Báo cáo khoa học: Dissecting the role of the N-terminal metal-binding domains in activating the yeast copper ATPase in vivo pptx

Therefore, we provide evidence of a dual role for the Ccc2 N-terminus, namely to receive copper from Atx1 and to convey copper to another domain of Ccc2, thereby activating the ATPase..

domains in activating the yeast copper ATPase in vivo

Isabelle Morin1,2,3,*, Simon Gudin1,2,3, Elisabeth Mintz1,2,3and Martine Cuillel1,2,3

1 CEA, DSV, iRTSV, Laboratoire de Chimie et Biologie des Me´taux, Grenoble, France

2 CNRS, UMR 5249, Grenoble, France

3 Universite´ Joseph Fourier, Grenoble, France

Introduction

Copper is an essential element for life because of its

ability to switch between Cu(I) and Cu(II) in vivo Yet,

because of its redox properties, copper can also be

toxic when present in excess Accordingly, the

intracel-lular concentration of this metal is tightly regulated by

a cascade of cytosolic and membrane proteins that

interplay to deliver copper to specific targets, namely

the cytosolic superoxide dismutase, the mitochondrial

cytochrome c oxidase and secreted enzymes maturing

in the trans-Golgi network (TGN) [1] In humans, the

Menkes and Wilson proteins are copper ATPases

embedded in the TGN membranes that play an essen-tial role in copper delivery to the secretory pathway and in copper balance These ATPases receive copper from the metallo-chaperone Atox1 and assure its fur-ther transport into the TGN where the metallic cation

is inserted as a co-factor in various secreted enzymes [2] Among them, ceruloplasmin, a multicopper ferroxi-dase secreted by hepatocytes, carries almost all of the copper circulating in the plasma Two genetic disorders

of copper metabolism, called Menkes and Wilson dis-eases, are caused by deficient copper ATPases and

Keywords

Atx1; Ccc2; complementation; Cu(I);

domain–domain interactions

Correspondence

M Cuillel, CEA ⁄ iRTSV ⁄ LCBM, 17 rue des

Martyrs, 38054 Grenoble Cedex 9, France

Fax: +334 38 78 54 87

Tel: +334 38 78 96 51

E-mail: martine.cuillel@cea.fr

*Present address

Comparative Genomics Centre, James Cook

University, Townsville, Australia

(Received 24 March 2009, revised 28 May

2009, accepted 16 June 2009)

doi:10.1111/j.1742-4658.2009.07155.x

In yeast, copper delivery to the trans-Golgi network involves interactions between the metallo-chaperone Atx1 and the N-terminus of Ccc2, the P-type ATPase responsible for copper transport across trans-Golgi network membranes Disruption of the Atx1–Ccc2 route leads to cell growth arrest

in a copper-and-iron-limited medium, a phenotype allowing complementa-tion studies Coexpression of Atx1 and Ccc2 mutants in an atx1Dccc2D strain allowed us to study in vivo Atx1–Ccc2 and intra-Ccc2 domain– domain interactions, leading to active copper transfer into the trans-Golgi network The Ccc2 N-terminus encloses two copper-binding domains, M1 and M2 We show that in vivo Atx1–M1 or Atx1–M2 interactions activate Ccc2 M1 or M2, expressed in place of the metallo-chaperone Atx1, were not as efficient as Atx1 in delivering copper to the Ccc2 N-terminus How-ever, when the Ccc2 N-terminus was truncated, these independent metal-binding domains behaved like functional metallo-chaperones in delivering copper to another copper-binding site in Ccc2 whose identity is still unknown Therefore, we provide evidence of a dual role for the Ccc2 N-terminus, namely to receive copper from Atx1 and to convey copper to another domain of Ccc2, thereby activating the ATPase At variance with their prokaryotic homologues, Atx1 did not activate the Ccc2-derived ATPase lacking its N-terminus

Abbreviations

MBD, metal-binding domain; TGN, trans-Golgi network.

highlight the need for copper, as well as its toxicity

[3,4]

The yeast Saccharomyces cerevisiae provides an

excellent model with which to characterize copper

delivery to the secretory pathway in further detail

because the proteins involved in copper homeostasis

are highly conserved from yeasts to humans The first

step occurs at the membrane level where a reductase

converts extracellular Cu(II) into Cu(I); the latter then

enters the cell via the high-affinity transporter Ctr1 [5]

and is distributed to specific intracellular targets

Delivery to the secretory pathway is mediated by the

metallo-chaperone Atx1, the homologue of Atox1, and

by the copper ATPase Ccc2, homologous to the

Menkes and Wilson proteins [6,7] In the TGN lumen,

copper binds to and activates the multicopper oxidase

Fet3 [8], the homologue of the human ceruloplasmin

Fet3 is then secreted towards the plasma membrane as

a complex with the permease Ftr1 [9] This complex, in

which Ftr1 transports Fe(II) across the plasma

mem-brane and Fet3 transforms it into Fe(III), is

responsi-ble for high-affinity iron uptake in the absence of

siderophores [5,10,11] Therefore, copper-dependent

interactions between Atx1 and Ccc2 and further

cop-per transfer into the TGN lumen by Ccc2 are required

to activate high-affinity iron uptake in yeast (Fig 1A)

