Báo cáo khoa học: Cdc37 maintains cellular viability in Schizosaccharomyces pombe independently of interactions with heat-shock protein 90 doc

Con-served residues within the N-terminal domain, serines Keywords Cdc37; fission yeast; heat-shock protein 90 Hsp90; molecular chaperone; pombe Correspondence P.. pombe Cdc37, lacking t

pombe independently of interactions with heat-shock

protein 90

Emma L Turnbull, Ina V Martin* and Peter A Fantes

Institute of Cell Biology, School of Biological Sciences, University of Edinburgh, UK

Cdc37 is a molecular chaperone that was identified in

two different ways First, cdc37 was identified during a

screen for Saccharomyces cerevisiae mutants that arrest

with a cell division cycle (cdc) phenotype [1] and

secondly as a 50 kDa protein from chick cells called

p50 associated with the client v-src, that was

subse-quently shown to share sequence homology with S

cere-visiaeCdc37 [2] Cdc37 has been found to associate with

client proteins involved in a range of cellular processes

including cell cycle regulation, DNA and protein

syn-thesis and signal transduction (for review see [3])

Many protein clients rely on chaperones for activation,

folding and protection from degradation Client

pro-teins of Cdc37 are predominantly protein kinases such

as Cdk4 [4–6] and Raf1 [7] which bind the N-terminal domain of Cdc37 [8] Cdc37 has been identified in high molecular mass complexes in association with a wide variety of clients and other co-chaperones [6,9–11] Structurally, three domains of human Cdc37 were defined by limited proteolysis and peptide analysis and are referred to as the N-terminal, middle and C-ter-minal domains [8] At present there is no known role for the C-terminal domain, whereas functions for the N-terminal and middle regions have been identified The N-terminal domain of Cdc37 is the region most highly conserved among species and has been found to bind the client protein kinase, eIF2a kinase [8] Con-served residues within the N-terminal domain, serines

Keywords

Cdc37; fission yeast; heat-shock protein 90

(Hsp90); molecular chaperone; pombe

Correspondence

P A Fantes, Institute of Cell Biology,

School of Biological Sciences, Mayfield

Road, University of Edinburgh, Edinburgh

EH9 3JR, UK

Fax: +44 131 651 3331

Tel: +44 131 650 5669

E-mail: p.fantes@ed.ac.uk

*Present address

Institute of Physiology, RWTH Aachen,

Uniklinikum, Pauwelsstr 30, 52074 Aachen,

Germany

(Received 11 May 2005, revised 17 June

2005, accepted 20 June 2005)

doi:10.1111/j.1742-4658.2005.04825.x

Cdc37 is a molecular chaperone that interacts with a range of clients and co-chaperones, forming various high molecular mass complexes Cdc37 sequence homology among species is low High homology between yeast and metazoan proteins is restricted to the extreme N-terminal region, which

is known to bind clients that are predominantly protein kinases We show that despite the low homology, both Saccharomyces cerevisiae and human Cdc37 are able to substitute for the Schizosaccharomyces pombe protein in

a strain deleted for the endogenous cdc37 gene Expression of a construct consisting of only the N-terminal domain of S pombe Cdc37, lacking the postulated heat-shock protein (Hsp) 90-binding and homodimerization domains, can also sustain cellular viability, indicating that Cdc37 dimeriza-tion and interacdimeriza-tions with the cochaperone Hsp90 may not be essential for Cdc37 function in S pombe Biochemical investigations showed that a small proportion of total cellular Cdc37 occurs in a high molecular mass complex that also contains Hsp90 These data indicate that the N-terminal domain of Cdc37 carries out essential functions independently of the Hsp90-binding domain and dimerization of the chaperone itself

Abbreviations

cdc, cell division cycle; Cdk4, cyclin dependent kinase 4; CKII, casein kinase II; 5FOA, 5¢ fluoro-2¢-deoxyuridine; GST, glutathione

S-transferase; Hsp, heat-shock protein; HA, influenza hemagglutinin epitope.

14 and 17 in S cerevisiae [12] and serine 13 in rat [13]

