Gene clusters for b-lactam antibiotics and control of their expression: why have clusters evolved, and from where did they originate?

Paloma Liras; Juan F. Martín^*

Biotechnological Institute, Scientific Park of Leon, Spain, and Microbiology Area, Faculty of Biological and Environmental Sciences, University of Leon, Spain

Corresponding author

While b-lactam compounds were discovered in filamentous fungi, actinomycetes and gram-negative bacteria are also known to produce different types of b-lactams. All b-lactam compounds contain a four-membered b-lactam ring. The structure of their second ring allows these compounds to be classified into penicillins, cephalosporins, clavams, carbapenens or monobactams. Most b-lactams inhibits bacterial cell wall biosynthesis but others behave as b-lactamase inhibitors (e.g., clavulanic acid) and even as antifungal agents (e.g., some clavams). Due to the nature of the second ring in b-lactam molecules, the precursors and biosynthetic pathways of clavams, carbapenems and monobactams differ from those of penicillins and cephalosporins. These last two groups, including cephamycins and cephabacins, are formed from three precursor amino acids that are linked into the a-aminoadipyl-L-cysteinyl-D-valine tripeptide. The first two steps of their biosynthetic pathways are common. The intermediates of these pathways, the characteristics of the enzymes involved, the lack of introns in the genes and bioinformatic analysis suggest that all of them should have evolved from an ancestral gene cluster of bacterial origin, which was surely transferred horizontally in the soil from producer to non-producer microorganisms. The receptor strains acquired fragments of the original bacterial cluster and occasionally inserted new genes into the clusters, which once modified, acquired new functions and gave rise to the final compounds that we know. When the order of genes in the Streptomyces genome is analyzed, the antibiotic gene clusters are highlighted as gene islands in the genome. Nonetheless, the assemblage of the ancestral b-lactam gene cluster remains a matter of speculation. [Int Microbiol 2006; 9(1):9-19].

Key words: b-lactam antibiotics · antibiotic biosynthesis · metabolic regulation · gene clusters · microbial evolution.

RESUMEN

Complejos génicos ("clusters") de antibióticos b-lactámicos y control de su expresión: ¿por qué se han originado, y de dónde proceden?

Las β-lactamas fueron descubiertas en hongos filamentosos, pero se sabe que los actinomicetos y algunas bacterias gram negativas también producen diferentes tipos de β-lactamas. Todas las b-lactamas contienen un anillo de cuatro miembros. La estructura del segundo anillo permite clasificarlas en penicilinas, cefalosporinas, clavamas, carbapenemas o monobactamas. La mayoría de estos compuestos inhiben la síntesis de la pared celular bacteriana, pero algunos se comportan como inhibidores de β-lactamasas (por ejemplo, el ácido clavulánico), o incluso como agentes antifúngicos (algunas clavamas). Debido a la naturaleza del segundo anillo en la molécula de las β-lactamas, los precursores y las vías biosintéticas de clavamas, carbapenemas y monobactamas son diferentes de los de penicilinas y cefalosporinas. Las moléculas de estos dos grupos, incluyendo cefamicinas y cefabacinas, están formadas por tres aminoácidos precursores que se unen para formar el tripéptido α-aminoadipil-L-cisteinil-D-valina. Las primeras dos etapas de la biosíntesis de las cefamicinas y las cefabacinas son comunes. Los intermediarios de estas vías, las características de las enzimas que intervienen en ellas, la falta de intrones en los genes y el análisis bioinformático sugieren que se originaron a partir de un complejo génico ("cluster") ancestral de origen bacteriano, que fue transferido horizontalmente en el suelo desde los microorganismos productores a los no productores. Las cepas receptoras adquirieron fragmentos del complejo génico bacteriano original y ocasionalmente insertaron en él nuevos genes; dichos genes, una vez modificados, adquirieron funciones nuevas y dieron lugar a los nuevos compuestos finales que conocemos. Cuando se analiza el orden de los genes en el genoma de Streptomyces, los complejos génicos de la síntesis de antibióticos destacan como islas en el genoma. Sin embargo, la forma en que se ensambló el complejo génico ancestral para β-lactamas sigue siendo motivo de conjeturas. [Int Microbiol 2006; 9(1):9-19].

Palabras clave: antibióticos b-lactámicos · biosíntesis de antibióticos · regulación metabólica · complejos génicos ("clusters") · evolución microbiana.

RESUMO

Complexos génicos ("clusters") de antibióticos b-lactámicos e controle de sua expressão: por que se originaram, e de onde procedem?

As b-lactamas foram descobertas em fungos filamentosos, mas sabe-se que os actinomicetos e algumas bactérias gram-negativas também produzem diferentes tipos de b-lactamas. Todas as b-lactamas possuem um anel de quatro membros. A estrutura do segundo anel permite classificar estes compostos em penicilinas, cefalosporinas, clavamas, carbapenemas ou monobactamas. A maioria das b-lactamas inibem a síntese da parede celular bacteriana, mas algumas se comportam como inibidoras de b-lactamases (por exemplo, o ácido clavulânico) ou inclusive como agentes fungicidas (algumas clavamas). Devido à natureza do segundo anel da molécula das b-lactamas, os precursores e as vias biosintéticas de clavamas, carbapenemas e monobactamas são diferentes das de penicilinas e cefalosporinas. As moléculas nestes dois grupos, incluindo cefamicinas e cefabacinas, são formadas por três aminoácidos precursores que se unem para formar o tripéptido a-aminoadipil-L-cisteinil-D-valina. As primeiras duas fases da biossíntese dos cefamicinas e cefabacinas são comuns. Os intermediários destas vias, as características das enzimas que intervêm nelas, a falta de íntrons nos genes e a análise bioinformática sugerem que todas elas se originaron a partir de um complexo génico ("cluster") ancestral de origem bacteriano, que foi transferido horizontalmente no solo dos microorganismos produtores aos não produtores. As cepas receptoras adquiriram fragmentos do complexo génico bacteriano original e ocasionalmente inseriram nele novos genes, os quais, uma vez modificados, adquiriram funções novas e deram lugar aos novos compostos finais que conhecemos. Quando se analisa a ordem dos genes nos genomas de Streptomyces, os complexos génicos para sínteses de antibióticos destacam-se como ilhas no genoma. No entanto, a forma em que se encaixou o primeiro complexo génico ancestral para b-lactamas continua sendo motivo de conjeturas. [Int Microbiol 2006; 9(1):9-19].