Atx1 is a 73-amino-acid cytosolic protein featuring a

babbab ferredoxin-like fold with a metal-binding site

located in its first loop This site comprises the

consen-sus sequence MXCXXC, whose two cysteines are

involved in binding Cu(I) [12,13] Ccc2 and the

mam-malian copper ATPases all possess a cytosolic

N-termi-nus made of between two and six metal-binding

domains (MBD) and each of these domains is similar

to Atx1 in terms of structure, consensus sequence and

copper-binding site [14] The Ccc2 N-terminus (Met1–

Ser258) is composed of two MBDs, designated M1

(Met1–Lys80) in tandem with M2 (Ser78–Gly151)

Yeast two-hybrid experiments have shown that Atx1

interacts with each in the presence of copper [6,15,16]

In particular, it has been demonstrated in vitro that

copper can be exchanged between Atx1 and the first

MBD expressed separately [17]

Several aspects of copper delivery into the TGN

have been studied Under conditions where only

high-affinity transporters can ensure uptake, i.e in

copper-and iron-limited media, deletion of ATX1 results in a

deficient high-affinity iron uptake and this leads to

growth arrest [15] Consequently, atx1D deletion

strains have been used as a tool to investigate whether

homologous proteins would replace Atx1 and ensure

copper transport into the TGN [18–22] Deletion of

the CCC2 gene also results in a deficient high-affinity

iron uptake and leads to growth arrest in copper- and iron-limited media Thus, ccc2D deletion strains have been used in complementation assays to assess the transport of copper inside the TGN by Ccc2 [8] and human homologues [23,24], or to investigate the role

of various amino acids in these proteins [25–28] Although these studies provide useful information about the functionality of the ATPase, they are all interpreted independently of the metallo-chaperone Indeed, studying the Atx1–Ccc2 route in vivo remains challenging, because of the existence of an Atx1-inde-pendent pathway for copper delivery to the TGN [15] This pathway was first discovered in atx1D, in which overexpression of Ccc2 allows growth in a copper- and iron-limited medium The current hypothesis involves the plasma membrane transporter Ctr1 which would undergo endocytosis and deliver copper directly to Ccc2, bypassing the need for Atx1 (Fig 1A)

We designed a sensitive complementation assay to study in vivo copper-dependent interactions between Atx1 or its homologues and Ccc2 The technical diffi-culty arising from activating Atx1-independent copper transfer to Ccc2 was first solved by expressing Ccc2 at

a very low level in an atx1Dccc2D deletion strain We show that the presence of one MBD on the Ccc2 N-terminus (M1 or M2) is necessary and sufficient to receive copper from Atx1 and to activate the ATPase, leading to copper transport into the TGN The high sensitivity of our new complementation assay allowed

us to determine that M1 and M2 expressed as indepen-dent cytosolic proteins were less efficient than Atx1 in transferring copper to the Ccc2 N-terminus However, these independent MBDs were able to deliver copper directly to another copper-binding site in Ccc2, whose presence was revealed by truncation of the N-terminus Our results suggest a dual role for the N-terminus of Ccc2, able to receive copper from Atx1, but also to transfer copper to another copper-binding site whose location in Ccc2 is discussed

Results and Discussion

A sensitive complementation assay to assess copper transfer from Atx1 to Ccc2 in vivo The aim of this study was to investigate in detail cop-per delivery to the Ccc2 N-terminus by the metallo-chaperone Atx1 and its further transfer into the TGN lumen As an alternative to the yeast two-hybrid sys-tem, we chose to coexpress Atx1 and Ccc2, the two partner proteins, in an atx1Dccc2D deletion strain and

to assess growth restoration on a copper- and iron-lim-ited medium, referred to as the selective medium

(0.1–1 lm copper and 100 lm iron chelated by 1 mm

ferrozine) Such low concentrations of copper and iron

in the selective medium ensure that Ctr1 and the Fet3–

Ftr1 complex are the only effective transporters for

copper and iron uptake by the cells [5,8] Bearing in

mind that Fet3 is essential for cell growth and that its

activity requires copper, the atx1Dccc2D deletion strain

cannot grow unless Atx1 and Ccc2 are coexpressed

and exchange copper, which then reaches Fet3 in the

TGN (Fig 1A, plan arrows) In this section, we report

the design of a new complementation assay to dissect the role of Atx1, M1 and M2 in activating Ccc2 in vivo (the strains and plasmids needed for this assay are described in Table 1)

Studying interactions between Atx1 and Ccc2 in vivo was challenging because of the presence of an Atx1-independent copper-delivery pathway to the TGN, activated by Ccc2 overexpression [15] (Fig 1A, dotted arrows) Accordingly, expression of Ccc2 using a multicopy vector pY–Ccc2 (PMA1 promoter) restored

act cat

Cytoplasm

Golgi lumen

C

D

E B

E

E-P

(2)

