A Positive Feedback Loop between Protein Kinase CKII and Cdc37 Promotes the Activity of Multiple Protein Kinases*

We report here the identification ofCDC37, which encodes a putative Hsp90 co-chaperone, as a multicopy suppressor of a temperature-sensitive allele (cka2-13 ts) of the CKA2 gene encoding the α′ catalytic subunit of protein kinase CKII. Unlike wild-type cells, cka2-13 cells were sensitive to the Hsp90-specific inhibitor geldanamycin, and this sensitivity was suppressed by overexpression of either Hsp90 or Cdc37. However, onlyCDC37 was capable of suppressing the temperature sensitivity of a cka2-13 strain, implying that Cdc37 is the limiting component. Immunoprecipitation of metabolically labeled Cdc37 from wild-type versus cka2-13 strains revealed that Cdc37 is a physiological substrate of CKII, and Ser-14 and/or Ser-17 were identified as the most likely sites of CKII phosphorylationin vivo. A cdc37-S14,17A strain lacking these phosphorylation sites exhibited severe growth and morphological defects that were partially reversed in a cdc37-S14,17E strain. Reduced CKII activity was observed in both cdc37-S14A andcdc37-S17A mutants at 37 °C, and cdc37-S14Aor cdc37-S14,17A overexpression was incapable of protectingcka2-13 mutants on media containing geldanamycin. Additionally, CKII activity was elevated in cells arrested at the G1 and G2/M phases of the cell cycle, the same phases during which Cdc37 function is essential. Collectively, these data define a positive feedback loop between CKII and Cdc37. Additional genetic assays demonstrate that this CKII/Cdc37 interaction positively regulates the activity of multiple protein kinases in addition to CKII.

exceptional cases Tyr) (5) in an acidic environment (2). CKII phosphorylates a broad spectrum of endogenous substrates involved in transcription, translation, signal transduction, and other functions.
Cdc37 was initially isolated in a genetic screen for mutants defective in progression through Start (6). Cdc37 function is required for the proper association of Cdc28 (yeast CDK1) with both G 1 and mitotic cyclins, thus demonstrating a role for Cdc37 in G 2 /M progression as well (7). Additional cdc37 mutants generated by Dey et al. (8) arrest in both G 1 and G 2 /M when shifted to restrictive temperature. CDC37 interacts genetically with several different protein kinases that are involved in diverse cellular roles. The list of genetic interactors of CDC37 include CDC28 (7), KIN28 (9), mammalian v-Src when expressed in yeast (8), MPS1 (10), STE11 (11), and CAK1 (12).
Consistent with the genetic results in yeast, mammalian CDC37 was found to encode the p50 subunit of an Hsp90 molecular chaperone complex that exhibits specificity for protein kinases (13). Although Hsp90 recognizes a diverse collection of client proteins, including steroid receptors, protein kinases, and others, Cdc37/Hsp90 complexes appear to interact exclusively with the catalytic subunit of protein kinases, with one notable exception (14,15). Binding sites for the protein kinase client and for Hsp90 have been mapped to the N-and C-terminal regions of Cdc37, respectively (13,15,16), providing further support for the notion that Cdc37 functions as a kinasetargeting co-chaperone of Hsp90. Cdc37 has also been demonstrated to possess chaperone activity on its own, independent of Hsp90 (17).
We show here that the CKII/Cdc37 connection is unique among the known interactions between Cdc37 and protein kinases because the highly conserved N terminus of Cdc37 contains an evolutionarily conserved CKII consensus site that is phosphorylated by CKII in vivo. Further, CKII and Cdc37 constitute a positive feedback loop. Although Hsp90 inhibition reduces CKII function in vivo, Cdc37 and not Hsp90 is the limiting factor in maintaining CKII function. We also present genetic evidence that CKII-mediated phosphorylation of Cdc37 is essential for the ability of the latter to maintain the activity of multiple protein kinases involved in diverse cellular functions. This reveals a previously unsuspected role for CKII in regulating the activity of diverse protein kinases (via Cdc37 phosphorylation).

EXPERIMENTAL PROCEDURES
Strains and Growth Media-S. cerevisiae strains used in this study are listed in Table I. Yeast strains were grown in rich glucose medium (YPD: 1% yeast extract, 2% peptone, and 2% glucose) or in supplemented minimal medium (18) at different temperatures as indicated in figure legends. Eviction of URA3-marked plasmids was accomplished by plating cells on supplemented minimal medium containing 0.75 mg/ml 5-fluoroorotic acid (Diagnostic Chemicals Ltd.). Sporulation was carried out in liquid sporulation medium (1% potassium acetate, 0.1% yeast extract, and 0.5% glucose). Geldanamycin (GA), 1 purchased from Sigma, was dissolved in dimethyl sulfoxide and added to warm medium soon after autoclaving. Escherichia coli strain DH5␣ (Clontech) was grown in Luria broth containing 50 g/ml ampicillin.
