Eur. J. Biochem. 225, 677-685 (1994) 0 FEBS 1994

Expression, assembly and secretion of a fully active plant ferredoxin-NADP' reductase by Saccharomyces cerevisiae Jorgelina OTTADO, Adr ib K. ARAKAKI, Nora B. CALCATERRA and Eduardo A. CECCARELLI Molecular Biology Section, Departamento de Ciencias BiolCigicas, Facultad de Ciencias Bioquimicas y FarmacCuticas, Universidad Nacional de Rosario, Rosario, Argentina

(Received March 29/August 11, 1994) - EJB 94 0437/2

The flavoprotein ferredoxin-NADP' reductase catalyzes the final step of the photosynthetic electron transport i.e., the reduction of NADP' by ferredoxin. Expression and secretion of this enzyme was examined in Saccharomyces cerevisiae using a cDNA cloned from a pea library [New- man, B. J. & Gray, J. C. (1988) Plant Mol. Biol. 10, 511-5201. Two pea library cDNA sequences were employed, one corresponding to the mature enzyme and the other containing, in addition, the sequence of the transit peptide that directs ferredoxin-NADP' reductase to the chloroplast. These sequences were introduced into a yeast shuttle vector in frame with the mating factor a1 secretion- signal coding region under the control of its natural mating factor a1 promoter. Saccharomyces cerevisiae cells transformed with the recombinant plasmids were able to synthesize and secrete fully active pea ferredoxin-NADP' reductase. In both cases, a 35-kDa polypeptide was the major product. N-terminal sequencing of the secreted proteins indicates processing at position - 1 with respect to the N-terminus of the pea mature enzyme. Yeast cells transformed with plasmid encoding the ferre- doxin-NADP' reductase precursor secrete four-times more ferredoxin-NADP' reductase to the me- dium than cells transformed with the plasmid encoding the mature form of the enzyme. Ferredoxin- NADP' reductases purified from culture medium showed structural and enzymatic properties that were identical, within the experimental error, to those of native plant ferredoxin-NADP' reductase. The overall results indicate that pea ferredoxin-NADP' reductase can be properly folded and its prosthetic group assembled in the yeast endoplasmic reticulum, and that its natural transit peptide favors its secretion.

Ferredoxin-NADP' reductase is a FAD-containing pro- tein present in chloroplasts and cyanobacteria [l]. It catalyzes the last step of the photosynthetic electron-transport chain, the electron transfer between reduced ferredoxin and NADP', with formation of the NADPH necessary for CO, fixation through the Benson-Calvin cycle and other biosyn- thetic pathways [l]. This enzyme is an hydrophilic protein of about 35 kDa containing 1 mol non-covalently bound FAD/ monomer [1, 21. It is encoded in the cell nucleus and synthe- sized in cytoplasmic ribosomes as a precursor containing a 5-kDa transit peptide at the N-terminus [3, 41. The precursor is transported into chloroplasts, where the transit peptide is cleaved by a specific peptidase [5]. The mature holoprotein then binds tightly to the stroma surface of the thylakoid membrane via a 17.5-kDa polypeptide [6]. In addition to its role in NADP' photoreduction, ferredoxin-NADP+ reductase is able to catalyze in vitro the oxidation of NADPH by sui- table electron acceptors like potassium ferricyanide,

Correspondence to E. A. Ceccarelli, Area de Biologia Molecu- lar, Departamento de Ciencias BiolCigicas, Facultad de Ciencias Bio- quimicas y FarmacCuticas, Universidad Nacional de Rosario, Sui- pacha 531, (2000) Rosario, Argentina

Abbreviations. Hsp60, 60-kDa heat-shock protein; HsplO, 10- kDa heat-shock protein; ER, endoplasmic reticulum ; PhMeSO,F, phenylmethylsulfonyl fluoride ; MFal, mating factor a1, Hsp70, 70-kDa heat-shock protein.

Enzyme. Ferredoxin-NADP' oxidoreductase (EC 1.18.1.2).

dichloroindophenol, tetrazolium salts (diaphorase activity) or oxidized ferredoxin coupled to cytochrome c reduction (cyto- chrome c reductase activity) [ l , 7-91. These reverse reac- tions have been widely employed to investigate the catalytic mechanism of the enzyme [l]. The primary structures of sev- eral plant and cyanobacterial ferredoxin-NADP' reductases have been deduced from cDNA sequences [3, 4, 10-131. However, limited information concerning plastid import, pro- cessing and in vivo assembly of the enzyme is at present available.

Recently, systems were developed for the expression of both the spinach [14] and the pea ferredoxin-NADP' reduc- tases [15- 171 in Escherichia coli. The recombinant enzymes accumulated to high levels in the bacterial host. It was found that E. coli cells could assemble the prosthetic group FAD to yield an active holoenzyme that showed kinetic and spectral properties identical to ferredoxin-NADP' reductases isolated from plant tissues [14-171. With the aid of GroESL E. coli mutants we have demonstrated that folding and FAD assem- bly of ferredoxin-NADP' reductase expressed in E. coli was shepherded by bacterial chaperonins, i.e. the 60-kDa heat- shock protein (Hsp60) and lOkDa heat-shock protein (HsplO) [17]. It therefore seemed likely that a similar process should occur in chloroplasts involving the chaperonin60 and chaperoninlo homologous proteins [17, 181. The findings that such a diverse foreign protein can be expressed as active holoenzyme in E. coZi suggest that the assembly and incorpo-

678

ration of FAD in ferredoxin-NADP' reductase does not re- quire specific recognition of the apoprotein acceptor by the host and that prosthetic group assembly may occur by a pro- cess common to all organisms.

In this report we addressed these questions by expressing in Saccharomyces cerevisiae mature ferredoxin-NADP' re- ductase and its precursor fused in frame to the yeast a-factor leader sequence [19], that had proved to be useful for the export of foreign proteins [20-221. In that way we forced the enzyme to be folded in the endoplasmic reticulum (ER) lumen, where no Hsp60-Hsp10 have been identified [23,24].

The results presented indicate that in this heterologous system ferredoxin-NADP' reductase incorporates FAD and folds properly during the secretion process, resulting in the accumulation of a fully active enzyme in the culture medium. Secretion was higher for the ferredoxin-NADP' reductase precursor than for the mature enzyme, indicating that the presence of the natural transit peptide favors translocation of the protein to the yeast ER.