H 2 O Pi

(4)

(1)

nCulum

A

Ctr1

Ctr1

Ctr1

Golgi

ATP

ADP

Fet3

Atx1

Fet3

Ftr1

Ccc2 Ctr1

Ctr1

Ctr1

Golgi

Plasma membrane

ATP

ADP

Fet3

Atx1

Fet3

Ftr1

Ccc2

Fig 1 Ccc2 function, enzymatic cycle and schematic representations (A) Copper delivery to the secretory pathway in yeast Black dots rep-resent Cu(I), white dots reprep-resent Fe(II) Both Cu(I) and Fe(II) are produced by Fre1, a membrane reductase that is not reprep-resented here to simplify the scheme Ctr1, high-affinity Cu(I) transporter at the plasma membrane; Atx1, cytosolic metallo-chaperone; Ccc2, copper ATPase

at the TGN membrane; Ftr1, high-affinity Fe(II) transporter; Fet3, multicopper ferroxidase The Fet3–Ftr1 complex is secreted to the plasma membrane Under conditions where only high-affinity transporters are active, i.e in copper- and iron-limited media, the physiological pathway comprises Atx1, which delivers Cu(I) to Ccc2 (plain arrow) However, when Ccc2 is overexpressed, an endocytosis-dependent pathway brings copper to Ccc2 independent of Atx1 (dotted arrow) (B) The enzymatic cycle of Ccc2 (1) The ATPase binds Cu(I) at its transport site,

a step which induces a conformational change and (2) phosphorylation from ATP; (3) the energy is used to open the transport site on the other side of the membrane and to release Cu(I) into the TGN lumen; (4) the aspartyl-phosphate bound is hydrolyzed (C) Wild-type Ccc2 A schematic representation of Ccc2 structure that comes from the calcium ATPase structure [40] Empty circles represent known Cu(I)-binding sites N-terminus includes M1 [white hexagon with the Cu(I)-binding sequence CASC] and M2 (grey hexagon with CGSC) The membrane domain comprises eight helices and the transport site (CPC) is on helix 6 The catalytic loop (cat), where phosphoryl transfer occurs from ATP to the phosphorylation site (Asp627 indicated by the star) when Cu(I) is bound to the transport site, is localized between helices 6 and

7 The actuator domain (act), between helices 4 and 5, participates in dephosphorylation of the catalytic loop (D) Ccc2-like proteins bearing only one functional domain, either M1 or M2 (E) A Ccc2-like protein unable to bind copper at its N-terminus In (D) and (E), all the constructs are named according to their N-terminus Intact copper-binding motifs are denoted CXXC SXXS refers to domains whose cysteines are changed to serines Stripes recall their inability to bind Cu(I).

the growth of atx1Dccc2D in the selective medium,

even though Atx1 was absent (Fig 2A, lane 2) We

reasoned that lowering Ccc2 expression would impair

the Atx1-independent pathway of copper delivery to

the TGN Thus, CCC2 was cloned in the monocopy

plasmid pC, generating pC–Ccc2 (PMA1 promoter),

but growth was still restored in the absence of Atx1

(Fig 2A, lane 3) While trying to replace the PMA1

promoter with that of CCC2 [7] in pC–Ccc2, we

obtained a plasmid denoted pX–Ccc2 with neither

promoter Although pX–Ccc2 was obtained through

serendipitous discovery, it did not restore the growth

of atx1Dccc2D in the selective medium (Fig 2A, lane

4) unless Atx1 was coexpressed (Fig 2A, lane 5) No

complementation was obtained when the Atx1

copper-binding site was impaired as in Atx1ss (Fig 2A, lane

6) Growth restoration of atx1Dccc2D in the selective

medium using pC–Atx1 and pX–Ccc2 is therefore

likely to reflect interactions between Atx1 and Ccc2 In

all cases, growth was recovered by supplementation

with 350 lm iron, because iron at high concentrations

is imported into the cell by low-affinity transporters [8] Immunodetection of Ccc2 showed that pX–Ccc2 provides a very low expression of Ccc2, weaker than the expression level detected in YPH499, the wild-type strain (Fig 2B)

Two-hybrid experiments [6,16] suggested that Atx1 interacts with both domains of the Ccc2 N-terminus (M1 and M2) Our new complementation assay using

pX was further assessed by studying copper transfer from Atx1 to each MBD Accordingly, we chose to express Atx1 together with Ccc2 proteins bearing only one intact MBD We first generated pX–M1ssM2Ccc2

in which the copper-binding cysteines of M1 are chan-ged to serines (i.e M1 cannot bind copper) (Fig 1D) Coexpression of Atx1 with M1ssM2Ccc2 restored the growth of atx1Dccc2D in the selective medium, suggest-ing that copper transfer occurred in vivo between Atx1 and M2 (Fig 2C, lane 3) Growth restoration was dra-matically reduced when Atx1 was not coexpressed

Table 1 Yeast strains and plasmids.

