The glucose-6-phosphate-isomerase reaction is essential for normal glucose repression in Saccharomyces cerevisiae

Transcription

1 Eur. J. Biochem. 214, (1993) 0 FEBS 1993 The glucose-6-phosphate-isomerase reaction is essential for normal glucose repression in Saccharomyces cerevisiae Laurens N. SERKSTRA', Herman H. W. SLLJE', John M. A. VERBAKEL' and C. The0 VERRPSZ Department of Molecular Cell Biology, University of Utrecht, The Netherlands Unilever Research Laboratorium Vlaardingen, The Netherlands (Received January 26Debruary 25, 1993) - EJB /6 Wild-type Saccharomyces cerevisiae and a strain carrying a deletion in the glucose-6-phosphateisomerase gene (pgil) were grown in carbon-limited continuous cultures on a mixture of fructose and galactose. Pulses of glucose, fructose and galactose were given to these cultures to investigate whether the pgil strain was capable of normal glucose repression. Glucose and galactose pulses inhibited fructose consumption and thus glycolysis in the pgil strain by a combination of competition between glucose and fructose at the uptake and/or phosphorylation level and inhibition of fructose uptake and/or phosphorylation by glucose 6-phosphate. Fructose pulses administered to the pgil strain transiently decreased the glycolytic flux downstream of fructose-l,6-bisphosphate. Transcriptional induction of the PDCl gene (encoding pyruvate decarboxylase) was observed after glucose or galactose pulses were applied to the pgil strain, demonstrating that metabolism of these sugars beyond glucose 6-phosphate is dispensable for PDCl induction. Fructose also induced PDCl transcription, indicating that intracellular sugars could act as trigger for PDC induction or, alternatively, that two inductors are present. n contrast to the wild-type transcriptional inhibition of the glucose-repressible genes, HXKl and GAL1 0 (encoding hexokinase isoenzyme 1 and uridine diphosphoglucose 4-epimerase, respectively) did not occur upon addition of glucose or fructose to the pgil mutant. Transcriptional repression was observed after application of the fructose pulse when the yeast had resumed metabolism of fructose. These results demonstrate that the initial signal for catabolite repression is not generated by high sugar concentrations or high concentrations of intermediates; moreover a simple role for the hexokinases can also be excluded. The absence of an increased glycolytic flux in the pgil mutant after administration of the sugar pulses while the concentrations of sugar and glycolytic intermediates were high, suggests that the initial signal for glucose repression could be linked to an increased glycolytic flux. The occurrence of PDCl induction in the pgil strain while GALOHXKZ repression is absent, demonstrates that the initial signals for catabolite induction and catabolite repression are different. The addition of glucose to derepressed cultures of Saccharomyces cerevisiae inhibits the transcription of glucoserepressible genes, e.g SUC2 (encoding invertase), HXKl (gene encoding hexokinase isoenzyme ) and the GAL (encoding the enzymes of the Leloir-pathway [l]) within 2-3 min [2]. This results in the long-term repression of enzyme synthesis known as catabolite repression [3]. Moreover, catabolite inactivation and the glucose-dependent inactivation of Correspondence to J. F. Deij, Management Support and Communication, Unilever Research Laboratorium, P. 0. Box 114, NL AC Vlaardingen, The Netherlands Fax: Abbreviations. SUC2, HXKl, HXK2, PDC, GAL10 and ACT1 are the genes encoding invertase, hexokinase isoenzyme, hexokinase isoenzyme 11, pyruvate decarboxylase, uridine diphosphoglucose 4-epimerase and actin, respectively ; pgil, glucose-6-phosphateisomerase deletion strain; GAL-genes, genes encoding the enzymes of the Leloir pathway. Enzymes. Glucose-6-phosphate isomerase (EC ); hexokinase (EC ); invertase (EC ); fructose-l,6-bisphosphatase (EC ); 6-phosphofructo-2-kinase (EC ); maltase (EC ); pyruvate decarboxylase (EC ); uridine diphosphoglucose 4-epimerase (EC ). proteins, e.g. the galactose pennease [4] and fructose-l,6- bisphosphatase [5], also occurs. nvolvement of the RAS CAMP pathway in the catabolite-inactivation process has been suggested [6, 71 but this pathway seems not to be involved in the transcriptional regulation of glucose-repressible genes [8]. The only parts of the signal-transduction pathway leading to glucose repression which have been identified are the nuclear components which interact with DNA 13, 91. These factors, however, are part of the general transcription machinery regulating a variety of induciblekepressible genes [3, 10, 111. The only transcription factor identified so far which appears specific for the glucose-signalling pathway is MGl (multicopy inhibitor of Gall promoter) [12], for which consensus binding sites have been identified in a number of glucose-regulated genes [2, 13, 141. n addition to catabolite repression and catabolite inactivation, catabolite induction (e.g. an increase in PDCl gene transcription) [ 151 and catabolite activation (e.g. increased 6-phosphofructo-2-kinase-specific activity) [ 161 occurs upon glucose addition to derepressed yeast cells and it is not known whether separate initial signals and signal-transduction pathways exist for these responses.

