Cytosolic Chaperonin Is Up-regulated during Cell Growth

The chaperonin containing t-complex polypeptide 1 (CCT) is a heterooligomeric molecular chaperone assisting in the folding of actin, tubulin, and other cytosolic proteins. The expression levels of CCT subunits varied among seven mouse cell lines tested but showed a close correlation with growth rate. Both the CCT protein and mRNA levels in the human promyelolytic cell HL60 decreased concomitant with growth arrest during differentiation. More rapid decrease in CCT level occurred when the mouse interleukin (IL)-3-dependent myeloid DA3 cells were starved for IL-3. Readdition of IL-3 caused rapid resumption of CCT synthesis during synchronous growth: the maximum CCT protein and mRNA levels were observed at G1/S transition through early S phase. The turnover rate of CCT was nearly constant regardless of growth. Gel filtration and immunoprecipitation analyses indicated that CCT in vivo is associated with tubulin at early S phase, but not at G0/G1 phase. These results demonstrated that CCT expression is strongly up-regulated during cell growth especially from G1/S transition to early S phase and is primarily controlled at the mRNA level. CCT appears to play important roles for cell growth by assisting in the folding of tubulin and other proteins.

. A fixed arrangement (14) and hierarchy in ATP binding activity (15) of these subunits in the torus-like complex have been proposed.
CCT is known to assist in the folding of actin and tubulin in the presence of ATP in vitro (16 -21) and to bind newly synthesized actin and tubulin in vivo (22)(23)(24). In budding yeast, null mutations of CCT subunit genes are lethal, and temperaturesensitive mutants of individual CCT subunits are affected in assembly of tubulin and/or actin (reviewed in Ref. 25). Several other proteins were also reported as possible substrates of CCT. In vitro, ␣-transducin (20) and firefly luciferase (19,26) can be folded with the assistance of CCT. Unfolded cofilin (27), the 22-kDa peroxisomal membrane protein (28), and cyclin E (29) are also able to bind CCT. In addition, CCT may be involved in vesicular trafficking (30), neurofilament formation during neurite development (31), and a Ras-mediated signal transduction pathway in fungi (32).
Despite the essential roles of CCT in protein folding, little is known about the regulation of CCT expression. Here, we report that the CCT expression is closely correlated with cell growth and is markedly enhanced at early S phase of the cell cycle in mouse and human cultured cells. The CCT up-regulated around early S phase is associated with tubulin in vivo. From these observations, we discuss the role of CCT in cell growth and cell cycle progression.
For synchronization, DA3 cells were arrested at G 0 /G 1 phase by starvation for IL-3 for 16 h in 10% FBS/RPMI 1640, and 7TD1 cells were arrested by starvation for IL-6 for 20 h. Synchronized cell growth was induced by addition of these growth factors.
Differentiation of Human HL60 Cells-Promyelocytic leukemia HL60 cells (obtained from ATCC) were routinely maintained in 10% FBS/RPMI 1640. Differentiation of HL60 cells was induced by addition of 1.4% dimethyl sulfoxide (Me 2 SO), 1 M retinoic acid (RA), 1 mM sodium butyrate, or 100 nM 12-O-tetradecanoylphorbol-13-acetate (TPA). Directions and progression of differentiation were monitored by analyzing morphological changes, expression of cell-surface CD11b antigen, and induction of two different cytosolic esterase activities using ␣-naphthyl acetate and naphthol AS-D chloroacetate as substrates (35). Proportion of cells undergoing apoptotic cell death was analyzed by two-color flow cytometry using the FACSCalibur system (Becton-Dickinson, San Jose, CA) after staining by terminal deoxynucleotidyltrans-ferase-mediated dUTP nick-end labeling (TUNEL) method using an Apoptosis Detection System, Fluorescein kit (Promega, Madison, WI).
