INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Penicillin, a classical secondary metabolite found in some filamentous fungi, and cephalosporin, another widely used antibiotic, belong to the group of -lactam antibiotics, which are still of great importance for the treatment of infectious diseases (37). The hydrophilic cephalosporins are produced by some gram-positive and gram-negative bacteria as well as by filamentous fungi, e.g., Acremonium chrysogenum (Cephalosporium acremonium). The hydrophobic penicillins, on the other hand, are synthesized only by some fungi, most notably Penicillium chrysogenum and Aspergillus (Emericella) nidulans (reviewed in references 1, 6, 13, 34, and 37). The biosynthesis of penicillin and cephalosporin starts from the amino acid precursors L--aminoadipic acid (AA), L-cysteine, and L-valine. It is catalyzed by the enzymes -(L--aminoadipyl)-L-cysteinyl-D-valine synthetase, isopenicillin N synthase, and acyl coenzyme A:isopenicillin N acyltransferase, which are encoded by acvA (pcbAB), ipnA (pcbC), and aat (penDE), respectively. The genes have been cloned and sequenced. They are organized into a cluster (reviewed in references 1, 6, and 37) (Fig. 1A). acvA and ipnA are divergently transcribed (reviewed in references 1, 6, 34, and 37) and, in A. nidulans, are separated by an 872-bp intergenic region (27, 31, 35) (Fig. 1A).

View larger version (8K): [in this window] [in a new window]   FIG. 1.   (A) Penicillin biosynthesis gene cluster of A. nidulans (reviewed in references 1 and 6). (B) acvA-uidA and ipnA-lacZ gene fusions carrying the full-length intergenic region between acvA and ipnA (strain FLIRT), which was integrated in a single copy at the chromosomal argB gene locus (39). Roman numerals indicate the chromosomes on which the genes are located.

Since the biochemistry of penicillin biosynthesis is well understood, it currently represents the most advanced model system for studying the regulation of the biosynthesis of a secondary metabolite. In particular, penicillin biosynthesis in A. nidulans has been studied because this fungus has a well-defined sexual cycle, facilitating genetic analyses (3, 10, 11, 43). In contrast, the deuteromycete P. chrysogenum, which is used for the industrial production of penicillin, does not.

Since penicillin is synthesized from amino acid precursors, we believed it conceivable that amino acids play an important role in the regulation of its biosynthesis. Evidence for this hypothesis came from the observation that in both A. nidulans and P. chrysogenum, the addition of L-lysine to fermentation medium led to reduced penicillin titers (8, 12). Since AA is a branch-point metabolite between the lysine and penicillin biosynthetic pathways, lysine inhibition was suggested to operate at one or more steps of the lysine pathway (28). Consistent with this assumption was the notion that in P. chrysogenum, one target for lysine regulation is homocitrate synthase, which catalyzes the initial reaction of lysine biosynthesis. Lysine was found to cause both feedback inhibition of homocitrate synthase activity and repression of its synthesis (14, 26). In agreement with these findings, it was shown that the AA pool available for penicillin production was reduced in lysine-grown mycelia (23, 38).

In A. nidulans, lysine led to reduced expression of both acvA and ipnA gene fusions (8), indicating an additional and more direct effect on the expression of penicillin biosynthesis genes. Furthermore, in P. chrysogenum, the addition of DL-methionine to the medium (final concentration, 20 mM) led to a three- to fourfold increase in the production of cephalosporin C (16, 44). The increased production of this -lactam compound was paralleled by increased steady-state levels of mRNAs of cephalosporin biosynthesis genes pcbAB (acvA), pcbC (ipnA), cefEF and, to a slight extent, cefG (44). This finding suggests that the methionine effect was mediated via the expression of biosynthesis genes. In addition, Lara et al. (25) reported an inductive effect of L-glutamate on penicillin biosynthesis in P. chrysogenum.

Because of these limited data, we wished to study comprehensively whether the presence of different amino acids in the medium affected A. nidulans penicillin biosynthesis and, in particular, the expression of penicillin biosynthesis genes acvA and ipnA. Furthermore, a molecular mechanism which accounts for the repressing effects of some amino acids is elucidated.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