and human [14] (equivalent to S cerevisiae serine 14),

have been identified as important sites of

phosphoryla-tion by casein kinase II (CKII) Phosphorylaphosphoryla-tion of

these conserved serine residues by CKII is required for

Cdc37 activity [12,13] There is evidence that Cdc37

and CKII maintain each other’s activity in a feedback

loop of activation [12] Phosphorylation of these serine

residues is important for client interactions, as the

unphosphorylated form of human Cdc37 was found to

have significantly reduced binding affinity towards

several client kinases [13] Cdc37 has been found to

display a range of chaperone activities towards bound

clients Cdc37 can facilitate the assembly of protein

kinases such as cyclin dependent kinase 4 (Cdk4) and

its partner cyclin D into complexes [5] Cdc37 can also

promote an activation competent state of the client

in vitro by cooperating with other co-chaperones such

as heat-shock protein (Hsp) 70 and Hdj1 [15] Cdc37

interacts with a range of clients and co-chaperones,

such as Hsp90, forming a variety of heterocomplexes

There are several lines of evidence which indicate

that Cdc37 functions in part with Hsp90 by delivering

client protein kinases to this cochaperone Cdc37 and

Hsp90 have been found in the same high molecular

mass (450 kDa) complex associated with the client

Cdk4 in NIH-3T3 cells [6] A complex consisting of

the interacting domains of yeast Hsp90 and human

Cdc37 has been crystallised and its structure

deter-mined [16] Amino acids 164–170 and 204–208 of

human Cdc37 were found to form a hydrophobic

patch that interacts with the N-terminal region of yeast

Hsp90 [16] Human Cdc37 binds both the N-terminal

domains and the adjacent linker regions of the Hsp90

dimer [17] Cdc37 binds to Hsp90 as a dimer [18] at

a 1 : 1 molar ratio [17] Cdc37 preferentially binds a

non-ATP bound form of Hsp90 and suppresses ATP

turnover [18] After Cdc37 has been released from the

tertiary complex with Hsp90 and the client, ATP

turn-over by Hsp90 is carried out as a two step process,

promoting conformational changes of the Hsp90–client

complex [19] Studies of the interaction between Cdc37

and Hsp90 are more advanced in mammalian systems

due to the unstable nature of the tertiary complex in

yeast systems [20] A genetic interaction between

Cdc37 and Hsp90 has been observed in S cerevisiae,

in that mutations compromised for Cdc37 and Hsp90

function are synthetically lethal [15] Identification of

biochemical interactions between Hsp90 and Cdc37 in

yeast systems is limited In S cerevisiae an interaction

between Hsp90 and Cdc37 has been shown using

recombinant glutathione S-transferase (GST)-Cdc37 in

pull-down experiments [21] and in the yeast two-hybrid

assay, using a mutant form of Hsp90 in which ATP hydrolysis was inhibited [22]

The fission yeast Schizosaccharomyces pombe has been used as a model eukaryote for the investigation

of a variety of cellular processes, notably cell cycle control and the responses to stress Little is known about Cdc37 in S pombe The cdc37 gene is essential for viability [23], and depletion of the Cdc37 protein in shut-off experiments led to heterogeneous cell pheno-types, indicating an involvement in several cellular roles that have not been elucidated A temperature conditional cdc37 mutant was isolated as a suppressor

of hyperactivation of the stress-activated mitogen acti-vated protein kinase pathway [24], and a direct interac-tion between Cdc37 and the client kinase Spc1⁄ Sty1 was demonstrated We set out to identify which domains of S pombe Cdc37 were essential for func-tion We generated a series of truncation mutants of cdc37 and expressed them in a cdc37Dstrain to ascer-tain their ability to compensate for loss of wild-type Cdc37 Surprisingly, we discovered that expression of the N-terminal domain alone can sustain cellular via-bility These truncated proteins do not contain the pos-tulated Hsp90-binding domain, suggesting that binding

of the cochaperone Hsp90 by Cdc37 is not required for cellular viability These data indicate that Cdc37 has an essential role, independent of interactions with Hsp90 However, biochemical investigations reveal that

a small proportion of total Cdc37 protein is associated with the cochaperone Hsp90 in a high molecular mass complex

Results

Human and S cerevisiae Cdc37 are functional homologues of S pombe Cdc37

Alignment of Cdc37 homologues from human,

S cerevisiae and S pombe show low overall sequence identity (Fig 1A) Despite low overall sequence homology, specific regions of Cdc37 are more highly conserved The N-terminal domain of Cdc37 is the most highly conserved region and is involved in client interactions [8] In the N-terminal 40 amino acids there

is 80% identity between the S pombe and S cerevisiae sequences and 50% identity between the S pombe and human proteins To investigate conservation of Cdc37 function between species, plasmids encoding human,

S cerevisiae and S pombe Cdc37 were introduced into the S pombe strain ED1526 and expressed from pREP81 in the plasmid shuffle assay (see below) (Fig 1B) Note that expression of wild-type S pombe Cdc37 from pREP81 generates a level of Cdc37

protein very similar to endogenous (data not shown).

S cerevisiae CDC37 expression was able to maintain

cellular viability (Fig 1B) This observation suggests

that there is functional equivalence between yeast

Cdc37 proteins At wild-type expression levels, human

Cdc37 was unable to sustain cellular viability

(Fig 1B), although increased expression from pREP1

restored cellular viability (Fig 1C) These data suggest

that human Cdc37 is a functional homologue of

S pombeCdc37, although the human protein may act

inefficiently in S pombe

Affinity purified S pombe Cdc37 antibody

To investigate Cdc37 in S pombe a polyclonal

anti-body was raised in rabbit and affinity purified The

specificity of this antibody was tested by western blot

analysis against GST, S pombe whole cell protein

extracts and GST-Cdc37 The antibody recognized

GST-Cdc37 and a 64 kDa protein from S pombe

tein extracts (Fig 2A) To verify that the 64 kDa

pro-tein is indeed Cdc37, S pombe Cdc37 antibodies were

depleted from the antiserum by preincubation with

GST-Cdc37 conjugated to glutathione beads Western

blot analysis on GST, S pombe protein extracts and GST-Cdc37 using depleted serum results in a loss of signal against S pombe protein extracts and recombin-ant Cdc37 (Fig 2A) The predicted molecular mass of