[ 3 H]Thymidine Incorporation Assay of Growing Cells-Cultured cells were inoculated at a concentration of 1 ϫ 10 5 cells/ml in appropriate media produced by replacing normal FBS with dialyzed FBS (cutoff size of 2000), and cultured for 8 h. After addition of [ 3 H]thymidine (37 kBq/ml; NEN Life Science Products) to the media, cells were further cultured for 1 h and then lysed in 0.5 N NaOH. After neutralization of the lysate with HCl, proteins were precipitated in 10% trichloroacetic acid. Following three washes in 5% trichloroacetic acid, radioactivity in the precipitate was measured using a liquid scintillation counter.
Analysis of Cell Cycle by DNA Content-Cells were stained with propidium iodide (PI) using a CycleTEST PLUS DNA reagent kit (Becton-Dickinson) and analyzed by the FACSCalibur flow cytometry system.
Western Blotting-Cells were lysed in 40 mM Hepes-NaOH buffer (pH 7.5) containing 1% Triton X-100, 5% glycerol, 5 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride on ice, and the lysate was centrifuged at 12,000 ϫ g for 15 min at 4°C. The supernatant was recovered and protein concentration was determined by the method of Bradford (38), using bovine serum albumin as a standard. The extracted proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) using 7.5 or 10% polyacrylamide gels (39), and blotted onto polyvinylidene difluoride filters according to the method of Towbin et al. (40) using Mini-PROTEAN II equipment (Bio-Rad). After blocking with 5% (w/v) skim milk in phosphate-buffered saline containing 0.05% Tween 20, the filters were incubated with appropriate first antibody and then with alkaline phosphatase-conjugated goat antibody against rabbit, rat, or mouse immunoglobulins (BIOSOURCE International, Camarillo, CA) in phosphate-buffered saline/Tween 20. The specific binding of the antibodies was visualized by developing with tetrazolium bromochloroindolyl phosphate and nitro blue tetrazolium in 50 mM sodium carbonate buffer (pH 9.8) containing 1 mM MgCl 2 . The resulting bands were scanned with a flatbed scanner (JX-330; Sharp, Mahwah, NJ) and analyzed using the public domain NIH Image program (U. S. National Institutes of Health, Bethesda, MD).
Metabolic Labeling of Proteins and Immunoprecipitation-Cells were cultured in methionine (Met)-and cysteine (Cys)-free RPMI 1640 medium containing 10% dialyzed FBS and [ 35 S]Met/[ 35 S]Cys mixture (NEN Life Science Products Inc.), at 1.5 MBq/ml for 15 min, or at 7.4 MBq/ml for 4 h, depending on purpose. The labeled proteins in these cultured cells were chased in medium containing 10 mM nonradiolabeled Met and Cys for appropriate periods and extracted using RIPA buffer (1% Nonidet P-40, 0.1% sodium deoxycholate, 100 mM NaCl, 4 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.2 milliunits/ml aprotinin, 50 mM HEPES-NaOH, pH 7.5). After removing proteins that bind to protein G-Sepharose by absorption (Amersham Pharmacia Biotech), aliquots (200 l) of the supernatant was mixed with 5 g of anti-CCT␣ antibody (clone 22b) and incubated at 4°C for 1 h. Protein G-Sepharose beads were then added to the mixture, incubated at 4°C overnight on a rotator, and washed repeatedly in RIPA buffer. Proteins bound to the resin were extracted by boiling in SDS-PAGE sample buffer and separated by SDS-PAGE using 10% polyacrylamide gels. Radioactivity of bands was counted using a BAS2000 image analysis system (Fuji Film, Tokyo, Japan).
Gel Filtration Chromatography-Cellular proteins were extracted as described above (see "Western Blotting" section). A Superose 6 column (1 ϫ 30 cm) (Amersham Pharmacia Biotech) was equilibrated with a buffer consisting of 40 mM Hepes-NaOH (pH 7.2), 5 mM EDTA, 1 mM dithiothreitol, and 5% glycerol. The extracted proteins were applied to the column at a flow rate of 0.5 ml/min with an FPLC system (Amersham Pharmacia Biotech). Protein concentration was monitored by absorbance at 280 nm and fractions (0.5 ml each) of the eluate were collected. Aliquots of the fractions were analyzed by Western blotting.

