© 1993 Oxford University Press Nucleic Acids Research, 1993, Vol. 21, No. 3 635-639

Erythromycin and 5S rRNA binding properties of thespinach chloroplast ribosomal protein CL22

Pierre Carol, Claude Rozier, Ester Lazaro1, Juan P.G.Ballesta1 and Regis Mache*Laboratoire de Biologie Moleculaire Vegetale, U.J.Fourier BP 53X, 38041 Grenoble Cedex, Franceand 1Centro de Biologia Molecular 'Severo Ochoa' (CSIC-UAM), Canto Blanco, 28049 Madrid, Spain

Received September 25, 1992; Revised and Accepted December 18, 1992

ABSTRACT

The spinach chloroplast ribosomal protein (r-proteln)CL22 contains a central region homologous to theEscherichla coll r-proteln L22 plus long N- and C-terminal extensions. We show In this study that theCL22 combines two properties which In E.coll ribosomeare split between two separate proteins. The CL22which binds to the 5S rRNA can also be linked to anerythromycin derivative added to the 50S ribosomalsubunit. This latter property Is similar to that of theE.coll L22 and suggests a similar localization In the 50Ssubunit. We have overproduced the r-protein CL22 anddeleted forms of this protein in E.coll. We show thatthe overproduced CL22 binds to the chloroplast 5SrRNA and that the deleted protein containing the N- andC-termlnal extensions only has lost the 5S rRNAbinding property. We suggest that the centralhomologous regions of the CL22 contains the RNAbinding domain.

INTRODUCTION

In its overall structure and function the chloroplast ribosomeresembles the procaryotic ribosome. It is composed of 30S and50S subunits containing four rRNA species highly homologousto their Escherichia coli counterparts and approximately 55ribosomal proteins (r-proteins). About one third of the chloroplastr-proteins are encoded by the chloroplast genome (1, 2). Theremaining two thirds are encoded by the nucleus, synthesizedin the cytosol and transported to the chloroplasts. Most of thechloroplast r-proteins are homologous to the E.coli r-proteins.However, some of them have a reshaped primary structure,possess peptidic extensions that can increase by up to 100% thesize of the protein (3) or are drastically shortened, as observedfor the r-protein CS-S1 (4). There also exists a class of'chloroplast-specific' r-proteins which do not possess homologywith any of the bacterial r-proteins (5—8). The consequences ofthese reshaped or chloroplast-specific r-proteins on the overallproperties and function of the chloroplast ribosome are notknown, although the shortened CS1 r-protein and the E.coli SIplay a similar role in the initiation of translation (4).

To better understand the possible changes in the chloroplastr-protein's structure and function, we have studied the propertiesof the spinach chloroplast r-protein CL22. This protein is encodedby the chloroplast genome (9), except in pea and probably inother plants of the same family where the gene is located in thenucleus (10). This r-protein is an example of a reshaped proteinthat possesses large peptidic extensions compared to the E.colir-protein L22. Previous studies have shown that unlike itsbacterial homologue, the CL22 is a 5S rRNA binding protein(9, 11).

The E.coli 122 does not bind 5S rRNA but is an early assemblyprotein present in the core particle of the 50S subunit (12). L22binds to erythromycin (13) and is one of the polypeptides thatare affected by mutations conferring resistance to erythromycin(14). As a consequence of its procaryotic origin, the chloroplasttranslational apparatus exhibits a sensitivity to antibiotics similarto that of the bacteria (15). We therefore address the question,does the spinach CL22 also bind to erythromycin?

We demonstrate in this paper that the spinach CL22 binds toa labelled derivative of erythromycin, suggesting that this proteinis located nearby the peptidyltransferase center like its bacterialhomologue (13). We also wanted to localize the domainresponsible for the specific 5S rRNA binding property withinthe CL22 sequence. We overproduced the CL22 in E.coli andtested in vitro its ability to bind chloroplast 5S rRNA after deletionof the central part of the protein and of the C-terminus peptidicextension. The results obtained suggest that the rRNA bindingproperty is located in the central part of the protein.