S pombeCdc37 is 56 kDa, but by SDS ⁄ PAGE gel it runs at  64 kDa S cerevisiae Cdc37 has a predicted molecular mass of 58.4 kDa, but was found to run on

an SDS⁄ PAGE gel at  68 kDa [25] Taken together these data indicate that the observed 64 kDa protein corresponds to S pombe Cdc37 which the antibody specifically recognizes

The C-terminal domain of S pombe Cdc37 is not essential for in vivo function

Biochemical investigations using limited proteolysis and peptide analysis have defined three discrete domains in human Cdc37 (p50); an N-terminal domain consisting of amino acids 1–126, a middle region com-posed of residues 128–282 and a C-terminal domain

of amino acids 283–378 [8] By aligning human and

S pombe Cdc37 sequences we were able to map these regions onto the yeast protein as indicated in Fig 2B

To identify the functional domains of Cdc37, we

A

Fig 1 Comparison of Cdc37 homologues (A) Alignment of Cdc37 protein sequences from human, S cerevisiae and S pombe Black boxes indicate identical amino acids amongst all three Cdc37 homologues Grey boxes denote identical amino acids between two Cdc37 homo-logues (B) Human, S cerevisiae and S pombe Cdc37 were expressed from pREP81 (wild-type levels) in the S pombe strain ED1526 by plasmid shuffle to determine their ability to sustain cellular viability (C) Human, S cerevisiae and S pombe Cdc37 overexpression (from pREP1) in the plasmid shuffle S pombe strain ED1526 to determine the ability of Cdc37 homologues to rescue an S pombe cdc37D.

expressed truncation mutants of Cdc37 in S pombe

ED1526 and tested them for function by plasmid

shuffle assay Truncation mutants Cdc37(1–428),

Cdc37(1–412), Cdc37(1–385), Cdc37(1–360) and

Cdc37(1–351), deleted in the C-terminal domain, were

able to compensate for loss of full length Cdc37

(Fig 2C) These data indicate that the C-terminal

domain is not essential for Cdc37 function in S pombe

The shorter truncation mutants, Cdc37(1–321),

Cdc37(1–273), Cdc37(1–264) and Cdc37(1–250), were

unable to support cellular viability (Fig 2C) Cells

expressing these truncations phenotypically resembled

those observed in Cdc37 depletion experiments,

com-prising morphologically heterogeneous nondividing

cells [23] Surprisingly, the mutants Cdc37(1–190) and

Cdc37(1–220) were able to support a low level of

growth when expressed at wild-type levels (Fig 2C),

although these truncations lack most of the middle and

all of the C-terminal domains, including the postulated Hsp90-binding and dimerization regions [16] Mutants with truncations extending into the N-terminal domain, Cdc37(1–155), Cdc37(1–120), Cdc37(1–60) and Cdc37(1–37), were unable to promote colony formation

at any temperature (Fig 2C) Mutants of S pombe Cdc37 deleted from the N-terminus for the first 20 and 40 amino acids were unable to maintain cellular viability at low or high expression levels (data not shown)

The Cdc37 mutants, Cdc37(1–321), Cdc37(1–273), Cdc37(1–264) and Cdc37(1–250), truncated within the middle domain, might be unable to sustain cellular viab-ility if the mutant proteins were unstable and present at reduced levels This has been observed for the S cerevis-iaemutant cdc37-1 which is truncated at codon 360 [25] The level of Cdc37 protein in each of the truncation mutants was assayed Whole cell protein extracts were

A

B

C

D

Fig 2 The C-terminus is dispensable for Cdc37 function in S pombe (A) Verification of the specificity of the anti-S pombe Cdc37 IgG (Upper panel) western blots of GST, native S pombe protein extracts and GST-Cdc37 were carried out with S pombe Cdc37 antibody (Lower panel) Anti-serum depleted of Cdc37 antibodies (see text) was used in western blots against GST, native S pombe protein extracts and GST-Cdc37 (B) The protein sequence of human (p50) and S pombe Cdc37 were aligned (Fig 1A) Structural domains defined in human Cdc37 by limited proteolysis and peptide analysis [8] were mapped onto the S pombe Cdc37 protein sequence The Hsp90 binding and dimerization domains of human Cdc37 identified by crystallization studies [16] were also mapped onto S pombe Cdc37 by alignment Boxes with horizontal stripes indicate the location of the postulated Hsp90 binding domain and the box with diagonal stripes denotes the homo-dimerization domain (C) Truncation mutants of S pombe Cdc37 were expressed from pREP81 in the S pombe plasmid shuffle strain ED1526 to determine their ability to sustain cellular viability (D) Protein levels of Cdc37 truncation mutants were compared to endogenous Cdc37 by western blot of total cellular protein extracts with the anti-S pombe Cdc37 IgG Asterisk indicates proteolytic truncation of endo-genous Cdc37.