Growth-dependent Expression of CCT in Mouse Cultured
Cells-Previously, we found that CCT subunits are highly expressed in mouse tissues containing rapidly propagating cells such as the testis and spleen (13), suggesting that CCT expression may be correlated with cell growth. To study this correlation further, we analyzed CCT expression levels in several lines of mouse cultured cells that exhibit widely different growth rates. Cellular proteins were extracted from these cells, and expression levels of the eight different CCT subunits were determined by Western blotting with specific antibodies to individual subunits (Fig. 1). The mouse cell lines used in this experiment were: FM3A (mammary carcinoma), DA3 (myeloid cells), L5187Y (lymphoma), 7TD1 (B-cell hybridoma), J774A.1 (macrophage-like cells), LLC1 (lung carcinoma), and L929 (fibroblasts). Under the culture conditions used, FM3A, 7TD1, DA3, and L5187Y cells grew in suspension, whereas LLC1, L929, and J774A.1 grew as adherent cells. The CCT expression levels determined by Western blotting (Fig. 1A) varied widely among these cells, and were in the following order from lower to higher: To examine the correlation between CCT levels and growth rates, [ 3 H]thymidine incorporation into DNA was determined with these cells. The radioactivity incorporated was correlated well with the levels of CCT subunit expression in these cells (Fig. 1B), with correlation coefficients of 0.89 to 0.97 (Fig. 1D). We also analyzed the proportion of S phase cells in the cultures by flow cytometry after DNA staining with propidium iodide and found a reasonable correlation between % S phase cells and CCT content (data not shown). Since the folding of tubulin and actin is known to be assisted by CCT (16 -18, 21-24), the amounts of these proteins expressed in these cells were determined. In contrast to CCT, the amounts of three known substrates of CCT, actin, ␣and ␤-tubulin, showed no significant correlations with the growth rates ( Fig. 1, C and D).
Although the CCT expression levels varied among the cell lines, the expression patterns of subunits were very similar among the eight subunit species (Fig. 1, A and B). To examine whether these subunits formed complexes, gel filtration analysis of cell extracts prepared from J774A.1, L929, DA3, and FM3A was carried out. More than 99% of CCT subunits (each approximately 60 kDa) were eluted as symmetrical peaks at about 700 kDa as judging from Western blot analysis of eluates, for any of the four cell lines used (data not shown). These results indicate that most CCT subunits in vivo exist as a large complex regardless of cell growth. CCT complexes in growing cells are likely to be formed with almost identical subunit composition, although it can be changed in special circumstances such as differentiated neurites (42). The constant ratio of subunits is consistent with the fixed arrangement of different CCT subunits in the complex as proposed by Liou et al. (14), and implies that there is a tight control mechanism for adjust-ing the expression levels of different CCT subunits in growing mammalian cells.
Decrease in CCT Expression Level Concomitant with Growth Arrest during Human HL60 Cell Differentiation-Next, we analyzed CCT expression levels in cells undergoing growth arrest during differentiation. We used HL60 (35), a multipotent promyelolytic cell that can differentiate into different types of cells in response to a variety of chemical reagents: Me 2 SO and RA induce granulocytic differentiation, whereas sodium butyrate and TPA induce monocytic and macrophage-like differentiation, respectively. The differentiation induced by these chemicals was confirmed by morphological changes, expression of cell-surface CD11b antigen, and induction of two esterases as reported previously (35). In the course of differentiation induced by Me 2 SO, sodium butyrate, or TPA treatment, cell numbers doubled during the initial 24 h but showed little increase thereafter ( Fig. 2A). Cell proliferation was arrested after 2 to 3 days upon RA treatment. Analysis of [ 3 H]thymidine incorporation also confirmed the decrease of DNA synthesis during differentiation (Fig. 2B). In addition, flow cytometry analysis of cells stained by the TUNEL method revealed that only a small fraction of the cells (less than 10%) were apoptotic during differentiation induced by any of the above chemicals (data not shown).