Classical and novel b-lactam families

The b-lactam antibiotics, like many other secondary metabolites, have highly unusual chemical structures, very different from those of classical primary metabolites [31]. Understanding the biosynthesis and the molecular genetics of these compounds has been a challenge for microbiologists over the last few decades, and only a few of the impressive array of secondary metabolites is known at the molecular level. A comprehensive study of these compounds will surely require many more decades of research.

The structure of the b-lactam antibiotic penicillin consists of a bicyclic penam nucleus formed by a b-lactam ring and a thiazolidine ring containing a sulfur atom and an acyl side-chain bound to the amino group present at C-6. The two rings are produced by Penicillium and Aspergillus species, as well as a few other ascomycetes [24]. A second compound with b-lactam structure, produced by Acremonium chrysogenum, was discovered in 1955 [38] and chemically characterized as cephalosporin C, a molecule in which the five-membered thiazolidine ring of penicillin is replaced by a six membered dihydrothiazine ring forming the cephem nucleus (Fig. 1). Cephalosporin C has an a-aminoadipyl side-chain attached at the C-7 amino group, which is identical to that of penicillin N but different from those of other penicillins. Both b-lactam compounds, penicillin and cephalosporin C, are of great clinical interest as inhibitors of peptidoglycan biosynthesis in bacteria.

Over the last three decades, a wide array of compounds belonging to the same family of molecules has been discovered using new targets and screening techniques. Some of these compounds are modified cephalosporins, such as the cephamycin family, in which the cephem nucleus contains, in addition to the a-aminoadipyl side-chain, a methoxyl group at C-7. This methoxyl group renders the cephamycin structure insensitive to hydrolysis by most b-lactamases. Cephamycins are produced by different actinomycetes species. In other b-lactams, as in the family of cephabacins, produced by gram-negative bacteria, a formyl group is frequently present at C-7, and different peptides are attached to the C-3 carbon of the dihydrothiazine ring in the cephem nucleus. All the above-mentioned groups, produced either by filamentous fungi or by bacteria, have a common mode of action, similar precursors, and partially overlapping biosynthetic pathways [1]. In addition to these classical b-lactam compounds, many non-conventional b-lactam structures have been discovered and characterized since 1970. Some of them, including carbapenems and nocardicins, also inhibit peptidoglycan biosynthesis. Others, such as clavulanic acid, are weak antibiotics but potent b-lactamase inhibitors, or have antifungal activity (i.e., some clavams). These non-conventional b- lactams contain a b-lactam ring and they usually have a distinct bicyclic structure. The second ring in the molecule of clavulanic acid and other clavams is an oxazolidinic ring that includes oxygen instead of a sulfur atom, as occurs in classical b-lactams (Fig. 1). The members of carbapenem and the olivanic acid family have a carbapenem ring containing a carbon atom instead of sulfur. Thienamycin is the model structure of this family. Finally, there are many compounds with monocyclic structure that contain only the b-lactam ring and different side-chains (they are known as monobactams). Some monobactams, as is the case of the nocardicins, are produced by actinomycetes [2], but others, such as sulfacezin, are produced by proteobacteria [19].

Non-conventional b-lactam antibiotics are molecules of scientific and industrial interest, and their biochemistry and molecular genetics are being actively investigated by several groups [10,30,36,39,45,49]. The precursors and biosynthetic pathways of non-conventional b-lactam antibiotics are different from those of classical b-lactams. The enzymes, genes, and gene organization are also different [17]. For example, the specific enzyme involved in the formation of the b-lactam ring in non-conventional b-lactams is not a dioxygenase, as occurs in the classical b-lactam antibiotics, but more closely resembles asparaginases or asparagine synthetases [3], suggesting that those compounds should have a different evolutionary origin. Up to now, non-conventional b-lactam antibiotics have never been found to be produced by eukaryotic cells. The genetic information encoding enzymes for the biosynthesis of the different b-lactam families is restricted to a few taxonomic groups [1]. This raises the question of the evolutionary origin of these genes (see below).

Penicillin, cephalosporin and cephamycin biosynthesis pathways

A brief description of the biochemical pathways leading to classical b-lactam antibiotic biosynthesis is provided here. Several previous reviews have described the specific steps in greater detail [27,29,32]. The formation of b-lactam compounds proceeds through "early biosynthetic steps", "intermediate steps", "late ('decorating') steps," and reactions for the formation of the specific precursors. Depending on the compound formed in the pathway, more or fewer steps will be required-penicillin biosynthesis being the shortest pathway, and cephamycin and cephabacins pathways being the longest ones.

Three aminoacids, L-a-aminoadipic acid, L-cysteine, and L-valine are always the precursors of the basic structure of all classical b-lactam antibiotics (Fig. 2). Of these three precursor amino acids, L-valine and L-cysteine are common, whereas L-a-aminoadipic acid is a non-proteinogenic amino acid that must be synthesized by a specific pathway. In fungi, a-aminoadipic acid is an intermediate of the lysine biosynthesis pathway [4]. In addition, lysine is catabolized to a-aminoadipic acid in Penicillium chrysogenum by: (i) an w-aminotransferase, encoded by the oat1 gene, which is induced by lysine, and (ii) by a reversal of the lysine biosynthesis pathway catalyzed by the enzymes saccharopine dehydrogenase/saccharopine reductase [13,35]. Gene oat1 is present in neither penicillin nor cephalosporin gene clusters [37], and only one gene encoding a saccharopine dehydrogenase, potentially involved in the a-aminoadipic acid biosynthesis, has been located in the amplified region containing the penicillin biosynthesis cluster in P. chrysogenum.

In b-lactam-producing actinomycetes, lysine is converted into a-aminoadipic acid semialdehyde by lysine-6-aminotransferase (LAT) [5,54]. This enzyme and its encoding gene (lat) are only found in b-lactam-producing microorganisms, and the presence of this enzyme activity or positive hybridization with the lat gene are good indicators for detecting novel b-lactam producers. The LAT-reaction product, a-aminoadipic semialdehyde, cyclizes spontaneously to form piperideine-6-carboxylate (PCD) [16] and later is oxidized to a-aminoadipic acid by a piperideine-6-carboxylate dehydrogenase, encoded by the pcd gene [43]. Both lat and pcd are located in the cephamycin gene cluster in Streptomyces clavuligerus.

