Mol Gen Genet (1992) 235:242-246
OIG'G
{3 Springer-Verlag 1992
Leucine and serine induce mecillinam resistance in Escherichia coli Philippe Bouloc, Daniel Vinella, and Richard D'Ari Institut Jacques Monod, CNRS, Universit6 Paris 7, 2, place Jussieu, F-75251 Paris Cedex 05, France Received March 27, 1992 / Accepted June 11, 1992
Summary. We have previously shown that resistance to the ~-lactam mecillinam in Escherichia coli can be brought about by a high ppGpp pool, as observed under conditions of partial amino acid starvation and RelAdependent induction of the stringent response. We show here that our E. coli wild-type strain, which is sensitive to mecillinam on minimal glucose plates, becomes resistant in the presence of L-leucine or L-serine (or cysteine, which inactivates the antibiotic). The resistance, which is not a transient effect and does not depend on the physiological state of the cells when plated, is specific for mecillinam and is reversed by the presence of isoleucine and valine in the medium. At least in the case of serine, the resistance is RelA-dependent. We conclude that the presence of leucine and serine in the growth medium cause partial starvation for isoleucine/valine, leading to induction of the stringent response and concomitant resistance to mecillinam. Key words: Mecillinam - Penicillin-binding protein 2 Stringent response - L-leucine - L-serine
Introduction The harmonious growth of a cell requires tight homeostatic regulation of virtually all metabolic pathways and strict coupling amongst the various macromolecular biosyntheses. The bacterium Escherichia eoli has provided a convenient model system for unravelling many of these regulatory mechanisms and interactions. Our present state of knowledge, although far from complete, provides paradigms for a number of regulatory circuits ensuring both balanced growth and rapid responses to environmental changes. One such system, the stringent response, helps provide homeostatic regulation of the protein synthesizing machinery, adjusting the cell's protein synthetic capacity to its ability to form aminoCorrespondence to: P. Bouloc
acyl-tRNA precursors (Cashel and Rudd 1987). The central regulator of the stringent response is the stringent factor, coded for by the relA gene. This protein, which is associated with ribosomes, catalyses the synthesis of the nucleotide pppGpp whenever an uncharged cognate tRNA occupies the A site of an actively translating ribosome. The nucleotide pppGpp is rapidly hydrolysed to ppGpp, and this nucleotide, in turn, is an RNA polymerase effector which reduces the transcription of most genes coding for elements of the protein synthesizing machinery, including all ribosomal RNA and protein genes. Thus, a low charge ratio for any tRNA species rapidly results in a lowered rate of ribosome production. In addition to regulating ribosome production, the stringent response is also involved in the regulation of bacterial envelope synthesis. The penicillin-binding proteins (PBPs) are integral membrane proteins responsible for the synthesis of the rigid peptidoglycan layer, ppGpp has been reported to be a negative effector of some step in peptidoglycan synthesis and is involved in autolysin regulation (Ishiguro and Ramey 1976, 1978; Goodell and Tomasz 1980; Vanderwel and Ishiguro 1984; Kusser and Ishiguro 1987). Furthermore, cells deprived of an essential element have a relA-dependent "phenotypic resistance" to a broad range of [3-1actam antibiotics, which are associated with peptidoglycan changes (Kusser and Ishiguro 1987; Tuomanen and Tomasz 1991). More recently, we have shown that ppGpp specifically alters the cell's requirement for PBP2 (Vinella et al. 1992). This PBP plays a crucial role in maintaining the rod shape of E. coli (Spratt and Pardee 1975). Inactivation of PBP2, whether by the specific antibiotic mecillinam or by mutation of the structural gene pbpA, results in the formation of spherical cells which normally stop dividing, increase in volume and die (Lurid and Tybring 1972; Park and Burman 1973; Matsuhashi et al. 1974; James et al. 1975; Spratt 1975, 1977; Ogura et al. 1989; Vinella et al. 1992). The lethality associated with PBP2 inactivation can be avoided in certain genetic backgrounds, including mutants possessing a partially defective aminoacyl-tRNA synthetase, which causes a high
243 ppGpp pool; these mutants are resistant to mecillinam (Vinella et al. 1992). We showed that this partial induction of the stringent response is in fact responsible for the cells' ability to tolerate PBP2 inactivation: introduction of a relA mutation restored mecillinam sensitivity to these t R N A synthetase mutants, and a multicopy plasmid, prelA', coding for a hyperactive (p)ppGpp synthetase, conferred mecillinam resistance on wild-type strains (Vinella et al. 1992). We concluded that PBP2 is normally an essential protein whose inactivation results in blockage of some vital process other than cell wall elongation, possibly cell division, but that a high ppGpp pool short-circuits this blockage, either by interrupting the communication between PBP2 and the vital process concerned, or by otherwise bypassing the latter. It has long been known that relA mutants, incapable of synthesizing ppGpp in response to aminoacyl-tRNA limitation, are hypersensitive to physiological situations such a~snutritional shifts down, in which there is an initial starvation period. In addition, reIA mutants exhibit a series of as yet unexplained sensitivities to certain apparently anodyne metabolites. Alf61di and Kerekes (1964) showed that relA strains growing exponentially in rich medium, although able to adapt to minimal glucose medium, have greatly reduced colony-forming ability on glucose plates supplemented with leucine, serine, cysteine, isoleucine or phenylalanine. Uzan and Danchin (1976, 1978) later showed that relA mutants are hypersensitive to serine, or to serine + glycine + methionine (SMG), and further showed that addition of SMG to exponentially growing cultures of relA + strains causes a temporary inhibition of growth, accompanied by starvation for isoleucine and valine and induction of the stringent response. It was suggested that relA mutants, being unable to induce the stringent response, somehow get locked in a physiological impasse in certain conditions (Alf61di and Kerekes 1964). It is reasonable to suppose that all conditions to which reIA mutants are hypersensitive cause induction of the stringent response, since (p)ppGpp synthesis is the only known activity of the RelA protein. We have shown that induction of the stringent response enables the cell to tolerate inactivation of PBP2, by mecillinam or by mutation, without blocking cell division or other vital processes (Vinella et al. 1992). This led us to examine the mecilliham sensitivity of a wild-type E. coli strain under physiological conditions in which reIA mutants have reduced colony-forming ability. We show here that wild-type E. coli is in fact permanently resistant to mecillinam under such conditions. Furthermore, the resistance is reIA +-dependent, indicating that certain metabolites in the growth medium can cause lasting partial induction of the stringent response in wild-type cells. Our results indicate, surprisingly, that the presence in the growth medium of certain amino acids such as leucine or serine can significantly reduce the rate of ribosome synthesis.
