Independent Regulation of Human D-type Cyclin Gene Expression during GI Phase in Primary Human T Lymphocytes*

Cyclins and cyclin-dependent kinases are critically involved in controlling cell cycle progression in virtually all cells. The recent identification of candidate G1 cyclins in mammalian cells has been a major ad- vance in this field, but the exact functions of these cyclins are unknown. The expression of three D-type cyclins (Dl, D2, and D3) was investigated in primary human T lymphocytes as these cells were induced to leave Go, traverse G1, and enter S phase by T cell- specific mitogens. Go phase T cells expressed low levels of cyclin D2, but not cyclin D3. Treatment of these cells with phytohemagglutinin and 12-0-tetradeca-noylphorbol-13-acetate in the presence of fetal calf serum resulted in rapid induction of cyclin D2 RNA in early GI and slower induction of cyclin D3 in late GI. Cyclin D l was not detected in T cells under any con-dition tested. Treatment of T cells with hydroxyurea to arrest cells at Gl/S did not block induction of either D2 or D3. However, arrest of cells in mid GI with deferoxamine blocked D3 expression without affecting D2. Cyclosporin A blocked the induction of both cyclin D2 and D3. Polyclonal antisera were prepared in rabbits against both cyclin D2 and cyclin D3 glutathione S-transferase fusion proteins and used to examine cyclin in T cells. to to that were to These results indicate that striking differences exist in the induction and regulation of two candidate cyclins in human T cells and suggest that these cyclins could participate in multiple cell cycle checkpoints during Go, G1, or S phase.

Cyclins and cyclin-dependent kinases are critically involved in controlling cell cycle progression in virtually all cells. The recent identification of candidate G1 cyclins in mammalian cells has been a major advance in this field, but the exact functions of these cyclins are unknown. The expression of three D-type cyclins (Dl, D2, and D3) was investigated in primary human T lymphocytes as these cells were induced to leave Go, traverse G1, and enter S phase by T cellspecific mitogens. Go phase T cells expressed low levels of cyclin D2, but not cyclin D3. Treatment of these cells with phytohemagglutinin and 12-0-tetradecanoylphorbol-13-acetate in the presence of fetal calf serum resulted in rapid induction of cyclin D2 RNA in early GI and slower induction of cyclin D3 in late GI. Cyclin D l was not detected in T cells under any condition tested. Treatment of T cells with hydroxyurea to arrest cells at Gl/S did not block induction of either D2 or D3. However, arrest of cells in mid GI with deferoxamine blocked D3 expression without affecting D2. Cyclosporin A blocked the induction of both cyclin D2 and D3. Polyclonal antisera were prepared in rabbits against both cyclin D2 and cyclin D3 glutathione S-transferase fusion proteins and used to examine cyclin D2 and D3 proteins in [36S]methionine-labeled T cells. Protein levels were found to correlate closely with RNA levels for both cyclins. No detectable histone H1 kinase activity could be precipitated with either cyclin. However, several cellular proteins were observed to coprecipitate with the cyclins, including several proteins that were observed to associate only with D3. These results indicate that striking differences exist in the induction and regulation of two candidate G1 cyclins in human T cells and suggest that these cyclins could participate in multiple cell cycle checkpoints during Go, G1, or S phase.
Passage through the cell cycle is a highly regulated process involving ordered expression of a series of cell cycle control genes. Growth factors or other external events that affect the rate of cell proliferation are ultimately likely to act by controlling the expression or function of the products of these genes. In all eukaryotic cells, transition through Gz/M is controlled by a 34-kDa serine/threonine protein kinase originally identified as the product of the CDC28 gene of Saccha-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ll TO whom correspondence should be addressed: Div. of Tumor Immunology, Dana-Farber Cancer Institute, 44 Binney St., Boston, MA 02115. Tel.: 617-632-3360;Fax: 617-632-4388. romyces cereuisiae (Reed et al., 1985) and the homologous cdc2 gene of Schizosaccharomyces pombe (Simanis and Nurse, 1986). The kinase activity of ~34'~'' oscillates through the cell cycle, peaking abruptly at G2/M . The activity of ~34'~'' kinase is regulated by several events, including phosphorylation on multiple residues by several kinases (Draetta and Beach, 1988), and by association with one of a family of positive regulatory subunits known as cyclins . B-type cyclins associate with and are required for the activation of ~3 4 '~'~ which induces entry into mitosis (Draetta et al., 1989). A-type cyclins also associate with p34'd'2 in Gz and may play a role a t G2/M or earlier in the cell cycle (Pines and Hunter, 1990a).