Early steps in the formation of b-lactam antibiotics

Two enzymatic steps are common to all b-lactam producers and result in the formation of isopenicillin N, the first compound in the pathway with antibiotic activity. The first enzyme of the pathway is d-(L-a-aminoadipyl)-L-cysteinyl-D-valine (ACV) synthetase (ACVS), a non-ribosomal peptide synthetase [31]. ACV synthetases are very large multifunctional proteins (about 460 kDa) encoded by 11-kb intron-free genes named pcbAB. This enzyme uses ATP to sequentially activate the three amino-acid substrates to form aminoacyl-adenylates, then binds them to the enzyme as thioesthers, epimerizes the L-valine to the D-valine configuration, and finally links the three amino acids to form the peptide L-d(a-aminoadipyl)-L-cysteinyl-D-valine. The peptide is released from the enzyme by the action of an internal thioesterase, another of the activities of the ACVS complex. ACVSs have three well-conserved domains, specific to activate each amino acid [31]. Several research groups are attempting to construct hybrid pcbAB genes containing genetic domains from other peptide synthetases in order to obtain modified b-lactam antibiotics.

The second enzyme in the pathway is isopenicillin N (IPN) synthase (also named ACV cyclase), encoded by the pcbC gene. IPN synthases are intermolecular dioxygenases that require Fe²⁺, molecular oxygen, and ascorbate for their activities. These enzymes remove four hydrogens from the ACV tripeptide, forming directly the bicyclic structure of isopenicillin N. The cyclase of P. chrysogenum has been crystallized [46], and knowledge of the enzyme's protein structure has been an important step in obtaining basic information to produce, by mutagenesis, modified cyclases able to recognize other peptides. The process of direct formation of the byciclic structure of isopenicillin N differs from that of the other non-conventional b-lactam antibiotics, which first form the b-lactam ring and then, using a different enzyme, cyclize the monocyclic intermediate to form the second ring (oxazolidinic or carbapenem) present in the molecule, although it is probable that the two steps are coupled in vivo. Monobactam producers lack the ability to close the second ring of the nucleus.

In addition to the pcbAB and pcbC genes common to bacteria and filamentous fungi, penicillin producers (i.e., Penicillium, Aspergillus) contain a third gene in the penicillin gene cluster, penDE. This gene is of eukaryotic origin (it contains three introns) and encodes an isopenicillin N acyltransferase, an enzyme that hydrolyzes the a-aminoadipic side-chain of the isopenicillin N and introduces instead a chain from an acyl activated compound, most frequently phenylacetyl-CoA to produce penicillin G. This gene is not present in either cephalosporin-C or cephamycin-producing microorganisms.

Intermediate steps

Isopenicillin N is converted into the D-isomer, penicillin N, by all cephalosporin and cephamycin producers. This conversion is carried out in a single step in bacterial strains. The enzyme responsible for this epimerization was first purified from actinomycetes [26,56], and the gene encoding its activity, cefD, was found to be located in the cephamycin gene cluster [6,23]. The bacterial cefD-encoded protein is a pyridoxal phosphate-dependent enzyme of about 43 kDa. However, attempts to find a homologous cefD gene and a clear epimerase activity in A. chrysogenum failed for many years. Purification of the A. chrysogenum "epimerase" proved to be difficult, non-reproducible, and unreliable. In 2002, a significant breakthrough in our understanding of cephalosporin formation occurred when we found that the epimerization reaction was different in eukaryotic and prokaryotic microorganisms (Fig. 3). The epimerization of isopenicillin N in A. chrysogenum is encoded by two linked genes, cefD1-cefD2, located in the early cephalosporin gene cluster. Transcriptional studies on A. chrysogenum revealed the presence of two transcripts in the region downstream of pcbC [55]. Sequencing of the region confirmed the presence of two open reading frames (ORF) separated by a bidirectional promoter region. The first, cefD1, has five introns and encodes a 71-kDa protein with homology to fatty acid acyl-CoA synthetases. The second, cefD2, has one intron and encodes a protein with high homology to a-methylCoA racemases of eukaryotic origin. Disruption of either of these ORFs results in lack of cephalosporin C production, loss of isopenicillin N epimerase activity, and the accumulation of isopenicillin N in the culture. The proposed epimerization in fungi includes three biochemical steps: (i) CefD1 converts isopenicillin N into isopenicillinyl N-CoA; (ii) CefD2 isomerizes the compound into penicillinyl N-CoA (Fig. 3); and (iii) penicillinyl N-CoA is probably released from the enzyme by a thioesterase. Indeed, ACVS contains a thioesterase domain, although it is unclear whether it works on the penicillinyl N-CoA. Another gene encoding a separate thioesterase has been located 1 kb downstream of cefG, in the late cephalosporin gene cluster, but it has not yet been characterized.

The next step in the cephalosporin/cephamycin pathway is expansion of the five-membered thiazolidine ring of penicillin N to a six-membered dihydrothiazine ring. The enzyme responsible for this important conversion is the deacetoxycephalosporin C (DAOC) synthase commonly known as expandase. This protein is an intermolecular dioxygenase very similar to ACV cyclase. It requires Fe²⁺, molecular oxygen, and a-ketoglutarate to form DAOC and succinic acid. The expandase does not recognize the isomer isopenicillin N, penicillin G, or deacylated 6-aminopenicillanic acid (6-APA) as substrates. Expandase from S. clavuligerus has been crystallized, and the gene cefE was introduced in Penicillium chrysogenum, leading to the biosynthesis of adipyl-7-aminodeacetoxycephalosporanic acid (adipyl-7-ADCA) and adipyl-7-ACA, compounds that can be transformed into the economically relevant 7-ADCA and 7-ACA intermediates (reviewed in [11]).