Under the conditions described above, the levels of two CCT subunits (␣ and ␤) examined by Western blotting decreased during treatment with the chemicals (Fig. 2, C-F). These results are consistent with the idea that CCT expression is closely correlated with cell growth rate. The time course of the decrease in expression level was similar between the two subunits but different among the inducers utilized. The CCT levels decreased to less than 5% by TPA treatment for 4 days, but only to 20 -50% after 5 days with the other chemicals (Fig. 2, E  and F). The rapid decrease in CCT induced by TPA is probably related to a rapid decrease in DNA synthesis (Fig. 2B). It is notable that the decrease in CCT subunit levels had already started within 24 h of treatment with all the inducers used, whereas cell number doubled during this period. These observations suggest that the synthesis of CCT subunits is reduced prior to growth arrest determined by counting cell numbers. The levels of CCT mRNAs encoding ␣ and ␤ subunits also decreased during treatment with all the inducers: the mRNA levels had started to decrease within 12 h of treatment (Fig. 3). Thus, the CCT mRNA levels decrease before the onset of growth arrest of these cells, suggesting that cells rapidly reduce transcription of CCT genes prior to growth arrest. Taken together, these results indicate that CCT expression is tightly coupled with cell growth and is mainly controlled at the mRNA level.
In contrast to the changes in CCT levels, no significant changes were observed in ␣-tubulin level except for cultures treated with TPA (Fig. 2G). Upon TPA treatment, HL60 cells can differentiate to macrophage-like cells without cell division or new DNA synthesis (43), accompanied by phosphorylation of tubulin and microtubule-associated proteins followed by destabilization and depolymerization of microtubule architecture (44). Since TPA treatment also leads to rapid reduction of CCT subunit level, tubulin and CCT appears to be coregulated in these cells. The level of ␣-tubulin mRNA decreased rapidly in response to all of the inducers tested (Fig. 3), similarly to CCT subunit mRNA level, suggesting that transcription of these mRNAs may also be co-regulated. On the other hand, the amount of actin showed little change during differentiation induced by any of these chemicals (Fig. 2H) and the actin mRNA showed little change or slow decrease depending upon the chemicals used (Fig. 3).
Turnover of CCT in Cultured Cells-Since the level of CCT showed marked changes concomitant with cell growth as described above, turnover of CCT may also play important roles in regulating the amounts of CCT in the cell. For this experiment, we employed mouse DA3 cells which requires IL-3 for growth (33). When DA3 cells that had been grown with IL-3 were cultured in the absence of IL-3, the amounts of all CCT subunits decreased rapidly (Fig. 4). After IL-3 starvation for 16 h, the amounts of CCT subunits were reduced to approximately half of the levels in normally growing cells (Fig. 4B) although the rate of decrease is not completely identical between subunits (turnover rates may be slightly different). Even after 32 h of starvation, readdition of IL-3 restored growth (data not shown) concomitant with coordinated synthesis of CCT subunits (Fig. 4A). A similar phenomenon was observed using another growth factor-dependent cell line, 7TD1 (data not shown), which requires IL-6 for growth (34). These results suggested that CCT subunits are rapidly degraded during starvation leading to growth arrest. The amount of ␤-tubulin was reduced significantly during starvation but that of actin remained unaffected (Fig. 4). The labeled CCT complex was analyzed by immunoprecipitation using anti-CCT␣ monoclonal antibody followed by SDS-PAGE (Fig. 5A). Since eight different subunits of approximately 60 kDa are contained in the CCT hexadecamer complex (4), they appear as a cluster of bands at 53-65 kDa by SDS-PAGE (11,16,17,45). The half-life of labeled CCT in DA3 cells was estimated to be 6.0 h in the presence of IL-3 and 6.8 h in the absence of IL-3 (Fig. 5B), indicating that the turnover rate of CCT is little affected by growth conditions. Another cell line, FM3A, which grows faster and shows higher CCT contents (Fig. 1), showed a similar CCT half-life (7.5 h) under normal growth conditions (Fig. 5). These results indicate that CCT undergoes rapid degradation with nearly constant rates regardless of growth conditions and probably also of cell type.
