Archives of Physiology and Biochemistry, 1996, 104, B31-B56 B31

REVUE TRIMESTRIELLE (1), MARS 1996DRIEMAANDELIJKS TIJDSCHRIFT (1), MAART 1996

SOCIÉTÉ BELGE DE BIOCHIMIE ET DE BIOLOGIE MOLÉCULAIREBELGISCHE VERENIGING VOOR BIOCHEMIE EN MOLECULAIRE BIOLOGIE

162e RÉUNION, Antwerpen (UIA), 2 MARS 1996

Bio-computing

MEMBERS PROTECTEURSBESCHERMENDE LEDEN

AMERSHAM BelgiumANALISAPPLIED BIOSYSTEMSBEUN-DE RONDEBIO-RAD LABORATORIESBOEHRINGER INGELHEIM BIOWHITTAKERBOEHRINGER-MANNHEIM (Belgium)CANBERRA PACKARDCARL ZEISS N.V./S.A.CERESTAR gruppo FerruzziCURRENT BIOLOGYDEVOS-FRANÇOISDU PONT DE NEMOURSFISONS INSTRUMENTSHEWLETT PACKARDJANSSEN PHARMACEUTICALABAZ-SANOFILAMECOLIFE TECHNOLOGIESMEDATOM EUROPAMEDECINE/SCIENCE-JOHN LIBBEY EUROTEXTMERCK-BELGOLABONEW BRUNSWICK SCIENTIFICOMNILABOPHARMACIA LKBPLEUGERPROMEGASIGMA-ALDRICHSMITHKLINE BEECHAM BIOLOGICALSUCBVAN HOPPLYNUSWATERS

1. CONFÉRENCES

S. WODAK (UL Bruxelles)Can the protein fold be predicted from the amino acidsequence?

A. DANCHIN (Institut Pasteur, Paris)Bacterial genomes in silico.

2. COMMUNICATIONS

(1) Publication subsidiée par le Ministère de l’Education Nationaleet de la Culture.Publikatie gesubsidieerd door het Ministerie van NationaleOpvoeding en Cultuur.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B32 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Effect of cAMP- and cGMP-analogues on the enzymaticactivity of the nucleoside diphosphate kinase of rat C6glioma

K. ANCIAUX, D. ROYMANS and H. SLEGERS (Department of Bio-chemistry, University of Antwerp, U.I.A., Universiteitsplein 1,B-2610 Wilrijk-Antwerpen, Belgium)

The nucleoside diphosphate kinase (NDPK, nm23) ap-pears to be involved in processes related to differentiation,proliferation and metastasis (Okabe-Kado et al., 1995). Theenzyme transfers the γ-phosphate of NTP to NDP by a ping-pong reaction mechanism involving a histidine-phosphorylatedintermediate. Recently also autophosphorylation of serine-44 in nm23-H1 has been reported (McDonald et al., 1993).Mutant studies revealed that this autophosphorylation requiresthe active-site histidine (Bominaar et al., 1994).

We studied the effect of cAMP- and cGMP-analogues (8-ClcAMP, 8-BrcAMP, N6-monobutyryl-cAMP, dibutyryl-cAMP, 8-BrcGMP and dibutyryl-cGMP) on NDPK autophos-phorylation and on its ATP/GDP phosphotransferase activity.C6 glioma cells are induced to differentiate by the formercAMP-analogues but not by the cGMP-analogues.

DbcAMP, the most potent inducer of differentiation inC6, reveals a weak inhibition of the NDPK-activity. Howeverthis analogue increases the intracellular cAMP concentra-tion. Cyclic AMP concentrations above 100 µM inhibit theserine-autophosphorylation, but the ATP/GDP phosphotransferis only affected by mM amounts of cAMP. Strelkov et al.(1995)proposed a competition between cAMP and ATP binding inthe active site of NDPK.

Chloro cAMP is an efficient inhibitor of both the serine-autophosphorylation and the phosphotransferase activity ofthe enzyme. The latter cAMP-analogue is a differentiationinducer of C6 and Cho-Chung (1989) demonstrated that itinhibits cell growth by substituting for cellular cAMP. Wealso observed that µM concentrations of cGMP, 8-BrcGMPand 8-BrcAMP inhibit the ATP/GDP phosphotransferase ac-tivity. The effect of these analogues on the serine-autophos-phorylation and on the induction of differentiation in C6 isnot known.

The data presented show that differentiation of C6 is notunequivocally related to inhibition of the NDPK phospho-transferase reaction. It is not unlikely that inhibition of theserine-autophosphorylation of NDPK inactivates other (non-enzymatic) functions of the enzyme.

K.A. is a fellow of the Belgian NFWO (Kom op tegen Kanker)

ReferencesBOMINAAR, A.A., TEPPER, A.D. & VÉRON, M. (1994) Febs Letters, 353, 5-8.CHO-CHUNG, Y.S. (1989) J.Natl.Cancer Inst., 81, 982-987.MCDONALD, N.J., DE LA ROSA, A., BENEDICT, A., FREIJE, J.M.P., KRUTSCH, H.

& STEEG, P. (1993) J.Biol.Chem., 268, 25780-25789.OKABE-KADO, J., KASUKABE, T., HOZUMI, M., HONMA, Y., KIMURA, N., BABA,

H., URANO, T. & SHIKU, H. (1995) Febs Letters, 363, 311-315.STRELKOV, S.V., PERISIC, O., WEBB, P.A. & WILLIAMS , R.L. (1995) J.Mol.Biol.,

249, 665-674.

Playing around with electrostatic effects on the isoelec-tric pH and the pK a of the Catalytic Residue His-102 ofthe recombinant ribonuclease from Bacillus amylolique-faciens (Barnase)

K. BASTYNS1, M. FROEYEN1, J.F. DIAZ 1, G. VOLCKAERT2 andY. ENGELBORGHS1 (1Laboratory of Chemical and BiologicalDynamics, K.U.Leuven, Celestijnenlaan 200 D, B-3001 Leuven,Belgium and 2Laboratory of Gene Technology, K.U.Leuven, Wil-lem de Croylaan 42, B-3001 Leuven, Belgium).

Barnase, the guanine-specific ribonuclease of Bacillusamyloliquefaciens, was subjected to mutations in order toalter the electrostatic properties of the enzyme. In this study(Bastyns et al., 1996) mutations are made with the goal tointroduce an extra charge in the neighbourhood of the cata-lytically important His-102. This goal can be achieved byreplacing Ser-85 by Glu or Asp. In the case of the Ser85Glumutant, the COO- is at a distance of 5 Å from the His-102(Nε3)and 11 Å from the negative charge of the Pi of the dinucle-otide GpG (own modelling). This operation can also beachieved by the simultaneous mutation of Asp-86 into anAsn, with the goal to keep the total charge of the proteinconstant. The negative charge of Asp-86 is at 9 Å from His-102 (Nε3).

A similar set of mutations was made using Asp instead ofGlu at position 85. The results show that this is very impor-tant for polymeric substrates at pH 8.0 and 6.2, but much lessfor dinucleotides and for the cyclic intermediate at pH 6.5.For all mutants the pI was determined using the technique ofisoelectric focussing and calculated on the basis of the Tan-ford-Kirkwood theory.

When Glu was used to replace Ser-85 the correlationbetween the experimental and the calculated values is per-fect. However, in the Ser-85-Asp mutant the experimental pIdrop is bigger than the calculated one, and in the doublemutant (Ser-85-Asp and Asp-86-Asn) the compensation isnot achieved.

The effect of the mutations on the pKa of His-102 can bedetermined from the pH dependence of the kcat/KM for thehydrolysis of dinucleotides e.g. GpC. The effect can also becalculated using the the method of Honig. In this case theagreement is very good for the Glu-mutants and the singleAsp-mutant, but less for the double Asp-mutant. The globalstability of the Asp-mutants is, however, the same as that ofthe wild type, as shown by stability studies using urea dena-turation. Molecular dynamics calculations, however, showthat in the double Asp-mutant His-102 (H+) swings out of hispocket to make a hydrogen bridge with Gln-104 which shouldcause an additional pKa rise.

The effect of the Glu-mutations was also tested on all thekinetic parameters for GpC and the cyclic intermediate G>pat pH 6.5, for RNA at pH 8.0, and for poly(A) at pH 6.2. Theeffect of the mutations is rather limited for the dinucleotideand the cyclic intermediate, but a strong increase of the KM isobserved in the case of the single mutant (extra negativecharge) with polymeric substrates.

These results indicate that the extra negative charge has astrong destabilizing effect on the binding of the polymericsubstrates in the ground state and the transition state com-plex. A comparison with the structure of bound tetranucle-otides (Buckle & Fersht, 1994) shows that the extra negativecharge points towards the P2 site.

This work is supported by the Onderzoeksraad K.U.Leuven.

ReferencesBASTYNS, K., FROEYEN, M., DIAZ, F., VOLCKAERT, G. & ENGELBORGHS, Y. (1996)

Proteins, in press.BUCKLE, A.M. & FERSHT, A.R. (1994) Biochemistry 33, 1644-1653.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B33SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Conformation and folding of the novel protein h3

F. BOLLENGIER and A.MAHLER (Farmacologie, Faculteit G&F,V.U.B., B-1090 Brussel, Belgium)

To be biologically active, proteins must adopt specificfolded 3-dimensional tertiary structures. Yet the genetic in-formation for the protein specifies only the primary struc-ture, the linear sequence of amino acids in the polypeptidebackbone.

Denaturant urea-gradient-electrophoresis and urea-gel-chromatography provide an excellent tool for studying con-formational transitions. Consequently, to investigate the con-formational behaviour of the novel ubiquitous h3, isolatedand biochemically defined by us in the last decade (Bol-lengier & Muller, 1980,1988; Bollengier et al., 1983,1989),we applied both these techniques.

The results were as follows: 1) in urea-gradient Superosechromatography a progressive unfolding of the protein wasobserved between 3- and 5 M urea, with an intermediatephase of coexistence of the unfolded and folded state. Theaverage 23-kDa Mr of h3 monomer as calculated from stand-ard proteins, shifted to Mr 46 kDa as the unfolding proceed-ed. The influence of urea on the elution volume stronglysuggests that the native conformation of h3 is much morecompact than the fully unfolded state; indeed we alreadyreported that 25% of the h3 protein is folded as β-pleatedsheets (Bollengier et al., 1989), a percentage that argues infavour of a compact structure; 2) in urea-gradient-PAGE acomplex but reproducible pattern was obtained with a varietyof h3 preparations. The protein displayed one band whosemobility slowly and smoothly decreased between 3- and 5 Murea and multiple parallel bands of lesser mobility at start 0M urea, whose mobility steeply, but not abruptly , decreasedbetween 2- and 3 M urea toward a plateau. At this level spurformation was also noticed, suggesting that a fraction of theh3 molecules could not refold rapidly at the transition re-gion.

When reduced and alkylated h3 proteins were submittedto urea-gradient-electrophoresis, very similar patterns wereobtained for all specimens under investigation: a smooth,gradual and slow unfolding in the urea-gradient was observed,indicating a rapid equilibration of conformational states. At6- and 8 M urea dissociation into several parallel bands wasobserved for all the h3 proteins, although to a different de-gree; this behaviour is suggestive for intrinsic microhetero-geneity.

Globally, the results are consistent with a rather abrupttransition of folded to unfolded state of the native protein,and a smooth gradual low-key unfolding when the proteinsare reduced and alkylated. The similar behaviour in urea-gel-chromatography and urea-gradient-PAGE of the different spe-cies-h3 shows that they do not only share a common back-bone structure (Bollengier et al., 1993, Seddiqi et al., 1994),but also conformational properties.

ReferencesBOLLENGIER, F. & MAHLER, A. (1980) Neuropeptides 1, 119-135.BOLLENGIER, F., FORIERS, A., LAUWEREYS, H. & MAHLER, A. (1983) Neuropep-

tides 3, 243-254.BOLLENGIER, F. & MAHLER, A. (1988) J. Neurochem. 50, 1210-1214.BOLLENGIER, F., BEECKANS, S., MAHLER, A. & KANAREK, L. (1989) J. Neuro-

chem. 52, 1123-1126.BOLLENGIER, F., MAHLER, A., ANDRIES, R. & BOURGAIN, R. (1993) Arch. int.

Physiol. Biochim. 101, 63-69.SEDDIQI, N. et al. (1994) J. Mol. Evol. 39, 655-660.

Comparative analysis of multiple protein sequence align-ment methods

P. BRIFFEUIL, G.BAUDOUX, I. RERINGSTER, E. DEPIEREUX andE. FEYTMANS (Laboratoire de Biologie Moléculaire et Structu-rale, Facultés Universitaires Notre-Dame de la Paix, rue deBruxelles 61, B-5000 Namur, Belgium)

An essential step in modelling the three-dimensional struc-ture of a protein with unknown structure using a homologoussequence of known tertiary structure, is the alignement of thesequences of the proteins. Homologous proteins tend to main-tain sequence similarity within core domains and active sitessuch that it is often possible to identify structurally con-served regions (SCRs) between these proteins even whenoverall sequence similarity is low. The SCRs are all morereliably predicted from multiple sequence alignments. In simplecases, the quality of the alignment is excellent as judged bythe ability to correctly align corresponding domains of knowntertiary structure. However, in case of low homology, thealignment remains a big stumbling block (Shortle, 1995).The purpose of this study is to evaluate the ability of differ-ent multiple alignment methods to correctly identify the SCRswithin a family of proteins.

Alignments produced automatically by six different soft-wares BLOCKMAKER (Henikoff & Henikoff, 1991), CLUS-TALW (Thompson et al., 1994), MAP (Huang, 1994), MSA(Lipman et al., 1989), PIMA (Smith & Smith, 1992) andMATCH-BOX (Depiereux & Feytmans, 1992) are comparedto the published structure alignments of 13 different proteinfamilies. Alignments were performed using default parame-ters provided by the authors.The comparison of the performances of each alignment method,is based on two criteria:

- the rate of success(%)= (A/B)x100,- the rate of confidence(%)= (A/C)x100,

where,A = number of residues correctly aligned by a given method.B = total number of residues aligned in the structure align-

ment,C = total number of residues aligned in a sequence alignment

The optimal-alignment method would reproduce exactlythe structure alignment (corresponding to 100% of successand 100% of confidence). The rate of confidence indicatesthe accuracy of the predicted SCRs.

Results. The rate of success drops dramatically whenlow similarity sequences are aligned.CLUSTALW, MAP and MSA produce quite comparable re-sults, far better than PIMA, in terms of both success andconfidence. MATCH-BOX provides better success and con-fidence than BLOCKMAKER. Although the rate of successof MATCH-BOX is slightly lower than CLUSTALW, MAPand MSA, the MATCH-BOX method displays the best confi-dence rate especially when low similarity sequences arealigned.

These results suggest that the biologist-user of multipleprotein-sequence-alignment methods should exercise cautionin alignment of low similarity sequences. Under these condi-tions, MATCH-BOX provides more reliable results.

ReferencesDEPIEREUX, E. & FEYTMANS, E. (1992) CABIOS 8, 501-509.HENIKOFF, S. & HENIKOFF, J.G. (1991) Nucleic Acids Res. 19, 6565-6572.HUANG, X. (1994) CABIOS 10, 227-235.LIPMAN, D.J., ALTSCHUL, S.F. & KECECIOGLU, J.D. (1989) Proc Natl. Acad. Sci.

USA 86, 4412-4415.SHORTLE, D. (1995) Nature Struct. Biol. 2, 91-93.SMITH, R.F. & SMITH , T.F. (1992) Prot. Eng. 5, 35-41.THOMPSON, J.D., HIGGINS, D.G.& GIBSON, T.J. (1994) Nucleic Acids Res. 22,

4673-4680.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B34 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Diethylstilbestrol-induced renal carcinogenesis in maleSyrian hamster: biochemical comparison with α2-micro-globulin nephropathy in male rat

R. BROHÉE1, R.WATTIEZ1, F.JOURNÉ1, J.A.HEUSON-STIENNON2 and P.FALMAGNE 1. (1Service de Chimie Biologique, Fa-culté des Sciences and 2Service d’Histologie, Faculté deMédecine, Département de Biologie et de Pathologie Cel-lulaire, Université de Mons-Hainaut, Belgium)

Diethylstilbestrol (DES), like other synthetic and naturalestrogens, is able to induce renal tumors in male Syrian ham-sters (Llombart-Bosch & Peyro, 1975). The nephropathy whichpreceeds the induction of these neoplasms seems to sharemany similarities with a renal disease, called α2-microglo-bulin nephropathy, developed by male rats exposed to vari-ous chemical compounds, such as unleaded gazoline, d-limonene and 1,4-dichlorobenzene. The α2-microglobulin ne-phropathy is also involved in the development of renal tu-mors in male rats.

The biochemical mechanisms responsible for this dis-ease in male rats are well known: nephrotoxic compoundsbind to α2-microglobulin (α2µ), a major microprotein in kidneyand urine of male rat, and inhibit its lysosomal degradationin proximal tubules of the kidney. α2µ therefore accumulatesas hyaline droplets that are typical of this nephropathy(Borghoff et al.,1990). Kidneys of male hamsters chronicallyexposed to DES displayed in proximal tubule cells the samekind of intracellular droplets consistant with the develop-ment of a hyaline droplet nephropathy. It has therefore beensuggested that similar mechanisms could be involved in theinitiation of renal tumors by DES in male hamsters.

In this study, we assessed the presence of an a2m-likeprotein, or of any protein able to bind DES, in the kidney ofmale hamster. This problem has been approached by the frac-tionation of the soluble kidney fraction of male rats (as posi-tive control), male hamsters, treated or not with DES, femalehamsters (as negative control) and urine of male hamsters,using molecular gel filtration.

The components obtained were characterised by their pro-tein content, molecular weight (SDS-PAGE) and ability tospecifically bind to anti-rat a2m antibodies (western blot-ting) and to 3H(DES).

Cytosolic kidney fraction of male rats displayed two ma-jor microproteins ( Mr 16 and 18 kDa ) that were specificallyrecognized by anti-rat α2µ antibodies; in contrast, the resultsobtained with hamsters failed to show any major micropro-tein in cytosolic renal fractions as well as in urine, except a14-kDa protein identified as lysozyme by automated Edmandegradation.

Moreover, in the experimental conditions used, there wasno specific recognition of any renal or urinary protein by theanti- rat a2m antibodies in hamsters.

Finally, analysis of 3H(DES) binding showed that no re-nal or urinary protein specifically bind to DES.

The results strongly suggest that, in contrast to rats, malehamsters are unable to develop an α2-microglobulin ne-phropathy, as it was previously shown for the human (Borghoff& Lagarde,1993).

This work was supported by a grant FNRS Televie to F. JOURNE.

ReferencesBORGHOFF, S.J., SHORT, B.G. & SWENBERG, J.A. (1990) Annu. Rev. Pharmacol.

Toxicol. 30, 349-367.BORGHOFF, S.J. & LAGARDE, W.H. (1992) Toxicol.Appl.Pharmacol. 103, 539-

548.LLOMBART-BOSCH, A. & PEYRO, A. (1975) Europ.J.Cancer 11, 403-412.

Protein microsequencing by automated EDMAN degra-dation. Cysteine identification after alkylation withacrylamide in the microsequencer reaction cartridge

M. CARION, R. WATTIEZ and P. FALMAGNE (Laboratoire deChimie Biologique, Université de l’Etat à Mons, B-7000 Mons,Belgium).

Cysteine is not easy to identify during protein sequenceanalysis by automated Edman degradation because its phe-nylthiohydantoin (PTH) derivative is unstable, losing H2S toform PTH-dehydroalanine, which in turn degrades further toa variety of unidentifiable products (Edman & Henschen,1975). The problem of PTH-Cys instability has been over-come by the alkylation of the sulfhydryl group with acryla-mide to form PTH-Cys-S-propionamide (PTH-Cys-S-PAM),a more stable thioether (Brune, 1992). First, the PTH-Cys-S-PAM derivative was synthesized to allow its identification byreverse-phase high-pressure liquid chromatography (HPLC).The synthesis of the PTH-Cys-S-PAM derivative includedtwo successive steps : firstly, the synthesis of the Cys-S-propionamide (Cys-S-PAM) derivative by reaction, undermildly alkaline conditions, between L-cysteine and acryla-mide (Friedman & Cavins, 1965; Cavins & Friendman, 1968);secondly, the synthesis of the PTH-Cys-S-PAM derivative byreaction between the Cys-S-PAM derivative and phenylisothi-ocyanate (PITC) (Edman & Henschen, 1975). The differentproducts were synthesized and characterized by infrared- andultraviolet spectroscopy and nuclear magnetic resonance.

The cysteine alkylation was then carried out on a proteinin solution (α-lactalbumin), and next, on this same proteinadsorbed onto a glass-fiber disk previously processed withpolybrene or onto a polyvinylidene difluoride (PVDF) mem-brane, directly in the reaction cartridge of the protein micro-sequencer (in situ alkylation). The automation of the chemi-cal modification was obtained by adding a «reduction-alkyla-tion» programme in the principal programme of the proteinmicrosequencer.

Chemical modification of the phenylthiohydantoin deri-vative of cysteine with acrylamide complied with the manydifferent requirements of the protein microsequence analy-sis. Indeed, the in situ cysteine alkylation by acrylamide didnot significantly affect the initial and repetitive yields of theprotein microsequence analysis. On the other hand, the PTH-Cys-S-PAM derivative was stable during the different stepsof the automated Edman degradation, and its elution profilewas characterized by a sharp, distinct absorption peak thatwas well separated from all the other PTH-amino acid stand-ards. Finally, the PTH-Cys-S-PAM derivative was easily syn-thesized.

In summary, the in situ cysteine sulfhydryl alkylation byacrylamide is a reliable method that can be used during pro-tein microsequence analysis by automated Edman degrada-tion to quantitatively identify the cysteine residues. This alkyla-tion performed directly in the microsequencer (in situ) al-lows the identification of cysteine in the protein concentra-tion range of the order of the picomole.

