the presence of galactose, but not glucose. Accordingly, as can be seen in the upper panel of Figure 3 , all transformants formed colonies on galactose-containing agar plates. In the presence of glucose, however, only cyclin EL-expressing DL1 cells could grow, whereas cyclin ET and cyclin ES were unable to rescue the triple cln deficiency. It has been shown that strong over-expression of cyclin EL is toxic in S.cerevisiae (27 ). Therefore our experiments were performed in the presence of 1 mM methionine. Under those conditions the Met25 promoter is repressed and gives rise to low levels of expression (37 ). However, to exclude the possibility that the failure of cyclin ET and ES to rescue triple cln mutant yeast cells was due to an insufficient expression level, the same experiment was carried out in the absence of methionine, which led to high levels of expression for all three cyclin E variants (EL, ES and ET), as determined by Westen blot analysis (27 ; data not shown). However, even this strong over-expression of cyclin ET or cyclin ES was insufficient to enable the DL1 cells to proliferate in the presence of glucose (Fig. 3 , lower panel). Therefore, it can be concluded that cyclin ET is unable to complement a G1 cyclin deficiency in S.cerevisiae. Furthermore, these results clearly show that the presence of an intact cyclin box is insufficient for cyclin E to function as a G1 cyclin in S.cerevisiae. This observation suggests that cyclin ET might be unable to interact with cdk partners. This notion was indeed confirmed by a yeast two-hybrid experiment. While a lex A-cyclin EL hybrid protein showed efficient binding to a VP16-cdk2 fusion protein, there was no evidence for any interaction of the corresponding lex A-cyclin ET hybrid and the VP16-cdk2 protein in the same assay (data not shown).
Since the cyclin box is the only domain shared by all known cyclin proteins, this domain has been implicated in the binding of cyclins to their catalytic cdk partners. Indeed, the cyclin box of cyclin A is not only required for binding of cdc2 and cdk2 (40 ,41 ), but even seems to be sufficient (41 ). As cyclin ET is unable to function as a G1 cyclin in S.cerevisiae but contains an intact cyclin box, it might be able to interfere with the function of cyclin EL in a trans-dominant negative manner. To test this hypothesis, cyclin ET was co-expressed with cyclin EL in DL1 yeast cells (Fig. 4 ). To ensure strong over-expression of cyclin ET relative to cyclin EL in these experiments, cyclin ET was expressed from the GPD promoter (pRS415GPD) whereas cyclin EL expression was under the control of the ADH promoter (pRS416ADH), whose transcriptional activity is ~70-fold weaker than that of the GPD promoter (38 ). Despite this over-expression, cyclin ET (or cyclin ES) was unable to interfere with the ability of cyclin EL to complement the G1 cyclin deficiency of the DL1 yeast cells. In summary, these results show that cyclin ET is neither able to act as a G1 cyclin in S.cerevisiae nor to antagonize the rescue of triple cln yeast mutants by cyclin EL.
The expression pattern of cyclin ES during serum stimulation of WI-38 fibroblasts and differentiation of promyelocytic HL-60 cells is identical to that of cyclin EL (27 ,34 ), indicating that at least under those conditions the alternative splicing of cyclin ES is not regulated. In order to investigate whether the expression of cyclin ET may follow a different pattern, we studied the expression of cyclin E during the cell cycle. In a first set of experiments, cyclin ET expression was analysed during the G0 -> S progression of serum-stimulated WI-38 fibroblasts (Fig. 5 ). In agreement with previous studies (9 ,27 ), the expression of all three cyclin E mRNA isoforms was low in G0 (Fig. 5 a), was induced after 9 h, reached peak levels at 16 h, i.e. around entry into S phase (Fig. 5 b), and declined as the cells proceeded through G2 and M. Although the overall expression pattern of all three cyclin E isoforms was similar, induction of cyclin ET was already detectable 2 h after serum stimulation. In order to verify and extend this observation, the 0-2 h time interval was analysed in more detail. As can be seen in Figure 6 , an increase in the level of cyclin ET RNA was detectable as early as 20 min post-stimulation. After 2 h a 4-fold induction relative to cyclin EL was observed. In contrast, the expression of cyclin ES relative to cyclin EL remained constant during this period. The induction of cyclin ET was inhibited in the presence of actinomycin D (Fig. 6 c), indicating that this effect was dependent on de novo synthesis of ET mRNA and not simply due to increased stability of the cyclin ET transcript. These results show that the alternative splicing of human cyclin E RNA is not a constitutive process, but instead is regulated during G0 -> S progression.