et al. 1977) and UTH1038 (Goldschmidt et al. 1970). A relA1 derivative, GC3698 (Vinella et al. 1992), was constructed by Plvir mediated transduction. The lrp::TnlO allele was transduced into GC2553 from strain NEW26 (Lin et al. 1990), kindly given us by E. Newman. Other prototrophic strains were NEW1 (Lin et al. 1990), 594 (Campbell 1961), JM101 (Messing 1979) and W3110 (Bachmann 1987). Media and culture conditions. Rich medium was LB broth and minimal medium was M63 (Miller 1972), supplemented with 0.4% glucose, 1 gg/ml thiamine and amino acids at 100 gg/ml. All amino acids (except glycine) were L stereoisomers. Casamino acids, treated with activated charcoal, was used at 0.4%. Mecillinam, generously provided by Laboratoires Leo (France), was used at 10 gg/ml. Nitrocefin was purchased from Laboratoires Glaxo (France). Solid media contained 1.5% Difco agar. Cysteine solutions (1% in water) and cysteine-containing media were prepared less than 24 h before use. Liquid cultures were grown in shaking water baths; all experiments were carried out at 37 ° C.
Results and discussion
Mecillinam resistance in the presence of certain amino acids
The mecillinam sensitivity of E. coli in minimal glucose medium is strain dependent. Spratt reported that his strain was resistant under these conditions (Barbour et al. 1981), whereas our wild-type E. coli K12 strain, GC2553, is sensitive (Bouloc et al. 1988, 1989). We examined its mecillinam sensitivity on glucose plates containing different supplements by plating an overnight glucose-grown culture. The results showed clear mecillinam
Table 1. Mecillinam resistance of wild-type Escherich& coli in the presence of certain amino acids Supplementsa
Efficiencyof plating - Mec + Mec
None Leucine Serine SMG Cysteine Arginine Histidine Lysine Methionine Phenylalanine Proline Threonine Tryptophan Tyrosine Casamino Acids
1.00 1.08 1.01 0.95 0.99 0.98 1.11 1.15 0.97 1.06 1.05 1.04 0.98 1.04 1.02
2.8 x 10-3 0.94 0.21 1.06 0.79 1.4 x 10- 3 1.8 x 10-3 2.4 x 10-3 2.3 × 10-3 4.3 x 10- 3 1.6 x 10.3 0.6 x 10-3 3.7 x 10-3 2.8 x 10-3 1.3 x 10-3
Materials and methods
Bacterial strains. Our wild-type E. coli K12 strain was GC2553, an F - @ ) - prototroph also called FB8 (Bruni
Mec, Mecillinam; SMG, serine+ methionine + glycine a Amino acids were added at 100 ~tg/ml, mecillinam was used at 10 gg/ml
244 Table 2. Effectof isoleucine and valine on mecillinamresistance Supplements None Ile+Val Leu÷ Ile+Val Ser+ Ile+Val SMG+Ile+Val
Efficiencyof plating - Mec + Mec 1.00 1.07 0.98 0.82 1.05
5.2 x 10.3 4.5 x 10 3 4.4 x 10.4 9.7 x 10.3 9.1 x 10 2
Mec, Mecillinam;SMG, serine+methionine÷glycine
resistance on plates containing leucine, serine, SMG or cysteine (Table 1). As has previously been observed in work with mecillinam-resistant mutants, growth in the presence of mecillinam was slow, with the single exception of the cysteine-supplemented plates, on which colonies appeared equally rapidly in the presence and absence of mecillinam. No mecillinam resistance was observed in the presence of isoleucine or phenylalanine, to which relA mutants were previously reported to be sensitive (Alf61di and Kerekes 1964), or in the presence of amino acids not known to inhibit relA strains (Table 1). Certain amino acids clearly protect E. coli against mecillinam in minimal medium. Casamino acids, however, which contain leucine, serine, methionine and glycine, did not afford protection (Table 1), even if 100 gg/ ml additional leucine was added (data not shown). The sensitivity of relA mutants to SMG is relieved by isoleucine and valine (Uzan and Danchin 1978), which are also present in Casamino acids. We therefore evaluated the effect of isoleucine and valine on the mecillinam resistance of wild-type E. coli in the presence of leucine, serine or SMG. The presence of Ile + Val suppressed the protective effect of leucine, serine and SMG (Table 2). Cysteine again proved to be different: the presence of 100 gg/ml cysteine in glucose Casamino acids mecillinam plates permitted colony formation (efficiency of plating 0.6).