ReferencesBRUNE, D.C. (1992) Anal. Biochem. 207, 285-290.CAVINS, J.K. & FRIEDMAN, M. (1968) J. Biol. Chem. 243, 3357-3360.EDMAN, P. & HENSCHEN, A. (1975) in Protein Sequence Determination (NEEDLE-

MAN, S., ed.), pp. 212-236, Springer-Verlag, New York.FRIEDMAN, M. & CAVINS, J.K. (1965) J. Am. Chem. Soc. 87, 3672-3682.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B35SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Characterization of Fau rearrangements in radiation-in-duced osteosarcomas of CF-1 mice

D. CASTEELS1, L. MICHIELS2 and J. MERREGAERT1 (Laboratoryof Molecular Biotechnology, Department of Biochemistry, Uni-versity of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, 2 Dr.Willems-Instituut, Universitaire Campus, B-3610 Diepenbeek,Belgium)

The Fau gene (FBR-MuSV associated ubiquitiously ex-pressed gene) is the cellular homologue of the fox sequencein the Finkel-Biskis-Reilly Murine Sarcoma Virus (FBR-MuSV). This retrovirus was originally isolated from a [90Sr]-induced bone tumor of the XG/f mouse strain and it is capa-ble of inducing osteosarcomas in susceptible mice (Finkel etal., 1975). The Fau gene encodes a hybrid protein of 133amino acids consisting of the S30 protein, a protein of thesmall ribosomal subunit, fused at the amino terminus to anubiquitin-like protein (FUBI)(Michiels et al., 1993; Olvera& Wool, 1993).

In three out of eight primary [90Sr]-induced osteosarco-mas of CF-1 mice an additional Fau fragment has been iden-tified. This extra fragment is also present in normal Balb/cmice but absent in normal CF-1 mice (Michiels et al., 1993).We characterized this extra Fau fragment in Balb/c as a novelretropseudogene, Fau-ps3 (Casteels et al., 1995). It is 90%identical to the mouse Fau cDNA sequence and contains thestart- and stopcodon, and the polyadenylation signal. Thepoly-A tail (9bp), the loss of introns and the direct repeat(AAAAAATCTTCCTTTCA/GG/A) flanking the processedtranscript, typical for retropseudogenes, were found at theexpected positions (Vanin, 1985). The regions flanking theFau-ps3 sequence are unknown.

When PCR was performed with primers in the regionflanking the Fau-ps3 insertion, several fragments were de-tected. When blotted and hybridized with the the Fau cDNAprobe the expected band (1200 bp) containing the Fau-ps3retropseudogene was detected in Balb/c and the three tu-mours. No band of this length was seen when normal CF-1genomic DNA was used. We cloned and sequenced the 1200bpband from one of the three tumours and found it to be 99%identical to the Fau-ps3 gene of Balb/c. This implicates thatthe Fau-ps3 sequence is derived from a common ancestorand that de novo formation of retropseudogenes in the radia-tion-induced osteosarcomas is not a consequence of irradia-tion. The differences found between the individual CF-1 micecan be explained by the fact that CF-1 is an outbred strainwhich allows heterozygosity.

The same PCR reaction also detected an additional bandin all samples. This band was of the same length as the frag-ment without the Fau-ps3 insertion. To confirm this, the CF-1 fragment was cloned and sequenced and it indeed turnedout to be the sequence of the flanking regions of the Fau-ps3retropseudogene. The insertion site was, as expected, the sameas the flanking direct repeats of Fau-ps3 (AAAAAATCTTC-CTTTCAG). Whether this fragment and the Fau-ps3 frag-ment are allelics or not, remains to be proven by chromosom-al localization. Since inbred mouse strains, such as Balb/care expected to be homozygous, it is most likely that they arelocated on different loci on the chromosomes.

ReferencesCASTEELS, D., POIRIER, C., GUÉNET, J.-L. & MERREGAERT, J. (1995) Genomics,

25, 291-294.FINKEL , M.P., REILLY , C.A. JR. & BISKIS, B.O. (1975) Front.Radiat.Ther.Oncol.,

10, 28-39.MICHIELS, L., VAN DER RAUWELAERT, E., VAN HASSELT, F., KAS, K. & MERRE-

GAERT, J. (1993) Oncogene, 8, 2537-2546.OLVERA, J. & WOOL, I.G. (1993) J. Biol. Chem., 268, 17967-17974.VANIN , E.F. (1985) Annu. Rev. Genet., 19, 253-272.

The pyrH-encoded UMP kinase appears to be involved inthe pyrimidine-specific regulation that modulates promoteractivity of the Escherichia coli carAB operon encodingcarbamoylphosphate synthetase

D. CHARLIER1,4, A. KHOLTI2, D. GIGOT2, M. ROOVERS3,4, N.HUYSVELD3,4 and N. GLANSDORFF1,3,4. (1Research Instituteof the COOVI-CERIA, Ave. E. Gryson 1, B-1070 Brussels, Bel-gium, 2Microbiology ULB, 3Microbiology VUB and 4Vlaams In-teruniversitair Instituut voor Biotechnologie)

The expression of the pyrimidine biosynthetic genes andoperons is non-coordinately regulated by multiple, differentcontrol mechanisms including UTP-sensitive transcriptionattenuation for pyrBI and pyrE and translational control basedon a nucleotide-pool-dependent selection of alternative tran-scription starts for pyrC and pyrD (Neuhard & Nygaard,1996). The carAB operon, encoding the unique carbamoyl-phosphate synthetase (CPSase) of E. coli, presents neither ofthese control mechanisms; however, the dual role of CPSasein the biosynthesis of arginine and the pyrimidines is reflect-ed in the cumulative repression of the tandem pair of promot-ers governing the expression of the carAB operon (Piette etal., 1984). Repression, by arginine, of the downstream pro-moter P2 is now well documented (Roovers et al., 1988,Charlier et al., 1992) but though a number of different trans-acting elements including the integration host factor (IHF)(Charlier et al., 1993) and CarP/PepA (Roovers et al., 1988;Charlier et al., 1995a & b) have been shown to be requiredfor pyrimidine-specific regulation of the upstream promoterP1, our vision of how pyrimidines exert their control on thecarAB operon is still uncomplete. Indeed, IHF and CarP/PepA seem to play an architectural role in the assembly ofthe pyrimidine-specific-nucleo-protein regulatory complexwhile the real sensor of changes in the nucleotide pools hadnot yet been identified.

It now appears that this sensor could be UMP kinase, theproduct of the E. coli pyrH gene. Indeed, we have isolated anew, particular type of pyrH mutant that bears a quasi normallevel of UMP kinase activity but yet is impaired in pyrimi-dine-specific regulation of the carAB promoter P1 as demon-strated by enzyme assays and quantitative S1-nuclease-map-ping of car transcripts. This mutation was cloned, sequencedand shown to consist of a single C to A transversion thatconverts alanine 94 of the pyrH sequence into glutaminicacid. Overexpression of the mutant pyrH allele by cloninginto a multicopy vector does not restore normal pyrimidineregulation of the car promoter P1, though the UMP kinaselevels of such transformants exceed the one of a wild typepyrH+ strain by about 10-fold. These results strongly suggestthat UMP kinase fulfills, besides its catalytic role in UMPphosphorylation, an additional, previously unrecognized func-tion directly involved in pyrimidine - specific modulation ofpromoter activity.

This work was supported by the Belgian Fund for Joint Basic Research (con-tract n° 2.9007.92).

ReferencesCHARLIER, D., ROOVERS, M., VAN VLIET, F., BOYEN, A., CUNIN, R., NAKAMURA ,

Y., GLANSDORFF, N. & PIÉRARD, A. (1992) J. Mol. Biol. 226, 367-386.CHARLIER, D., ROOVERS, M., GIGOT, D., HUYSVELD, N., PIÉRARD, A. &

GLANSDORFF, N. (1993) Mol. Gen. Genet. 237, 273-286.CHARLIER, D., GIGOT, D., HUYSVELD, N., ROOVERS, M., PIÉRARD, A. &

GLANSDORFF, N. (1995a) J. Mol. Biol. 250, 383-391.CHARLIER, D., HASSANZADEH, G., KHOLTI, A., GIGOT. D., PIÉRARD, A. &

GLANSDORFF, N. (1995b) J. Mol. Biol. 250, 392-406.NEUHARD, J. & NYGAARD, P. (1996) in E.coli and S.typhimurium: Cellular &

Molecular Biology. (NEIDHARDT et al., eds.) American Society for Micro-biology. Washington D.C.

PIETTE, J., NYUNOYA, H., LUSTY, C., CUNIN, R., WEYENS, G., CRABEEL, M.,CHARLIER, D., GLANSDORFF, N. & PIÉRARD, A. (1984) Proc. Natl. Acad.Sci. USA 81, 4134-4138.

ROOVERS, M., CHARLIER, D., FELLER, A., GIGOT, D., HOLEMANS, F., LISSENS, W.,PIÉRARD, A. & GLANSDORFF, N. (1988) J. Mol. Biol. 204, 857-865 (1988).

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B36 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Expression of the DNA-binding domain of the androgenand glucocorticoid receptor as fusion proteins with glu-tathione S-transferase

CLAESSENS, F., SCHOENMAKERS, E., ALEN P., PEETERS, B., 1VER-HOEVEN, G. and ROMBAUTS, W. (Afdeling Biochemie, 1Labo-ratorium voor Experimentele Geneeskunde en Endocrinologie,Campus Gasthuisberg, Herestraat 49, Leuven, Belgium)

Androgen response elements (ARE) are DNA sequences whichinteract directly with the androgen receptor, and which are ableto confer an androgen responsiveness to a heterologous promot-er. Several potential AREs have been defined by means of tran-sient transfection experiments. However, the demonstration ofbinding by the androgen receptor has been difficult. De Vos et al.(1991) have first described a fusion protein with the DNA-bind-ing domain of AR (AR-DBD) fused to protein A. This proteinhas extensively been used in gel retardations and DNase I foot-printing analyses (De Vos et al., 1991; Claessens et al., 1993;Devos, 1995). However, large amounts of protein have to beadded before DNA-binding can be detected. Furthermore, cleav-ing off the protein A moiety from the DBD has proven to bedifficult.

We have earlier reported that one ARE from the probasingene promoter has a high affinity for the AR-DBD but not for theGR-DBD (PB-ARE-2; Claessens et al., 1995). To investigatewhich of the differences in the DBDs of AR versus GR are re-sponsible for this specificity, we need a more versatile techniquefor the production of mutant DBDs suitable for use in gel retar-dation experiments.

Because of the difficulties we met with the protein A fusionproteins, the cDNA fragments coding for the AR and GR-DBDswere amplified by PCR and inserted downstream of the IPTG-inducible glutathione S-transferase (GST) cassette in the pGEX-2TK vector (Pharmacia). The resulting fusion proteins bind DNAin a sequence-specific manner, as demonstrated by competitionexperiments and gel retardation assays. Like the protein A-fu-sion proteins, the GST-fusion proteins display a high affinity forthe C3(1)ARE (originating from the C3(1) intron), whereas thePB-ARE-2 is recognised with high affinity by the AR-DBD- butnot by the GR-DBD-GST fusion protein.

In a further search for AR-specific sequences, we substitutedthe T by a G at position +2 (underscored) in the sequence of theC3(1) ARE (5’AGTACGTGATGTTCT-3’). This element has ahigh affinity for the AR-DBD, but not for the GR-DBD. Theabsence of binding by GR and PR to a GRE containing the samemutation has also been reported by Nordeen et al. (1990).

In order to exclude protein-protein interactions by the GST-portion of fusion proteins interfering with the DNA-binding theseexperiments were repeated with the DBDs cleaved off by thrombindigestion. The observed binding characteristics were identical tothose observed for the uncleaved fusion proteins.

This expression system will now be used in mutation analy-ses of the AR- and GR-DBDs, in search for an explanation forthe specificity of AR for the PB-ARE-2 sequence.

Supported by grants from “Fonds voor Geneeskundig en WetenschappelijkOnderzoek” and “Geconcerteerde Onderzoeksactie”; F.C. is holder of a post-doctoral fellowship of the “Nationaal Fonds voor Wetenschappelijk Onder-zoek”. We thank R. De Bruyn , R. Bollen and V. Feytons for excellent techni-cal assistance. P.A. is holder of a scholarship of the “Vlaams Instituut voor debevordering van het Wetenschappelijk-Technologisch Onderzoek in de Indus-trie (I.W.T.)”.

ReferencesCLAESSENS, F., CELIS, L., DE VOS, P., PEETERS, B., HEYNS, W., VERHOEVEN, G.

& ROMBAUTS, W. (1993) Biochem. Biophys. Res. Comm. 191, 688-694.CLAESSENS F., ALEN, P., DE VOS, P., VERHOEVEN, G., PEETERS, B. & ROMBAUTS,

W. (1995) Arch. int. Biophys. Biochim. 103, B5DE VOS, P., CLAESSENS, F., WINDERICKX, J., VAN DIJCK, P., CELIS, L., PEETERS,

B., ROMBAUTS, W., HEYNS, W. & VERHOEVEN, G. (1991) J. Biol. Chem.266, 3439-3443.

DEVOS, A. (1995) Doctoral thesis at the Katholieke Universiteit van LeuvenNORDEEN, S.K., SUH, B.J., KÜHNEL, B. & HUTCHISON, C.A. III (1990) Mol.

Endocrinol. 12, 1866-1873.

Characterization of an androgen responsive unit upstreamof the mouse GPX-5 gene

CLAESSENS, F., 1LAREYRE, J.-J., 1DREVET, J., 1DUFAURE, J.-P.and ROMBAUTS, W. (Afdeling Biochemie, Campus Gasthuis-berg, Herestraat 49, 3000 Leuven, Belgium and 1Laboratoire deBiologie Cellulaire, Université Blaise Pascal, CNRS URA 1940,63177Aubière Cedex, France)

The epithelial cells of the epididymis of the mouse se-crete a number of proteins under androgen control. The cDNAsequence of one of these, previously called arMEP (andro-gen-regulated mouse epididymal protein) was shown to behomologous to the glutathione peroxidases, and is thereforecalled GPX-5 (Ghyselinck et al., 1990).

In order to study the transcriptional regulation of GPX-5, the coding gene was cloned by Ghyselinck et al. (1993).The promoter region (from position -167 to +24) of this sin-gle-copy gene contains several consensus sequences for gen-eral transcription factors like cEBP, NF1, Sp1 and a possibleTATA-box, but no good candidate androgen-responsive ele-ment (ARE) is present. This promoter region is indeed capa-ble of directing transcription of a CAT gene, but no androgenresponses could be induced (Ghyselinck et al., 1993).

Subsequently, restriction fragments originating from theupstream region were cloned in front of a tk-promoter linkedto a luciferase gene, and screened for androgen-responsive-ness in transient co-transfection experiments. Constructs con-taining the sequence between -1797 and -577, show andro-gen-inducible transcription.

We have performed DNase I footprinting experiments todetermine the exact binding site of the androgen receptor.We observed a protection of the sequence 5’-TGGTAGGAT-ACATGTTCTCTCTC-3’, which was abolished by additionof 100-fold excess of the C3(1)ARE (Claessens et al., 1993).The affinity of this new ARE, as measured in gel retardation,was very low.

In order to establish the possible binding of other tran-scription factors, the same restriction fragment was subject-ed to DNase I digestion in the presence of nuclear extractsfrom liver. A clear protection of the sequence 5’-GTTTAT-CAGATTGGCAGAGTTAAGAGGA-3’ was found. Thiselement contains a consensus sequence for binding of factorsbelonging to by the GATA (underlined), as well as the NF-1family (in bold). A hundredfold excess of NF-1 consensusoligonucleotides prevents the formation of the footprint, in-dicating that indeed an NF-1-like factor is binding.

In conclusion, we have established binding of the andro-gen-receptor, and an NF-1-like factor to an upstream andro-gen responsive unit of the GSPX-5 gene. The combination ofan ARE with an NF-1-binding site was also found in theARUs of the C3(1) intron (Celis et al., 1991), the crp1 and 2promoters (Devos, 1995), and the mvdp-promoter (Claessenset al., 1995).

Supported by grants from Fonds voor Geneeskundig en Wetenschappelijk Onder-zoek and Geconcerteerde Onderzoeksactie; F.C. is holder of a postdoctoralfellowship of the Nationaal Fonds voor Wetenschappelijk Onderzoek. Wethank R. De Bruyn , R. Bollen and V. Feytons for excellent technical assist-ance. P.A. has a scholarship of the Vlaams Instituut voor de bevordering vanhet Wetenschappelijk-Technologisch Onderzoek in de Industrie (I.W.T.).

ReferencesCELIS, L., CLAESSENS, F., PEETERS, B., HEYNS, W., VERHOEVEN, G. & ROM-

BAUTS, W. (1993) Mol. Cell. Endocrinol. 94, 165-172.CLAESSENS, F., CELIS, L., DE VOS, P., PEETERS, B., HEYNS, W., VERHOEVEN, G.

& ROMBAUTS, W. (1993) Biochem. Biophys. Res. Commun. 191, 688-694.CLAESSENS, F., DARNE, C., FABRE, S., MANIN , M., PEETERS, B., ROMBAUTS, W.,

VERHOEVEN, G., VEYSSIERE, G. & JEAN, CL. (1995) Arch. int. Physiol.Biochim. 103, B5.

DEVOS, A. (1995) Doctoral thesis at the K.U.Leuven, BelgiumGHYSELINCK, N.B., JIMENEZ, C., LEFRANÇOIS, A.M. & DUFAURE, J.-P. (1990)

Mol. Endocrinol. 4, 4-15.GHYSELINCK, N.B., DUFAURE, I., LAREYRE, J.-J., RIGAUDIERE, N., MATTEI, M.-

G. & DUFAURE, J.-P. (1993) Mol. Endocrinol. 7, 258-272.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B37SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Effect of phosphate on the thermostability of native andchimeric yeast alcohol dehydrogenases

DE BOLLE, X., PAQUET, J.-Y., VINALS, C., DEPIEREUX, E.,VANDENHAUTE, J. & FEYTMANS, E. (Unité de Recherche enBiologie Moléculaire, Facultés Universitaires Notre-Dame de laPaix, B-5000 Namur, Belgium)

Yeast alcohol dehydrogenases (ADH) I and II are tetra-meric enzymes catalyzing the reversible oxidation of ethanolto acetaldehyde, using NAD+ as coenzyme. The two isozymesADH I and ADH II differ by their thermostability (Schimpfes-sel, 1968). Seven chimeric ADH I - ADH II proteins (CADH1or ADH1-[1-258]-ADH2-[259-347] to CADH7 or ADH1-[1-57]-ADH2-[58-347]) have been produced (De Bolle et al.,1995). The thermal stability of CADH1 is insensitive to re-duction by dithiothreitol, unlike ADH I (dithiothreitol is ableto reduce a disulfide bridge in ADH I). The substitutionCys277→Ser would be responsible for this difference be-tween ADH I and CADH1. ADH I and CADH1 are thermo-labile.

The dissociation propensity and the thermal stability ofreduced ADH I and CADH1 were studied. Increasing CADH1concentration or phosphate concentration stabilize the tetra-meric form of CADH1, as shown by gel permeation analysis.This suggests that dissociation is reversible, and that the dis-sociation constant can be affected by the presence of phos-phate. The thermal stability of CADH1 enzyme is stronglydependent upon salt nature: the t50 in 50 mM phosphate buff-er is 44°C while it is 35°C in MOPS (3-[N-morpholino]-propanesulfonate) buffer, at pH 7.0. The presence of a low-affinity phosphate-binding site is expected at the interfacebetween subunits in the tetrameric CADH1, as previouslysuggested for ADH I (Hayen, 1995).

The low thermostabilities of reduced ADH I and CADH1are proposed to be due to their high propensity to dissocia-tion. Among the seven substitutions between ADH I andCADH1, three (Val259→Cys, Cys277’Ser and Gln283’His)are located in a potential inter-subunit contact zone, in athree-dimensional model of yeast ADH I dimer based on thehorse liver ADH crystallographic structure. The higher sta-bility of CADH1 in the presence of phosphate (∆t50 of 9°Cwhen compared to MOPS), which is not observed with re-duced ADH I, could be due to His283. A positively chargedhistidine residue could interact with phosphate at the inter-subunit region, thereby modifying dissociation propensity andthermal stability.

The effects of phosphate buffers on the thermostability ofnative and chimeric ADH were systematically investigated.It was shown that phosphate buffers at high concentrationstabilize enzymes with intermediate thermostability (t50~60°C), but destabilize enzymes with high thermal stability(t50 ~70°C). The effects of the cations K+ or Na+ are negligi-ble in most of the tested conditions. The effects of the phos-phate buffers on enzymes with intermediate and high thermalstability are compatible with the classic “Hofmeister effect”.At high temperatures, the aggregation could be promoted byhigh concentration of phosphate, which would explain thelower thermal stability in the presence of high concentrationof phosphate.

ReferencesDE BOLLE, X., VINALS, C., PROZZI, D., PAQUET, J.-Y., LEPLAE,R., DEPIEREUX, E., VANDENHAUTE, J. & FEYTMANS, E. (1995)Eur. J. Biochem. 231, 214-219.HAYEN, P. (1995) PhD Thesis in Biological Sciences, FUNDP,Namur.SCHIMPFESSEL, L. (1968) Biochim. Biophys. Acta 151, 317-329.

Identification of two Physarum Casein Kinase II enzymesthat phosphorylate fragmin, an endogenous actin-bind-ing protein

V. DE CORTE, J. GETTEMANS, Y. DE VILLE, 1E. WAELKENS, 1P.AGOSTINIS and J. VANDEKERCKHOVE. (Flanders Institute ofBiotechnology, Department of Biochemistry, University Ghent,Gent, B-9000 Belgium and 1Laboratory for Biochemistry, KU-Leuven, Campus Gasthuisberg,, B-3000 Leuven, Belgium)

Casein kinase II (CK II) isolated from different cell typesis a 130-kDa heterotetramer resulting from the combinationof two catalytic subunits (α and/or α’) and two regulatorysubunits (β) (Pinna, 1990). Fragmin is a Ca2+-dependent ac-tin-binding protein (ABP) that is structurally very similar tothe amino-terminal half of gelsolin (Ampe & Vandekerck-hove, 1987). In vitro, it stimulates actin polymerization, bindsto the fast growing end of F-actin filaments and is able tosever (break) actin filaments (Gettemans et al., 1995).