MATERIALS AND METHODS Plasmid construction

A PstI- EcoRI fragment (1.3 kbp) containing the coding sequence for pea ferredoxin-NADP' reductase precursor and the 3' flanking region (kindly provided by Dr J. C. Gray, University of Cambridge, England [4]) was inserted into compatible sites of pUC19. Plasmids containing inserts were selected and named pCV101. A 1305-bp fragment containing the coding sequence for pea ferredoxin-NADP' reductase and a section of the pUC19 polylinker were obtained by di- gestion of pCVlOl with HindIII and EcoRI and inserted into the HindIII site of pUC119 with the help of an EcoRI- HindIII linker, generating plasmid pCV1010. The EcoRI- HindIII 52-bp linker was obtained by digestion of pUC119 with both enzymes, isolated by agarose electrophoresis and electroelution, followed by purification by Nacs prepac (BRL). The HindIII-Hind111 insert purified from pCVlOl0 was ligated to the multicopy yeast shuttle vector pES (2p circle, LEU 2) [22]. Following transformation into competent JM109 E. coli cells, plasmids containing inserts were se- lected by a colony-filter hybridization technique using a 32p- labeled ferredoxin-NADP' reductase cDNA as a probe. Plas- mids carrying the desired orientation were further selected by restriction analysis and the junctions were verified by DNA sequencing. The resulting recombinant plasmid (pYPF, Fig. 1) contains the yeast a-factor leader coding region fused in frame through a 9-bp linker (3 amino acids) to the fourth codon of the pre-ferredoxin-NADP+ reductase apoenzyme. (Fig. 2).

For the preparation of pYMF (Fig. l ) , plasmid pCV105 1151 was used. Briefly, pCV105 contains a cDNA insert en- coding the entire sequence of mature pea ferredoxin-NADP' reductase and the last two residues of the transit peptide [4], fused in frame to the initial 15 codons of P-galactosidase in plasmid pUC9 and a pUCl9 BamHI-Sac1 19-bp linker. An 11-bp fragment of pCV105 corresponding to the coding se- quence of the /J-galactosidase was deleted by digestion with H i n d and SmaI and religation. Recombinant plasmids carry- ing the desired deletion were screened by the loss of a BamHI site and named pAC1. The complete cDNA encoding the entire sequence of mature ferredoxin-NADP' reductase (1178 bp) plus a region of the pUC9 and pUC19 polylinkers was inserted into the HindIII site of pUC119 using the

52-bp EcoRI-Hind111 linker described above, generating the plasmid pAC2. A 1230-bp fragment was purified after re- striction of plasmid pAC2 with HindIII and ligated to the multicopy shuttle vector pES [22]. Plasmids carrying inser- tions were selected by colony-filter hybridization and ana- lyzed for the desired orientation by restriction analysis. The linker region of the construction was further verified by DNA sequencing. The resulting recombinant plasmid pYMF con- tains the yeast a-factor leader coding region fused in frame through a 24-bp linker (8 amino acids) to the first codon of the mature ferredoxin-NADP' reductase cDNA. (Fig. 2).

Recombinant DNA techniques were performed by stan- dard procedures [2S, 261. E. coli JM109 [25] and S. cerevis- iae GRF-18 (Muta leu2-3 leu2-112 his3-11 his3-15 can1 mal) [22] were used throughout this work. Transformations of S. cerevisiae cells were carried out by means of published meth- ods [26]. Transformed yeast cells were grown at 30°C in YPD (2% yeast extract, 1% bacto-peptone, 2% dextrose) or complete minimal medium [26] containing 0.17% yeast ni- trogen base without amino acids, 0.13% complete amino acids mix without leucine, 0.5% ammonium sulfate and 2% dextrose.

Detection of ferredoxin-NADP+-reductase-secreting yeast strains

After colony growth on solid complete minimal medium, individual clones were transferred to YPD plates and grown for 24 h at 30°C. Plates were then exposed to a medium containing 50 mM Tris/HCl, pH 8.0,O.g mM NADP', 0.05% (mass/vol.) nitroblue tetrazolium, 3 mM glucose 6-phosphate and 1 U glucose-6-phosphate dehydrogenase and kept in the dark until color appearance. Under these conditions, plates remained pale yellow and clones secreting active ferredoxin- NADP' reductase stained deep purple. Ferredoxin-NADP' reductase secreted to the solid medium was also observed as a blue-purple halo surrounding secreting clones.

Secretion, detection and purification of recombinant ferredoxin-NADP reductase from yeast

S. cerevisiae harboring either the pYPF or pYMF were grown for 24 h with gentle shaking at 30°C in 5 ml complete minimal medium. Cells were collected by centrifugation and resuspended in 10 ml YPD medium and grown in the same conditions for 72 h. All subsequent steps were carried out at 4°C. Cultures were cleared by centrifugation (3000 g for 5 min). Supernatants were supplemented with 0.1 mM phenylmethylsulfonyl fluoride (PhMeS0,F) and brought to 50% saturation of ammonium sulfate. Following removal of a small amount of precipitated material by centrifugation, the supernatant solutions were brought to 75% saturation of anmonium sulfate. Proteins precipitated were collected by centrifugation (10000 g for 10 min.). Pellets were dissolved in SO0 pl 50 mM Tris/HCl, pH 8.0, 0.1 mM PhMeSO,F, then dialyzed against 50 mM Tris/HCI, pH 8.0, 0.1 mM PhMe- SO,F, 1 mM EDTA (buffer A). For further purification the fractions were loaded onto a DEAE-cellulose column. The column was washed with two volumes of buffer A and eluted using a linear gradient from 0 to 0.4 M NaCl in buffer A. Fractions showing diaphorase activity were pooled, dialyzed extensively against buffer A and lyophilized. The enzyme was then gel filtrated through a FPLC Supherose-12 column equilibrated in buffer A. Samples were stored at -20°C. For large-scale preparations the above procedure was scaled up

679

proportionally starting with 200 ml liquid culture. Samples were analyzed by SDSPAGE followed by immunoblotting. Extracts (20-80 pl) were denatured by incubation in 50 mM Tris/HCl, pH 8.0, 2 mM EDTA, 0.1 % 2-mercaptoethanol, 10% (by vol.) glycerol and 2% SDS for 3 min in a boiling water bath and separated by SDS/PAGE in 15% gels as de- scribed by Laemmli [27]. After electrophoresis, the samples were electrotransferred to Z-probe membrane (Bio-Rad) and immunoblot analysis was carried out using rabbit antisera raised against spinach ferredoxin-NADP' reductase.