A. nidulans strains. The fungal strains used in this study are listed in Table 1.

Media and fermentation conditions. Fermentations were carried out with Aspergillus minimal medium (AMM) or fermentation medium (FM) essentially as described previously (7, 39). Seed cultures (20 ml of AMM or FM in 250-ml flasks) contained 4% (wt/vol) glucose as the carbon source. These cultures were inoculated with 1 ml of conidial suspension and incubated for 24 h at 26°C. Experimental cultures (20 ml of AMM or FM in 250-ml flasks) contained 4% (wt/vol) lactose as the carbon source and were supplemented with or without one of the different L-amino acids. The final amino acid concentrations in AMM and FM cultures were 10 and 50 mM, respectively. Experimental cultures were inoculated with 1 ml of the seed culture suspension. They were incubated for 48 h at 26°C because the measured effects on gene expression were most pronounced after this incubation time. The effects of exogenously added L-amino acids on the expression of acvA-uidA and ipnA-lacZ gene fusions were measured with the well-defined AMM. For this purpose, mycelia were harvested by filtration through Mira cloth (Calbiochem). Crude protein extracts were obtained by grinding the mycelia to a fine powder in liquid nitrogen. For preparation of cell extracts, a method previously described in detail was followed (7). In AMM, fungi hardly produce any penicillin, so penicillin production was determined with FM (see below). When required, biotin (0.3 µg/ml) or p-aminobenzoic acid (15 µg/ml) was added to the flasks.

-GLU and -GAL activity assays. The expression of acvA-uidA and ipnA-lacZ gene fusions was monitored by determining -glucuronidase (-GLU) and -galactosidase (-GAL) activities, respectively. They were measured in mycelial extracts from cells grown in AMM with or without L-amino acids (final concentration, 10 mM) essentially as previously described (7). For detection of -GAL activity in a plate assay, 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal; 50 µg/ml) was added to the AMM agar.

Determination of protein concentrations. Protein concentrations were determined according to the method of Bradford (5).

Activity assays for acidic and alkaline phosphatases. For determination of the activities of acidic and alkaline phosphatases, a plate assay was applied essentially according to the method of Dorn (15).

Penicillin bioassay. Centrifuged fermentation supernatant fluid (100 µl) from experimental cultures grown for 48 h in FM with or without amino acids (final concentration, 50 mM) was analyzed in a penicillin bioassay. This assay, in which Bacillus calidolactis C953 was used as an indicator organism, was carried out as described previously (7).

Genetic techniques. Sexual crosses and characterization of the resulting progeny were performed according to the methods of Pontecorvo et al. (33). For construction of A. nidulans FLIRT-2-3, strains G191 and FLIRT (Table 1) were crossed. Ascospores were plated on AMM agar plates with glucose as the carbon source and supplemented with p-aminobenzoic acid and X-Gal. One of the progeny was designated FLIRT-2-3. It was auxotrophic for p-aminobenzoic acid (pabaA1) and stained the agar blue on AMM-X-Gal agar plates with glucose as the carbon source, indicating an ipnA-lacZ gene fusion.

Reproducibility. Data represent the means of results from single experiments. A single experiment consisted of two different flasks containing the same strain and incubated under the same conditions. The standard deviations (SD) calculated for -GLU and -GAL specific activities from these flasks were less than 10%. SD calculated for the penicillin titer were higher (less than 15%) because of the use of FM, which results in greater variability of values. AMM cannot be used to quantify the penicillin titer because hardly any penicillin is detectable in this medium. All experiments were repeated at least three times. Their results were essentially identical. The SD calculated from the means of results of these independent experiments were less than 15%.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of adding various amino acids to the medium on the expression of penicillin biosynthesis genes acvA and ipnA and on penicillin production. In order to comprehensively test the effects of most of the proteinogenic amino acids on the expression of penicillin biosynthesis genes, A. nidulans FLIRT was used. This strain contains the full-length intergenic region of 872 bp between A. nidulans genes acvA and ipnA fused in frame to Escherichia coli reporter genes uidA and lacZ, resulting in the generation of acvA-uidA and ipnA-lacZ gene fusions (Fig. 1B). These gene fusions were integrated in a single copy at the chromosomal argB gene locus, allowing the precise determination of the expression of both genes simultaneously within one strain (7, 39) (Fig. 1B). Because strain FLIRT carries the complete intergenic region fused to reporter genes, this strain was designated wild type with respect to the expression of the acvA and ipnA gene fusions. Strain FLIRT was grown in AMM with lactose as the carbon source (see Materials and Methods). The expression of the acvA-uidA and ipnA-lacZ gene fusions was monitored by determination of -GLU and -GAL specific activities, respectively. L-Amino acids were added to the medium at a final concentration of 10 mM. The results are shown in Fig. 2.