MATERIALS AND METHODS

Photo-labelling of ribosomal subunits

Chloroplast ribosomal subunits were prepared from youngspinach leaves (Spinacia oleracea, var. GSant d'hiver) (16). Thesubunits were pelleted by ultra-centrifugation and resuspendedin a buffer containing 20 mM Tris-HCl, pH 7.6, 10 mMMgCl2> 100 mM NH4CI, 6 mM /3-mercaptoethanol.

The photoreactive erythromycin derivative (TED) used in theseexperiments was synthetised as previously indicated and waslabelled with 123Iodine (17).

* To whom correspondence should be addressed

636 Nucleic Acids Research, 1993, Vol. 21, No. 3

Ribosomal subunits were dialysed against 10 mM borate, pH7.5, 100 mM KC1, 10 mM MgCl2 at 4°C and were incubatedin the dark with IED concentration of 0.1 to 1.5 /tM, at 37°Cfor 20 min. The samples were cooled to 0°C, placed in aborosilicate glass tube in a refrigerated bath at 4°C and irradiatedfrom a distance of 5 cm with a 125 W medium pressure mercurylamp. Ribosomal subunits treated in the same way but notirradiated were used as control.

The labelled proteins were fractionated through a 12.5% PAGEgel in the presence of SDS (sodium Dodecyl Sulfate) (18) andthen visualized by autoradiography. Alternatively, the proteinswere fractionated by reverse phase HPLC (19) using theconditions suitable for chloroplast r-proteins (20).

The amount of IED non-covalently bound to the 50S ribosomalsubunits was determined by measuring the radioactivity in thepellet after ultra-centrifugation of the IED-treated subunits. Thecovalently bound IED was determined as follows: irradiated 50Ssubunits were precipitated with acetone, filtered through a glassfiber (GF/C, Whatman), and rinsed with 80% acetone. Theradioactivity retained on the filter was measured using ascintillation counter.

Cloning of the rpl22 gene and deleted genesThe spinach chloroplast rpl 22 gene coding for the chloroplastr-protein CL22 was cloned from the SalI-9 spinach chloroplastDNA fragment (9). Standard methods (21) were used for theconstruction of the following plasmids:— pUC-L22 and pBL22 plasmids: the 1507 bp HindHI-EcoRIfragment containing the entire rpl 22 sequence was inserted intopUC18 and Bluescript KSII, resulting in the plasmids pUC-L22and pBL22, respectively (Fig. 2).— pETl 1-L22 plasmid: a Hpall fragment containing the rpl 22gene and a portion of the rps3 gene was inserted at the BamHIsite of the pETl 1 vector (22). Protruding ends were filled in andblunt end ligated. The correct orientation of the rptll gene,respective to the transcription start was checked by restrictionmapping.— pETl 1-AHC plasmid. pUC-L22 was digested by EcoRV andXhol, ends were partially filled-in and the remaining singlestrands were digested by the action of SI nuclease. The plasmidwas then ligated. Reading frame was checked by double strandsequencing using the dideoxy chain termination method (21) anda T7 DNA polymerase sequencing kit (Pharmacia). The Hpall-Sau3A fragment containing the AHC coding sequence was theninserted at the BamHI site of pETll as indicated above.

C-terminal deletions: the 3' end of the HindTfl-EcoRI fragmentcontaining the rptll gene was progressively digested byexonuclease IQ and nuclease SI. The fragments were theninserted separately into Bluescript KSII plasmids. The size ofthe deletion was analysed by sequencing. Clones deleted withnucleotides 10 to 250 in the rp[22 coding sequence were selected.The plasmid with a 3 '-terminal deletion of 28 codons in the CL22coding frame overproduced a deleted protein which wasdesignated AC30.