Activity measurements and other procedures Ferredoxin-NADP'-reductase-dependent diaphorase ac-

tivity was determined by means of published methods [7]. Following addition of the ferredoxin-NADP' reductase sam- ple, the reactions were monitored spectrophotometrically following ferricyanide reduction at 420 nm (e420 = 1 mM-' . cm-I). An alternative method was developed for the quanti- tative determination of ferredoxin-NADP' reductase activity. The assay is well suited to the rapid analysis of multiple samples reducing the sample volume to a minimum, and was employed essentially for enzyme detection in liquid culture or during enzyme purification steps. The assay has 40-times more sensitivity to that of ferricyanide and was carried out as follows: 60 pl of sample were mixed with 100 p1 of a medium containing 50 mM Tris/HCl, pH 8.0, 0.8 mM NADP', 0.05% (mass/vol.) nitroblue tetrazolium, 3 mM glu- cose 6-phosphate and 0.2 U glucose-6-phosphate dehydroge- nase. After incubation at 37°C for 10 min, 1 ml N,N-dimeth- ylformamide was added and absorbance recorded at 490 nm ( E ~ ~ ~ = 18.5 mM-' . cm-') [8]. 1 U ferredoxin-NADP' re- ductase activity is defined as the mount of enzyme that cata- lyzed the conversion of 1 pmol substrate/min at 30°C. Ferre- doxin-dependent cytochrome c reductase activity was mea- sured according to Shin [9] using 10 pM spinach ferredoxin plus 50 pM horse heart cytochrome c.

For the determination of the N-terminal sequences of re- combinant ferredoxin-NADP' reductases, 0.05 mg of the purified enzymes was subjected to preparative SDSPAGE and transferred to poly(viny1idene difluoride) membranes [28]. Sequence analysis was then performed using an Ap- plied Biosystems 477/A protein sequencer at LANAIS-PRO laboratories, Consejo Nacional de Investiguciones Cientificas y Tkcnicas and Universidud Nucional de Buenos Aires, Ar- gentina.

For the quantitative determination of the ferredoxin- NADP' reductase activity present in the cytosolic fraction, yeast cells were disrupted using glass beads (0.5-mm diame- ter) as described [26] and cytoplasm was obtained by centrif- ugation (1 5 000 gX 15 min). Intrinsic diaphorase contaminat- ing activities were observed in cytoplasm from yeast cells transformed with the parental plasmid pES. A method was developed to avoid unwanted contaminating on activities (Arakaki, A. Ottado, J. and Ceccarelli, E., unpublished re- sults) thus malung possible the proper determination of the desirable ferredoxin-NADP'-reductase-specific activity. Essentially, the procedure is based on the capture of ferre- doxin-NADP' reductase onto 1 .3-cm2 nylon disks uniformly coated with a specific anti-(ferredoxin-NADP' reductase) antibody. Samples containing the enzyme to be measured were incubated with IgG-coated filters disks during 2 h at room temperature under agitation. Filters were washed five times with 137 mM NaCl, 2.7 mM KC1, 4.3 mM Na,H- PO,,1.4 mM KH2P04, pH 7.3 and kept on ice in the same

buffer until activity was measured. Subsequently, the en- zymic reaction was performed on the filter using nitroblue tetrazolium as described under this section. The standard curves were performed diluting the indicated amount of puri- fied enzymes in 0.3 ml blocking buffer. A direct correlation was observed between the amount of enzyme and the activity measured. No enzyme inhibition was observed with the anti- bodies bound to the filters.

For the determination of kinetic parameters, steady-state kinetic data were fitted to the theoretical curves using Sig- maplot software (Jandel Scientific).

Immunochemical quantitative determination of recombi- nant ferredoxin-NADP' reductase was canied out using a slot-blot apparatus and Z-probe membranes (Bio-Rad). As standard we used purified recombinant ferredoxin-NADP' reductase expressed in E. coli, mixed with the corresponding yeast extract from non-transformed cells in order to apply to each slot a constant amount of S. cerevisiae soluble protein. Anti-(ferredoxin-NADP' reductase) was used as first anti- body. The alkaline-phosphatase-conjugated second antibody as well as substrates were from Promega, and were used ac- cording to manufacturer's instructions. Densitometric scan- ning was carried out using a Shimadzu CS-9000 scanning densitometer.

Protein concentration was determined by a dye-binding assay [29]. Absorption spectra were recorded in a Gilford Response spectrophotometer.

Northern blot analysis

Yeast cells were grown in YPD to a density of 4X107 cells/ml. Cells were harvested, washed once with lysis buffer (0.2 M Tris/HCl, pH 7.5,0.5 M NaC1, 10 mM EDTA, 20 mM aurin tricarboxylic acid), suspended in lysis buffer and total RNA was extracted using phenol/chloroform and glass beads (0.5-mm diameter). RNA was ethanol precipitated at -20°C for 12 h. Pellets were resuspended in 100 p1 10 mM Tris pH 7.8, 1 mM EDTA, supplemented with 1 p1 RNAse-free DNAse (2.5 mg/ml), 100 U RNAsin, 1 pl 1 M MgC12, and incubated for 20 min at 37 "C. RNA was extracted once with phenol/chloroform, ethanol precipitated and resuspended in 20 p1 water. A total of 30 pl total RNA was electrophoresed on a denaturing agarose gel, transferred to a Hybond-N filter and probed with the '2P-labeled Sac1 -EcoRI ferredoxin- NADP' reductase fragment [4].