2 122 Experiments in which the long-term repression of invertase and/or maltase were analyzed under a variety of growth conditions in different glycolytic mutants have identified a number of possible signals for glucose-induced responses. Galactose does not repress invertase synthesis [3], which indicates that metabolism beyond glucose 6-phosphate is not essential for glucose repression. Furthermore, non-metabolizable glucose analogues (2-deoxyglucose, glucosamine) do exhibit glucose repression [17, 181, suggesting a specific role for high intracellular glucose and glucose-6-phosphate levels. Since mutants in the gene encoding hexokinase isoenzyme 1 (HXK2) showed constitutive expression of glucose-repressible genes [19], a role for hexokinase isoenzyme 1 in glucose repression was implied. t was postulated that this enzyme contained two domains; one regulatory and one catalytic [19]. Subsequent experiments, concentrating on the identification of these domains [20], revealed that the catalytic activity of hexokinase 1 was closely correlated with the extend of glucose repression of invertase activity. Therefore, it was suggested that the role of the hexokinases in glucose repression could be in the phosphorylation of a substrate (possibly glucose) which could subsequently act as the trigger for glucose repression [21, 221. Another hypothesis is that the ratios or fluxes of glycolytic intermediates could trigger glucose repression. Moreover, the available amount of ATP, phosphate and the redox potential of the cell or cell organelles may be important. n an attempt to elucidate the signal(s) for glucose repression and induction we have studied the expression of glucose-regulated genes in a glucose-6-phosphate-isomerasedeletion strain because such a strain excludes certain metabolites as initial triggers for glucose repression. The transcript levels of glucose-responsive genes were measured in preference to the enzyme activities of invertase or maltase in order to reduce the chance of non-specific interference from processes not directly related to the initial signal for glucose repression. The glucose-6-phosphate isomerase deletion strains (pgil) are unable to grow on either fructose, glucose or galactose as a sole carbon source and therefore a combination of carbon sources must be provided to supply both glucose 6-phosphate (e.g. glucose or galactose) and a carbon source which can enter the citric acid cycle (e.g. fructose or glycerol) [23, 241. Glucose concentrations higher than 0.1% are inhibitory for growth of the pgil strain [23, 241 and, therefore, it is impossible to cultivate this strain under identical conditions to the wild-type strain in batch cultures. Continuous cultures offer the possibility to grow the pgil and wild-type strains under identical, standardized conditions. Therefore, the pgil and wild-type strains were cultivated in carbon-limited continuous cultures at a dilution rate of 0.1 h-' on a mixture of fructose and galactose. Subsequently, pulses of fructose, glucose or galactose were given to these cultures in order to determine the effect of the pgil deletion on the transcript levels of glucose repressible or inducible genes. MATERALS AND METHODS Strains and growth conditions Strain ENY. WA-1A (MATa, ura3-52, leu2-3, 112, trpl- 289,his3-deltu 1, MAL2-8", MAL3, SUC3) and the construction of the pgil deletion strain EB23-1 have been described [24]. The strains were grown on previously described minimal medmm [l, 251 with the following modifications. 4 mgl-' folic acid, 0.8 gl-' nicotinic acid, 0.4 gl-' p-aminobenozoic- acid and 0.2 gl riboflavine were added to the vitamin solution (a 1OOOX stock solution) and 300mgl-' leucine, 50 mgl-' histidine and 50 mgl-' tryptophan were added to the medium. n the case of the wild-type strain, the medium was supplemented with 50 mgl-' uracil. The limiting carbon source was a mixture of 10 gl-' fructose and 2 gl-' galactose for the wild-type strain, while the pgil strain was cultivated on ' fructose and 4 gl-' galactose as limiting carbon sources. Both strains were grown under carbon limitation at 30 C in a 2-1 fermenter connected to an Applikon AD11020 controller unit. The airflow and stirrer speed were set at 2 lmin-' and 250 rpm respectively, while the oxygen tension in the medium was always above 50%. The ph was automatically controlled at 5.0 by the addition of 2 M NH,OH. Carbon dioxide production, oxygen consumption and ethanol production were measured on-line by the connection of the headspace of the fermenter to a VG gas-analysis mass-spectrometer MMS-80 (VG nstruments). Sugar pulses The wild-type and pgil deletion strains were grown at a steady-state dilution rate of 0.1 h-'. To the steady-state cultures of the wild-type strain, 60 mm sugar pulses were given. 