Overproduction of the r-protein CL22 and mutantsPlasmids pETll-L22 and pETl 1-AHC carrying the CL22 andAHC coding sequences respectively were introduced into theE.coli strain BL21(DE3) (22). The constructs inserted into theBluescript plasmids were used for overproduction in the E.colistrain XLl-Blue. Bacteria were grown in liquid selective mediumto ODgooon, = 0.5. and overproduction was induced for 4 h by

the addition of IPTG (Isopropyl-/3-D-thiogalactoside) up to 1 mM.Bacteria were harvested, washed at 4°C with the extraction buffer(20 mM Tris HC1, pH 7.6, 50 mM KC1, 1 mM EDTA, 10%v/v glycerol, 14 mM /3-mercaptoethanol, 0.05 mM PMSF(Phenylmethylsulphonyl fluoride) and disrupted by sonication.The overproduced proteins were in the pellet obtained bycentrifugation at 10 000Xg. The pellet was washed with 2 Murea and the aggregated overproduced proteins were solubilizedin the extraction buffer containing 8 M urea. After centrifugation,the solubilized proteins contained in the supernatant wereprecipitated by 10% TCA. The precipitate was pelleted andredissolved in electrophoresis buffer. After electrophoresis(Fig. 3A) the proteins were blotted onto a nitrocellulose sheetfor binding experiments or individually electroeluted for dot blotassays.

Semi-purified proteins were used for UV cross-linkingexperiments (Fig. 4). These were produced by solubilisingaggregates of overproduced protein in extraction buffer plus 1.5%N-Lauryl Sarcosinate (Sigma), then dialised against 1L ofextraction buffer (23).

Proteins were quantified by the method of Bradford (24).

5S rRNA-protein bindingChloroplast 5S rRNA was synthesized in vitro using the 5S rRNAgene cloned in pGEM linearized at its HindTTI site (Rozier,unpublished results), T7 RNA polymerase from a RNAtranscription kit (Stratagene) and [a32P]UTP. The extrasequences of the 5S rRNA are of 7 and 16 nucleotides at the5'- and 3'-end, respectively. The synthesized 5S rRNA waspurified by electrophoresis and renatured for 5 min at 50°C inthe presence of 10 mM Mg2+ before cooling at roomtemperature.

For dot blots, the purified proteins eluted from the gel (seeabove) were blotted onto a nitrocellulose filter, renatured in thebinding buffer I (20 mM Tris-HCl, pH 7.6, 100 mM KC1, 0.1mg/ml Bovine Serum Albumin), and incubated with radioactive5S rRNA (50 000 cpm/ml) in the same buffer as previouslydescribed (11, 25). These conditions allow a specific recognitionof 5S rRNA binding proteins (11, 25).

For the UV cross-Unking experiment, proteins were incubatedin the presence of radioactive chloroplast 5S rRNA (106 cpm)in 50 /tl of binding buffer H (20 mM Tris-HCl, pH 7.6, 10mM MgCl?, 100 mM NH4CI) for 10 min at 37°C in the wellof a microtitration plate. The sample was then irradiated at 4°C,2 cm beneath a UV lamp at 254 nm for 15 min. After irradiation,the mixture was then treated with 1/jg of RNAase A for 30 minat 37°C. The proteins were collected by precipitation with acetoneand were fractionated by SDS-PAGE. The radioactive proteinswere visualized by autoradiography.

RESULTSIdentification of the r-proteins labelled by an erythromycinderivativeCovalent and non-covalent binding of the iodinated erythromycinderivative (IED) to SOS chloroplast ribosomal subunits wasdetermined. The specificity of the binding was verified bycompetition with cold erythromycin, as shown on Fig. 1A. Weconclude that the IED binding is due to the interaction of theerythromycin with the 50S subunits and that the UV treatmentused for covalent binding does not significantly alter the bindingproperties of the complex. This latter conclusion allows the search

Nucleic Acids Research, 1993, Vol. 21, No. 3 637

o

Su

120 -,

100

60 .