Considerably less is known about cell cycle control in G1. In S. cereuisiae, ~3 4 "~'~' is required both for transition at G2/ M and also for passage through Start, the G1 restriction point determining commitment to DNA synthesis (Pines and Hunter, 1990b). Genetic screening studies have identified three genes encoding cyclin-like proteins in S. cereuisiae (CLNl, CLN2, and CLN3), the function of which is required to pass Start and which associate with and activate p3PdCz8 at Start (Richardson e t al., 1989). These CLN proteins are thus "GI cyclins" with apparently redundant functions, since expression any one of the three is permissive for cell growth (Wittenberg et al., 1990). p34'd'2 is expressed in higher eukaryotic cells, and human p34'd'2 can replace yeast p34 both at Gz/M and at G1/S . However, the role of cells is unknown, and there is conclusive evidence only for M phase-promoting activity. In primary human T lymphocytes, we have previously shown that expression of the ~3 4 '~'~ gene is not detectable until early S phase (or very late G1 phase), suggesting that p34'd'2 function is not required during GI, although it could be required in S or at Gl/S (Furukawa et al., 1990). ~34'~"'-related kinases, such as p33 cdk2, have been identified in human cells (Tsai et al., 1991), leading to the suggestion that the "Start" function of ~3 4 '~'~ observed in yeast might be fulfilled by a ~34'~'~-related kinase in higher cells.
Recently, however, several potential mammalian GI cyclins have been identified by three different approaches; Xiong et al. (1991), Lew et al. (1991), and Koff et al. (1991 isolated three novel types of cyclin genes (C-, D-, and E-type) by using human cDNA libraries to complement CLN-deficient yeast; Motokura et al. (1991) identified PRADl (D-type cyclin) as a candidate oncogene rearranged in human parathyroid adenomas; and Matsushime et al. (1991) identified CYLI (a Dtype cyclin now termed Dl) as a colony stimulating factor-linducible gene in murine macrophages and two other related D-type cyclin genes (cyclin D2 and cyclin D3) by homology with CYLI. CYLI is the murine equivalent of the human D cyclin isolated by Xiong et al. and Motokura et al. In macro-P34'dc2 if any, in regulating G1 progression of higher eukaryotic 4113 phages, the CYL proteins were found to associate with a CDC2-related protein, although the identity and functions of any associated kinases have not been determined (Matsushime et al., 1991). Human cyclin E was shown to interact with human ~34'~'' and cdk2 to perform Start in yeast (Koff et al., 1991) and to form a complex with cdk2 during the G1 phase of the human cell cycle (Koff et al., 1992). Taken together, these data indicate that cyclins C, D, and E are each candidate G1 cyclins for human cells and could regulate one or more stages of cell cycle progression. In addition to the immortal or transformed cell lines used t o study the mammalian cell cycle, primary T lymphocytes have proven to be a useful model of normal cells. During the process of differentiation, T cells spontaneously arrest in Go and may remain quiescent for long periods of time until exposed to specific antigen or mitogens. Previous studies have indicated that the transitions from Go to GI and from GI to S are accompanied by an orderly sequence of interdependent events (Crabtree, 1989). For example, transition from Go to G1 is marked by transcriptional activation of the IL-2 receptor and IL-2 genes (Crabtree, 1989). Subsequent Gl events and initiation of DNA synthesis are dependent on induction of IL-2 receptor and on a supply of IL-2 from autocrine or external supply (Kumagai et al., 1988). T cells have a prolonged GI phase compared to immortal or neoplastic cells (typically 24-30 h for T cells).