Multicopy Suppression Screen-The temperature-sensitive strain YDH13 (Table I) was transformed via the spheroplast method (18) with 10 -20 g of an S. cerevisiae genomic library in multicopy vector YEp24 (19) and plated on CM medium lacking uracil and leucine. Plates were incubated at 23°C for 24 h to allow accumulation of gene products expressed from YEp24 and then placed at 35°C (2°C above the maximum permissive temperature of YDH13). YEp24-based suppressor plasmids were isolated by transforming E. coli strain SK1108 to ampicillin resistance with DNA-containing extracts prepared from each suppressor strain (20). That suppression was plasmid-linked was confirmed by retransformation of YDH13, and the relevant gene was identified by similarly testing various subclones. Among the genes identified in the screen was CDC37. The correct assignment in this case was confirmed by constructing a frameshift mutation (an additional GATC sequence) at the BglII site in the CDC37 open reading frame. This construct is predicted to encode a severely truncated product of 120 residues and fails to suppress YDH13.
Strains carrying the above mutations as the only CDC37 allele were constructed as follows. CDC37 (in pRS316) was introduced into a diploid yeast strain (gift from David Morgan) carrying a LEU2-marked deletion in one copy of the CDC37 gene (7). The resultant strain was sporulated, and haploid progeny that were auxotrophic for leucine and uracil (i.e. cdc37 null and rescued by wild-type CDC37 on pRS316) were selected for further manipulation. After selection of a haploid strain with the a mating type, cells were transformed with CDC37, cdc37-S14,17A, cdc37-S14A, cdc37-S17A, and cdc37-S14,17E (all on pRS314). Transformants were plated twice in succession on minimal medium containing 5-fluoroorotic acid to evict the URA3-marked plasmid, yielding strains YSB11 through YSB15, respectively (Table I). All strains were tested for growth at a range of temperatures from 23 to 38°C.
Overexpression of CKII and Hsc/p82-The ␣ and ␤ subunits of Drosophila melanogaster CKII were expressed from the single-copy vector pBM272 under control of the GAL1,10 promoter, as described previously (24). Overexpression of Hsc82 or Hsp82 was achieved with multicopy YEp24 plasmids (gift of Kevin Morano) expressing each gene under control of its own promoter (25).
Western Blotting-Five A 600 units of cells were pelleted by centrifugation, resuspended to a total volume of 100 l by addition of sterile deionized water, and stored at Ϫ80°C. Thawed cell suspensions were mixed rapidly with 100 l of boiling 2ϫ SDS sample buffer, vortexed, and incubated at 100°C for 5 min. After heat treatment, the samples were centrifuged at 11,000 ϫ g for 10 min, and the clarified supernatants were stored at Ϫ20°C. Protein concentration was determined with the BCA protein assay (Pierce) using bovine serum albumin as a standard, and equal amounts of protein were electrophoresed in a 10% SDS-polyacrylamide gel and transferred to polyvinylidene difluoride membrane (Immobilon). Immunodetection was performed using the Amplified Alkaline Phosphatase Immun-Blot kit (Bio-Rad). Mouse monoclonal anti-Cdc37 (gift from Avrom Caplan, Mt. Sinai School of Medicine, New York, NY) was used at 1:1500 dilution, rabbit polyclonal anti-Cdc37 at 1:3000 dilution (gift of Steve Reed, The Scripps Research Institute, La Jolla, CA), mouse anti-phosphotyrosine antibody 4G10 (Upstate Biotechnology, Inc.) at 1:1000 dilution, and rabbit anti-Fpr3 (gift from Jeremy Thorner, University of California, Berkeley, CA) at 1:1500 dilution.
Metabolic Labeling and Immunoprecipitation-Strains were grown in YPD to mid-log phase at 26°C, and aliquots containing 5 A 600 units of cells were removed, labeled with [ 32 P]orthophosphate (500 Ci/sample) for 2.5 h at 26 or 37°C, harvested, and washed in water. Protein extracts were prepared as above, except that extracts were adjusted to 1% Nonidet P-40, 0.5% deoxycholate (to trap SDS in mixed micelles) prior to incubation with 1 l of affinity-purified rabbit polyclonal antibody against Cdc37 (gift of David Morgan, University of California, San Francisco, CA) (7). Immunoprecipitation was performed essentially as described by Cardenas et al. (26). Immunoprecipitated proteins were resolved by electrophoresis in an SDS-polyacrylamide gel (10%), and labeled Cdc37 was visualized with a Molecular Dynamics PhosphorImager.