In the course of these experiments, a 70-kDa protein coprecipitated with the CCT complex was observed (indicated by asterisks in Fig. 5). Since HSP70 and/or HSC70 were previously reported to be co-purified with CCT complex (11,27,45), we tested the precipitates by Western blotting using antibodies specific to either HSP70 or HSC70. Indeed, the 70-kDa band was recognized by anti-HSC70 (but not by anti-HSP70) antibody (data not shown). It should be noted that the HSC70 associated with CCT showed a turnover rate similar to CCT. These results probably indicate that HSP70 family proteins (e.g. HSC70 under nonstress conditions) cooperate with CCT in assisting the folding of cytosolic proteins in vivo, as is the case in vitro (19, 26, 46). Cell Cycle-dependent Expression of CCT Subunits-To study the correlation between CCT expression and cell growth in detail, DA3 cells were synchronized and the expression of CCT was analyzed during the cell cycle. Upon IL-3 starvation for 16 h, cell growth was arrested uniformly at G 0 /G 1 phase as monitored by flow cytometry with propidium iodide staining of DNA (0 h in Fig. 6A). These cells showed no significant apoptotic cell death during starvation (data not shown). After readdition of IL-3, DNA contents changed in a well synchronized manner (Fig. 6A) in accordance with the phases of the cell cycle as confirmed by Western blot analysis of cyclins (Fig. 6, D and  G). The cells entered G 1 /S transition state at 9 h, S phase at 12 h, and G 2 /M phases at 15-18 h after readdition of IL-3. The S phase of the second cycle started at approximately 24 h although in a less synchronized manner.

FIG. 2. Changes in levels of CCT and substrate proteins expressed in HL60 cells during granulocytic (induced by Me 2 SO (DMSO) or RA), monocytic (by sodium butyrate), and macrophage-like (by TPA) differentiation in culture.
The levels of CCT subunits in the synchronized cells were determined by Western blotting (Fig. 6, B and E). CCT subunits were increased by severalfold relative to the initial low level (20 -50% of normally growing cells): it started to increase at 6 h after IL-3 addition, and continued to increase until 12 h. After reaching the peak at 12 h, CCT level began to decrease at 12-15 h. This was followed by a second increase which started at 18 -21 h and continued until 24 h. Both the peaks of CCT level observed at 12 and 24 h coincide with early S phase,

FIG. 4. Changes in CCT subunit levels in DA3 cells during growth arrest by IL-3 starvation and recovery of growth by readdition of IL-3. DA3 cells were starved for IL-3 for 32 h, and then cultured in the presence of IL-3 (50 units/ml) for 12 h.
A, at the intervals indicated, cells were collected, and extracted proteins (2.5 g/lane) were analyzed by Western blotting using antibodies against CCT subunits, ␤-tubulin, and actin. B, bands shown in panel A (except after IL-3 readdition) were quantified and plotted against time of starvation. indicating that the abundance of CCT subunits changes during the cell cycle with a maximum level at early S phase. Synthesis rates of CCT subunits during the cell cycle were then analyzed by pulse labeling for 15 min followed by immunoprecipitation with an antibody against CCT␣ (Fig. 7). These results indicated that maximum synthesis occurs at 12 and 21-24 h corresponding to early S phase. The total protein synthesis determined from trichloroacetic acid-precipitable radioactivity started much earlier (with the first peak at 6 h; Fig. 7B) than CCT synthesis, suggesting that the proteins synthesized at the early stage (0 -6 h) are less dependent on CCT-assisted folding and that some of the proteins highly expressed around early S phase require abundant CCT for folding. In agreement with the results of CCT synthesis, the levels of mRNAs encoding CCT subunits determined by Northern blotting (Fig. 8) started to increase within 3 h and reached the first peak at 9 h. This was followed by a decrease at 12-15 h and a second increase at 18 -21 h.