View larger version (50K): [in this window] [in a new window]   FIG. 2.   Effects of various L-amino acids (10 mM) on the expression of the acvA-uidA and ipnA-lacZ gene fusions in strain FLIRT in minimal medium (AMM). Data obtained without the addition of amino acids (0) were set at 100. The relative values given represent the means of data obtained in single experiments. SD were less than 10%. See the text for details.

View larger version (38K): [in this window] [in a new window]   FIG. 3.   Penicillin titers after the addition of amino acids to the medium. Cultures of A. nidulans FLIRT in FM were supplemented with 50 mM histidine (H), valine (V), lysine (K), or methionine (M). Control cultures did not contain amino acids (0). Data represent the means of results of a single experiment. SD were less than 15%.

Deletion of specific DNA fragments of the intergenic region between acvA-uidA and ipnA-lacZ reduced the repressing effect of histidine and valine but not of lysine and methionine on the expression of both gene fusions. To analyze the effects of amino acids at the molecular level, we used A. nidulans strains which carried deletions in the intergenic region between the acvA-uidA and ipnA-lacZ gene fusions (Fig. 4). Like the wild-type strain FLIRT, these strains contained various deletion constructs integrated in a single copy at the chromosomal argB gene locus (39). The strains were cultivated in AMM with lactose as the carbon source. L-Amino acids were added to the medium at a final concentration of 10 mM. The results are shown in Fig. 4.

View larger version (21K): [in this window] [in a new window]   FIG. 4.   Determination of cis-acting DNA elements involved in the effects of exogenously added L-amino acids on the expression of acvA-uidA and ipnA-lacZ. Cultures were grown in AMM with a 10 mM concentration of the indicated L-amino acid. The expression of acvA-uidA and ipnA-lacZ gene fusions was determined as -GLU and -GAL specific activities, respectively. Data obtained without the addition of L-amino acids for each strain were set at 100. The relative values given for the deletion clones represent the means of results of three complete experiments. SD were less than 15%. Deletions are indicated by broken lines. The four PACC binding sites bound in vitro by PACC (42) were designated 1 (ipnA1), 2 (ipnA2), 3 (ipnA3), and 4AB (ipnA4AB) (17). Sites that were found to be functional in vivo for the expression of ipnA-lacZ (17) and/or for acvA-uidA expression are marked by hatched boxes. Site 3, which seems to be of major importance for both ipnA-lacZ expression (17) and acvA-uidA expression, is marked by a filled box. WT, wild-type strain FLIRT; nd, not determined.

View larger version (45K): [in this window] [in a new window]   FIG. 5.   Effects of L-amino acids on acvA-uidA and ipnA-lacZ expression and extracellular pH in both mutant strain PacC5-5 and the wild type. A. nidulans FLIRT (wild type for pacC) (WT) and PacC5-5 were cultivated for 48 h with or without (0) different L-amino acids at a concentration of 10 mM in AMM. The expression of acvA-uidA and ipnA-lacZ gene fusions was determined as -GLU and -GAL specific activities, respectively. The values given represent the means of data obtained with single experiments. SD are shown. The experiments were repeated three times. Their results were essentially identical. SD of values calculated from the means of these independent experiments were less than 15%.

A pacC5 mutation led to a higher level of acvA-uidA expression and abolished its repression by histidine and valine but not by lysine and methionine. To analyze the possible involvement of PACC in the effects of amino acids on acvA expression and, in general, whether PACC regulated acvA expression, a pacC5 mutant strain (PacC5-5) which carries a constitutively active PACC protein (42) was analyzed. For this purpose, we generated by sexual crossing a strain which contained acvA-uidA and ipnA-lacZ gene fusions integrated in a single copy at the chromosomal argB gene locus and the pacC5 mutation. The presence of the pacC5 mutation was tested by the loss of acidic phosphatase activity as described by Dorn (15). Accordingly, several progeny from a cross of strains FLIRT-2-3 (acvA-uidA and ipnA-lacZ gene fusions) and G0156 (pacC5) had no acidic phosphatase activity, had a higher level of ipnA-lacZ expression, and produced about twice the amount of penicillin as the wild type (Table 2, shown for strain PacC5-5). These results agreed well with previous results which had shown that pacC5 mutant strains produced larger amounts of penicillin and had higher steady-state levels of both ipnA transcripts and ipnA-lacZ expression (19, 36). Interestingly, the expression of acvA-uidA gene fusions was increased as well (Table 2), suggesting that PACC also regulates the expression of acvA. One of the resulting strains, PacC5-5, was used to analyze the effects of amino acids on gene fusions in a pacC5 background. As shown in Fig. 5, in contrast to those in cultures of wild-type strain FLIRT, pH values in the culture broth of mutant strain PacC5-5 were all similar irrespective of amino acid additions. For acvA-uidA expression, the inhibitory effects of histidine and valine were not detectable in the pacC5 mutant strain (PacC5-5), whereas the repressing effects of lysine and methionine were enhanced. These results were essentially the same as those obtained for deletion clone 183-312 (Fig. 4), which had one of the PACC binding sites (ipnA3) deleted. Taken together, these data suggest that at least the effects of histidine and valine on acvA-uidA expression are due to a decrease in the ambient pH as the result of the metabolism of these amino acids.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our data suggested that the acvA gene of A. nidulans, like ipnA (19, 30, 42), is regulated by the pH-dependent transcriptional factor PACC. In addition, we demonstrated that most amino acids, when added to the medium, affect the expression of gene fusions consisting of the penicillin biosynthesis genes acvA and/or ipnA fused to reporter genes. The repressing effects of histidine and valine on acvA-uidA expression are most likely due to the action of PACC and therefore to the pH of the culture broth.