In this study, we have examined the expression and regulation of the human D cyclin genes in T cells and examined the role of IL-2 in their regulation. The results indicate an unexpected degree of diversity, suggesting that the functions of the D cyclins may be distinct and non-redundant in these cells.

MATERIALS AND METHODS
Cell Preparation and Culture-Leukocytes were obtained from normal healthy volunteers by leukopheresis. All tissue samples were obtained after informed consent of donors and under Institutional Review Board-approved protocols. Mononuclear cells were isolated by Ficoll/Hypaque gradient centrifugation (Pharmacia LKB Biotechnology Inc.). After depletion of adherent cells on plastic dishes, T lymphocytes were isolated by erythrocyte rosetting. The erythrocyte rosette positive fraction contained < 5% monocytes or B lymphocytes, as assessed by flow cytometric analysis. T cells were cultured at 2-5 X lo6 cells/ml in RPMI 1640 medium (GIBCO) supplemented with 10% heat-inactivated fetal calf serum (GIBCO), 2% L-glutamine, and 1% penicillin/streptomycin a t 37 "C with 5% COZ, unless otherwise indicated. Endotoxin content of fetal calf serum was 5 5 pg/ml as determined by Limulus amoebocyte assay. The T cells were stimulated by different combinations of the following agents: 1 pg/ml phytohemagglutinin-P (Wellcome), 0.3 ng/ml phorbol 12-myristate 13-acetate (Calbiochem), 0.12 pg/ml calcium ionophore (Calbiochem), and 1 pg/ml anti-CD28 monclonal antibody (gift from K. Sugita). Some cultures also contained 1 pg/ml cyclosporin A (Sandoz Pharmaceuticals), 100 p~ deferoxamine (CIBA-Geigy), 100 mM hydroxyurea (Sigma), or 30 pg/ml cycloheximide (Sigma). [3H]Thymidine incorporation was measured after incubating cultures with 0.2 mCi of [3H]thymidine (2 Ci/mmol, Du Pont-New England Nuclear) for 16 h. DNA histogram analysis was performed after staining of cell nuclei with propidium iodide as previously described (Furukawa et al., 1990).
Centrifugal Elutriation-5 X 10' T lymphocytes were stimulated by 1 pg/ml PHA, 0.3 ng/ml PMA, and 0.12 pg/ml ionomycin for 48 h. The cells were washed and resuspended in 5 ml of elutriation buffer (RPMI 1640,0.5% fetal bovine serum) and loaded into an elutriation rotor at a rotor speed of 2000 rpm. Twelve 100-ml fractions were collected by stepwise increases of the buffer flow rate from 10 to 27 ml/min as previously described (DeCaprio et al., 1992).
RNA Extraction and Northern Blotting-Total cellular RNA was extracted from fractions (2 X lo' cells) using the guanidium thiocyanate method. Samples (10 pg) were subjected to 1% MOPS/formaldehyde-agarose gel electrophoresis and blotted onto nitrocellulose membranes. The blots were hybridized with 32P-labeled cDNA probes. Cyclin Dl, D2, and D3 murine cDNAs containing plasmids have been described (Matsushime et al., 1991)  Generation of Antibodies to Cyclin 0 2 and 0 3 and Immunoprecipitation-Cyclin D2 and D3 were expressed as glutathione S-transferase fusion proteins in Escherichia coli transformed by plasmids pGEX3XN9 or pGEX3XN2, respectively (from Dr. Charles Sherr). The fusion proteins were affinity-purified by glutathione S-transferase-agarose beads from bacterial lysates as described (Kaelin et al., 1991). Rabbits were immunized with a series of five biweekly subcutaneous injections of purified fusion proteins to obtain high titer antisera. A polyclonal antibody to cdk2 was prepared by subcloning a 1.0-kilobase BamHI-EcoRI fragment of cdk2 cDNA (a gift from Dr. Ed Harlow, Massachusetts General Hospital, Boston, MA) into pGEXZT, producing a glutathione S-transferase fusion protein in E. coli as above, and immunizing a BALB/c mouse once weekly for 4