Fragmin is phosphorylated by a protein kinase that ispresent in Physarum polycephalum, Dictyostelium discoideum,Saccharomyces cerevisiae and pig spleen. Using several pro-tein and peptide substrates and a variety of common kinasemodulators, we were able to identify this enzyme as CK II.Interestingly, Physarum microplasmodia contain two CK IIactivities. Both enzymes phosphorylate fragmin at an identi-cal site (-Gly-Gly-Ser-Asp-Leu-Glu-) that corresponds withthe consensus CK-II phosphorylation motif: Ser/ThrXXA(A=acidic residue) (Meggio et al., 1994).

We partially purified these enzymes by sequential chro-matography on DEAE-cellulose, hydroxyapatite, mono Q ion-exchange and Superdex 200 gel filtration using fragmin assubstrate for screening of the column fractions. The first CKII (referred to as kinase 1) eluted from gel filtration with amolecular mass of 45-55 kDa, indicating one catalytic subu-nit. The second Physarum CK-II (referred to as kinase 2)was found to display a Mr of 300-350 kDa, compatible with amultimeric structure. Immunocharacterization of kinase 1 withchicken polyclonal CK II antibodies revealed that it consist-ed of a single subunit migrating at the position of α’ follow-ing SDS-PAGE. These data indicate that the Physarum ki-nase 1 represents a naturally occurring monomeric form ofCK II devoid of regulatory subunits.

This work was supported by grants from the Belgian National Science Fund(NFWO), the Concerted Research Action of the Flemish Community (GOA)and the Flemish Institute for Science and Technology (VLAB-COT).

ReferencesAMPE, C. & VANDEKERCKHOVE, J. (1987) EMBO J. 6, 4149-4157.GETTEMANS, J., DE VILLE, Y., WAELKENS, E., & VANDEKERCKHOVE, J. (1995) J.

Biol. Chem. 270, 2644-2651.MEGGIO, F., MARIN, O. & PINNA , L.A. (1994) Cell. Mol. Res. 40, 401-409.PINNA, L.A. (1990) Biochim. Biophys. Acta 1054, 267-284.A

rchi

ves

of P

hysi

olog

y an

d B

ioch

emis

try

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

McM

aste

r U

nive

rsity

on

11/2

6/14

For

pers

onal

use

onl

y.

B38 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Toward an integrated tool to manage sequencing projects

P. DÉHAIS1, J. COPPIETERS2 and M. VAN MONTAGU1 (1Labora-torium voor Genetica, via the Department of Genetics, affiliatedto the Flanders Interuniversity Institute for Biotechnology, Uni-versiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium;and 2European Molecular Biology Laboratory, EBI-Hinxton Hall,Hinxton CB10 1RQ, United Kingdom)

Since a few years a lot of genome sequencing projectshave begun implying a collaboration between laboratories allover the world. Some laboratories are involved in clone li-brary production and management, whereas others deal withthe sequencing itself of genome parts. Our tool is intendedfor this second set of laboratories.

This tool has first been developed for our own needs:several sequencing projects in our laboratory take place inthe context of ESSA projects (European Scientists Sequenc-ing Arabidopsis). The main problem the team had to face wasdue to the shotgun method adopted. With this kind of strate-gy, a sequence has first to be split in a random set of frag-ments that are short enough to be sequenced automatically.The advantage of doing so is that these two steps can be donevery fast. Unfortunately, even if the sequenced fragments areobtained quickly, the way they were combined to form thewhole sequence is lost as well as in which direction (reversecomplement or not) they have been sequenced. The standardway to reconstruct the whole sequence is to take advantageof potential overlapping between fragments, but this is anNP-hard problem.

Some solutions have been proposed: GCG (Staden, 1980;Gleeson & Staden, 1991), FAB (Kececioglu & Myers, 1989),Sequencher (Ahern, 1992), and TIGR (Sutton et al., 1995),but none seems to provide both quality and user-friendlyinterface, either because they do not allow direct manipula-tion of traces (“direct” signal output from sequencer with thefour-base-content prediction curves of the sequence), or donot produce an optimal arrangement. Nevertheless, the bestalgorithm is FAB, according to Miller and Powell (1994). Itis the reason why we have chosen and enhanced this algo-rithm to fit with users’ wishes. The tool we have developedcan read direct output files from A.L.F. and A.B.I. sequencer,showing traces allowing users to correct the prediction (sig-nal too high or too low to be interpreted, bad predictions,vector insertion…).

All these modifications lead to a problem of versions whenmore than one person is involved in the same project. It is thereason why we have linked this tool to a database managementsystem (Sybase) to keep track of all these modifications. Tohave means to go back further in case of inconsistency of dataand impossibility to find proper contigs (back to the cloningreactions, for instance, if infection during cloning phase hasbeen detected), we also store in this database all informationrelated to reactions upstream from the automated sequencingphase (cloning, DNA preparation…) in this database.

Finally, we have in mind to store some other informationto provide an overview of the whole sequencing project, tak-ing into account the results of other teams. This can be usefulto build a sequencing strategy. The next step of this develop-ment is to try new algorithms (Gleizes & Hénaut, 1994; Idury& Waterman, 1995; Myers, in preparation) and take advan-tage of “new” technics (parallelism and object-oriented data-bases).

ReferencesAHERN, K. (1992) Biotech. Software 9, 8-11.GLEESON, T.J. & STADEN, R. (1991) Comput. Appl. Biosci. 7, p. 398.GLEIZES, A. & HÉNAUT, A. (1994) Comput. Appl. Biosci. 10, 401-408.IDURY, R.M. & WATERMAN, M.S. (1995) J. Comput. Biol. 2, 291-306.KECECIOGLU, J. & MYERS, E. (1989) Techn. Rep. TR 89-5, University of Arizona.MILLER, M.J. & POWELL, J.I. (1994) J. Comput. Biol. 1, 257-269.STADEN, R. (1980) Nucleic Acids Res. 8, 3673-3694.SUTTON, G.G., WHITE, O., ADAMS, M.D. & KERVALAGE, A.R. (1995) Genome

Sci. Technol. 1, 9-19.

Allosteric properties of Escherichia coli carbamoylphos-phate synthetase: characterization of mutant enzymes ei-ther hypersensitive or totally insensitive towards the al-losteric inhibitor UMP

S. DELANNAY 1,4, C. TRICOT2, D. CHARLIER2,3, V. STALON1 andA. PIÉRARD1,2 (1Laboratoire de Microbiologie, Université Li-bre de Bruxelles, 2Research Institute of the CERIA-COOVI,3Vlaams Interuniversitair Instituut voor Biotechnologie, Ave. E.Gryson 1, B-1070 Brussels and 4Institut Supérieur Industriel duHainaut, Bd Solvay 31, B-6000 Charleroi, Belgium)

Carbamoylphosphate (CP), a common precursor of the arginineand pyrimidine pathways, is in E. coli synthesized by a uniquecarbamoylphosphate synthetase (CPSase), the product of the car-AB operon according to the reaction: glutamine + HCO3- + 2Mg-ATP + H2O Æ Glu + 2Mg-ADP + CP + Pi. The catalytic activity ofCPSase is highly modulated by the antagonistic effects of twoactivators, ornithine and IMP, and of a potent inhibitor, UMP (re-viewed in Cunin et al., 1986). The binding sites for these allostericeffectors, which mainly act by altering the affinity of the enzymefor Mg-ATP, are apparently confined to the 20-kDa C-terminal partof the large subunit of the enzyme (product of gene carB, Rubio etal., 1991). Their precise location and the detailed organization ofthe functional domains of CPSase are however not known. In theframework of the analysis of the structure-function relationships ofthe E. coli CPSase we have undertaken the study of mutants affect-ed in the response to these allosteric effectors.

Among a wide variety of non biauxotrophic carB mutants,mutants sensitive towards arginine (ars) or uracil (urs) have beenselected (Mergeay et al., 1974), some of which have been sequencedand partially characterized (Delannay et al., 1995a). Two of thesemutants, carB36 (ars) and carB803 (urs) proved to bear two a.a.substitutions each (respectively ser 743 by asn plus gly 824 by aspand ala 182 by val plus ser 948 by phe). We have constructedrecombinant CPSases bearing a single a.a. substitution with theaim to determine more precisely the function of each one of thefour residues affected. The properties of the single substitutionmutant enzymes were determined by use of the reaction couplingwith aspartate carbamoyltransferase (Delannay et al., 1995b) andcompared to those of the double substitution mutants and of thewild-type.

The two single substitution mutants (S743N and G824D) aswell as the original double ars mutant carB36 produce a CPSaseaffected in its sensitivity towards ornithine; the affinity is decreasedby a factor 9 in all three enzymes. In the presence of UMP theapparent affinity of the G824D enzyme for Mg-ATP is stronglyreduced (S0.5 is increased 5-fold as compared to the wild-type); theenzyme is therefore extremely sensitive to inhibition by UMP. Asevere increase in the Vmax for the overall reaction is observed inthe presence of ornithine. The double CPSase mutant shares thischaracteristic as well.

As described previously (Delannay et al., 1995a) the urs carB803mutant (A182V and S948F) is totally insensitive towards purine andpyrimidine nucleosides. This effect can now be attributed to the solesubstitution of ser 948 by a phe; indeed, this single mutant CPSaseexhibits the same allosteric properties as the double mutant whereasthe most N-proximal mutation (A182V) affects the apparent affinityof the enzyme for bicarbonate only.

Our results therefore confirm the importance of the C-terminalpart of the large subunit of CPSase in the allosteric control. More-over they strongly suggest that the binding site for ornithine isdistinct from that for UMP and IMP.

This work was supported by the Belgian Fund for Joint Basic Research (con-tract n° 2.9007.92).

ReferencesCUNIN, R., GLANSDORFF, N., PIÉRARD, A. & STALON, V., (1986) Microbiol. Rev.

50, 314-352.DELANNAY , S., KOSLOVA, E., CHARLIER, D., STALON, V. & PIÉRARD, A. (1995a)

Arch. Physiol. Biochim. 103, B8.DELANNAY , S., STALON, V., CHARLIER, D. & PIÉRARD, A. (1995b) Arch. Physi-

ol. Biochim. 103, B9.MERGEAY, M., GIGOT, D., BECKMAN, J., GLANSDORFF, N. & PIÉRARD, A. (1974)

Molec. Gen. Genet. 133, 299-316.RUBIO, V., CERVERA, J., LUSTY, C., BENDALA, E. & BRITTON, H. (1991) J. Biol.

Chem. 261, 11320-11327.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B39SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

The application of RAPD markers for the identificationof Rhododendron species

J. DE RIEK, S. DE SCHEPPER, M. MERTENS, E. VAN BOCKS-TAELE, J. HEURSEL and M. DE LOOSE (Department of Ap-plied Plant Genetics and Breeding, Centre for Agricultural Re-search, 9820 Merelbeke, Belgium)

Molecular markers have the potential to reveal similari-ties or differences in the genome of related plant species.Randomly amplified polymorphic DNA (RAPD) markers aregenerated by the polymerase chain reaction. Short oligonu-cleotide primers anneal on several sites in the genomic DNAand serve as a starting site for Taq-polymerase, amplifyingthese fragments. After different cycles of amplification, thefragments are visualised by gel electrophoresis. Polymor-phic fragments (markers) originate from the presence or lackof a primer-binding site or from inserts/deletions betweentwo primer-binding sites. RAPD markers can be of great helpfor the identification of plant species or varieties and to makethe breeding process more efficient. Molecular markers canbe determined at regular intervals over the chromosomes.This enables an objective and accurate identification of vari-eties since chromosomes are more homogeneously covered.Moreover, by carrying out crossings and segregation experi-ments, DNA markers can be isolated that are linked withimportant genes for breeding purposes.

As a first step to implement the above mentioned in Azal-ea breeding, RAPD patterns were assessed for six Rhododen-dron species. Some of these species were frequently used inthe past in crosses that made the current commercial Azaleaassortment (R. indicum, R. mucronatum, R. scabrum and R.simsii), other are related species (R. kiusianum and R. noria-kianum).

Genomic DNA was extracted from young leaves, groundto powder in liquid nitrogen, with a guanidine thiocyanatebased lysis buffer. The DNA was further purified using standardprotocols. Ten different decamer primers (Operon Technolo-gies Inc.) were evaluated for their capacity to regenerate RAPDpolymorphism’s. PCR conditions were as described by DeLoose et al. (1993). PCR products were separated on a TAEbuffered 1.5% agarose gel and were stained with ethidiumbromide. All primers allowed the generation of polymorphicDNA fragments. For the six most informative primers – i.e.those generating most of the polymorphism’s between thespecies – the DNA fingerprints were combined and analysedwith the GELCOMPAR software. In this way, the DNA fin-gerprint for each species consisted of approximately 50 loci.Cluster analysis was performed on these data, using the UPG-MA algorithm with the Jaccard coefficient. The six specieswere grouped in a dendrogram showing the correlation be-tween distinct groups. The highest correlation, approximate-ly 62%, was observed between R. scabrum and R. mucrona-tum. This value is still low enough to allow a distinct separa-tion between the two species. This shows that identificationof Rhododendron species can be effectively performed usingRAPD markers, resulting from a limited set of primers. Thegrouping based on a DNA fingerprint with a limited numberof bands will seldom be indicative for phylogenetic relation-ships between species.

In further research, RAPDs in combination with the morepowerful AFLP technique will be used to generate a libraryof DNA fingerprints discriminative for Azalea varieties.

ReferenceDE LOOSE, M., VAUTERIN, L., REHEUL, D. & VAN BOCKSTAELE, E. (1993)

Identification of rye-grass varieties using RAPD markers. Med. Fac. Land-bouww. Univ. Gent, 58/4b.

Multiple sequence alignment strategies

P. DE RIJK and R. DE WACHTER (Department of Biochemistry,University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Bel-gium)

For the maintenance of a large database of aligned ribos-omal RNA (rRNA) sequences several programs and strate-gies were developed. Since sequence alignment is a basictask in the comparative study of any molecule, these pro-grams are generally applicable. Alignments can be used amongothers to deduce structural models for related sequences, andto create phylogenetic trees.

The raw sequence data containing rRNA are obtainedfrom the Belgian EMBNet Node. Using tkConvers, the rRNAsequences and their reference data are filtered out and con-verted to the formats used by the program DCSE (DedicatedComparative Sequence Editor). They are then distributed overthe large and small subunit rRNA databases.

DCSE (De Rijk & De Wachter, 1993) is a tool to helpobtaining multiple sequence alignments. By the extensiveuse of dynamic memory allocation almost any size of pro-tein, DNA or RNA alignment can be edited. tkDCSE has aneasy-to-use, yet flexible and programmable X-windows in-terface with a complete on-line hyper text help. The programis easily extensible, and the interface can be adopted to spe-cific needs. External programs can be seamlessly integrated.

In DCSE, sequences can be aligned using a combinationof automatic and manual methods. A new sequence is usuallyautomatically aligned to a closely related sequence or set ofsequences (profile alignment) that is already aligned to allother sequences in the database. In this way, the new pair-wise alignment is incorporated within the existing sequencealignment. The automatic alignment can then be manuallyrefined simply by shifting characters or blocks of characters.The colour display of different characters gives a strong visu-al clue to the correctness of an alignment. Pattern searching,sequence grouping and refined protection mechanisms alsoaid in attaining an optimized alignment.

Comparative study of a molecule showing a particularspatial structure, such as proteins or ribosomal RNA, is of-ten an iterative process. The higher-order structure of a mol-ecule can to a certain extent be deduced from a multiplesequence alignment. The alignment in its turn can be fine-tuned by taking into account the structure of the studied mol-ecule. tkDCSE has special support for this kind of studies.Both RNA and protein secondary structure can be incorpo-rated for each sequence within the alignment by using spe-cial symbols. Structural elements can also be coloured differ-ently. An incorporated RNA structure can be checked forconsistency. Other functions, such as searching a comple-ment, or the creation of a base-pairing matrix can help infinding a plausible structure model.

tkDCSE is currently available for IRIX on Silicon Graphicsmachines, Solaris 2 on Suns, and Linux on PCs. The homepage for DCSE is available on our WWW server at URLhttp://rrna.uia.ac.be/~peter/dcse/. The LSU and SSU data-bases are available from http://rrna.uia.ac.be/.

ReferenceDE RIJK, P. & DE WACHTER, R. (1993) Comp. Appl. Biosci. 9, 735-740.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B40 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Investigation of the role of L-lactate dehydrogenase Thr246:structural and functional characterisation of the T246Wmutant

DEVOS, D., VINALS, C., DE BOLLE, X. and FEYTMANS, E.(Laboratoire de Biologie Moléculaire Structurale, Unité de Re-cherche en Biologie Moleculaire, FUNDP, 61 rue de Bruxelles,5000 Namur, Belgium)

The homotetrameric enzyme L-lactate dehydrogenase (L-LDH) catalyzes the reversible reduction of pyruvate into L-lactate, regenerating the NAD+ coenzyme. The enzyme L-LDH is highly stereospecific and is totally unable to catalyzethe oxidation of D-lactate. The molecular study of this spe-cificity should allow us to increase our knowledge of therelationship between structure and function in enzyme cata-lytic sites.

For this purpose, we investigated the role played by theThr246 residue in the catalysis reaction of L-LDH. Thr246 isan important active site residue (Bur et al., 1989) conservedin all known L-LDH sequences. The X-ray structure of theBacillus stearothermophilus L-LDH suggests that the hydroxylgroup of Thr246 can form a hydrogen bond with the carboxylgroup of the substrate analog oxamate (Wigley et al., 1992).It has also been found that the hydroxyl group of Thr246plays a crucial role in promoting substrate inhibition(Sakowicz et al., 1992). Furthermore, the lateral chain of theThr246 is assumed to reduce the volume of this part of theactive site. Another possible explanation of the Thr246 cata-lytic role is that it could activate the dihydronicotinamidenucleus of the NADH cofactor.

To investigate the role played by the Thr246, we con-structed several mutant L-LDH three-dimensional models bymolecular mechanics and dynamics, starting from the modelof rabbit muscle L-LDH. The analysis of this model led us toassume that the introduction of a Trp in position 246 does notmodify the conformation of the essential residues involved inthe catalytic reaction.

From this basis, we introduced the T246W mutation inthe nucleotidic sequence of the rmL-LDH. The kinetic char-acterisation of the mutant enzyme allowed us to investigatethe role of Thr246 in catalysis. The optimal pH for catalysisis unmodified, the pyruvate kinetic constants are slightly al-tered, and no lactate oxidation activity was detected. Also,the NADH KM is higher in the mutant enzyme. These resultsconfirm the role of the Thr246 in NADH stabilisation andhighlight the role of the Thr246 in lactate binding.

ReferencesBUR, D., CLARKE, T., FRIESEN, J.D., GOLD, M., HART, K.W., HOLBROOK, J.J.,

JONES, J.B., LUYTEN, M.A., WILKS, H.M. (1989). Biochem. Biophys. Res.Com. 161: 59-63.

SAKOWICZ, R., KALLWASS, H., PARRIS, W., GOLD, M., JONES, J. (1992). Biochem.Biophys. Res. Commun. 182: 1309-1312.

WIGLEY, D.B., GAMBLIN , S.J., TURKENBURG, J.P., DODSON, E.J., PIONTEK, K.,MUIRHEAD, H., HOLBROOK, J.J. (1992). J. Mol. Biol 223(1): 317-335.

Molecular cloning and expression of the Physarum actin-fragmin kinase (AFK) reveals a novel type of protein ki-nase

L. EICHINGER, M. SCHLEICHER, J. VANDEKERCKHOVE1 and J.GETTEMANS1 (Institute for Cell Biology, Ludwig-Maximilians-University, 80336 München, Fed. Rep. Germany and *FlandersInstitute of Biotechnology, Department of Biochemistry, Univer-sity of Gent, Ghent, B-9000 Belgium).

The actin-fragmin kinase (AFK), isolated from the uni-cellular myxomycete Physarum polycephalum, phosphorylatesactin at residues Thr203 and Thr202 only when actin is asso-ciated with fragmin (Gettemans et al., 1992), an endogenousCa2+-dependent actin-binding protein. Phosphorylation of actinin the actin-fragmin complex results in: 1) inhibition of itsactin-nucleating activity (= inhibition of formation of newactin microfilaments) and 2) the fact that the F-actin (+)-endcapping activity of the complex becomes dependent on Ca2+

(Gettemans et al., 1995).The amino-acid sequence of tryptic peptides and V8 AFK

peptides (Gettemans et al., 1993; Waelkens et al., 1995) wereused for designing degenerate oligonucleotide primers.Through PCR, Physarum cDNA library screening and RACE(Rapid Amplification of cDNA Ends), the cDNA clone cod-ing for the full sequence of the AFK was isolated. Analysis ofthe primary structure revealed no significant similarity withpreviously published kinase sequences. However, we identi-fied 6 ‘kelch’ repeats that comprise nearly the complete C-terminal half of the enzyme. Based on the similarity with thekelch repeats from viral neuraminidase we predict that the C-terminal half of the AFK assumes a β-propeller (β-flower)configuration. A linker region, characterized by a high con-tent of Ser and Pro residues, separates the N-terminal fromthe C-terminal half of the molecule. In addition, 6 nearlyperfect P-loops were identified that may be involved in nu-cleotide binding.

To prove the intrinsic phosphorylating activity of the AFK,we expressed recombinant enzyme 1) in a rabbit reticulocytelysate using the pT7-7 expression vector with full lengthAFK cDNA or 2) after transformation into E. coli. Inductionof enzymatic activity was followed by measuring phospho-rylation of actin-fragmin in the lysates. We purified the re-combinant protein from E.coli to homogeneity and confirmedits identity by N-terminal sequencing on Pro-Blott. Takentogether, the results suggest that the AFK may represent theprototype of a new family of protein kinases.

This work was supported by grants from the Concerted Research Action of theFlemish Community (GOA), the Flanders Institute for Science and Technolo-gy (VLAB-COT) and the Flanders Institute of Biotechnology (VIB) to J.V.and J.G., and by grants from the Deutsche Forschungsgemeinschaft and theEuropean Union to M.S.

ReferencesGETTEMANS, J., DE VILLE, Y., VANDEKERCKHOVE, J. & WAELKENS, E. (1992)

EMBO J. 11, 3185-3191.GETTEMANS, J., DE VILLE, Y., VANDEKERCKHOVE, J. & WAELKENS, E. (1993)

Eur. J. Bioch. 214, 111-119.GETTEMANS, J., DE VILLE, Y., WAELKENS, E. & VANDEKERCKHOVE, J. (1995) J.