(Fig 2C, lane 4), or when Atx1ss was coexpressed (data not shown) To investigate whether copper transfer would also occur between Atx1 and M1 in our system,

we tried to generate M1M2ssCcc2, but this protein was not expressed using either pX–M1M2ssCcc2 or pC–M1M2ssCcc2 Alternatively, another construct denoted M1Ccc2 was created in which M2 was deleted and replaced by M1 (Fig 1D) Coexpression of Atx1 with M1Ccc2 restored the growth of atx1Dccc2D in the selective medium, suggesting that copper transport occurs between Atx1 and M1 (Fig 2C, lane 5)

These results are in agreement with the conclusions drawn from two-hybrid experiments [6,16] and in addi-tion show that Atx1–M1 or Atx1–M2 copper-depen-dent interactions activate the ATPase in vivo, leading

to copper transfer in the TGN lumen and growth res-toration in the selective medium Our findings also agree with those obtained with both human ATPases, whose fifth or sixth MBD (of six) is sufficient for normal transport activity of the protein in Chinese hamster ovary cells [29,30]

Atx1 homologues and copper transfer to the Ccc2 N-terminus

In a previous study, some structural homologues of Atx1 (namely MerP, a mercury-binding protein from Cupriavidus metallidurans CH34, Ntk, the MBD of the cadmium ATPase from Listeria monocytogenes, and CopZ, the metallo-chaperone from Bacillus subtilis), were found to be able to deliver copper to endogenous Ccc2 in an atx1D strain [22] This study also showed that M1 and M2, expressed as independent proteins denoted Mbd1 and Mbd2, could also play the role of Atx1 Given the higher sensitivity of our new expres-sion system, we reasoned that expressing these Atx1 homologues would allow us to compare their ability to transfer copper to Ccc2 For this purpose, Mbd1, Mbd2, Mbd3 (M1 in tandem with M2 expressed as an independent protein), MerP, CopZ and Ntk were coex-pressed in atx1Dccc2D with Ccc2 and growth was assessed in the selective medium (Fig 3A) On the whole, Mbd1, Mbd2 and Mbd3 were less efficient than the other homologues in transferring copper to Ccc2, a feature that was not observed in our previous study [22] Coexpression of Mbd1 or Mbd3 with Ccc2 showed a quite similar growth, in agreement with structural data obtained in vitro for the whole N-termi-nus of Ccc2 showing unfolded M2 [31] However, coexpression of Ccc2 and Mbd2 on its own was more efficient than coexpression with Mbd1, however, it still did not reach the growth levels obtained by coexpres-sion with Atx1, CopZ, Ntk or MerP

1 2 3 4 5 6 7

Cell dilution

pC-Atx1

pX-Ccc2

pY-Ccc2 pC-Ccc2

pX-Ccc2 pX-Ccc2 pC-Atx1ss

pC-Atx1

+

+

+

+

+

+

+

YPH499

*

atx1

Δ ccc2

Δ

Cell dilution

pX-Ccc2

pX-M1ssM2Ccc2

pX-M1Ccc2 pX-M1Ccc2

+

+

+

+

pC-Atx1

1 2 3 4 5 6

A

B

C

Fig 2 Design of a sensitive complementation system to study

Atx1–Ccc2 interactions in vivo (A) The growth of atx1Dccc2D

cotransformed with indicated plasmids was assessed on the

selec-tive medium (1 l M copper, 100 l M iron and 1 m M ferrozine) or the

iron-supplemented medium (1 l M copper, 350 l M iron and 1 m M

ferrozine) Plasmids are indicated on the left, pC– stands for empty

plasmids, required to provide accurate selection markers (B)

Expression of Ccc2 using the pX vector Immunodetection of

endogenous Ccc2 in the wild-type strain YPH499 (1) and

immun-odetection of Ccc2 provided by the pX– vector in the presence of

Atx1 (pC–Atx1) in atx1Dccc2D (2) The proteins are indicated by *.

(C) The growth of atx1Dccc2D cotransformed with indicated

plasmids (and empty plasmids when required) was assessed on

the selective (100 l M ) and iron-supplemented (350 l M ) media.

As we replaced Ccc2 by M1ssM2Ccc2, the growth

pattern was similar (Fig 3B) Interestingly, when

M1Ccc2 was expressed in place of Ccc2 neither Mbd1

nor Mbd2 seemed efficient in restoring copper transfer

(Fig 3C) Taken together, these experiments suggest

that interactions between the independent MBDs and

one or the other MBD tethered to the membrane

pro-tein are poorly efficient or even not efficient in

restor-ing copper transfer Recallrestor-ing that copper transfer

from Atx1 to M1 or M2 is thought to occur via a

copper-mediated interaction [6,32], one possible

expla-nation is that M1 and M2 do not form homo- or

hetero-dimers in the presence of copper This may be a

special feature of copper ATPase MBDs, as opposed

to the metallo-chaperones, because Atox1, Atx1 and

CopZ are known to form dimers in the presence of

copper [33–35] Such an inability to form a dimer

would be beneficial to wild-type Ccc2, avoiding a

dead-end heterodimer between M1 and M2 and

favouring their interaction with Atx1 to collect copper

These results prompted us to investigate further the

role of the N-terminus in the overall function of Ccc2

In particular, we investigated whether Mbd1 or Mbd2 could deliver copper to some other site in Ccc2, apart from the N-terminus

Is the N-terminus dispensable in vivo?