40 -

20 -

0

Non covaJent bindingCovalent binding

Erythromyon concentration (|iM)

Stained Gel Autoradiography

kDa

43 —

20 —

14 —

Autoradiography Stained Gel

fa-.

uv

kDa

— 6/

— (0

— SO

—14

Figure 1. Identification of the 50S subunit chloroplast r-proteins labelled by theradioactive pbocoactivable iodinated erythromycin denvative (JED). A. Competitionfor the binding of the IED to the SOS subunh. Increasing amount of colderythromycin was added to the IED (ljiM) in the presence of chloroplast SOSsubunit. The concentration of the cold erythromycin is indicated in micromolar.The quantity of retained radioactivity is expressed in percent of the value obtainedwhen no cold erythromycin is added (10-12 000 cpm). B. Proteins labelled bythe IED were separated by SDS-PAGE after binding of the ribosomal 50S subunitparticles with IED, then irradiation (+) or not ( - ) with UV. Coomassie bluestaining (left) and autoradiography (right) of the gd are shown. C. Hectrophoreticseparation of HPLC-fractions containing radioactive r-proteins. Coomassie bluestaining (right) and autoradiography (left) are shown. The triangles indicate theposition of CL22 (black) and CS-L21 (white). Molecular size of markerpolypeptides are reported on the side of stained gels.

for the 50S r-proteins which are in close contact with theantibiotic, after stabilizing the interaction by the UV treatment.

When r-proteins were isolated from the labelled 50S subunitsand fractionated by SDS-PAGE, two labelled bands wereobserved (Figure IB). The major band corresponds to a proteinof 22-25 kDa. A second protein of 15-18 kDa is also foundwith a lower level of radioactivity. For identification purposes,the labelled proteins, the proteins extracted from irradiated 50Sribosomal subunits were fractionated by HPLC and each fractionsubjected to electrophoresis. Here again, two proteins of 23 kDaand 18 kDa respectively were found (Fig. 1C). These werenamed according to the system used in our laboratory (16, 20),the 23 kDA protein corresponds to CL22, the product of the rpl22 gene (9, 11), and the 18 kDa protein is named CS-L21, itsgene is as yet known.

The peptidic extensions of the spinach r-protein CL22 alone,do not bind to the chloroplast 5S rRNAIn addition to the erythromycin binding property which is similarfor chloroplast and E.coli L22 r-proteins, spinach CL22 is a 5SrRNA binding protein. We wanted to know if this 5S rRNAbinding property was correlated to the presence of the largepeptidic extensions which double the size of the spinach CL22compared to its E.coli protein homologue (Fig. 2).

Using the previously characterised chloroplast encoded rpU2gene (9) several constructs were made in order to overproduceall or part of the CL22 r-protein in E.coli (Fig. 2). One of theconstructs allowed the production of a protein termed AHC witha deletion in its central region homologous to E.coli L22. Thisprotein contains the two fused N- and C-terminal extensions,resulting in an 101 amino acid residue polypeptide. Deletion ofthe 3' end of the rptll gene gave overproduction of a proteinmissing 30 amino acids in the C-terminal end (AC30). Constructswith extended deletions at the C- or N-terminus have also beentested for overproduction but no overproduced protein wasobtained (not shown).

The overproduction of proteins in E. coli was verified by invivo labelling with 35S-Methionine and SDS-PAGE (not shown).Fig. 3 A shows an electrophoretic analysis of overproduced CL22and AHC proteins obtained after solubilization of the aggregateswith 8M urea (see Materials and Methods). The electroelutedproteins were incubated with ^P-labelled chloroplast 5S rRNA.Fig. 3B shows that the CL22 protein only binds to the 5S rRNA.No radioactive 5S rRNA is bound to the AHC protein.

The binding properties of CL22 were verified by UV cross-linking of the 5S rRNA to overproduced protein (Fig. 4). Weshowed that AC30 is still able to bind to the 5S rRNA (Fig. 4).From these experiments we conclude that the peptidic extensionsalone are not responsible for the 5S rRNA binding property ofthe r-protein CL22.