Morphological Examinations-Strains were grown to mid-log phase at 23°C and fixed by addition of formaldehyde to a final concentration of 3.7%. Cell morphology was visualized using Nomarski optics. For generation of growth curves, saturated cultures of the relevant strains were inoculated into fresh YPD at a starting concentration of 3 ϫ 10 5 cells/ml. Cultures were grown at 23°C with vigorous shaking at 450 rpm. To determine cell number, aliquots were removed at the indicated times, fixed with 3.7% formaldehyde, and counted in a hemacytometer.
In Vivo Assay for CKII Activity-Steady-state Tyr phosphorylation of Fpr3 in the absence of Ptp1 function was used as an assay for CKII activity in vivo (5). Deletion of PTP1 was carried out by the PCR method of Wach et al. (27). Two primers were used to amplify the kanamycin resistance gene from pFA-6A (gift of Peter Philippsen, Biozentrum, Switzerland) together with flanking sequences derived from the immediately upstream and downstream regions of PTP1. The forward primer used was 5Ј-CAAGAAAAGCGTTTGTTAGTAGTAGTTTGCGACAGTG-GAGCGTACGCTGCAGGTCGAC-3Ј, and the reverse primer used was 5Ј-TATTGAAAAAATGCACAATTAGGAACTTTTATATAGCTGTATC-GATGAATTCGAGCTCG-3Ј (PTP1 flanking sequences underlined). Disruptions were verified by PCR.
To assay steady-state Tyr phosphorylation of Fpr3, isogenic wildtype and mutant strains were routinely grown in minimal medium to mid-log phase at 23°C. For experiments involving GAL-regulated overexpression of Drosophila CKII, strains were grown at 30°C in 2% raffinose medium lacking uracil. Early log phase cultures were shifted to 37°C for 3 h, adjusted to 2% galactose to induce CKII expression, and incubated for an additional 3 h. To assay Tyr phosphorylation of Fpr3 at different cell cycle stages, cultures of YSB48 (Table I) were first grown in YPD to mid-log phase at 25°C. Cultures were then adjusted to 10 g/ml ␣-factor (4 h), 0.1 M hydroxyurea (5 h), or 30 g/ml benomyl (4 h) to promote arrest in the G 1 , S, or G 2 /M phases of the cell cycle, respectively. Tyr-phosphorylated Fpr3 was detected by Western blotting as described above.

Isolation of Cdc37 as a Multicopy Suppressor of cka2-13
Mutants-To identify genes which interact genetically with CKII in S. cerevisiae, we utilized strain YDH13 (Table I) to screen for multicopy suppressors of the temperature-sensitive cka2 allele, cka2-13 ts (28). The screen was carried out at the minimum restrictive temperature of 35°C, and one of the suppressor plasmids identified encoded the protein kinase-specific (co-)chaperone, Cdc37 (Fig. 1A). Multicopy CDC37 also suppressed the slow growth rate at permissive temperature and flocculation displayed by YDH13 (data not shown). CDC37 suppressed four other cka2 ts alleles (data not shown), as well as two alleles of the CKA1 gene, encoding the ␣ subunit of S. cerevisiae CKII. 2 Because Cdc37 has been shown to act as a chaperone either by itself (17) or in concert with Hsp90 (13), we assessed the effect(s) of GA, an Hsp90-specific inhibitor, on cka2-13 mutants. GA is a member of the benzoquinoid ansamycin family of antibiotics that binds Hsp90 within a conserved pocket that constitutes the nucleotide binding site of Hsp90, and inhibits Hsp90 ATPase activity (29 -31). GA has been shown to indirectly inhibit a variety of oncogenic tyrosine kinases and other substrates of Hsp90 including mineralocorticoid receptor and glucocorticoid receptor (32)(33)(34)(35)(36). Refolding of denatured firefly luciferase by the Hsp90 chaperone complex is also inhibited by treatment with ansamycins (37). GA has been shown to inhibit Hsp90 function in vivo in yeast as well (25). In contrast to human cell lines, which are sensitive to GA at nanomolar concentrations, the growth of wild-type yeast is unimpaired even at concentrations of 2 mM GA (25). We have also found similar results with strains wild-type for CKII activity (data not shown).