RESULTS

Construction of the yeast secretion vectors pYMP and pYMF

The construction of vectors used for secretion in yeast of mature pea ferredoxin-NADP' reductase and its precursor is outlined in Fig. 1. Both plasmids contain the complete 5' non-coding region that includes the sequence for a-factor promoter function in vivo, the region coding for the 83- amino-acid signal peptide and the first spacer peptide (amino acids 84-89) of the mating factor a1 (MFnl) gene, fused in frame to the fourth codon of the ferredoxin-NADP' reduc- tase precursor (plasmid pYPF) or to the first codon of the mature ferredoxin-NADP' reductase genes (plasmid pYMF). The 3' flanking regions of the insertions contain the tran-

680

pCV101 4.1 kbp

pCV1 01 0 4.5 kbD

1,305 up

U D - 3.2 kbp

H

y - b g E

HC B SM m t

/ E H H S

-1 - L_ , 1,178 bp 3L up

H

. H

1,230 bp

S

3.9 kbp 0

E E

Fig. 1. Construction of pea ferredoxin-NADP' reductase secretion plasmids pYPF and pYMF. pCVlOl is a pUC19 plasmid with a PstI-EcoRI 1.3-kb pair insert that contains cDNA sequences encoding pea ferredoxin NADP' reductase precursor. pCV105 is a pUC9 plasmid with a SacI-EcoRI 1.2-kbp insert that contains cDNA sequence encoding mature the pea ferredoxin NADP' reductase. The hatched areas within recombinant plasmids represent DNA coding for pea ferredoxin NADP' reductase; the gray areas, sequences coding for yeast a-factor; the stippled areas, the 2 pm FLP gene; the solid areas represent bacterial plasmid regions used as linkers; and the open areas of pES, pYPF and pYMF, yeast plasmids DNA. The arrows inside the plasmids indicate MFal and yeast LEU2 gene approximate transcription origins and direction. H, HindIII; P, PstI; E, EcoRI; HC, HincII; B, BamHI; SM, SmaI; S, SacI.

scription termination and the polyadenylaton signals of the yeast 2-ym FLP gene. The regions fused to codon 89 of the MFal gene are indicated in Fig. 2.

Expression and secretion of the recombinant proteins by S. cerevisiae cells

Plasmids pYPF and pYMF were used to transform S. cer- evisiae cells (strain GRF-18). Secretion of active ferredoxin- NADP' reductase was initially screened on culture plates. Yeast cells from individual clones of the original trans- formants carrying plasmids pYPF or pYMF were grown in selective medium laclung leucine then plated over YPD me- dium. After growth into single colonies, diaphorase activity was detected on the plates using nitroblue tetrazolium as sub- strate as described in Materials and Methods. Out of 300 analyzed colonies transformed with plasmids pYPF or pYMF, 297 (99%) stained deep purple indicating that they were able to secrete active pea ferredoxin-NADP' reductase to the culture medium. A blue-purple halo surrounding se- creting clones was also observed.

The nature of the ferredoxin-NADP' reductases secreted to liquid medium was investigated by SDSPAGE and immu- noblotting (Fig. 3). A major immunoreactive peptide of an apparent molecular mass of 35 kDa was detected in culture supernatants after 72 h growth of yeast cells bearing plasmid pYPF or pYMF. Yeast cells transformed with the parental

plasmid pES showed neither immunoreactive peptides nor diaphorase activity in the culture medium. The apparent mo- lecular masses of the polypeptides were similar to mature ferredoxin-NADP' reductase expressed in E. coli (Fig. 3). Faint bands with higher molecular masses were observed in the Western blots presumably resulting from unprocessed secreted proteins or by glycosylation during secretion [30].

Yeast cells transformed with plasmid encoding the ferre- doxin-NADP' reductase precursor secrete four-times more reductase to the medium than those transformed with the plasmid encoding the mature form of the enzyme. As shown in Table 1, the amounts of enzyme secreted, measured by enzyme activity, or by slot-blot and immunoreaction, were about 50 UA (0.42 mgA) for pYPF and 11 U/1 (0.10 mgA) for pYMF in YPD medium at 30°C. Results obtained by immunodetection and enzyme activity are in agreement, indi- cating that most or all of the secreted proteins are folded into an active conformation, and that unfolded or inactive ferredoxin-NADP' reductases are not secreted or are de- graded during the secretion process. Ferredoxin-NADP' re- ductase secretion was temperature dependent, with an increase of 250% for supernatants of cells grown at 20"C, with respect to yeast cells grown at 30°C. At 37°C a de- fective cell growth was observed and no secretion was de- tected. However, secretion in minimal medium was undetec- table throughout the range of temperatures tested (data not shown).

P W F a-FACTOR PUC 9 pUC 19 MATURE FERREDOXIN-

I I I NADP+ REDUCTASE - - - AAA ADA GAG GCT GAA GCT TGG CTG CAG GTC GGG TAC CGA GCT CAG GTT ACT - - -

K R E A E A W L Q V G Y R A Q V T

84 89 1 2 3

+ 0 0

PYPF a -FACTOR puc 9 PRECURSOR FERREDOXIN-

I I NADP+ REDUCTASE - - - AAA ADA GAG GCT GAA GCT TGC ATG CCT GCA GTA ACA - - -

K R E A E A C M P A V T

84 89 4 5 6

+ U U

Fig. 2. Nucleotide and amino acid sequences at the joining loci between mating factor a1 and ferredoxin-NADP' reductase. Numerals below the amino acids indicate their position with respect to the a-factor initial methionine and the position in the mature and precursor forms of the enzyme. Solid arrows show the processing site recognized by the KEX2 gene product [31] and open arrows those for the STE13 product [21].

Table 1. Expression and secretion of active ferredoxin-NADP reductase by S. cerevisiae. Diaphorase activity was measured using nitroblue tetrazolium as an electron acceptor. Immunochemical determination of recombinant ferredoxin-NADP' reductase were carried out as described in Materials and Methods. Cytoplasm activity (U) and mass (mg) represent quantities of recombinant enzyme detected in yeast cell cytoplasm obtained from 1 1 culture medium. pYPF/pYMF represent ratios between secreted precursor and mature enzyme.