On immunoprecipitation by anti-CCT␣ antibody, pulse-labeled proteins of 70 and 50 kDa were coprecipitated with CCT subunits appearing at 53-65 kDa (Fig. 7A). The 70-kDa protein is HSC70 as described above, whereas the 50-kDa protein is probably tubulin because their molecular masses are similar, tubulin is known to be copurified with CCT (22,37) and association between tubulin and CCT can also be detected using DA3 cells as described in the next section. Since the 50-kDa protein coprecipitated with CCT gave signals comparable with CCT by labeling for 15 min (Fig. 7A) but not by labeling for 4 h (Fig. 5A), association between this protein and CCT is probably transient. This agrees with the previous observation indicating that the half-life of CCT association with newly synthesized tubulin in vivo is 5-10 min (22). It should be noted that the band intensity of the 50-kDa protein changed during cell cycle in the manner nearly identical to that of CCT.
Coordinate Expression and Association of Tubulin with CCT-In the above experiments with DA3 cells, we observed that the cellular levels of ␣and ␤-tubulin decreased during IL-3 starvation and increased by 2-to 3-fold upon readdition of IL-3, although ␣-tubulin was more periodic than ␤-tubulin (Fig.  6, C and F). The level of ␣-tubulin mRNA, which had been greatly reduced during starvation, increased slowly after IL-3 readdition (Fig. 8). On the other hand, the actin level showed little change during starvation but slightly up-regulated (25% increase) upon readdition of IL-3. The small extent of changes may be due to very large amount of actin pool in the cell because the ␤-actin mRNA reduced during starvation, recovered very rapidly within 3 h after IL-3 readdition. These results suggested that the expression of tubulin and/or actin was partly co-regulated with CCT in DA3 cells To further examine the physiological significance of growthand cell cycle-dependent expression of CCT, we analyzed the CCT and substrate proteins in the synchronized DA3 cells by a combination of gel filtration chromatography and immunoprecipitation. Proteins extracted from the synchronized cells were applied onto a Superose 6 column, and the eluted fractions were analyzed by Western blotting using monoclonal antibodies against CCT␣, ␣-tubulin, or actin (Fig. 9). When proteins from the G 0 /G 1 -arrested cells (starved for IL-3 for 16 h) were analyzed, no significant correlations were observed between the elution peaks of CCT and those of actin or tubulin (Fig. 9A). However, at G 1 /S transition state (9 h after IL-3 readdition), a small but significant amount of tubulin was co-eluted with CCT complex around the 670-kDa marker (Fig. 9B). At early S phase (12 h after IL-3 readdition), a much greater portion of tubulin was co-eluted with CCT (Fig. 9C). We also determined the amounts of ␤-tubulin in these fractions and the same results were obtained (data not shown). To confirm the cell cycle-dependent binding of CCT to tubulin, the CCT peak fractions around 670 kDa were pooled and immunoprecipitated with anti-␣-tubulin antibody, and the precipitates were analyzed by Western blotting with anti-CCT antibody (Fig. 9D). A weak band of 60 kDa was observed with the fractions from G 1 /S transition state (lane 2) and the signals became much stronger at early S phase (lane 3), in contrast to the sample from G 0 /G 1 phase that gave no signal (lane 1). These results clearly indicate that the CCT complex in vivo associates with tubulin at early S phase, but not at G 0 /G 1 phase.
Since co-fractionation of CCT with a small amount of actin was observed in fractions at approximately 400 kDa (Fig. 9, A-C), we tested whether CCT bound actin in these fractions by immunoprecipitation with anti-actin antibody. However, no CCT was detected in the precipitates (data not shown) and thus the actin eluted in these fractions is not bound to CCT. These results suggest that actin occupies CCT for a period much shorter than tubulin. In agreement with this view, Sternlicht et al. (22) estimated the half-lives of CCT-tubulin and CCT-actin complexes in Chinese hamster ovary cells to be 5-10 min and less than 2-3 min, respectively, and Thulasiraman et al. (24) estimated the latter to be 50 s.