Biol. Chem. 270, 2644-2651.WAELKENS, E., GETTEMANS, J., DE CORTE, V., DE VILLE, Y., GORIS, J., VANDEKER-

CKHOVE, J. & MERLEVEDE, W. (1995) Advan Enzyme Regul. 35, 199-227.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B41SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Cold adaptation of enzymes: the case of an antarctic fish(Paranotothenia magellanica) trypsin

S. GENICOT and Ch. GERDAY (Laboratory of Biochemistry, Uni-versity of Liège, Sart Tilman B6, 4000 Liège, Belgium)

Temperature is an external environmental factor whichplays an important role in the process of the adaptation oforganisms as it affects both the structure and the stability ofproteins. As shown by Hochachka and Somero (1973), en-zymes from ectotherms living in cold environments, such asAntarctic sea waters, are better catalysts than their mesophilicand homeothermic counterparts. Up to now, little is knownabout the molecular basis of the adaptation of these enzymesto low temperature.

As part of a study devoted to the conformational peculi-arities of cold enzymes, we have isolated and characterisedtrypsin from an Antarctic fish. Over the temperature range 0to 30°C, this enzyme displays a much higher physiologicalefficiency than that of homologous enzymes isolated frommesophilic and thermophilic organisms (Genicot et al., 1988).This is a common characteristic of enzymes produced bypsychrophilic organisms. A model of the three-dimensionnalstructure of the Antarctic fish enzyme was built, based on thehigh structural isology existing within the chymotrypsin ser-ine protease family. The program QUANTA (Molecular Simu-lation, Inc., Burlington, MA, USA), implemented on a Sili-con Graphic workstation was used for the modelling. Tenserine proteases, the tertiary structure of which has been solvedat a resolution of 2Å or better, were selected as referencestructures. Amongst them, bovine trypsin was used as prima-ry reference structure as it shows the higher sequence identi-ty (62%) with the fish enzyme. The main chain coordonates,i.e. the N, (α, C and O atoms positions of the Antarctic fishmodel were also copied from the bovine enzyme. Where iden-tical in sequence to the bovine trypsin or to one of the otherreference proteases, side-chain conformations were directlycopied into the fish model. Non homologous side-chain con-formations were built using the principle of the maximal over-lap (Summers et al., 1987) or the minimum perturbation ap-proach (Shi et al., 1985). The coordonates of the fish modelwere finally minimised using the Adopted Basis Newton Raph-son algorithm.

Among the 85 amino-acid substitutions observed betweenthe fish enzyme and the bovine one, 77 involve surface resi-dues. Such changes are not expected to change the tertiaryfold to a great extent. Nevertheless, the lack of one salt bridgeand several hydrogen bonds and the overall decrease in hy-drophobicity are some characteristics of the fish enzyme thatcould enhance its flexibility as compared to the bovine one.

ReferencesGENICOT, S., FELLER, G. & GERDAY, CH. (1988) Comp. Biochem. Physiol.

90B3, 601-609.HOCHACHKA, P.W. & SOMERO, G.N. (1973) in Strategies of Biochemical Adap-

tations, SAUNDERS, W.B., ed.SHI, H.H., BRADY, J. & KARPLUS, M. (1985) Proc. Natl. Acad. sci. U.S.A. 82,

1697-1700.SUMMERS, N.L., CARLSON, W.D. & KARPLUS, M. (1987) J. Mol. Biol. 196, 175-

198.

Is genetics of proto-eucaryotes possible in Escherichia coli?

J.P. GRATIA (Laboratoire de Microbiologie, ULB-CP 614, B-1070Brussels, Belgium)

Knowledge in matter of sexuality in bacteria is almostrestricted to the F-mediated conjugation and transfer of plas-mids, transposons or non-replicative DNA fragments in E.coli K12. However, complete zygotes can be formed in Ba-cillus subtilis followed by co-replication of parental chromo-somes as in fecondation by fusion of haploid germinal cells.Moreover, whereas in B. subtilis, cell fusion requires artifi-cial conversion to protoplasts and induction by PEG (Hotch-kiss & Gabor, 1980), in the E. coli K12 « Ufr » strain previ-ously described (Gratia, 1994), a fusion-like process natural-ly occurred after a short-time contact: sectored colonies wereformed on a suitable medium at a rate reaching 10 %. Thetwo forms were differentiated by parental markers locatedthroughout the whole chromosome from 1 min to 92 min.The partners of the so-called IMP+ strain (IMP for illegiti-mate mating property) were genetically labelled strains of E.coli, Shigella flexneri or even Salmonella typhimurium.

Fusion was cytologically visualized using mecillinam-rounded cells (not sphaeroplasts!). In broth with added suit-able inorganic salts, cocci formed aggregates and becamegiant syncytia-like spheres. Electron micrographs of thinsections showed remnants of membranes inside a dense cyto-plasm containing several nucleoids. (Alternatively, coccalcells were brought into contact with untreated bacteria in-duced to elongate: significant binding between spheres andfilaments was observed). Giant cells transferred to penicil-lin-free medium were unable to recover their rod shape inone step. Instead, they evolved through a long series of trans-formations including monstrous branched forms. These wereinhibited to septate properly at the branching points and re-sembled the ftsZ mutants described by Bi and Lutkenhaus(1992). It seems that fusion had altered septation. The fre-quency of identified zygotes remained the same whether ornot bacteria were exposed to mecillinam.

Like B. subtilis exfusants, the E. coli (or Shigella) clonesissued from a zygote were either recombinants or non-com-plementing diploids. In the latter case, both parental chro-mosomes were able to replicate but only one was expressed.Functional genetic complementation between chromosomescarrying heterologous auxotrophic mutations was only tran-sient. At varying rates, a parental type switched from onemode of expression to the other whatever the number of markersand the distances between them. A recessive mutation wasapparent, in spite of heteroallelic conditions, without segre-gation having occurred.

The question arises as to the relationships between bothobservations, fusion and chromosome inactivation. Such asituation might have existed in ancestors of haploid proca-ryotes and diploid eucaryotes. It appears to be appropriate tostudy mutation in diploids and to analyse factors influencingthe mutation rate, directly or indirectly. Factors which « in-duce » a mutation in the case of a non-complementation (theX-chromosome in mammals is another example of it) mightin fact trigger the reactivation of a silent chromosome carry-ing a pre-existent mutation.

Part of this work was supported by the Fonds DEFAY. The author thanks Dr.D. DEKEGEL for electron microscopy.

ReferencesBI, E. & LUTKENHAUS, J. (1992). J. Bacteriol. 174, 5414-5423.GRATIA, J.P. (1994). Res. Microbiol. 145, 309-325.HOTCHKISS, R. & GABOR, M. (1980). Proc. Natl. Acad. Sci. USA 77, 3553-3557.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B42 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Cytokine profile of human myelin basic protein reactiveT cells

G. HERMANS1,2, P. STINISSEN1, L. HAUBEN1, J. RAUS1,2 andJ. ZHANG1,2 (1MS Research and Immunology Dept., Dr. L. Wil-lems-Instituut and 2Limburgs Universitair Centrum, Universi-taire Campus, B-3590 Diepenbeek, Belgium)

Myelin basic protein (MBP) reactive T cells have been im-plicated in the pathogenesis of multiple sclerosis (MS) (Zhang,1995). In experimental autoimmune encephalomyelitis, theanimal model of MS, the cytokine profile of myelin specific T-cell clones determines their pathogenic potential as only Th1myelin specific T-cell clones are able to induce the disease(Ando et al., 1989). So far, it has not been clear whether MBPreactive T cells in MS display a similar Th1 phenotype.

In the initial setting, a panel of human MBP-reactive T-cellclones were stimulated with MBP and supernatants were har-vested after a predetermined optimal incubation time of 72hours. The cytokines IL-2, IL-4, IL-6, IL-10, IFNg and TNFawere quantified using commercially available ELISA kits (Bio-source Inc., Camarillo, California). In parallel experiments, thestimulation indexes were determined by standard 3H-thymi-dine uptake assays. The proinflammatory cytokines IFNγ andTNFα were produced by virtually all clones tested, irrespec-tive of whether the clones were derived from MS patients orcontrol subjects. In contrast, IL-4 was produced by most MS-derived clones (27/49), but only by 5 out of 23 clones derivedfrom control subjects (χ2 = 5.77, P < 0.02). Furthermore, MS-derived clones produce higher concentrations of all cytokinestested as compared to controls, with the notable exception ofIL-6. Of the 72 clones tested, 71 displayed a cytokine profilenot matching typical murine Th1/2 patterns. However, consid-erable coregulation was found between the Th1-associated cy-tokines IFNg, TNFa and IL-2 but not with or between the Th2-associated IL-4, IL-6 and IL-10. Remarkably, the level of se-cretion of IFNγ and TNFα correlated well with the prolifera-tive responses to MBP. In contrast, an inverse correlation wasobserved between proliferative response and secretion of theinhibitory cytokine IL-10.

In a second experiment, the cytokine profile of short-termMBP reactive T-cell lines was studied in order to avoid a possi-ble bias in cytokine profile due to long-term culture. PBMCfrom 19 MS patients and 17 healthy controls were plated atlimiting dilution conditions and stimulated with MBP. A totalof 4320 short-term lines were tested for cytokine secretion inresponse to MBP. Therefore, the cultures were split and resti-mulated either with autologous MBP-pulsed PBMC or hen egglysozyme (HEL) pulsed PBMC as an irrelevant control anti-gen. 531 lines produced significantly more IFNg and/or IL-4 inresponse to MBP than HEL after 48 hours of incubation andwere therefore considered as MBP reactive. More lines wereMBP-reactive in MS patients as compared to healthy controls(22.3% of all wells in MS vs 11.8% in healthy controls). Anincreased number of MS-derived lines produce IL-4 (χ2

= 13.3,P < 0.002). The MS-derived lines also produce higher amountsof IL-4 (78 pg/ml vs 48 pg/ml).

In conclusion, MBP reactive T cells in MS patients canproduce IFNg and TNFa in larger quantities as compared tohealthy controls, and IL-4 production was more prominent inboth T-cell clones and primary lines derived from MS patients.

We acknowledge the technical assistance of K. Engelen, D. Mercken, C. Bock-en and E. Smeyers, and thank Dr. R. Medaer for providing patient bloodsamples. This work was supported by grants from the ‘Nationaal Fonds voorWetenschappelijk Onderzoek’ (NFWO), ‘Stichting Wetenschappelijk Onder-zoek in Multiple Sclerose’ (WOMS) and the ‘Nationale Loterij’. GH is a recip-ient of a fellowship grant from the ‘Vlaams Instituut ter Bevordering van hetWetenschappelijk-technologisch Onderzoek in de Industrie’

ReferencesANDO, D. G., CLAYTON, J., KONO, D., URBAN, J. L., & SERCARZ, E. E. (1989)

Cell. Immunol. 124, 132-143.ZHANG, J. (1995) in Current Neurol. Appel, S. H. ed., vol. 15, pp. 115-155.

Mosby-Year Books, St. Louis.

Bacillus subtilis phage 2C thermosensitive mutants, sub-stituting hydroxymethyluracil by thymine in their DNA:effect on transcription during the lytic cycle

P. HOET, G. DAXHELET and P. GILLET (Intern. Inst. of Cellul.Pathology, Unit of Microbial Pathogenesis, University of Lou-vain Medical School, B-1200-Brussels, Belgium)

Phage 2C is a lytic phage of B. subtilis, having hy-droxymethyluracil (hmUra) replacing thymine (T) in its ge-nome (Hoet et al., 1992). Little is known of the metabolicadvantages of this substitution, in particular for the cascadeof regulatory factors ensuring efficient transcription of early,middle and late promoters (Stewart, 1993). Here we reporton the isolation of thermosensitive 2C mutants, which couldbe subdivided in three classes: 1) mutants having 75% re-placement of hmUra by thymine in the restrictive conditions;2) mutants unable to inhibit B. subtilis DNA synthesis in thecourse of infection; 3) mutants with impaired viral DNA bio-synthetic machinery (DNA- mutants).

Different techniques led to these conclusions: differen-tial binding of chromofluorophores with hmUra- and T-con-taining DNA (Daxhelet et al., 1989; 1990), quantitation ofviral DNA by hybridization and buoyant density analysis oflabelled DNA in CsCl gradients. Transcription during infec-tion by a phage having DNA with partial replacement by Twas reduced three-fold. Restriction fragments carrying earlypromoters were transcribed with delay under the restrictiveconditions. These experiments show clear transcriptional ad-vantages for hmUra-containing DNA.

ReferencesDAXHELET, G.A., COENE, M.M., HOET, P.P. & COCITO, C.G. (1989) Anal. Bio-

chem. 179, 401-403.DAXHELET, G.A., KOHNEN, M.M., COENE, M.M. & HOET. P.P. (1990) Anal.

Biochem. 190, 116-119.HOET, P.P., COENE, M.M. & COCITO, C.G. (1992) Annu. Rev. Microbiol. 46, 95-

116.STEWART, C.R. (1993) In Bacillus subtilis and other Gram-positive bacteria

SONENSHEIN, A.L., HOCH, J.A. & LOSICK, R., eds. American Society forMicrobiology, Washington, D.C. pp. 813-829.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B43SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

The IL-6/IL-6R binding interface. Proposed mechanismof interaction

M. KALAI 1,2, F.A. MONTERO-JULIAN 3, J. GRÖTZINGER4, V. FON-TAINE1, P. VANDENBUSSCHE1, R. DESCHUYTENEER2, A.WOLLMER4), H. BRAILLY 3 and J. CONTENT1,2. (1Institut Pas-teur, Département de Virologie, Bruxelles, 2Université Libre deBruxelles, Faculté de Médecine, Laboratoire de Microbiologie,Bruxelles, Belgium. 3Immunotech S.A., Marseille, France and4Institut für Biochemie, RWTH Aachen Klinikum, Aachen,Germany)

Interleukin-6 (IL-6) is a multifunctional cytokine pro-duced by a variety of cell types. It binds first to an α-subunitreceptor, called IL-6R. The IL-6/IL-6R complex recruits twoβ-subunit, called gp130. These form a covalently linked dim-er that transduces the signal. An abnormal expression of IL-6can be related to a variety of diseases. Some of these aremultiple myeloma, rheumatoid-arthritis, post-menopausal os-teoporosis, chronic autoimmune diseases, Castleman’s dis-ease and AIDS (Akira et al., 1993).

The interaction between IL-6 and IL-6R is the initial andmost specific step in the IL-6 signalling pathway. Under-standing its mechanism at the aminoacid level may help inthe development of small and simple IL-6-inhibiting mole-cules. We decided to study the IL-6/IL-6R-binding interfaceby molecular modelling and site-directed mutagenesis.

Our model and previous results (Krütten et al., 1990;Fontaine et al., 1991,1993; Kalai et al., 1995), have suggest-ed that the IL-6/IL-6R complex could be considered as a newprotein, representing an activated form of the cytokine. Thecentre of this new protein is the interaction area between thetwo molecules consisting of an hydrophobic region predictedto account for most of the binding free energy. This hydro-phobic core is shielded by hydrophilic residues, also respon-sible for recognition. Following this hypothesis, we mutatedin the IL-6 and IL-6R molecules, a series of residues predict-ed by our model to participate in this interaction area and tointeract with each other. We studied the capacity of the mu-tants to form an IL-6/IL-6R complex and to transduce the IL-6 signal. Our results show that the aromatic/hydrophobic aminoacids Phe74 and Phe78 of IL-6 are necessary for IL-6R bind-ing and for signal transduction, probably due to the interac-tion with Tyr188, Phe248 and Tyr249 of IL-6R. They alsoshow that the hydrophilic residues Glu296 and Glu297 in theIL-6R are crucial for the formation of the IL-6/IL-6R com-plex and suggest that they interact with arginines 179 and182 of IL-6. The results demonstrate that our model’s predic-tions were correct and strongly support our hypothesis.

We would like to thank L. De Wit and E. Bautens for their technical assistance.M. Kalai is a recipient of a fellowship from “Université Libre de Bruxelles”.

ReferencesAKIRA, S., TAGA, T. & KISHIMOTO, T. (1993). Adv. Immunol. 54, 1-78.FONTAINE, V., BRAKENHOFF, J., DEWIT, L., ARCONE, R., CILIBERTO, G. & CON-

TENT, J. (1991). Gene 104, 227-234.FONTAINE, V., SAVINO , R., ARCONE, R., DEWIT, L., BRAKENHOFF, J., CONTENT, J.

& CILIBERTO, G. (1993). Eur. J. Biochem. 211, 749-755.KALAI , M., MONTERO-JULIAN , F. A., GRÖTZINGER, J., WOLLMER, A., MORELLE,

D., BROCHIER, J., ROSE-JOHN, S., HEINRICH, P. C., BRAILLY , H. & CON-TENT, J. (1995). Eur. J. Biochem. (Submitted).

KRÜTTEN, A., ROSE-JOHN, S., MÖLLER, S., WROBLOWSKI, B., WOLLMER, A.,MÜLLBERG, J., HIRANO, T., KISHIMOTO, T. & HEINRICH, P. C. (1990). FEBSLett. 262, 323-326.

Differential binding of ligands to bovine profilin isoforms

A. LAMBRECHTS, J.-L. VERSCHELDE, V. JONCKHEERE, M. GOE-

THALS, J. VANDEKERCKHOVE and C. AMPE (Flanders Inter-university Institute for Biotechnology, Department of Biochem-istry, Faculty of Medicine, University of Ghent, B-9000 Ghent,Belgium)

Profilin is an ubiquitous small actin-binding protein thatcan enhance either actin polymerisation or depolymerisationdepending on the presence of other actin-binding proteins(Pantaloni & Carlier, 1993). Other ligands of profilin arephosphatidylinositol 4,5 bis-phosphate (PIP2) and poly(L-proline). Both are important in signal transduction and thisleads to a model in which profilin has a function as mediatorin signaling to the cytoskeleton. The two mammalian iso-forms have a complementary expression level in differenttissues (Honoré et al., 1993) and similar affinity for actin(Lambrechts et al., 1995). Here we present data on the differ-ential binding of the bovine isoforms to PIP2 and proline-richsequences.

Several proteins containing oligoproline stretches in theirsequence are known. Among them are the cyclase-associatedprotein (CAP), the vasodilator-stimulated phosphoprotein(VASP) and a surface protein of the cyctotoxic pathogen Lis-teria monocytogenes (ActA). These proteins have previouslybeen postulated as ligands of profilin (Vojtek et al., 1991;Theriot et al., 1994; Reinhard et al., 1995). We assayed bind-ing of proline-rich peptides derived from these proteins, tobovine profilin I and II using fluorescence spectroscopy. Pro-filin II binds more strongly to peptides containing at least 12to 13 proline residues than does profilin I. This suggests thatCAP and VASP are potential profilin II ligands. Shorter pep-tides, such as those in ActA, show no binding to either of theprofilins, indicating that ActA is probably not a ligand.

When we assayed profilin-binding to PIP2 by microfiltra-tion we observed that profilin I has a higher affinity for thisphospholipid than has profilin II.

These results were corroborated by circular-dichroismmeasurements. We used this technique to assess tertiary struc-ture changes in the profilin isoforms upon binding of eitherPIP2 or a proline-rich peptide. The spectrum of profilin Ishows a large decrease in ellipticity after PIP2 binding, whilein the presence of the peptide the spectrum almost does notchange. In contrast, PIP2 induces only a small change in theprofilin II molecule while binding of proline-rich peptideinduces a considerable change in the molecule. This result isin agreement with molecular dynamics calculations whichalso predict larger changes in the hypothetical poly(L-pro-line) binding site of profilin II.

We thank Prof. M. Rosseneu for use of the CD spectropolarimeter. C. Ampe isa Research Associate of the Belgian National Fund for Scientific Research.This work was supported by EC grant CI1-CT93-0049 and grant 3.0008.94 ofthe F.G.W.O. to C. Ampe and grant GOA-91/96-3 to J. Vandekerckhove.

ReferencesHONORÉ, B., MADSEN, P., ANDERSEN, A. & LEFFERS H. (1993) FEBS Lett. 330,

151-155.LAMBRECHTS, A., VAN DAMME, J., GOETHALS, M., VANDEKERCKHOVE, J. &

AMPE C. (1995) Eur. J. Biochem. 230, 281-286.PANTALONI , D. & CARLIER, M.-F. (1993) Cell 75, 1007-1014.REINHARD, M., GIEHL, K., ABEL, K., HAFFNER, C., JARCHAU, T., HOPPE, V.,

JOCKUSCH, B. & WALTER, U. (1995) EMBO J. 14, 1583-1589.THERIOT, J., ROSENBLATT, J., PORTNOY, D., GOLDSCHMIDT-CLERMONT, P. &

MITCHISON, T. (1994) Cell 76, 505-517.VOJTEK, A., HAARER, B., FIELD, J., GERST, J., POLLARD, T., BROWN, S. & WIGLER,

M.(1991) Cell 66, 497-505.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B44 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Application of expression and secretion signals of the sub-tilisin inhibitor gene from Streptomyces venezuelae for ef-ficient extracellular production of mouse tumor necrosisfactor α

E. LAMMERTYN1, L. VAN MELLAERT1, A. VAN BROEKHOVEN2

and J. ANNE1 (1Rega Instituut, K.U.Leuven, Minderbroeders-straat 10, B-3000 Leuven and 2 Innogenetics N.V., IndustrieparkZwijnaarde 7, Box 4, B-9052 Gent, Belgium)

In addition to their occurrence in plant and animal cells,subtilisin inhibitors seem to be present in a wide variety ofStreptomyces species which secrete these compounds in largeamounts. This feature renders the expression and secretionsignals of these inhibitors particularly interesting for the ex-tracellular production of other homologous and heterologousproteins. Promising results have already been obtained usingthe STI1 from S. longisporus for the secretion of CD4 T cellreceptor derivatives (Fornwald et al., 1993).