In vitro studies of several ATPases have shown that their N-terminus modulates their activity [36,37] The current hypothesis involves domain–domain interac-tions between the N-terminus and other domains of the ATPase Indeed, in vitro studies of the purified N-terminus and catalytic loop of the Wilson ATPase have shown that these two domains interact in the absence of copper and that their interaction is dimin-ished by copper binding to the N-terminus [38] Such a domain–domain interaction has been proposed to hin-der access to the membrane transport site and there-fore to decrease the apparent affinity of the Wilson ATPase for copper [39] Similar conclusions arose from the study of CadA, a cadmium ATPase expressed

in Sf9 cells [36,40] It is currently thought that these inhibitory domain–domain interactions are relieved by binding of the metal to the N-terminus of the ATPase

To further study the role of the Ccc2 N-terminus in the overall function of the ATPase in vivo, we engi-neered DMBDCcc2 (Fig 1E), a protein in which the N-terminus is truncated from both M1 and M2 As shown by the complementation assays in Fig 4A, coexpression of DMBDCcc2 with Atx1 did not restore the growth of atx1Dccc2D in the selective medium Using green fluorescent protein fusions, the correct intracellular location of DMBDCcc2GFP was demon-strated (Fig 4B) Therefore, mislocation could not explain this nongrowth phenotype and the enzymatic activity of DMBDCcc2 had to be verified

Copper ATPases such as Ccc2 and its human homo-logues are P-type ATPases, whose enzymatic cycle has been well described (Fig 1B) [25,41,42] As for all P-type ATPases, the energy for the active transport of copper is delivered by phosphoryl transfer from ATP

to the aspartate belonging to the consensus sequence DKTGT in the catalytic loop A tight coupling between this chemical reaction and ion transport across the protein is ensured by phosphoryl transfer requiring copper to be bound to the so-called transport site [43] Once phosphorylated, the ATPase undergoes

a conformational change and copper can dissociate towards the TGN lumen The aspartyl-phosphate is then hydrolysed and the ATPase is ready for a new cycle According to the known structures of other P-type ATPases, the transport site is located among the transmembrane helices, as indicated in Fig 1C

Iron (µ M )

Cell dilution

pC-Atx1

pC-pC-Mbd1

pC-Mbd2

pC-Mbd3

pC-CopZ

pC-Ntk

pC-MerP

pX-Ccc2 pX-M1ssM2Ccc2 pX-M1Ccc2

ATPases

Fig 3 Evaluation of the Atx1 analogues efficiency in transferring

copper to Ccc2 N-terminus The growth of atx1Dccc2D

cotrans-formed with indicated plasmids was assessed on the selective

medium Proteins used as potential metallo-chaperones are

indi-cated on the left side (A) The potential metallo-chaperones are

expressed together with Ccc2; (B) Ccc2 is replaced by

M1ssM2Ccc2; (C) Ccc2 is replaced by M1Ccc2.

[44] In the case of Ccc2, copper needs to bind to the

CPC motif located at the sixth helix to allow the

phos-phorylation reaction, and thereby the release of copper

in the TGN lumen [25] When copper cannot bind to

the transport site, the ATPase cannot be

phosphory-lated by ATP and is considered inactive

DMBDCcc2, Ccc2 and its negative control, the

non-phosphorylatable mutant Asp627Ala, were

overpro-duced in Sf9 cells [25] Membrane fractions were used

to measure phosphoenzyme formation from

radioac-tive ATP in the presence of contaminating copper, an assay which also evaluates whether the transport site has bound copper (Fig 1B) The membrane fractions were mixed with [32P]ATP[cP], the reaction was acid-quenched at different times, samples were loaded onto acidic gels and the radioactivity incorporated at

110 kDa (Ccc2) or 94 kDa (DMBDCcc2) was evalu-ated Ccc2 displayed a strong and transient signal, similar to the one obtained for DMBDCcc2, in com-parison with the background, evaluated with Asp627Ala (Fig 4C) In conclusion, DMBDCcc2 was well localized in the cell and active in vitro but could not deliver copper to the TGN because the N-terminus was missing and Atx1–MBD copper-dependent inter-actions could no longer occur to activate the ATPase

We propose that, as described for the Wilson pro-tein, the N-terminus interacts with another domain of Ccc2 (the catalytic loop for example) and this interac-tion inhibits the ATPase activity Truncainterac-tion of the N-terminus alleviated these interactions in DMBDCcc2, thereby facilitating the access of copper

at the transport site in vitro, as shown previously for CadA [36,40] Consequently, because Cu(I) is not available in the cytosol [45], a suitable copper carrier may be necessary to deliver copper to DMBDCcc2

in vivo This is why we investigated whether Mbd1 or Mbd2 could directly deliver copper to DMBDCcc2