As shown in Fig. 1B, cka2-13 mutants were sensitive to 35 M GA at 23°C (permissive temperature). Moreover, this sensitivity was suppressible by CDC37 overexpression. As might be expected, overexpression of the yeast homologs of Hsp90, HSC82 or HSP82 (which are the targets of GA in yeast; Refs. 25 and 31), also suppressed the GA sensitivity of cka2-13 mutants. A mutant allele of the yeast heat shock factor, hsf1-583, was marginally sensitive to GA at this temperature (used as a positive control) (25), whereas yeast that are wild-type for CKII were unaffected. CDC37 overexpression suppressed the GA sensitivity of cka2-13 mutants at temperatures from 21 to 34°C (lowest and highest temperatures tested; data not shown). The minimum restrictive temperature for cka2-13 mutants is 35°C, and the fact that they do not grow even at 21°C in the presence of GA emphasizes the importance of Hsp90 function for the survival of cka2-13 mutants. Consistent with the reported effects of GA on protein kinases in mammalian cells, treatment of cka2-13 cells with GA was found to cause decreased steady-state levels of CKII-dependent phosphorylation of Fpr3 ( Fig. 1C; see below for a discussion of this assay). We therefore conclude that, upon prolonged incubation of cka2-13 cells with GA, CKII activity drops below a critical threshold causing these cells to die. Cdc37 and Not Hsp90 Is the Limiting Factor in Maintenance of CKII Function-CDC37 overexpression has been shown to suppress the temperature-sensitivity of hsp90 mutants in an allele-specific manner (17), raising the possibility that Cdc37 enhances Hsp90 function in vivo in S. cerevisiae. Alternatively or in addition, Cdc37 might function independently of Hsp90 to protect one or more critical targets from denaturation in an hsp90 mutant background. Given the GA sensitivity of cka2-13 mutants, we sought to determine whether suppression of cka2-13 by CDC37 was mediated via Hsp90. To explore the general relationship between Cdc37 and Hsp90 in yeast, we made use of the previously reported observation that hsf1-583 cells are sensitive to GA because of reduced Hsp90 expression (25,38). If Cdc37 enhances Hsp90 function, CDC37 overexpression might be expected to suppress the GA sensitivity of an hsf1-583 mutant. As shown in Fig. 2A, overexpression of CDC37 did not enhance the ability of hsf1-583 mutants to grow in the presence of GA at either 25 or 30°C. This result suggests that CDC37 overexpression does not enhance Hsp90 function (or expression) in vivo. To determine whether the ability of CDC37 to suppress the GA sensitivity of cka2-13 mutants might be mediated via an increase in Hsp90 function, we tested whether direct overexpression of HSC82 or HSP82 could rescue the temperature sensitivity of a cka2-13 strain. As shown in Fig. 2B, neither gene is able to suppress this mutant under conditions where CDC37 is an effective suppressor.
The above results make it unlikely that CDC37 overexpression suppresses the GA sensitivity of a cka2-13 strain by enhancing Hsp90 function. One possibility is that Cdc37 is itself an independent chaperone of CKII, a conclusion consistent with the ability of purified Cdc37 to protect CKII activity in vitro (17). However, this interpretation does not explain the fact that cka2-13 mutants are sensitive to GA (Fig. 1B), which implies that Hsp90 function is required for CKII activity. An alternative possibility is thus that Hsp90 and Cdc37 are both required for CKII activity, possibly as a functional complex, but that Cdc37 is the limiting component (overexpression of Cdc37 but not Hsp90 suppresses the temperature sensitivity of a cka2-13 strain). A similar conclusion has been reached regarding the nature of the Cdc37/Hsp90 relationship with a temperature-sensitive mutant of the Hck kinase in human cells (39). Because our results were consistent with the possibility that Cdc37 is the limiting factor in the maintenance of CKII function, we decided to characterize further the nature of the CKII/ Cdc37 relationship. Cdc37 Is a Physiological Substrate of CKII-Alignment of Cdc37 sequences from diverse organisms revealed an evolutionarily conserved CKII phosphorylation motif near the N terminus of the protein (see Fig. 4A). All available Cdc37 sequences conserve a potential CKII phosphorylation site at Ser-14 (S. cerevisiae numbering), and the fungal sequences share a second potential site at Ser-17. Residue 17 represents the important ϩ3 determinant of the Ser-14 site, and this is potentially satisfied by a phospho-Ser residue in the fungal species or by a glutamate in the others. Consistent with these facts, yeast Cdc37 has been shown to be a substrate of mammalian CKII in vitro (17).
The above data suggest that Cdc37 may be an in vivo substrate of CKII. To test this possibility, we examined Cdc37 phosphorylation in vivo via metabolic labeling and immunoprecipitation in wild-type versus CKII-deficient strains. Western blotting revealed that these strains had comparable Cdc37 levels (data not shown). As shown in Fig. 3 (lanes 1-4), Cdc37 is radiolabeled in vivo in wild-type cells and also in a cka1⌬ CKA2 strain, at both permissive and non-permissive temperature. In contrast, incorporation into Cdc37 is strongly diminished in a cka1⌬ cka2 ts strain at permissive temperature, and essentially abolished at non-permissive temperature (Fig. 3,   lanes 5 and 6). The simplest interpretation of these data is that Cdc37 is a direct substrate of CKII in vivo.
