Relief of cyclin A gene transcriptional inhibition during ... - Nature

Oncogene (1997) 14, 2575 ± 2583  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Relief of cyclin A gene transcriptional inhibition during activation of human primary T lymphocytes via CD2 and CD28 adhesion molecules Ariane Plet, Xavier Huet, MicheÁle AlgarteÂ, Jocelyne Rech, Jean Imbert, Alexandre Philips and Jean Marie Blanchard 1 Institut de GeÂneÂtique MoleÂculaire de Montpellier, CNRS, UMR 5535, BP 5051, 1919 route de Mende, 34033 Montpellier cedex 1, France; 2INSERM U119, 27 Boulevard LeõÈ Roure, 13009 Marseille, France

Cyclin A transcription is cell cycle regulated and induced by cell proliferative signals. To understand the mechanisms underlined in this regulation in normal human cells, we have analysed in vivo protein-DNA interactions at the Cyclin A locus in primary T lymphocytes. Stimulation of puri®ed T lymphocytes by a combination of monoclonal antibodies directed at CD2 and CD28 adhesion molecules gives rise to a long lasting proliferation in the absence of accessory cells. Cyclin A was observed after 4 days of costimulation with anti CD2+CD28 whereas stimulation by anti CD2 or anti CD28 alone was not eective. In vivo genomic DMS footprinting revealed upstream of the major transcription initiation sites, the presence of at least three protein binding sites, two of which were constitutively occupied. They bind in vitro respectively ATF-1 and NF-Y proteins. The third site was occupied in quiescent cells or in cells stimulated by anti CD2 or anti CD28 alone. The mitogenic combination of anti CD2+ anti CD28 released the footprint as cells were committed to proliferation. Consistent with theses results, nuclear extracts prepared from quiescent cells formed a speci®c complex with this element, whereas extracts prepared from cells treated with anti CD2+ anti CD28 failed to do so after cells entered a proliferative state. Keywords: cyclin A; primary T lymphocytes adhesion

Introduction The mechanisms leading to transcriptional modulation of proliferation associated genes involved in the control of the transition between the G1 and the S phases of the cell division cycle are not well understood in primary cells. Only few informations have been gathered about cell cycle speci®c binding of transcription factors in vivo. We have undertaken a thorough analysis of the in vivo structure of the promoterproximal region of cyclin A gene through the use of high resolution genomic footprinting. It is now currently proposed that cell progression through the dierent phases of the cell division cycle is controlled by a family of cyclin-dependent protein kinases (cdk). Several types of cyclins have been identi®ed in higher eukaryotes and cyclin A is synthesized at the onset of DNA replication as well as during the G2/M transition. By binding to respectively Correspondence: JM Blanchard Received 7 November 1996; revised 14 February 1997; accepted 17 February 1997

the cdk2 and cdc2 (cdk1) kinases, the A-type cyclin has been implicated in the control of both the dependence of mitosis on S-phase completion, as well as that of mitosis (Swenson et al., 1986; Lehner and O'Farrell, 1989, 1990; Pines and Hunter, 1990a; Tsai et al., 1991, Walker and Maller, 1991). Cyclin A is found in cellular activities which promote both replication (Cardoso et al., 1993; D'Urso et al., 1990; Pagano et al., 1992a; Pines and Hunter, 1990b; Resnitzky et al., 1995; Sobczak-Thepot et al., 1993), and transcription (for reviews see: LaThangue, 1994; Nevins, 1992; Weinberg, 1995). In the former case, it has been shown associated with replication complexes and is likely to directly participate in the phosphorylation of some of the replication factors (Dutta & Stillman, 1992; Xiong et al., 1993). In the latter case, its association with the E2F/DRTF1 family of transcription factors has been proposed to play a central role in cell cycle transcriptional control (LaThangue, 1994). A large body of evidence suggests that cyclin-dependent kinases participate in the regulation of the activity of this transcription complex. This is believed to occur either directly, by the phosphorylation of E2F by cyclin A-cdk2 kinase, or indirectly through the phosphorylation of a group of proteins structurally related to RB, the retinoblastoma susceptibility gene product, by cyclins D, E, A associated with cdk4, cdk6 or cdk2 kinases. Consistent with these ideas, cyclin A is found complexed with E2F in late G1 and during S phases, recruiting the catalytic kinase subunit to the DNAbinding complex (Cao et al., 1992; Devoto et al., 1992; Helin and Harlow, 1993; Li et al., 1993; Pagano et al., 1992b; Shirodkar et al., 1992; Xu et al., 1994). In contrast to the expanding wealth of information about the regulation of the expression of molecules involved in cell cycle control in established cell lines, only few studies have been devoted to normal human primary cells. Primary T cell culture oers a good opportunity to study the transition from a quiescent to a proliferative state upon peripheral stimulation. These cells, which are normally in a Go state, can be quantitatively committed to proliferation when receiving an appropriate combination of mitogenic stimuli such as CD28-speci®c monoclonal antibodies (mAbs) in association with either CD2 or CD3 mAbs. Under such circumstances, several of the physiological aspects of T lymphocyte activation are mimicked. CD2+CD28 costimulation induces a strong activation of both IL2Ra gene transcription and mRNA stabilization, which results in the high level of expression of the receptor at the cell surface. This is followed by a wave of cell divisions after a 3 ± 4 days time lag, responsible for a long lasting (3 ± 4 weeks) and monocyte-independent T