Independence of mecillinam resistance on 9rowth phase The sensitivities reported for reIA mutants under shiftdown conditions were only observed in the exponential phase of growth (Alf61di and Kerekes 1964). We therefore examined the effects on mecillinam resistance of the physiological state of the culture when plated. Strain GC2553 was grown to exponential phase in glucose minimal medium, then plated on glucose plates supplemented with leucine, serine, SMG or cysteine, with or without mecillinam. With no amino acid supplement, the survival in the presence of mecillinam was 4 x 10-3, whereas with each of the supplements, the efficiency of plating was greater than 50%. Thus the mecillinam resistance observed in the presence of leucine, serine, SMG or cysteine does not depend on the physiological state of the bacteria when plated. When the same supplements were added in liquid culture, growth slowed down for some time, then re-
sumed, as previously reported for serine and SMG (Uzan and Danchin 1978). One hour after the additions, the cultures were observed under the phase contrast microscope. No morphological changes were apparent; in particular, the cells remained rod-shaped, indicating that PBP2 function was not impaired. Leucine, serine, SMG and cysteine, when added to a culture of wild-type cells, clearly cause a shock, presumably resulting in induction of the stringent response in relA + strains. It was conceivable that the mecillinam resistance observed was related to this shock and that the addition of mecillinam to the medium at the same moment in some way froze the cells in a special "shocked" physiological state with a high ppGpp pool. To test this possibility, the cells were allowed first to adapt to these metabolites in liquid culture, then examined for resistance to mecillinam. Cultures were grown overnight (> 6 generations) in glucose medium supplemented with leucine, serine, SMG or cysteine, then plated on glucose mecillinam plates containing the same supplements. In all cases, the efficiency of plating was greater than 15 %. We conclude that cells growing in the presence of these metabolites are permanently locked in a physiological state which permits PBP2 inactivation, even many generations later, presumably via partial induction of the stringent response (see below). The presence of leucine in the growth medium radically alters the rate of expression of a number of E. coli genes. It has recently been shown that these genes form part of a global response, the "leucine/Lrp regulon", under the control of the /eucine-responsive regulatory protein (Lrp), a transcriptional activator/repressor coded for by the lrp gene (Newman et al. 1992). In Irp mutants, many of the member operons behave as though the exogenous leucine concentration were high; in particular, expression of the leu operon is low (Lin et al. 1992), and indeed these strains appear to be partially starved for leucine (Lin et al. 1990). It therefore seemed possible that lrp mutants cultivated in the absence of leucine would be resistant to mecillinam. An lrp::TnlO insertion was transduced into the wild-type strain GC2553, which was then plated on glucose mecillinam plates. The transduced strain showed the same sensitivity (survival 1.7 x 10 -3) as the lrp + parent strain, but became resistant to mecillinam in the presence of exogenous leucine, again like the parent strain. It was noticeable that the colony size of both lrp and lrp + cells was smaller on leucine-supplemented plates than on glucose alone. The growth rate of the wild-type strain GC2553 growing exponentially in glucose medium with or without supplements was measured. With no amino acids, the doubling time was 51 min; leucine increased the doubling time by 17% and serine increased it by 10%, leading to the paradoxical result that the cells grow more slowly in a richer medium. SMG, although not increasing the doubling time, provided, as we have seen, total protection against mecillinam (Table 1).
245 Involvement of the stringent response The mecillinam resistance observed in the presence of certain amino acids seemed to reflect the induction of the stringent response, apparently by creating partial starvation for isoleucine and/or valine. To verify the role of the stringent response in the mecillinam resistance observed, we tested a relA1 derivative of strain GC2553. The sensitivities of relA mutants were originally observed only under shift-down conditions (Alf61di and Kerekes 1964). To avoid a nutritional shift-down, a saturated overnight glucose culture was plated on glucose medium containing various supplements, with or without mecillinam. Our relA strain proved sensitive to leucine and SMG, even in the absence of a nutritional shift-down, so the relA dependence of the protection afforded by these metabolites could not be tested. However, the relA mutant was able to grow on serine and cysteine plates [efficiency of plating (e.o.p.) 60%]: serine did not confer mecillinam resistance (e.o.p. 1 x 10 -3) while cysteine, on the other hand, conferred resistance.
Generality of the phenomenon We tested four other prototrophic K12 strains, NEW1, 594, JM101 and W3110, to see whether the presence of L-leucine conferred mecillinam resistance. In all cases, however, the strains were mecillinam resistant in unsupplemented glucose medium, like the strain used by Barbour et al. (1981). We have shown that our "wildtype" strain GC2553 in fact carries a mutation in the 91 rain region of the genetic map, which provides mecilliham sensitivity unsupplemented glucose medium. Furthermore, the mutation can be lethal in certain conditions, and this lethality is suppressed by the introduction of aryS or alaS alleles or the plasmid prelA', all of which confer mecillinam resistance in rich media (Vinella et al. 1992). This work will be fully described in a forthcoming publication (D. Vinella, P. Bouloc and R. D'Ari, manuscript in preparation). We further investigated whether the resistance conferred by leucine, serine, SMG and cysteine was limited to mecillinam (which specifically inhibits PBP2) or constitutes a general protection against 13-1actam antibiotics. For this purpose an overnight glucose culture of strain GC2553 was platted on glucose plates containing various supplements, with or without 10 gg/ml ampicillin, which binds to all of the PBPs. Leucine, serine and SMG did not protect the cells against ampicillin (e.o.p.< 10-4). This shows that we are not dealing with the phenomenon of phenotypic resistance. Cysteine, in contrast, provided essentially full protection (e.o.p. 0.3).
antibiotics, independently of the cells (Markowitz and Williams 1985), and we confirmed that 80 gg/ml cysteine in aqueous solution rapidly hydrolyses the colour indicator nitrocefin, a cephalosporin analogue. It is perhaps worth pointing out that this highly reactive amino acid is spontaneously oxidized to cystine or cysteic acid after several days in aqueous solution and thus is not normally present in reduced form in rich media. Leucine, serine and SMG specifically confer resistance to mecillinam. The protection is reversed by the presence ofisoleucine and valine and, at least in the case of serine, requires a functional relA + gene product. We therefore conclude that the presence of leucine, serine or SMG in the culture medium results in partial starvation for isoleucine/valine and relA + dependent induction of the stringent response. The fact that addition of leucine, serine or SMG initially induces the stringent response was not unexpected, since relA mutants are hypersensitive to such shocks. The surprising finding was that, although relA + strains readily adapt to the presence of leucine, serine or SMG and resume growth, they apparently continue to be partially starved for isoleucine/ valine and the stringent response remains partially induced, conferring permanent mecillinam resistance. In the presence of leucine or serine, the doubling time is significantly longer than in unsupplemented glucose medium. It is also longer in the relA strain (once the cells have adapted to leucine or serine), as would be expected if growth is limited by isoleucine/valine availability, although in this case the stringent response cannot be induced and the bacteria remain sensitive to mecillinam. The wild-type strain was also resistant to mecillinam in the presence of SMG and, since isoleucine + valine reversed this resistance, the situation again resembled one of partial starvation. However, the cells actually grew faster in SMG than in unsupplemented medium. This may be related to our previous observation that there is no simple correlation between growth rate and mecilliham resistance (Bouloc et al. 1989; Vinella et al. 1992). Through the fortuitous discovery of an E. coli strain that is particularly sensitive to mecillinam in minimal glucose medium, we have discovered a mechanism of physiological regulation that is masked in some other strains. Our results show that glucose medium supplemented with leucine or serine, although richer than unsupplemented medium, affords a lower growth rate. This paradox may reflect the nature of the natural environments of E. coli, which tend to provide either very little in the way of nutrients (the outside world) or a mixture containing all amino acids (the gut). Media rich in only one or a few specific amino acids are perhaps so rare that E. coli has not learned to profit maximally from them.