To confirm the importance of the CKII phosphorylation site FIG. 1. Cdc37 and Hsp90 augment CKII function. A, multicopy suppression of cka2-13 by CDC37. YDH6 (cka1⌬ CKA2; row 1), YDH13 (cka1⌬ cka2-13 ts ; row 2), and YDH13 carrying multicopy CDC37 (row 3) were spotted in triplicate on YPD and grown at the indicated temperatures. B, suppression of GA sensitivity of cka2-13 mutant by CDC37 or HSC/P82 overexpression. Empty vector (YEp24) or multicopy plasmids expressing CDC37, HSC82, or HSP82 were transformed into YDH13, spotted in duplicate on YPD or YPD supplemented with 35 M GA, and grown at 23°C. NSY-B (hsf1-583) and YDH6 were used as controls (bottom row). All cells were spotted at 5000 cells/spot. C, GA-mediated inhibition of CKII activity in a cka2-13 strain. YSB48 (cka1⌬ CKA2 ptp1) and YSB49 (cka1⌬ cka2-13 ts ptp1) were grown in minimal medium at 23°C to mid-log phase, transferred to YPD with or without 100 M GA, and incubated for 3 h. Immunoblots of cell lysates were probed with anti-Tyr(P) antibody to assay for levels of phosphorylated Fpr3 as a measure of CKII activity in vivo. Blots were stripped and reprobed with anti-Fpr3.

FIG. 2. Cdc37 and not Hsp90 is the limiting factor in the maintenance of CKII function. A, overexpression of Cdc37 does not enhance
Hsp90 function. NSY-B (hsf1-583) cells were transformed with empty plasmid (pRS426) or CDC37 on a multicopy plasmid. The resulting strains were spotted in duplicate on YPD without or with 35 M GA at 5000 and 500 cells/spot and incubated at 25 or 30°C for 3 days. B, Cdc37 but not Hsp90 suppresses a CKII temperature-sensitive mutant. YDH13 (cka1⌬ cka2-13) cells were transformed with empty vector (YEp24) or multicopy plasmids encoding CDC37, HSP82, or HSC82. The resulting strains were streaked on YPD and incubated at 35°C for 3 days .   FIG. 3. Cdc37 is a physiological 1, 3, 5, 7, and 8) or 37°C (lanes 2, 4, and 6). Total cell extracts were immunoprecipitated with a polyclonal antibody against Cdc37, and immunoprecipitated material was analyzed by gel electrophoresis and phosphorimaging. Lanes 1 and 2, YPH499 (CKA1 CKA2 CDC37); lanes 3 and 4, YDH6 (cka1⌬ CKA2 CDC37); lanes 5 and 6, YDH13 (cka1⌬ cka2-13 CDC37); lanes 7 and 8, YRM 14.0 (CKA1 CKA2 cdc37-1) transformed with a single-copy plasmid expressing CDC37 or cdc37-S14,17A, respectively. The cdc37-1 allele contains a premature termination codon and does not produce a 70-kDa Cdc37 signal (7). motif on Cdc37, we compared the in vivo phosphorylation of wild-type Cdc37 with that of a Cdc37-S14,17A mutation (see below). As shown in Fig. 3 (lanes 7 and 8), in vivo labeling of the mutant is dramatically reduced compared with that of wildtype Cdc37. Although we cannot formally exclude the possibility of a conformational change that affects CKII phosphorylation elsewhere on the molecule, these results strongly suggest that Ser-14 and/or Ser-17 constitute a major site of Cdc37 phosphorylation in vivo.
CKII Phosphorylation Site Mutants of CDC37-To probe the functional significance of Cdc37 phosphorylation by CKII, mutant alleles encoding non-phosphorylatable (cdc37-S14,17A), semi-phosphorylatable (cdc37-S14A and cdc37-S17A), and quasi-phosphorylated (cdc37-S14,17E) derivatives of CDC37 were constructed (Fig. 4A) and expressed from their native promoters in a CDC37 null background. Western blotting with anti-Cdc37 antibody revealed that all of the mutants are expressed at levels comparable to that of wild-type (data not shown). The cdc37-S14,17A strain was the most severely affected, displaying an extremely slow growth rate (doubling time of 10 h versus 90 min for the isogenic wild-type control) and an elongated, enlarged morphology, even at 23°C (Fig. 4, B and C). The cdc37-S14A strain was more severely affected for both growth and morphology than the cdc37-S17A mutant (Fig. 4, B and C), supporting the idea that the evolutionarily conserved CKII site at Ser-14 is more important for Cdc37 function than Ser-17. The cdc37-S14,17A and cdc37-S14A strains were found to be temperature-sensitive for growth at 30 and 37°C, respectively, whereas cdc37-S17A was not sensitive for growth at any of the temperatures tested (data not shown). A cdc37-1 allele, which encodes a C-terminal truncation of the last third of the protein (7), had an essentially wild-type morphology under the same conditions (Fig. 4C), underscoring the importance of the CKII phosphorylation sites in Cdc37.