made for truncation mutants expressed from pREP81 in

the cdc37+ strain ED1090 and equal amounts loaded

onto SDS polyacrylamide gels Western blot analysis

allowed comparison in each mutant of the levels of the

Cdc37 truncation protein with that of endogenous full

length protein All truncation mutants except for

Cdc37(1–155) yielded a truncated protein that was

detected by western blot with the Cdc37 antibody

(Fig 2D) In other experiments (not shown) we tested

the possibility that some truncated proteins, particularly

those truncated within the middle domain, might be

insoluble and therefore unable to contribute to essential

Cdc37 function(s) We prepared native extracts of

strains expressing various Cdc37 constructs and

frac-tionated them by centrifugation into supernatant and

pellet fractions which were then analysed by western

blotting However the truncated proteins showed no

increase in the proportion of insoluble fraction

com-pared with the full length Cdc37

All mutants deleted within the middle and C-terminal

domains were detected by the S pombe Cdc37 antibody

at levels approximately equal to endogenous Cdc37

However, Cdc37(1–190) and Cdc37(1–220) were

detec-ted by the antibody at reduced levels compared to

endogenous Cdc37 This may be due to low expression

levels, reduced stability of mutant proteins or poor

recognition by the S pombe Cdc37 antibody whose

recognition epitope(s) has not been fully characterized

Therefore, phenotypes observed for Cdc37(1–190) and

Cdc37(1–220) may arise from reduced protein levels

Overexpression of the N-terminal domain of Cdc37 is sufficient for cellular viability Several Cdc37 truncations were overexpressed from pREP1 to determine the effect of increasing mutant protein levels on cellular viability Overexpression

of the mutants Cdc37(1–321), Cdc37(1–273), Cdc37(1–264) and Cdc37(1–250), truncated in the middle domain, resulted in a wild-type phenotype (Fig 3A) in contrast to the inviability of strains expressing the same truncations at wild-type levels (Fig 2C) These data indicate that the truncated pro-teins have reduced function but this is compensated by increased expression so that the overall level of func-tion is above the threshold level for cellular viability Overexpression of the shorter truncation mutants, Cdc37(1–190) and Cdc37(1–220), which lack the mid-dle and C-terminal domains, resulted in growth com-parable to expression of pREP1-cdc37 (Fig 3A) Expression of these truncation mutants at wild-type levels was previously shown (Fig 2C) to support limi-ted growth, most likely due to their reduced protein levels This is confirmed by the complete restoration of viability by expression of these truncations at levels substantially greater than endogenous Cdc37 from pREP1 (Fig 3B) Proteolysis of Cdc37 was observed

in these experiments, most notably for Cdc37(1–190) and Cdc37(1–220) (Fig 3B) Overexpression of the truncation Cdc37(1–155), which lacks part of the N-ter-minal domain, was unable to support cellular viability

A

B

Fig 3 Overexpression of the N-terminal

domain of Cdc37 sustains cellular viability in

S pombe (A) Cdc37 truncation mutants

were overexpressed from pREP1 in the

S pombe plasmid shuffle strain ED1526 to

determine ability to sustain cellular viability.

(B) The truncation mutants

155, 190 and

pREP1-cdc37-220 were expressed in the S pombe

wild-type strain ED1090 to compared to protein

levels against endogenous Cdc37 by

west-ern blot of total cell protein extracts Protein

levels were assayed by western blot with

the anti-S pombe Cdc37 IgG Asterisk

indi-cates proteolytic truncation of endogenous

Cdc37.

(Fig 3A) Cdc37(1–155) protein was detected by

west-ern blot with the S pombe Cdc37 antibody and was

found to be present at a greater level than the

endo-genous full length protein (Fig 3B) In summary, the

data presented in Figs 2 and 3 show that expression of

the full N-terminal domain of Cdc37 at levels greater

than endogenous is sufficient for full cellular viability

in S pombe There is a clear distinction between

domains that are essential and those that are

dispen-sable for Cdc37 function in vivo in S pombe The

defi-ning boundary appears to be between the N-terminal

and middle domain The middle and C-terminal

domains that contain the postulated Hsp90 and

homo-dimerization domain are not essential for Cdc37

func-tion in S pombe provided the level of expression is

sufficient This points towards Cdc37 carrying out

essential functions that are independent of

Hsp90-bind-ing and homodimerization

A five amino acid in-frame insertion within the

middle domain of S pombe Cdc37 abolishes

function

Mutants Cdc37-I120, Cdc37-I252, Cdc37-I386 and

Cdc37-I422 generated by in vitro pentapeptide

muta-genesis described in Experimental procedures contain

five amino acid insertions commencing at residues 120,

252, 386 and 422, respectively (Fig 4A) The mutants

were expressed from pREP81 in the plasmid shuffle

assay (Fig 4B) Expression of Cdc37-I120, Cdc37-I386

and Cdc37-I422 resulted in a wild-type phenotype,

supporting growth comparable to pREP81-cdc37

expression, showing that these insertions in the N- and

C-terminal domains do not dramatically affect Cdc37 function In contrast, the mutant Cdc37-I252 was unable to support cellular viability and no growth was observed (Fig 4B) Cells appeared sick, being hetero-geneous in phenotype, characteristic of depletion of Cdc37 [23] According to the alignment of p50 and