weeks. Cultures of T lymphocytes were labeled in vivo with [35S]
methionine (1200 Ci/mmol; Du Pont-New England Nuclear) for 3 h in methionine-free RPMI 1640 medium (Flow Laboratories), and lysed in buffer containing 20 mM Tris-HC1 (pH 8.0), 137 mM NaC1, 10% glycerol, 1.0% Nonidet P-40 (Sigma), 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.15 unit/ml aprotonin, 10 pg/ml leupeptin, 1 mM sodium orthovanadate, and 100 mM NaF (Sigma). The cell lysates were incubated with 5 pl of preimmune or immune serum and protein A-Sepharose beads (Sigma) for 2 h at 4 "C. The beads were washed 5 times in the same buffer, suspended in sample buffer as described (DeCaprio et al., 1992), heated a t 95 "C for 5 min, and the eluted proteins applied to SDS-polyacrylamide gels and detected by autoradiography. In some experiments, the immune complexes were collected on protein A beads, washed 5 times with lysis buffer, suspended in kinase buffer (50 mM Tris, p H 7.4,lO mM MgClZ, 1 mM dithiothreitol, 10 p~ ATP), and incubated with [Y-~'P]ATP (24 nmol/ p l , 7000 Ci/mmol, Du Pont-New England Nuclear) for 20 min a t 30 "C, in the presence or absence of histone H1 (1 pg) as an exogenous substrate.

RESULTS
The Expression of D Cyclins Is Cell Cycle-dependent in T Cells-Resting T lymphocytes were isolated from the peripheral blood of normal donors, and induced to enter the cell cycle by exposure to a combination of phytohemagglutinin (PHA), phorbol 12-myristate 13-acetate (PMA), and ionomycin, at 1 wg/ml, 0.3 ng/ml, and 0.12 pg/ml, respectively.
Under these conditions, T lymphocytes enter G1 phase in 2-4 h, enter S phase after approximately 18-24 h, and reach G2/ M by 36-48 h (Furukawa et al., 1990). After stimulation, aliquots of cells were harvested at various time points, analyzed for DNA content by flow cytometry, and for cyclin mRNA expression by Northern blot analysis (Fig. 1). Of the six cyclins studied, only cyclin C and cyclin D2 mRNA were detected in resting (Go) cells, although in both cases the level of RNA was low. Cyclin D l was not detected at any time of the cell cycle in T lymphocytes. Cyclin A, cyclin C, cyclin D2, cyclin D3, and cyclin E were each inducible by >IO-fold following treatment with mitogens, but the kinetics of induction varied markedly. In resting T cells, cyclin D2 mRNA was present in a very low amount, whereas cyclin D3 mRNA was undetectable. Cyclin D2 expression increased >5-fold within 2 h after stimulation, reached a plateau after 24 h, and did not change thereafter. Cyclin D3 could first be detected typically at 12-16 h after stimulation, reached maximal levels after approximately 30 h, and remained stable thereafter.

C.
D.  (June rt nl., 1987). ' I ' cells were cultured with PHA, I'MA. and ionornycin in the presence o r ahsence of CsA for 38 h. Cvclin D.7 and cvclin E rnRNAs were Mocked hv the presence of CsA. CsA also hlocked the induction of cyclin D2 and cvclin C mRNA, which is usuallv ohserved hetween 2 and 24 h (Fig. 3). This experiment suggests that although cvclin D2 and cyclin C are detectahle in resting, (ill T cells, further induction and expression of these cyclins in late G I or S phase is at least partly I L-2-dependent.

E.