Recombinant Activity in enzyme from -

cytoplasm at 20 "C culture medium at ~

20 "C 30 "C

UA mgll U/I mgfl u/1 mgfi pYPF pYMF

1.6? 0.2 1.3X10-* 120 t 24 1.00+-0.18 50-C 10 0.42 2 0.10 1.3 2 0.4 1.0x 32-C 6 0.26 t 0.05 l l t 4 0.10 5 0.03

pYPF/pYMF 3.75 3.85 4.54 4.2

kDa 1 2 3 4 94 - 67 - 42 -

-FNR 30 - 20 -

Fig. 3. Secretion of ferredoxin-NADP reductase and its precur- sor protein in S. cerevisiae cells. Analysis of proteinaceous compo- nents of yeast culture supernatants secreting ferredoxin-NADP' re- ductase. 1 ml of supernatant was precipitated with 55% saturated ammonium sulfate, dialyzed, lyophilized and resuspended in 100 p1 20 mM Tris/HCl. Samples were applied to a SDS/15% polyacrylam- ide gel, electrophoresed at 20 mA for 3 h and immunoblotted. Lane 1 ,20 pl pES; lane 2,20 p1 p W F ; lane 3,80 pl pYMF; lane 4,0.2 pg purified pea ferredoxin-NADP' reductase expressed in E. coli. FNR indicates the position were ferredoxin-NADP' reductase migrates.

The natural ferredoxin-NADP' reductase transit peptide: may favor secretion by precluding enzyme folding or by fa- cilitating unfolding after translation. This prompted us to in- vestigate whether the mature enzyme folds in the yeast cyto- plasm thus reducing its export to ER. Using rabbit anti-(fer- redoxin-NADP' reductase) IgG we quantified specific dia- phorase activity, avoiding unwanted intrinsic contaminating activities. Soluble active holoenzyme was found in cyto- plasmic extracts of cells transformed by both plasmids pYPF and pYMF in the range 1.3-1.6 UA yeast culture (approxi- mately 12 pg acive enzyme/l). These results were consistent with slot-blot determinations of ferredoxin-NADP' reductase content of the same fractions. No activity was detected in cells transformed with the yeast parent vector pES. The amount of cytoplasmic ferredoxin-NADP' reductase repre- sent 1% (plasmid pYPF) and 4% (plasmid pYMF) with re- spect to the secreted enzyme. The overall results indicate that although there is no remarkable intracellular accumulation of ferredoxin-NADP' reductase, cytoplasmic enzyme activity was undoubtedly observed in both cases.

To investigate wether differences in secretion were due to reasons other than the presence of the transit-peptide se-

682

1 2 3

2 5 s - 18s -

Fig. 4. Northern analysis. Total RNA from yeast cells (30 pg) was analyzed as described in Materials and Methods using a 32P-labeled SacI-EcoRI ferredoxin-NADP' reductase cDNA as a probe. Lane 1, pYPF; lane 2, pYMF and lane 3, pES. The positions of 18s and 25s ribosomal RNA markers are shown on the left.

quence of the pea ferredoxin-NADP' reductase, two dif- ferent experiments were carried out. Plasmid stability was studied by plating yeast cells on selective and non-selective solid media after growth under conditions used for secretion. The percentage of plasmid loss was nearly identical for all of them: pES (42?4%), pYPF (52+5%) and pYMF (48 ?4%), suggesting that this is not the reason for the dif- ferences observed in secretion. Moreover, total cell RNA was prepared from ferredoxin-NADP'-reductase-secreting yeast cells and analyzed, Fig. 4 shows the result of hybridization of the RNA blots with a 32P-labeled ferredoxin-NADP' re- ductase cDNA probe. The size of both transcripts is about 2 kb. The size expected for full transcripts starting in the MFal gene, going through the ferredoxin-NADP' reductase cDNAs, and terminating at the end of the polyadenylation signals of the yeast 2 pm FLP gene are 2.2 kb for pYPF and 2.1 kb for pYMF. The fact that single transcripts of the expected sizes were observed, and at similar levels, indicates that transcription of a full-length RNA occurs with equiva- lent efficiency in both plasmids. In addition, no minor bands were observed accounting for anomalous processing of the RNA.

Purification and properties of the secreted ferredoxin-NADP' reductase

Ferredoxin NADP' reductases secreted by yeast trans- formed with plasmids pYPF and pYMF were purified to near homogeneity as described in Materials and Methods, starting with 200 ml of supernatant. The elution profiles of recombi- nant enzymes in DEAE-cellulose were essentially identical to those of pea ferredoxin-NADP' reductase and of recombi- nant ferredoxin-NADP' reductase expressed in E. coli (data not shown). Purification yields were approximately 15 % of the initial crude supernatants. These enzymes were more than 95% pure, as judged by SDS/PAGE (data not shown). The enzyme's purity was also analyzed measuring the A2751A456 absorbance ratios (Table 2). The observed values (8.1 for pYPF and 8.4 for pYMF) are like that found for crystalline ferredoxin-NADP' reductase 191. These enzymes were able to catalyze the reverse electron-flow reaction from NADPH to a suitable electron acceptor like ferricyanide, 2,6-dichloro- indophenol or nitroblue tetrazolium salts. The kinetic param- eters for diaphorase activity using ferricyanide as acceptor and ferredoxin-dependent cytochrome c reductase activity

were calculated for both recombinant enzymes (Table 2). To further characterize the purified secreted enzymes, N-termi- nal amino acid sequence analysis were performed. The re- sulting sequences are shown in Table 2, indicating N-termini at position - 1 for both proteins with respect to the N-termi- nus of the mature pea enzyme. Our results indicate that pro- cessing of the secreted proteins by yeast takes place near the ferredoxin NADP' reductase natural processing site, as seen in E. coli [15]. The deduced molecular mass for a pea ferre- doxin-NADP ' reductase starting at residue - 1 and ending at its C-terminus is 34.8 kDa. The overall results indicate that the ferredoxin-NADP' reductase precursor or the mature en- zyme expressed in s. cerevisiae is proceseed and that they are able to bind properly the prosthetic group and to acquire an active conformation during their passage through the ER and the Golgi apparatus.