A role for M1 and M2 in delivering copper to another binding site in Ccc2

In the first section of this study, we showed that Mbd1 and Mbd2 (M1 and M2 expressed as independent proteins) were barely able to deliver copper to the N-terminus To evaluate the possibility that a native N-terminal MBD delivers copper to a domain other

A

B

C

Fig 4 DMBDCcc2 is active in vitro but does not restore copper transport to the TGN (A) The growth of atx1Dccc2D cotransformed with the indicated plasmids was assessed on the selective (100 l M ) and iron-supplemented (350 l M ) media (B) Localization of green fluorescent protein (GFP)-tagged proteins BY4741ccc2D cells transformed with pC–Ccc2GFP or pC–DMBDCcc2GFP were grown until the exponential phase in liquid cultures Ccc2GFP is localized partly at the vacuole membranes, as originally observed [7] A simi-lar pattern was observed with DMBDCcc2 The bar indicates 5 lm (C) Phosphorylation assay Ccc2, its non-phosphorylatable mutant Asp627Ala and DMBDCcc2 were expressed in Sf9 cells Mem-brane preparations were incubated in ice-cold buffer (20 m M bis-Tris propane pH 6.0, 200 m M KCl, 5 m M MgCl 2 ), phosphorylated with 5 l M radioactive ATP (1500 cpmÆpmol)1) and samples were taken at different times, as indicated All phosphoenzyme levels were compared with the value measured with the wild-type Ccc2

at 1 min (j) Ccc2, (r) Asp 627 Ala, (h) DMBDCcc2.

than the N-terminus, we coproduced Mbd1 or Mbd2

as independent proteins together with DMBDCcc2 in

atx1Dccc2D Mbd1ss was also generated where the

cysteines from the copper-binding site have been

chan-ged to serines (i.e Mbd1ss cannot bind copper) Unlike

Atx1 (Fig 5A, lane 2), Mbd1 and Mbd2 coexpressed

with DMBDCcc2 restored the growth of atx1Dccc2D in

the selective medium (Fig 5A, lanes 4,5) No growth

restoration was observed with Mbd1ss, showing that

copper binding to Mbd1 or Mbd2 is a necessary step

to allow copper translocation into the TGN by

DMBDCcc2 (Fig 5A, lane 3) In addition, no growth

restoration was obtained when CopZ or MerP was

coexpressed with DMBDCcc2 (data not show)

Inter-estingly, CopZ has an electrostatic potential surface

similar to that of Mbd1 and Mbd2 [22], but was still

unable to transfer copper to DMBDCcc2, suggesting

that electrostatic interactions are not the driving force for copper transfer to DMBDCcc2 HA-tagged Mbd1, Mbd1ss, Mbd2 and Atx1 were also coexpressed with DMBDCcc2 to evidence their expression by immunode-tection (Fig 5B) Thus, our data disclose special prop-erties of Mbd1 and Mbd2 which enable both to transfer copper to DMBDCcc2, although this transfer does not occur with Atx1

Therefore, we can conclude that both M1 and M2, when expressed as independent cytosolic proteins, are able to bind copper in the cytosol and transfer this metal to a copper-binding site in Ccc2, out of the N-terminus Although the site that receives copper from Mbd1 or Mbd2 has not yet been identified in DMBDCcc2, we can assume that when tethered to Ccc2, M1 and M2 deliver copper to the same site The identity of this copper-binding site remains unknown but one good candidate is the membrane transport site, which comprises a CPC motif in the sixth helix, whose cysteines are essential for ATPase activity [25] Therefore, unless another copper-binding site is disclosed in Ccc2, we propose that the copper ion bound to either N-terminal MBD is transferred directly to the membrane transport site, thereby acti-vating the ATPase The possibility of a still unidenti-fied protein participating in copper transfer between the MBDs and the membrane transport site cannot

be excluded here However, activation of the Wilson ATPase by Atox1 in vitro suggests that no other pro-tein is necessary to transport copper [46] To con-clude, our results suggest that the N-terminus of Ccc2

is not only involved in receiving copper from Atx1, but also plays a crucial role in the overall function of the ATPase by delivering copper to another site belonging to the rest of the protein It is an impor-tant link in the cascade that delivers copper to the ATPase transport site, immediately before the active transport step occurs

A model for the Atx1–Ccc2 route to the TGN under iron- and copper-limited conditions Our study, in synergy with data gathered from the literature, allows us to propose a mechanism regarding copper transport to the Golgi lumen, enhancing the crucial role of Atx1 and the Ccc2 N-terminus This model is described in Fig 6 and shows step-by-step copper transport to the Golgi