In an effort to hetter define the complex hlocking effects of CsA on cvclin 112 expression, we examined the effects of adding exogenous 11,-2, and also triggered T cells to produce 11,-2 through a CsA-independent pathway (cross-linking of t h e cell surface antigen CD28) (June r ! nl., 1987(June r ! nl., , 1990). Expression of rvclin 112 m R N A was inhihited I, ? CsA in hoth PHAand l'MA/ionomvcin-stin~~~lated ' I ' cells (data not shown). Addition of' 11,-2 at time zero of the culture restored expression of cyclin I)2 mHNA. However, in the presence of anti-CD28 and I'MA, CsA had no effect on cvclin 112 expression. 'I"Since we found no evidence that thc I) cyclins "cycle" at the RNA level in 7' cells. we examined exprcwion of'cylin D2 and 113 proteins (Fig. 5 , r l and I{ ). I'olyclonal nntihotlics were raised against cyclin-glut;lthione S -I ransferast. fusion proteins and used to immunoprecil)it;lte cycling f r o m "'Slaheled, mitogen-stimulated T cells. A small amount o f cyclin D2 was detected in resting T cells. Both cyclin I)? : t n d I):\ levels increased >IO-folcl kinetically in paralkl tvith 1tX.A levels (Fig. 1). Cyclin D2 levels appeared t o peak :it (;,/S and then decline, suggesting that this cyclin may c y c k through mechanisms involving protein metabolism rat hrr t h;tn t ranscriptional events. Studies of cyclin (' and E proteins have not heen done. We also examined the immune con1plwc.s [or proteins which might coprecipitate with either cyclin I)? o r D3. When compared with preirnmune antisera, hoth 112 and D3 were found t o coprecipitate with sevrral proteins. liot h I) cvclins precipitated with a p21 (Fig. 5.4 ). ('yclin 113 {vas also found to precipitate with several high molecular mass proteins (140-240 kDa) (Fig. 5B).
frnrnunc Cornplrx Kinas(. Assays-In order to determine if either cyclin I12 and D3 coprecipitated with a kinase capable o f phosphorylating histone H1, an immune complex kinase nssay was performed with the anti-D2 and D3 antisera described ahove. Neither cyclin D: ! or 113 were found to coprecipitate with any histone H1 kinase activity (Fig. 6). However, a polyclonal antihody against cdk2 precipitated an histone H1 kinase that was inducible in T cells (Fig. 6).

DISCOSSION
In human and other higher eukaryotic cells, identification o f the genes that control passage through G I is of particular interest because virtually all external events that rekqllate proliferation act primarily or exclusively during (Pardee, 1989). In hematopoietic cells, for example, growth factors such as interleukin-2 (11,-2) or colony stimulating factor-1 are required to induce commitment of responsive cells to enter S phase, but are not again necessary until the next G I phase, a situation analogous to Start in yeast,. Similarly, growth inhihitory cytokines such as transfnrming growth factor /f or inter- A and H has heen well documented in mammalian cells (Pines and Hunter, 1991 ), the kinetics o f expression and reptlation of other cyclin genes has not been extensively investicated. In this study, we have examined expression o f various candidate G I cyclins in primary human T lymphocytes as they were  (Shipp and Heinherz, lW7;('rahtree, 1989). I t is likely that the order of inductinn o f such genes is determined, a t least in part, hy when a particular gene product is first needed within the cycle.
The data presented here indicate that the kinetics ofexpres-

Cyclin D Expression in T Cells
ping. These results also suggest that the functions of each cyclin in Go, G1, and S phase may be a t least partially distinct.
For example, it is unlikely that cyclin D3, cyclin A, and cyclin E play an important role in Go or early G1, since their earliest detectable expression is in late G1 or S. The product of the retinoblastoma susceptibility gene, pRB, is an attractive candidate as a substate for a GI cyclin-associated kinase (De-Caprio et al., 1989;Lyn et al., 1991). We have previously shown that pRb is phosphorylated in at least three stages in primary T cells, with the first phosphorylation occurring in mid GI (about 12 h) (DeCaprio et al., 1992). This phosphorylation occurs before detectable expression of cyclin D3, E, A, and B, and is not blocked by deferoxamine treatment,' which blocks expression of cyclin D3, E, A, and B. However, definitive studies on where in the cell cycle these cyclins are important will require identification of their functions. In order to better define the stage of the cell cycle during which each cyclin first appeared, T cells were blocked at different stages by various agents, allowing one to temporally order the induction of cyclin gene expression with more precision. Hydroxyurea, which arrests cells at Gl/S by reversibly inhibiting DNA synthesis (Adams and Lindsey, 1967), did not block expression of any of the cyclins studied here, including the "last" cyclin to appear, cyclin A. These results are interpreted to mean that these cyclins are induced in S phase or earlier and that completion of DNA synthesis is not required. Since hydroxyurea did not help distinguish among the cyclins, two other agents that arrest T cells earlier in the cell cycle were examined. Deferoxamine, which blocks cells in mid-G1 phase (Lederman et al., 1984;Terada et al., 1991), was found to have markedly differential effects, blocking induction of cyclin D3, cyclin A, and cyclin E, but not cyclin D2 or cyclin C. These results were largely consistent with the kinetic analysis of these cyclins, which suggested that cyclin D2 and cyclin C were induced much earlier than the other cyclins.