S pombe Cdc37 shown in Fig 2B, the Cdc37-I252 insertion is located at the edge of the postulated Hsp90 binding domain located in the six helix bundle

of the middle domain [16] and may disrupt structure in this region

A small fraction of Cdc37 occurs in a high molecular mass complex

Hsp90 and Cdc37 from mammalian lysates were found

by size exclusion chromatography to occur in a range

of high molecular mass fractions consistent with obser-vations that various proteins associate with these chap-erones [6,9–11] We carried out size exclusion chromatography to determine whether S pombe Cdc37 occurred in a high molecular mass complex Recom-binant Cdc37 was initially studied to establish the elu-tion pattern of the chaperone alone The majority of recombinant Cdc37 was found to elute at around

200 kDa (Fig 5A) By analogy human Cdc37 is

 50 kDa, but in its native state exists as a dimer [16] and is structurally elongated which might affect its apparent size Similar factors may be responsible for the unexpected elution profile of S pombe Cdc37 Size exclusion chromatography of S pombe whole cell pro-tein extracts showed the majority of Cdc37 also eluted

at around 200 kDa (Fig 5A) A small proportion of Cdc37 protein eluted as a high molecular mass com-plex(es) at  669 kDa, while no recombinant Cdc37 eluted at this position These data suggest that in vivo,

a small fraction of Cdc37 interacts stably with other proteins to form a high molecular mass complex

Cdc37 and Hsp90 interact in high molecular mass complexes

We have been unable to identify any interaction between

S pombe Hsp90 and Cdc37 by coimmunoprecipitation from unfractionated native cell extracts of S pombe, pull-down using recombinant proteins or yeast two-hybrid assay A new technique was employed, using size exclusion chromatography to isolate, from cell extracts

of S pombe, fractions containing the high molecular mass complex of Cdc37 and probing these to identify an interaction with Hsp90 The elution pattern of the Cdc37 high molecular mass complex and Hsp90-influ-enza hemagglutinin epitope (HA) from the size

exclu-A

B

Fig 4 A mutational insertion within the middle domain abolishes

Cdc37 function (A) The location in Cdc37 of in-frame insertions of

five codons generated by in vitro pentapeptide mutagenesis is

shown schematically (B) Expression of in-frame insertion mutants

at wild-type levels in the S pombe strain ED1526 by plasmid

shuf-fle to determine the ability of these mutants to sustain cellular

via-bility in a cdc37D.

sion chromatography overlapped We then asked

whe-ther Cdc37 and Hsp90-HA were stably associated in the

high molecular complex To generate adequate material

for analysis of the Cdc37 high molecular mass

com-plex(es), preparative size exclusion chromatography was

carried out on a Sephacryl S-300 column (Fig 5B)

Fractions containing Cdc37 in high molecular mass

complex(es) were pooled and used as a source for

immunoprecipitations Immunoprecipitations using the

S pombeCdc37 antibody also precipitated Hsp90-HA

(Fig 5C), indicating an interaction between the two

chaperones in the high molecular mass complex

How-ever, the reverse immunoprecipitation using the HA

antibody did not yield Cdc37, perhaps because the

antibody may not have access to the HA epitope in the

complex

S pombe Cdc37 and Hsp90 interact genetically

In the previous section we show biochemically that a

small fraction of Cdc37 stably associates with Hsp90

Further evidence for an interaction comes from genetic interactions between the genes that encode for Hsp90 and Cdc37 The Hsp90 temperature-sensitive mutant swo1-26 [26] was crossed to each of the four cdc37 temperature-sensitive mutants, and each double mutant was found to be synthetically lethal at temperatures permissive for the single mutants (data not shown) A different genetic interaction between these two chaper-ones is shown by suppression of the temperature-sensi-tive mutant cdc37-13 lethality by increased expression

of Hsp90 (Fig 6A) In the converse experiment, increased expression of S pombe Cdc37 at low or high levels did not suppress the lethality of swo1-26 (data not shown) Double mutants of Hsp90 and Cdc37 tem-perature-sensitive genes in S pombe may be unable to sustain cellular viability at the permissive temperature either because Cdc37 and Hsp90 carry out important functions together (for instance, in a physical complex which does not form in the double mutant) or because they both have essential independent roles In the lat-ter case, the synthetic lethal defect may arise because

A

B

C

Fig 5 A small proportion of total cellular Cdc37 occurs in a high molecular mass complex associated with Hsp90 (A) An extract prepared from ED1537 cells was fractionated by analytical size exclusion chromatography on a Superose 6 column (lower panels) In a parallel experi-ment, recombinant S pombe Cdc37 expressed in E coli was run on an identical column (upper panel) The resulting fractions were western blotted and probed with antibodies specific for HA (to detect tagged Hsp90) or Cdc37 as indicated, to determine the distribution patterns of the two chaperones across the molecular mass range (B) Large scale preparative size exclusion chromatography of protein extracts from