60 mm fructose pulses and 30 mm glucose or galactose were given to the carbon-limited continuous culture of the pgil strain. Samples Samples for the determination of residual sugars, intracellular metabolites, dry mass and nlrna analysis were taken from the fermenter as previously described [2, 261. Extracts for metabolite determination and for mrna analysis were performed as previously described [2,26]. The determination of the residual-sugar concentrations and intracellular metabolites were also performed as described in [2, 261. Northern-blot analysis For the detection of the mrna of HXKl, GALO, PDCl and ACT1, previously described oligonucleotides were used [2]. The separation of 5 pg total RNA, blotting of the gel, the hybridisation procedure and washing conditions have been described [2]. n all experiments, the actin mrna level was used as an internal control for the amount of RNA loaded onto the gel by means of double hybridizations. Northern blots were quantified by densitometry on a LKB Ultrascan XL. Different exposure times were used in order to obtain reliable densitometry data. RESULTS Metabolism after administration of the sugar pulses The wild-type and pgil strain were grown in galactose/ fructose-limited continuous cultures at a steady-state dilution rate of 0.1 h-'. Either glucose, fructose or galactose pulses, administered to the wild-type strain, resulted immediately in an increased glycolytic flux which was accompanied by a simultaneous increase in carbon dioxide and ethanol production (Fig. 1A). Essentially, these results are identical to those obtained for a commercial yeast strain [2]. Pulses of glucose and galactose administered to the pgil strain resulted in an immediate inhibition of glycolysis

3 mrnolh-' PPm *O r rnmoih-' B h k a v h E, L1 v U g * Time (h) mmolh l-., , _ PPm Tirne(h) FRUCTOSE h L: - E, n v U 8 3 U 250 g - 0 t a s W Time (h) Fig. 1. Metabolism of the sugars after administration as a pulse to the wild-type and the pgil strain. The carbon dioxide production (-) and oxygen consumption (-.-) and ethanol production (...) after the addition of (A) glucose to the wild-type strain; results obtained for fructose and galactose are essentially identical. (B) Addition of galactose to the pgil strain (carbon dioxide production after a glucose pulse: --); (C) addition of fructose to the pgil strain. (Fig. 1B). nhibition of glycolysis was most pronounced when glucose was pulsed (reduction in glycolytic flux to 10-20% of the initial value) as compared with the galactose pulse (reduction in glycolytic flux to 30-40% of the initial value). n both cases, metabolism could be partially restored by the addition of fructose (Fig. B). A fructose pulse, applied to the pgil strain, had an inhibitory effect on glycolysis for the first min, after which metabolism was resumed and resulted in enhanced glycolysis which was indicated by the ethanol production and enhanced carbon dioxide production (Fig. 1C). Residual-sugar concentrations Growth of both the pgil and wild-type strains at a dilution rate of 0.1 h-' on fructose/galactose resulted in similar residual-sugar concentrations of 1 mmoll-' fructose and 0.18 mmoll-' galactose. n both the wild-type and the pgil strains, a glucose pulse resulted in inhibition of fructose and galactose consumption, which was indicated by the increase in the residual concentrations of these sugars (Fig. 2A). The wild-type strain immediately started the consumption of the pulsed glucose (approximately 6 mmolg-'h-') while, in the pgil strain, consumption of glucose was low (approximately 0.9 mmolg-'h-'). A fructose pulse had little effect on the consumption of galactose by the pgil strain, while inhibition of galactose utilization was more pronounced in the wild-type strain (Fig. 2B). Enhanced consumption of fructose was directly observed for the wild-type strain (approximately 5.2 mmol g-lh-'), while fructose utilization was low in the pgil strain during the first 20 min after the pulse but resumed after min (approximately 4 mmolg-'h-'). A galactose pulse resulted in an increase in the residualfructose concentration for the pgil deletion strain in contrast to the wild-type strain for which only a small effect was