Cyclin A expression in human primary T lymphocytes A Plet et al

2576

cell proliferation. This is associated with the prolonged secretion of cytokines, either T-speci®c, such as IL-2 and IFN-g, or normally synthesized by accessory cells, such as IL-1a, TNF-a and CSF-1 (Olive et al., 1994). It had previously been shown that cyclin A gene was transcriptionally activated by mitogenic signals and the major promoter-proximal elements characterized by transient transfection experiments into transformed or immortalized cell lines (Barlat et al., 1993; Henglein et al., 1994; Schulze et al., 1995; Zwicker et al., 1995; Huet et al., 1996). Using dimethylsulphate (DMS) genomic footprinting we show here that, like for the established cell lines examined so far, at least three putative protein binding sites are occupied in vivo. Two of them, which bind in vitro proteins of the CREB/ ATF and NF-Y families, are constitutively occupied in vivo. The third site, proposed to be involved in the down regulation of the promoter in quiescent cells, is occupied in quiescent but not in CD2+CD28 costimulated proliferating cells. Stimulations by either CD2 or CD28 mAbs alone, conditions known to be insucient to give rise to a mitogenic signal, did not result in a relief of the footprint.

Results

b

a

c

d

MW (kD) 96 —

1

2

3

4

5

0 0

4 –

1 +

4 +

14 +

64 —

Cyclin A expression in primary human T lymphocytes When puri®ed T lymphocytes were stimulated with a combination of anti CD2+anti CD28 mAbs, incorporation of radioactive thymidine was observed after a 1 day time lag and increased rapidly to reach a maximum around 8 ± 10 days of costimulation (Figure 1a). This resulted in a more than 10-fold increase in cell number (Figure 1c) which was spread on a 3 ± 4 week period. In contrast, stimulation by anti CD2 or anti CD28 mAb alone did not give rise to a measurable level of thymidine incorporation (Figure 1b), nor to an increase in cell number. Similarly, CD2+CD28 costimulation activated the secretion of IL-2 which peaked around 4 days, as well as the expression of IL2Ra chain of the receptor at the cell surface, which lasted for at least 3 weeks, with more than 80% of the cells scoring positive (data not shown, Costello et al., 1993). Medium alone, single anti CD2 or anti CD28 stimulations, were ineective in inducing a detectable level of secreted IL-2, neither an increase of IL-2Ra chain at the cell surface (data not shown). Cyclin A expression was monitored by a Western blot analysis of nuclear extracts prepared from either quiescent or stimulated T lymphocytes. In a way parallel to that of radioactive thymidine incorporation, the protein began to be detectable after 2 days of CD2+CD28 costimulation, was clearly visible at 4 days, and its accumulation was maximum between 12 and 14 days (Figure 1d). Again, medium alone, anti CD2 or anti CD28 single stimulations were ineective in inducing cyclin A. A motif required for the repression of cyclin A promoter in quiescent ®broblasts is active in a T cell context It has previously been shown that cyclin A expression during the G1/S transition results from a transcriptional activation of its gene (Barlat et al., 1993;

46 — days in culture αCD2 + αCD28

Figure 1 Induction of T lymphocyte proliferation and cyclin A by the costimulation with monoclonal antibodies anti CD2 and anti CD28. (a) DNA synthesis in stimulated T lymphocytes. [3H]thymidine incorporation was measured in cells treated with antibodies for the time indicated as described in Materials and methods. (b) 3H-thymidine incorporation measured after 4 days in absence (0) or in presence of single (CD2, CD28) or double (CD2+28) stimulation. (c) Growth curve of T cells induced by the combination of both antibodies. (d) Immunoblot analysis of protein lysates prepared from cells either newly puri®ed from peripheral blood (lane 1) or cultured in absence (lane 2) or in presence of the proliferative monoclonal antibodies for the time indicated (lanes 3 ± 5)

Henglein et al., 1994; Schulze et al., 1995; Zwicker et al., 1995; Huet et al., 1996). Comparison of both the human and mouse cyclin A promoters has revealed the presence of a stretch of conserved nucleotides within which several putative protein binding sites have been underlined. Several recent reports, based on transient transfection experiments carried out in immortalized or transformed cell lines have stressed the importance of a cyclic AMP responsive element (CRE) in mediating cyclin A gene responses to both cAMP, TGF-b or contact inhibition in human Hs27 (Desdouets et al., 1995), hamster CCL39 ®broblasts (Barlat et al., 1995), as well as in bovine aortic endothelial cells (BAEC) (Yoshizumi et al., 1995). Interestingly, another GC-rich element, named the `cell cycle responsive element' (CCRE) has also been demonstrated to be instrumental in the response of cyclin A promoter to serum stimulation in quiescent ®broblasts (Huet et al., 1996). Mutation of this site rendered the cyclin A promoter constitutive and cell cycle-irresponsive.