DISCUSSION

Erythromycin is an antibiotic known to block translationalelongation in the procaryotic ribosome of E. coli by interactingwith essential components present in the peptidyl transferasecenter, but does not inhibit the peptidyl transferase reaction itself(26). In E.coli, erythromycin resistance can result from mutationsin three genes, two of which are known to affect the r-proteinsL4 and L22 and the third affects rRNA processing (27).Erythromycin derivatives have been shown to interact with L22(13). These results suggest that the L22 participates somehowin the peptidyl transferase center.

In the spinach chloroplast ribosome the CL22 homologue tothe E.coli 122, is a 5S rRNA binding protein (9, 11). Since the5S rRNA is located in the central protuberance of the 50Sribosomal subunit (28), it raises the question of the location ofCL22 in the chloroplast 50S subunit and of its participation inthe peptidyl transferase activity. Results reported in this papershow that the CL22 is labelled by an erythromycin derivative,suggesting the presence of an erythromycin binding site in theCL22 and possibly a localization in the ribosomal subunit similarto its bacterial homologue in the vicinity of the peptidyl transferasecenter (13). The CL22 contains a central region of equal lengthand with homology to the E. coli L22. We suggest that this regionis responsible for the erythromycin derivative binding andpossibly contributes to the architecture of the peptidyl transferasecenter.

638 Nucleic Acids Research, 1993, Vol. 21, No. 3

A.Sail Hindi11 Hpall .

1 1 ^L^4s$\/)?s 19 ^ N ( — I

I

CL22 ]

AHC |

AC 30

Tr EcoRV

^N HC

[291 99

s

Xhoi

1• ^

B58

sy

s|29ll7|58fe

| 2 9 | 99 | 2 8 |

«Sau3A HpallEcoRi Sail

Figure 2. Schematic representation of the overcxpressed coding sequences. A.The spinach chloroplast DNA fragment Sal 1-9 used for the constructions. Thecoding sequence of the genes present on this fragment are shown. Only usedrestrictions sites are shown. B. Proteins derived from the chloroplast rpl 22 codingsequence and overproduced in E.coli. N, HC and C represent the N-terminal,the central homologous and C-terminal domains respectively. The number ofaminoacid residues of each domain is indicated.

.1 M

We have shown here that in contrast to the E. coli 50S subunita second r-protein, named CS-L21, is labelled by theerythromycin derivative used in this experiment. The primarystructure of the CS-L21 protein (named according to its size,ref. 16) is not known. It would be interesting to check if CS-L21is homologous to the E.coli L4, which is altered in a mutantresistant to erythromycin (27), or to the E.coli L15 or L27 whichcan be labelled by derivatives of the macrolides group ofantibiotics, of which erythromycin belongs (29). Interestingly,in the green alga Chlamydomonas reinhartii there is an alteredchloroplast r-protein CL6, that is nuclear encoded and isassociated with resistance to erythromycin (30). It is thereforepossible that the C. reinhartii CL6 protein is equivalent to thespinach CS-L21 protein.

We have also studied the localization of the 5S rRNA bindingproperty of the CL22. Especially, we searched for the possiblerole of the large peptidic extensions as RNA binding domain.These extensions were good candidates for this this we speculatedthat the RNA binding domain could be separated from theerythromycin-binding domain as is the case in the E.coli 5OSsubunit. We have overexpressed in E.coli CL22 with a centraldeletion. The resulting protein is unable to bind to the 5S rRNA.Furthermore, the deletion of the last 30 amino acids of the C-tenninal extension does not affect RNA binding. No sequenceor structual similarity between the peptidic extensions and the80 amino acids RRM (RNA Recognition Motif) or any otherknown RNA-binding domain (31) could be detected (not shown).Therefore, we suggest that the central part of the protein isnecessary for the RNA binding and that the peptidic extensionsare not independant RNA-binding domains. The deletion of thecentral domain could affect the tertiary folding of the protein andprevent the RNA-binding function of the peptidic extensions. Itcould also be possible that the peptidic extensions of CL22participate cooperatively in RNA binding to the central part ofthe protein. In both cases the 30 C-terminal amino acids wouldnot be necessary. Nevertheless, we speculate that the RNAbinding domain of the CL22 is located in the central parthomologous to E. coli L22, in domains specific to the chloroplastprotein. The CL22 central part has 35 % amino acid identity withits E.coli homologue. It does not possess homology to any oneof the E.coli 5S rRNA binding proteins (the L5, L18 and L25).