Conclusions
Protection by cysteine clearly occupies a special place: it is relA-independent, is not reversed by isoleucine plus valine and extends to other ~-lactams. The mechanism of cysteine protection is, in fact, via chemical attack of the
Acknowledgements. We thank Elaine Newman for stimulatingdiscussions and Antonia Kropfinger for bibliographic research. Our research project on the mechanisms of mecillinam resistance was recognised by the Laboratoires Glaxo in the award of the Glaxo "PBP" Prize to P.B. This work was financedin part by the Association pour la Recherche sur le Cancer (contract no. 6696).
246
References Alf61di L, Kerekes E (1964) Neutralization of the amino acid sensitivity of R C rd Escherichia coll. Biochem Biophys Acta 91:155-157 Bachmann BJ (1987) Derivations and genotypes of some mutant derivatives of Eseherichia coli K-12. In: Neidhardt FC, Ingraham JL, Low KB, Magasanik B, Schaechter M, Umbarger HE (eds) Escherichia eoli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, DC, pp 1190-1219 Barbour AG, Mayer LW, Spratt BG (1981) Mecillinam resistance in Escherichia coli: dissociation of growth inhibition and morphologic change. J Infect Dis 143:114-120 Bouloc P, Jaff6 A, D'Ari R (1988) Preliminary physiologic characterization and genetic analysis of a new Escherichia eoli mutant, lov, resistant to mecillinam. Rev Infect Dis 10:905-910 Bouloc P, Jaff~ A, D'Ari R (1989) The Escherichia coli lov gene product connects peptidoglycan synthesis, ribosomes and growth rate. EMBO J 8 : 317-323 Bruni CB, Colantuoni V, Sbordone L, Cortese R, Blasi F (1977) Biochemical and regulatory properties of Escherichia coli K-12 hisT mutants. J Bacteriol 130:4-10 Campbell A (1961) Sensitive mutants to bacteriophage lambda. Virology 14:22 Cashel M, Rudd KE (1987) The stringent response. In: Neidhardt FC, Ingraham JL, Low KB, Magasanik B, Schaechter M, Umbarger HE (eds) Escherichia coli and Salmonella typhimurium : cellular and molecular biology. American Society for Microbiology, Washington, DC, pp 1410 1438 Goldschmidt EP, Cater MS, Matney TS, Butler MA, Greene A (1970) Genetic analysis of the expression of the histidine operon in Salmonella typhimurium. Genetics 66:219-229 Goodell W, Tomasz A (1980) Alteration of Escherichia coli murein during amino acid starvation. J Bacteriol 144:1009-1016 Ishiguro EE, Ramey WD (1976) Stringent control of peptidoglycan biosynthesis in Escherichia coli K-12. J Bacteriol 127:1119-1126 Ishiguro EE, Ramey WD (1978) Involvement of the relA gene product and feedback inhibition in the regulation of UDP-Nacetylmuramyl peptide synthesis in Escheriehia coli. J Bacteriol 135:766-774 James R, Haga JY, Pardee AB (1975) Inhibition of an early event in the cell division cycle of Escherichia coli by FL1060, an amidinopenicillanicacid. J Bacteriol 122:1283-1292 Kusser W, Ishiguro EE (1987) Suppression of mutations conferring penicillin tolerance by interference with the stringent control mechanism of Escherichia coli. J Bacteriol 169:43964398 Lin RT, D'Ari R, Newman EB (1990) The leucine regulon of Escherichia coli K-12: a mutation in rblA alters expression
of L-leucine-dependent metabolic operons. J Bacteriol 172: 45294535 Lin RT, D'Ari R, Newman EB (1992))~ placMu insertions in genes of the leucine regulon: extension of the regulon to genes not regulated by leucine. J Bacteriol 174:1948-1955 Lund F, Tybring L (1972) 6-13Amidinopenicillanic acid A new group of antibiotics. Nature New Biol 236:135-137 Markowitz SM, Williams DS (1985) Effect of L-cysteine on the activity of penicillin antibiotics against Clostridium difficile. Antimicrob Agents Chemother 27:419421 Matsuhashi S, Kamiryo T, Blumberg PM, Linnet P, Willoughby E, Strominger JL (1974) Mechanism of action and development of resistance to a new amidino penicillin. J Bacteriol 117:578-587 Messing J (1979) A multipurpose cloning system based on singlestranded DNA bacteriophage M13. Recomb DNA Tech Bull 2(2): 43 Miller JH (1972) Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Newman EB, D'Ari R, Lin RT (1992) The leucine-Lrp regulon in E. coli: a global response in search of a raison d'etre. Cell 68:61%619 Ogura T, Bouloc P, Niki H, D'Ari R, Hiraga S, Jaff6 A (1989) Penicillin-BindingProtein 2 is essential in wild-type Escherichia coli but not in lov or cya mutants. J Bacteriol 171:3025-3030 Park JT, Burman L (1973) FL1060 A new penicillinwith a unique mode of action. Biochem Biophys Res Commun 51:863-868 Spratt BG (1975) Distinct penicillinbinding proteins involved in the division, elongation, and shape of Escherichia coli K-12. Proc Natl Acad Sci USA 72:2999-3003 Spratt BG (1977) A comparison of the binding properties of two 6 13-amidinopenicillanic acid derivatives which differ in their physiological effects on Escherichia coli. J Antimicrob Chemother 11:161-166 Spratt BG, Pardee AB (1975) Penicillin-bindingprotein and cell shape in E. coli. Nature 254:515-517 Tuomanen E, Tomasz A (1991) Mechanism of phenotypic tolerance of nongrowing pneumococci to beta-lactam antibiotics. Scand J Infect Dis 74:102-112 Uzan M, Danchin A (1976) A rapid test for the relA mutation in E. coll. Biochem Biophys Res Commun 69:751-758 Uzan M, Danchin A (1978) Correlation between the serine sensitivity and the derepressibility of the ilv genes in Eseherichia coli relA- mutants. Mol Gen Genet 165:21-30 Vanderwel D, Ishiguro EE (1984) Properties of cell wall peptidoglycan synthesized by amino acid deprived relA mutants of Escherichia coli. Can J Microbiol 30:1239 1246 Vinella D, D'Ari R, Bouloc P (1992) Penicillin-bindingprotein 2 is dispensable in Escherichia coli when ppGpp synthesis is induced. EMBOJ 11:1493 1501 C o m m u n i c a t e d by R. D e v o r e t
OIG'G
{3 Springer-Verlag 1992
Leucine and serine induce mecillinam resistance in Escherichia coli Philippe Bouloc, Daniel Vinella, and Richard D'Ari Institut Jacques Monod, CNRS, Universit6 Paris 7, 2, place Jussieu, F-75251 Paris Cedex 05, France Received March 27, 1992 / Accepted June 11, 1992
Summary. We have previously shown that resistance to the ~-lactam mecillinam in Escherichia coli can be brought about by a high ppGpp pool, as observed under conditions of partial amino acid starvation and RelAdependent induction of the stringent response. We show here that our E. coli wild-type strain, which is sensitive to mecillinam on minimal glucose plates, becomes resistant in the presence of L-leucine or L-serine (or cysteine, which inactivates the antibiotic). The resistance, which is not a transient effect and does not depend on the physiological state of the cells when plated, is specific for mecillinam and is reversed by the presence of isoleucine and valine in the medium. At least in the case of serine, the resistance is RelA-dependent. We conclude that the presence of leucine and serine in the growth medium cause partial starvation for isoleucine/valine, leading to induction of the stringent response and concomitant resistance to mecillinam. Key words: Mecillinam - Penicillin-binding protein 2 Stringent response - L-leucine - L-serine