Four lines of evidence support these conclusions. (i) The expression of acvA-uidA gene fusions was significantly increased in pacC5 mutant strains. This finding is consistent with the observation that the steady-state levels of acvA mRNA were increased at an alkaline ambient pH, a characteristic of genes regulated by PACC (18). (ii) The growth of A. nidulans with histidine and valine led to a lower ambient pH. At this pH, most or at least some of the PACC molecules are inactive (30, 42). At an alkaline ambient pH, however, PACC acts as an activator for the expression of both ipnA (19) and acvA (18; this study). Therefore, histidine and valine led to the acidification of the medium, caused the accumulation of the inactive form of PACC, and resulted in reduced acvA-uidA expression. (iii) In the deletion strain 183-312, the addition of histidine and valine caused a lower pH but did not lead to a reduction in acvA-uidA expression. In this deletion strain, one of the four PACC binding sites identified in vitro, ipnA3, spanning nt 265 to 270 (17, 42) (Fig. 4), was missing. Taken together, these findings suggest that in this deletion strain, the expression of the acvA-uidA gene fusion could not be further repressed by acidification because the absence of PACC binding site ipnA3 made it almost PACC independent. These results also imply that PACC binding site ipnA3 is the major site for PACC regulatory activity of acvA-uidA expression. PACC binding sites at nt 149 to 154 (ipnA1) and 174 to 179 (ipnA2) contributed less, and those at nt 509 to 514 (ipnA4A) and 524 to 529 (ipnA4B) apparently did not contribute to the activation of acvA-uidA expression by PACC (Fig. 4). These findings agreed well with those of Espeso and Peñalva (17), who had precisely analyzed the functionality of the PACC binding sites identified in vitro for ipnA-lacZ expression in vivo. The authors found that site ipnA3 was most important for PACC-dependent ipnA expression, whereas sites ipnA2 and ipnA4 were less important, although site ipnA2 was bound with the highest affinity by a glutathione S-transferase-PACC fusion protein in vitro. Binding site ipnA1 was apparently not required for PACC-dependent ipnA expression (17) (Fig. 4). Hence, binding site ipnA3 seems to be of major importance for PACC-dependent expression of both gene acvA and gene ipnA. When this site was lacking (strain 183-312), the repressing effects of histidine and valine were no longer detectable. (iv) In a pacC5 mutant strain (PacC5-5) with a constitutively active PACC protein, no repression by histidine or valine could be detected. This result demonstrates the importance of PACC for repression by these two amino acids.

The pH of the culture supernatant of the pacC5 mutant strain (PacC5-5) was higher than that of the wild-type strain. Therefore, it was conceivable that due to the increased ambient pH, the mutant had a reduced uptake of some amino acids, thus avoiding repressing intracellular concentrations of histidine, valine, or derivatives. However, this situation was unlikely, because in strains with the wild-type pacC gene and with gene fusions having deletions of certain PACC binding sites, there was hardly any detectable repression either. The only difference between these deletion strains and wild-type strain FLIRT was the deletion of nucleotides in the intergenic region between the acvA-uidA and ipnA-lacZ gene fusions. This fact supports the assumption that the repressing effects of histidine and valine on acvA-uidA expression were mediated, at least in part, by an ambient pH via the transcriptional factor PACC. This fact is also consistent with the result that exogenously added threonine led to a slightly increased ambient pH and to increased acvA-uidA expression in the wild type, suggesting that PACC activity was increased under these conditions.