In order to con®rm this observation in T lymphocytes, we resorted to the use of a human T cell line (Kit225) which recapitulates some of the properties of normal T lymphocytes, but which are more amenable to transfection experiments. Puri®ed, quiescent primary T lymphocytes turned out to be, at least in our hands, very dicult to transfect, and only strong promoters such as the human cytomegalovirus early promoter gave reproducible results. Kit225 cells are absolutely dependent upon the presence of IL-2 for their proliferation. This is illustrated in Figure 2a where cells grown in serum alone (upper panel) or serum+IL-2 (lower panel) have been sorted according to their DNA content. In the absence of the cytokine almost 80% of the cells were found in the G1 phase of the cell cycle, with only few cells entering S phase (6.4%). IL-2 treatment gave rise to a ®ve- to sixfold increase in cells in the S phase (35.8%) together with a twofold increase in cells in the G2/M phase (28.7%). Kit225 cells were transfected with a luciferase expression vector containing a DNA fragment spanning nucleotides 7177 to +100 relative to the 3' most initiation site (Barlat et al., 1995; Huet et al., 1996). Stimulation of quiescent cells by IL-2 gave rise to a four- to ®vefold stimulation of the wild type vector (pGL2-cycA), a ®gure commensurate with the increase in S phase cells, whereas mutation of the CCRE (pGL2-cycAm) abolished this eect (Figure 2b). Moreover, most of this eect was accounted for by a relief of the inhibition observed in the absence of IL-2.

in quiescent cells or in cells stimulated with anti CD2 or anti CD28 mAb alone (Figure 3a, right panel, last lane). This is illustrated in a more quantitative way in Figure 3b. The average of the signals corresponding to

An in vivo footprint is present on a motif involved in cell cycle control of cyclin A promoter only in quiescent T lymphocytes To analyse in more details the mechanism governing the inducibility of cyclin A gene transcription, we performed an in vivo footprint analysis of the region containing the conserved upstream promoter sequence. Quiescent primary T lymphocytes were either stimulated with anti CD2, anti CD28 or a combination of both mAbs. At each indicated time of treatment, cells were exposed to dimethylsulphate (DMS) and genomic DNA was processed for the detection of hypomethylated guanine residues as described (Garrity and Wold, 1992; Barlat et al., 1995). In cells either quiescent or stimulated for a 24 h period (Figure 3a, left panel) with any combination of mAbs, a clear protection was visible on both the CRE and the GC-rich (CCRE) motifs together with a weaker but signi®cant protection of a NF-Y site (Figure 3a). As a control that lymphocytes have already evolved towards proliferation, changes in proteins binding to the NF-kB site present in the promoter of the early gene coding for IL-2Ra was examined on the same DNA preparations (Figure 4). A clear protection was observed as soon as 6 h post costimulation which corresponded to nuclear translocation of NF-kB as detected by gel shift experiments (data not shown) (Algarte et al., 1995). However, whereas the CRE and the NF-Y sites were protected whatever the time point, or the mAbs combination selected to stimulate T lymphocytes, the CCRE motif became exposed only when cells were costimulated with a mitogenic combination of CD2+CD28 mAbs at times when proliferation began to take place (96 h). The same site remained protected

Figure 2 Presence of a transcriptional negative regulatory element in cyclin A promoter revealed by transfection into an IL-2-dependent T cell line. (a) FACS analysis performed on Kit225 cells grown in the presence of fetal calf serum either alone (top panel) or supplemented with IL-2 (lower panel) as described in Materials and methods. Numbers refer to the proportion of each cell subpopulation in the various phases of the cell cycle. (b) quiescent cells were transfected with the pGL2-CycA (wt) or pGL2-CycAm (mCCRE) containing respectively the wild type and the mutated promoters. Luciferase activity was measured 72 h later, either in the absence of IL-2 (open rectangles), or after 24 h of treatment with IL-2 (closed rectangles). Results, corresponding to two experiments carried out in triplicates, were normalized to b-galactosidase activity expressed from a cotransfected pSV-lacZ vector as described (Barlat et al., 1995; Huet et al., 1996)

2577


CD2+CD28

CD2

unstimulated

in vitro

CD2+CD28

CD28

CD2

in vitro

a

unstimulated

2578

➝ ➝

CRE

➝

*

➝ ➝

NFY

➝ ➝ ➝ ➝

CCRE

*

24 h

96 h

b

Figure 3 Identi®cation of in vivo protein binding sites in proliferating versus resting cells. (a) Primary T cells were cultured in medium alone (unstimulated), stimulated with only one monoclonal antibody (CD2, CD28) or stimulated with the mitogenic combination of both of them (CD2+CD28). Cells were treated with DMS after 1 (left panel) or 4 days (right panel) of treatment, and their DNA was prepared for in vivo footprinting as described. Genomic footprinting was run with the control DMS/piperidintreated naked DNA (in vitro) prepared from T lymphocytes of the same donor. Arrows, protected G residues. Note that there is


a

probes

3➝ 2➝ 1➝

24 h

48 h

b

72 h

stimulating antibody:

96 h

SRF 4 ** ➝ ➝ ➝

NF-κB

2

3

4

5

6

Figure 4 DMS/LM-PCR in vivo footprinting of the NF-kB and the SRF-binding sites in the human IL-2Ra gene. Lane 1: in vitro methylated DNA of peripheral blood T lymphocytes; lane 3 ± 6: in vivo methylated DNA of stimulated T cells harvested after 24, 48, 72 and 96 h (respectively lanes 3 to 6) of anti CD2+CD28 treatment. Arrows, protected G residues; asterisks, hypersensitive sites

WT CREm NFYm CCREm

–

3➝ ➝ 2➝ 1➝

probes

1

–

CD2+CD28, 24 h

CD2 + CD28

unstimulated

in vitro

probes

CD2+CD28, 96 h

Nuclear extracts prepared from quiescent T lymphocytes were reached with a series of probes spanning nucleotides 774 to +34 relative to the 3' most transcription initiation site, and containing individual mutations of respectively, the CRE, the NF-Y and the CCRE. Three major retarded complexes were clearly visible (Figure 5a). Mutation of the CRE abolished the formation of the fast migrating compex 1, and decreased as well as the intensity of the slower migrating complex 3. Mutation of the NF-Y site decreased by a factor of three the intensity of band 2, probably allowing the detection of another band migrating at the same position as band 2. A slight reinforcement of both bands 1 and 3 was also observed. Finally, mutation of the CCRE did not

CD2+CD28, 24 h

Proteins binding in vitro to the CCRE of cyclin A promoter are present only in nuclear extracts prepared from quiescent T lymphocytes

lead to an apparent variation in band intensity. This last result could be interpreted by the masking of the complex formed on the CCRE by that formed with the NF-Y site, the two complexes having identical mobilities. The latter possibility was suggested in some experiments where the resolution was high enough to visualize two closely migrating bands (data not shown). We then used these probes with extracts prepared from T lymphocytes stimulated by various combinations of mAbs. When the wild type probe was used, stimulation with anti CD2+CD28 gave rise to a

CD2+CD28, 96 h

each DMS-sensitive bands of the CCRE and the CRE have been plotted after normalization relative to a set of invariant bands. Only the mitogenic combination of mAbs (CD2+CD28) gave rise to a complete relief of the footprint present on the CCRE, whereas anti CD2 alone was inecient. The invariant footprint on the CRE is provided as an internal control.

WT

NFYm

Figure 5 Protein complexes formed in vitro with large probes covering the region conserved between mouse and human (774, +34). (a) Gel shifts were carried out with nuclear extracts from resting T lymphocytes and either with a wild type probe (WT) or a probe bearing a mutated site at the various indicated positions (CREm, NFYm, CCREm). Only retarded bands are shown. (b) Nuclear extracts were prepared from either unstimulated T cells (7) or cells stimulated with a combination of anti CD2 plus anti CD28 (CD2+CD28) and harvested after 24 h or 4 days. Gel shifts were carried out with either wild type probe (left panel) or probe bearing a mutated NF-Y site (right panel). Gels were dried and quanti®ed using a phosphorimager

more DNA in the lane CD2 for the sample prepared at 96 h (right panel). (b) Quantitative analysis of the sensitivity to DMS of the CRE and the CCRE. The signals corresponding to the various G residues indicated by arrows in A were scanned and their intensities divided by that of the corresponding positions of the in vitro samples. The results corresponding to 96 h of stimulation by the indicated stimuli were normalized to the intensities of the invariant bands indicated by stars, and plotted as averages. In vitro refers to naked DNA methylated in vitro. `Quiescent', `CD2' and `CD2+CD28' refer respectively to DNAs methylated in vivo when cells were either quiescent, stimulated with anti CD2, or stimulated with anti CD2+anti CD28. Closed bars: CCRE; Open bars; CRE

2579


progressive loss of bands 1 and 3, whereas band 2 persisted, suggesting the constitutive binding of some of these factors (Figure 5b). Interestingly, when the probe mutated on the NF-Y site was used, a progressive loss of all three bands was observed with the transient appearance of a fourth type of complex between bands 2 and 3. These observations could tentatively be explained by an anchoring role for NFY, which could stabilize the binding of factors to the other sites nearby. Because of the complexity of the patterns obtained with the large probes, we next switched to oligonucleotides containing individual sites (Figure 6). The NF-Y oligonucleotide showed a prominent, slowly migrating complex, together with a set of fast migrating bands. The former corresponded to a constitutive and speci®c binding of factor(s) whatever the origin of the extract and was therefore used as an internal standard of gel loading. The faster migrating bands were not considered as speci®c as they were equally competed by both a wild type (Figure 6) and a mutated (data not shown) NF-Y oligonucleotides. An antibody directed at the B subunit of NF-Y displaced quantitatively the retarded band into a slower migrating complex. A