We report here on the isolation of the subtilisin inhibitorfrom S. venezuelae (VSI) (Van Mellaert et al., 1996) and theapplication of its expression and secretion signals for theextracellular production of mouse tumor-necrosis factor byS. lividans. The levels of secreted biologically active mTNFobtained in this way reached the highest values thus far ob-tained for mammalian proteins produced by Streptomycesspp. This success is probably due to a combination of anefficient promoter, a well-suited signal peptide and the char-acteristics of the heterologous protein itself. The vsi-P pro-moter showed a higher capacity to initiate mRNA transcrip-tion than ermE-P1a and aph-P1, two strong Streptomycespromoters. Several mutations were introduced in the posi-tively charged N-region of the VSI signal peptide resulting insignal peptide mutants containing 0, 1, 2, 3 (wild type), 4 and5 positive charges. Analysis of secreted mTNF levels in allthese cases revealed that the production could still dramati-cally be improved by reducing the +3 charge of the wild typesignal peptide to +2. Further analysis of these mutants at themRNA level are being carried out.

L.V.M. is a postdoctoral fellow of IWT

ReferencesFORNWALD, J.A., DONOVAN, M.J., GERBER, M., KELLER, J., TAYLOR, D.P., ARCURI,

E.J. & BRAWNER, M.E. (1993). Biotechnology 11, 1031-1036.VAN MELLAERT, L., LAMMERTYN, E., PROOST, P., SABLON, E., SCARCEZ, T., VAN

BROEKHOVEN, A., VAN DAMME, J. & ANNE, J. (1996) J. Bacteriol. (submit-ted)

Computational drug design of new HIV-1 protease inhi-bitors

LEBON, F., 1V INALS, C., 1FEYTMANS, E. and DURANT, F.(Laboratoire de Chimie Moléculaire Structurale, et 1Labora-toire Biologie Moléculaire Structurale, Facultés UniversitairesNotre-Dame de la Paix, 61 rue de Bruxelles, Namur, Belgium)

Extensive work on the HIV-1 protease has been conduct-ed to synthesize bioavailable and low-molecular-weight in-hibitors (Kim et al., 1995). In this work, we propose a newstrategy to design non peptide low-molecular-weight HIV-1PR inhibitors and to evaluate in a first approximation their invitro affinity.

Our strategy started with the analysis of the interactionsbetween existing co-crystals of HIV-1 PR and several peptid-ic inhibitors (Lebon, 1994). These led to the generation of ageneral but simple pharmacophore in which we included thewater molecule which is essential for the binding of the in-hibitor into the active site (Lam et al., 1994).

The minimal elements necessary to observe inhibitionwere submitted to the Cambridge Structural Database for ageometric search (CSD Quest 3D algorithm). The small frag-ments that emerged from this search could be used as a startfor the design of new HIV-1 PR inhibitors.

Interaction energies were then evaluated in order to dif-ferentiate candidate structures. The potential energy of inter-action chosen is the simple Lennard-Jones function and thecoulombic representation of the electrostatic interactions.

We established a relationship between the potential ener-gy of interaction between receptor and ligands and their af-finity in vitro. This simple method of interaction has beenused previously among acyclic peptidomimetic inhibitors(Holloway et al., 1995). We have extended this work to avariety of known inhibitors of very different chemical na-ture.

From the CSD search, small important fragments werethen docked into the active site. The interaction with theenzyme was optimized through molecular mechanics (ener-gy minimization using steepest descent, followed by conju-gate gradients algorithms until the maximum derivative < 1kcal mol-1Å-1 and ∆E < 1 kcal mol-1, forcefield CVFF). Func-tional groups necessary to compensate the hydrophobic pocketswere branched to the core of the inhibitor in order to get thebest potential energy of interaction.

Based on this new strategy, we succeeded in designingnew de novo inhibitors with a theoretical affinity of 10-9 M.

ReferencesHOLLOWAY, M.K., WAI , J.M., HALGREN, T.A., FITZGERALD, P.M.D., VACCA,

J.P., DORSEY, B.D., LEVIN, R.B., THOMPSON, W.J., CHEN, L.J., DESOLMS,S.J., GAFFIN, N., GHOSH, A.K., GUILIANI , E.A., GRAHAM, S.L., GUARE,J.P., HUNGATE, R.W., LYLE, T.A., SANDERS, W.M., SUCKER, T.S., WIGGENS,M., WISCOUNT, C.M., WOLSERSDORF, O.W., YONG, S.D., DARKE, P.L., ZUGDY,J.A. (1995) J. Med. Chem. 38, 305-317.

K IM, E.E., BAKER, C.T., DWYER, M.D., MURCKO, M.A., RAO, B.G., TUNG, R.D.& NAVIA , M.A. (1995) J. Am. Chem. Soc. 117, 1181-1182.

LAM , P. Y. S., JADHAV, P. K., EYERMANN, C. J., HODGE, C. N., RU, Y., BACHE-LER, L. T., MEEK, J. L., OTTO, M. J., RAYNER, M. M., WONG, Y. N., CHANG,C. H.; WEBER, P. C., JACKSON, D. A.; SHARPE, T. R. & ERICKSON-VIITANEN , S. (1994) Science, 263, 380-384.

LEBON, F. Mémoire de licence (1994) Facultés Universitaires N-D de la Paix,Namur.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B45SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Ornithine carbamoyltransferase from obligate psy-chrophilic bacteria

Z. LIANG1,3, M. DEMAREZ2, C. LEGRAIN2, M. BAETENS1,3,N. GLANSDORFF 1,2,3, H.J. RÜGER 4 and T.L. TAN4 ( (1Micro-biology, Vrije Universiteit Brussel, 2Research Institute ofthe CERIA, 1 Ave E. Gryson, B-1070 Brussels, Belgium,3Vlaams Interuniversitair Instituut voor Biotechnologieand 4Institute for Polar and Marine Research, D-2850Bremerhaven, Germany).

Ornithine and aspartate carbamoyltransferases (OTCaseand ATCase respectively) have become paradigms for inves-tigations on structure-function relationships in organismsadapted to extremes of temperature and pressure. We haveexamined OTCase in three Vibrio-like, deep-sea obligate psy-chrophilic strains (2674, 2825 and 2693). The strains growin marine broth or in minimal synthetic medium supplement-ed with glucose and vitamins, with generation times of about6 h in marine broth at their optimal temperature (6 °C). Allthree OTCases display a distinctly psychrophilic activity profile(optimal temperature between 17 and 20 °C, still 20 to 50%activity at 0 °C); OTCases from strains 2674 and 2693 arehowever relatively thermostable (critical temperature about58 °C for 15 min exposition in crude extracts) whereas the2825 OTCase is very labile (critical temperature about 23°C). The three enzymes are therefore excellent material forfurther comparative investigations on the structural determi-nants of catalysis and stability over the whole range of tem-perature extending from psychrophiles to hyperthermophiles(the OTCase of Pyrococcus furiosus, optimal growth tempe-rature 102 °C, is being analyzed in our laboratory as well).The OTCase gene from strain 2693 (argF) was cloned bycomplementing an argF,I E. coli auxotroph; the recombinantenzyme has an activity profile (maximum at 17 °C) identicalto that of the native enzyme but is quite unstable in the E.coli background. From preliminary purification experiments,the native protein would appear to be a trimer composed ofidentical subunits of MW 33 kDa (as calculated from the 301residues identified in the cognate open reading frame). Thegene appears to be part of an operon encoding at least argB,argF and argG (in that order), a novel type of linkage inbacterial arginine regulons. The synthesis of 2693 OTCaseis repressed by arginine, which corroborates its involvementin arginine biosynthesis rather than catabolism.

The gene itself appears homologous to all known OT-Cases (30 to 37% identity, 55 to 61% similarity). Compara-tive modelization of OTCase from strain 2693 and from P.furiosus on the pattern of the 3D structure of the Pseudomonasaeruginosa homologous enzyme (Villeret et al., 1995) mayindicate features involved in psychrophily, while mere in-spection of known OTCase sequences does not suggest cha-racteristics outstanding in this respect. However the enzymeappears peculiar by its rather low affinity to carbamoylphos-phate and to the bisubstrate analogue d-N-phosphonoacetyl-L-ornithine. By analogy with the phaseolotoxin-insensitiveOTCase from Pseudomonas syringae (Staskawicz et al., 1980),this suggests that residues critical for the catalytic activity ofthe enzyme may differ from their counterparts in other OT-Cases.

Supported by grants from the FKFO, the EEC and the IWT.

ReferencesSTASKAWICZ, B.J., PANOPOULOS, N.J. & HOOGENRAAD, N.J. (1980) J. Bacte-

riol. 142, 720-723.VILLERET, V., TRICOT, C., STALON, V. & DIDEBERG, O. (1995) Proc. Natl. Acad.

Sci. USA 92, 10762-10766.

Hen egg white lysozyme: a temperature dependence studyof the folding process

A. MATAGNE1, S.E. RADFORD AND C.M. DOBSON (Oxford Cen-tre for Molecular Sciences, University of Oxford, New Chemis-try Laboratory, Oxford OX1 3QT, UK; 1Present address: Labo-ratoire d’Enzymologie, University of Liège, Institut de ChimieB6, B-4000 Sart Tilman, Belgium)

The folding pathway of lysozyme has been investigatedby diverse techniques, including pulse labelling hydrogenexchange methods, combined with nuclear magnetic reso-nance spectroscopy (Radford et al., 1992) and electrospraymass spectrometry (ESI-MS; Miranker et al., 1993), circulardichroism in the near and far UV (Chaffotte et al., 1992;Radford et al., 1992) and fluorescence measurements using arange of fluorescent probes (Itzhaki et al., 1994). Coupledwith rapid mixing methods, these various spectroscopic tech-niques have allowed a detailed model for the folding of henlysozyme to be proposed (Dobson et al., 1994; Radford &Dobson, 1995).

We have undertaken the study of the temperature depend-ence of this process. The folding kinetics of lysozyme weremonitored by stopped-flow optical methods over a tempera-ture range from 2˚C to 50˚C and the change of the variousrate constants was measured as a function of the temperature.The results show that simple Arrhenius behaviour is not ob-served, and the overall rate of folding decreases above 40˚C.Furthermore, hydrogen exchange labelling experiments, com-bined with ESI-MS, indicate that the stability of the transientintermediate on the major folding pathway is dependent oftemperature. We bring evidence that the “fast” and “slow”phases detected during stopped-flow absorbance, intrinsic flu-orescence and far UV CD experiments are related to the for-mation of that a-domain intermediate. These experiments shedlight on the different factors determining the folding proper-ties of the protein.

We would like to thank E. Chung, G. Howarth, and Dr. C. Robinson for per-forming the mass spectroscopy experiments. A.M. was the recipient of anEMBO long term fellowship.

ReferencesCHAFFOTTE, A.F., GUILLOU, Y. & GOLDBERG, M.E. (1992) Biochemistry 31,

9694-9702.DOBSON, C.M., EVANS, P.A. & RADFORD, S.E. (1994) Trends Biochem. Sci. 19,

31-37.ITZHAKI , L.S., EVANS, P.A., DOBSON, C.M. & RADFORD, S.E. (1994) Bioche-

mistry 33, 5212-5220.MIRANKER, A., ROBINSON, C.V., RADFORD, S.E., APLIN, R.T. & DOBSON, C.M.

(1993) Science 262, 896-900.RADFORD, S.E., DOBSON, C.M. & EVANS, P.A. (1992) Nature 358, 302-307.RADFORD, S.E. & DOBSON, C.M. (1995) Phil. Trans. R. Soc. Lond. B348, 17-

25.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B46 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

IMGT (ImMunogGeneTics database), an integrated sys-tem for the antigen-receptor nucleic sequences and theirannotations

I. MOUGENOT1,2, P. DÉHAIS3 AND T. LIBOUREL1 (1Laboratoired’Informatique, de Robotique et de Microélectronique, Montpellier; 2Lab-oratoire d’Immunogénétique Moléculaire, Montpellier, France and 3Lab-oratorium voor Genetica, via the Department of Genetics, affiliatedto the Flanders Interuniversity Institute for Biotechnology, UniversiteitGent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium)

Here, we report the experience established during the creation of aninteractive system applied to the specific problems of immunogenetics (spe-cialized branch of genetics). The success of “genome projects” such as thehuman genome HGP (Pearson & Söll, 1991) and sequencing projects of themodel organisms Escherichia coli and Caenorhabditis elegans (Sulston etal., 1992) is strictly related to the computer science and inevitably requires aclose collaboration between biologists and computer scientists. In this con-text, we built a specialized database for antigen-receptor nucleic sequencesdedicated to the biological community.

This creation is completely justified by the complex specific geneticmechanisms which govern antigen receptor nucleic sequences synthesis andneed a special expert evaluation. In this way, various communities collabo-rate: LIRMM (Laboratoire d’Informatique, de Robotique et de Microélec-tronique, Montpellier, France), LIGM (Laboratoire d’ImMunoGénétiqueMoléculaire, Montpellier, France), CNUSC (Centre National UniversitaireSud de Calcul, Montpellier, France), EMBL (European Molecular BiologyLaboratory, Cambridge, England).

The challenge is to adapt the models available in computer science(database conceptual models, formal grammars) to the specific area of bio-logical data with the objective of structuring these data. The methodology(Mougenot, 1995; Mougenot et al., 1995) presented here is based on theinteraction of concepts stemming from computer science (databases, soft-ware engineering, artificial intelligence, human – computer interaction) andbiology (antigen receptor genetics).

The methodology includes several steps. (i) The analysis of the biolog-ical domain under consideration: the aim here is to extract the knowledgefrom the biological sequences. We proceeded by the design of an ontology(in the context of knowledge, sharing an ontology means a description of theconcepts and relationships that can exist in the domain; Gruber, 1993):definition of a consensual vocabulary which describes sequences subregionsand a key word tree. (ii) Formal specifications allow the translation of thesequence-specific constraints. The genetic sequences can be considered as acollection of motifs (or patterns) of which the composition and position aremore or less determined. We have written a formal BNF-type grammar(Mougenot, 1995) which takes into account this particularity. (iii) The con-ceptual model (Déhais & Mougenot, 1994) (database scheme) integrates the“core data” (nucleic sequence, citations, taxonomy,..) and annotations (def-initions, comments, key words, feature descriptions,..). (iv) For the imple-mentation of our project, we have decided to use acknowledged standardsfor both information storage and user interface. Thus, we have chosen SY-BASE as database management system because of its robustness, its per-formance, and especially for its relational model (Codd, 1970) and its querylanguage SQL (Date, 1988) which are both wide spread.

The grammar-based syntactic analyzer is implemented in Prolog. Weuse the X-Windows environment (Scheifler et al., 1986) for its graphicalcapabilities, its portability and its network transparency for the user. Fur-thermore, IMGT is available on the Web (http://www.ebi.ac.ik/contrib/imgt/top_ imgt.html). (v) The phase of expert validations is particularly impor-tant. The results obtained have to be validated by the experts (biologists). Ifnecessary, the conception can be refined.

The novelty of the work is twofold. On the one hand, we propose amodel to represent the nucleic sequences of antigen receptor emerging fromthe confrontation on the different formalisms existing (entity association,relational, object-oriented) and the biological thought. On the other hand,the biological expert evaluation expressed through the “annotations” is ana-lyzed in order to arrive at a consensus on their standardization. This consen-sus then allows their direct integration into the model. Note that while ourapproach was done in the specific field of immunogenetics, it can easily begeneralized to many other biological domain.

ReferencesCODD, E.F. (1970) Commun. ACM 13, 377-387.DATE, C.J. (1988). A guide to the SQL standard. Addison Wesley, Reading,

MA.DÉHAIS, P. & MOUGENOT, I. (1994) Proc. HICSS Conference (Hawai) 5, 25-34.GRUBER, T. (1993) Toward principles for the design of ontologies used for

knowledge sharing. Technical report, Standford Knowledge Systems Lab-oratory, Palo Alto, CA, USA.

MOUGENOT, I. (1995) LIGM-DB un système de gestion intégré des séquencesnucléiques des récepteurs d’antigènes et de leurs annotations. Thèse,Université des Sciences et Techniques du Languedoc, Montpellier II,Montpellier France.

MOUGENOT, I., DÉHAIS, P. & LIBOUREL, T. (1995) Proc. DASFAA Conference,Singapore 95, 333-341.

PEARSON, P. & SÖLL, D. (1991) FASEB J. 5, 35-39.SCHEIFLER, R.W., GETTYS, J. & NEWMAN, R. (1986). X-Windows System. Dig-

ital Press, Bedford, MA.SULSTON, J. et al. (1992) Nature 356, 37-41.

Hydrophobic cluster analysis of the major structural pro-tein of pestivirus

G. MUYLDERMANS, M. CHAMEKH, L. WYNS and R. HAMERS

(1Laboratorium voor Ultrastructuur, 2Algemene Biologie, VrijeUniversiteit Brussel, Vlaams Interuniversitair Instituut voor Bio-technologie, Paardenstraat 65, B-1640 Sint-Genesius-Rode, Bel-gium)

The genus Pestivirus belongs to the Flaviviridae family(Westaway et al., 1985). In this genus, three viruses : Classi-cal Swine Fever Virus (CSFV), Bovine Viral Diarrhea Virus(BVDV) of cattle and Border Disease Virus (BDV) of sheepshow strong structural and serological similarities (Darby-shire, 1960). RNA sequence data revealed approximately90% aminoacid homology between viruses of the same spe-cies compared with 70% homology between different spe-cies (Muyldermans et al.,1993). The most variable geneticregions were found in the genes encoding E2, the major anti-genic structural protein of the pestivirus. Comparison ofpestiviral E2 sequences revealed that the cysteine residuesand the N-glycosylation sites are conserved. However, noconserved or non-conserved regions were detected. The mu-tations in aminoacid sequences of the E2 protein were dis-tributed throughout the protein.

We used a hydrophobic cluster analysis in order to evalu-ate structural domains and to identify the conserved structur-al domains that are essential to maintain the overal structureof the protein. These results were related to some knownantigenic determinants and residues important for neutrali-zation of the virus.

ReferencesDARBYSHIRE, J.H. (1960) Vet. Rec. 72: 331.MUYLDERMANS, G., SAN GABRIEL, M.C., CAIJ, A., DE SMET, A., & HAMERS, R.

(1993) Arch Virol. 132: 429-435WESTAWAY, E.G., BRINTON, M.A., GAIDAMOVISCH, S.Y.A., HORZINEK, M.C.,

IGARASHI, A., KAÄRIAINEN , L., LVOV, D.K., PORTERFIELD, J.S., RUSSELL,P.K. & TRENT, D.W. (1985) Intervirology 24: 125-139

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B47SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Homotropic and heterotropic interactions in Pseudomonasaeruginosa catabolic ornithine carbamoyltransferase: as-sociation of concerted homotropic cooperative interac-tions and local heterotropic effect

NGUYEN, V.T.1, TRICOT, C.2 and STALON V.1 (1Laboratoire deMicrobiologie, Faculté des Sciences, Université Libre de Bru-xelles, Brussels and 2Institut de Recherches du CERIA, B-1070Brussels, Belgium)

Pseudomonas aeruginosa has an anabolic (argF) and a cata-bolic (arcB) carbamoyltransferases. Despite extensive similar-ities of sequence, these enzymes function unidirectionaly invivo. The anabolic OTCase participates in arginine synthesisand catalyzes in vivo the thermodginamically favoured reac-tion, the synthesis of citrulline and Pi. The catabolic OTCasepromotes the reverse reaction in vivo. Homotropic cooperativi-ty and low affinity for carbamoylphosphate prevent catabolicOTCase to perform the anabolic reaction in vivo. The catabolicenzyme activity is heterotropically regulated by spermidine, aninhibitor, and by AMP which acts as an activator.

Analysis of the catabolic OTCase crystal structure (Villeretet al., 1995) revealed that N and C terminal residues of thepolypeptide chain are involved in trimer-trimer contacts withinthe dodecamer. Therefore these residues are probably involvedin the signal communication pathways for homotropic interac-tions occuring between the carbamoylphosphate catalytic sites,and also for heterotropic interactions induced by effectors suchas spermidine or AMP. In order to understand the molecularbasis of the homotropic and heterotropic interactions in OT-Case, structural studies must be correlated with additional in-formations on the kinetics of the wild-type and mutant en-zymes. The possible role for these residues in the binding ofeffectors has to be investigated.

Amino acid substitution of glutamate 105 (located in thehydrophobic core of the carbamoylphosphate domain, adjacentto the active site) by alanine produces an enzyme which isdevoid of homotropic interactions between catalytic sites forcarbamoylphosphate and which is blocked in an active confor-mation. This altered form of the enzyme is still sensitive to theallosteric effectors AMP and spermidine; their action do notrestore homotropic cooperativity.

N-terminal amino acids are involved in trimers interfaceswithin the dodecamer and therefore are expected to play a rolein the signal transmission pathway of both homotropic andheterotropic interactions and/or in the binding of effectors. Re-placement of the 27 N-terminal residues of the wild-type en-zyme by the homologous 26 amino acids of the anabolic OT-Case from E. coli (a trimeric and a non modulable enzyme)induces a modification of quaternary structure, resulting in atrimeric enzyme which retains reduced homotropic cooperativ-ity. Activation by AMP and inhibition by spermidine of thischimaeric enzyme do not affect CP homotropic cooperativity.The activator reduces only the concentration of substrate athalf maximum velocity while spermidine acts in the inverseway.

Another chimaeric enzyme consisting of the 42 N terminalamino acid residues of the E. coli anabolic OTCase followedby the C-terminus of the catabolic OTCase shows no responseat all towards AMP: this indicates that the N-terminal region ofthe polypeptide chain is involved in AMP binding; yet thismodified enzyme still exhibits CP homotropic cooperativityand spermidine sensitivity. These observations indicate that ho-motropic and heterotropic interactions are disconnected in themutant enzyme and must involve different mechanisms. On theother hand, the extend of stimulation by phosphate is consider-ably reduced for this chimaeric enzyme when compared to thewild-type enzyme, indicating that, besides its binding to thecatalytic site, phosphate may also act at a regulatory site identi-cal to or overlapping the AMP allosteric site.

ReferencesVILLERET, V., TRICOT, C., STALON, V. & DIDEBERG, O. (1995) Proc. Natl Acad.

Sci. USA 92 : 23, 10762-10766.