The addition of lysine and methionine also resulted in a decrease in both the ambient pH and the expression of the acvA-uidA and ipnA-lacZ gene fusions. Although this pattern was similar to that observed for histidine and valine, the effects of methionine and lysine seemed to be mediated via different mechanisms, although some contribution of PACC could not be entirely excluded. This conclusion was drawn from the observation that, in contrast to those of histidine and valine, the repressing effects of lysine and methionine on acvA-uidA expression were even increased in deletion clone 183-312, lacking PACC binding site ipnA3 (Fig. 4). Furthermore, in strain PacC5-5, the constitutively active PACC5 protein could not overcome the repressing effects of lysine and methionine (Fig. 5), making the involvement of PACC unlikely. So far, the regulatory mechanisms that mediate the repressing effects of these amino acids have not been elucidated. For methionine, the analysis of deletion clones indicated that a major cis-acting DNA element responsible for its effect on acvA-uidA expression is located between nt 433 and 569. For lysine, no DNA region which could be involved has yet been identified. The effects of lysine and methionine in the medium are mediated by regulatory mechanisms which appear to differ even between these two amino acids. It is interesting to note that lysine and methionine are closely related to the precursor amino acids AA and cysteine, respectively.

Velasco et al. (44) noted several consensus CANNTG sequences in the intergenic region of A. chrysogenum cephalosporin biosynthesis genes pcbAB (acvA) and pcbC (ipnA), whose transcription was increased after the addition of methionine. Members of the basic region-helix-loop-helix protein family recognize such a consensus motif. Some of these proteins are involved in the transcriptional control of the sulfur network in Saccharomyces cerevisiae (4, 9, 29, 40, 41). Therefore, Velasco et al. (44) suggested that a basic region-helix-loop-helix protein might mediate the effect of methionine on the cephalosporin biosynthesis genes. For penicillin biosynthesis, the involvement of a similar mechanism has not been clarified yet. However, there is no consensus CANNTG motif in the region spanned by nt 433 to 569 of the intergenic region between acvA and ipnA, which was found to be important for the methionine effect (Fig. 4). Therefore, an understanding of the effects of exogenously added methionine at the molecular level awaits further studies.

The analysis of deletion clones implied the involvement of PACC in the regulation of ipnA-lacZ expression by amino acids. This implication was contradicted, however, by results obtained with strain PacC5-5, in which, in contrast to acvA-uidA expression, ipnA-lacZ expression was still repressed by histidine and valine. It is conceivable that the different effects on the two gene fusions in strain PacC5-5 were due to the pacC5 mutation. Small changes in the ambient pH enable the mutant as well as the wild-type PACC protein to regulate the expression of acvA. This finding may be due to the organization of the bidirectionally oriented promoter region between acvA and ipnA. For the expression of ipnA, the pH signal may need to be more pronounced. It is tempting to speculate that acvA expression is subject to more sensitive regulation than ipnA expression. This hypothesis is supported by the observation in this study that acvA was regulated to a greater extent by amino acids than was ipnA. This idea is also consistent with the finding that the expression of acvA is rate limiting for penicillin production (24) and thus represents a critical step in penicillin biosynthesis.

It was also conceivable that a GCN4-like factor (reviewed in reference 21) was involved in the effects of exogenously added amino acids. In A. nidulans, the expression of the argB gene, encoding ornithine transcarbamoylase, was increased when cells were starved for certain amino acids (20, 32), indicating that general amino acid control is present in this fungus. This result was further supported by the findings of Wanke et al. (45), who recently cloned a GCN4 homolog of A. niger. Some of the amino acid effects reported here could thus be due to the action of a GCN4 analog in A. nidulans. Thus, the addition of excess leucine could lead to reduced amounts of isoleucine and valine by feedback inhibition, causing activation of the GCN4 system (reviewed in reference 21). Furthermore, exogenously added aminotriazole, which causes histidine starvation and thus presumably induces GCN4-analogous regulators, was found to increase penicillin production in certain P. chrysogenum strains (22). It was suggested that this increase was due to the induction of lysine biosynthesis by histidine starvation, leading to increased AA levels and thus to increased penicillin titers. However, no correlation between the increased expression of penicillin biosynthesis genes and ornithine transcarbamoylase specific activity, which was measured as an indicator of the activity of a GCN4-analogous regulator, was detected for any of the amino acids analyzed here (data not shown). This finding suggests that for the effects described here, the participation of a GCN4-like factor is unlikely.

In summary, some of the effects of exogenously added amino acids on the biosynthesis of penicillin are mediated via the pH-dependent transcriptional regulator PACC. This finding may have implications for other biosynthetic pathways which are regulated by exogenously added amino acids.