Late steps

The final step in cephalosporin C biosynthesis is the conversion of DAC to cephalosporin C by a DAC-acetyltransferase that uses acetyl-CoA as donor of the acetyl group. This 49-kDa enzyme, encoded by the cefG gene [18], is evolutionarily similar to O-acetyl homoserine acetyl transferases. The cefG gene contains two introns and is linked to cefEF, but in the opposite orientation. Bioinformatic analysis of cefG revealed 55% identical residues with the met2 genes of Aspergillus fumigatus and Ascobolus immersus. The met2 genes encode O-acetyl homoserine acetyl transferase activity involved in sulfur-containing amino-acid biosynthesis. However, mutations disrupting cefG do not lead to methionine auxotrophy, which suggests that there is a separate met2 in Acremonium chrysogenum. A detailed phylogenetic analysis of the CefG protein revealed that it is, indeed, a member of the serine/homoserine acetyl transferase class. However, it seems to be a specialized enzyme, different from that of primary metabolism, that has evolved to form part of the "late" cephalosporin gene cluster. The weak nature of the cefG promoter causes a bottleneck of this enzymatic step in the pathway. Replacing the cefG promoter by the glyceraldehyde-3-phosphate dehydrogenase promoter (gpd) of Aspergillus nidulans or the glutamate dehydrogenase (gdh) promoter of P. chrysogenum improved cephalosporin C production by two- to three-fold .

In cephamycin-producing actinomycetes, two sequential enzymatic reactions-an hydroxylation and the transfer of a methyl group to the hydroxyl present at C-7-are carried out by a CmcI-CmcJ complex. This complex co-purifies by immunoaffinity chromatography and binds S-adenosylmethionine and deacetylcephalosporin C, in contrast to the isolated CmcI and CmcJ purified proteins, which do not have binding activity [8,12]. In addition, the product of the cmcH gene carries out a carbamoyl transfer reaction [9]. The final steps of cephabacin biosynthesis have been deduced from the presence of several genes in the cephabacin gene cluster encoding non-ribosomal peptide synthetases. Their protein products are probably involved in the formation of the lateral chain in C-3 [21,50].

Gene clusters for b-lactam antibiotic biosynthesis: possible transmission by horizontal transfer

Genes for b-lactam biosynthesis are clustered in all producer strains, whether they are eukaryotic or prokaryotic microorganisms (Fig. 4). With minor differences between strains, the pcbC-pcbAB genes are always grouped and are located next to the penDE gene in penicillin producers. In A. chrysogenum, the "early" gene cluster, located in chromosome VII (4.6 Mb), contains the genes pcbAB and pcbC, encoding the enzymes for the first two steps of the pathway; cefD1 and cefD2, responsible for the epimerization of isopenicillin N; and cefT, which encodes a transmembrane protein needed for putative exportation of the antibiotic. The "late" gene cluster, located in chromosome I (2.2 Mb), contains the genes cefEF and cefG, whose protein products are involved in the final steps of cephalosporin biosynthesis. A similar gene organization, pcbAB-pcbC-cefD2, is found in Kallichroma tethys, a wood-inhabiting marine fungus phylogenetically related to A. chrysogenum [20,33].

The largest b-lactam clusters are those of cephamycin- and cephabacin-producing bacteria [28]. In S. clavuligerus, the cephamycin C gene cluster is adjacent to the clavulanic acid gene cluster. The entire supercluster of cephamycin C-clavulanic acid extends for about 50 kb. This organization of the biosynthetic genes of both antibiotics in a supercluster occurs also in other clavulanic-acid-producing strains, including Streptomyces jumonjinensis and Streptomyces katsurahamus.

Analysis of the distribution of the amino-acid sequence of the enzymes involved in b-lactam biosynthesis (ACV synthetase, IPN synthase, CefE, CefF) in bacteria reveals that their amino-acid identities range from 70 to 75%, while their identity to homologous proteins in b-lactam-producing eukaryotes ranges from 57 to 60%. These percentages of identity between prokaryotic and eukaryotic homologous proteins are higher than those of primary metabolism enzymes, including enzymes for amino-acid biosynthesis. This finding led to the proposal that either horizontal gene transfer took place in the soil about 370 million years ago [57] or multiple gene transfer events occurred from bacteria to fungi [1], conferring upon the latter the ability to produce b-lactam antibiotics to several genera of fungi. This theory is also supported by the lack of introns in pcbAB (despite its large size, 11.5 kb), pcbC, and cefEF, while specific genes absent in bacterial b-lactam clusters but present in penicillin and cephalosporin biosynthesis clusters in Penicillium and Acremonium, such as penDE, cefG, cefD1, and cefD2, contain introns.

The presence of genes for b-lactam antibiotics, and therefore the ability to produce these compounds, confer a major ecological advantage to soil fungi, which are thus able to protect nutrient sources that would otherwise be used by fast-growing bacteria. In prokaryotes, the advantage of producing b-lactams is less clear. Since most bacteria are sensitive to b-lactams, producer bacteria might therefore be killed by their own antibiotics; however, b-lactam-producing bacteria are somehow less sensitive to b-lactams. Frequently, b-lactam clusters include genes for b-lactamases and penicillin-binding proteins (PBP) [7,41]. Class A b-lactamase, present in the cephamycin cluster of S. clavuligerus, is a typical penicillinase, inactive against both cephalosporins and cephamycins. S. clavuligerus is a natural cephalosporin-resistant strain (up to 1 mg cephalosporin/ml), but it is sensitive to 1000-fold less penicillin G. Cephalosporin resistance correlates with the ability of secreting cephamycin or cephalosporin intermediates [34] by cephamycin producers. It seems that cephamycin C is of ecological importance to this strain in natural habitats in which it competes with cephamycin-sensitive bacteria. The b-lactamase present in the cephamycin cluster protects the producer strain against intermediate penicillins formed in the pathway and against penicillins released either by secretion (intermediate penicillins are indeed partially secreted) or by cell lysis, but it is not active against cephamycins. The cluster of cephabacin biosynthesis genes in Lysobacter lactamgenus also contains a class A b-lactamase, but in this case the enzyme resembles typical cephalosporinases. The substrate specificity of this "cephalosporinase-like" enzyme and its role in the biosynthesis of cephabacins therefore deserve to be studied in more detail.