4 124 galactose (mmoll-1) fructose (mmoll-'j -6 galactose (mmoll-l) 08 B O ' Time (min) L------Jo OO fructose (mmoll-l j Fig. 2. Residual sugar concentrations after administration of sugar pulses to the wild-type and pgil strains. The residual galactose (0, 0) and fructose (U, 0) concentrations after the addition of (A) glucose, (B) fructose and (C) galactose. Wild-type (0, U); pgil (0, 13). The residual-sugar concentrations have a standard error of <lo%. observed (Fig. 2C). Galactose consumption immediately increased after the galactose pulse had been applied to the wild-type strain (approximately 4.5 mmolg-'h-') while galactose consumption was low in the pgil strain (approximately 1.3 mmolg'h-'). Glycolytic intermediates The effects of sugar pulses administered to the wild-type strain were essentially as described previously for a commercia1 baker's yeast strain [2]. Both glucose 6-phosphate and fructose 6-phosphate concentrations transiently increased after the addition of glucose and these concentrations re- turned to their normal level within 2-3 min. Pulses of fructose or galactose gave no increase in the glucose 6-phosphate level but the fructose pulse gave a slight increase in fructose 6-phosphate levels. An initial decrease in the ATP concentration was observed after administration of either glucose, fructose or galactose Dukes. but the ATP level returned to the initial valug within 21-3 min. The fructose-l,6-bisphosphate concentration increased from 7 umolc' to about 20 umoln-' v " within 1 rnin of glucose or f;uctose pulse administration while, after a galactose pulse, the concentration increased from 7 ymolg-' to 13 pmolg-'. No decrease in the fructose- 1,6-bisphosphate concentration was observed during the first 10 min after these pulses had been administered.

5 125 FruGp, ATP, Fru(1.6)P2 (prnolg-) GlcGp (ymolg-') 10, " Northern-blot analysis When fructose or glucose were administered to the wildtype strain, the levels of GAL10 and HXKl transcripts were reduced within 3 min. The transcripts were undetectable within 15 min and 7.5 min, respectively. These results are essentially as described previously for a commercial baker's yeast strain [2]. After the addition of galactose, a 2-3-fold induction of the GAL10 mrna could be seen after 60 min, while a decrease in the HXKl mrna was observed. PDCl induction occurred after the glucose (9-10-fold), fructose (5-6-fold) or galactose (2-3-fold) pulses (Fig. 4A). A glucose pulse administered to the pgil strain resulted in a reduction of the GAL10 transcript levels to 50% of the initial value within min. This effect was also observed after the addition of galactose. After glucose or galactose pulses had been administered, no apparent reduction in the level of the HXKl transcript occurred. Following the fructose pulse, repression of the GAL10 and HXKl transcript levels was observed after 30 min and 15 min, respectively. However, repression of transcription was not complete. nduction of PDCl transcription was observed after the glucose (4-5-fold), fructose (5-fold) or galactose (7 -%fold) pulses to the pgil strain (Fig. 4B). 0' Fig. 3. Glycolytic intermediates after administration of sugar pulses to the pgil strain. Glucose 6-phosphate (Glc6P; O), fructose 6-phosphate (Fru6P; U), fructose-l,6-bisphosphate (Fru(l,6)P2; 0) and ATP (0) concentrations after a galactose (A) and fructose (B) pulse. Results from the glucose pulse are identical to the results obtained for galactose. The glycolytic intermediates have a standard error of < 25 %. The galactose and glucose pulses, administered to the pgil strain, resulted in an increase in the glucose 6-phosphate concentration and a decrease in the fructose 6-phosphate, fructose-1,6-bisphosphate and ATP concentrations (Fig. 3A). The fructose pulse administered to the pgil strain caused a decrease in the glucose 6-phosphate concentration and a slight decrease in the fructose 6-phosphate concentration. n addition, the ructose-1,6-bisphosphate concentration increased, while the ATP level decreased (Fig. 3B). DSCUSSON The addition of