DISCUSSION The results presented demonstrate that the prosthetic

group FAD can be incorporated into a pea apoferredoxin- NADP' reductase when the foreign protein is expressed and secreted by S. cerevisiae. In addition, we report here that secretion is strongly enhanced by the presence of the natural ferredoxin-NADP' reductase transit peptide in the fusion protein. We have constructed two recombinant plasmids con- taining fusions between the yeast a-factor leader region (sig- nal peptide, pro-segment and first spacer peptide) and cDNAs encoding mature pea ferredoxin-NADP' reductase and its precursor protein (Figs 1 and 2). In both cases, trans- formed S. cerevisiae cells secreted a soluble protein with comparable electrophoretic mobility to the product obtained by transgenic expression of pea ferredoxin-NADP' reductase in E. coli ([15] and Fig. 3). The secreted proteins are pro- cessed at amino acid - 1 of the natural mature form, as deter- mined by N-terminal sequencing (Table 2) corresponding to a calculated molecular mass of 34868 Da [4]. The overall results indicate that processing has occurred during or following secretion. Since we have examined polypeptides secreted through the cell wall to the culture medium we can- not eliminate the possibility that processing has occurred in- side the ER or after the protein has reached the outside. The products of genes KEX2 [31] and STE13 [21] have been identified as responsible for a-pheromone maturation (Fig. 2). It has been observed that production of a-pheromone hybrid protein precursors from multicopy plasmids in yeast results in incomplete processing by the dipeptidyl aminopep- tidase A (product of the STE13 gene) 1211. However, in our experiments no unprocessed polypeptides were observed in Western blots. Considering that several sites suitable for pro- tease digestion are found in the N-terminal region of the re- combinant fusion proteins, processing to its mature form could have taken place in the culture medium by vacuolar proteases, perhaps released by an undetectable cell lysis. The proteolytic processing observed in precursor and mature fer- redoxin NADP' reductase fusion proteins expressed in E. coli [15] and secreted in S. cerevisiae (this work) occurs in the same region as in the natural pea enzyme 141, suggesting the presence of a specific recognition site for proteolysis. However, it is more likely that the apparent specific process- ing in the recombinant enzymes is the consequence of a tryp- sin-like digestion over a highly exposed region in the junc- tion between the transit peptide and the mature protein.

The amount of recombinant enzyme secreted to the me- dium, measured by specific diaphorase activity or immunore-

683

Table 2. Characterization of recombinant ferredoxin-NADP' reductases. Potassium ferricyanide reduction was determined using the diaphorase assay of Zanetti [7]. Ferredoxin reduction was determined by cytochrome-c-coupled reaction according to Shin [9]. Absorption spectra were determined at 5-8 FM ferredoxin-NADP' reductase in 20 mM sodium phosphate, pH 7.5. N-terminal sequences were deter- mined as described in Materials and Methods. Wild-type values refer to those obtained from purified pea ferredoxin-NADP' reductase.

Enzyme source K,Fe(CN), reduction Ferredoxin N-terminal AdA456 reduction sequence V P A D P H KZ rn V,,

mM U/mg mol e- . s-' . mol-'

pYPF 140 L 12 8 2 2 228? 12 38.2 pYMF 136 2 10 9 2 3 237 2 10 36.5 Wild type 1035 4 1 4 t 2 2462 8 40.5

AQVTTEAXAKV 8.1 AQVTTEAXAKV 8.4 QVTTEAPAKV" 8.5

a From Newman, B. J. and Gray, J. C. [4].

action, is about l.OOmg/l (pYPF at 20°C) and 0.26 mg/l (pYMF at 20"C), (Table 1). In both cases, secretion was re- duced by growing yeast cells at 30°C (0.42 mg/l for pYPF and 0.10 mg/ml for pYMF). However, the secretion ratio pYPF/pYMF was about 4 at both temperatures tested. Plas- mid loss was nearly identical for the pES parental plasmid and for the recombinant pYPF and pYMF during the secre- tion experiment. In addition, similar levels of transcripts pro- duced by both plasmids were detected, and no minor bands were observed accounting for anomalous processing of the RNA (Fig. 4).

Most of the secretory proteins are translocated during their translation to the rough endoplasmic reticulum [32]. However, different experimental evidence has shown that the yeast prepro-a-factor, as with other precursors, crosses the ER membrane after synthesis is completed [30, 331. In chlo- roplasts, no ribosomes attached to the outer membrane have been found [34]. During its import to chloroplast, ferredoxin- NADP' reductase may be using a similar mechanism as the yeast a-factor during its transport to ER. To make the trans- port possible the already synthesized protein has to be un- folded, or in a loosely folded conformation. Protein translo- cation has been shown to involve cytoplasmic protein factors homologues of the highly conserved 70-kDa heat-shock pro- tein (Hsp7O) family in ER, in the mitochondrion and the chloroplast [23]. Our results clearly demonstrate higher lev- els of ferredoxin-NADP' reductase secretion by yeast cells transformed with pYPF in comparison with pYMF, and that this phenomenon is related to the presence of the ferredoxin- NADP' reductase transit-peptide region in the fusion protein. One possible hypothesis is that the ferredoxin-NADP' reduc- tase precursor is more easily maintained in an unfolded or loosely folded conformation than the mature protein or that the fusion protein containing the mature ferredoxin-NADP' reductase can stably fold in the cytoplasm and subsequently affect its translocation to the ER. In this context, it has been shown that the bovine flavoprotein NADPH-adrenodoxin re- ductase can be actively expressed in the yeast cytoplasm [35]. Moreover, it has been observed that the chloroplast 5- enolpyruvylshikimate-3-phosphate synthase precursor has the ability to function catalytically as an enzyme and to bind the competitive inhibitor glyphosate with the same affinity as the mature enzyme. The rate of import of this precursor into chloroplast in the presence of glyphosate is significantly reduced, probably because the complex adopts a conforma- tionally rigid state which is less likely to be unfolded during import [36]. Wienhues et al. [37] have investigated the im- port to yeast mitochondria of a fusion protein containing the

N-terminal third of the precursor to cytochrome b, and of the entire dihydrofolate reductase. The hybrid protein was found to be as efficiently imported and processed as cytochrome b,. When yeast cells expressing b,-dihydrofolate reductase received aminopterin, a folate analogue known to stabilize the tertiary structure of dihydrofolate reductase, the fusion protein was arrested in the import pathway. These observa- tions demonstrate the post-translational character of the pro- tein import to mitochondria in yeast and the requirement for precursor unfolding. This unfolding may be carried out by the cytoplasmic Hsp70, as suggested by experiments with the E. coli Hsp70 homologue DnaK which has shown unfoldase functions [38]. Natural optimized evolution of the transit peptide may account for capabilities of correct identification of the import machinery and facilitation of protein unfolding.