S pombe ED1537 cells was carried out on a Superose 12 column and western blot analysis with the anti-S pombe Cdc37 and anti-HA IgGs identified the elution patterns of Cdc37 and Hsp90 (C) Immunoprecipitation reactions were carried out using antirat (control), anti-S pombe Cdc37 and anti-HA IgGs on fractions from the Superose 12 column containing the Cdc37 high molecular mass complex.

the cumulative effect of the loss of both chaperones

results in chaperone activity falling below a critical

threshold Hsp90 and Cdc37 may carry out the same

or similar independent functions, being able to

com-pensate for one another in some instances, shown by

the ability of Hsp90 to partially rescue the

tempera-ture-sensitive mutant cdc37-13

Discussion

Cdc37 sequence homology between different species is

low, but our results show that human and S cerevisiae

Cdc37 are functional homologues of Cdc37 in

S pombe Human Cdc37 was less efficient than the

S cerevisiae protein in sustaining cellular viability,

whereas overexpression of the human homologue was

required to rescue the S pombe cdc37D The structural

domains of human Cdc37 have been defined [8] and

were mapped onto S pombe Cdc37 to investigate the

functional regions of this chaperone protein An

inter-esting new result is that expression of the N-terminal

domain of S pombe Cdc37 is sufficient for cellular

viability These truncation mutants lack the postulated

Hsp90-binding and homodimerization domains,

indica-ting that these functions are not essential for Cdc37

activity in S pombe Interestingly, an interaction

between Cdc37 and Hsp90 was detected both

biochem-ically and genetbiochem-ically Size exclusion chromatography

showed that a small proportion of total cellular Cdc37

is found in a high molecular mass complex in associ-ation with Hsp90, indicating that these two chaperones interact in a nonessential manner in S pombe

Our observations are consistent with those of Lee

et al [27], showing that the C-terminal domain of Cdc37 is completely dispensable for function In

S cerevisiae the truncation mutant Cdc37(1–355) which lacks the latter part of the middle domain and the entire C-terminal domain was unable to restore cel-lular viability in a cdc37Dstrain whether expressed at low or high level [27] However, similar S pombe mutants, Cdc37(1–428), Cdc37(1–412), Cdc37(1–385) and Cdc37(1–351), truncated around the putative Hsp90-binding and homodimerization domains, were able to sustain cellular viability when overexpressed Lethality at wild-type expression levels of these trunca-tion mutants was most likely the result of reduced function, as protein levels were not compromised and there was no difference in the relative amounts of sol-uble and insolsol-uble Cdc37 in the truncation mutants One explanation for this phenomenon is that the N-terminal domain may be titrated away from carry-ing out essential roles by the aberrant middle domain attempting (for example) to interact with Hsp90 or homodimerize Alternatively, folding may be disrupted, negatively affecting the protein structure, as these mutants are truncated in the six a-helix bundle identi-fied in human Cdc37 [16] Disruption of the protein structure in this region might affect essential N-ter-minal interactions between Cdc37 and client proteins Whatever the reason, it appears that Cdc37 proteins with a defective middle domain are more compromised than mutants entirely lacking it Perhaps significantly, the mutant I252, in which five amino acid residues are inserted within this region, is not viable

We have shown that truncation mutants, Cdc37(1–190) and Cdc37(1–220), lacking the majority

of the middle and all of the C-terminal domains, sup-ported limited growth at wild-type levels Overexpres-sion of these truncations, increasing the protein abundance above endogenous Cdc37 levels, was able

to sustain cellular viability This differs from the obser-vations in S cerevisiae which showed that in a cdc37Dstrain overexpression of the truncation mutants Cdc37(1–148) and Cdc37(1–239), also truncated around the N-terminal and middle domain boundary, enabled slow growth in a temperature dependent man-ner [27] We have demonstrated that in S pombe, the middle and C-terminal domains are completely dispen-sable for cellular viability provided protein levels are not a limiting factor Truncations within the N-ter-minal domain were unable to sustain cellular viability

A

B

Fig 6 Hsp90 expression rescues a cdc37 temperature-sensitive

mutant at the restrictive temperature (A) Serial dilutions of the

temperature-sensitive mutant cdc37-13 with increased expression

levels of Hsp90 incubated on yeast extract for four days at the

per-missive (28
C) and nonpermissive (36
C) temperatures (B) Cells

of cdc37-13 with increased expression of Hsp90 plated on yeast

extract as serial dilutions and incubated for 4 days at 36
C.