similar supershift was introduced by an antibody directed at the A subunit of the same factor (data not shown), thus identifying this sequence as an in vitro high anity site for NF-Y. Similarly, analysis on proteins binding to the CRE probe revealed an invariant binding whether extracts were prepared from quiescent or stimulated cells. We thus show in Figure 6 only results obtained with extracts from quiescent cells. The more prominent speci®c retarded complexes were supershifted as expected by broad speci®city antibodies reacting with all members of the CREB/ATF families. When monoclonal antibodies directed at speci®c members of these families were used, only anti-ATF-1 gave rise to a strong supershift. A speci®c complex was also formed with the CCRE probe and extracts from quiescent as well as for 24 h stimulated cells. However, extracts from cells stimulated for longer times failed to form stable complexes in vitro. As the CCRE core sequence is GC-rich, we tried competitions with unlabelled nucleotides corresponding respectively to SP1 and E2F consensus sites without any success. At the moment, the factor(s) binding to the cyclin A CCRE is still unknown and its characterization will await further work.

+ –

– +

– –

+ –

CREM

– – wt mut –

CREB

– –

ATF–1

antibodies competitor

– –

– –

– –

CCRE CREB/ATF

NFY

CD2+CD28, 96 h

– –

–

CD2+CD28, 24 h

– –

CD28, 24 h

– –

CD28, 24 h

stimulating antibodies – competitor – αYb –

CD2+CD28, 96 h

Discussion CD2+CD28, 24 h

2580

–

–

–

CRE Figure 6 In vitro protein binding to the cyclin A NF-Y, CCRE or CRE oligonucleotides. Nuclear extracts were prepared as indicated in the legend to Figure 5b. For each probe, a competition assay was performed with 100-fold molar excess of unlabeled corresponding oligonucleotide (competitor) added in the reaction mixture. wt and mut refer respectively to the wild type and mutated version of the oligonucleotide. When indicated, extracts were incubated with antibodies directed respectively against the Yb subunit of NF-Y (aYb), the CREB/ATF family, ATF-1, CREB or CREM, 1 h prior to the addition of the probe

Numerous studies conducted in immortalized or transformed cell lines have shown that cyclin A mRNA accumulation is cell cycle regulated by a transcriptional activation of its gene in the late G1 phase of the cell division cycle (Barlat et al., 1992; Henglein et al., 1994; Schulze et al., 1995; Zwicker et al., 1995; Huet et al., 1996). We have therefore undertaken the analysis of sequences and factors involved in the control of the expression of this gene under conditions close to physiological ones. To this aim, we have analysed the in vivo chromatin structure at the cyclin A locus in primary T human lymphocytes. Comparison of the sequences of the human and mouse promoters had revealed the presence of a region of striking homology containing a cAMP responsive element (CRE), a NF-Y binding site, and a GC-rich motif (Zwicker et al., 1995; Schulze et al., 1995; Huet et al., 1996). Our previous work on cyclin A down regulation of TGF-b had pointed to the CRE as an important mediator of the inhibitory eect of the cytokine (Barlat et al., 1993, 1995). However, a promoter, mutated on the CRE, was still able to respond to serum stimulation albeit at a lower eciency. DMS footprinting of proteins binding in vivo to this locus in T lymphocytes stimulated with appropriate combinations of mAbs directed at cell surface molecules, revealed the presence of several protected regions near the major transcription initiation sites. Whereas the CRE and the NF-Y sites were occupied whichever combination of mAbs selected, the occupancy of the GC-rich site was speci®cally observed in quiescent cells or in cells stimulated with nonmitogenic mAbs. Conversely, when lymphocytes were exposed to a mitogenic anti CD2+CD28 costimulation, the footprint present on the CCRE was released at late times of stimulation when cells had entered a proliferative programme. Occupancy of the CCRE was thus observed only when cyclin A gene was