In addition, the cluster of genes for b-lactam biosynthesis contains genes encoding proteins that appear to be involved in antibiotic secretion from cells. Such proteins are membrane proteins of the major facilitator family (MFs). A. chrysogenum contains a gene, cefT1, located downstream of pcbAB and in the opposite orientation, encoding a protein with 12-transmembrane domains (TMs) bearing the characteristic motifs of Drug:H+ antiporters of the 12-TMs class. Targeted inactivation of cefT1 does not affect significantly cephalosporin biosynthesis. This might be due to the presence of an additional cefT3 gene similar to cefT1 located downstream of cefD1. Amplification of cefT1 results in a 100% increase in cephalosporin C production, which supports the importance of these genes in antibiotic secretion or in the regulation of antibiotic biosynthesis [34]. Current information does not exclude the possibility that either CefT1 or homologous transmembrane proteins are involved in the transduction across the membrane of signal molecules controlling antibiotic biosynthesis rather than in the secretion of b-lactam antibiotics.

Similar cmcT genes are also present in the cephamycin C clusters of S. clavuligerus and Amycolatopsis lactamdurans. Both genes encode 14-TMs proteins of the MFs family that are 73% identical among themselves. Hybridization studies using probes internal to cmcT gave positive hybridization with total DNA from all the b-lactam producers tested (Streptomyces cattleya 8057, S. griseus 3851, S. lipmanii 3584) but not with the b-lactam non-producer S. lividans JI1326. Transmembrane proteins of the 14-TMs class are also present in the sequenced genomes of the actinomycetes Streptomyces coelicolor (SCO2309) and Streptomyces avermitilis (SAV5867), but they are only 48% identical to cmcT, which supports the conclusion that CmcT proteins of the cephamycin clusters are specific cephamycin exporters. A significant number of cmcT-related genes encoding proteins of the MFs family have been described in different antibiotic-producing actinomycetes [31]. The closest cmcT relative (55% amino-acid identity) known so far is the entT (Q9KHJ5) gene of Streptomyces maritimus present in the enterocin (a polyketide antibiotic) gene cluster.

Amplification of the penicillin gene cluster in penicillin-overproducing strains

Hybridization studies of total DNA from penicillin low-producing, medium-producing, and high-producing strains using probes for the pcbC, pcbAB or penDE genes showed that high-producing strains have more copies of the three genes. A 106.5-kb DNA region was found to be amplified in tandem repeats (five to six copies in the improved producing strain P. chrysogenum AS-P-78, and up to fourteen in the industrial P. chrysogenum E1 strain [14]. Amplification always occurs in chromosome I (Fig. 5). P. notatum ATCC 9478 (the strain isolated by Fleming), and the wild-type P. chrysogenum NRRL1951 contain a single copy of the 106.5-kb region. That region is bounded by a hexanucleotide sequence, TTTACA/T, which appears to be a site-specific sequence for recombination. Some high-producing strains, such as P. chrysogenum E1, contain more copies in tandem of a shorter 56.8-kb DNA region (internal to the 106.5-kb region). This 56.8-kb region carries the penicillin biosynthesis genes and is bounded either by the TTTACA/T sequence or its complementary TGTAAA [14]. It seems that this amplification arose through the mutagenesis process used in the screening programs by researchers in industry to obtain high-yield-producing strains. The loss of that 56.8-kb DNA fragment in strain P. chrysogenum npe10, obtained in our laboratory, and in several mutants isolated in industrial companies, results in penicillin-non-producing mutants [14]. This type of iterated DNA amplification has not been found in cephalosporin-producing A. chrysogenum, nor in the penicillin-producing Aspergillus nidulans.

Transcriptional analysis of the 56.8-kb amplified region revealed the presence of at least eight transcripts expressed in penicillin-producing conditions. The entire 56.8-kb region has been sequenced, and computer analysis showed a total of 16 ORFs, including the three structural genes involved in penicillin biosynthesis (F. Fierro and J.F. Martín, unpublished results). Studies to determine whether ORFs present in the amplified region are related to penicillin biosynthesis are underway. One of these genes encodes a saccharopine dehydrogenase, an enzyme involved in the conversion of a-aminoadipic acid into lysine, and might play a role in penicillin biosynthesis by contributing to the formation of a-aminoadipic acid from lysine by the reverse lysine pathway.

Production of penicillin by fungi growing on food products

Regulation of b-lactam biosynthesis

Transcriptional analysis of the penicillin gene cluster indicates that the pcbC and pcbAB genes are expressed from a bidirectional promoter region of 1013 bp (see Fig. 4). The two genes show a similar pattern of temporal expression and regulation, which suggests their coordinated expression. The third penicillin gene (penDE) carries its own promoter. These promoter regions are under the control of a variety of regulatory mechanisms. For example, in P. chrysogenum, alkaline pH results in positive regulation of the three promoters. Multiple regulatory sequences have been described in the pcbAB-pcbC bidirectional promoter region [22]; seven Pac sequences recognized by the pH-dependent transcriptional regulator PacC [53]; six CreA sites for binding of the general carbon catabolite regulatory protein CreA; six NRE sequences, putatively involved in nitrogen repression; and six CCAAT boxes for the binding of the wide domain trimeric regulator AnCF [51]. By coupling the bidirectional pcbAB-pcbC promoter to reporter genes followed by sequential deletion of the promoter, three boxes (A, B, C) were defined as essential for optimal expression of the reporter genes [22]. Boxes A and B formed clear protein–DNA complexes, as demonstrated by changes in their electrophoretic mobility. Deletion of box A decreases transcriptional activity of the promoter by 41%. A palindromic heptanucleotide sequence, TTAGTAA, is the binding site for the transcriptional activator PTA1. Box A also contains four of the putative CreA binding sites, which have the consensus sequence SYGGRG. Box B contains a CreA site, three putative recognition sites for PacC, and is bordered by two NRE consensus sequences. However, little is known about the nature of the modulator proteins-such as PTA1-that control the expression of this promoter.

Regulation in prokaryotic b-lactam producers is completely different than in eukaryotes. In cephamycin-producing S. clavuligerus, a gene present in the cluster, ccaR, encodes a SARP-type protein (Streptomyces antibiotic regulatory proteins). These proteins are specific activators of antibiotic biosynthesis gene clusters in Streptomyces. Disruption of ccaR results in mutants unable to produce clavulanic acid and cephamycin C; the production of both compounds is restored by re-transformation with the entire gene [42]. Mobility-shift experiments indicate that CcaR binds both to its own promoter and to the bidirectional cefD-cmcI promoter. Binding of CcaR to the cefD-cmcI bidirectional promoter, which controls early, intermediate, and late steps of cephamycin biosynthesis, increases the expression of the genes involved [47]. The clavulanic-acid cluster contains the gene claR, encoding a LysR-type regulatory protein [40,44]. Mutants disrupted in claR do not produce clavulanic acid but are still able to produce cephamycin C. Binding of CcaR to promoters of genes involved in clavulanic acid has never been found, but quantitative-PCR expression of claR in ccaR-null mutants indicates that its expression is low (M.T. López-García, unpublished results), which suggests that a cascade of regulatory proteins connect CcaR and claR.