at low or high expression levels Overexpression of

Cdc37(1–155), truncated into the N-terminal domain,

yielded protein at a level greater than endogenous

Cdc37, suggesting loss of essential Cdc37 function was

the limiting factor

Although Cdc37 truncation mutants lacking the

postulated Hsp90-binding domain can sustain cellular

viability in S pombe, we have identified an interaction

biochemically and genetically between these two

molecular chaperones A small fraction of total

S pombe Cdc37 forms a high molecular mass

com-plex that was also found to contain Hsp90

Identifica-tion of an interacIdentifica-tion between Hsp90 and Cdc37 has

been problematic due to the instability of the

inter-action and the small amount of Cdc37 that is present

in the high molecular mass complex(es) Biochemical

interactions between Hsp90 and Cdc37 have been

found to be very salt labile [28] and more unstable in

yeast than in mammalian systems [20] Our data show

that Cdc37 does interact with Hsp90 in S pombe, but

the presence of the postulated Hsp90-binding domain

is not essential for cellular viability It is possible that

Cdc37 carries out essential functions independently of

Hsp90 binding and that these roles do not require

homodimerization of Cdc37 The N-terminal domain

of Cdc37 may be involved in chaperone activities

independently of other co-chaperones as human

Cdc37 has been shown to play an crucial role in

pro-moting complex assembly between cyclin-dependent

kinases and their cyclin partners [5] Alternatively,

Cdc37 may interact with other co-chaperones possibly

through its N-terminal domain as S cerevisiae Cdc37

has been found to maintain clients in an activation

competent state in association with the co-chaperones

Hsp70 and Hdj1 [15] Yet another possibility is that

normally the Cdc37–client complex interacts with

Hsp90, perhaps presenting the client to Hsp90 When

only the N-terminal domain of Cdc37 is expressed, the

Cdc37–client complex forms but is unable to interact

in the usual way with Hsp90, but does so by random

encounters in the cytoplasm This would be expected

to be an inefficient process, and indeed truncation

mutants expressing only the N-terminal domain grow

poorly unless the expression level is increased

Experimental procedures

Alignment of Cdc37 protein sequence

Cdc37 protein sequences were aligned with vector ntitm

(Invitrogen Ltd., Paisley, UK) using BLOSUM62MT2

mat-rix and the output was displayed by genedoc (http://

www.psc.edu/biomed/genedoc/)

Antibodies

An anti-Cdc37 IgG was raised in rabbit against full length recombinant S pombe Cdc37 and affinity purified against recombinant GST-Cdc37 Anti-HA 12CA5 monoclonal antibodies (Roche Applied Science, Lewes, UK), antirabbit IgG HRP-linked antibody (Amersham Biosciences UK Ltd., Little Chalfont, UK), antimouse IgG HRP-linked antibody (Amersham) and antirat IgG HRP-linked anti-body (Amersham) were used as appropriate

Cloning and expression vectors

S pombe cdc37[23] was cloned into pREP vectors (pREP1, pREP81) for expression in S pombe pREP1 is a strong, thiamine-repressible promoter, while pREP81 retains the thiamine-repressibility but expression is 80–100-fold less [29] For expression in S pombe, the complete open reading frame of the human p50cdc37 cDNA was subcloned follow-ing PCR amplification from plasmid pET16b-p50 (kind gift from C Prodromou, Institute of Cancer Research, London, UK) into the Nde1-Xma1 sites of pREP1 and pREP81 vec-tors Similarly, S cerevisiae CDC37 was amplified from the plasmid E119 [30] by PCR and ligated into the Nde1-Xma1 sites of the vectors pREP1 and pREP81 S pombe cdc37+ was PCR amplified and cloned into the BamH1-EcoR1 sites

of the vector pGEX1 (Amersham) for expression in Escherichia coli Truncations of S pombe cdc37 were gener-ated by PCR mutagenesis introducing a stop codon fol-lowed by the restriction site Xma1 for cloning into the sites Nde1-Xma1 site of the pREP vectors S pombe swo1+(the gene encoding Hsp90 in S pombe) was a kind gift from

K Belaya (Department of Genetics, University of Cam-bridge, The Wellcome Trust⁄ Cancer Research UK Gurdon Institute, UK) who PCR amplified it from the cosmid c926 (Wellcome Trust Sanger Institute, Cambridge, UK) and ligated it into Nde1-Xma1 sites of pREP vectors The DNA sequences of all cloned inserts were verified by DNA sequencing

Random mutational analysis of S pombe cdc37 GPStm

-LS Linker Scanning System (NEB Ltd., Hitchin, UK) (#E7102S) was used to randomly mutate cdc37 by

in vitropentapeptide transposition introducing 15 bp inser-tions TnsABC* Transposase was used to insert a trans-poson derived from GPS5 (Transprimer-5 donor plasmid) randomly into target cdc37 fragments previously excised from pPRE81 with Nde1-Xma1 and purified Fragments of cdc37 containing insertions were identified by gel electro-phoresis, cut out of the gel and purified, then ligated into the restriction sites Nde1 and Xma1 of pREP81 The trans-poson transprimer was removed by restriction digestion with Pme1 followed by recircularization of plasmids by DNA ligase Resulting plasmids were sequenced to

deter-mine the identity and location of the nucleotide insertion.