The kinetic parameters of the secreted ferredoxin- NADP' reductase establish that the protein has incorporated a functional prosthetic group FAD and that regions involved in the interaction of substrates and in electron transfer are folded as in the native enzyme. Moreover, the values ob- served for cytochrome c reductase activity indicate that the recombinant enzymes are able to bind ferredoxin and that they are correctly folded. The mechanism of assembly of the prosthetic group FAD in flavoproteins is not currently understood. We have recently demonstrated that in E. coli, assembly of pea ferredoxin-NADP' reductase requires GroE (Hsp60-Hsp10) molecular chaperones [17]. Mutants used in those experiments were only chaperonin deficient and the Hsp70 homologue DnaK was not affected. It is not known whether other proteins directly participate in the assembly of FAD or in the recognition of acceptor apoproteins. In other cell compartments, but not in the ER, different chaperones have been identified directly involved in protein folding, i.e. Hsp60-Hsp10 in chloroplast and mitochondria and TPC-1- like chaperones in cytoplasm [23]. No Hsp60-Hsp10 have been identified in the ER. The yeast ER-localized Hsp70, called Kar2p, (BiP in mammalian cells) is needed for the transport of proteins across the membrane acting through a direct interaction with the protein being transported [24, 391. The roles that are emerging for cytosolic, mitochondrial, chloroplast and ER Hsp70 are similar. They can assist protein translocation across membranes and are involved in the path- way of protein folding by other chaperones [23]. However, it remains unclear whether or not Hsp70 alone are capable of properly folding a protein. Our present data demonstrate that the ER is capable of properly folding a protein that in other systems is folded with Hsp60-Hsp10 assistance. One possibility is that an unidentified Hsp60-Hsp10 homologue

684

is present inside the ER. Another possibility is that Kar2p and/or Sec63p (which have a DnaJ-like domain) [40] and/or other proteins could be acting as a refolding device for the incoming polypetide, as was previously proposed [24]. An- other interesting implication of our results is that FAD is available inside the yeast ER. We cannot discriminate at this point whether the nucleotide is already present in the ER or if there is a co-transport with the flavoprotein.

We have previously expressed pea ferredoxin-NADP+ re- ductase E. coli showing that the folding and assembly was Hsp60-Hsp10 dependent [15 - 171. We have now expressed and secreted ferredoxin-NADP' reductase in yeast. The re- sults presented here, in which FAD is correctly incorporated into pea ferredoxin-NADP' reductase in the ER indicate that this process may be common to all organisms and probably equivalent in all cellular compartments. Further studies on the folding and assembly of ferredoxin-NADP' reductase are currently in progress.

We are grateful to Dr NCstor Carrillo for many helpful sugges- tions and critical reading of the manuscript. We thank Dr John C. Gray (University of Cambridge, England) for the generous gift of the original pea cDNA clone for ferredoxin NADP' reductase, Dr Ana Clara Schenberg Frascino (Universidade de Sa'o Paulo) for pro- viding the plasmid pES and the yeast strain GRF-18, and Dr Mirtha Biscoglio and Mario Feldman (Universidad de Buenos Aires) for their generous help in the determination of N-terminal sequences. Supported by Grants from Fundacidn Antorchas (Buenos Aires, Ar- gentina). EAC and NBC are staff members of Consejo Nacional de lnvestigaciones Cientficas y Tkcnicas.

REFERENCES 1. Carrillo, N. & Vallejos, R. H. (1987) Ferredoxin-NADP' oxi-

doreductase, in The ligth reactions. Topics in photosynthesis (Barber, J., ed.) vol. 8, pp. 527-560, Elsevier, Amsterdam, New York, Oxford.

2. Shin, M. & Amon, D. I. (1965) Enzymic mechanisms of pyri- dine nucleotide reduction in chloroplasts, J Biol. Chem. 240,

3. Jansen, T., Reilaender, H., Steppuhn, J. & Herrmann, R. G. (1988) Analysis of cDNa clones encoding the entire precur- sor-polypeptide for ferredoxin-NADP ' reductase from spin- ach, Curl: Genet. 13, 517-522.

4. Newman, B. J. & Gray, J. C. (1988) Characterisation of a full- length cDNA clone for pea ferredoxin-NADP' reductase, Plant Mol. Biol. 10, 511 -520.

5. Grossman, A. R., Bartlett, S. G., Schmidt, G. W., Mullet, J. E. & Chua, N.-H. (1 982) Optimal conditions for post-translational uptake of proteins by isolated chloroplasts, J. Bid. Chem. 257,1558-1563.

6. Vallejos, R. H., Ceccarelli, E. A. & Chan, R. L. (1984) Evidence for the existence of a thylakoid intrinsic protein that binds ferredoxin-NADP' oxidoreductase, J. Biol. Chem. 259, 8048 - 8051.

7. Zanetti, G. (1976) A lysyl residue at the NADP' binding site of ferredoxin-NADP' reductase, Biochim. Biophys. Acta 445,

8. Zanetti, G. (1981) The reduction of iodonitrotetrazolium chlo- ride by ferredoxin-NADP' reductase: A new tool for the char- acterization of the spinach chloroplast flavoprotein, Plant Sci. Lett. 23, 55-61.

9. Shin, M. (1 971) Ferredoxin-NADP' reductase, Methods Enzy- mol. 23, 440-447.

10. Yao, Y., Tamura, T., Wada, W., Matsubara, H. & Kodo, K. (1984) Spirulina ferredoxin-NADP' oxidoreductase. The complete amino acid sequence, JBiochem. (Tokyo) 95,1513- 1516.

11. Michalowski, 0. B., Schmitt, J. M. & Bohnert, H.-J. (1989) Expression during salt stress and nucleotide sequenee of

1405 - 141 1.

14-24.

cDNA for ferredoxin-NADP' reductase from Mesembryan- themum crystallium, Plant Physiol. (Bethesda) 89, 817- 822.

12. Fillat, M. F., Bakker, H. A. C. & Weisbeek, P. J. (1990) Se- quence of the ferredoxin-NADP'-reductase gene from Ana- baena PCC7119, Nucleic Acids Res. 18, 7161.

13. Schluchter, W. M. & Bryant, D. A. (1992) Molecular charac- terization of ferredoxin-NADP' oxidoreductase in Cyano- bacteria. Cloning and sequence of the petH gene of Synecho- coccus sp PCC-7002 and studies on the gene product, Bio- chemistry 31, 3092-3102.