glucose, fructose or galactose to the fructosefgalactose-limited continuous culture of the wild-type strain led to an increase in the glycolytic flux and the fructose and glucose pulses reduced the levels of transcripts of glucose-repressible genes within min. This decrease in transcript levels can be caused by inhibition of transcription andor increased degradation of the specific mrna species. The relative importance of these processes in glucose repression of specific genes is not known. For example, it has been reported that there is no contribution of mrna degradation to the glucose repression of the iso-cytochrome c gene [27, 281 while others claim that the expression level of the iron-protein subunit of succinate dehydrogenase is regulated by mrna degradation in addition to transcriptional inhibition [29]. Both genes are reported to be subject to glucose repression [27, 291. nduction of the PDCl gene transcript was observed in all cases, although to a lesser extend when galactose was pulsed. These results agree with previously reported results for a commercial baker's yeast strain [2]. Metabolism of the different sugars after pulses applied to the pgil strain was severely distorted. After glucose or galactose addition, a decrease in the glycolytic flux (as measured by the oxygen consumption and carbon dioxide production rates) was observed. Fructose addition restored metabolism in both cases, demonstrating that this decrease in the glycolytic flux was caused by a decreased fructose utilisation. Subsequent measurement of the levels of glycolytic intermediates and residual fructose concentrations showed that this decreased fructose metabolism was probably caused by a combination of competition at the uptake level between glucose and fructose and by an inhibition of the fructose uptake andfor phosphorylation by glucose 6-phosphate. These phenomena are probably the cause of the inhibitory effect of high glucose concentrations on the growth of the pgil strain in batch cultures. A transient decrease in the glycolytic flux and a simultaneous decrease in the steady-state concentration of ATP were observed when fructose was pulsed to the pgil strain. The increasing fructose-1,6-bis-

6 126 Fig. 4. Northern-blot analysis after application of pulses to the wild-type and the pgil strains. Northem-blot analysis of the GALO, HXKl and PDCl transcripts with actin as a control after application of glucose, fructose or galactose pulses to the wild-type (A) and pgil (B) strain. Northern blots were quantified by densitometry (except for HXKl which could not be scanned due to unsufficient discrimination between the HXKl and ACT1 bands). The amount of specific mrna is given beneath each lane as a percentage of the initial value at t = 0 min.

7 127 phosphate concentration observed after the fructose pulse indicated that the fructose uptake and/or phosphorylation was not inhibited. This demonstrates that the block in glycolysis must be downstream of the phosphofmctokinase reaction. The cause of this block in glycolysis could not be determined from the available data. nduction of PDCl in the pgil strain was observed with all carbon sources, which demonstrates that metabolism beyond glucose 6-phosphate is not essential for transcriptional induction of PDCl. This suggests that intracellular sugars could act as the initial trigger for PDCl induction. Alternatively, it could be that two inductors are present; glucose 6- phosphate and another downstream of fructose 6-phosphate. The latter seems more likely because induction is delayed after administration of a fructose pulse to the pgil strain. nduction appears to occur when the glycolytic flux increases and ethanol production is observed. Transcriptional induction of PDCl by the latter is not glucose specific and related to the fermentative metabolic state of the yeast [26]. The pulses to the pgil strain clearly showed that the conversion of glucose 6-phosphate to fructose 6-phosphate is essential for normal glucose-repression in yeast. Although no inhibition of glucose-repressible genes was observed when glucose was pulsed, the glucose 6-phosphate concentration increased. These results rule out intracellular glucose metabolites as the initial triggers for glucose repression, which is in contrast to experiments using non-metabolizable glucose analogues [18, 191. Recent