Figure 3.5S rRNA binding property of the overproduced CL22 and AHC proteins.A. Coomassie blue staining of the aggregated proteins electrophoresed aftersolubilization with 8M urea (see Materials and Methods). Lane 1: AHC, whitetriangle; lane 2: CL22, black triangle. B. Proteins were eluted from the gel, blottedon a nitrocellulose filter then incubated with radioactive 5S rRNA. Lanel,overproduced AHC; lane 2, over produced CL22; lane 3, HPLC purified CL22from chloroplast ribosomes.

kDa

43 —

30 —

CL 2?

?0 —

14 —

kDa

43 —

30 —

?0 —

14 —

1 2 3

Figure 4. UV cross-Unking of the overproduced CL22 and AC30 proteins withradioactive 5S rRNA. Autoradiography of the UV cross-linked proteins fractionatedby SDS^AGE A. 200ng of semi-purified proteins (see Materials and Methods)were mixed with radioactive 5S rRNA then treated with UV light. Lane 1, CL22;lane 2, AC 30; lane 3, control CL22 with no UV treatment. The position of CL22(indicated by an arrow) and of the AC3O deleted protein, corresponds to thatof the overproduced proteins (not shown) and to that of HPLC purified CL22protein UV-crosslinked to the 5S rRNA in the same conditions. B. Comparisonof the CL22- and AC30-5S rRNA binding ability using an equal amount (50ng) of semi-purified protein (see above). Lane 1, CL22 control not treated withUV; lane 2, CL22; lane 3, equivalent amount of CL22 and AC30.

Thus, we searched for chloroplast specific motives that are sharedby other chloroplast CL22 proteins. We found a motif(MPYRACYP) containing tyrosine residues which are absent inthe E.coli L22 and are conserved in other chloroplast CL22proteins (32). These tyrosine residues could be important forRNA binding as shown in some RNA-binding proteins (33).However, mutation of tyrosine residues conserved in thechloroplast CL22 and absent in the E.coli L22 sequence do notchange the RNA-binding property of CL22 (data not shown),suggesting that the RNA-binding domain may be scattered in theCL22 sequence, as shown for the E.coli r-protein S8 (34).

Nucleic Acids Research, 1993, Vol. 21, No. 3 639

If the CL22 binds to the 5S rRNA in vivo similarly to in vitro,then we must speculate that CL22 takes the place of some ofthe proteins homologous to the bacterial 5S rRNA binding,namely the L5, LI8 and L25 (35). None of these proteins hasbeen identified yet in the chloroplast ribosome hence it is notknown if such proteins exist, or if they have been lost in thechloroplast ribosome. However the properties of the CL22suggest a reorganisation of the chloroplast 50S ribosomal subunitcompared to E.coli.

We have shown here for the first time that a chloroplastribosomal protein can combine two properties divided intodifferent proteins in the E. coli ribosome. Together with previousworks (4, 6), we have shown that some chloroplast r-proteinscan possess specific properties that may prove to be an interestinglink with the function or regulation of translation in thechloroplast.

28. WalkczekJ., Schfllkr.D., Staffler-Mellicke.M., Brimacombe.R. andStSffler.G. (1988) EMBO J., 7, 3571-3576.

29. Tejedor.F. and Ballesta.J.P.G. (1986) Biochem., 26, 652-656.30. DavidsoaJ.N., HansonJvl.R. and Bogorad,L. (1974) MoL Gen. Genet., 132,

119-129.31. Kenan.D.J., Query.C.C and KeeneJ.D. (1991) Trends. Biochem., 16,

214-219.32. Carol.P. (1992) Ph D. Thesis, University de Grenoble I, France.33. Merril.B.M., Stone.F., Cobianchi.F., Wilson.S.H., and Williams.K.R.