It is also notable that the change of CCT subunit abundance during cell cycle followed a time course similar to that of cyclin E (specific to G 1 /S transition and S phase; Fig. 6, E and G). Cyclin E is known to be involved in G 1 /S phase transition (47,  (29) reported that CCT plays an essential role in cyclin E maturation in yeast and is associated with cyclin E in human cells (29). Although we analyzed cyclin E using the Superose 6 fractions, little or no co-elution of cyclin E with CCT was observed in the present system (data not shown). CCT may interact with cyclin E transiently as with actin.

Growth-and Cell Cycle-dependent Expression of CCT-We
presented several lines of evidence indicating that the expression levels of cytosolic chaperonin CCT are closely correlated with growth rates of mouse and human cells: faster growing cells express CCT subunits at higher levels. This applies to a wide variety of cells derived from different origins and showing various distinctive characteristics (fibroblasts, carcinoma, lymphoma, myeloma etc; Fig. 1). Such a correlation is likely to apply also during the differentiation of cells, since the levels of CCT subunits and their mRNAs decrease with growth arrest during differentiation of the promyelotic cell line HL60 (Figs. 2  and 3). The same applies to cells under growth arrest by growth factor starvation ( Fig. 4 and similar results from the IL-6-dependent cell line 7TD1). These results indicate that CCT expression is down-regulated at G 0 /G 1 phase of the cell cycle.
These observations are consistent with the following findings reported previously: mouse tissues abundant in growing cells (e.g. testis, spleen, thymus, and bone marrow) and cultured cells express much higher levels of CCT subunits and their mRNAs than those poor in growing cells (e.g. heart, kidney, and lung) (13,49,50). Mouse preimplantation embryos begin to synthesize CCT subunits at an early cleavage stage (51,52). In post-implantation embryos, Tcp-1/Ccta mRNA abundantly expressed at early stages decreases with development (49). Levels of CCT subunits decrease in a subunit-dependent manner during the RA-induced neuronal differentiation of the mouse em-bryonal carcinoma P19 cells (42). In human keratinocyte cell line K14, CCT synthesis is down-regulated by agents affecting cell growth and differentiation such as TPA, interferon ␥, and okadaic acid (36). Even in an evolutionarily distant organism, Dictyostelium discoideum, the Tcp-1/Ccta mRNA level decreases rapidly during development (53).
However, CCT expression unrelated to cell growth appears to occur under special circumstances. When tubulin is rapidly synthesized and assembled, CCT expression is up-regulated even after growth arrest (e.g. during spermatogenesis (54) and neurite development (31) of mammals, and cilia generation of Tetrahymena (55,56)). Apart from such cases, the primary cause for promoting CCT expression seems to be continuous cell growth.
Since degradation rates of CCT are rapid (t1 ⁄2 ϭ 6.0 -7.5 h) and do not seem to vary greatly among different cell types (Fig.  5), CCT subunit levels in vivo can decrease rapidly once their synthesis is stopped (Fig. 4). It is tempting to speculate that such instability represents part of the mechanism for regulating CCT levels. Excess CCT should be removed from the cytosol when cell growth is arrested because they can be deleterious to the cell when substrate proteins are not synthesized. In this connection, CCT expression levels vary by more than 100-fold between the mouse testis and heart (13).
The growth-dependent CCT expression can be explained by preferential CCT synthesis during the period from G 1 /S transition to early S phase. The CCT levels reach a maximum at this time in DA3 cells (Fig. 6), whereas the degradation rate of CCT is almost constant in these cells regardless of growth rate (Fig. 5). Maximum synthesis of CCT subunits occurs at almost the same time as maximum CCT level (Fig. 7), and is controlled at the mRNA level (Fig. 8). It is clear that the CCT expression is up-regulated not only by release from G 0 or quiescent phase because up-regulation of CCT around early S phase also occurs FIG. 9. Association of tubulin with CCT in DA3 cells at early S phase. DA3 cells were starved for IL-3 for 16 h, and IL-3 was added to obtain synchronized cultures. After the starvation for 16 h (G 0 /G 1 phase; panels A and E), or after readdition of IL-3 for 9 h (G 1 /S transition; panels B and F) or 12 h (early S phase; panels C and G), cells were collected and lysed. Proteins extracted from 1 ϫ 10 7 cells were applied onto a Superose 6 column. A-C, the fractions eluted from the column were collected and analyzed by Western blotting with antibodies against CCT␣, ␣-tubulin, or actin (fraction numbers 15 to 34, from left to right). Filters were developed until the same total intensity of ␣-tubulin was obtained for each panel. in the second round of cell cycle (Figs. 6 -8) and slowly growing cells which are not at G 0 phase express less abundant CCT than rapidly growing cells (Fig. 1). These observations indicate that CCT subunit expression levels are tightly controlled during the cell cycle and this control is primarily exerted at the mRNA level.