Introduction The harmonious growth of a cell requires tight homeostatic regulation of virtually all metabolic pathways and strict coupling amongst the various macromolecular biosyntheses. The bacterium Escherichia eoli has provided a convenient model system for unravelling many of these regulatory mechanisms and interactions. Our present state of knowledge, although far from complete, provides paradigms for a number of regulatory circuits ensuring both balanced growth and rapid responses to environmental changes. One such system, the stringent response, helps provide homeostatic regulation of the protein synthesizing machinery, adjusting the cell's protein synthetic capacity to its ability to form aminoCorrespondence to: P. Bouloc
acyl-tRNA precursors (Cashel and Rudd 1987). The central regulator of the stringent response is the stringent factor, coded for by the relA gene. This protein, which is associated with ribosomes, catalyses the synthesis of the nucleotide pppGpp whenever an uncharged cognate tRNA occupies the A site of an actively translating ribosome. The nucleotide pppGpp is rapidly hydrolysed to ppGpp, and this nucleotide, in turn, is an RNA polymerase effector which reduces the transcription of most genes coding for elements of the protein synthesizing machinery, including all ribosomal RNA and protein genes. Thus, a low charge ratio for any tRNA species rapidly results in a lowered rate of ribosome production. In addition to regulating ribosome production, the stringent response is also involved in the regulation of bacterial envelope synthesis. The penicillin-binding proteins (PBPs) are integral membrane proteins responsible for the synthesis of the rigid peptidoglycan layer, ppGpp has been reported to be a negative effector of some step in peptidoglycan synthesis and is involved in autolysin regulation (Ishiguro and Ramey 1976, 1978; Goodell and Tomasz 1980; Vanderwel and Ishiguro 1984; Kusser and Ishiguro 1987). Furthermore, cells deprived of an essential element have a relA-dependent "phenotypic resistance" to a broad range of [3-1actam antibiotics, which are associated with peptidoglycan changes (Kusser and Ishiguro 1987; Tuomanen and Tomasz 1991). More recently, we have shown that ppGpp specifically alters the cell's requirement for PBP2 (Vinella et al. 1992). This PBP plays a crucial role in maintaining the rod shape of E. coli (Spratt and Pardee 1975). Inactivation of PBP2, whether by the specific antibiotic mecillinam or by mutation of the structural gene pbpA, results in the formation of spherical cells which normally stop dividing, increase in volume and die (Lurid and Tybring 1972; Park and Burman 1973; Matsuhashi et al. 1974; James et al. 1975; Spratt 1975, 1977; Ogura et al. 1989; Vinella et al. 1992). The lethality associated with PBP2 inactivation can be avoided in certain genetic backgrounds, including mutants possessing a partially defective aminoacyl-tRNA synthetase, which causes a high
243 ppGpp pool; these mutants are resistant to mecillinam (Vinella et al. 1992). We showed that this partial induction of the stringent response is in fact responsible for the cells' ability to tolerate PBP2 inactivation: introduction of a relA mutation restored mecillinam sensitivity to these t R N A synthetase mutants, and a multicopy plasmid, prelA', coding for a hyperactive (p)ppGpp synthetase, conferred mecillinam resistance on wild-type strains (Vinella et al. 1992). We concluded that PBP2 is normally an essential protein whose inactivation results in blockage of some vital process other than cell wall elongation, possibly cell division, but that a high ppGpp pool short-circuits this blockage, either by interrupting the communication between PBP2 and the vital process concerned, or by otherwise bypassing the latter. It has long been known that relA mutants, incapable of synthesizing ppGpp in response to aminoacyl-tRNA limitation, are hypersensitive to physiological situations such a~snutritional shifts down, in which there is an initial starvation period. In addition, reIA mutants exhibit a series of as yet unexplained sensitivities to certain apparently anodyne metabolites. Alf61di and Kerekes (1964) showed that relA strains growing exponentially in rich medium, although able to adapt to minimal glucose medium, have greatly reduced colony-forming ability on glucose plates supplemented with leucine, serine, cysteine, isoleucine or phenylalanine. Uzan and Danchin (1976, 1978) later showed that relA mutants are hypersensitive to serine, or to serine + glycine + methionine (SMG), and further showed that addition of SMG to exponentially growing cultures of relA + strains causes a temporary inhibition of growth, accompanied by starvation for isoleucine and valine and induction of the stringent response. It was suggested that relA mutants, being unable to induce the stringent response, somehow get locked in a physiological impasse in certain conditions (Alf61di and Kerekes 1964). It is reasonable to suppose that all conditions to which reIA mutants are hypersensitive cause induction of the stringent response, since (p)ppGpp synthesis is the only known activity of the RelA protein. We have shown that induction of the stringent response enables the cell to tolerate inactivation of PBP2, by mecillinam or by mutation, without blocking cell division or other vital processes (Vinella et al. 1992). This led us to examine the mecilliham sensitivity of a wild-type E. coli strain under physiological conditions in which reIA mutants have reduced colony-forming ability. We show here that wild-type E. coli is in fact permanently resistant to mecillinam under such conditions. Furthermore, the resistance is reIA +-dependent, indicating that certain metabolites in the growth medium can cause lasting partial induction of the stringent response in wild-type cells. Our results indicate, surprisingly, that the presence in the growth medium of certain amino acids such as leucine or serine can significantly reduce the rate of ribosome synthesis.