In addition to the region for CcaR binding, the promoter of ccaR contains a 26-bp ARE sequence located 890 bp upstream of the initial ATG start codon. ARE sequences (consensus sequence TNANAWACNNACYNNNCGGTTTKTTT) have been reported to be binding sites for Brp proteins (butyrolactones receptor proteins) [15]. The ARE sequence of S. clavuligerus is functional and binds a Brp protein expressed in Escherichia coli from the S. clavuligerus brp gene [48]. However, in addition to Brp, the sequence is a binding site for other proteins, as shown by gel mobility assays using cell extracts from a brp-null mutant of S. clavuligerus. The role of butyrolactones in Streptomyces antibiotic production has been reported in the case of streptomycin, pristinamycin, and tylosin [52], but butyrolactones do not appear to be synthesized in S. clavuligerus or other cephamycin C producers.

Concluding remarks

The presence of gene clusters for b-lactam antibiotic biosynthesis confers an ecological advantage to b-lactam-producing fungi or bacteria living in the soil. It appears that different fungal strains received large intron-less DNA fragments of the b-lactam gene cluster from soil bacteria, and subsequently attached other fungal genes (containing introns) to the fragment of the original bacterial antibiotic gene clusters to produce new and more potent antibiotics, such as penicillin G and cephalosporin C. These final compounds are of the utmost interest in the pharmaceutical industry. The penicillin-overproducing strains used industrially contain multiple copies of the penicillin gene cluster, which was amplified by random mutagenesis and selection during strain improvement programs. A worrying aspect of b-lactam production by fungi is that these gene clusters are expressed in the mycobiota involved in the ripening of cured meat products (e.g., salami), which results in a potential problem in that selection of b-lactam-resistant bacterial strains is thereby favored. The regulation of b-lactam biosynthesis is beginning to be elucidated and is clearly different in bacterial and fungal b-lactam producers.

Acknowledgements

This research was supported by grants BIO2003-3274 (P. Liras) and BIO2000-0060-P4-03 (J.F. Martín) from the CICYT, Madrid (Spain).

References

1. Aharonowitz Y, Cohen G, Martín JF (1992) Penicillin and cephalosporin biosynthetic genes: structure, organization, regulation, and evolution. Annu Rev Microbiol 46:461-495        [ Links ]

2. Aoki H, Sakai H, Kohsaka M, Konomi T, Hosoda J (1976) Nocardicin A, a new monocyclic beta-lactam antibiotic. I. Discovery, isolation and characterization. J Antibiot 29:492-500        [ Links ]

3. Bachmann BO, Li R, Townsend CA (1998) b-Lactam synthetase: a new biosynthetic enzyme. Proc Natl Acad Sci USA. 95:9082-9086        [ Links ]

4. Casqueiro J, Gutiérrez S, Bañuelos O, Hijarrubia MJ, Martín JF (1999) Gene targeting in Penicillium chrysogenum: disruption of the lys2 gene leads to penicillin overproduction. J Bacteriol 181:1181-1188.         [ Links ]

5. Coque JJR, Liras P, Láiz L, Martín JF (1991) A gene encoding lysine 6-aminotransferase which forms the b-lactam precursor a-aminoadipic acid is located in the cluster of cephamycin biosynthetic genes in Nocardia lactamdurans. J Bacteriol 173:6258-6264        [ Links ]

6. Coque JJR, Liras P, Martín JF (1993) Characterization and expression in Streptomyces lividans of cefD and cefE gene from Nocardia lactamdurans: the organization of the cephamycin gene cluster differs from that in Streptomyces clavuligerus. Mol Gen Genet 236:453-458        [ Links ]

7. Coque JJR, Liras P, Martín JF (1993) Gene for a b-lactamase, a penicillin-binding protein and a transmembrane protein are clustered with the cephamycin biosynthetic genes in Nocardia lactamdurans. EMBO J 12:631-639         [ Links ]

8. Coque JJR, Enguita FJ, Martín JF, Liras P (1995) A two-protein component 7a-cephem-methoxylase encoded by two genes of the cephamycin C cluster converts cephalosporin to 7-methoxycephalosporin C. J Bacteriol 177:2230-2235         [ Links ]

9. Coque JJR, Pérez-Llarena FJ, Enguita FJ, Fuente JL, Martín JF, Liras P (1995) Characterization of the cmcH genes of Nocardia lactamdurans and Streptomyces clavuligerus encoding a functional 3'-hydroxymethylcephem O-carbamoyltransferase for cephamycin biosynthesis. Gene 162:21-27         [ Links ]

10. Coulthurst SJ, Barnard AML, Salmond GPC (2005) Regulation and biosynthesis of carbapenem antibiotics in bacteria. Nat Rev Microbiol 3:295-306        [ Links ]

11. Díez B, Mellado E, Rodríguez M, Fouces R, Barredo JL (1997) Recombinant microorganisms for industrial production of antibiotics. Biotechnol Bioeng 55:216-226        [ Links ]

12. Enguita FJ, Liras P, Leitão AL, Martín JF (1996) Interaction of the two proteins of the methoxylation system involved in cephamycin C biosynthesis. J Biol Chem 271:33225-33230         [ Links ]

13. Esmahan C, Alvarez E, Montenegro E, Martín JF (1994) Catabolism of lysine in Penicillium chrysogenum leads to formation of 2-aminoadipic acid, a precursor of penicillin biosynthesis. Appl Environ Microbiol 60:1705-1710         [ Links ]

14. Fierro F, Barredo JL, Díez B, Gutierrez S, Fernández FJ, Martín JF (1995) The penicillin gene cluster is amplified in tandem repeats linked by conserved hexanucleotide sequences. Proc Natl Acad Sci USA 92:6200-6204        [ Links ]