Design of new potential inhibitors for HIV1 protease di-merisation

NOLLEVAUX , F., VINALS, C., DE BOLLE, X. AND FEYTMANS, E.(Unité de Recherche en Biologie Moléculaire, Laboratoire deBiologie Moléculaire Structurale, Facultés Universitaires No-tre-Dame de la Paix, rue de Bruxelles 61, B-5000 Namur, Bel-gium).

The homodimeric aspartyl-protease of the human immu-nodeficiency virus 1 (HIV1) is essential for the maturation ofnew viral particles. The inhibition of the dimerisation of thisenzyme is able to prevent the processing of gag and gag-polviral polyproteins and so interrupt the viral life cycle (Kohlet al., 1988). The interface between the two subunits includesmainly an extended four-stranded b-sheet which consists ofthe interdigitating N- and C- terminal residues of both mono-mers (Weber, 1990).

Two compounds that can be potential inhibitors of theHIV1 protease dimerisation have been designed by molecu-lar modelling. The first one is composed by the two terminalpentapeptides of a monomer (Pro1-Leu5 and Cys95-Phe99)chemically linked between the amine group of Pro1 and theSH group of Cys95. The interpeptidic link has been extendedin the second inhibitor by a cysteine before Cys95 and by ahistidine before Pro1.

The starting structures of the monomer/inhibitor and in-hibitor/inhibitor complexes have been constructed on the ba-sis of the cartesian coordinates of the HIV1 protease dimer(PDB code: 1HHP, Spinelli et al., 1991). These initial struc-tures, as well as the protease dimer, have been submitted toin vacuo molecular dynamics calculations (100 ps, 600K,dielectric constant = r) using the CVFF forcefield (Dauber-Osguthorpe et al., 1988).

The potential energy of interaction has been computedfor each inhibitor associated with a HIV1-protease mono-mer. These values are only 15% (25 kcal mol-1) lower thanthe interaction energy of the dimer β-sheet. The interactionenergy of an inhibitor/inhibitor complex is neverthelesssimilar to that of a monomer/inhibitor complex. This indi-cates a possible formation of inhibitors dimers that can inter-fere with the inhibition of HIV1 protease dimerisation. Thereis no significant difference between the computed interactionenergies for the two compounds proposed.

ReferencesDAUBER-OSGUTHORPE, P., ROBERTS, V.A., OSGUTHORPE, D.J., WOLFF, J., GENEST,

M. & HAGLER, A.T. (1988) Proteins : Structure, Function and Genetics 4,31-47.

KOHL, N.E., EMINI , E.A., SCHLEIF, W.A., DAVIS L.J., HEIMBACH, J.C., DIXON,R.A.F., SCOLNICK, E.M., & SIGAL, I.S. (1988) Proc. Natl. Acad. Sci. USA85, 4686-4690.

SPINELLI, S., LIU, Q.Z., ALZARI , P.M., HIREL, P.H. & POLJAK, R.J. (1991) Bio-chimie 73, 1391-1396.

WEBER, I.T. (1990) J. Biol. Chem. 265, 10492-10496.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B48 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Topology prediction of Brucella abortus Omp2b porin

PAQUET, J.-Y., VINALS, C., DE BOLLE, X., TIBOR, A., DEVOS,D., DENOEL, P., LETESSON, J.-J. and FEYTMANS, E. (Unitéde Recherche en Biologie Moléculaire, Facultés UniversitairesNotre-Dame de la Paix, rue de Bruxelles 61, B-5000 Namur,Belgium)

Porins are transmembrane proteins, located in the Gram-negative bacteria outer membrane. They form a β-barrel thatallows diffusion of small hydrophilic solutes towards the peri-plasm. The E. coli porin tridimensional (3-D) structures arewell-characterized. In the intracellular pathogenic bacteriaBrucella abortus, two outer-membrane proteins (Omp2a etOmp2b), sharing no sequence homology with known porins,show porin-like properties (i.e. pore-forming activity andtrimeric structure; Marquis & Ficht, 1993). Omp2a and Omp2bhave 96% identities over their shared lenght but Omp2a is 41amino acids shorter becaused of a deletion in the middle ofthe gene (Ficht et al., 1989). Only Omp2b has been shown tobe expressed in B. abortus. In vivo, porins probably mediateexchanges between Brucella sp. and its environment and theirregulation might contribute to Brucella sp. survival in infect-ed cells.

Here we describe a model of Omp2b topology based onmultiple sequence alignment, secondary structure predictionand threading.

We first performed a multiple sequence alignment ofOmp2b and Omp2a from various Brucella species and twoRhizobium sp. outer-membrane proteins. We then submittedthe alignment to the Predict Protein Server PHD (Rost &Sander, 1994), that has been previously tested for the se-condary structure prediction of porins of known structure.The predicted secondary structure of Omp2b contains 16 β-strands separated alternatively by short turns and long loops.Using the PHD-based threading Predict Protein Server, theE. coli PhoE porin of known structure was found to have asimilar secondary structure pattern. From this sequence/struc-ture alignment the transmembrane b-strand orientations werepredicted. As in the case of porins of known structure, thelarge loops are oriented to the exterior of the cell while theshort turns point towards the periplasm. Using the Gibbsmotif-sampling algorithm (Neuwald et al., 1995), the posi-tion and orientation of the β14 β-strand could be confirmed.

According to the topological model of Omp2b, a RGDmotif is located on the external loop L6, a position compati-ble with the proposed role of this RGD motif in the adher-ence of B. abortus to mononuclear phagocytes (Campbell etal., 1994).

The Omp2a deletion of 41 amino acids would remove theβ6 and β7 predicted β-strands and Omp2a would thereforeform a 14 b-strand barrel which might explain the functionaldifferences between Omp2a and Omp2b.

The Omp2b topological model will be tested by site-spe-cific deletion mutagenesis, and a 3-D model will be elaborat-ed in order to investigate the Omp2b structure/function rela-tionship.

J.Y. Paquet is a recipient of a fellowship from “Fonds pour la Formation à laRecherche dans l’Industrie et dans l’Agriculture”.

ReferencesCAMPBELL, G.A., ADAMS, L.G. & SOWA, B.A. (1994) Vet. Immunol. Immuno-

path. 41, 295-306.FICHT, T.A., BEARDEN, S.W., SOWA, B.A. & ADAMS, L.G. (1989) Infection and

Immunity 57, 3281-3291.MARQUIS, H & FICHT, T.A. (1993) Infection and Immunity 61, 3785-3790.NEUWALD, A.F., LIU, J.S., LAWRENCE, C.E. (1995) Protein Sci. 4, 1618-1632.ROST, B. & SANDER, C. (1994) Proteins 19, 55-72.

Purification and characterisation of L-methionine- S-me-thyltransferase from green barley malt

M.J. PIMENTA1, C. LIEGEOIS1, J.P. DUFOUR2 and Y. LARON-

DELLE 1 (1Unité de la Biochimie de la Nutrition, UniversitéCatholique de Louvain, Place Croix du Sud 2/Bte 8, B-1348Louvain-la-Neuve et 2Food Science Department, OTAGO Uni-versity, P.O. Box 56, Dunedin, New Zealand)

S-methylmethionine (SMM) is a modified amino acidpresent in different plants. Its physiological role is currentlystill speculative. SMM is produced from methionine and S-adenosylmethionine in a reaction catalysed by the S-adeno-syl-L-methionine: L-methionine S-methyltransferase (EC2.2.1.12). A rapid, simple and specific method has been ap-plied to measure this enzyme activity (Pimenta et al., 1995).

The enzyme has been identified in several plants and re-cently purified to apparent homogeneity and characterisedfrom leaves of Wollastonia biflora (James et al., 1995).

The purpose of our work is to purify and characterize thisenzyme in barley malt, since SMM is of utmost importancefor technological points of view, as it is the major precursorof the dimethylsulfide (DMS) off-flavour in beer (Anness &Bamforth, 1982).

In this communication we present a 480-fold purificationof the enzyme from green malt by PEG precipitation, anion-exchange chromatography, and affinity chromatography onadenosine agarose as well as a kinetic analysis of substrateinteraction and product inhibition patterns. This study allowsus to propose a mnemonical mechanism, with ordered fixa-tion of substrates (first S-adenosylmethionine and secondmethionine) and ordered liberation of products (first S-ade-nosylhomocysteine and second SMM), including a comple-mentary step with a ’’dead-end’’ complex (E-S-adenosylme-thionine-SMM). A molecular mass of 200 kDa was obtainedfor the enzyme using gel filtration.

From these results we can conclude that the methyltrans-ferase from green barley malt is relatively different from thatof Wollastonia biflora with respect to molecular mass (200 kDa-450 kDa) and to kinetic mechanism (mnemonical-classicalordered Bi Bi).

The research carried out by M.J. Pimenta was supported by the EuropeanCommunity (AIR program).

ReferencesANNESS, B.J. & BAMFORTH, C.W. (1982) J. Inst. Brew. 88, 244-252.JAMES, F., NOLTE, K.D. & HANSON, A. (1995) J. Biol. Chem. 270, 167-169.PIMENTA, M.J., VANDERCAMMEN, A., DUFOUR, J.P. & LARONDELLE, Y. (1995)

Anal. Biochem. 225, 167-169.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B49SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Chaodynamics and biomatics, new ways to compute bio-logical parameters

P. J. QUERINJEAN (BIOCLUB® n.p.m.a. B-1340 Ottignies-LLN(Belgium)

Chaodynamics (Andrey, 1989; Querinjean, 1993, Quer-injean & Andrey, 1994) represents a new way to computebiological parameters. If one analyzes individual time-seriesof a given parameter, short to medium term predictability,i.e. up to several months, of the health and disease evolutionmay be obtained.

Oscillatory parameters exhibit a periodical variation char-acteristic of a healthy situation. Several months before anyother sophisticated diagnostic methodology, a chaotic varia-tion might be detected by adequate softwares.

Applications of this approach has already given interest-ing results in the interpretation of ECG and EEG. It seemsthat without disposing of similar number of data, some sero-logical significant parameters may predict the evolution ofsome diseases such as cancers. The search for the most sig-nificant parameters in different pathological situations opensthe way to another understanding of the health and diseaseprocesses.

This new biocomputing technique applied to serologicaldata presents several attractive features:- statistically, the full meaning of a test is obtained by com-

parison of the subject’s own new value with the previousdata of the same parameter, reducing the number of out-liers. The subject is his own auto-reference. No new so-phisticated technologies are required as the existing onesallow a long term data recording and the iterative chaody-namical calculation will be carried out as an automatic up-dating system ;

- economically, only a few simple tests are sufficient to givea clear individually predictive answer for many cancer typesand the cost of numerous biological analyses carried outwhen the patient enters the hospital will largely cover theprice of annual or semi-annual simple tests for a long peri-od of time ;

- psychologically, the patient, if educated appropriately, willdedramatize several deadly pathologies, reducing the healthcare costs even further.

This computer approach will help to develop and confirmseveral models elaborated as biomatical explanations of so-called “spontaneous” cancer remission and genesis process-es (Querinjean, 1992) leading to a systemic health approachas requested by many health experts (Salomon, 1991).

ReferencesANDREY, L. (1989) Med. Hypoth., 28, 143-144.QUERINJEAN, P.J. (1992), Proc. 13th Int. Congr. on Cybernectics,, 894-898.QUERINJEAN, P.J. (1993) Proc. XVIth UICC Int. Cancer Congr., Delhi, Mon-

duzzi, Bologna, vol. 4, 2971-2974.QUERINJEAN, P.J. & ANDREY, L. (1994), Proc. 2nd Sun, Moon and Living

Matter Int. Workshop , Bratislava, pp. 58-62.QUERINJEAN, P.J. (1994), Proc. 2nd Sun, Moon and Living Matter Int.

Workshop, Bratislava, 105-108.SALOMON, J.C. (1991), Le tissu déchiré, Propos sur la diversité des cancers, p.

19, Seuil, Paris.

Orientation of apolipophorin-III a-helices in dimyristoyl-phosphatidylcholine discs

V. RAUSSENS1, V. NARAYANASWAMI 2, E. GOORMAGHTIGH1, R.O.RYAN2 and J.-M. RUYSSCHAERT1 (1Free University ofBrussels, Campus Plaine CP 206/2, B-1050 Brussels and 2Uni-versity of Alberta, Edmonton, Alberta, Canada T6G 2S2)

Apolipophorin-III (apoLp-III) is a hemolymph proteinfrom the sphynx moth Manduca sexta that exists as a globu-lar lipid-free protein or as a lipid surface-associated apolipo-protein. Apolp-III is a polypeptide of 166 amino acids whichexists as a monomer in solution and is rich in a-helix (Cole etal., 1987). In comparison with the three-dimensional struc-ture of the apoLp-III from L. migratoria (Breiter et al., 1991),the M. sexta apoLp-III likely contains five long amphipathica-helices in a bundle connected by short loops. Associationof apoLp-III with dimyristoylphosphatidylcholine vesiclespresumably leads to a conformational change which exposesthe hydrophobic faces of the helices to the lipids, leading tothe formation of uniform lipid discs. These discs contain 6molecules of apoLp-III. Geometrical calculations based onthese data have allowed the presentation of a model wherethe a-helices of the apoLp-III orient perpendicular to the phos-pholipid chains and surround the lipid disc (Wientzek et al.,1994).

In order to provide experimental evidence, we used po-larized Fourier-transform-attenuated total-reflection infraredspectroscopy to gain information about the secondary struc-ture (Goormaghtigh et al., 1990) and orientation (Goor-maghtigh & Ruysschaert, 1990) of apoLp-III with respect tothe lipid bilayer. The results on the secondary structure of theprotein in the apoLp-III:DMPC complexes show a mainly a-helical structure (70-80%) and are in good agreement withthe CD measurements (Ryan et al., 1993). The results ob-tained from the linear dichroic spectra of the protein providethe first experimental evidence of an orientation of the apoLp-III helical domains perpendicular to the lipid acyl chains, atopology opposite to that described for the apoA1 (Brasseuret al., 1990; Hefele Wald et al., 1990).

ReferencesBRASSEUR, R. DE MEUTTER, J., VANLOO, B., GOORMAGHTIGH, E. & RUYSS-

CHAERT, J.-M. (1990) Biochim. Biophys. Acta 1043, 245-252.BREITER, D.R., KANOST, M.R., BENNING, M.M., WESENBERG, G., LAW, J.H.,

RAYMENT, I. & HOLDEN, H.M. (1991) Biochemistry 30, 603-608.COLE, K.D., FERNANDO-WARNAKLASURIYA , G.J.P., BOKUSKI, M.S., FREEMAN,

M., GORDON, J.I., CLARK, W.A., LAWW, J.H. & WELLS, M.A. (1987) J.Biol. Chem. 262, 11794-11800.

GOORMAGHTIGH, E. & RUYSSCHAERT, J.-M. (1990) in “Molecular descriptionof biological membrane components by computer aided conformationalanalysis”, R. BRASSEUR ed., CRC Press, 285-329.

GOORMAGHTIGH, E., CABIAUX , V. & RUYSSCHAERT, J.-M. (1990) Eur. J. Bio-chem. 193, 409-420.

HEFELE WALD, J., GOORMAGHTIGH, E., DE MEUTTER, J., RUYSSCHAERT, J.-M. &JONAS, A. (1990) J. Biol. Chem. 265, 20044-20050.

RYAN, R.O., OIKAWA , K. & KAY, C.M. (1993) J. Biol. Chem. 268, 1525-1530.WIENTZEK, M., KAY, C.M., OIKAWA , K. & RYAN, R.O. (1994) J. Biol. Chem.

269, 4605-4612.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B50 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Gene Prediction in Arabidopsis thaliana: Genomic Se-quences

P. ROUZÉ1,2, S. ROMBAUTS2, G. VAN LAERE2, L. VAN WIE-MEERSCH2 AND M. VAN MONTAGU2 (1Laboratoire associé del’Institut National de la Recherche Agronomique (France), Uni-versiteit Gent, B-9000 Gent, Belgium and 2Laboratorium voorGenetica, via the Department of Genetics, affiliated to the Flan-ders Interuniversity Institute for Biotechnology, Universiteit Gent,K.L. Ledeganckstraat 35, B-9000 Gent, Belgium)

The identification of genes inside genomic sequences in not astraightforward task for higher-eukaryote nuclear genomes as it maybe in procaryotes or even in yeast where the information content isdense and where no or very few introns disrupt the coding sequen-ces. The most common and easiest way to find genes is to searchsequences databases for homologs. This approach is limited by theamount and diversity of the sequences available today. This is ofcourse also valid for Arabidopsis thaliana, the model plant chosenfor exhaustive sequencing of its 100-Mb genome, where the analy-sis of expressed sequence tags (ESTs) have shown that at best 40%of them have homologs (Höfte et al., 1993)

Therefore, programs have been developed to predict genes fromthe intrinsic knowledge of the genomic sequence, most of them forthe human genome. Whatever their ability to predict genes in ge-nomes for which they were developed, they mostly failed to do soconfidently on A. thaliana sequences. This comes from variations ingenome ‘style’ between organisms, which imply that genome-spe-cific tailoring is a prerequisite for reliable gene prediction.

We first made use of GeneMark (GM), a program looking forcoding potential by statistics based on Markov chain models (Boro-dovsky et al., 1994), using At specific transition matrices. GM wasdeveloped and shown to be very successful for gene prediction inprocaryotes. Early trials with GM on actual Arabidopsis genomiccontigs have shown its ability to identify genes, shown as clusters ofpredicted exons in different frames of the same strand, separated byintergenic sequences shown as areas with no prediction. One of thestrong points of GM is the very confident positioning of each pre-dicted exon in a particular frame. Nevertheless, GM has clear draw-backs: small exons are not predicted (or give a weak signal), whichis a concern for the At genome where small exons occur frequently,and the predicted exons boundaries are badly defined.

NetPlantGene (NPG) was thus developed together with SørenBrunak’s team to predict splice sites location, using a neural-net-work program similar to NetGene previously developed for verte-brates sequences (Brunak et al., 1991) for which a dataset of A.thaliana genes checked for consistency and non-redundancy hasbeen used for training and testing. NPG is now available as a e.mailserver on Internet for exon/intron prediction in A. thaliana and,more generally, in dicots genomic sequences (see Rouzé et al., 1996).

Other gene prediction programs have more recently been adapt-ed to Arabidopsis sequences, and also made available on Internet ase.mail and www servers. These are the Gene-Finder (GF) (Solovyevet al., 1995) and the GRAIL (Xu et al., 1994, also available as X-client) suite of programs. Their individual performances have beenchecked, and compared to NPG (for splice sites prediction) and toGM (for exons predictions). Briefly, NPG and GF perform clearlybetter than GRAIL for splice sites prediction. Gene modelling of-fered by GRAIL and GF is still poor, both programs missing abouthalf of the exons, with the additional problem, for GRAIL, of falseexon prediction and of wrong border assignment.

These programs are tested in Ghent for the A. thaliana genomicsequences obtained in the frame of the ongoing EU ESSA genomeproject. Further improvements will be made to integrate GM andNPG which have complementary features under a common tool,and to adapt more tightly the prediction to the different types of A.thaliana genes, after their classification.

ReferencesBRUNAK, S., ENGELBRECHT, J. & KNUDSEN, S. (1991) J. Mol. Biol. 220, 49-65.BORODOVSKY, M., KOONIN, E.V. & RUDD, K.E. (1994) Trends Biochem. Sci. 19,

309-313.HÖFTE, H. et al. (1993) Plant J. 4, 1051-1061 [5, 611].ROUZÉ, P., TOLSTRUP, N. & BRUNAK, S. (1996) Arch. Physiol. Biochem., this

issue.SOLOVYEV, V.V., SALAMOV, A.A. & LAWRENCE, C.B. (1995) in ISMB-95: Proceed-

ings Third International Conference on Intelligent Systems for MolecularBiology (RAWLINGS, C., CLARK, D., ALTMAN , R., HUNTER, L., LENGAUER, T.& W ODAK, S. eds). AAAI Press, Menlo Park, CA, pp. 367-375.

XU, Y., MURAL, R.J. & UBERBACHER, E.C. (1994) Comput. Appl. Biosci. 10,613-623.

NetPlantGene, a neural network program combining lo-cal and global information for efficient prediction of splicesites in Arabidopsis thaliana genomic sequences

P. ROUZÉ1, N. TOLSTRUP2 AND S. BRUNAK2 (1Laboratoire asso-cié de l’Institut National de la Recherche Agronomique (France),Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent,Belgium and 2Center for Biological Sequence Analysis, DTU, B-206, DK-2800 Lyngby, Denmark)

The primary transcript of most nuclear protein-coding genesof plants are interrupted by introns, in the same order of magni-tude as observed in metazoa. Albeit far less is known specificallyabout splicing in plants, compared to vertebrates and yeasts, it isbelieved that the mechanism is largely similar. Nevertheless, plantintrons have peculiarities: the splicing consensus appears moredegenerate than in vertebrates, there is no branch-point consen-sus, plant introns are A/U rich, and should be such to be spliced,there is no polypyrimidine tract at the 3’ end of plant introns, butoften poly-T tracts. Finally, it has been shown that vertebratepre-mRNAs are usually not spliced in plants. This explains thefailure of prediction programs built on human genome data topredict plant splice sites confidently.

Using an approach similar to NetGene used for prediction ofhuman splice sites (Brunak et al., 1991), we choose to develop aneural network program trained with, and adapted to, the Arabi-dopsis thaliana genome, that was called NetPlantGene (NPG).

Great efforts were taken to prepare a high quality dataset,from 999 A. thaliana entries from release 87 of Genbank. Weselected only the entries having full-length CDS and containingat least two introns, without conflicts or uncertainties in featuresor descriptions. The entries dealing with more than one genewere split, and checked as above. Every one of the remaining167 entries were then checked for data consistency: uninterrupt-ed CDS, appropriate size of introns, local compliance of thesplice sites to the general consensus rule (exon/GT-intron-AG/exon). Doing this, a number of suspicious entries were found,that turned out, after looking at each of them, to be errors (orheavy suspicion of errors) for most of them, with the exceptionof a few authentic 5’ intron borders (1%) for which GT werereplaced by GC. Most of the errors were bad numbering of splicesites, with equal shifting on both sites, resulting from intronborder localization using uncritically alignment with cDNA, forwhich there is often several solutions. As a whole, 45 out of 878donor sites, and 44 out of 880 acceptor sites were found corrupt.