First, we hypothesized the presence of inhibitory domain–domain interactions involving the Ccc2 N-ter-minus, as represented in Fig 6A1, and similar to those demonstrated for the Wilson ATPase in vitro [38,39] These interactions were functionally evidenced by the

pY-Atx1

pY-Mbd1

pY-Mbd1ss

pY-Mbd2

pC-ΔMBDCcc2

pC- ΔMBDCcc2

pC- ΔMBDCcc2

pC-ΔMBDCcc2

Cell dilution

A

+

+

+

+

+

B

pY

A 1 M

- Y p

A 2 b M -pY

A 1 t A

p

A 1 M -pY

A s d M

pC-ΔMBDCcc2

chaperones

ATPase

1

2

3

4

5

atx1 Δ ccc2Δ

atx1 Δ ccc2Δ

Fig 5 M1 and M2 domains produced as independent proteins

(Mbd1 and Mbd2) are able to deliver copper to DMBDCcc2 (A) The

growth of atx1Dccc2D cotransformed with indicated plasmids was

assessed on the selective (100 l M ) and iron-supplemented

(350 l M ) media (B) Immunodetection of HA-tagged proteins

expressed in atx1Dccc2D using pY–Atx1HA, pY–Mbd1HA,

pY–Mbd2HA or pY–Mbd1ssHA together with pC–DMBDCcc2.

Cys to Ser mutations in all MBDs The -SH to -OH

substitutions made the N-terminus unable to bind

cop-per, but still able to interact with the catalytic loop,

leading to a conformation that was inactive Recent

structural data from the Archaeoglobus fulgidus copper

ATPase are in agreement with such interactions [47]

To alleviate these inhibitory domain–domain

interac-tions in wild-type Ccc2, copper transfer from Atx1 or

any other efficient metallo-chaperones to M1 or M2

domains of Ccc2 is essential (Fig 6A1 and A2) We

demonstrate here that one, and only one, functional

copper-binding site on the Ccc2 N-terminus is required

for efficient copper transfer from Atx1 to Ccc2

(Fig 6B where M1 is impaired and unable to bind

copper) Another way to disrupt these inhibitory

domain–domain interactions in Ccc2 is to delete the

N-terminus, as in DMBDCcc2 In vitro, DMBDCcc2 is

active and phosphorylatable by ATP in the presence of

copper Copper binding at the membrane transport site

CPC (i.e distinct from the N-terminus copper-binding

sites) induces phosphoryl transfer from ATP to

Asp627 (Fig 1B) (for a description of the Ca2+-ATPase

transport cycle based on recent structural data, see Olesen et al [44]) However, in vivo, Cu(I) is not freely available [45] and DMBDCcc2 did not restore copper homeostasis because the N-terminus was not present

to receive copper from Atx1 (Fig 5) The suitable cop-per carriers that were found here are Mbd1 and Mbd2, the two domains of the N-terminus expressed

as independent proteins, which could directly deliver copper to another binding site in DMBDCcc2 (Fig 6C1 and C2)

We conclude that the copper cargo, initially deliv-ered by Atx1 and bound to the Ccc2 N-terminus, is next transferred to another binding site in the protein through domain–domain interactions Interestingly, a recent report on in vitro copper transfer from the metallo-chaperone CopZ to the solubilized copper ATPase CopA from A fulgidus reaches the opposite conclusion [48] CopA possesses one MBD in its N-terminus and another in its C-terminus and although both MBDs are able to receive copper from CopZ this does not activate the ATPase Activation is obtained by direct delivery of copper by CopZ to CopA transport site in the membrane

Conclusion The complementation studies reported here have led

us to propose a model for in vivo copper delivery to the secretory pathway in S cerevisiae which takes into account in vitro studies performed on the human Wilson ATPase [38,39,46] Our model depicts pro-tein–protein and intraprotein domain–domain interac-tions leading to copper transfer from Atx1 to the core of the copper ATPase These interactions ensure that no release of free copper occurs in the cytosol during the transfer of the ion to the secretory path-way The assay designed here should be useful for designing a functional Atox1–Menkes ATPase or Atox1–Wilson ATPase pathway for copper delivery into the TGN in yeast This could enable further investigation of the effect of mutations in Atox1 or

in the Menkes or Wilson ATPase N-terminus that may impair copper transfer and result in disorders of copper homeostasis

Experimental procedures

Yeast strains Genetic modifications were performed on the YPH499 strain of S cerevisiae (a gift from Pierre Thuriaux, CEA,

disrupted in YPH499 using a deletion cassette bearing the

A

Fig 6 Model for the copper route to the TGN (A) Wild-type Ccc2:

(1) Atx1 can deliver Cu(I) to M1 or M2; (2) M2 has bound Cu(I); (3)

Cu(I) at the transport site (CPC) induces phosphorylation and

trans-port into the TGN (B) One MBD is necessary and sufficient to

receive copper from Atx1 and to activate the ATPase In this

repre-sentation, Atx1 delivers Cu(I) to the M2 domain of M1ssM2Ccc2.

(C) DMBDCcc2 receives Cu(I) from Mbd1 (or Mbd2) (1) and

trans-ports Cu(I) into the TGN (2).

TRP1 gene as a selection marker generating atx1D [22].