(1988) J. BM. Oem., 7, 3307-3313.34. Wower.1.0., Kowdslri.M.P., Sears.L.E. and Zimmermarm.R.A. (1992) J.

Bact., 4, 1213-1221.35. Garrctt.R.A. and Noiler.H.F. (1979) J. MoL BM., 132, 637-648.

ACKNOWLEDGMENTS

This work was supported by grants from the EMBO (short termfellowship) to P.C., from the Direccion General de PoliticaCientifica (Spain) (PB87-045) J.P.B. and E.L. and from theFundacion Ramon Aceres (institutional grant to the Centra deBiologia Molecular).

REFERENCES

1. Dorne.A.M., Eneas-FilhoJ., Heizmann.P. and Macbe.R. (1984) Mol.Gen.Genet., 193, 129-134.

2. Sugiura.M. (1989) Ann. Rev. CelL Biol., 5, 51-70.3. Mache.R. (1990) Plant Science, 72, 1-12.4. Franzetti.B., Carol.P. and Mache.R. (1992) J. Biol. Chem., 267,

19075-19081.5. GantU.S. (1988)Cu/7\ Genet., 14, 519-528.6. Zhou.D.X. and Mache.R. (1989) MoL Gen. Genet., 219, 204-208.7. Johnson,C.H.,Kruft,V. and Subramanian.A.R. (1990)/ BioL Chem.,265,

12790-12795.8. Carol.P., Li,Y.F. and Mache.R. (1991) Gene, 103, 139-145.9. Zhou.D.X., Quigley.F., Massenet.O. and Mache.R. (1989) MoL Gen.

Genet., 216, 439-445.10. GantU.S., BaMauf.S.L., Calie.P.J., Weeden.N.F. and PalmerJ.D. (1991)

EMBOJ., 10, 3073-3078.11. Toulrifimpa.R., Romby,P., Rozier.C., Ehresmann.C, EhresmannB. and

Mache.R. (1989) Biochem., 28, 5840-5846.12. Herold.M. and Nierhaus.K.H. (1987) / BioL Chem., 18, 8826-8833.13. Arevalo.M.A., Tejedor,F., Polo.F. and BallestaJ.P.G. (1988) J. Biol.

Chem., 263, 58-63.14. Pardo.D. and Rosset.R. (1977) Mol. Gen. Genet., 156, 267-271.15. Vazquez.D. (1979) MoL BioL Biochem. Biophys., 30, 1-312.16. Mache.R., Dome.A. M. and Marti Batlle.R (1980) MoL Gen. Genet., YJ1,

333-338.17. Arfvalo.M.A., Tejedor.F.. Polo.F. and BallestaJ.P.G. (1989) / Med.

Chem., 32, 2200-2204.18. Laemlli, U.K. (1970) Nature, 227, 680-685.19. Kerlavage,A.R., Hasant,T. and Cooperman,B.S.(1983) / BioL Chem., 258,

6313-6318.20. Thomas.F. (1987) PhD. Thesis, University de Grenoble I, France.21. SambrookJ., Frisch.E.P. and Maniatis.T. (1989) Molecular Cloning: A

Laboratory Manual. Cold Spring Harbor University Press, Cold SpringHarbor.

22. Studier.F.W., Rosenberg.A.H., Dunn.A.H. and Dubbendorff.J.W. (1990)Methods Enzymol., 185, 60-89.

23. Frankel.S., Sohn.R. and Leiwand.L. (1991) Proc. Natl Acad. USA, 88,1192-1196.

24. Bradford.M. M. (1976) Anal. Biochem. 72, 248-254.25. Rozier.C. and Mache.R. (1984) Nucleic Adds Res. 12, 7293-7304.26. Cundliffe.E. (1987) Biochimie, 69, 863-869.27. Wittmann,H.G., Stdffler.G., Apirion.D., Rosen.L., TanakaJC, Tamaki.M.,

Tanaka^., Dekio.S., Otaka.E. and Osawa.S. (1973)MoL Gen. Genet., 127,175-189.