The results with hydroxyurea and deferoxamine were extended by studies with cyclosporin A, which induces arrest early in G1 by inhibiting IL-2 transcription (Emmel et al., 1989;June et al., 1989). The induction of each of cyclins was blocked by CsA. However, since cyclin C and D2 are apparently expressed in resting T cells (prior to IL-2 gene expression) there may be both IL-2-dependent and independent regulation of these two genes. Overall, the order of cyclin gene induction in primary T cells is cyclin D2, cyclin C, cyclin D3, cyclin E, cyclin A, and cyclin B. Our kinetic data would suggest that cyclin D3 and cyclin E could play a role at the mammalian equivalent of Start in yeast, or during S phase.
In contrast, cyclin D2 or cyclin C could play roles at either the Go/Gl transition or early in G1.
Surprisingly, when we separated unsynchronized, rapidly proliferating cultures of primary T cells by centrifugal elutriation, none of the cyclins studied showed any dramatic level of cycling at the RNA level. In multiple experiments, cyclin D3, cyclin C, and cyclin E RNA expression tended to be highest in S/GZ/M phases, however, expression was readily detected during all phases of the cell cycle. Although cyclin A and B show considerable variation through the cell cycle a t the level of RNA, protein, or both, at least one other cyclin, CLN 3, does not cycle in yeast (Wittenberg et al., 1990). Cyclin E RNA was the most cyclical in our experiments, consistent with previous observations (Koff et al., 1991). The expression of cyclin D2 and D3 proteins were also studied in T cells. Cyclin D2 appeared to cycle, peaking at approximately G1/S, at least through the first cell cycle after stimulation.
Elutriated samples were not studied, however, and whether F. Ajchenbaum and J. Griffin, unpublished observations. or not cycling is maintained is not known.
Although not studied in detail here, preliminary studies of the induction of cdc2-related kinases in T cells suggest that there are some parallels with induction of the cyclin genes.
p3Pdk2 mRNA was not detectable in resting T cells and was not induced until 24 h after stimulation. We and others have previously shown that ~34'~'' mRNA first appears in T cells at about 24 h and is not blocked by hydroxyurea (Furukawa et al., 1990;Terada et al., 1991). Deferoxamine blocked both cdk2 and cdc2 mRNA ( Fig. 2 and data not shown). If cyclin D2 and cyclin C interact with a cdc2-related kinase in early G1, our results suggest that ~34'~"' or cdk2 are unlikely candidates. Cyclin A can associate with and activate ~34'~'' in many species, including clam (Rosenthal et al., 1980), Drosophila (Lehner and O'Farell, 1989), Xenopus (Draetta et al., 1989), and humans (Draetta and Beach, 1988;Hunter, 1989,1990a). Cyclin A and cyclin E can also associate with p3Pdk2, a ~34"~'~-related kinase (Koff et al., 1991;Tsai et al., 1991). The kinetics of induction of cyclin A, cyclin E, cdk2, and cdc2 are similar, again consistent with the notion that the kinetics of induction may reflect similarities in gene regulation.