14. Aliverti, A., Jansen, T., Zanetti, G., Ronchi, S., Herrmann, R. G. & Curti, B. (1990) Expression in Escherichia coli of ferre- doxin-NADP' reductase from spinach. Bacterial synthesis of the holoflavoprotein and of an active enzyme form lacking the first 28 amino acid residues of the sequence, Eur: J Bio- chem. 191, 551-555.

15. Ceccarelli, E. 4., Viale, A. M., Krapp, A. R. & Carrillo, N. (1991) Expression, assembly, and processing of an active plant ferredoxin-NADP' oxidoreductase and its precursor in Escherichia coli, J. Biol. Chem. 266, 14283-14287.

16. Serra, E. C., Carrillo, N., Krapp, A. R. & Ceccarelli, E. A. (1 993) One-step purification of plant ferredoxin-NADP' oxi- doreductase expressed in Escherichia coli as fusion with glu- tathione S-transferase, Protein Express Pur$ 4, 539-546.

17. Carrillo, N., Ceccarelli, E. A,, Krapp, A. R., Boggio, S., Fer- reyra, R. G. & Viale, A. M. (1992) Assembly of plant ferre- doxin NADP' oxidoreductase in Escherichia coli requires GroE molecular chaperones, J. Bid. Chem. 267, 15537- 15541.

18. Tsugeki, R. & Nishimura, M. (1993) Interaction of homologous of Hsp70 and Cpn60 with ferredoxin-NADP' reductase upon its import into chloroplasts, FEBS Lett. 320, 198-202.

19. Kurjan, J. & Herskowitz, I. (1982) Structure of a yeast phero- mone gene (MFa): A putative a-factor precursor contains four tandem copies of mature a-factor, Cell 30, 933-943.

20. Brake, A. J. (2990) a-Factor leader-directed secretion of heterol- ogous proteins from yeast, Methods Enzymol. 185, 408-421.

21. Bitter, G. A,, Chen, K. K., Banks, A. R. & Lai, P. (1984) Secre- tion of foreign proteins from Saccharomyces cerevisiae di- rected by a-factor gene fusions, Proc. Natl Acad. Sci. USA 81, 5330-5334.

22. Astolfi Filho, S., Galembeck, E. V., Faria, J. B. & Schenberg Frascino, A. C. (1986) Stable yeast transformants that secrete functional a-amylase encoded by cloned mouse pancreatic cDNA, Bio-technology (NY) 4, 311 -315.

23. Craig, E. A., Gambill, B. D. & Nelson, R. J. (1993) Heat shock proteins : Molecular chaperones of protein biogenesis, Micro- biol. Rev. 57, 402-414.

24. Larriba, G. (1993) Translocation of proteins across the mem- brane of the endoplasmic reticulum: A place for Saccharo- myces cerevisiae, Yeast 9, 441 -463.

25. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1989) Molecular cloning: a laboratory manual, 2nd edn, Cold Spring Harbor Laboratory, Cold Spring Harbor NY.

26. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seid- man, J. G., Smith, J. A. & Struhl, K. (1987) Currents proto- cols in molecular biology, John Wiley & Sons, New York.

27. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the head of the bacteriophage T4, Nature 227, 680-685.

28. Matsudaira, P. (1987) Sequence from picomole quantities of proteins electroblotted onto polyvinylidene difluoride mem- branes, J. Biol. Chem. 262, 10035-10038.

29. Sedmak, J. & Grossberg, S. (1977) A rapid, sensitive and versa- tile assay for protein using Coomassie brilliant blue (3-250, Anal. Biochem. 79, 544-552.

30. Rothblatt, J. A. & Meyer, D. 1. (1986) Secretion in yeast: translocation and glycosylation of prepro-a-factor in vitro can occur via an ATP-dependant post-translational mechanism.

31. Brennan, S. O., Peach, R. J. & Bathurst, I. C. (1990) Specifity of yeast kex2 protease for variant human proalbumins is iden-

EMBO J. 5, 1031-1036.

685

tical to the in vivo specifity of the hepatic proalbumin conver- tase, J. Biol. Chem. 265, 21 494-21 497.

32. Walter, P., Gilmore, R. & Blobel, G. (1984) Protein transloca- tion across the endoplasmic reticulum, Cell 38, 5-8.

33. Waters, M. G. & Blobel, G. (1986) Secretory protein transloca- tion in a yeast cell-free system can occur post-translationally and requires ATP hydrolysis, J. Cell Biol. 102, 1543-1550.

34. Chua, N.-H. & Schmidt, G. W. (1978) Post-translational trans- port into intact chloroplasts of a precursor to the small subunit of ribulose 1 ,5-biphosphate carboxylase, Proc. Nut1 Acad. Sci.

35. Akiyoshi-Shibata, M., Sakaki, T., Yabusaki, Y., Murakami, H. & Ohkawa, H. (1991) Expression of bovine adrenoxin and NADPH-adrenoxin reductase cDNAs in Succharomyces cere- visiae, DNA Cell Biol. 8, 613-621.

36. della-Cioppa, G. & Kishore, G. (1988) Import of a precursor protein into chloroplasts is inhibited by the herbicide glypho- sate. EMBO J. 7, 1299-1305.

USA 75, 6110-6114.

37. Wienhues, U., Becker, K., Schleyer, M., Guiard, B., Tropschug, M., Horwich, A. & Neupert, W. (1991) Protein folding causes an arrest of preprotein translocation into mitochondria in vivo, J. Cell Biol. 115, 1601-1609.

38. Skowyra, D., Georgopoulos, C. & Zylicz, M. (1990) The E. coli dnaK gene product, the hsp70 homolog, can reactivate heat- inactivated RNA polymerase in an ATP-dependent manner, Cell 62, 939-944.

39. Sanders, S. L., Whitfield, K. M., Vogel, J. P, Rose, M. D. & Schekman, R. W. (1992) Sec61p and Bip directly facilitate polypeptide translocation into the ER, Cell 69, 353-365.

40. Sadler, I., Chiang, A., Kurihara, T., Rothblatt, J., Way, J. & Silver, P. (1989) A yeast gene important for protein assembly into the endoplasmic reticulum and the nucleus has homology to DnaJ, an Escherichia coli heat shock protein, J. Cell Biol. 109, 2665-2675.