experiments in which we pulsed 2-deoxyglucose to a galactose-limited continuous culture revealed that glucose repression only occurred after one hour of 2-deoxyglucose excess (unpublished results) when the yeast had stopped metabolism. This clearly demonstrates that 2-deoxyglucose is not a normal trigger for glucose-induced signalling in S. cerevisiae and, therefore, this glucose analogue should not be used to analyze the initial signal for glucose repression. n addition, our data exclude the simple model in which the hexokinases produce a phosphorylated fructose or glucose metabolite which then triggers catabolite repression [21, 221 because this model predicts that activation of the signal-transduction pathway should also have occurred in the pgil strain when glucose was pulsed. The occurrence of PDCl induction without repression of GAL1 O/HXK transcription demonstrates that the initial stimuli which trigger the signal-transduction pathways resulting in catabolite induction and catabolite repression are different in S. cerevisiae. The presence of high concentrations of extracellular glucose and intracellular glucose 6- phosphate in combination with a low glycolytic flux did not produce a signal for glucose repression in the pgil strain. Moreover, fructose only repressea transcription of glucoseregulated genes when the glycolytic flux was increased (ethanol production) and not when the amount of extracellular fructose, fructose-1,6-bisphosphate and fructose 6-phosphate were high. Ethanol production itself has already been excluded as the initial signal for glucose repression in other experiments [26]. Taken together, the results suggest that the initial signal for glucose repression is closely linked to the enhanced glycolytic flux after administration of the sugar pulses. Our results, however, cannot locate the position where the glycolytic flux is measured, but possible locations are the sugar-uptake system, sugar phosphorylation or the ratio of metabolites. Since a correlation exists with the catalytic activity of the hexokinases and the glucose repression of invertase and maltase, it is conceivable that the glycolytic flux is sensed in a complex containing the hexokinases which, when the glycolytic flux is high, phosphorylates a substrate which acts as the initial trigger for glucose repression. Currently we are performing experiments to test this hypothesis. The authors thank E. Boles and Dr F. K. Zimmermann for the gift of the yeast strains and Dr J. W. Chapman for critically reading the manuscript. REFERENCES 1. Douglas, H. C. & Hawthorne, D. C. (1964) Genetics 49, Sierkstra, L. N., Nouwen, N. P., Verbakel, J. M. A. & Verrips, C. T. (1992) Yeast 8, Gancedo, J. M. (1992) Eul: J. Biochem. 206, Holzer H. (1976) Trends Biochem. Sci., Gancedo, C. (1971) J. Bacteriol. 107, Ramos, J. & Cirillo, V. P. (1989) J. Bacteriol. 171, Rittenhouse, J., Moberly, L. & Marcus, F. (1987) J. Biol. Chem. 26, Hubbard, E. J. A., Yang, X. & Carlson, M. (1992) Genetics 130, Trumbly, R. J. (1992) Molec. Microbiol. 6, Peterson, C. L. & Herskowitz,. (1992) Cell 68, Keleher, C. A,, Redd, M. J., Schultz, J., Carlson, M. &Johnson, A. D. (1992) Cell 68, Nehlin, J. 0. & Ronne, H. (1990) EMBO J. 9, Mercado, J. J., Vincent, 0. & Gancedo, J. M. (1991) FEBS Lett. 291, Nehlin, J. O., Carlberg, M. & Ronne, H. (1991) EMBO J. 10, Schmitt, H. D., Ciriacy, M. & Zimmermann, F. K. (1983) Molec. Gen. Genet. 192, Thevelein, J. M. (1988) Exp. Mycol. 12, Gancedo, C. & Gancedo J. M. (1985) Eur. J. Biochem. 148, Furst, A. & Michels, C. A. (1977) Molec. Gen. Genet. 155, Entian, K. D., Hilberg, F., Opitz, H. & Mecke, D. (1985) Mol. Cell. Biol. 5, Ma, H., Bloom, L. M., Zhu, Z., Walsh, C. T. & Botstein, D. (1989) Mol. Cell. Biol. 9, Ma, H., Bloom, L. M., Walsh, C. T. & Botstein, D. (1989) Mol. Cell. Biol. 9, Rose, M., Albig, W. & Entian, K. D. (1991) Eul: J. Biochem. 199, Maitra, P. K. (1971) J. Bacteriol. 107, Ciriacy, M. & Breitenbach, M. (1979) J. Bacteriol. 139, Boles, E. & Zimmermann, F. K. (1992) Curr Genet 21, in the press. 26. Sierkstra, L. N., Verbakel, J. M. A. & Vemps, C. T. (1992) J. Gen. Microbiol. 138, Zitomer, R. S., Montgomery, D. L., Nichols, D. L. & Hall, B. D. (1979) Proc. Natl Acud. Sci. USA 76, Zitomer, R. S. & Nichols, B. D. (1978) J. Bacteriol. 135, Lombardo, A,, Cereghino, G. P. & Scheffler,. E. (1992) Mol. Cell. Biol. 12,