15. Folcher M, Gaillard H, Nguyen LT, Nguyen KT, Lacroix P, Bamas-Jacques N, Rinkel M, Thompson CJ (2001) Pleiotropic functions of a Streptomyces pristinaespiralis autoregulator receptor in development, antibiotic biosynthesis, and expression of a superoxide dismutase. J Biol Chem 276:44297-44306        [ Links ]

16. Fuente JL de la, Rumbero A, Martín JF, Liras P (1997) D-1-Piperideine-6-carboxylate dehydrogenase, a new enzyme that forms a-aminoadipate in Streptomyces clavuligerus and other cephamycin C-producing actinomycetes. Biochem J 327:59-64         [ Links ]

17. Gunsior M, Breazeale SD, Lind AJ, Ravel J, Janc JW, Townsend CA (2004) The biosynthetic gene cluster for a monocyclic beta-lactam antibiotic, nocardicin A. Chem Biol 11:927-938        [ Links ]

18. Gutiérrez S, Velasco J, Fernández FJ, Martín JF (1992) The cefG gene of Cephalosporium acremonium is linked to the cefEF gene and encodes a deacetylcephalosporin C acetyltransferase closely related to homoserine O-acetyltransferase. J Bacteriol 174:3056-3064        [ Links ]

19. Imada A, Kitano K, Kintaka K, Muroi M, Asai M (1981) Sulfazecin and isosulfazecin, novel b-lactam antibiotics of bacterial origin. Nature 289:590-591        [ Links ]

20. Kim CF, Lee SKY, Price J, Jack RW, Turner G, Kong RYC (2003) Cloning and expression analysis of the pcbAB-pcbC b-lactam genes in the marine fungus Kallichroma tethys. Appl Environ Microbiol 69: 1308-1314        [ Links ]

21. Kimura H, Izawa M, Sumino Y (1996) Molecular analysis of the gene cluster involved in cephalosporin biosynthesis from Lysobacter lactamgenus YK90. Appl Microbiol Biotechnol 44:589-596        [ Links ]

22. Kosalková K, Marcos AT, Fierro F, Hernando-Rico V, Gutierrez S, Martín JF (2000) A novel heptameric sequence (TTAGTAA) is the binding site for a protein required for high level expression of pcbAB, the first gene of the penicillin biosynthesis in Penicillium chrysogenum. J Biol Chem 275:2423-2430         [ Links ]

23. Kovacevic S, Tobin MB, Miller JR (1990) The b-lactam biosynthesis genes for isopenicillin N epimerase and deacetoxycephalosporin C synthetase are expressed from a single transcript in Streptomyces clavuligerus. J Bacteriol 172:3952-3958        [ Links ]

24. Laich F, Fierro F, Martín JF (2002) Production of penicillin by fungi growing on food products: identification of a complete penicillin gene cluster in Penicillium griseofulvum and a truncated cluster in Penicillium verrucosum. Appl Environ Microbiol 68:1211-1219        [ Links ]

25. Laich F, Fierro F, Martín JF (2003) Isolation of Penicillium nalgiovense strains impaired in penicillin production by disruption of the pcbAB gene and application as starters on cured meat products. Mycol Res 107:717-726        [ Links ]

26. Láiz L, Liras P, Castro JM, Martín JF (1990) Purification and characterization of the isopenicillin N epimerase from Nocardia lactamdurans. J Gen Microbiol 136:663-671        [ Links ]

27. Liras P (1999) Biosynthesis and molecular genetics of cephamycins. Cephamycins produced by actinomycetes. Ant Leeuw 75:109-124         [ Links ]

28. Liras P, Rodríguez AR, Martín JF (1998) Evolution of genes for b-lactam antibiotics: a model for evolutive combinational assembly of new b-lactams. Int Microbiol 1:271-278        [ Links ]

29. Martín, JF (1992) Clusters of genes for the biosynthesis of antibiotics: regulatory genes and overproduction of pharmaceuticals. J Ind Microbiol 9:73-90        [ Links ]

30. Martín JF (1998) New aspects of gene and enzymes for b-lactam antibiotic biosynthesis. Appl Microbiol Biotechnol 50:1-15         [ Links ]

31. Martín JF, Gutiérrez S, Aparicio JF (2000) Secondary metabolites. In: Lederberg J (ed) Encyclopedia of Microbiology, vol. 4, 2nd ed. Academic Press, San Diego, CA, pp 213-236        [ Links ]

32. Martín JF, Gutiérrez S (1995) Genes for b-lactam antibiotic biosynthesis. Ant Leeuw 67:181-200        [ Links ]

33. Martín JF, Ullán RV, Casqueiro FJ (2004) Novel genes involved in cephalosporin biosynthesis: The three-component isopenicillin N epimerase system. In: Brakhage AA (ed) Advances in Biochemical Engineering/Biotechnology 88:91-110        [ Links ]

34. Martín JF, Casqueiro J, Liras P (2005) Secretion systems for secondary metabolites: how producer cells send out messages of intercellular communication. Curr Opin Microbiol 8:282-293        [ Links ]

35. Martín de Valmaseda EM, Campoy S, Naranjo L, Casqueiro J, Martín JF (2005) Lysine is catabolized to 2-aminoadipic acid in Penicillium chrysogenum by an w-aminotransferase and to saccharopine by a lysine 2-ketoglutarate reductase. Characterization of the w-aminotransferase. Mol Genet Genomics 274:272-282        [ Links ]

36. McGowan SJ, Barnard AML, Bosgelmez G, Sebaihia M, Simpson NJL, Thomson NR, Todd DE, Welch M, Whitehead NA, Salmond GPC (2005) Carbapenem antibiotic biosynthesis in Erwinia carotovora is regulated by physiological and genetic factors modulating the quorum sensing-dependent control pathway. Mol Microbiol 55:526-545         [ Links ]

37. Naranjo L, Lamas-Maceiras M, Ullán RV, Campoy S, Teijeira F, Casqueiro J, Martín JF (2005) Characterization and expression studies of the oat1 gene of Penicillium chrysogenum encoding an omega-aminotransferase: Induction by L-lysine, L-ornithine and L-arginine and repression by ammonium. Mol Genet Genomics 274:283-294        [ Links ]