In apparent contradiction to our results, Dittmar et al. (57) recently reported G 2 /M phase-specific expression of CCT (no CCT protein at G 1 and S phases) using Ehrlich ascites tumor cells separated by centrifugal elutriation, although its mRNA was detectable by Northern blotting throughout the cell cycle. The cell line used was different from those examined here. However, this is unlikely to be the reason for the different conclusions since we obtained consistent results using many kinds of cell lines. In addition, large amounts of signals on CCT can be detected in non-M phase cells by indirect immununofluorescence of monkey CV-1 (45) and human HEp-2 (36) cells. Thus, the basis of this contradictory result remains unclear unless Ehrlich ascites tumor cells are an exceptional case.
Possible Role of CCT in Cell Growth and Cell Cycle Progression-It is well known that CCT can assist the folding of actin and tubulin in vitro (16 -18, 20) and that CCT binds newly synthesized actin and tubulin in vivo (22)(23)(24). In the present study, coregulation of CCT and tubulin was often observed (Figs. 2-4 and 8) although it is not always (Fig. 1). Since a fraction of tubulin was found to associate with the CCT complex in early S phase cells, but not in G 0 /G 1 -arrested cells (Fig.  9), one of the CCT role for cell growth seems to be assisting in the folding of tubulin. On the other hand, the amount of actin showed little change during differentiation or growth factor starvation but was slightly up-regulated during synchronous recovery (Figs. 2, 4, and 6). Although these results do not exclude the requirement of CCT for actin folding, actin may require less abundant CCT than tubulin to fold the same number of molecules in vivo since the actin synthesized in vivo is known to occupy CCT more transiently than tubulin (22,24) and the actin bound to CCT was below the level of detection in the present experiments (Fig. 9).
Although the seven kinds of cells used in this study show various growth rates and contain different levels of tubulin and actin under normal growth conditions, the expression levels were not significantly correlated with growth rate (Fig. 1). These results indicate that the growth-dependent up-regulation of CCT does not always occur in cells abundant in tubulin or actin, and thus suggest that the contribution of the abundant CCT in rapidly growing cells is not restricted to the folding of tubulin and actin. Many other targets of CCT-assisted folding have been suggested: ␣-transducin (20) and firefly luciferase (19,26) can be folded by assistance of CCT in vitro. CCT binds unfolded cofilin (27), the 22-kDa peroxisomal membrane protein (28) and cyclin E (29). Recently, Thulasiraman et al. (24) detected more than 70 different CCT-bound polypeptides in vivo by two-dimensional gel electrophoresis of immunoprecipitated proteins. These observations indicate that CCT can interact with many proteins other than tubulin and actin. It is likely that CCT assists in the maturation of proteins up-regulated at G 1 /S transition and/or early S phase.
One strong candidate of such proteins is cyclin E, because this protein is known to be a G 1 cyclin involved in G 1 /S phase transition (47,48) in association with the cyclin-dependent kinase Cdk2 (58). CCT is essential for de novo synthesis of cyclin E in yeast, and cyclin E becomes associated with CCT before interacting with Cdk2 in human cells (29). Thus, CCT abundance from G 1 /S transition to early S phase may contribute to effective production of active cyclin E. In support of this idea, the time courses of CCT and cyclin E expression during synchronous culture of DA3 cells are very similar to each other (Fig. 6), although we could not detect the binding of CCT to cyclin E probably due to transient association of these proteins, as described above.