The chromosomal CCC2 gene was then disrupted in atx1D

using a deletion cassette inserting the URA3 gene into the

selected on Drop Out medium without uracil or tryptophan

and their genomic DNA was prepared according to Zhang

et al [49] PCR analysis confirmed the correct insertion of

deletion cassettes in the yeast genome The double mutant

strain was designated atx1Dccc2D

Expression constructs

Soluble proteins

Genomic DNA was used to amplify the ATX1 gene The

metal-binding domains of Ccc2, Mbd1 (Met1–Lys80),

Mbd2 (Ser78–Gly151) and Mbd3 (Met1–Gly151) were

generated by PCR amplification of the corresponding

seg-ments of CCC2 cDNA [25] Mbd1ss (i.e Mbd1 bearing

C13S and C16S mutations), was generated by PCR

start-ing from Mbd1 For expression in atx1Dccc2D, the

cod-ing sequences were inserted into mono- or multicopy

plasmids containing the PMA1 promoter and the ADC1

terminator and all bearing LEU2 as selection marker

(Table 1) The monocopy plasmid (pC-) was derived from

pRS315 [50]; the multicopy plasmid (pY-), from pYep181

[51] Site-directed mutagenesis was used to fuse the

HA-coding sequence at the end of Atx1, Mbd1, Mbd1ss

and Mbd2 coding sequences in the multicopy plasmid pY

(Table 1)

Expression constructs for mutants of Ccc2

Plasmids encoding the modified Ccc2 proteins described in

Fig 1 were generated with the QuickChange

site-mutagen-esis kit (Stratagene, Agilent Technologies, Massy, France)

using pSP72CCC2 as the template [25] In all cases, the

presence of designed mutations and the absence of

fortu-itous mutations were verified by automated DNA

sequenc-ing CCC2 was first cloned in a multicopy plasmid (pY)

and in a monocopy plasmid (pC) under the control of the

bear-ing HIS3 as selection marker (Table 1) The monocopy

plasmid was derived from pRS313 [50], the multicopy

plasmid from pYep181 in which the selection marker was

changed to HIS3 [51] In an attempt to clone the CCC2

promoter in place of the PMA1 promoter in pC–Ccc2, a

fortuitous ligation occurred between SacII and StuI sites

generating a plasmid with neither CCC2 nor PMA1

pro-moters This plasmid was called pX–Ccc2, where pX

stands for extra-low-copy expression plasmid, and used

for extra-low expression of other constructs (Table 1)

When the enzymatic activity of DMBDCcc2 protein had

to be evaluated, the mutant was cloned in the pFastBac

plasmid for expression in Sf9 cells [25] When the location

of the copper ATPase had to be verified, site-directed

mutagenesis was used to fuse the green fluorescent pro-tein-coding sequence at the end of Ccc2 and DMBDCcc2

in the monocopy plasmid pC

Expression in yeast and complementation assay atx1Dccc2D was co-transformed with one plasmid express-ing Atx1 or an Atx1-like protein and another expressexpress-ing wild-type or modified Ccc2 In some experiments, empty plasmids with the HIS3 or the LEU2 markers were used Transformants were selected on DO–Leu–His–Trp–Ura plates Randomly selected clones were grown for 3 days at

30C in copper- and iron-limited medium to assess growth restoration by expression of the proteins of interest (105 and 104 cells per drop) [52] The medium designated selec-tive contained 1.7 gÆL)1Yeast Nitrogen Base without copper

or iron (Qbiogene, MP Biochemicals, Illkirch, France),

5 gÆL)1 ammonium sulfate, 20 gÆL)1 glucose, 0.6 gÆL)1

(pH 6.1), 1 lm CuSO4, 100 lm NH4(FeSO4)2.6H2O and

1 mm ferrozine Higher concentrations of iron in this medium [350 lm NH4(FeSO4)2.6H2O] allow yeast cells with

an Atx1–Ccc2-deficient pathway to incorporate iron and therefore to grow The iron-supplemented medium was systematically used as a control after each transformation of atx1Dccc2D with modified Atx1 and⁄ or Ccc2 proteins For each complementation assay, we used three different clones obtained from at least two different transformations Plates were photographed after 3 days of growth at 30C

Yeast proteins extracts: expression level analysis

washed twice with ice-cold water They were then suspended

2 mm EDTA) supplemented with antiproteases [one tablet

of complete EDTA-free protease inhibitor (Roche Diagnos-tics, Meylan, France) and 1 mm phenylmethanesulfonyl fluoride] After disruption with glass beads, the lysate was centrifuged for 10 min at 1000 g to pellet cell debris and nuclei The resulting supernatant was spun down for 60 min

at 100 000 g and 4C and either the pellet was suspended in buffer A for membrane protein determination or the super-natant was concentrated using a 5000 Da molecular mass

Brumath, France) for HA-tagged metallo-chaperone detec-tion Protein concentration was determined using DC protein assay (Bio-Rad, Marnes-la-Coquette, France) For Ccc2 and Ccc2-like proteins, 30 lg of total protein was

blotting Proteins were immunodetected using the BM Chemiluminescence Western Blotting kit (Roche Diagnos-tics), Ccc2 with the polyclonal anti-Ccc2GB [25] and the metallo-chaperones with the anti-HA-peroxidase system (Roche Diagnostics)