Finally, the dataset was checked for redundancy, and similargenes removed. This cleaned dataset, comprising 144 entries inwhich 146 genes with 764 donor and 766 acceptor sites (Korninget al., 1996) was used for 2/3 for training the network, and for 1/3 for testing it.

The networks used were of the perceptron type, one operat-ing on a larger sequence window sensing coding and non codingregions, and one on a narrower window operating for donor andacceptor sites, separately. The local and global sequence net-works exploit different mechanisms to detect the splice-site lo-cation, some of them being apparent after analysis of the net-works hidden units. The final prediction makes use of the outputof the two types of network, the global network controlling thethreshold of the local networks, as developed for NetGene.

The prediction performance of the global network was foundto compare favorably when compared to GeneMark on the testset. NPG as a whole was found to offer high confident predictionfor most 5’ splice sites, and to predict at the same confidencelevel a lower fraction of the 3’ sites. On the test set, NPG wasshown to perform clearly better than the ACeDB GeneFinderprogram (P. Green, unpublished data), after having provided thisprogram the training set used by NPG, and on a limited set ofnew sequences better than the GRAIL-2 splice-site prediction,and to a similar extent compared to the ASPL program from theGene-Finder suite (see Rouzé et al., 1996).

ReferencesBRUNAK, S., ENGELBRECHT, J. & KNUDSEN, S. (1991) J. Mol. Biol. 220, 49-65.KORNING, P.G., HEBSGAARD, S.M., ROUZÉ, P. & BRUNAK, S. (1996) Nucleic

Acids Res., in press.ROUZÉ, P., ROMBAUTS, S., VAN LAERE, G., VAN WIEMEERSCH, L. & VAN MON-

TAGU, M. (1996) Arch. Physiol. Biochem., this issue.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B51SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Arginine specific repression in Bacillus stearothermophi-lus: characterization of the promoter-operator region ofthe argCJBD operon

A. SAVCHENKO1, D. CHARLIER2,5, M. DION1, P. WEIGEL1, J-N.HALLET1, S. BAUMBERG3, N. GLANSDORFF2,4,5 and V.SAKANYAN 1 (1Laboratoire de Biotechnologie, Faculté des Sci-ences et des Techniques, Université de Nantes, rue de laHoussinière 2, F-44072 Nantes Cedex 03, France, 2ResearchInstitute of the COOVI-CERIA, Av. E. Gryson 1, B-1070 Brus-sels, Belgium, 3Department of Genetics, University of Leeds,Leeds LS2 9JT, United Kingdom, 4Erfelijkheidsleer en Micro-biologie, Vrije Universiteit Brussel, Brussels, Belgium and 5VlaamsInteruniversitair Instituut voor Biotechnologie)

Mechanisms of gene regulation have not yet been studiedextensively in thermophilic bacteria and consequently infor-mation on protein-DNA interactions in these organisms isstill scarce. Arginine biosynthesis can be regarded as a mod-el system to study gene regulation in thermophilic bacteriasince a large body of data is already available for their meso-philic counterparts (Glansdorff, 1996) and since the path-way has already been investigated in some of their archaealand bacterial representatives (Van de Casteele et al., 1990;Sakanyan et al., 1992; Legrain et al., 1995). We have shownpreviously that the argCJBD gene cluster of B. stearother-mophilus is subjected to arginine specific repression (Sak-anyan et al., 1993).

By colony hybridization we have cloned the promoterproximal part of the B. stearothermophilus argC gene andthe adjacent control region of the cluster on a 4.1-kb PstI-SalI fragment. The argC gene is preceeded by a promoterregion displaying the -10 (TATAAT) and -35 (TTGATT) se-quences recognised by the major s factors of B. subtilis andE. coli, the startpoint of which was determined by primerextension. Overlapping the promoter we identified a sequencesimilar to the arginine operators of B. subtilis and E. coli.Construction of an argC-lacZ gene fusion allowed us to showthat the argC promoter of B. stearothermophilus is stronglyinhibited by the heterologous B. subtilis arginine repressorAhrC when introduced in E. coli cells. The E. coli argininerepressor ArgR however can not repress the argC promoter.This is not surprising since it was already known that B.subtilis AhrC interacts with the E. coli operators rather effi-ciently but in contrast, E. coli ArgR does not inhibit the argCpromoter of B. subtilis (Smith et al., 1989). This may be dueto stringent requirements of the ArgR repressor for an exactspacing between adjacent ARG boxes in E. coli operators(Charlier et al., 1992).

By in vitro mobility-shift and DNase-I footprinting ex-periments we have shown that the B. stearothermophilus argCoperator-like sequence tightly and specifically interacts withpure hexameric AhrC repressor in the presence of arginineonly. It appears therefore very likely that in B. stearother-mophilus the expression of the argCJBD operon is modulat-ed by a repressor that is the thermophilic homolog of AhrC.

ReferencesCHARLIER, D., ROOVERS, M., VAN VLIET, F., BOYEN, A., CUNIN, R., NAKAMURA ,

Y., GLANSDORFF, N. & PIÉRARD, A. (1992) J. Mol. Biol. 226; 367-386.GLANSDORFF, N. (1996) in E. coli and S. typhimurium: Cellular & Molecular

Biology. NEIDHARDT et al. eds. American Society for Microbiology. Wash-ington DC.

LEGRAIN, C., DEMAREZ, M., GLANSDORFF, N. & PIÉRARD, A. (1995) Microbiol-ogy 141, 1093-1099.

SAKANYAN , V., KOCHIKYAN, A., METT, I., LEGRAIN, C., CHARLIER, D., PIÉRARD,A. & GLANSDORFF, N. (1992) J. Gen. Microbiol. 138, 125-130.

SAKANYAN , V., CHARLIER, D., LEGRAIN, C., KOCHIKYAN, A., METT, I., PIÉRARD,A. & GLANSDORFF, N. (1993) J. Gen. Microbiol. 139, 393-402.

SMITH, M.C.M., CZAPLEWSKI, L., NORTH, A.K., BAUMBERG, S. & STOCKLEY,P.G. (1989) Mol. Microbiol . 3, 23-28.

VAN DE CASTEELE, M., DEMAREZ, M., LEGRAIN, C., GLANSDORFF, N. & PIÉRARD,A. (1990) J. Gen. Microbiol. 136, 1177-1183.

Molecular genetic analysis of Charcot-Marie-Tooth neu-ropathy type 2 (CMT2) families with chromosome1p35-36 and 3q13-22 markers

P. SPOELDERS1, A. LÖFGREN1, P. DE JONGHE1, E. NELIS1, J.-J.MARTIN2, C. VAN BROECKHOVEN1 and V. TIMMERMAN 1

(1Laboratories for Neurogenetics and 2Neuropathology, BornBunge Foundation, Department of Biochemistry, University ofAntwerp (UIA), B-2610 Antwerpen, Belgium)

Charcot-Marie-Tooth disease (CMT) is the most com-mon inherited motor and sensory neuropathy. The neuronalform of this disorder is referred as Charcot-Marie-Tooth dis-ease type 2 (CMT2). CMT2 is usually inherited as an auto-somal dominant trait with a variable age of onset. Nerveconduction velocities are slightly reduced or normal. CMT2was shown to be genetically heterogeneous.

The gene responsible for one form of CMT2 (CMT2A)maps to the distal part of the short arm of chromosome 1(Ben Othmane et al.,1993). This locus seems to be a minorlocus for CMT2 since in most CMT2 families linkage of thedisease with 1p35-36 markers can be excluded. Recently, asecond locus that predisposes to CMT2 (CMT2B) has beenlocated on the long arm of chromosome 3 (Kwon et al.,1995).

Highly polymorphic genetic markers were used to per-form a genetic linkage analysis in 11 unrelated CMT2 fami-lies. Two-point linkage analyses were performed using theMLINK programme of the FASTLINK computer package.

Genotype analysis was performed at the chromosome 1p35-36 loci D1S244 (AFM220yf4), D1S160 (MIT-MS48), D1S228(AFM196xb4) and D1S170 (MIT-COS37). Only one familyshowed suggestive evidence for linkage to 1p35-36(Timmerman et al.,1996). Next we analysed linkage withchromosome 3 markers D3S1769 (CHLC.GATA8DO2.431),D3S1551 (AFM198yc1) and D3S1744 (vsheffie.16.gata3c02).Positive, non-significant lod score results were obtained inthree families with D3S1769 and in two families with D3S1744.One family showed suggestive evidence for linkage withD3S1551 (z = 1.272 at θ = 0.00). Additional short tandemrepeat polymorphic markers need to be analysed in order toevaluate linkage with the 3q13-22 region.

ReferencesBEN OTHMANE, K., MIDDLETON, L.T., LOPREST, L.J., WILKINSON,K.M., LENNON,

F., ROZEAR, M.P., STAJICH, J.M., GASKELL, P.C., ROSES, A.D., PERICAK-VANCE, M.A. & V ANCE, J.M. (1993) Genomics 17: 370-375

KWON, J.M, ELLIOTT, J.L, YEE, W., IVANOVICH , J., SCAVARDA, N.J., MOOl-SIN-

TONG, P.J. & GOODFELLOW, P.J. (1995) Am. J. Hum. Genet. 57: 853-858TIMMERMAN , V., DE JONGHE, P., SPOELDERS, P., SIMOKOVIC, S., LÖFGREN, A.,

NELIS, E., VANCE, J., MARTIN, J.-J. & VAN BROECKHOVEN, C. (1996) Neu-rology (in press)

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B52 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Use of chimeric aspartate transcarbamylases to define therole of discrete structural elements in allosteric regulation

Ch. SWARUPA RANI , F. VAN VLIET, R. CUNIN, M. WALES andJ.R. WILD (Laboratorium voor Erfelijkheidsleer en Microbio-logie, Vrije Universiteit Brussel , Research Institute of the CE-RIA and Vlaams Interuniversitair Instituut voor Biotechnologie,Brussels, Belgium and 1Dept of Biochemistry and Biophysics,Texas A&M University, College Station, USA)

Aspartate transcarbamylase (ATCase) catalyzes the first stepof de novo pyrimidine synthesis. In Escherichia coli and otherenteric bacteria, this step is a major target for the control of themetabolic flow in the pathway. Therefore, ATCase is a highlyregulated enzyme with its genetic expression and catalytic activ-ity modulated by the concentration of various nucleotide effec-tors. The best studied ATCase is that of E. coli (E.C 2.1.3.2).This enzyme consists of two trimers of identical catalytic chains(c3) held in association through their interaction with three dim-ers of regulatory chains (r2) bearing the binding sites for nucle-otide effectors [2(c3):3(r2)] (Honzatko et al., 1982). E. coli AT-Case exhibits homotropic cooperative interactions between cata-lytic sites for the binding of the substrate aspartate; E. coli AT-Case also shows heterotropic interactions between regulatoryand catalytic sites. Its catalytic activity is subject to feedback-inhibition by CTP and UTP and competing activation by ATP.

The systematic study of the ATCases of a variety of entericbacteria has revealed a conservation of the [2(c3):3(r2)] quater-nary structure. These enzymes present however considerabledifferences in the cooperativity of substrate binding and in theirallosteric regulation by nucleotide effectors Previous studies onhybrid enzymes established that the regulatory subunits deter-mine the type of allosteric regulatory properties (Wales et al.,1993). A comparison of the primary sequences of regulatorypolypeptides from ATCases showing divergent patterns of allo-steric regulation revealed three major blocks of higher sequencedivergence . Two of these regions are involved in contacts withthe catalytic subunits in E. coli ATCase.

We exchanged blocks of divergent amino-acid residues be-tween E. coli, S. marcescens and P. vulgaris ATCases. We reporthere on the influence of divergent sequences at the C-terminalregion of the regulatory polypeptide. In E. coli ATCase, thisregion is part of one of the interfaces (rl-c4) between the regula-tory and catalytic polypeptides and its integrity is essential forboth homotropic cooperativity and heterotropic regulation bythe nucleotides (Newton & Kantrowitz, 1990; Xi et al., 1990; DeStaercke et al., 1995). We found that the rl-c4 interface consti-tutes an essential element for the recognition and the transmis-sion of the ATP regulatory signal in both native and chimericATCases. The rl-c4 interactions formed by the residues of the146 to 154 region of the regulatory chain could be a direct targetfor the ATP regulation, but these interactions might also propa-gate the ATP signal through the interface towards the aspartate-binding site. The specific aminoacid composition of this regionand its interactions with the aminoacid residues in the 240s loopof c4 appear to modulate the amplitude of the enzymes responseto ATP. In the P. vulgaris ATCase, other structural features be-sides the C-terminal region of the regulatory chain seem to con-tribute to the very strong activation of the enzyme by ATP.

The participation of the C-terminal region of the regulatorychain in the regulation by CTP is much less clear. In the ATCaseswhere CTP acts as an heterotropic activator, its signal appears tofollow a route distinct from that of ATP. It was discovered in thisstudy that UTP, in the presence of CTP, sends a specific inhibito-ry signal to the catalytic sites. This signal appears to be differentin the various ATCases investigated, involving the C-terminalregion of the regulatory chain in the E.coli and S. marcescensATCases, but not in P. vulgaris ATCase.

ReferencesDE STAERCKE, L., VAN VLIET, F., XI, X-G., RANI , C.S., LADJIMI , M., JACOBS, A., TRINIOLLES,

F., Herve, G., Cunin, R. (1995) J. Mol. Biol. 246, 132-143.HONZATKO et al. (l982) J. Mol. Biol. 160, 219-263.NEWTON & K ANTROWITZ (1990) Proc. Natl Acad. Sci. USA, 87, 2307-2316.WALES et al. (1993) in Biocatalyst Design for Stability and Specificity, M.E. HIMMEL & G.

GEORGIOU, eds., ACS Symposium Series No 516, American Chemical Society pp.195-209.

XI et al. (1990) J. Mol. Biol. 216, 375-384.

In vitro comparative studies of the interactions of gen-tamicin and azithromycin with membrane lipids: contri-bution to the understanding of the molecular mechanismby which they cause a lysosomal phospholipidosis in cul-tured cells

D. TYTECA1, L. BELAABIDIA 1, J.P. MONTENEZ1, F. VAN BAM-

BEKE1, M.P. MINGEOT-LECLERCQ1, J. PIRET1, R. BRASSEUR2

and P.M. TULKENS1 (1Unité de Pharmacologie Cellulaire etMoléculaire, Université Catholique de Louvain, UCL 73.70avenue Mounier 73, 1200 Bruxelles and 2Centre de BiophysiqueMoléculaire Numérique, Faculté des Sciences Agronomiques deGembloux, Passage des Déportés 2, 5030 Gembloux, Belgium)

Aminoglycoside antibiotics such as gentamicin (Gm) in-duce a lysosomal phospholipidosis in vivo (kidney proximaltubular cells) and in vitro (cultured fibroblasts; Aubert-Tulkenset al., 1979). More recently, we described that azithromycin(Az), a newly developped macrolide antibiotic also causes alysosomal phospholipidosis in fibroblasts (Montenez et al,1993).

Gm and Az are both cationic drugs but have otherwisevery different chemical properties. We therefore have com-pared the inhibitory potency of Gm and Az towards lysosom-al phospholipase activity in vitro, in correlation with theircapacity to bind to and to cause aggregation of negatively-charged liposomes. Tests were made with liposomes made ofa mixture of cholesterol, phosphatidylcholine, sphingomye-lin, and phosphatidylinositol. Enzyme inhibition was obtainedin a range of concentrations of 10 to 250 µM, which is largelybelow what is presumably reached in the lysosomes of cul-tured cells. For Gm, this inhibition was correlated with itselectrostatic binding to the membrane (thus decreasing thenegative charge density necessary for optimal phospholipaseactivity). Gm also caused a massive aggregation of liposomes,which could prevent the access of the enzyme to its substrateand contributes to its inhibitory effect. Az also bound to lipo-somes, but through both electrostatic and hydrophobic forces(binding was indeed only moderately modulated by ionicstrength and by negative charge). Moreover, Az caused noaggregation of liposomes, suggesting that it cannot bridgetogether adjacent membranes. These experimental data wererationalized by a computer-aided conformational analysiswhich showed that Gm is located at the lipid/water interface,and could therefore establish a close contact between adja-cent membrane constituants, whereas Az is deeply insertedin the hydrophobic domain, and therefore totally masked fromthe membrane surface.

Our data therefore suggests that the molecular mecha-nisms by which Gm and Az inhibit phospholipase activityare different. This work opens interesting perspectives in thecomprehension of the interaction between drugs and mem-branes in relation with the cellular alterations they may in-duce and, indirectly, in the unravelling of the mechanism andregulation of phospholipase activity.

ReferencesAUBERT-TULKENS, G., VAN HOOF, F. & TULKENS, P.M. (1979) Lab. Invest. 40,

481-491.MONTENEZ, J.P., CARLIER, M.B., GARCIA-LUQUE, I. & TULKENS, P.M. (1993) in

33th ICAAC (New Orleans, LA) abstract 310.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B53SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Genomic analysis of a new type of primate T-lymphotro-pic virus: PTLV-L

M. VAN BRUSSEL, P. GOUBAU, J. DESMYTER and A.M. VAN-DAMME. (Rega Institute for Medical Research and UniversityHospitals, Minderbroedersstraat 10, B-3000 Leuven, Belgium)

The human T-lymphotropic viruses (HTLV-I and HTLV-II) are complex, slowly evolving retroviruses associated withneurological disorder and leukaemia. Phylogenetically, theseviruses are linked to the simian T-lymphotropic viruses. Thewhole group is referred to as primate T-lymphotropic viruses(PTLV). They have a similar genomic structure as the moredistantly related bovine leukemia virus (BLV). Recently, weisolated a primate T-lymphotropic virus from a wild-bornHamadryas baboon with a highly divergent HTLV-like anti-body pattern. Initially, screening of a cDNA library from avirus-producing cell line, with a 32P-labelled PCR fragment,resulted in the isolation of a 1802-bp cDNA fragment. Se-quence analysis of this fragment, spanning the partial envand pX region of the viral genome, revealed that this simianvirus differs as much from the known PTLV types -I and -IIas both differ from each other. Sequence and phylogeneticanalysis, both support the distinction this virus as a new typeof primate T-lymphotropic virus, provisionally called PTLV-L (Goubau et al., 1994). The sequence of the complete viralgenome was obtained by further screening of the cDNA li-brary combined with an “extra long” PCR strategy. The pro-viral genome spans 8919 bp, having about 63 % and 66%nucleotide similarity with PTLV-I and PTLV-II respectively.The viral genome encodes the retroviral Gag Pol, and Envstructural proteins and the more PTLV/BLV-specific Tax andRex regulatory proteins. By sequence analysis, we identifiedseveral putative splice donors and acceptors on the PTLV-Lgenome (Van Brussel et al., 1996). In addition, two putativeORFs were found downstream env, in one of the most diver-gent regions in the PTLV genome. RNA-PCR analysis showedthe existence of splice junction between the splice donor inthe env region, and a splice acceptor situated in the proximalpX region. The messenger generated by this event potentiallyencodes a new viral protein having poor similarity with theproteins of PTLV-I and PTLV-II. These data further justifythe distinct classification of this virus, not only in terms ofsequence divergence but also in terms of its different genom-ic structure.

ReferencesGOUBAU, P., VAN BRUSSEL, M., VANDAMME , A.-M., LIU, H. & DESMYTER, J.

(1994). Proc. Natl. Acad. Sci. USA 91, 2848-2862.VAN BRUSSEL, M., GOUBAU, P., DESMYTER, J. & VANDAMME A.-M. (1996). J.

Gen. Virol. 77, 347-358.

Substitution rate calibration of nucleotide sequence align-ments and application to phylogenetic tree construction

Y. VAN DE PEER and R. DE WACHTER (Departement Biochemie,Universiteit Antwerpen (UIA), Universiteitsplein 1, B-2610 Ant-werpen, Belgium)

The fast accumulation of sequence data permits the con-struction of large evolutionary trees, containing hundreds ofsequences. For certain classes of tree construction methods,such as maximum parsimony and maximum likelihood, thisis problematic due to the enormous number of possible treetopologies that have to be evaluated. Some distance methods,like neighbor-joining, do not suffer from this problem. How-ever, the reliability of phylogenetic trees based on distancedata is highly dependable on the accurate estimation of theevolutionary distances from the observed sequence dissimi-larities. Since we do not have an exact historical record ofevents that took place in the evolution of the sequences, cor-rect estimation of the evolutionary distance is not straightfor-ward. As a result, the estimation of evolutionary distances isgenerally based on a simplified model of substitution. Themajor drawback of most of these models is that they do nottake into account differences in substitution rates among thedifferent sites of a molecule. Nevertheless, it has been shownbefore that the use of an unrealistic substitution model canlead to serious artefacts in tree topologies. Recently, we de-veloped a method called ‘substitution-rate calibration’ to com-pute the relative substitution rate of the individual nucle-otides in a sequence alignment (Van de Peer et al., 1993,1996). From the resulting distribution of substitution rates, anew equation can be derived that describes a more realisticrelationship between dissimilarity and evolutionary distancethan equations derived from a substitution model that doesnot take into account different substitution rates.

Application of this new method to ribosomal RNA se-quences leads to tree topologies that are significantly im-proved. In particular sequences with a very high evolutionaryrate, bound to cluster erroneously, are now clustered correct-ly due to the improved estimation of their evolutionary dis-tance to other sequences. Furthermore, specific evolutionaryrelationships can be demonstrated for organisms for whichthe evolutionary position was previously unknown or highlyquestionable. For example, on the basis of ‘substitution-ratecalibration’ we could convincingly demonstrate that the plastidsof the Chlorarachnida (a group of amoeboid algae) havebeen derived from an endosymbiosis with a green alga, a factthat was previously suggested but could not be proven on thebasis of the sequence data.

ReferencesVAN DE PEER, Y., NEEFS, J-M., DE RIJK, P. & DE WACHTER, R. (1993) J. Mol.

Evol. 37, 221-232.VAN DE PEER, Y., VAN DER AUWERA, G. & DE WACHTER, R. (1996) J. Mol.

Evol. (in press).