transcriptionally silent. This suggested the possibility of the binding of a transcriptional inhibitor to this region. This prompted us to look for proteins binding to a CCRE probe in vitro in a standard gel shift assay. A speci®c binding was observed with extracts prepared from quiescent cells and not from cells stimulated with anti CD2+CD28 at times where cells were actively growing. As a control, the NF-Y probe showed a constitutive binding. A recent report (Zwicker et al., 1996) evidenced the cell cycle regulation of the B-myb E2F site occupation in vivo in serum-stimulated NIH3T3 cells. Despite the weak similitudes between the human cyclin A CCRE and the murine B-myb E2F sites (respectively GTCGCGGG and CTTGGCGG), both sites present a loss of protection in vivo that coincided with the onset of transcription of the respective genes. Based upon similar patterns of in vitro and in vivo methylated guanine protection, the authors proposed that the protein complex interacting in vivo was indeed E2F. However, no electromobilityshift assays performed in parallel with nuclear extract from cell-cycle synchronized NIH3T3 cells were presented. Our data do not allow us to conclude whether the CCRE repressor is one of the known E2F complexes. Further experiments are required to elucidate whether the factors at play in the two experimental models are either identical or only similar in their consequence of in vivo occupancy. While this work was completed, two reports (Schulze et al., 1995; Zwicker et al., 1995) pointed to the CCRE as an important element in controlling the cell cycle behaviour of human cyclin A promoter, but diverged on the nature of the protein(s) binding in vitro to this sequence. The former group (Schulze et al., 1995) proposed that members of the E2F/DP family together with pockets proteins were binding to the CCRE, whereas the latter group (Zwicker et al., 1995) did not succeed in characterizing proteins interacting with this sequence, and even alluded to the fact that recombinant E2F-1 was not binding in vitro to the CCRE. In a previous work (Huet et al., 1996), we had shown that mutation of the GC-rich sequence led to a 10-fold increase in the reporter activity and almost totally impaired its cell cycle-dependent activation, when assayed by transient transfection, thus con®rming its negative regulatory role. Interestingly, the CCRE sequence was not sucient to endow an enhancer-less promoter with a cell cycle-responsiveness. This result thus suggests a cooperation between the CCRE and other sequences, either promoter-proximal such as the CRE or the NF-Y, or more distal and which remain to be de®ned. The presence of a constitutive in vivo footprint on both the CRE and the NF-Y sites nearby points to their possible involvement in modulating the CCRE activity. The apparent discrepancy between the constitutive occupancy of the CRE and the absence of complex in gel shift performed with the mutated large probe and extracts from mitogenically stimulated cells, again points to a cooperation between the binding sites in this region. They are several possible mechanisms, which are not mutually exclusive. First of all, the structure of the entire region, such as its curvature, could be modi®ed by the binding of the dierent factors. An eect visible only with large probes spanning the entire region. The same factors could also transiently attract an auxiliary protein not

necessarily directly contacting DNA, which could explain the transient appearance of band 4. Finally, we propose that the NF-Y factor could organize the binding of the nearby factors. Such a situation is not unprecedented and previous report have pointed to the possibility that NF-Y might facilitate in vivo recruitment of upstream DNA binding transcription factors (Wright et al., 1994; Coustry et al., 1995). The resolution of this problem will await the puri®cation of the factor binding to the CCRE and the establisment of an in vitro assay to study these multiple interactions. Materials and methods Cell culture and transfection Peripheral lymphocytes were isolated from voluntary, healthy donors and puri®ed as described (Cerdan et al., 1991). Purity of T cell preparations (499%) was checked by immuno¯uorescence labelling with various mAbs speci®c respectively for monocytes, B and NK cells. Contamination by monocytes was also checked by the absence of proliferation upon optimal stimulation with PHA. Cells were maintained in RPMI 1640 supplemented with 10% fetal calf serum (FCS), and activated with the indicated antibodies. Kit225, an interleukin-2 (IL-2)dependent human T cell line (Hori et al., 1987), was maintained in RPMI 1650 supplemented with 10% fetal calf serum, 2 mM L-glutamine and 1 nM recombinant IL-2 (Eurocetus). Cells were growth-arrested by culturing them for 48 h in IL-2-free medium. When needed, cells were sorted according to their DNA content with a FACScan (Becton Dickinson) after labelling with propidium iodide as described (Huet et al., 1996). After a 24 h period of growth in low IL-2 (50 U/ml), 26106 cells were washed three times with PBS, concentration to 107 cells/ml in serum and IL-2-free medium, and transfected by exposition for 5 h to 2 mg DNA and 8 mg DMIERC (GIBCO). Fifteen percent fetal calf serum was then added, and cells grown in the absence of IL-2 for an additional 48 h. Luciferase activity was monitored 24 h later, after addition of IL-2 to a ®nal concentration of 360 U/ml. In vivo genomic footprint analysis 107 cells were concentrated in 1 ml of RPMI containing 10% FCS and 20 mM HEPES (pH 7.3), then exposed for 1 min to 5 ml/ml dimethylsulphate and the reaction quenched with 2% b-mercaptoethanol, followed by washing in cold PBS. After extraction of genomic DNA and Piperidine cleavage of methylated Guanines, methylation interference was monitored using a ligation mediated PCR (LM-PCR) as described (Garrity and Wold, 1992; Barlat et al., 1995). The following oligonucleotides were used: for the cyclin A locus: -GCGGGAGGAGCGTAGAGCCC -AGCCCAGGAGCCGCGAGCTG -CAGGAGCCGCGAGCTGCGCG for the IL-2Ra locus: -TGACTCCTGAGGACGTTA -GGGGAGGACTCAGCTTATGAAGTGC -CAGCTTATGAAGTGCTGGGTGAGACCACT The last oligonucleotides of each set were labelled with T4 polynucleotide kinase and [g32P]ATP, and radioactive, elongated DNA was fractionated on a 6% polyacrylamide sequencing gel.