38. Newton GG, Abraham EP (1955) Cephalosporin C, a new antibiotic containing sulphur and d-alpha-aminoadipic acid. Nature 75:548        [ Links ]

39. Núñez LE, Méndez C, Braña AF, Blanco G, Salas JA (2003) The biosynthetic gene cluster for the b-lactam carbapenem thienamycin in Streptomyces cattleya. Chem Biol 10:301-311        [ Links ]

40. Paradkar AS, Aidoo KA, Jensen SE (1998) A pathway-specific transcriptional activator regulates late steps of clavulanic acid biosynthesis in Streptomyces clavuligerus. Mol Microbiol 27: 831-843         [ Links ]

41. Pérez-Llarena FJ, Martín JF, Coque JJR, de la Fuente JL, Galleni M, Frère J-M, Liras P (1996) The bla gene of the cephamycin cluster of Streptomyces clavuligerus encodes a class A b-lactamase of low enzymatic activity. J Bacteriol 179:6035-6040         [ Links ]

42. Pérez-Llarena FJ, Liras P, Rodríguez-García A, Martín JF (1997) A regulatory gene (ccaR) required for cephamycin and clavulanic acid production in Streptomyces clavuligerus: Amplification results in overproduction of both b-lactam compounds. J Bacteriol 179:2053-2059        [ Links ]

43. Pérez-Llarena FJ, Rodríguez-García A, Enguita FJ, Martín JF, Liras P (1998) The pcd gene encoding piperideine-6-carboxylate dehydrogenase involved in biosynthesis of a-aminoadipic acid is located in the cephamycin cluster of Streptomyces clavuligerus. J Bacteriol 180: 4753-4756        [ Links ]

44. Pérez-Redondo R, Rodríguez-García A, Martín JF, Liras P (1998) The claR gene of Streptomyces clavuligerus, encoding a LysR-type regulatory protein controlling clavulanic acid biosynthesis, is linked to the clavulanate-9-aldehyde reductase (car) gene. Gene 211:311-321        [ Links ]

45. Reeve AM, Breazeale SD, Townsend CA (1998) Purification, characterization, and cloning of an S-adenosylmethionine-dependent 3-amino-3-carboxypropyltransferase in nocardicin biosynthesis. J Biol Chem 273:30695-30703        [ Links ]

46. Roach PL, Clifton IJ, Fülöp V, Harlos K, Barton GJ, Hajdu J, Andersson I, Schofield CJ, Baldwin JE (1995) Crystal structure of isopenicillin N synthase is the first from a new structural family of enzymes. Nature 375:700-704        [ Links ]

47. Santamarta I, Pérez-Redondo R, Lorenzana LM, Martín JF, Liras P (2005) Different proteins bind to the butyrolactone receptor protein ARE sequence located upstream of the regulatory ccaR gene of Streptomyces clavuligerus. Mol Microbiol 56:824-835 (Erratum: 57:1175-1176)        [ Links ]

48. Santamarta I, Rodríguez-García A, Pérez-Redondo R, Martín JF, Liras P (2002) CcaR is an autoregulatory protein that binds to the ccaR and cefD-cmcI promoters of the cephamycin C-clavulanic acid cluster in Streptomyces clavuligerus. J Bacteriol 184:3106-3113         [ Links ]

49. Sleeman MC, Schofield CJ (2004) Carboxymethylproline synthase (CarB), an unusual carbon-carbon bond-forming enzyme of the crotonase superfamily involved in carbapenem biosynthesis. J Biol Chem 279:6730-6736         [ Links ]

50. Sohn YS, Nam DH, Ryu DDY (2001) Biosynthetic pathway of cephabacins in Lysobacter lactamgenus: molecular and biochemical characterization of the upstream region of the gene clusters for engineering of novel antibiotics. Metab Eng 3:380-392        [ Links ]

51. Steidl S, Papagiannopoulos P, Litzka O, Andrianopoulos A, Davis MA, Brakhage AA, Hynes MJ (1999) AnCF, the CCAAT binding complex of Aspergillus nidulans, contains products of the hapB, hapC, and hapE genes and is required for activation by the pathway-specific regulatory gene amdR. Mol Cell Biol 19:99-106        [ Links ]

52. Stratigopoulos G, Gandecha AR, Cundliffe E (2002) Regulation of tylosin production and morphological differentiation in Streptomyces fradiae by TylP, a deduced g-butyrolactone receptor. Mol Microbiol 45:735-744        [ Links ]

53. Tilburn J, Sarkar S, Widdick DA, Espeso EA, Orejas M, Mungroo J, Peñalva MA, Arst HN Jr (1995) The Aspergillus PacC zinc finger transcription factor mediates regulation of both acid- and alkaline-expressed genes by ambient pH. EMBO J 14:779-790        [ Links ]

54. Tobin MB, Kovacevic S, Madduri K, Hoskins JA, Skatrud PL, Vining LC, Stuttard C, Miller JR (1991) Localization of the lysine e-aminotransferase (lat) and d-(L-a-aminoadipyl)-L-cysteinyl-D-valine synthetase (pcbAB) genes from Streptomyces clavuligerus and production of lysine e-aminotransferase activity in Escherichia coli. J Bacteriol 173: 6223-6229        [ Links ]

55. Ullán RV, Casqueiro J, Bañuelos O, Fernández FJ, Gutiérrez S, Martín JF (2002) A novel epimerization system in fungal secondary metabolism involved in the conversion of isopenicillin N into penicillin N in Acremonium chrysogenum. J Biol Chem 277:46216-46225         [ Links ]

56. Usui S, Yu CA (1989) Purification and properties of isopenicillin N epimerase from Streptomyces clavuligerus. Bioch Biophys Acta 999: 78-85         [ Links ]

57. Weigel BJ, Burgett SG, Chen VJ, Skatrud PL, Frolik CA, Queener SW, Ingolia TD (1988) Cloning and expression in Escherichia coli of isopenicillin N synthetase genes from Streptomyces lipmanii and Aspergillus nidulans. J Bacteriol 170:3817-3826        [ Links ]

Corresponding author
J.F. Martín
Instituto de Biotecnología (INBIOTEC)
Parque Científico de León
Av. Real, 1 ]]> 24006 León, Spain
Tel. +34-987291505. Fax +34-987291506
E-mail: degjmm@unileon.es

Received 15 December 2005
Accepted 23 January 2006