To address the question of whether the alternative splicing of cyclin ET is also regulated during normal cell cycle progression, HaCat cells were synchronized at G2/M by successive hydroxyurea and nocodazole blocks. After release from the cell cycle arrest, cyclin E expression was monitored by RT-PCR. As shown in Figure 7 b, the synchronized cells underwent mitosis within 2 h after release from the G2/M block. After 12 h most cells were in S phase and entered the next cell cycle between 20 and 24 h with a somewhat decreased synchrony. As expected, expression of cyclin EL was low in the G2/M-arrested HaCat cells (Fig. 7 a), reached peak levels around the G1 -> S-transition (4-fold induction after 6 h), declined as cells entered the G2 and M phases and was induced again in the subsequent G1 phase. Whereas the expression of cyclin ES paralleled that of cyclin EL, cyclin ET showed a clearly different pattern. Two hours after release from the G2/M arrest, i.e. in early G1, cyclin ET expression was already maximally induced (~6-fold), remained constant at a lower level during G1 and S, decreased further as the cells proceeded through G2 and M and increased again in the subsequent early G1 phase. These results clearly show that the pattern of cyclin ET expression differs from those of cyclin EL and ES during both the G0 -> S and G1 -> S progressions.
In view of the unique expression pattern of cyclin ET during cell cycle entry and progression, we finally asked the question how is cyclin ET expression regulated in cells exiting from the cell cycle during terminal differentiation? As a model system we used the promyelocytic HL-60 cell line, which can be induced by either RA or TPA to differentiate into granulocyte-like or macrophage-like cells respectively (30 ,31 ). Unexpectedly, in both differentiated cell types cyclin ET expression was dramatically reduced (Fig. 8 ). To exclude an in vitro artefact, fresh granulocytes and monocytes from peripheral blood samples were also analysed. As shown in Figure 8 , this analysis yielded a very similar result, i.e. a relatively strong repression of cyclin ET. To investigate this finding in greater detail, cyclin ET expression was monitored at various time points after induction of differentiation by either RA (Fig. 9 ) or TPA (Fig. 10 ). Regardless of the inducing agent, cyclin EL expression remained relatively constant during the early time points and declined in the late stages of differentiation. In contrast, cyclin ET expression was initially induced, raising the possibility that progression through a certain cell cycle stage in G1, where cyclin ET expression is induced, might be necessary to enter terminal differentiation. As shown in Figure 8 , cyclin ET expression strongly decreased in the late stages of differentiation. A quantitative evaluation of the data showed, however, that cyclin ET was specifically repressed in the macrophage-like cells (Fig. 10 c), whilst the decrease in cyclin ET expression paralleled that of cyclin EL during differentiation of HL-60 cells into granulocyte-like cells (Fig. 9 c). These results indicate that expression of cyclin ET during terminal differentiation is regulated in a lineage-specific fashion. Interestingly, cyclin ES was clearly induced in the late stages of both HL-60 cell differentiation pathways, showing that expression and/or splicing of cyclin ES also occurs in a specific and regulated manner, at least in certain instances. This observation is in apparent contradiction of the results of a previous study (34 ), which showed that expression of cyclin ES during terminal differentiation of HL-60 cells paralleled that of cyclin EL. However, because of the very similar sizes of cyclin ES and ET, the expression levels measured for cyclin ES presumably reflect the combined expression of both isoforms, thereby blurring the differential regulation. In summary, our results show that expression of cyclin ET, and to some extent of cyclin ES, is specifically regulated during terminal differentiation in vitro and in vivo.
In this study we have identified a new alternatively spliced isoform of cyclin E, cyclin ET. Expression of cyclin ET clearly differs from that of cyclin EL, being induced early during cell cycle entry and progression and being strongly repressed during terminal differentiation of myeloid cells. Cyclin ET is unable to function as a G1 cyclin in yeast, although it contains an intact cyclin box. These results show that the cyclin box, although required, is insufficient for the function of cyclin E. At present it is, however, only possible to speculate about the biological function of cyclin ET. Cyclin E has been detected in several complexes with various other proteins, like p107, E2F, p21 and p27 (for reviews see 2 ,3 ,5 -7 ) and cyclin ET may be able to bind to one or more of these proteins, thereby interfering with the formation or normal function of these complexes. Furthermore, the unique expression pattern of cyclin ET during cell cycle entry and progression strongly suggests a specific function for cyclin ET in these processes. In addition, the cell type-specific expression pattern of cyclin ET and cyclin ES during myeloid differentiation might indicate a role for cyclin E splice variants in terminal differentiation, similar to observations made with D-type cyclins (42 ). Apart from cyclin ET, two further spliced isoforms of human cyclin E have been described (23 ,27 ). The present study, however, shows for the first time that alternative splicing of human cyclin E is not a constitutive process but is regulated during cell cycle progression and differentiation. Therefore, the alternative splicing of cell cycle regulatory molecules may represent an additional mechanism of cell cycle control in higher eukaryotes. The fact that the existence of an alternatively spliced variant has also been described recently for the cyclin E catalytic partner cdk2 (43 ) strongly supports this hypothesis.
We are grateful to E.Nalbatow for excellent technical assistance and to Dr M.Krause and S.Bischofsberger for synthesis of oligonucleotides. This work was supported by the Dr Mildred Scheel Stiftung für Krebsforschung and the DFG (Mu601/8-1). D.M. was the recipient of a fellowship from the Boehringer Ingelheim Fonds.
REFERENCES