2581


2582

Gel mobility shifts The large probes covering the region conserved between mouse and human (774, +34), were synthesized by radioactive PCR ampli®cation using as starting material the plasmids containing the various indicated mutations. For each individual site, the following double stranded oligonucleotide corresponding to respectively, the cyclin A CRE, the NF-Y and the CCRE sites were used: CRE: 5'-GGGTCGCCTTGAATGACGTCAAGGCCGCGA NF-Y: 5'-GAGCGCTTTCATTGGTCCATTT CCRE: 5'-GGGATTTCAATAGTCGCGGGCTACTTGAA When indicated, the following mutated oligonucleotides were also used: mCRE: 5'-GGGTCGCCTTCAACGACCTCGAGGCCGCGA mNF-Y: 5'-GAGCGCTTGAATTCGTCCATTT mCCRE: 5'-GGGATTTCAATAGTCGAATTCTACTTGAA The annealed oligonucleotides described above, were either end-labelled with T4 polynucleotide kinase and [g32P]ATP (3000 Ci/mmole, Amersham), or labelled by ®lling in of the recessive ends with Klenow fragment of DNA polymerase I and [a32P]dCTP (3000 Ci/mmole, Amersham). For nuclear extracts, T cells were washed in cold Trisbuered saline then resuspended in 0.4 ml of 10 mM HEPES (pH 7.8), 10 mM KCl, 2 mM MgCl2, 1 mM dithiothreitol, 0.1 mM EDTA, 0.1 mM phenylmethylsulphonyl ¯uoride, 0.3 mg/ml leupeptine, for 15 min on ice. Twenty-®ve ml of 10% Nonidet-P40 was then added, and after a 15 s vigorous agitation, cells were spun for 15 s at 13 000 r.p.m. (Sorval minicentrifuge). Nuclei were resuspended in 50 ml 50 mM HEPES (pH 7.8), 50 mM KCl, 300 mM NaCl, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulphonyl ¯uoride, 10% (v/v) glycerol for 20 min and then centrifuged for 5 min at 13 000 r.p.m. The cleared supernatants were stored at 7808C after estimation of the protein concentration according to Bradford (Bio-Rad). Two mg of nuclear extracts were used in the presence of either 1 mg poly(dI-dC) (CRE and NF-Y) of dA-dT (CCRE) and radioactive probe (0.1 ± 1 ng) in 20 ml of the same buer. Complexes were resolved in a 5% polyacrylamide (acrylamide : bisacrylamide, 29 : 1) gel run in 0.256TBE at 48C. Proliferation assay Puri®ed T lymphocytes were grown in a 96-well microtiter plate (20 000 cells/well) in a ®nal volume of 200 ml RPMI

1640 medium supplemented with 10% FCS, and [3H]thymidine (Amersham, 0.05 mCi/well) added at the indicated times for 18 h. Cells were then collected on Whatman glass ®lters and incorporated radioactivity estimated by liquid scintillation counting. All assays were performed in triplicate. Antibodies The following antibodies were used for T lymphocyte stimulation: mAb 39C1.5 and 6F10.3 (anti CD2), or mAb 248 (anti CD28). CD2 mAb was used puri®ed at a concentration of 10 mg/ml and CD28 mAb as a 1/400 dilution of an ascites ¯uid. All other antibodies were from Santa Cruz Biotechnology. CREB/ATF, mouse monoclonal 25C10G, ATF-1, mouse monoclonal IgA C41-5.1; CREB-1 and CREM-1, rabbit polyclonal IgG, respectively C-21 and X-12. Protein analysis Proteins were obtained by direct lysis of the cells in Laemmli's loading buer containing 6 mM urea and equal amounts were resolved by polyacrylamide gel electrophoresis. After transfer onto a nitrocellulose membrane according to standard procedures, immunodetection was carried out with an anity-puri®ed rabbit antibody directed against a bacterially expressed human cyclin A (Barlat et al., 1993), using the Amersham chemiluminescence detection system. Proteins were quanti®ed prior to electrophoresis using the Bradford assay and then by staining the membrane with Ponceau red.

Acknowledgements We are grateful to C Benoist, D Mathis and R Mantovani for their generous gift of NF-Y antibodies. This work was funded by institutional grants from CNRS, INSERM and UniversiteÂ Montpellier II, as well as by contracts from Association pour la Recherche sur le Cancer, Ligue Nationale Contre le Cancer and RhoÃne-Poulenc Rorer. XH is supported by the Ligue Nationale contre le Cancer and AP by the BIOMED2 programme of the European Community.

References Algarte M, Lecine P, Costello R, Plet A, Olive D and Imbert J. (1995). EMBO J., 14, 5060 ± 5072. Barlat I, Fesquet D, Brechot C, Henglein B, Dupuy DA, Vie A and Blanchard JM. (1993). Cell Growth Dier., 4, 105 ± 113. Barlat I, Henglein B, Plet A, Lamb N, Fernandez A, McKenzie F, Pouyssegur J, Vie A and Blanchard JM. (1995). Oncogene, 11, 1309 ± 1318. Cao L, Faha B, Dembski M, Tsai LH, Harlow E and Dyson N. (1992). Nature, 355, 176 ± 179. Cardoso MC, Leonhadt H and Nadal GB. (1993). Cell, 74, 979 ± 992. Cerdan C, Martin Y, Brailly H, Courcoul M, Flavetta S, Costello R, Mawas C, Birg F and Olive D. (1991). J. Immunol., 146, 560 ± 564. Costello R, Lipcey C, Algarte M, Cerdan C, Baeurle PA, Olive D and Imbert J. (1993). Cell Growth Dier., 4, 329 ± 339.