Four mutants obtained contained in-frame insertions

resulting in the insertion of five amino acids into the Cdc37

protein Stop codons were inserted by the five codon

inser-tion in four mutants, each adding only 1–3 extra amino

acids after the initial insertion site Six mutants, Cdc37(1–

412), Cdc37(1–385), Cdc37(1–321), Cdc37(1–273), Cdc37(1–

264) and Cdc37(1–252), contained insertions that generated

a frameshift This resulted in an extra 10, 10, 8, 10, 9 and 7

amino acids, respectively, being added followed by a stop

codon

Site directed mutagenesis of S pombe cdc37

to generate truncation mutants

Truncation mutants were generated by PCR amplification

using a mutagenic oligonucleotide at the 3¢ end to introduce

a stop codon followed by the restriction site Xma1

Trunca-tions of cdc37 were cloned into Nde1-Xma1 of pREP vectors

and sequenced

Yeast strains

S pombe strains used were ED1090 (ura4-D18 leu1-32),

ED1560 (swo1-26 leu1-32) [26] was a kind gift from P Russell

(The Scripps Research Institute, La Jolla, CA, USA)

ED1537 (swo1 : 2HA-ura4+ura4-D18 leu1-32), a gift from

J Jimenez (Laboratorio Andaluz de Biologia, Universidad

Pablo de Olavide, Sevilla, Spain), expresses HA-tagged

Hsp90 from the endogenous swo1 (Hsp90) locus The strain

was used in biochemical experiments to investigate

interac-tions between Hsp90 and Cdc37 The S pombe strain

ED1526 [23] used for plasmid shuffle is deleted for

endog-enous cdc37 and kept alive by pREP82-cdc37+ We also used

temperature-sensitive mutants of cdc37; cdc37-681 [24],

cdc37-184, cdc37-13 and cdc37-J (E L Turnbull and P A

Fantes, unpublished observations) Medium (minimal

medium and yeast extract medium) for standard growth

con-ditions was used as described in [31] S pombe was grown at

32
C except for temperature-sensitive mutants which were

grown at the permissive temperature of 28
C and the

restrictive temperature of 36
C

Assay of mutant cdc37 genes in a cdc37D strain

by plasmid shuffle in S pombe

ED1526 cdc37::his1+ ade6 ura4-D18 leu1-32 his1-102

pREP82-cdc37+ was transformed with leu+ plasmids

(pREP1 or pREP81) expressing wild-type or mutant cdc37

Cells were precultured overnight in MM containing adenine

and uracil The attenuance at 600 nm was adjusted to 0.5

and serial dilutions were spotted onto MM + adenine +

uracil plates with and without 5¢ fluoro-2¢-deoxyuridine

(5FOA) Plates were incubated at 25, 28, 32 and 36
C for

4 days 5FOA selects against ura4+ cells but allows ura4) cells to grow Strains carrying a functional cdc37 construct expressed from pREP1 or pREP81 are able to grow in the absence of pREP82-cdc37+and are 5FOA resistant Strains with a nonfunctional cdc37 construct, cannot lose the pREP82-cdc37+and are unable to grow on 5FOA medium

Recombinant protein purification GST fusion proteins of S pombe Cdc37 were produced in

E coliBL21 cells Cells were lysed by sonication in 200 mm Tris [pH 8], 5 mm EDTA and 5 mm EGTA Recombinant protein was absorbed onto Glutathione Sepharosetm

4B (Amersham) in buffer containing 500 mm NaCl, 0.5% (v⁄ v) NP-40, 50 mm Tris pH 7.6, 5 mm EDTA, 5 mm EGTA and 1· Complete Inhibitors (Roche) Recombinant protein was eluted in 200 mm Tris (pH 8), 5 mm EDTA, 5 mm EGTA and 50 mm glutathione Samples were dialysed against

150 mm NaCl, 20 mm Tris pH 7.6, 1 mm EDTA and 1 mm EGTA Protein concentration was then determined by Brad-ford Assay (Bio-Rad Laboratories Ltd., Hemel Hempstead, UK)

Size exclusion chromatography

A culture of S pombe strain ED1537 at A600of 0.5 was har-vested and whole cell lysate extracted using glass beads and vortexing in lysis buffer [150 mm NaCl, 0.5% (v⁄ v) NP-40,

50 mm Tris (pH 7.5), 10% (w⁄ v) glycerol, 10· Complete pro-tease inhibitors, 20 mm molybdate] The insoluble debris was removed by centrifugation at 4
C for 15 min at 20 000 g.

Protein concentration was determined by Bradford Protein Assay (Bio-Rad) Size exclusion chromatography was carried out on either a Superose 6 column (for analytical prepar-ation) or a Sephacryl S-300 HR 26⁄ 60 column (for prepara-tive analysis) in SEC buffer [20 mm Hepes (pH 7.9), 3 mm MgCl2, 150 mm KCl, 10% (w⁄ v) glycerol, 1 mm dithio-threitol] and maintained at 4
C Fraction samples were then run using SDS⁄ PAGE for analysis by western blot

Immunoprecipitation The high molecular mass fractions containing Cdc37 from the Sephacryl S-300 column were pooled and quantified by Bradford Protein Assay (Bio-Rad) Equal amounts of pro-tein were used in each immunoprecipitation experiment Protein A Sepharosetm beads CL-4B (Amersham) were incubated with anti-(S pombe Cdc37) IgG, anti-HA IgG or anti-rat IgG (Amersham) for 30 min at 4
C Immunopreci-pitations were carried out at 4
C for 2 h Immunoprecipi-tates were washed four times with 1 mL of lysis buffer and resuspended in 2· SDS gel loading buffer Samples were run on polyacrylamide gels and western blotted using

anti-S pombeCdc37 and anti-HA IgGs