et al. 1977) and UTH1038 (Goldschmidt et al. 1970). A relA1 derivative, GC3698 (Vinella et al. 1992), was constructed by Plvir mediated transduction. The lrp::TnlO allele was transduced into GC2553 from strain NEW26 (Lin et al. 1990), kindly given us by E. Newman. Other prototrophic strains were NEW1 (Lin et al. 1990), 594 (Campbell 1961), JM101 (Messing 1979) and W3110 (Bachmann 1987). Media and culture conditions. Rich medium was LB broth and minimal medium was M63 (Miller 1972), supplemented with 0.4% glucose, 1 gg/ml thiamine and amino acids at 100 gg/ml. All amino acids (except glycine) were L stereoisomers. Casamino acids, treated with activated charcoal, was used at 0.4%. Mecillinam, generously provided by Laboratoires Leo (France), was used at 10 gg/ml. Nitrocefin was purchased from Laboratoires Glaxo (France). Solid media contained 1.5% Difco agar. Cysteine solutions (1% in water) and cysteine-containing media were prepared less than 24 h before use. Liquid cultures were grown in shaking water baths; all experiments were carried out at 37 ° C.
Results and discussion
Mecillinam resistance in the presence of certain amino acids
The mecillinam sensitivity of E. coli in minimal glucose medium is strain dependent. Spratt reported that his strain was resistant under these conditions (Barbour et al. 1981), whereas our wild-type E. coli K12 strain, GC2553, is sensitive (Bouloc et al. 1988, 1989). We examined its mecillinam sensitivity on glucose plates containing different supplements by plating an overnight glucose-grown culture. The results showed clear mecillinam
Table 1. Mecillinam resistance of wild-type Escherich& coli in the presence of certain amino acids Supplementsa
Efficiencyof plating - Mec + Mec
None Leucine Serine SMG Cysteine Arginine Histidine Lysine Methionine Phenylalanine Proline Threonine Tryptophan Tyrosine Casamino Acids
1.00 1.08 1.01 0.95 0.99 0.98 1.11 1.15 0.97 1.06 1.05 1.04 0.98 1.04 1.02
2.8 x 10-3 0.94 0.21 1.06 0.79 1.4 x 10- 3 1.8 x 10-3 2.4 x 10-3 2.3 × 10-3 4.3 x 10- 3 1.6 x 10.3 0.6 x 10-3 3.7 x 10-3 2.8 x 10-3 1.3 x 10-3
Materials and methods
Bacterial strains. Our wild-type E. coli K12 strain was GC2553, an F - @ ) - prototroph also called FB8 (Bruni
Mec, Mecillinam; SMG, serine+ methionine + glycine a Amino acids were added at 100 ~tg/ml, mecillinam was used at 10 gg/ml
244 Table 2. Effectof isoleucine and valine on mecillinamresistance Supplements None Ile+Val Leu÷ Ile+Val Ser+ Ile+Val SMG+Ile+Val
Efficiencyof plating - Mec + Mec 1.00 1.07 0.98 0.82 1.05
5.2 x 10.3 4.5 x 10 3 4.4 x 10.4 9.7 x 10.3 9.1 x 10 2
Mec, Mecillinam;SMG, serine+methionine÷glycine
resistance on plates containing leucine, serine, SMG or cysteine (Table 1). As has previously been observed in work with mecillinam-resistant mutants, growth in the presence of mecillinam was slow, with the single exception of the cysteine-supplemented plates, on which colonies appeared equally rapidly in the presence and absence of mecillinam. No mecillinam resistance was observed in the presence of isoleucine or phenylalanine, to which relA mutants were previously reported to be sensitive (Alf61di and Kerekes 1964), or in the presence of amino acids not known to inhibit relA strains (Table 1). Certain amino acids clearly protect E. coli against mecillinam in minimal medium. Casamino acids, however, which contain leucine, serine, methionine and glycine, did not afford protection (Table 1), even if 100 gg/ ml additional leucine was added (data not shown). The sensitivity of relA mutants to SMG is relieved by isoleucine and valine (Uzan and Danchin 1978), which are also present in Casamino acids. We therefore evaluated the effect of isoleucine and valine on the mecillinam resistance of wild-type E. coli in the presence of leucine, serine or SMG. The presence of Ile + Val suppressed the protective effect of leucine, serine and SMG (Table 2). Cysteine again proved to be different: the presence of 100 gg/ml cysteine in glucose Casamino acids mecillinam plates permitted colony formation (efficiency of plating 0.6).