Coustry F, Maity SN and de Crombrugghe B. (1995). J. Biol. Chem., 270, 468 ± 475. D'Urso G, Marraccino RL, Marshak DR and Roberts JM. (1990). Science, 250, 786 ± 791. Desdouets C, Matesic G, Molina CA, Foulkes NS, Sassone CP, Brechot C and Sobczak TJ. (1995). Mol. Cell. Biol., 15, 3301 ± 3309. Devoto SH, Mudryj M, Pines J, Hunter T and Nevins JR. (1992). Cell, 68, 167 ± 176. Dutta A and Stillman B. (1992). EMBO J., 11, 2189 ± 2199. Garrity PA and Wold BJ. (1992). Proc. Natl. Acad. Sci. USA, 89, 1021 ± 1025. Helin K and Harlow E. (1993). Trends Cell Biol., 3, 43 ± 46. Henglein B, Chenivesse X, Wang J, Eick D and Brechot C. (1994). Proc. Natl. Acad. Sci. USA, 91, 5490 ± 5494. Hori T, Uchiyama T, Tsudo M, Umadome, H, Ohno H, Fukuhara S, Kita K, and Uchino H. (1987). Blood, 70, 1069 ± 1072.


Huet X, Rech J, Plet A, VieÂ A and Blanchard JM. (1996). Mol. Cell. Biol., 10, 3789 ± 3798. LaThangue N. (1994). Trends Biochem. Sci., 19, 108 ± 114. Lehner CF and O'Farrell PH. (1989). Cell, 56, 957 ± 968. Lehner CF and O'Farrell PH. (1990). Cell, 61, 535 ± 547. Li Y, Graham C, Lacy S, Duncan AM and Whyte P. (1993). Genes Dev., 7, 2366 ± 2377. Nevins JR. (1992). Science, 258, 424 ± 429. Olive D, Pages F, Klasen S, Battifora M, Costello R, Nunes J, Truneh A, Ragueneau M, Martin Y, Imbert J, Birg F, Mawas C, Bagnasco M and Cerdan C. (1994). Fundament. Clin. Immunol., 2, 185 ± 197. Pagano M, Draetta G and Jansen DP. (1992a). Science, 255, 1144 ± 1147. Pagano M, Pepperkok R, Verde F, Ansorge W and Draetta G. (1992b). EMBO J., 11, 961 ± 971. Pines J and Hunter T. (1990a). Nature, 346, 760 ± 763. Pines J and Hunter T. (1990b). New Biol., 2, 389 ± 401. Resnitzky D, Hengst L and Reed SI. (1995). Mol. Cell. Biol., 15, 4347 ± 4352. Schulze A, Zerfass K, Spitkovsky D, Middendorp S, Berges J, Helin K, Jansen DP and Henglein B. (1995). Proc. Natl. Acad. Sci. USA, 92, 11264 ± 11268. Shirodkar S, Ewen M, DeCaprio JA, Morgan J, Livingston DM and Chittenden T. (1992). Cell, 68, 157 ± 166.

Sobczak TJ, Harper F, Florentin Y, Zindy F, Brechot C and Puvion E. (1993). Exp. Cell Res., 206, 43 ± 48. Swenson KI, Farrell KM and Ruderman JV. (1986). Cell, 47, 861 ± 870. Tsai LH, Harlow E and Meyerson M. (1991). Nature, 353, 174 ± 177. Walker RA. (1995). Cell, 354, 314 ± 317 Weinberg RA. (1995). Cell, 81, 323 ± 330. Wright KL, Vilen BJ, Itoh-Lindstrom Y, Criscitiello M, Cogswell P, Clarke JB and Ting JPY. (1994). EMBO J., 13, 4042 ± 4053. Xiong Y, Zhang H and Beach D. (1993). Genes Dev., 7, 1572 ± 1583. Xu M, Sheppard KA, Peng CY, Yee AS and Piwnica WH. (1994). Mol. Cell. Biol., 14, 8420 ± 8431. Yoshizumi M, Hsieh CM, Zhou F, Tsai JC, Patterson C, Perrella MA and Lee ME. (1995). Mol. Cell. Biol., 15, 3266 ± 3272. Zwicker J, Liu N, Engeland K, Lucibello FC and Muller R. (1996). Science, 271, 1595 ± 1597. Zwicker J, Lucibello FC, Wolfraim LA, Gross C, Truss M, Engeland K and Muller R. (1995). EMBO J., 14, 4514 ± 4522.

2583