Although cyclin C mRNA is present a t or before that of cyclin D2, the further induction of RNA induced by mitogenic stimuli occurs later than cyclin D2 induction, a t about 8-12 h. Thus, cyclin D2 is the first of these cyclin genes to be induced in T cells, several hours before the other cyclin genes respond to mitogens. The induction of cyclin D2 is blocked by cycloheximide treatment (data not shown), suggesting that cyclin D2 is not a classical "early response gene." The timing of induction of cyclin D2, and its partial IL-%independence suggests a potential role for this cyclin at the Go/G1 transition, a point in the cell cycle not previously associated with control by cyclins or cdc2-like kinases. Although cyclin D2 was expressed in the absence of IL-2 expression (in the presence of a cyclosporin A block), its further induction and expression in late Gl appeared to be dependent on a supply of IL-2. In studies not shown, we observed that activation of T cells by PMA alone or by ionomycin alone were not sufficient to promote IL-2 secretion and did not lead to cyclin D2 expression. Simultaneous exposure to both PMA and ionomycin bypasses the requirement for T cell receptor activation, triggers IL-2 expression, and induces cyclin D2 RNA.
Expression of a third D-type cyclin gene, cyclin D l , was not detected in T cells, consistent with the previous studies of Matsushime et al. (1991), who showed that cyclin D l was inducible in murine macrophages by colony stimulating factor-1 but was not present in murine T cells. This heterogeneity of CYL gene expression is unexpected from previous studies with other cyclins or from studies with cell cycle control genes in general, where conservation of function among different lineages is typical (Nurse, 1990). The reasons for the differential usage of D cyclins in T cells and macrophages are unknown. It is not due to differences between species since the human equivalent of cyclin Dl, PRADl, has been shown to be expressed in several human cell types (Matsushime et al., 1991;Withers et al., 1991), and in unpublished studies with normal murine T cells, we again observed induction of cyclin D2 and cyclin D3, but not cyclin Dl. It is possible that this differential usage reflects the diversity of positive and negative growth signals in various cell lineages that are operative in controlling cell cycle progression during G I . However, it will be necessary to identify specific functions of each D cyclin before the real significance of this observation can be evaluated. In preliminary studies with 24 different transformed human hematopoietic cell lines, cyclin D3 was found t o be widely expressed in both myeloid and lymphoid cell lines, whereas cyclin D2 was expressed only in a small fraction, without regard to whether the line was myeloid, T cell or B cell in origin. Since all of the lines were factor-independent, it is possible that the "phenotype" of the D cyclins expressed may be a reflection of the generally unknown mutations in signaling or regulatory pathways that have led to factor independence.
Using polyclonal antisera raised against cyclin D2 and D3, we could not detect histone H1 kinase activity in multiple experiments a t any time point between 0 and 48 h after stimulation, while control experiments demonstrated that anti-cdk2 precipitated an inducible histone H1 kinase in T cells (Fig. 6). This is consistent with older studies, which demonstrated that H1 kinase activity is minimal in Go and GI but is induced rapidly a t G2/M coincident with the activation of ~3 4 '~'~. If a kinase coprecipitates with any of the D cyclins, other substrates such as pRb would be potentially useful for immune complex assays. In fact, Matsushime et al. (1992) have recently identified a novel kinase, cdk4, which can form complexes with D cyclins. These complexes lack histone H1 kinase activity but can phosphorylate pRb in vitro.
We did, however, detect specific coprecipitation of several potentially interesting cellular proteins with both of the cyclins (Fig. 5 ) . It is noteworthy that the proteins which coprecipitate with cyclin D2 differ, in part, from the proteins which coprecipitate with cyclin D3, suggesting that the functions of these two cyclins in G, may be distinct. The identification of these coprecipitating proteins may be useful in elucidating the functions of the D cyclins in T cells.
Taken together, our results indicate that striking differences exist in the induction and regulation of two candidate GI cyclins in human T cells and suggest that these cyclins could participate in multiple cell cycle checkpoints during G1, S phase, and even potentially in Go. Detailed studies of gene regulation, identification of associated kinases, and identification of any other cellular proteins which interact with these cyclins, such as Rb or p107, are likely to be informative as to the exact role each of these cyclins play in regulating the growth of human T cells.