When compared with cdc37-S14,17A, the cdc37-S14,17E strain displayed a faster growth rate (doubling time of 4 versus 10 h) and less severe morphological defects (Fig. 4, B and C). Furthermore, the minimum restrictive temperature of cdc37-S14,17E was 35°C compared with 30°C for cdc37-S14,17A (data not shown). The ability of the double glutamate replacement to partially rescue the phenotype associated with the double alanine replacement is consistent with Ser-14 and Ser-17 serving as phosphorylation sites. The inability of cdc37-S14,17E to provide wild-type function might reflect the difference between a phospho-Ser versus a glutamate residue and/or a requirement for phosphoryl group turnover.
All of the CKII phosphorylation site alleles were recessive to wild-type (data not shown), indicating that they represent lossof-function mutations.
Because of the known relationship between Cdc37 and Hsp90, we also examined the growth of CKII phosphorylation site mutants of Cdc37 in the presence of the Hsp90 inhibitor GA. As shown in Fig. 5, a cdc37-S14A mutant is sensitive to GA (Fig. 5, vector), and this sensitivity is suppressed by overexpression of wild-type CDC37 as well as by cdc37-S17A. This mutant is also weakly suppressed by overexpression of cdc37-S14A itself, indicating dosage sensitivity of this mutant allele. In contrast, overexpression of cdc37-S14,17A not only fails to rescue the GA sensitivity of the cdc37-S14A mutant but results in a dominant-negative effect. These data confirm the importance of the CKII sites for Cdc37 function in vivo and support a potential interaction between Cdc37 and Hsp90 in yeast. They also corroborate the conclusions that Ser-14 is the more important of the two sites but that both are required for optimal function.
CDC37 and CKII Constitute a Positive Feedback Loop-The ability of CDC37 overexpression to suppress the temperature and GA sensitivity of a cka2-13 mutant (Fig. 1, A and B) prompted us to examine CKII activity in a strain with impaired Cdc37 function using the phosphorylation state of Fpr3, a well characterized substrate of CKII in S. cerevisiae, as an indicator of CKII activity in vivo (5). Fpr3 is the only known substrate of CKII phosphorylated on tyrosine (in addition to serine and probably threonine), and phosphorylation of Fpr3 (at Tyr-184) by CKII can thus be conveniently monitored by Western blotting of whole cell extracts with anti-phosphotyrosine antibody. Phosphorylation of Fpr3 at Tyr-184 is completely dependent upon functional CKII in vivo, but accumulation of Tyr-phosphorylated Fpr3 requires prior deletion of PTP1, which encodes a tyrosine phosphatase active against this site (5). PTP1 deletions had no effect on the growth, morphology, or temperature sensitivity of the strains used to monitor CKII activity (data not shown).
As shown in Fig. 6A, Tyr-184-phosphorylated Fpr3 represented a major band in a Western blot of a whole-cell extract prepared from a CDC37 ptp1 strain. An increase in the intensity of this band was observed upon overexpression of Drosophila CKII, as described previously (5). Relative to the CDC37 strain, extracts made from cdc37-S14A or cdc37-S17A cells had a lower steady state level of phosphorylated Fpr3, either with FIG. 4. CKII phosphorylation site mutants of CDC37. A, sequence alignment of the conserved N terminus of Cdc37 from various species (human, chicken, D. melanogaster, Caenorhabditis elegans, Schizosaccharomyces pombe, Candida albicans, and S. cerevisiae) plus S. cerevisiae Cdc37 phosphorylation site mutations. The conserved CKII motif is shown boxed, with the potentially modified Ser residues in bold. B, growth rates of CDC37 mutant strains. Strains (YSB11-YSB15) harboring the indicated CDC37 alleles were inoculated into YPD at a starting concentration of 3 ϫ 10 5 cells/ml and grown at 23°C. Cell number was monitored as described under "Experimental Procedures." C, morphology of CDC37 mutant strains. Strains harboring the indicated CDC37 alleles were grown at 23°C to mid-log phase, and photographs of cells were taken on a Zeiss IM 35 epifluorescence microscope fitted with Nomarski optics. SR672-1, carrying the C-terminal truncation allele cdc37-1, is included for comparison. or without overexpression of Drosophila CKII (the comparable behavior displayed by the S14A and S17A mutants in this experiment may reflect the stringent conditions employed, such that both mutants are severely impaired). The observed decrease in Fpr3 Tyr phosphorylation in the cdc37 mutant strains indicated that normal Cdc37 function is required to sustain CKII activity in vivo. Consistent with this conclusion, overexpression of cdc37-S14,17A or cdc37-S14A (unlike wildtype CDC37 overexpression) failed to suppress the GA sensitivity (Fig. 6B) or temperature sensitivity of a cka2-13 strain (data not shown). These results also implicate the CKII phosphorylation site(s) in the ability of Cdc37 to maintain CKII activity.