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B54 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Detailed quantitative analysis of the actin-binding site ofthe actin-sequestering protein thymosin β4

M. VAN TROYS, D. DEWITTE, M. GOETHALS, M.-F. CARLIER1,J. VANDEKERCKHOVE and C. AMPE (Flanders InteruniversityInstitute for Biotechnology, Department of Biochemistry, Facul-ty of Medicine, University of Ghent, B-9000 Ghent, Belgium and1Laboratory of Enzymology, CNRS, 91198 Gif-sur-Yvette, France)

Thymosinβ4 is a small actin-monomer-binding proteinthat plays an important role in the regulation of actin fila-ment turnover in all vertebrate cells. In a previous, qualita-tive study Vancompernolle et al. (1992) delineated the mini-mal site required for actin binding as residues 1-25. NMR-analysis of Tβ4 in aqueous solution showed that the NH2-terminal part of the actin binding site (residues 1-16) and thehexapeptide motif (residues 17-22) form separate structuralentities (Czisch et al., 1993). In this work we further clarifiedthe role of both entities by analysing the actin-binding activ-ity of a series of Tβ4 variants carrying aminoacid substitu-tions or deletions. Using chemical cross-linking assays, com-plex formation in native gels and polymerization inhibitionstudies, we prove that both the NH2-terminus and the motifare involved in actin binding and we identify the residues ofTβ4 establishing the contact. These quantitative data, in com-bination with circular-dichroism studies, allow to present amodel in which the NH2-terminal part of Tβ4 (residues 4-16)needs to adopt an α-helix for binding to actin. This α-helixinteracts through a patch of hydrophobic residues (6M-I-F12)located on one side of this helix. In addition, Tβ4 residueslysine 14 (in the α-helix) and lysine 18 (in the hexapeptidemotif) form salt bridges with acidic residues in actin.

The residues we identified as being important in the ac-tin-Tβ4 contact are conserved throughout the β-thymosin fam-ily. In addition, sequence alignment shows a similar patternof hydrophobic and charged residues in the COOH-terminalheadpiece domain of the villin family (Arpin et al., 1988).We performed peptide competition experiments the resultsof which indeed suggest that Tβ4 and the actin-bundling do-main of villin bind to (partly) overlapping sites on the actinmolecule.

We thank J.L. Verschelde for constructing computer images, J. Vandamme formass measurements and Prof. M. Rosseneu for use of the spectropolarimeterto perform the circular-dichroism experiments. C. Ampe is a postdoctoralfellow of the Belgian National Science Foundation (N.F.W.O.). This work wassupported by the Human Capital and Mobility Program of the European Com-munity and by EC grant CI1-CT93-0049 and grant 3.0008.94 of the FGWO toC. Ampe and grant GOA-91/96-3 to J. Vandekerckhove

ReferencesARPIN, M., PRINGAULT, E., FINIDORI, J., GARCIA, A., JELTSCH, J.-M., VANDEKERCK-

HOVE, J. & LOUVARD, D. (1988) J. Cell Biol. 107, 1759-1766.CARLIER, M.-F. & PANTALONI , D. (1994) Sem. Cell Biol. 5, 183-191.CZISCH, M., SCHLEICHER, M., HORGER, S., VOELTER, W. & HOLAK , T.A. (1993)

Eur. J. Biochemistry 218, 335-344.VANCOMPERNOLLE, K., GOETHALS, M., HUET, C., LOUVARD, D. & VANDEKERCK-

HOVE, J. (1992) EMBO J., 11, 4739-4746.VAN TROYS, M., DEWITTE, D., GOETHALS, M., CARLIER, M.-F., VANDEKERCKHOVE,

J. & AMPE, C. (1996) EMBO J. 15, in press.

Characterization of the human secretory component genepromoter

VERRIJDT, G.,1 CLAESSENS, F.,1 SWINNEN, J.,2 PEETERS, B., 1

VERHOEVEN, G.2 and ROMBAUTS, W.1 (1Afdeling Biochemie,2Laboratorium voor Experimentele Geneeskunde en Endocri-nologie, K.U.Leuven, Campus Gasthuisberg, Herestraat 49, 3000Leuven, Belgium)

The secretory component protein (SC) is responsible forthe transport of polymeric immunoglobulins (IgA & IgG) fromthe submucosal interstitial fluid through the epithelial cellsinto the secretions. Its synthesis by the epithelial cell is regu-lated by a number of factors (e.g. steroid hormones, interleuk-ines (-interferon-γ). In lacrymal epithelial cells of the rat, an-drogen stimulation causes a rise in SC mRNA concentration(Gao et al., 1995). In prostatic epithelial cells, SC synthesis isalso dependent on androgens (Stern et al., 1992). In HT-29cells (a cell line derived from colon carcinoma cells) howeverandrogens do not influence SC production, whereas interfer-on-γ induces an increase in steady state SC mRNA levels(Piskurich et al., 1993).

Screening of a human genomic library with an oligonucle-otide specific for the first exon of the SC gene (Krajci et al.1992), resulted in the isolation of 4 positive clones of which 3were unique. Two clones have a 4-kb XbaI fragment in com-mon which contains the promoter, the first exon and 474 bp ofthe first intron of the SC gene. This fragment was subclonedand its sequence characterized. The transcription initiation siteof the SC gene was determined using the 5'-RACE technique(Rapid Amplification of 5' cDNA Ends, GIBCO-BRL) on totalRNA extracted from benign prostate hypertropic cells. Fourpossible transcription initiation sites were found in an 8-bpregion encompassing the 5' end of the known exon 1 sequence.

No obvious TATA-box is found immediately upstream ofexon 1. The upstream region and first intron of the SC genecontain several consensus sequences for binding of AP-1,NFκ-B, the glucocorticoïid receptor (GR) and Ets-family tran-scription factors. Furthermore, a putative androgen-responsiveelement (ARE) (5'- TGAAGAtgcTGTTCT -3') is found inthe opposite strand at position - 226 to - 212 relative to themost proximal transcription initiation site.

Binding of a fragment of the androgen receptor (the DNA-binding domain + hinge region) to the putative SC ARE wasdemonstrated in gel retardation experiments.

The activity of the SC promoter is studied in transfectionexperiments: two genomic fragments (1.3 and 0.55 kb) con-taining the promoter were cloned upstream of the luciferasereporter gene in the pGL3 basic vector (Promega). LNCaP,T47-D, COS-7, IEC and HT-29 cell lines were transiently trans-fected with these constructs. Similar promoter activity wasdetected for both constructs.

In conclusion, we have cloned and characterized the hu-man SC gene and determined the region of transcription initi-ation. The transcriptional activity of our constructs, as meas-ured in transient transfection experiments, makes it now pos-sible to search for regulatory elements in the promoter region.

Supported by grants from the “Fonds voor Geneeskundig en WetenschappelijkOnderzoek.” V. G. is holder of a scholarship of the “Vlaams Instituut voor debevordering van het Wetenschappelijk-Technologisch onderzoek.” C.F. is holderof a grant of the “Nationaal Fonds voor Wetenschappelijk Onderzoek.” Wegratefully acknowledge the excellent technical assistance of H. De Bruyn, R.Bollen and V. Feytons.

ReferencesGAO, J., LAMBERT, R. W., WICKHAM, A., BANTING, G. & SULLIVAN , D. A.

(1995) J. Steroid. Biochem. Mol. Biol. 52, 239-249.KRAJCI, P., KVALE, D., TASKEN, K., BRANDTZAEG, P. (1992) Eur. J. Immunol.

22, 2309-2315.PISKURICH, J., FRANCE, J. A., TAMER, C. M., WILLMER, C. A., KAETZEL, C. S.,

KAETZEL, D. M. (1993) Mol. Immunol. 30, 413-421.STERN, J. E., GARDNE,R S, QUIRK, D. & WIRA, C. R. (1992) J. Reprod. Immu-

nol. 22, 73-85

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B55SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Threading-based prediction of the N-terminal domain ofthe α subunit of the GABAA receptor fold

1VINALS, C., 2LOPEZ-ROMERO, B., DURANT, F. and FEYTMANS,E. (1Laboratoire de Biologie Moléculaire Structurale et 2Labo-ratoire de Chimie Moléculaire Structurale, FUNDP, 61 rue deBruxelles, 5000 Namur, Belgium)

The GABAA receptor is a major type of pentameric re-ceptor for the essential neurotransmitter inhibitor j-aminobu-tyric acid (GABA) (Enna & Möhler, 1987). This receptor ismember of a superfamily of ligand-gated ion channels. Thereceptor complex is formed by the association of distinctsubunits, named α, β, γ, δ and ρ.

Benzodiazepines exert their therapeutic effects (anxio-lytic, hypnotic, anticonvulsant and myorelaxant properties)by binding to the ω modulatory sites of GABAA receptorcomplex, and thus affecting the chloride flux in the ion chan-nel (Haefely, 1989). The histidine 101 residue, located in theN-terminal extracellular domain of the a1 subunit, is sus-pected to be involved in the binding of benzodiazepine lig-ands (Wieland et al., 1992). Thus, it could be very informa-tive to have an insight of the structure of this domain in orderto understand the molecular determinants of the benzodi-azepine function.

The general fold of the N-terminal domain of the α subu-nit of the GABAA receptor (NTD_GABAA) has been predict-ed by using a combination of methods. The first step was themultiple alignment of sequences homologous to NTD_GABAA,using CLUSTALW (Thompson et al., 1994) and Matchbox(Depiereux & Feytmans, 1992) algorithms, in order to definethe conserved core regions. Based on this alignment, a pre-diction of the secondary structure was performed using theneural-network-based method PHD (Rost & Sander, 1994).The sequence (252 residues) contains a signal peptide, pre-dicted as an α-helix, followed by a second helix and 10 to 17β-strands. The sequence was then submitted to a threadingprocedure using the algorithms PROFIT (Sippl, 1993) andTHREADER (Jones, 1994). The selection of suitable foldswas made on the basis of 1) the percentage of aligned se-quence to structure, 2) the energy score and 3) the secondarystructure composition of the template. We selected three pos-sible folds: a single-domain β-propeller (1NSC), a two-do-main β-trefoil (1ABR) and a two-domain all-b trypsine-likefold (1ARB). Further analysis of the predicted contacts wasperformed using the CORRELATED MUTATIONS algorithm(Göbel et al., 1993). All these data were used to build severalrough models of NTD_GABAA, which could be useful todesign experimental work on the protein structure-functionrelationship.

ReferencesDEPIEREUX, E. & FEYTMANS, E. (1992) CABIOS 8, 501-509.ENNA, S. J. & MÖHLER, H. (1987) in Psychopharmacology: The Third Genera-

tion of Progress (MELTZER, H.Y., ed.) 3rd ed., pp. 265-272, Rowen Press,New York.

GÖBEL, U., SANDER, C., SCHNEIDER, R. & VALENCIA, A. (1994) Prot. Struct.Funct. Genet. 18, 309-317.

HAEFELY, W.E. (1989) Eur. Arch. Psychiatr. Neurol. Sci., 238, 294.JONES, D.T. (1994) Prot. Sci. 3, 567-574.ROST, B. & SANDER, C. (1994) Prot. Struct. Funct. Genet. 19, 55-72.SIPPL, M.J. (1993) J. Comp.-Aid. Molec. Des. 4, 473-501.THOMPSON, J.D., HIGGINS, D.G., GIBSON, T.J. (1994) Nucl. Ac. Res. 22, 4673-

4680.WIELAND, H.A., LÜDDENS, H. & SEEBURG, P.H. (1992) J. Biol. Chem. 267, 1426.

The broad-host-range plasmid pMOL96 contains a puta-tive transposon similar to Tn402 (Tn 5090) and Tn5053

A. WILMOTTE, A. HENNEN, D. VAN DER LELIE and M. MERGEAY

(VITO, Boeretang 200, 2400 Mol, Belgium)

The plasmid pMOL96 (pES1) was isolated exogenouslyby triparental mating from a soil polluted with PCBs andPAHs (Top et al., 1994). The host-range of this cryptic plas-mid of approximatively 60 kb comprises (α-, β-, and γ-pro-teobacteria. It does not hybridize with the replicon probes forclassical broad-host-range plasmids (Couturier et al., 1988)and probably belongs to a new incompatibility group tenta-tively called UbiD.

During the isolation of its replication origin, a deleted 17kb derivative was obtained in which a Kanamycin Genblock(Pharmacia) had been inserted. This derivative is transfer-deficient but can still be mobilized to Escherichia coli. A4.5-kb EcoRI fragment was cloned in pZERO (Invitrogen)and sequenced. The sequence revealed the presence of geneswith high similarity to tniB (putative ATP-binding protein),tniQ and tniR (resolvase) from Tn402 (Tn5090) (Rädström etal., 1994) and Tn5053 (Kholodii et al., in press). The resolu-tion site between tniQ and tniR was identical to the site ob-served in Tn402 (Tn5090) and Tn 5053. The presence of tniA(transposase) on a 2.5-kb EcoRI fragment was demonstratedby PCR using consensus primers designed for Tn402 (Tn5090)and Tn5053. The presence of the putative transposon on theoriginal plasmid pMOL96 was shown by PCR using the primersdesigned for the sequencing of the deleted derivative.

In Tn402 (Tn5090) the tniABQR genes are associatedwith an integron, a gene cassette encoding resistance to tri-methoprim (dhfrII c) and a multiresistance gene (qacE)(Rädström et al., 1994) whereas in Tn5053 homologous tnigenes are located close to an operon encoding resistance tomercury (Kholodii et al., in press). No integrase could bedetected by hybridization of pMOL96 with the PCR-productcontaining a 1-kb sequence from the integrase gene fromTn402 (Tn5090). In addition there was no hybridization be-tween pMOL96 and Tn4378, a mercury transposon (Diels etal., 1989). In the deleted derivative of pMOL96, the 3’ end oftniR is flanked by about 300 non-coding bases followed by aputative ORF of 129 bases which shows 64% of nucleic-acidsimilarity with the ORF6 from the cob operon of Pseudomonasdenitrificans. The function of ORF6 was unknown but theauthors hypothetized that it could belong to a transport sys-tem (Crouzet et al., 1991). This ORF of 129 bases is fol-lowed by about 100 non-coding bases after which the Kan-amycin Genblock is inserted. Present data thus indicates thatthe putative transposon from pMOL96 is neither associatedwith an integron nor with resistance genes for mercury butmight be placed in a different genetic context.

This work was supported by the BIOTECH program from the EuropeanUnion.

ReferencesCOUTURIER, M., BEX, F., BERGQUIST, P.L. & MAAS, W.K. (1988) Microbiol.

Rev. 52, 375-395.CROUZET, J., LEVY-SCHIL, S., CAMERON, B., CAUCHOIS, L., RIGAULT , S., ROUYEZ,

M.C., BLANCHE, F., DEBUSSCHE, L. & THIBAUT, D. (1991) J. Bacteriol.173, 6074-6087.

DIELS, L., SADOUK, A. & M ERGEAY, M. (1989) Toxicol. Environm. Chem. 23, 79-89.

KHOLODII, G., MINDIN, S.Z., BASS, I.A., YURIEVA, O.V., MINAKHINA , S.V. &NIKIFOROV, V.G. Mol. Microbiol. (in press).

RÄDSTRÖM, P., SKÖLD, O., SWEDBERG, G., FLENSBURG, J., ROY, P.H. & SUND-STRÖM, L. (1994) J. Bacteriol. 176, 3257-3268.

TOP, E., DE SMET, I., VERSTRAETE, W., DIJKMANS, R. & MERGEAY, M. (1994)Appl. Environm. Microbiol. 60, 831-839.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.

B56 SOCIÉTÉ BELGE DE BIOCHIMIE, ANTWERPEN UIA, 2 MARS 1996

Molecular dynamics simulation of pressure-induced con-formational changes in BPTI

B. WROBLOWSKI, J. FERNANDO DÍAZ , K. HEREMANS and Y.ENGELBORGHS (Laboratorium voor Chemische and BiologischeDynamica, Katholieke Universiteit Leuven, Celestijnenlaan 200D,B-3001 Leuven, Belgium)

The stability of proteins at high temperature has beenwidely studied using molecular dynamics techniques. Butvery few studies have been done at another denaturing condi-tion, i.e. high pressure (Kitchen et al., 1992, Brunne & vanGunsteren, 1993). We have performed a 800 ps moleculardynamic simulation of bovine pancreatic trypsin inhibitor(BPTI) (Marquart et al., 1983) in water, with coupling topressure baths at 1, 10000, 15000 and 20000 bar. This posterpresents a MD simulation (van Gunsteren & Berendsen, 1987)that shows the early stages of denaturation of a protein underhigh pressure conditions and the related changes in the sol-vent structure. The simulation reproduces quite well the ex-perimental behaviour of the protein under high pressures.The protein keeps its globular form, but adopts a slightlydifferent conformation with a nearly negligible reduction involume.

Some residues in the hydrophobic core become exposedto water and a large part of the secondary structure of theprotein, (60 % of the sheet structure and 40% of the helicalstructure) is denatured between 10 and 15 kbar. This is inqualitative agreement with experimental data (Goossens etal., 1996) that show a decrease of α helix content of BPTIbetween 8 and 14 kbar. A further increase of the pressureresults in freezing of the protein as deduced from the largedecrease of the mobility of the residues. The structure of thewater during the simulation changes from the normal waterstructure (ice Ih-like) to an ice VI-like structure with inter-penetration of the H-bond network, while keeping the liquidstate. The much higher compressibility of the water, com-pared to the protein, results in an increase of the packing ofthe water molecules and hence in a higher density of hydro-gen bond donors and acceptors. The number of water-waterand water-protein hydrogen bonds increases, whereas thenumber of internal protein hydrogen bonds decreases.Thishigher compressibility of the water compared with the pro-tein, which produces a change in the solvent properties, seemsto be the driving force of the high-pressure-induced confor-mational transition.

This work was supported by the Onderzoeksraad K.U.Leuven.

ReferencesBRUNNE, R.M & VAN GUNSTEREN, W.F. (1993) FEBS Lett 323, 215-217.GOOSSENS, K., SMELLER, L, FRANK, J. & HEREMANS,K. (1996) Eur. J. Biochem,

in press.KITCHEN, D.B., REED, L.H. & L EVY, R.M. (1992) Biochemistry 31, 10083-

10093.MARQUART, M., WALTER, J., DEISENHOFER, J., BODE, W. & HUBER, R. (1983)

Acta Crystallogr., Sect. B, 39, 480-490.VAN GUNSTEREN, W.F. & BERENDSEN, H.J.C. (1987) Biomos biomolecular soft-

ware b.v. Laboratory of Physical Chemistry University of Groningen.

The c-MYB leucine zipper motif mediates homodimer for-mation in vivo as shown by the yeast 2-hybrid system

YING XU1, B. PUNYAMMALEE 2 and M. CRABEEL1

(1Dpt of Microbology, Vrije Universiteit Brussel and VlaamseInteruniversitaire Instelling voor Biotechnologie, c/o ResearchInstitute of CERIA-COOVI, 1, ave. E. Gryson, B-1070 Brussels,Belgium, and 2Research Division, Natl Cancer Institute, RamaVI Road, 10400 Bangkok, Thailand)

c-Myb is a major regulator of hematopoiesis, possiblyalso involved in cell cycle control. The activity of this keytranscription factor is believed to be modulated by associa-tion with other, yet unidentified regulatory proteins. The leu-cine zipper (LZ) located in the “negative regulatory domain”of Myb is one of the motifs suspected to participate in pro-tein complex formation. Here we demonstrate the ability ofthis motif to homodimerise in vivo by using the yeast S.cerevisiae 2-hybrid system (Chien et al., 1991).

We had previously shown that Myb-LZ fused to the GAL4-DNA binding domain (on plasmid pXY4) was able on itsown to activate the two reporter genes (HIS3, dispensingaminotriazole resitance and lacZ, giving blue colonies on X-gal plates) of strain Y190, commonly used as host in theyeast 2-hybrid system: this suggested that Myb-LZ was ableto recruit a yeast transactivator through complex formationwith a compatible LZ domain. This high background expres-sion of the reporter genes in the sole presence of pXY4 pre-vents the use of the 2-hybrid system to test for Myb-LZ ho-modimerisation capacity and to select for possible regulatorypartners by screening a human cDNA library. In an effort toidentify the interfering yeast factor, we have started a sys-tematic screen of yeast mutants affected in one of the limitednumber of known transactivators of the basic-leucine zipperfamily. We have shown, first, that a full c-myb gene carryinga point mutation in LZ is partially deficient in MRE-depend-ent reporter gene activation, and second, that in the geneticcontext of a strain carrying a mutation in the activator-encod-ing gene YAP2, this effect disappears. We therefore crossedthis yap2-deleted strain with Y190 in order to construct re-combinants bearing the markers required for a 2-hybrid screen-ing test with pXY4. As expected, two types of segregantswere obtained, one where pXY4 on its own induced the twoUASGAL4-dependent reporter genes just as in Y190, and an-other in which the reporters activity was reduced. However,the reduced pXY4-induced background activity did correlateneither with the YAP2 genetic status (as identified by PCR),nor with a 2+: 2- segregating “slow growing” phenotype alsocharacterizing the yap2-deletion parent strain. Using one suchlow-background but as yet uncharacterized mutant strain as ahost, we could now demonstrate aminotriazole resistance spe-cifically induced by the addition to pXY4 of the pXY5 plas-mid (which encodes a fusion of the Myb-LZ to the GAL4activation domain), and not of pACT II (same plasmid butwithout LZ),while the addition of p XY14, equivalent to pXY5but bearing a partial deletion of the GAL4 activation domaininduces resistance to a lesser extent. These results indicatethat the Myb-LZ motif can function as a homodimerizationmotif in vivo , and confirm the in vitro data of Nomura et al.(1993) who suggest a mechanism of negative autoregulationof cMyb activity by homodimer formation through the leu-cine zipper.

ReferencesCHIEN, C., BARTEL, P., STERNGLANZ, R., & FIELDS, S. (1991) Proc. Natl. Acad.

Sci. USA 88, 9578-9582.NOMURA, T., SAKAI , N., SARAI , A., SUDO, T., KANEI-ISHII, C., RAMSAY, R.,

FAVIER, D., GONDA, T. & ISHII, S. (1993) J. Biol. Chem. 268, 21914-21923.

Arc

hive

s of

Phy

siol

ogy

and

Bio

chem

istr

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

cMas

ter

Uni

vers

ity o

n 11

/26/

14Fo

r pe

rson

al u

se o

nly.