Independence of mecillinam resistance on 9rowth phase The sensitivities reported for reIA mutants under shiftdown conditions were only observed in the exponential phase of growth (Alf61di and Kerekes 1964). We therefore examined the effects on mecillinam resistance of the physiological state of the culture when plated. Strain GC2553 was grown to exponential phase in glucose minimal medium, then plated on glucose plates supplemented with leucine, serine, SMG or cysteine, with or without mecillinam. With no amino acid supplement, the survival in the presence of mecillinam was 4 x 10-3, whereas with each of the supplements, the efficiency of plating was greater than 50%. Thus the mecillinam resistance observed in the presence of leucine, serine, SMG or cysteine does not depend on the physiological state of the bacteria when plated. When the same supplements were added in liquid culture, growth slowed down for some time, then re-
sumed, as previously reported for serine and SMG (Uzan and Danchin 1978). One hour after the additions, the cultures were observed under the phase contrast microscope. No morphological changes were apparent; in particular, the cells remained rod-shaped, indicating that PBP2 function was not impaired. Leucine, serine, SMG and cysteine, when added to a culture of wild-type cells, clearly cause a shock, presumably resulting in induction of the stringent response in relA + strains. It was conceivable that the mecillinam resistance observed was related to this shock and that the addition of mecillinam to the medium at the same moment in some way froze the cells in a special "shocked" physiological state with a high ppGpp pool. To test this possibility, the cells were allowed first to adapt to these metabolites in liquid culture, then examined for resistance to mecillinam. Cultures were grown overnight (> 6 generations) in glucose medium supplemented with leucine, serine, SMG or cysteine, then plated on glucose mecillinam plates containing the same supplements. In all cases, the efficiency of plating was greater than 15 %. We conclude that cells growing in the presence of these metabolites are permanently locked in a physiological state which permits PBP2 inactivation, even many generations later, presumably via partial induction of the stringent response (see below). The presence of leucine in the growth medium radically alters the rate of expression of a number of E. coli genes. It has recently been shown that these genes form part of a global response, the "leucine/Lrp regulon", under the control of the /eucine-responsive regulatory protein (Lrp), a transcriptional activator/repressor coded for by the lrp gene (Newman et al. 1992). In Irp mutants, many of the member operons behave as though the exogenous leucine concentration were high; in particular, expression of the leu operon is low (Lin et al. 1992), and indeed these strains appear to be partially starved for leucine (Lin et al. 1990). It therefore seemed possible that lrp mutants cultivated in the absence of leucine would be resistant to mecillinam. An lrp::TnlO insertion was transduced into the wild-type strain GC2553, which was then plated on glucose mecillinam plates. The transduced strain showed the same sensitivity (survival 1.7 x 10 -3) as the lrp + parent strain, but became resistant to mecillinam in the presence of exogenous leucine, again like the parent strain. It was noticeable that the colony size of both lrp and lrp + cells was smaller on leucine-supplemented plates than on glucose alone. The growth rate of the wild-type strain GC2553 growing exponentially in glucose medium with or without supplements was measured. With no amino acids, the doubling time was 51 min; leucine increased the doubling time by 17% and serine increased it by 10%, leading to the paradoxical result that the cells grow more slowly in a richer medium. SMG, although not increasing the doubling time, provided, as we have seen, total protection against mecillinam (Table 1).
245 Involvement of the stringent response The mecillinam resistance observed in the presence of certain amino acids seemed to reflect the induction of the stringent response, apparently by creating partial starvation for isoleucine and/or valine. To verify the role of the stringent response in the mecillinam resistance observed, we tested a relA1 derivative of strain GC2553. The sensitivities of relA mutants were originally observed only under shift-down conditions (Alf61di and Kerekes 1964). To avoid a nutritional shift-down, a saturated overnight glucose culture was plated on glucose medium containing various supplements, with or without mecillinam. Our relA strain proved sensitive to leucine and SMG, even in the absence of a nutritional shift-down, so the relA dependence of the protection afforded by these metabolites could not be tested. However, the relA mutant was able to grow on serine and cysteine plates [efficiency of plating (e.o.p.) 60%]: serine did not confer mecillinam resistance (e.o.p. 1 x 10 -3) while cysteine, on the other hand, conferred resistance.
Generality of the phenomenon We tested four other prototrophic K12 strains, NEW1, 594, JM101 and W3110, to see whether the presence of L-leucine conferred mecillinam resistance. In all cases, however, the strains were mecillinam resistant in unsupplemented glucose medium, like the strain used by Barbour et al. (1981). We have shown that our "wildtype" strain GC2553 in fact carries a mutation in the 91 rain region of the genetic map, which provides mecilliham sensitivity unsupplemented glucose medium. Furthermore, the mutation can be lethal in certain conditions, and this lethality is suppressed by the introduction of aryS or alaS alleles or the plasmid prelA', all of which confer mecillinam resistance in rich media (Vinella et al. 1992). This work will be fully described in a forthcoming publication (D. Vinella, P. Bouloc and R. D'Ari, manuscript in preparation). We further investigated whether the resistance conferred by leucine, serine, SMG and cysteine was limited to mecillinam (which specifically inhibits PBP2) or constitutes a general protection against 13-1actam antibiotics. For this purpose an overnight glucose culture of strain GC2553 was platted on glucose plates containing various supplements, with or without 10 gg/ml ampicillin, which binds to all of the PBPs. Leucine, serine and SMG did not protect the cells against ampicillin (e.o.p.< 10-4). This shows that we are not dealing with the phenomenon of phenotypic resistance. Cysteine, in contrast, provided essentially full protection (e.o.p. 0.3).
antibiotics, independently of the cells (Markowitz and Williams 1985), and we confirmed that 80 gg/ml cysteine in aqueous solution rapidly hydrolyses the colour indicator nitrocefin, a cephalosporin analogue. It is perhaps worth pointing out that this highly reactive amino acid is spontaneously oxidized to cystine or cysteic acid after several days in aqueous solution and thus is not normally present in reduced form in rich media. Leucine, serine and SMG specifically confer resistance to mecillinam. The protection is reversed by the presence ofisoleucine and valine and, at least in the case of serine, requires a functional relA + gene product. We therefore conclude that the presence of leucine, serine or SMG in the culture medium results in partial starvation for isoleucine/valine and relA + dependent induction of the stringent response. The fact that addition of leucine, serine or SMG initially induces the stringent response was not unexpected, since relA mutants are hypersensitive to such shocks. The surprising finding was that, although relA + strains readily adapt to the presence of leucine, serine or SMG and resume growth, they apparently continue to be partially starved for isoleucine/ valine and the stringent response remains partially induced, conferring permanent mecillinam resistance. In the presence of leucine or serine, the doubling time is significantly longer than in unsupplemented glucose medium. It is also longer in the relA strain (once the cells have adapted to leucine or serine), as would be expected if growth is limited by isoleucine/valine availability, although in this case the stringent response cannot be induced and the bacteria remain sensitive to mecillinam. The wild-type strain was also resistant to mecillinam in the presence of SMG and, since isoleucine + valine reversed this resistance, the situation again resembled one of partial starvation. However, the cells actually grew faster in SMG than in unsupplemented medium. This may be related to our previous observation that there is no simple correlation between growth rate and mecilliham resistance (Bouloc et al. 1989; Vinella et al. 1992). Through the fortuitous discovery of an E. coli strain that is particularly sensitive to mecillinam in minimal glucose medium, we have discovered a mechanism of physiological regulation that is masked in some other strains. Our results show that glucose medium supplemented with leucine or serine, although richer than unsupplemented medium, affords a lower growth rate. This paradox may reflect the nature of the natural environments of E. coli, which tend to provide either very little in the way of nutrients (the outside world) or a mixture containing all amino acids (the gut). Media rich in only one or a few specific amino acids are perhaps so rare that E. coli has not learned to profit maximally from them.