The genetic and biochemical results described here support the idea that CKII and Cdc37 exist in a positive feedback loop wherein CKII activates Cdc37 by phosphorylation, which then can activate/maintain CKII. Because Cdc37 is required at the G 1 and G 2 /M phases of the cell cycle (7,8), we asked whether CKII activity might peak during these same stages. To monitor CKII activity during the cell cycle, we arrested a strain that is wild-type for CKII in the G 1 , S, and G 2 /M phases of the cell cycle by treatment with ␣-factor, hydroxyurea, and benomyl, respectively. As shown in Fig. 6C, CKII activity toward Fpr3 is high in G 1 and G 2 /M phase-arrested cells and low in S phasearrested cells. A requirement for CKII in the G 1 and G 2 /M phases of the cell cycle has been documented previously (28), and another group has previously reported that CKII activity is most likely inhibited during S phase in human cells (using a substrate other than Fpr3) (40). Failure to phosphorylate Cdc37 could therefore contribute to the previously reported G 1 and G 2 /M arrest of temperature-sensitive CKII mutants (28).
The CKII/CDC37 Feedback Loop Helps Maintain the Activity of Multiple Protein Kinases-The CKII phosphorylation sites on Cdc37 are important for its ability to activate/maintain CKII activity (Fig. 6B). Because Cdc37 functions as a chaperone for a number of protein kinase catalytic subunits, we wished to determine whether phosphorylation of Cdc37 by CKII was important for these substrates as well. Several potential Cdc37 clients have been identified by genetic studies in S. cerevisiae, including Kin28 (9), Mps1 (10), and Cdc28 (7). Because Hsp90 and Cdc37 often share similar kinase substrates (17), we reasoned that temperature-sensitive mutants of KIN28, MPS1, and CDC28 might be sensitive to GA in a fashion similar to the cka2-13 strain (Fig. 1B).

FIG. 8. A positive feedback loop between CKII and Cdc37 positively regulates multiple protein kinases.
Cdc37 promotes the activity of diverse protein kinase clients including CKII, which in turn phosphorylates and activates Cdc37. P'ase indicates a potential protein phosphatase active against CKII-modified Cdc37.
FIG. 6. CKII phosphorylation sites on Cdc37 are essential for the ability of Cdc37 to maintain CKII function. A, cdc37-S14A and -S17A mutants exhibit reduced CKII activity. YSB41 (CDC37 ptp1), YSB40 (cdc37-S14A ptp1), or YSB72 (cdc37-S17A ptp1) transformed with empty vector (pBM272) or a plasmid expressing both subunits of D. melanogaster CKII were grown to mid-log phase in raffinose medium at 30°C, shifted to 37°C for 3 h, and then incubated in the presence of galactose for 3 h. Cell extracts were immunoblotted with anti-Tyr(P) antibody 4G10. A parallel blot was probed with anti-Fpr3. B, CKII phosphorylation sites on Cdc37 are essential for the latter to support CKII function. YDH13 (cka2-13) transformed with empty multicopy plasmid (pRS426) or multicopy plasmid expressing CDC37, cdc37-S14,17A, or cdc37-S14A were spotted at 5000 cells/spot on YPD or YPD supplemented with 35 M GA and incubated at 21°C for 3 days. C, CKII activity is elevated in cells arrested at G 1 or G 2 /M. Extracts prepared from asynchronous cultures of YSB48 (ptp1) or YSB48 cultures arrested in G 1 , S, or G 2 /M were probed with anti-P-Tyr antibody to assay steady state levels of phosphorylated Fpr3. Blots were stripped and reprobed with anti-Fpr3.
tively. All three mutants were found to be GA-sensitive at temperatures lower than their restrictive temperature (see Fig.  7), whereas their isogenic wild-type controls were insensitive (data not shown). As shown in Fig. 7, the GA sensitivity of kin28-ts3 and mps1-1 mutants was suppressed by overexpression of CDC37 but not by cdc37-S14,17A or cdc37-S14A. This is consistent with the idea that the CKII phosphorylation sites on Cdc37 are essential for its ability to protect Kin28 and Mps1 in addition to CKII and that these kinases are reduced in function/activity in both cdc37-S14,17A and cdc37-S14A.
The GA sensitivity of cdc28-109 was not suppressed by CDC37 overexpression (Fig. 7) at a variety of temperatures tested (nor was its temperature sensitivity suppressible by CDC37 overexpression; data not shown). However, we consider it unlikely that the CKII phosphorylation site mutants of CDC37 have adequate Cdc28 function because these mutants are partially suppressed by Cdc28 overexpression as well as by deletion of SWE1, which inhibits the mitotic form of Cdc28. 3 Moreover, cdc37-1 mutants have been shown to be limiting for Cdc28 function previously (7,12), and cdc37-1 mutants are synthetically lethal when combined with the cdc28-109 mutation (41). Cdc37 function thus is clearly required for Cdc28. However, in contrast to the other kinase mutants tested, Cdc37 function might not be limiting in cdc28-109 mutants. DISCUSSION We have presented biochemical and genetic evidence for a positive feedback loop between CKII and Cdc37. According to our model (see Fig. 8), CKII phosphorylates and activates Cdc37, which in turn promotes/maintains the activity of CKII. This model is supported by the following observations: 1) CDC37 functions as a multicopy suppressor of temperaturesensitive CKII alleles; 2) loss-of-function mutations in CDC37 result in reduced CKII activity toward a known physiological substrate, Fpr3, in vivo; 3) CKII phosphorylates Cdc37 in vivo at Ser-14 and/or Ser-17 (or at a site affected by mutation of these residues); 4) replacement of Ser-14 and Ser-17 with alanine results in severe phenotypic deficits that are partially reversed in the corresponding glutamate mutant; and 5) CKII phosphorylation site mutants of Cdc37 fail to suppress the GA sensitivity of a cka2-13 strain. Additionally, CKII activity peaks at the same stages of the cell cycle that Cdc37 function has been shown to be essential by several groups.
In addition to augmenting the activity of Cdc37 and CKII, the positive feedback loop between these two proteins also promotes the activity of additional protein kinases including, Mps1 (required for spindle pole duplication as well as the spindle checkpoint) (10), Kin28 (a C-terminal domain kinase of RNA polymerase II that mediates the interaction of polymerase II with capping enzymes during transcription of genes) (9,42), and Cdc28 (the yeast cell cycle engine) (7). At least two other kinases also represent substrates of the positive feedback loop, because the cdc37-34 allele (encoding a Ser-14 to leucine replacement) (14) has been shown to be defective in the maturation/activation of Ste11 (a MAP kinase involved in ␣-factor signaling in yeast) (11) as well as the mammalian oncoprotein v-Src (when expressed in yeast) (8). Additional Cdc37 clients such as Cak1 (12) may be dependent upon CKII-mediated phosphorylation of Cdc37 as well, but have not been tested explicitly. By regulating Cdc37 phosphorylation, CKII plays an important role in promoting the activity of multiple cellular kinases involved in diverse functions.
Because CKII, Cdc37, and the CKII phosphorylation site on Cdc37 are evolutionarily conserved, a similar mechanism that positively regulates multiple kinases (including possible or-thologs of yeast kinase clients of the CKII/Cdc37 feedback loop shown in Fig. 8) may be in place in higher organisms as well. Such a model also gives us a deeper appreciation of the molecular basis for the pleiotropic nature of CKII since, by phosphorylating and activating Cdc37 (which represents only one of the many CKII substrates identified so far), it can play a role in diverse signal cascades regulated by Cdc37 clients.
Cdc37 overexpression has been shown to induce tumors in mice (43,44). Although Cdc37 seems to cooperate with c-Myc and cyclin D1 in producing tumors, the biochemical mechanisms underlying its action most likely include multiple kinase substrates of Cdc37 that work in concert to promote proliferation. Proto-oncogenic kinases that Cdc37 has been shown to interact with include v-Src (8), Raf-1 (16), and CDK4 (13). CKII overexpression also produces tumors in mice (45,46), and the mechanism for this might involve its ability to activate Cdc37, which can then activate a host of other oncogenic kinases. It would be interesting to see whether mice overexpressing nonphosphorylatable Cdc37 would remain able to cooperate with c-Myc or cyclin D1 in promoting oncogenesis.
How does phosphorylation of Cdc37 by CKII regulate its function? At least two possibilities exist. Phosphorylation might regulate the interaction of Cdc37 with Hsp90 and/or other co-chaperones. Alternatively, it might affect the interaction(s) between Cdc37 and its protein kinase clients or other substrates. We have been unable to detect protein-protein interactions between yeast Cdc37 and any other protein kinase using a two-hybrid system, and others have had trouble isolating proteins that interact with yeast Cdc37 as well (12). However, a physical interaction between Cdc37 and Ste11 has been reported in yeast (11), and the highly conserved N-terminal half of Cdc37 (which contains the conserved CKII phosphorylation site) has been shown to interact with at least one protein kinase, Raf-1, in mammalian cells (16). Based on the latter observation, we suspect that the CKII phosphorylation-deficient mutants of CDC37 might be impaired in interactions with target protein kinases. Consistent with such a possibility, overexpression of the N-terminal half of yeast CDC37 has a dominant negative effect in a cdc37-S14A but not a CDC37 strain. 3 Because a feedback loop between CKII/Cdc37 regulates several other proteins as well and because CKII activity (against Fpr3) shows a cell cycle dependence, we suspect that there might also exist regulators (activators/inhibitors) of the feedback loop. Such gene(s) might also be responsible for modulation of CKII activity during the cell cycle.