Conclusions
Protection by cysteine clearly occupies a special place: it is relA-independent, is not reversed by isoleucine plus valine and extends to other ~-lactams. The mechanism of cysteine protection is, in fact, via chemical attack of the
Acknowledgements. We thank Elaine Newman for stimulatingdiscussions and Antonia Kropfinger for bibliographic research. Our research project on the mechanisms of mecillinam resistance was recognised by the Laboratoires Glaxo in the award of the Glaxo "PBP" Prize to P.B. This work was financedin part by the Association pour la Recherche sur le Cancer (contract no. 6696).
246
References Alf61di L, Kerekes E (1964) Neutralization of the amino acid sensitivity of R C rd Escherichia coll. Biochem Biophys Acta 91:155-157 Bachmann BJ (1987) Derivations and genotypes of some mutant derivatives of Eseherichia coli K-12. In: Neidhardt FC, Ingraham JL, Low KB, Magasanik B, Schaechter M, Umbarger HE (eds) Escherichia eoli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, DC, pp 1190-1219 Barbour AG, Mayer LW, Spratt BG (1981) Mecillinam resistance in Escherichia coli: dissociation of growth inhibition and morphologic change. J Infect Dis 143:114-120 Bouloc P, Jaff6 A, D'Ari R (1988) Preliminary physiologic characterization and genetic analysis of a new Escherichia eoli mutant, lov, resistant to mecillinam. Rev Infect Dis 10:905-910 Bouloc P, Jaff~ A, D'Ari R (1989) The Escherichia coli lov gene product connects peptidoglycan synthesis, ribosomes and growth rate. EMBO J 8 : 317-323 Bruni CB, Colantuoni V, Sbordone L, Cortese R, Blasi F (1977) Biochemical and regulatory properties of Escherichia coli K-12 hisT mutants. J Bacteriol 130:4-10 Campbell A (1961) Sensitive mutants to bacteriophage lambda. Virology 14:22 Cashel M, Rudd KE (1987) The stringent response. In: Neidhardt FC, Ingraham JL, Low KB, Magasanik B, Schaechter M, Umbarger HE (eds) Escherichia coli and Salmonella typhimurium : cellular and molecular biology. American Society for Microbiology, Washington, DC, pp 1410 1438 Goldschmidt EP, Cater MS, Matney TS, Butler MA, Greene A (1970) Genetic analysis of the expression of the histidine operon in Salmonella typhimurium. Genetics 66:219-229 Goodell W, Tomasz A (1980) Alteration of Escherichia coli murein during amino acid starvation. J Bacteriol 144:1009-1016 Ishiguro EE, Ramey WD (1976) Stringent control of peptidoglycan biosynthesis in Escherichia coli K-12. J Bacteriol 127:1119-1126 Ishiguro EE, Ramey WD (1978) Involvement of the relA gene product and feedback inhibition in the regulation of UDP-Nacetylmuramyl peptide synthesis in Escheriehia coli. J Bacteriol 135:766-774 James R, Haga JY, Pardee AB (1975) Inhibition of an early event in the cell division cycle of Escherichia coli by FL1060, an amidinopenicillanicacid. J Bacteriol 122:1283-1292 Kusser W, Ishiguro EE (1987) Suppression of mutations conferring penicillin tolerance by interference with the stringent control mechanism of Escherichia coli. J Bacteriol 169:43964398 Lin RT, D'Ari R, Newman EB (1990) The leucine regulon of Escherichia coli K-12: a mutation in rblA alters expression
of L-leucine-dependent metabolic operons. J Bacteriol 172: 45294535 Lin RT, D'Ari R, Newman EB (1992))~ placMu insertions in genes of the leucine regulon: extension of the regulon to genes not regulated by leucine. J Bacteriol 174:1948-1955 Lund F, Tybring L (1972) 6-13Amidinopenicillanic acid A new group of antibiotics. Nature New Biol 236:135-137 Markowitz SM, Williams DS (1985) Effect of L-cysteine on the activity of penicillin antibiotics against Clostridium difficile. Antimicrob Agents Chemother 27:419421 Matsuhashi S, Kamiryo T, Blumberg PM, Linnet P, Willoughby E, Strominger JL (1974) Mechanism of action and development of resistance to a new amidino penicillin. J Bacteriol 117:578-587 Messing J (1979) A multipurpose cloning system based on singlestranded DNA bacteriophage M13. Recomb DNA Tech Bull 2(2): 43 Miller JH (1972) Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Newman EB, D'Ari R, Lin RT (1992) The leucine-Lrp regulon in E. coli: a global response in search of a raison d'etre. Cell 68:61%619 Ogura T, Bouloc P, Niki H, D'Ari R, Hiraga S, Jaff6 A (1989) Penicillin-BindingProtein 2 is essential in wild-type Escherichia coli but not in lov or cya mutants. J Bacteriol 171:3025-3030 Park JT, Burman L (1973) FL1060 A new penicillinwith a unique mode of action. Biochem Biophys Res Commun 51:863-868 Spratt BG (1975) Distinct penicillinbinding proteins involved in the division, elongation, and shape of Escherichia coli K-12. Proc Natl Acad Sci USA 72:2999-3003 Spratt BG (1977) A comparison of the binding properties of two 6 13-amidinopenicillanic acid derivatives which differ in their physiological effects on Escherichia coli. J Antimicrob Chemother 11:161-166 Spratt BG, Pardee AB (1975) Penicillin-bindingprotein and cell shape in E. coli. Nature 254:515-517 Tuomanen E, Tomasz A (1991) Mechanism of phenotypic tolerance of nongrowing pneumococci to beta-lactam antibiotics. Scand J Infect Dis 74:102-112 Uzan M, Danchin A (1976) A rapid test for the relA mutation in E. coll. Biochem Biophys Res Commun 69:751-758 Uzan M, Danchin A (1978) Correlation between the serine sensitivity and the derepressibility of the ilv genes in Eseherichia coli relA- mutants. Mol Gen Genet 165:21-30 Vanderwel D, Ishiguro EE (1984) Properties of cell wall peptidoglycan synthesized by amino acid deprived relA mutants of Escherichia coli. Can J Microbiol 30:1239 1246 Vinella D, D'Ari R, Bouloc P (1992) Penicillin-bindingprotein 2 is dispensable in Escherichia coli when ppGpp synthesis is induced. EMBOJ 11:1493 1501 C o m m u n i c a t e d by R. D e v o r e t
